We are IntechOpen, the world's leading publisher of Open Access books Built by scientists, for scientists

3,800+

Open access books available

116,000+

International authors and editors

120M+

Downloads

Our authors are among the

Top 1%

most cited scientists

12.2%

Contributors from top 500 universities

Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI)

## Interested in publishing with us? Contact book.department@intechopen.com

Numbers displayed above are based on latest data collected. For more information visit www.intechopen.com

## **Meet the author**

Francisco Bulnes, doctorate in mathematics sciences by UNAM with doctoral dissertation: "Some Relations between the Vogan-Zuckerman Cohomological Induction and Langlands Classification". Active member of research doctoral seminar "Representations Theory of Real Reductive Lie Groups"(2000-2012). Many works published, badges and recognitions received for this

seminar. Mathematics professor in many universities along 20 years. Coordinator of mathematics (2000-2002). Specialist in: Mathematical Physics, Integral Geometry, Field Theory, and Infinite Lie Theory. Principal and scientific advisor in diverse international and national scientific and academic events in the mathematics area. Full member of ANS (American Nano Society). Author of many texts and specialized books all with ISBN of mathematics for scientist and engineers, and published for many universities inside and without from Mexico. Principal of Appliedmath (2005-2008). Author of diverse international and national referred Journal articles in many areas of pure and applied mathematics (http://www.scirp. org/journal/PaperInformation.aspx?paperID=20338, http://www.scirp. org/journal/apm/) for example. Quantum Electrodynamics expert by the (AIP) American Institute of Physics. Co-Principal of doctoral seminar of Lie groups in IM/UNAM and fellowships and other distinctions in the doctoral Seminar of research. International projects judge of mathematics to CONACYT/Mexico. Book-Editor for other mathematicians. Scientific consul in research on mathematics and advanced systems with some countries as Cuba, USA and Russia. Actually the Dr. Bulnes is researching on nanotechnology, on the classification problem of irreducible unitary representations and on one new theory of the Universe with MIT and Harvard groups. Editor-in-Chief of Journal on Photonics and Spintronics, of USA, http://www.researchpub.org/journal/jps/editorial%20board.html. Has been involved in several editorial advisory boards of journals in mathematics. Actually is Principal of Research TESCHA from 2009. His interest areas are: mathematical physics, cohomology of cycles (motivic and relative), integral geometry, K-theory, moduli spaces and nanotechnology.

## Contents



### X Contents


To my master *Felix Recillas Juarez* (1918-2010), *(PhD, Princeton, New Jersey, USA, 1948)*  with all gratitude and my recognition for their wisdom and friendship

## Preface

The following work is the result of diverse explains, courses and conferences dictated by the *Francisco Bulnes, PhD*, in many international academic places and diverse national forums along thirteen years inside the research in integral theories of generalized measures, integral operators cohomology and Integration on Haar spaces and on spaces of infinite dimension. Many of these talks, were given in Institute of Mathematics of National Autonomous University of Mexico (IM/UNAM), into the seminar of "*Representation Theory of Reductive Lie Groups*" (1999-2006) and after, in the Department of Research in Mathematics and Engineering of Technological Institute of High Studies of Chalco (DIMI/TESCHA) among the years 2004-2012.

The purpose is to present a complete course on these topics and to establish some orbital applications of the integral on Lie groups to the harmonic analysis and induced representations in representation theory.

Some other applications in this respect they will be given for some integral models of the cosmos.

> Department of Research in Mathematics and Engineering, TESCHA Institute of Mathematics UNAM, Mexico

## **Introduction**

## **I. 1. Introduction**

In the study of the theory of *irreducible unitary representations*, is necessary to analyze and demonstrate diverse results on integral orbital of functions belonging to the cohomology Hi (g, K; V V\*), and that it is wanted they belong to the *L2(G)-cohomology* of their reducible unitary representations called *discrete series*. Then is necessary consider the *Frèchet space* I(G), and analyze the *2-integrability* to the fibers of the space G/K, in spaces or locally compact components of G/K. For it will be useful the invariance of the corresponding measures of Haar under the actions of Ad(G), and the corresponding images of the *Harish-Chandra transform* on the space of functions Ia,b(G).

Likewise, we will obtain a space in *cuspidal forms* that is an introspection of the class of the discrete series in the whole space G.

This *harmonic analysis* in the context of the space in cuspidal forms is useful in the exploration of the behavior of *characters* for those(g, K)-*modules* Hi (g, K; V V\*) and also for the generalization of the *integral formula of Plancherel* on *locally compact spaces* of G.

The generalization of the Plancherel formula is useful for the study of the functions on symmetrical spaces.

## **I. 2. Generalized spheres on Lie groups**

We consider to G = L/Hg, a *homogeneous space* with origin o = {Hg}. Given goG, let LgO be the subgroup of G, letting go fix, that is to say; the subgroup of isotropy of G, in go.

Def. I. 2.1. A *generalized sphere* is an *orbit* Lgog, in G, of some point gG, under the subgroup of isotropy in some point goG.

In the case of a Lie group the generalized spheres are the left translations (or right) of their *conjugated classes*.

We assume that Hg, and each Lgo, is *unimodular*. But is considering Lgog = Lgo/(Lgo)g, such that (Lgo)g, be unimodular then the orbit Lgog, have an invariant measure determined except for a constant factor. Then are our interest the following general problems:

	- a. To determine a function f, on G, in terms of their orbital integrals on generalized spheres.

In this problem the essential part consist in the normalization of invariant measures on different orbits.

If is the case in that Hg, is compact, the problem A), is trivial, since each orbit Lgog, have finite invariant measure such that f(go)is given as the limit when g go, of the variation of f, on Lgog.

### **I. 2.1. Orbits**

Suppose that to every go*G*, exist an open set *L*g0-invariant Cgo*G*, containing go in their classes such that to each gCg0, the group of isotropy (*L*go)g, is compact. The invariant measure on the orbit *L*g0g (go*G*, gCgo) can be normalized consistently as follows: We fix a Haar measure dgo, on Lo (*H*g = *L*o). If go= g o, we have *L*g0 = g*L*og-1,and we can to carry on dgo, to the measure dgg0,on *L*go through of the conjugation z gzg-1 (z*L*o). Since dgo, is biinvariant, dggo,g, is independent of the election of g satisfying go = g o, the which is biinvariant. Since (*L*go)g, is compact, this have an only measure of Haar dggo,g, with total measure 1 and reason why dgo, and dggo,g, determine canonically an invariant measure on the orbit*L*g0g = *L*go/(*L*go)g.

Reason why also the following problem can to establish:

b. To express to f(go), in terms of the integralsLg0gf(p)d(p), gCgo. that is to say, the calculus of the orbital integrals on those measurable open sets called orbits.

### **I. 3. Invariant measures on homogeneous spaces**

Let *G*, a locally compact topological group. Then a left invariant measure on *G*, is a positive measure, dg, on *G*, such that

$$\int \text{cf(xg)dg} = \int \text{cf(g) dg} \tag{1.3.1}$$

x*G*, and all fCc(*G*). If *G*, is separable then is acquaintance (Haar theorem) that such measure exist and is unique except a multiplicative constant.

If *G*, is a Lie group with a finite number of components then a left invariant measure on *G*, can be identified with a left invariant n-form on *G* (where dim *G* = *n*). If , is a left invariant non-vanishing *n*-form on *G*, then the identification is implemented by the integration with regard to , using the canonical method of differential geometry. If *G*, is compact then we can (if is not specified) to use normalized left measures. This is those whose measure total is 1.

If dg is a left invariant measure and if x*G*, then we can define a new left invariant measure on *G*, x, as follows:

$$
\mu\_{\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\tiny\rm{\varepsilon}}}}}}}}}\mu\_{\rm{\tiny\rm{\rm{\tiny\rm{\rm{\tiny\rm{\varepsilon}}}}}}}\cdots}\mu\_{\rm{\rm{\tiny\rm{\rm{\tiny\rm{\varepsilon}}}}}}\cdots\mu\_{\rm{\rm{\tiny\rm{\rm{\varepsilon}}}}}\cdots\mu\_{\rm{\rm{\varepsilon}}}\tag{1.3.2}
$$

The uniquely of the left invariant measure implies that

$$
\mu\_{\rm t}(\mathbf{f}) = \delta(\mathbf{x}) \Big|\varepsilon \mathbf{f}(\mathbf{x} \mathbf{g}) \mathrm{d}\mathbf{g},\tag{1.3.3}
$$

with , a function of x, which is usually called the modular function of *G*. If , is identically equal to 1,then we say that *G*, is unimodular. If *G* is then unimodular we can call to a left invariant measure (which is automatically right invariant) invariant. It is not difficult affirm that , is a continuous homomorphism of *G*, in the multiplicative group of positive real numbers. This implies that if *G*, is compact then *G*, is unimodular.

If *G*, is a Lie group, the modular function of *G*, is given by the following formula:

$$\delta(\mathbf{x}) = |\det \, \mathrm{Ad}(\mathbf{x})| \,\tag{1.3.4}$$

where Ad, is the usual adjunct action of *G*, on their Lie algebra.

Let M, be a soft manifold and be , their form of volume. Let *G*, be a Lie group acting on M. Then (g\*)x = c(g, x)x, each g*G*, and xM. If is left as exercise verify that c satisfies the cocycle relationship

$$\mathbf{c}(\mathbf{g}\mathbf{h},\mathbf{x}) = \mathbf{c}(\mathbf{g},\mathbf{h}\mathbf{x})\mathbf{c}(\mathbf{h},\mathbf{x}) \; \forall \; \mathbf{h}, \; \mathbf{g} \in \mathbf{G}, \; \mathbf{x} \in \mathbf{M}, \tag{\text{I.3.5}}$$

We write as M f(x)dx, to Mf. The usual formula of change of variables implies that

$$\int \mathsf{M}\,\mathsf{f}(\mathsf{g}\,\mathsf{x}) |\mathsf{c}(\mathsf{g},\mathsf{x})| \,\mathsf{d}\mathsf{x} = \int \mathsf{c}\,\mathsf{f}(\mathsf{x}) \,\mathsf{d}\mathsf{x},\tag{\mathsf{I.3.6}}$$

to fCc(G), and gG.

2 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

spheres.

different orbits.

**I. 2.1. Orbits** 

the orbit*L*g0g = *L*go/(*L*go)g.

measure, dg, on *G*, such that

whose measure total is 1.

Reason why also the following problem can to establish:

b. To express to f(go), in terms of the integralsLg0gf(p)d(p), gCgo.

**I. 3. Invariant measures on homogeneous spaces** 

measure exist and is unique except a multiplicative constant.

on Lgog.

a. To determine a function f, on G, in terms of their orbital integrals on generalized

In this problem the essential part consist in the normalization of invariant measures on

If is the case in that Hg, is compact, the problem A), is trivial, since each orbit Lgog, have finite invariant measure such that f(go)is given as the limit when g go, of the variation of f,

Suppose that to every go*G*, exist an open set *L*g0-invariant Cgo*G*, containing go in their classes such that to each gCg0, the group of isotropy (*L*go)g, is compact. The invariant measure on the orbit *L*g0g (go*G*, gCgo) can be normalized consistently as follows: We fix a Haar measure dgo, on Lo (*H*g = *L*o). If go= g o, we have *L*g0 = g*L*og-1,and we can to carry on dgo, to the measure dgg0,on *L*go through of the conjugation z gzg-1 (z*L*o). Since dgo, is biinvariant, dggo,g, is independent of the election of g satisfying go = g o, the which is biinvariant. Since (*L*go)g, is compact, this have an only measure of Haar dggo,g, with total measure 1 and reason why dgo, and dggo,g, determine canonically an invariant measure on

that is to say, the calculus of the orbital integrals on those measurable open sets called orbits.

Let *G*, a locally compact topological group. Then a left invariant measure on *G*, is a positive

x*G*, and all fCc(*G*). If *G*, is separable then is acquaintance (Haar theorem) that such

If *G*, is a Lie group with a finite number of components then a left invariant measure on *G*, can be identified with a left invariant n-form on *G* (where dim *G* = *n*). If , is a left invariant non-vanishing *n*-form on *G*, then the identification is implemented by the integration with regard to , using the canonical method of differential geometry. If *G*, is compact then we can (if is not specified) to use normalized left measures. This is those

*G*f(xg)dg = *G*f(g) dg, (I. 3.1)

Let *H*, be a closed subgroup of G. Be M = G/H. We assume that G, have a finite number of connect components. A *G*-invariant measure, dx, on M is a measure such that

$$\int \mathbf{J} \mathbf{f}(\mathbf{g} \mathbf{x}) d\mathbf{x} = \int\_{\mathcal{M}} \mathbf{f}(\mathbf{x}) d\mathbf{x}, \quad \forall \ \mathbf{f} \in \mathbf{C}\_{\mathbf{c}}(\mathbf{G}), \ \mathbf{g} \in \mathbf{G} \tag{\text{I.3.7}}$$

If dx, comes of a form of volume on M, then (I. 3.7), is the same, which is equal to that c(g, x) = 1 gG, xM.

If M, is a soft manifold then is well acquaintance that M, have a form of volume or M, have a double covering that admit a form of volume. To rising of functions to the double covering (if it was necessary) one can integrate relatively to a form of volume on any manifold. Come back to the situation M = *G*/*H*, is not difficult demonstrate that M, admit a measure *G*invariant if and only if the unimodular function of *G*, restricted to *H*, is equal to the unimodular function of H. Under this condition, a measure G-invariant on M is constructed

as follow: Be g, the Lie algebra of *G*, and be h, the subalgebra of g, corresponding to *H*. Then we can to identify the tangent space of 1*H* to M with g/h. The adjunct action of *H*, on g, induces to action Ad , of *H*, on g/h. The condition mentioned to obtain an identity in (I. 3. 7) tell us that Ad (h) = 1 hH. Thus if *H*0, is the identity component of *H*, (as is usual) and if , is a element not vanishing of m(g/h)\*[3] (m = dim *G*/*H*) it is can to translate , to a form of *G*-invariant volume on *G*/*H*0.

Therefore for rising of functions of M, to *G*/*H*0, is had an invariant measure on M. But the Fubini theorem affirms that we can normalize dg, dh and dx, such that

$$\text{f.f.f(g)dg} = \int\_{C^{\text{H}}} \text{(}\text{f}\_{\text{H}}\text{f(gh) dh) d(gH)}, \quad \text{f.e.C.(G)}\tag{1.3.8}$$

Let *G*, be a Lie group with a finite number of connects components. Let *H*, be a closed subgroup of *G*, and let dh, be a selection of left invariant measure on *H*. The following result is used in the calculus of measures on homogeneous spaces.

**Lemma I. 2.1.**If f, is a compactly supported continuous function on *H*/G, (note the change to the right classes) then it exists, g, a continuous function supported compactly on such *G*, that

$$\text{ft(Hx)} = \bigcup \text{cg(hx)} \text{ dh},\tag{I.3.9}$$

This result is usually demonstrated using a "partition of the unit" as principal argument.

For details of demonstration see 1.

Let *G*, be a Lie group and be A, and B, subgroups in *G*, such that A, and B, are compact and such that G = AB. The following result is used to the study of induced representations and classes of induced cohomology.

**Lemma I. 2.2.**We asume that *G*, is unimodular. If da, is a left invariant measure on A, and db, is a left invariant measure on B, then we can elect an invariant measure, dg, on G, such that

$$\int \text{cf(g) dg} = \int\_{\Lambda \times \mathbb{R}} \text{f(ab) da db}, \quad \text{para} \quad \text{f} \in \text{C}\_{\text{f}}(\mathbb{G}) \tag{\text{I.3.10}}$$

*Proof*: Consult 2.

In the following section we will explain basic questions on invariant measures on homogeneous spaces. With it will stay clear the concept and use of normalized measures.

Let G, be a Lie group with Lie algebra g; let H, a closed subgroup with Lie algebra h g. Each xG, gives rise to an analytic diffeomorphism

(x) :gHxgH, (I. 3.11)

of G/H, onto itself. Let , denote the natural mapping of G, onto G/H, and put o = (e). If hH, (d(h))o, is an endomorphism of the tangent space (G/H)o. For simplicity, we shall write d(h), instead of (d(h))o, and d, instead of (d)e [4].

### **Lemma I. 2.3.**

$$\det(\mathrm{d}\,\tau(\mathrm{h})) = \frac{\det \mathrm{Ad}\_{\mathrm{G}}(\mathrm{h})}{\det \mathrm{Ad}\_{\mathrm{H}}(\mathrm{h})},\tag{\text{I. 3.12}}$$

hH.

4 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Fubini theorem affirms that we can normalize dg, dh and dx, such that

is used in the calculus of measures on homogeneous spaces.

induces to action Ad

*G*-invariant volume on *G*/*H*0.

For details of demonstration see 1.

classes of induced cohomology.

Each xG, gives rise to an analytic diffeomorphism

*Proof*: Consult 2.

tell us that Ad

as follow: Be g, the Lie algebra of *G*, and be h, the subalgebra of g, corresponding to *H*. Then we can to identify the tangent space of 1*H* to M with g/h. The adjunct action of *H*, on g,

, is a element not vanishing of m(g/h)\*[3] (m = dim *G*/*H*) it is can to translate , to a form of

Therefore for rising of functions of M, to *G*/*H*0, is had an invariant measure on M. But the

Let *G*, be a Lie group with a finite number of connects components. Let *H*, be a closed subgroup of *G*, and let dh, be a selection of left invariant measure on *H*. The following result

**Lemma I. 2.1.**If f, is a compactly supported continuous function on *H*/G, (note the change to the right classes) then it exists, g, a continuous function supported compactly on such *G*, that

This result is usually demonstrated using a "partition of the unit" as principal argument.

Let *G*, be a Lie group and be A, and B, subgroups in *G*, such that A, and B, are compact and such that G = AB. The following result is used to the study of induced representations and

**Lemma I. 2.2.**We asume that *G*, is unimodular. If da, is a left invariant measure on A, and db, is a left invariant measure on B, then we can elect an invariant measure, dg, on G, such that

In the following section we will explain basic questions on invariant measures on homogeneous spaces. With it will stay clear the concept and use of normalized measures.

Let G, be a Lie group with Lie algebra g; let H, a closed subgroup with Lie algebra h g.

(x) :gHxgH, (I. 3.11)

, of *H*, on g/h. The condition mentioned to obtain an identity in (I. 3. 7)

G f(g)dg = G/H H f(gh) dh d(gH), fCc(G) (I. 3.8)

f(Hx) = *G*g(hx) dh, (I. 3.9)

Gf(g) dg = <sup>A</sup>B f(ab) da db, para fCc(G) (I. 3.10)

(h) = 1 hH. Thus if *H*0, is the identity component of *H*, (as is usual) and if

*Proof*. *d*, is a linear mapping of g, onto (G/H)o, and has kernel h. Let m, be any subspace of g, such that g = h + m, (direct sum). Then *d*, induces an isomorphism of m, onto (G/H)o. Let Xm. Then AdG(h)X = dRh1dLh(X). Since *R*h = , hH, and *L*g= (g) , gG, we obtain

$$d\pi \circ \text{Adc(h)}\!\!\!\!\!\!\/ = \text{d}\pi(\text{h}) \cdot \text{d}\pi(\text{X}),\tag{\text{I.3.13}}$$

The vector AdG(h)X, decomposes according to g = h + m,

$$\text{Adv}(\mathbf{h})\mathbf{X} = \mathbf{X}(\mathbf{h})\_{\mathbb{H}} + \mathbf{X}(\mathbf{h})\_{\mathbb{H}} \tag{\text{I.3.14}}$$

The endomorphism

$$\mathbf{A} \mathbf{h} \colon \mathbf{X} \to \mathbf{X} \text{(fh)}\_{\mathsf{H}} \tag{\text{I.3.15}}$$

of m, satisfies

$$d\pi \circ \text{Ad}(\mathcal{X}) = d\pi(\mathbf{h}) \cdot d\pi(\mathcal{X}),\tag{\mathcal{I}.3.16}$$

Xm, so det Ah = det(d(h)). For other side,

$$\exp\text{ Adc(h)tT = h\exp tTh^{-1} = \exp\exp\text{Ada(h)tT,}}\tag{1.3.17}$$

for tR, Th. Hence AdG(h)T = AdH(h)T, so

$$\det \operatorname{Adc}(\mathbf{h}) = \det \operatorname{A}\_{\mathbf{h}} \det \operatorname{Ad} \mathfrak{u}(\mathbf{h}) \tag{\text{I.3.18}}$$

and the lemma is proved.

**Proposition I. 2.1.** Let m = dim G/H. The following conditions are equivalent:

i. G/H, has a nonzero G-invariant m-form ;

ii. det AdG(h) = det AdH(h), for hH.

If these conditions are satisfied, then G/H, has a G-invariant orientation and the G-invariant m-form , is unique up to a constant factor.

*Proof*. Let , be a G-invariant m-form on G/H, 0. Then the relation (h)\*= , [3] at the point *o*, implies det(d(h)) = 1, so ii), holds. For other side, let X1, , Xm, be a basis of (G/H)*o*, and let 1, , m, be the linear functions on (G/H)*o*, determined by <sup>i</sup> (Xj) = ij. Consider the element1m, in the Grassmann algebra of the tangent space (G/H)*o*. The condition ii), implies that det(d(h)) = 1, and the element 1m, is invariant under the linear transformation d(h). It follows that exists a unique G-invariant m-form , on G/H, such that *o*= 1m. If \*, is another G-invariant m-form on G/H, then \* = f, where fC(G/H). Owing to the G-invariance, f = *constant*.

Assuming i), let : p (x1(p), , xm(p)), be a system of coordinates on an open connected neighborhood U, of *o*G/H, on which , has an expression

$$\alpha \circ \iota = \mathsf{F}(\mathsf{x}\_{1}, \dots, \mathsf{x}\_{m}) \, \mathsf{d}\mathsf{x}\_{1} \wedge \dots \wedge \mathsf{d}\mathsf{x}\_{m}$$

With F > 0, The pair ((g)U, (g1)), is a local chart on a connected neighborhood of g oG/H. We put ((g1))(p) = (y1(p), , ym(p)), for p(g)U. Then the mapping

$$\mathsf{τ(g)}: \mathsf{U} \to \mathsf{τ(g)}\mathsf{U}\_{\mathsf{'}}$$

has expression

$$(\mathbf{y}\_1 \,\, \cdots \,\_{\prime} \mathbf{y}\_{\mathrm{m}}) = (\mathbf{x}\_{1\prime} \,\, \cdots \,\_{\prime} \mathbf{x}\_{\mathrm{m}}).$$

On (g)U, , has an expression

$$\mathbf{q}\_{\mathbf{\dot{\tau}}(\mathbf{\dot{\beta}})\mathbf{U}} = \mathbf{G}(\mathbf{y}\_1, \dots, \mathbf{y}\_m) \text{ фу} \wedge \dots \wedge \mathbf{dy}\_{\mathbf{m}\_\ell}$$

and since q =(g)\*(g)q, we have for qU (g)U,

$$\alpha\_{\mathbb{M}} = \mathbf{G}(\mathbf{y}\_1(\mathbf{q}), \dots, \mathbf{y}\_{\mathbb{M}}(\mathbf{q})) \text{ (dy} \\ \boldsymbol{\iota} \wedge \dots \wedge \mathbf{dy}\_{\mathbb{M}})\_{\mathbb{q}} = \mathbf{G}(\mathbf{x}(\mathbf{q}), \dots, \mathbf{x}\_{\mathbb{m}}(\mathbf{q})) \text{ (dx} \\ \boldsymbol{\iota} \wedge \dots \wedge \mathbf{dx}\_{\mathbb{m}})\_{\mathbb{q}}$$

Hence F(x1(q), , xm(q)) = G(x1(q), , xm(q)), and

$$F(\mathbf{x} \circ (\mathbf{q}), \dots, \mathbf{x}\_{\mathbf{m}}(\mathbf{q})) = F(\mathbf{y} \circ (\mathbf{q}), \dots, \mathbf{y}\_{\mathbf{m}}(\mathbf{q})) [\hat{\varepsilon}(\mathbf{y} \circ (\mathbf{q}), \dots, \mathbf{y}\_{\mathbf{m}}(\mathbf{q})) [\hat{\varepsilon}(\mathbf{x} \circ (\mathbf{q}), \dots, \mathbf{x}\_{\mathbf{m}}(\mathbf{q}))]],$$

which shows that the Jacobian of the mapping (o(g1)) o , is positive. Consequently, the collection ((g)U, o(g1))gG, of local charts turns G/H, into an oriented manifold and each (g), is orientation preserving. Then G-invariant form , now gives rise to an integral f, which is invariant in the sense that

$$\int\_{C\mathbb{H}} \mathbf{f} \mathbf{o} \mathbf{o} = \int\_{C\mathbb{H}} (\mathbf{f} \circ \mathbf{\tau}(\mathbf{g})) \mathbf{o}, \quad \forall \quad \mathbf{g} \in G.$$

However, just as the Riemannian measure did not require orientability; an invariant measure can be construct on G/H, under a condition which is slightly more general than (ii). The projectiveP2, will, for example, satisfy this condition whereas it does not satisfy (ii). We recall that a measure , on G/H, is said to be invariant (or more precisely *G*-invariant) if (f o (g)) = (f), for all g*G*.

**Theorem I. 2.1.** Let *G*, be a Lie group and H, a closed subgroup. The following relation is satisfied

$$|\det \operatorname{Adc}(\mathbf{h})| = |\det \operatorname{Ad} \omega(\mathbf{h})| \qquad \mathbf{h} \in \mathbf{H},\tag{\text{I.3.19}}$$

Is a necessary and sufficient condition for the existence of a G-invariant positive measure on G/H. This measure dg*H*, is unique (up to a constant factor) and

$$\text{lcf}(\text{g})\text{dg} = \big|\_{\text{C}\models\text{l}} \text{f}(\text{gh})\text{dh}\big|\text{dg}\big|, \quad \forall \text{ f} \in \text{C}\_{\text{f}}(\text{G}), \tag{\text{I.3.20}}$$

if the left invariant measures dg, and dh, are suitably normalized.

*Proof* [6], [1]

6 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

neighborhood U, of *o*G/H, on which , has an expression

Owing to the G-invariance, f = *constant*.

has expression

On (g)U, , has an expression

and since q =(g)\*(g)q, we have for qU (g)U,

Hence F(x1(q), , xm(q)) = G(x1(q), , xm(q)), and

f, which is invariant in the sense that

(g)) = (f), for all g*G*.

and let 1, , m, be the linear functions on (G/H)*o*, determined by <sup>i</sup>

*Proof*. Let , be a G-invariant m-form on G/H, 0. Then the relation (h)\*= , [3] at the point *o*, implies det(d(h)) = 1, so ii), holds. For other side, let X1, , Xm, be a basis of (G/H)*o*,

element1m, in the Grassmann algebra of the tangent space (G/H)*o*. The condition ii), implies that det(d(h)) = 1, and the element 1m, is invariant under the linear transformation d(h). It follows that exists a unique G-invariant m-form , on G/H, such that *o*= 1m. If \*, is another G-invariant m-form on G/H, then \* = f, where fC(G/H).

Assuming i), let : p (x1(p), , xm(p)), be a system of coordinates on an open connected

U = F(x1,,xm) dx1dxm, With F > 0, The pair ((g)U, (g1)), is a local chart on a connected neighborhood of g

(g) : U (g)U,

(y1, , ym) = (x1, , xm).

(g)U = G(y1,, ym) dy1dym,

q = G(y1(q),, ym(q))dy1dym)q = G(x1(q), , xm(q))dx1dxm)q,

F(x1(q), , xm(q)) = F(y1(q),, ym(q))[(y1(q),, ym(q))**/**(x1(q), , xm(q))], which shows that the Jacobian of the mapping (o(g1)) o , is positive. Consequently, the collection ((g)U, o(g1))gG, of local charts turns G/H, into an oriented manifold and each (g), is orientation preserving. Then G-invariant form , now gives rise to an integral

G/H fG/H (f o (g))g*G*. However, just as the Riemannian measure did not require orientability; an invariant measure can be construct on G/H, under a condition which is slightly more general than (ii). The projectiveP2, will, for example, satisfy this condition whereas it does not satisfy (ii). We recall that a measure , on G/H, is said to be invariant (or more precisely *G*-invariant) if (f o

oG/H. We put ((g1))(p) = (y1(p), , ym(p)), for p(g)U. Then the mapping

(Xj) = ij. Consider the

## **Integrals, Functional and Special Functions on Lie Groups and Lie Algebras**

### **II. 1. Spherical functions**

Let P = P, be the minimal parabolic subgroup of G, with the Langlands decomposition

$$P = {}^{0}\text{MAN}\_{\prime}\tag{\text{II. 1.1}}$$

If (, H ), is an irreducible unitary representation of 0M, and if (aC)\*, then (, H), can denote the corresponding representation in principal serie. H, is equivalent with IK() = H . Indeed, if H, is a representation of K, then exist G, such that a\*= a\*, to a\*, the dual algebra of the algebra a p, since always there is a maximal Abelian subalgebra in p. Then IK() = IK() = H. By the subcocient theorem to induced representations using the Casselman theorem, is possible to construct an operator that go from Homg, K(V, H ), to Homg, K(V, H), that define an unitary equivalence between the representations in H , and H. Then H , and H, are equivalent representations as representations of the group K.

If fH , that is to say, fL2(0M/K), then f(nak) = af(k), nN, aA, and kK.

If gG, and g = nak, with nN, aA, and kK, then we can write n(g) = n, a(g) = a, k(g) = k. The theory of real reductive groups implies that as functions on G, n, a, and k, are smooth functions. We denote as **1**, to the function on K, that is identically equal to 1.

Let 0, be the class of the trivial representations of K. Then is clear that

$$(\mathbb{H}^\mu)\chi(\gg) = \mathbb{C}\mathbb{L} \tag{11.1.2}$$

If (aC)\*, then we define , for

$$\Xi\_{\mu}(\mathbf{g}) = \lhd \pi\_{\mu}(\mathbf{g}) \mathbf{1}\_{\mu^{\mu}} \mathbf{1}\_{\mu} \text{ }, \tag{\text{II. 1.3}}$$

Said extended function to all the subgroup K, come given as

$$\Xi\_{\mu}(\mathbf{g}) = \left[ \kappa \mathbf{a} (\mathbf{k} \mathbf{g})^{\mu + \rho} \, \text{d} \mathbf{k}, \,\,\forall \, \mathbf{g} \in \mathbf{G} \tag{\text{II. 1.4}} \right]$$

where **1**(g) = a(g)and **1**(k) = a(kg), gG, kK.

**Proposition II. 1.1.** If sW(g, a), then <sup>s</sup>= , (aC)\*.

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

*Proof.* The demonstration of this result due to Harish-Chandra can require of some previous considerations. Let Cc(K\G/K), the space of all differentiable K-bi-invariants functions on G, of compact support on G. To demonstrate the proposition is enough demonstrate that if fCc(K\G/K), then

$$\int \text{cf}(\mathbf{g}) \Xi\_{\mu}(\mathbf{g}) \, \text{dg} = \int \text{cf}(\mathbf{g}) \Xi\_{\mathbb{H}}(\mathbf{g}) \, \text{dg} \,\tag{\text{II. 1.5}}$$

Calculating

$$\int \text{cf}(\mathbf{g}) \Xi\_{\mu}(\mathbf{g}) \, \text{dg} = \int\_{\mathcal{C}\_{\lambda}} \kappa \mathbf{f}(\mathbf{g}) \mathbf{a}(\mathbf{kg})^{\mu+\rho} \, \text{dg} \, \text{dk} = \int \text{cf}(\mathbf{g}) \mathbf{1}\_{\mu}(\mathbf{g}) \, \text{dg}\_{\mathcal{N}}$$

where f, is invariant left. Let S = An, and let ds, be a election of the left invariant measure on S, applying the measurability in real reductive groups. Then

$$\int \mathrm{cf}(\mathbf{g}) \mathbf{1}\_{\mu}(\mathbf{g}) \, \mathrm{d}\mathbf{g} = \int \mathrm{s}\_{\times} \, \mathrm{sf}(\mathrm{sk}) \mathbf{a}(\mathbf{s})^{\mu+\rho} \mathrm{d}\mathbf{s} \, \mathrm{d}\mathbf{k} = \int \mathrm{sf}(\mathbf{s}) \mathbf{a}(\mathbf{s})^{\mu+\rho} \mathrm{d}\mathbf{s} \, \mathrm{d}\mathbf{k}$$

due to the K-invariance of f. Now ds, can be normalized such that ds = adnda. Thus, calculating said integral, this can to take the form

$$\int\_{N \times \mathcal{A}} \mathbf{\hat{f}}(\mathbf{n}\mathbf{a}) \mathbf{a}^{\mu-\rho} \mathbf{d} \mathbf{n} \mathbf{d} \mathbf{a},$$

We call to Ff(a) = aNf(na)dn. Then we have demonstrated that

$$\int \mathrm{cf}(\mathbf{g}) \Xi\_{\mu}(\mathbf{g}) \, \mathrm{dg} = \int\_{\mathrm{Aff}} \mathbf{\hat{n}}(\mathbf{a}) \mathbf{a}^{\mu} \mathrm{da},\tag{\text{II. 1.6}}$$

Therefore to demonstrate the proposition is enough demonstrate that

$$\mathsf{F} \mathsf{(expH)} = \mathsf{F} \mathsf{(exp } \mathsf{s} \mathsf{H} \mathsf{)}, \forall \ \mathsf{H} \in \mathfrak{a}, \mathsf{f} \in \mathsf{C}^{\pi}(\mathsf{K} \backslash \mathsf{G}/\mathsf{K}), \text{ and } \mathsf{s} \in \mathsf{W}(\mathfrak{g}, \mathfrak{a}), \tag{\mathsf{II.1.7}}$$

Indeed, let +, be a positive roots space in (g, a), corresponding to n. Let 0, the corresponding set of simple roots. Let F = {}, be with 0, and let (PF, AF), be the corresponding parabolic pair of G. Then is factible the Langlands canonical decomposition

$$\mathbf{P} \mathbf{v} = {}^0 \mathbf{M} \mathbf{r} \mathbf{A} \mathbf{r} \mathbf{N} \mathbf{v} \tag{\text{II. 1.8}}$$

Let also F, defined by F(H) = (1/2)tr(ad HIaP) Ha. Let´s express to fP(am), fCc(K\G/K), as

$$\mathbf{f}^{\mathsf{P}}(\mathbf{a}\mathbf{m}) = \mathbf{a}^{-\mathsf{p}\mathsf{p}}\|\_{\mathsf{NF}} \mathbf{f}(\mathbf{n}\mathbf{a}\mathbf{m})\mathbf{d}\mathbf{n}\_{\mathsf{F}}.$$

 aAF, m0MF. In this sense, is necessary to take in count that dnF, has been elected of invariant measure on NF. We consider also that \*PF = P MF, Then \*PF, is a minimal parabolic subgroup of MF, with the respective Langlands decomposition to \*PF, to know,

$$\ast \mathbf{P\_F} = {}^0 \mathbf{M} \mathbf{A} {}^\* \mathbf{N\_F} \quad \ast \mathbf{N\_F} = \mathbf{N} \cap \mathbf{M\_F}$$

Normalizing the invariant measure, d\*nF, on \*NF, such that

Chapter II – Integrals, Functional and Special Functions on Lie Groups and Lie Algebras 11

$$\mathbf{dn} = \mathbf{dn} \mathbf{r} \mathbf{d}^\ast \mathbf{n} \mathbf{r}\_\prime \tag{\text{II. 1.9}}$$

Let's notice that fPCc (KF\MF/KF). Let \*Fg, be denoting the "Ff" to MF. Then is had that

$$\mathbf{F} \mathbf{i} = \, ^\ast \mathbf{F}\_{\mathcal{B}} \quad \text{ } \mathbf{g} = \mathbf{f}^P \text{ } \tag{11.1.10}$$

Now we can consider sH = Ad(k)H, with kKF. Therefore to demonstrate the proposition II. 1.1, only is enough demonstrate (II. 1. 10), under the set of endomorphic actions sH = Ad(k)H, taking the case when 0 = {}. Let p0 = pbe to the class , of the trivial representation. Let , be an automorphism of U(a), defined by

$$
\beta(\mathbf{H}) = \mathbf{H} + \mathfrak{o}(\mathbf{H})\mathbf{1}, \quad \forall \ \mathbf{H} \in \mathfrak{a}\_{\prime}
$$

Let 0 = P0, be. Then 0, is called the Harish-Chandra homomorphism.

**Lemma II. 1.1.** The following two sentences are equivalent:

i. 0(U(g)K) U(a)W (W = W(g, a)),

ii. <sup>s</sup>, sW, (aC)\*.

*Proof.* If THomK(V, (H)K),and if uU(g)K, then

$$(\mathbf{u}\mathbf{T})^\circ = \mathbf{T}^\circ \mathbf{P}\_{\mathbf{\color{red}{y},\color{red}{x},\color{red}{y}}}(\mathbf{u}),$$

In particular, (H)K, have infinitesimal character of the form (). This implies in special that u (p0(u)), uU(g)K, (aC)\*. If uU(g)a, then

$$\mathbf{u}\Xi\_{\mu}(\mathbf{1}) = \mathrm{Ad}(\mathbf{k})\mathbf{u}\ \bullet \Xi\_{\mu}(\mathbf{1})\,.$$

Therefore from the analyticity of the functional , the lemma is followed.

**Lemma II. 1.2.** If 0 = {}, then 0(U(g)K) = {hU(a) sh = h}.

*Proof.* Consider gI = [g, g]. Then dim a gI = 1. Let X1, , Xp, be an orthogonal base of pI = p gI, relative to B. Let Cp = (Xj)2. For other side K, act transitively on the unitary sphere of pI, and is clear that S(p)K, is the generated algebra for **1**, p *z*(g), and Cp. Then by the subquotient to (g, K)-modules, is known that

$$\text{Ker } \gamma\_0 = \mathsf{U}(\mathfrak{g})^{\mathbb{K}} \cap \mathsf{U}(\mathfrak{g})\mathsf{I}^{\square},$$

where

10 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

S, applying the measurability in real reductive groups. Then

We call to Ff(a) = aNf(na)dn. Then we have demonstrated that

Therefore to demonstrate the proposition is enough demonstrate that

calculating said integral, this can to take the form

Ff(expH) = Ff(exp sH), Ha, fCc

Normalizing the invariant measure, d\*nF, on \*NF, such that

fCc(K\G/K), then

Calculating

fCc(K\G/K), as

*Proof.* The demonstration of this result due to Harish-Chandra can require of some previous considerations. Let Cc(K\G/K), the space of all differentiable K-bi-invariants functions on G, of compact support on G. To demonstrate the proposition is enough demonstrate that if

Gf(g)(g) dg = <sup>G</sup> Kf(g)a(kg)dgdk = Gf(g)**1**(g) dg, where f, is invariant left. Let S = An, and let ds, be a election of the left invariant measure on

Gf(g)**1**(g) dg = <sup>S</sup> Kf(sk)a(s)dsdk = Sf(s)a(s)ds, due to the K-invariance of f. Now ds, can be normalized such that ds = adnda. Thus,

<sup>N</sup> Af(na)adnda,

Gf(g)(g) dg = AFf(a)a

Indeed, let +, be a positive roots space in (g, a), corresponding to n. Let 0, the corresponding set of simple roots. Let F = {}, be with 0, and let (PF, AF), be the corresponding parabolic pair of G. Then is factible the Langlands canonical decomposition

 PF = 0MFAFNF, (II. 1.8) Let also F, defined by F(H) = (1/2)tr(ad HIaP) Ha. Let´s express to fP(am),

f P(am) = aFNFf(nam)dnF, aAF, m0MF. In this sense, is necessary to take in count that dnF, has been elected of invariant measure on NF. We consider also that \*PF = P MF, Then \*PF, is a minimal parabolic subgroup of MF, with the respective Langlands decomposition to \*PF, to know,

\*PF = 0MA\*NF, \*NF = N MF,

Gf(g)(g) dg = Gf(g)<sup>s</sup>(g) dg, (II. 1.5)

da, (II. 1.6)

(K\G/K), and sW(g, a), (II. 1. 7)

$$\gamma\iota(\mathbb{U}(\mathfrak{g})^{\mathbb{K}}) = \gamma\iota(\operatorname{Symm}(\mathbb{S}(\mathfrak{p}\_{\mathbb{C}}))^{\mathbb{K}}),\tag{\text{II. 1.11}}$$

A simple calculing show that exist constants c1, and c2 0, such that

$$\gamma \circ (\mathsf{C}\_{\mathsf{b}}) = \mathsf{cl} \langle \mathsf{H}^2 + \mathsf{c} \rangle\_{\mathsf{b}}$$

The lemma is followed immediately.

**Theorem II. 1.1.** The following sucesion of algebra homomorphisms is exact:

$$\mathfrak{l} \xrightarrow{} \mathrm{U(g)}^{K} \cap \mathrm{U(g)}\\\mathrm{I}\_{\gamma} \xrightarrow{} \mathrm{U(g)}^{K} \xrightarrow{\gamma\_{0}} \mathrm{U(g)}^{W} \xrightarrow{} \mathrm{U(g)}^{W} \tag{\text{II. 1.12}}$$

With major pointity 0o Symm : S(pC)K U(a)W, is a linear bijection.

We conclude this chapter with an estimation on the functional the which is used in the Casselman theorem.

Consider

$$\mathfrak{a}^+ = \{ \mathbf{H} \in \mathfrak{a} \mid \alpha(\mathbf{H}) > 0, \,\forall \alpha \in \Phi^+ \},\tag{11.1.13}$$

Let A+ = exp a+, be. We consider furthermore that

$$\mathbf{G} = \mathbf{K} \mathbf{C} \mathbf{l} (\mathbf{A}^+) \mathbf{K} \tag{\text{II. 1.14}}$$

which is true by the general theory of real reductive groups. Thus if fC(K\G/K), then f is completely determined for their values on Cl(A+). Let

$$(\mathfrak{a}\_{\ast})^{+} = \{ \mu \in \mathfrak{a}\_{\ast} \mid \ (\mu, \alpha) > 0, \ \forall \alpha \in \Phi^{+} \},\tag{11.1.15}$$

Let W = W(g, a), be. If a\*, then there is an unique element in W Cl((a\*)+). We use the notation , to this element defined univocally. The theory of the real reductive groups implies that the intersection

$$\mathsf{W}\mu \cap \mathrm{Cl}((\mathfrak{a}^\*)^\*) \not\simeq \bigotimes\_r$$

Therefore is possible to demonstrate that if , Cl((a\*)+), and if s= , sW, then = . From lemma of Harish-Chandra explained in the general theory of (g, K)-modules, is deduced that s= Q, with Q = c, the sum on , and whose coefficients c 0. Thus

$$\iota(\beta,\langle \beta \rangle) = (\alpha,\langle \alpha \rangle) = (\langle \beta - \bigcirc, \alpha \rangle = (\beta, \alpha) - (\alpha, \bigcirc) \leq (\beta, \alpha) - (\bigcirc, \beta) \leq (\beta, \langle \beta \rangle)$$

In particular (Q, ) = 0. But then (, ) = (, ) + (Q, Q). Thus Q = 0.

**Lemma II. 1.3.** Let (aC)\*, then (a)aRe0(a), aCl(A+).

*Proof.* From the integral functional expression(a) = Ka(ka)dk, is deduced that (g)Re(g). Thus we can assume that a\*. The proposition II. 1. 1, implies that we can assume that = Let aA, be. Then

$$\mathbb{E}\_{\mu}(\mathbf{a}) = \left| \kappa \mathbf{a}(\mathbf{k}\mathbf{a})^{\mu+\rho} \mathbf{d} \mathbf{k} = \left| \kappa \mathbf{a}(n)^{2\rho} \mathbf{a}(\mathbf{k}(n)\mathbf{a})^{\mu+\rho} \mathbf{d}n,\right.\tag{\text{II. 1.16}} \right|$$

for the lemma to the invariant definition of dn [5], the measure on K. Then k(*n*)Na(*n*)-1*n*. Thus

$$\Xi\_{\mu}(\mathbf{a}) = \left| \text{va}(n)^{-\mu+\rho} \text{ a}(n\mathbf{a})^{\mu+\rho} \text{d}n,\right.\tag{11.1.17}$$

We note also that a(*n*a) = aa(a*n*a). Therefore has been demonstrated that

$$\Xi\_{\mu}(\mathbf{a}) = \mathbf{a}^{\mu+\rho} \| \varkappa \mathbf{a}(n)^{-\mu+\rho} \mathbf{a}(\mathbf{a}^{-1} n \mathbf{a})^{\mu+\rho} \text{dn},\tag{\text{II. 1.18}}$$

If aCl(A+), then considering the invariance of the measure dn, under the actions of the subgroup A, is obtained

$$(\mathbf{a}(n)^{-1}\mathbf{a}(\mathbf{a}^{-1}n\mathbf{a}))^{\mu} \le 1,\tag{11.1.19}$$

Thus

12 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

With major pointity 0o Symm : S(pC)K U(a)W, is a linear bijection.

a+ = {Ha (H) > 0,

Let A+ = exp a+, be. We consider furthermore that

completely determined for their values on Cl(A+). Let

deduced that s= Q, with Q = c, the sum on

assume that = Let aA, be. Then

In particular (Q, ) = 0. But then (, ) = (, ) + (Q, Q). Thus Q = 0.

**Lemma II. 1.3.** Let (aC)\*, then (a)aRe0(a), aCl(A+).

(a) = Ka(ka)dk = Na(*n*)

(a\*)+ = {a\* () > 0,

**Theorem II. 1.1.** The following sucesion of algebra homomorphisms is exact:

<sup>0</sup> 0 U( ) U( )I U( ) U( ) 0, gg g a <sup>K</sup>

We conclude this chapter with an estimation on the functional the which is used in the

which is true by the general theory of real reductive groups. Thus if fC(K\G/K), then f is

Let W = W(g, a), be. If a\*, then there is an unique element in W Cl((a\*)+). We use the notation , to this element defined univocally. The theory of the real reductive groups

W Cl((a\*)+) , Therefore is possible to demonstrate that if , Cl((a\*)+), and if s= , sW, then = . From lemma of Harish-Chandra explained in the general theory of (g, K)-modules, is

(, ) = (, ) = (Q, ) = (, ) (, Q) (, ) (Q, ) (, ),

*Proof.* From the integral functional expression(a) = Ka(ka)dk, is deduced that (g)Re(g). Thus we can assume that a\*. The proposition II. 1. 1, implies that we can

K W

(II. 1.12)

}, (II. 1.13)

}, (II. 1.15)

, and whose coefficients c 0. Thus

a(k(*n*)a)dn, (II. 1.16)

G = KCl(A+)K, (II. 1.14)

The lemma is followed immediately.

Casselman theorem.

implies that the intersection

Consider

$$\Xi\_{\mu}(\mathbf{a}) \le \mathbf{a}^{\mu}(\mathbf{a}^{\rho}) \text{va}(n)^{\rho} \text{ a}(\mathbf{a}^{-1} n\mathbf{a})^{\rho} \text{dn}) = \mathbf{a}^{\mu} \Xi \omega(\mathbf{a}),\tag{11.1.20}$$

For (II. 1. 18).

**Corollary II. 1.1.** If (aC)\*, and from aCl(A+), then (a) a.

*Proof.* 0, is an unitary representation of G (see elemental theory of representations in Hilbert space L2(G)). Thus 0 1. The result is followed by the lemma II. 1. 3.

### **II. 2. Geometrical transforms: Integral transforms on Lie groups (Radon transforms on generalized flag manifolds)**

Let G, be a connected reductive algebraic group over C, and g, the Lie algebra of G. The group G, acts on g, by the adjoint action Ad. Let h, be a Cartan subalgebra of g, , the root system in h\*, {IiI0}, a set of simple roots , the set of positive roots , the set of negative roots h\*Z= Hom(H, C ) h\*, the weight lattice, and W, the Weyl group. For , we denote by g, the corresponding root space and by h, the corresponding co-root. For iI0, we denote by siW, the reflection corresponding to i. For wW, we set *l*(w) = #(w ). Set = ½ , and define a (shifted) affine action of W, on h\*, by

$$\mathbf{w} \diamond \boldsymbol{\lambda} = \mathbf{w}(\boldsymbol{\lambda} + \boldsymbol{\phi}) - \boldsymbol{\phi},\tag{11.2.1}$$

For I I0, we set

$$\begin{aligned} \mathsf{M} &= \Delta \sqcap \supset \sum\_{\mathsf{i}\_{\mathsf{c}} \in \mathsf{I}} \mathsf{Z}\_{\mathsf{c}\mathsf{i}\_{\mathsf{c}}} \mathsf{Z}\_{\mathsf{c}\mathsf{i}\_{\mathsf{c}}} \quad \Lambda \mathsf{i}^{\mathsf{a}} = \Delta \mathsf{i} \cap \Delta^{+} \mathsf{i} \quad \mathsf{W} \mathsf{i} = \mathsf{<} \mathsf{s} \mathsf{i} | \ \mathsf{i} \in \mathsf{I} \triangleright \mathsf{C} \mathsf{W} \\\\ \mathsf{I} &= \mathsf{I} \oplus \big( \bigoplus\_{\alpha \in \Lambda^{\mathsf{I}}} \mathsf{g}\_{\alpha} \Big) \quad \mathsf{m} = \bigoplus\_{\alpha \in \alpha \cdot \mathsf{A}^{+} / \mathsf{A}^{\mathsf{a}}} \mathsf{g}\_{\alpha} \quad \mathsf{p} = \mathsf{I} \oplus \mathsf{m} \urarrow \end{aligned}$$

where

$$\begin{aligned} (\mathbb{I}\_{\mathbb{A}^\*} \mathbb{Z})\_0^{\mathbb{I}} &= \{ \boldsymbol{\chi} \in \mathbb{I}\_{\mathbb{A}^\*} \mathbb{Z} \, \middle| \, \boldsymbol{\chi}(\alpha \mathbf{v}\_{\boldsymbol{\cdot}}) \ge 0, \,\forall \; \mathbf{i} \in \mathcal{I} \} \subset \left( \mathbb{I}\_{\mathbb{A}^\*} \mathbb{Z} \right)^{\mathbb{I}} \end{aligned}$$
 
$$(\mathbb{I}\_{\mathbb{A}^\*} \mathbb{Z})\_{\mathbb{I}} = \{ \boldsymbol{\chi} \in \mathbb{I}\_{\mathbb{A}^\*} \mathbb{Z} \, \middle| \, \boldsymbol{\chi}(\alpha \mathbf{v}\_{\boldsymbol{\cdot}}) \ge 0, \,\forall \; \mathbf{i} \in \mathcal{I} \} \subset \left( \mathbb{I}\_{\mathbb{A}^\*} \mathbb{Z} \right)^{\mathbb{A}}$$

and also

$$\rho\_{\mathbb{I}} = (\sum\_{a \in \Lambda + /\Lambda \mathbb{I}} a) / \mathcal{Z}\_{\mathbb{I}}$$

We denote by wI, the longest element of WI. It is an element of WI, characterized by wI(<sup>I</sup> ) = <sup>I</sup> . Let LI, NI, and PI, be the subgroups of G, corresponding to lI, nI, and pI.

For (h\*Z)I, let VI(), be the irreducible LI-module with highest weight . We regard VI(), as a PI-module with the trivial action of NI, and define the generalized Verma module with highest weight , by

$$\operatorname{Mul}(\lambda) = \mathcal{U}l(\mathfrak{g}) \otimes\_{\mathfrak{q}^{(\mathfrak{gl})}} \operatorname{Vol}(\lambda),\tag{11.2.2}$$

Let L(), be the unique irreducible quotient of MI() (note that L(), does not depend on the choice of I, such that (h\*Z)I). Then any irreducible PI-module is isomorphic to VI(), for some (h\*Z)I, and we have dim VI() = 1, if and only if (h\*Z)0I. Moreover, any irreducible (g, PI)- module is isomorphic to L(), for some (h\*Z)I.

Let

$$\chi\_{\mathbb{L}} = \mathbf{G}/\mathbb{P}\_{\mathbb{L}\_{\nu}}$$

be the generalized flag manifold associated to I.

Consider the category equivalence given in the following proposition to quasi-G-equivariant *D*Z-modules:

**Proposition II. 2.1.** Assume that Z = G/H, where H, is a closed subgroup of G, and set x = *e*H ∈ Z.


The statement (i) is well-known (see [40]), and (ii) is due to Kashiwara [41].

Then the isomorphism classes of G-equivariant *O*XI-modules (resp. quasi-G-equivariant *D*XImodules) are in one-to-one correspondence with isomorphism classes of PI-modules (resp. (g, PI)-modules). For (h\*Z)I, we denote *O*XI(), the G-quivariant *O*XI-module corresponding to the irreducible PI-module VI(). We see easily the following.

**Lemma II. 2.1.** Let (h\*Z)I. The quasi-G-equivariant *D*XI-module corresponding to the (g, PI)-module MI(), is isomorphic to *DO*XI() = *D*XI *O*XI*O*XI().

We need the following relative version of the Borel-Weil-Bott theorem later (see Bott [42]).

**Proposition II. 2.2.** Let I J I0, and let : XI XJ, be the canonical projection. For (h\*Z)I, we have the following.

	- (w(2I))( ) > 0, for any <sup>J</sup> +. Then we have

$$\mathbf{R}\pi \colon (\mathbf{C}\ltimes(\lambda)) \vDash (\mathbf{w}(\lambda+\mathfrak{p}-2\mathfrak{p}\mathfrak{i})-(\mathfrak{p}-2\mathfrak{p}\mathfrak{i})) [-l(\mathbf{w}\mathfrak{w})-l(\mathfrak{w}\mathfrak{i})].$$

Let I, J I0, with I J. The diagonal action of G, on XIXJ, has a finite number of orbits, and the only closed one G(ePI, ePJ), is identified with XIJ = G/(PI PJ). We can consider the correspondence to *D*-modules

$$
\mathcal{X} \leftarrow \mathcal{S} \rightarrow \mathcal{Y}, \tag{\text{II.2.3}}
$$

with X = XI, Y = XJ, and S = X, then (II. 2. 3) take the form

$$\mathbf{X\_{l}} \xleftarrow{f} \mathbf{X\_{l\cap l}} \xrightarrow{\mathbf{g}} \mathbf{X\_{l'}} \tag{\text{II. 2.4}}$$

and the Radon transform R(*DO*XI()), for (h\*Z)I. Since f, and g, are morphisms of Gmanifolds then the functor

$$\mathbb{R}: \mathcal{D}^{\flat}(\mathcal{D}\times) \to \mathcal{D}^{\flat}(\mathcal{D}\times)\_{\prime\prime}$$

with the correspondence rule

$$\mathbf{R}(\mathcal{M}) = \mathbf{g} \mathbf{f}^{-1}(\mathcal{M})\_{\mathcal{M}}$$

called the Radon transform on algebraic *D*-modules, induces a functor

$$\mathcal{R}: \mathcal{D}\_{\mathcal{G}}^{\flat}(\mathcal{D}\_{\mathcal{X}\_{\parallel}}) \to \mathcal{D}\_{\mathcal{G}}^{\flat}(\mathcal{D}\_{\mathcal{X}\_{\parallel}})\_{\prime} \tag{\mathcal{II}.2.5}$$

Note that we have

14 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

(h\*Z)I = {h\*Z<sup>i</sup>

(h\*Z)0I = {h\*Z<sup>i</sup>

(g, PI)- module is isomorphic to L(), for some (h\*Z)I.

be the generalized flag manifold associated to I.

via the correspondence *M* →*M*(x).

modules via the correspondence *M* →*M*(x).

/ ( / 2, ) *I I* 

We denote by wI, the longest element of WI. It is an element of WI, characterized by

For (h\*Z)I, let VI(), be the irreducible LI-module with highest weight . We regard VI(), as a PI-module with the trivial action of NI, and define the generalized Verma module with

 MI() = *U*(g) *U*(pI) VI(), (II. 2.2) Let L(), be the unique irreducible quotient of MI() (note that L(), does not depend on the choice of I, such that (h\*Z)I). Then any irreducible PI-module is isomorphic to VI(), for some (h\*Z)I, and we have dim VI() = 1, if and only if (h\*Z)0I. Moreover, any irreducible

XI = G/PI,

Consider the category equivalence given in the following proposition to quasi-G-equivariant

**Proposition II. 2.1.** Assume that Z = G/H, where H, is a closed subgroup of G, and set

i. The category of G-equivariant *O*Z-modules is equivalent to the category of H-modules

ii. The category of quasi-G-equivariant *D*Z-modules is equivalent to the category of (g, H)-

Then the isomorphism classes of G-equivariant *O*XI-modules (resp. quasi-G-equivariant *D*XImodules) are in one-to-one correspondence with isomorphism classes of PI-modules (resp.

The statement (i) is well-known (see [40]), and (ii) is due to Kashiwara [41].

. Let LI, NI, and PI, be the subgroups of G, corresponding to lI, nI, and pI.

 

iI},

iI} (h\*Z)I,

where

and also

wI(<sup>I</sup> ) = <sup>I</sup> 

Let

*D*Z-modules:

x = *e*H ∈ Z.

highest weight , by

$$
\Omega\_{\mathsf{f}} \equiv \mathsf{CM}\_{\mathsf{f}^\*} \mathsf{J}(\mathsf{\chi}, \mathsf{l}), \text{ for } \quad \mathsf{\chi}, \mathsf{l} = \Sigma\_{\mathsf{aux}\Delta} + \mathsf{c}\_{\mathsf{hl}} \mathsf{ox}.\tag{\text{II. 2.6}}
$$

### **II. 3. Radon transform of quasi-equivariant D-modules**

Let h<sup>Z</sup> \* We describe our method to analyse R(*DO*XI()) = g\*f (*DO*XI()). By

$$(\underline{\mathbf{f}}^{-1}(\mathcal{D}\mathbf{\mathcal{O}}\boldsymbol{u}(\boldsymbol{\lambda})))(\mathbf{e}(\mathrm{Pr}\cap\mathrm{P})) \cong \mathcal{D}\mathcal{O}\mathbf{u}(\boldsymbol{\lambda}))(\mathrm{eP}) \cong \mathrm{M}(\boldsymbol{\lambda}),\tag{\mathrm{II.3.1}}$$

the quasi-equivariant *D*XIJ-module f(*DO*XI()), corresponds to the (g, PI PJ)-module MI()= *U*(g) *U*(pI)VI(), under the category equivalence given in proposition II. 2. 1. Set

$$
\Gamma = \{ \mathbf{x} \in \mathbb{W} \mid \mathbf{x} \text{ is the shortest element of } \mathbf{W} \mathbf{i}\_{\cap} \mathbf{x} \} \tag{\text{II. 3.2}}
$$

and

$$
\Gamma\_{\mathbf{x}} = \{ \mathbf{x} \in \Gamma \mid l(\mathbf{x}) = \mathbf{k} \},
\tag{\text{II. 3.3}}
$$

It is well-known that an element xWI, belongs to , if and only if x <sup>I</sup><sup>J</sup> I. This condition is also equivalent to

$$(\chi(\mathfrak{y} + \mathfrak{h}))(\mathfrak{a}^\vee) \ge 0, \ \forall \mathfrak{a} \in \Lambda^+ \mathfrak{h} \tag{\text{II.3.4}}$$

In particular, we have x o (h<sup>Z</sup> \* )IJ, for x.

By Lepowsky [43] and Rocha-Caridi [44] we have the following resolution of the finite dimensional lI-module VI():

$$0 \to \mathcal{N}\_{\mathbb{H}} \to \mathcal{N}\_{\mathbb{H}-1} \to \dots \to \mathcal{N} \to \mathcal{N} \to \mathcal{V} \\ \iota \to \operatorname{Vr}(\mathbb{X}) \to 0,\tag{11.3.5}$$

with *n* = dim lI/lIpJ, and

$$\mathcal{N}\_{\mathbf{k}} = \bigoplus\_{\mathbf{x} \in \Gamma\_{\mathbf{k}}} \mathcal{U}(\mathbf{l}\_{\mathbf{l}}) \otimes\_{\mathcal{U}(\mathbf{l}\_{\mathbf{l}} \cap \mathfrak{A}\_{\mathbf{l}})} \mathcal{V}\_{\mathbf{l} \cap \mathbf{J}}(\mathbf{x} \circ \lambda)\_{\mathbf{\prime}} \tag{\text{II.3.6}}$$

By the Poincaré-Birkhoff-Witt theorem we have the isomorphi

$$\mathcal{GL}(\mathfrak{l}\_{\mathbb{I}}) \otimes\_{\mathcal{U}(\mathfrak{l}\_{\mathbb{I}} \cap \mathfrak{p}\_{\mathbb{J}})} \mathcal{V}\_{\mathbb{I} \cap \mathbb{J}}(\mathbf{x} \circ \lambda) \equiv \mathcal{U}(\mathfrak{p}\_{\mathbb{I}}) \otimes\_{\mathcal{U}(\mathfrak{p}\_{\mathbb{I} \cap \mathbb{J}})} \mathcal{V}\_{\mathbb{I} \cap \mathbb{J}}(\mathbf{x} \circ \lambda),\tag{\text{II.3.7}}$$

Of *U*(lI)-modules, where n<sup>I</sup>J, acts trivially on VIJ(x o ). For other side, the action of nI, on I J I ( ) I J () V x( ) l p *<sup>U</sup>* <sup>l</sup> *<sup>U</sup>* , is trivial. Indeed, by [pI, nI] nI, we have

$$\mathfrak{m}\mathcal{L}l(\mathfrak{h}) = \mathcal{L}l(\mathfrak{h})\mathfrak{m}\_{\mathfrak{h}}$$

and hence

$$\mathfrak{n}\_{\mathsf{l}}(\mathcal{Q}l(\mathfrak{p}\_{\mathsf{l}})\otimes\_{\mathcal{Q}l(\mathfrak{p}\_{\mathsf{l}\circ\mathsf{l}})}\mathcal{V}\_{\mathsf{l}\circ\mathsf{l}}(\mathbf{x}\circ\lambda)) \subset \mathcal{Q}l(\mathfrak{p}\_{\mathsf{l}})\mathfrak{n}\_{\mathsf{l}}\otimes \mathcal{V}\_{\mathsf{l}\circ\mathsf{l}}(\mathbf{x}\circ\lambda) \subset \mathcal{Q}l(\mathfrak{p}\_{\mathsf{l}})\otimes \mathfrak{n}\_{\mathsf{l}}\mathcal{V}\_{\mathsf{l}\circ\mathsf{l}}(\mathbf{x}\circ\lambda) = 0,$$

for nI n<sup>I</sup>J. Thus we obtain the following resolution of the finite dimensional pI-module VI() (with trivial action of nI):

0 N'*n* N'*<sup>n</sup>* – 1 N'1 N'0 VI() 0, (II. 3.8)

with

16 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Let h<sup>Z</sup>

and

\*

condition is also equivalent to

In particular, we have x o (h<sup>Z</sup>

dimensional lI-module VI():

with *n* = dim lI/lIpJ, and

and hence

(x())(

\*

By the Poincaré-Birkhoff-Witt theorem we have the isomorphi

I J I ( ) I J () V x( ) l p *<sup>U</sup>* <sup>l</sup> *<sup>U</sup>* , is trivial. Indeed, by [pI, nI] nI, we have

**II. 3. Radon transform of quasi-equivariant D-modules**

We describe our method to analyse R(*DO*XI()) = g\*f

(f(*DO*XI()))(e(PI PJ))*DO*XI())(ePI) MI(), (II. 3.1)

the quasi-equivariant *D*XIJ-module f(*DO*XI()), corresponds to the (g, PI PJ)-module

<sup>x</sup> = {x*l*(x) = k}, (II. 3.3)

By Lepowsky [43] and Rocha-Caridi [44] we have the following resolution of the finite

0 N*n* N*<sup>n</sup>* – 1 N1 N0 VI() 0, (II. 3.5)

I J <sup>k</sup> <sup>k</sup> I ( ) I J <sup>x</sup> N () V x( ),

Of *U*(lI)-modules, where n<sup>I</sup>J, acts trivially on VIJ(x o ). For other side, the action of nI, on

nI*U*(lI) = *U*(lI)nI,

I J I I ( ) I J I I I J I I I J ( () V x () V x () V x ( )) ( ) ( ) 0, <sup>p</sup> n p *U UU <sup>U</sup>* p n p n

) > 0,

= {xWI x is the shortest element of WIJx}, (II. 3.2)

l p *<sup>U</sup>* <sup>l</sup> *<sup>U</sup>* (II.3.6)

I J I J I ( ) I J I ( ) I J () V x ( ) V x ( ) ( ), l p <sup>p</sup> *U U* l p *U U* (II. 3.7)

MI()= *U*(g) *U*(pI)VI(), under the category equivalence given in proposition II. 2. 1. Set

It is well-known that an element xWI, belongs to , if and only if x

)IJ, for x.

(*DO*XI()). By

<sup>I</sup><sup>J</sup>

<sup>I</sup>J, (II. 3.4)

I. This

$$\mathcal{N}\_{\mathbf{k}}^{\circ} = \bigoplus\_{\mathbf{x} \in \Gamma\_{\mathbf{k}}} \mathcal{U}(\mathfrak{p}\_{\mathbf{l}}) \otimes\_{\mathcal{U}(\mathfrak{p}\_{\mathbf{l} \cap \mathbf{J}})} \mathcal{V}\_{\mathbf{l} \cap \mathbf{J}}(\mathbf{x} \circ \lambda).$$

By tensoring *U*(g), to (II. 3. 8) over *U*(pI), we obtain the following resolution of the (g, PI J) module MI():

$$0 \to \tilde{\mathcal{N}}\_n \to \tilde{\mathcal{N}}\_{n-1} \to \dots \to \tilde{\mathcal{N}}\_1 \to \tilde{\mathcal{N}}\_0 \to \mathcal{M}\_1 (\mathbb{X}) \to 0,\tag{11.3.9}$$

with

$$\mathbf{N}\_{\mathbf{k}}^{r} = \bigoplus\_{\mathbf{x} \in \Gamma\_{\mathbf{k}}} \mathbf{M}\_{\mathbf{I} \cap \mathbf{J}}(\mathbf{x} \circ \mathbf{\hat{x}}),$$

Since the quasi-G-equivariant *D*XIJ- module corresponding to the (g, PIJ)-module MIJ(x o ) is *DO*XIJ(x o ), we have obtained the following resolution of the quasi-G-equivariant *D*XIJmodule f(*DO*XI()),

$$0 \to \mathcal{N}\_n \to \mathcal{N}\_{n-1} \to \dots \to \mathcal{N}\_1 \to \mathcal{N}\_0 \to \underline{\mathfrak{f}}^{-1} (\mathcal{D}\mathcal{O}\_{\mathcal{X}\_l}(\mathbb{A})) \to 0,\tag{11.3.10}$$

with

$$\mathcal{N}\_{\mathbf{k}} = \bigoplus\_{\mathbf{x} \in \Gamma\_{\mathbf{k}}} \mathcal{D} \mathcal{O}\_{\mathbf{X}\_{\mathbf{l} \sim \mathbf{J}}} (\mathbf{x} \circ \boldsymbol{\lambda}) \,. \tag{\text{II.3.11}}$$

The following results consist in to investigate on the *D*XIJ-module g\*(*DO*XIJ(x o )), for x. We first remark that

$$\underline{\mathbf{g}}\_{\*}\left(\mathcal{D}\mathcal{O}\_{\mathcal{X}\_{\mathbb{I}\times\mathbb{J}}}\left(\mathbf{x}\circ\boldsymbol{\lambda}\right)\right) = \mathcal{D}\_{\mathcal{X}\_{\mathbb{I}}}\otimes\_{\mathcal{O}\_{\mathcal{X}\_{\mathbb{J}}}}\mathrm{Rg}\_{\*}\left(\mathcal{O}\_{\mathcal{X}\_{\mathbb{I}\times\mathbb{J}}}\left(\mathbf{x}\circ\boldsymbol{\lambda} + \boldsymbol{\gamma}\_{\mathrm{I},\mathcal{I}}\right)\right),\tag{\text{II. 3.12}}$$

Indeed, by (II. 2.6) we have

I J J X I J X I J I J I J J X I J I J <sup>1</sup> <sup>J</sup> <sup>X</sup> X I J I J <sup>J</sup> J X X I J J I J J X I J L L X X XI J X X L X XI J X 1 X g g X X gX X X g x Rg x Rg x Rg g x Rg x Rg ( ( )) ( ( )) ( ( )) ( ( )) ( ( )) ( *D DD <sup>D</sup> D D D D O O O O O O O O O O O O O*<sup>J</sup> I, J ( )), x 

**Lemma II. 3.1.** Let (h\*Z)I, and x.


$$\operatorname{Reg}\_{\bullet}(\mathcal{O}\_{\mathbf{X}\_{\mathbf{I}\cap\mathbf{J}}}(\mathbf{x}\circ\boldsymbol{\lambda}+\boldsymbol{\gamma}\_{\mathbf{I},\mathbf{J}}))=\operatorname{\mathcal{O}}\_{\mathbf{X}\_{\mathbf{J}}}((\operatorname{\mathbf{y}x})\circ\boldsymbol{\lambda})[-(l(\operatorname{\mathbf{w}}\_{\mathbf{J}}\mathbf{y})-l(\operatorname{\mathbf{w}}\_{\mathbf{I}\cap\mathbf{J}}))] \,\tag{\Pi.3.13}$$

*Proof*. [45].

**Proposition II. 3.1.** Let (h\*Z)I. Then there exists a family {*M*(k) }k0, of objects of DGb(*D*XJ), satisfying the following conditions.


$$C(\mathbf{k})^\bullet \to \mathcal{M}(\mathbf{k})^\bullet \to \mathcal{M}(\mathbf{k}+\mathbf{l})^\bullet \xrightarrow{+1} \tag{11.3.14}$$

where

$$\mathcal{L}(\mathbf{k})^\bullet = \bigoplus\_{\mathbf{x} \in \Gamma\_{\mathbf{k}}(\lambda)} \mathcal{D}\mathcal{O}\_{\mathbf{X}\_{\mathbf{j}}}((\mathbf{y}\_{\mathbf{x}}\mathbf{x})\circ\lambda)[l(\mathbf{x})\text{-}\mathbf{m}(\mathbf{x})] \,. \tag{\text{II.3.15}}$$

*Proof*. For 0 k dim lI/lIpJ, define an object *N*(k) , of DGb(*D*XIJ), by

$$\mathcal{N}\left(\mathbf{k}\right)^{\bullet} = \left[ \dots \to 0 \to \mathcal{N}\_n \to \mathcal{N}\_{n \cdot 1} \to \dots \to \mathcal{N}\_k \to 0 \dots \right] \prime \tag{11.3.16} \qquad \text{(II.3.16)}$$

Where *N*j, has degree j (see (II. 3. 10) and (II. 3. 11) for the notation). For k >dim lI/lIpJ, we set *N*(k) = 0. By *N*(0) f (*DO*XI()), we have g\**N*(0) R(*DO*XI()). Set

*M*(k) = g\**N*(k) . Then the statements (i) and (ii) are obvius. Let us show (iii). Applying g\*, to the distinguished triangle

$$\mathcal{N}\_{\mathbf{k}}[\mathbf{k}] \to \mathcal{N}(\mathbf{k})^{\bullet} \to \mathcal{N}(\mathbf{k}+1)^{\bullet} \xrightarrow{+1} \succ\_{\mathbf{k}}$$

we obtain a distinguished triangle

$$\underline{\mathbf{g}}\_{\ast}\mathcal{N}\_{\mathbf{k}}[\mathbf{k}] \to \mathcal{M}(\mathbf{k})^{\bullet} \to \mathcal{M}(\mathbf{k}+1)^{\bullet} \xrightarrow{+1} \cdots$$

For (II. 3.11) and (II. 3.12) and Lemma 2. 3.1, we have

$$\underline{\mathbf{g}}\_{\star} \mathcal{N}\_{\mathbf{k}} = \bigoplus\_{\mathbf{x} \in \Gamma\_{\mathbf{k}}(\lambda)} \mathcal{D} \mathcal{O}\_{\mathbf{X}\_{\mathbf{l}}}((\mathbf{y}\_{\mathbf{x}} \mathbf{x}) \circ \lambda) [-\mathbf{m}(\mathbf{x})]\_{\lambda}$$

The statement (iii) is proved.

### **Theorem II. 3.1.** Let (h\*Z)I.

i. We have

18 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

XI J J

**Proposition II. 3.1.** Let (h\*Z)I. Then there exists a family {*M*(k)

<sup>J</sup> <sup>k</sup> X x <sup>x</sup>

( )

(*DO*XI()), we have g\**N*(0)

(k) y x )[ (x) - m(x)]


Where *N*j, has degree j (see (II. 3. 10) and (II. 3. 11) for the notation). For k >dim lI/lIpJ, we

<sup>k</sup>[k (k) (k 1) <sup>1</sup> ] , *NNN*

<sup>k</sup> g [k (k) (k 1) <sup>1</sup> ] ,

<sup>I</sup> <sup>k</sup> <sup>k</sup> X x <sup>x</sup> g y x) m x

( )

*NMM*

(( )[ ( )], *N D <sup>O</sup>*

(( ) ,

) = 0, for some J, then we have Rg\*(*O*XIJ(x o I, J)) = 0.

) 0, for any J. Take yWJ, satisfying (yx())(

Rg x I, J X J I J ( ( )) (( )[ ( ( ) ( ))], yx) w y w *l l O O* (II.3.13)

(k) (k) (k 1) <sup>1</sup> , *CMM* (II.3.14)

*C D <sup>O</sup> <sup>l</sup>* (II.3.15)

, of DGb(*D*XIJ), by

R(*DO*XI()). Set

*NN N n n* (II. 3.16)

. Then the statements (i) and (ii) are obvius. Let us show (iii). Applying g\*, to

) > 0,

}k0, of objects of DGb(*D*XJ),

**Lemma II. 3.1.** Let (h\*Z)I, and x.

J. Then we have

ii. Assume that (x())(

satisfying the following conditions.

R(*DO*XI()).

= 0. By *N*(0)

we obtain a distinguished triangle

= g\**N*(k)

the distinguished triangle

 = 0, for k > dim lI/lI pJ. iii. We have a distinguished triangle

*Proof*. For 0 k dim lI/lIpJ, define an object *N*(k)

f

For (II. 3.11) and (II. 3.12) and Lemma 2. 3.1, we have

i. If (x())(

*Proof*. [45].

i. *M*(0)

ii. *M*(k)

where

set *N*(k)

*M*(k)

for any

$$\sum\_{p} (-1)^{p} [\mathcal{H}^{p} (\mathcal{D} \mathcal{D} \mathcal{O}\_{\mathbf{X}\_{j}} (\lambda))]] = \sum\_{\mathbf{x} \in \Gamma(\lambda)} (-1)^{l(\mathbf{x}) - m(\mathbf{x})} [\mathcal{D} \mathcal{O}\_{\mathbf{X}\_{j}} (\mathbf{y}\_{\mathbf{x}} \mathbf{x}) \circ \lambda]\_{\mathsf{y}} \tag{\text{II. 3.17}}$$

in the Grothendieck group of the category of quasi-G-equivariant *D*XJ-modules [23].


$$\mathcal{R}(\mathcal{D}\mathcal{D}u(\lambda)) = \mathcal{D}\mathcal{D}u(\text{(y.x)}\text{ o}\lambda)[l(\text{x}) - \text{m}(\text{x})].\tag{\text{II. 3.18}}$$


$$\{\Phi \colon \mathcal{D}\mathsf{C}\mathsf{x}(\{\mathsf{w}\mathsf{w}\mathsf{u}\_{\mathsf{v}}\}) \circ \lambda\} \to \mathsf{H}^{0}(\mathsf{R}(\mathcal{D}\mathsf{C}\mathsf{x}(\lambda))),\tag{\mathsf{II.3.19}}$$

Moreover, , is an epimorphism if *l*(x) > m(x), for any x()\{e}.

*Proof*. The statements (i), (ii) and (iii) are obvious from proposition 2. 3. 1. The statement (iv) follows from proposition 2. 3. 1, and the fact that any locally free *O*X-module *L*, we have H*p*(R(D*L*)) = 0 for any *p*<0 . We assume that , satisfies the assumption in (v). Then we have e(), and ye = wJwIJ. Hence (v) follows from proposition 2. 3. 1.

Consider the following technical lemma, over classes of root spaces around of orbits with elements ye = wJwIJ.

### **Lemma II. 3.2.**


$$\{\mathbf{y} \ge \mathbf{x} \mid \mathbf{x} \in \Gamma(\lambda)\} = \{\mathbf{w} \in \mathbb{W} | \text{Wi} \mid (\mathbf{w}(\lambda + \rho))(\mathbf{a}^\vee) > 0, \text{ for any } \mathbf{a} \in \Lambda^+\}\}\tag{\Pi. 3.20}$$

And we have

$$l(\mathbf{x}) - \mathbf{m}(\mathbf{x}) = l(\mathbf{y}\_{\mathbf{x}}) + l(\mathbf{x}) - \mathbb{H}(\boldsymbol{\Delta}^{+})\,\mathrm{(\Delta)} = l(\mathbf{y}\_{\mathbf{x}}\mathbf{x}) - \mathbb{H}(\boldsymbol{\Delta}^{+})\,\mathrm{(\Delta)},\tag{11.3.21}$$

*Proof*. [45].

For (h\*Z)I, we set

$$\Xi(\lambda) = \{ \mathbf{w} \in \mathbf{W} | \mathbf{W} \mathbf{i} \mid \{ \mathbf{w}(\lambda + \rho) \} (\mathbf{a}^\vee) > 0, \text{ for any } \alpha \in \Lambda^+ \}, \tag{\Pi. 3.22}$$

Using the lemma 2. 3. 2, above we can reformulate the theorem 2. 3. 1, as follows:

Theorem II. 3.2. Let (h\*Z)I.

i. We have

$$\sum\_{p} (-1)^{p} [\mathcal{H}^{p} (\mathcal{D} \mathcal{O} \mathcal{O}\_{\mathbb{X}\_{\mathbb{L}}} (\lambda))] = (-1)^{\# (\Lambda\_{\mathbb{L}}^{+} : \Lambda\_{\mathbb{L}})} \sum\_{\mathbf{w} \in \Xi(\mathbb{L})} (-1)^{l(\mathbf{w})} [\mathcal{D} \mathcal{O}\_{\mathbb{X}\_{\mathbb{L}}} (\mathbf{w} \circ \lambda)] \,\!\!/ \tag{\text{II. 3.23}}$$

in the Grothendieck group of the category of quasi-G-equivariant *D*XJ-modules [23].


$$\Phi: \mathcal{D}\mathsf{CD}\mathfrak{x}(\text{((w.w.)})\text{o }\lambda) \to \mathsf{H}^{0}(\mathsf{R}(\mathcal{D}\mathsf{CD}\mathfrak{x}(\lambda))),\tag{\text{II.3.25}}$$

Moreover , is a epimorphism if *l*(x) > #( J\I), for any w()\{wJwIJ}.

## **Orbital Integrals on Reductive Lie Groups**

## **III. 1. Orbital integrals on reductive Lie groups [39]**

Let G, be of inner type. Let

$$\mathbf{K}^{\prime\prime} = \langle \mathbf{k} \in \mathbf{K} \mid \det(\mathbf{I} - \text{Ad}(\mathbf{k})|\mathbf{}) \neq \mathbf{0} \rangle,\tag{11.1}$$

If h, is of Cartan belonging to g, be H = exp(h). Let

$$\mathbf{H}' = \langle \mathbf{h} \in \mathbf{H} \mid \det(\mathbf{I} - \mathbf{A} \mathbf{d}(\mathbf{h}) | \mathbb{L}) \neq 0 \rangle,\tag{11.2}$$

Let

20 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Using the lemma 2. 3. 2, above we can reformulate the theorem 2. 3. 1, as follows:

J I I J

( 1) [ ( ( ( )))] ( 1) ( 1) [ ( ], *<sup>p</sup> <sup>l</sup>*

*D D <sup>p</sup>*

J\I), for any w(), then we have H*p*(R(*DO*XI())) = 0.

in the Grothendieck group of the category of quasi-G-equivariant *D*XJ-modules [23].

X X w H R w ) #( : ) ( )

( )

J\I)], (II. 3. 24)

J\I, then there exist a canonical morphism

*DO*XJ((wJwIJ)o ) H0(R(*DO*XI())), (II.3.25)

J\I), for any w()\{wJwIJ}.

) > 0, for any

w

*O O* (II. 3.23)

J}, (II. 3.22)

() = {wWJWI (w())(

Theorem II. 3.2. Let (h\*Z)I.

*p*

ii. If () = , then R(*DO*XI()) = 0.

iii. If (), consists of a single element w, then iv. R(*DO*XI()) = *DO*XJ(w o )[*l*(w) #(

) < 0, for any

Moreover , is a epimorphism if *l*(x) > #(

i. We have

v. If *l*(w) #(

vi. If ()(

$$\mathbf{G}[\mathbf{H}'] = (\mathbf{g} \mathbf{h} \mathbf{g}^{\perp} \mathbf{e} \mathbf{H} \big| \ \mathbf{h} \in \mathbf{H}', \mathbf{g} \in \mathbf{G}),\tag{\text{III.3}}$$

Then GH', is an open subset of G. Indeed, that be an open subset of G, it means that GH' Int(G) . Indeed, Int(G) Aut(G) G. The relation of their closed subgroups is transitive in G. Then for definition

$$\text{Int}(\mathbf{G}) = \langle \mathbf{g} \in \mathbf{G} \mid \text{d}\mathbf{I}\hat{\mathbf{g}} = \text{Ad}(\mathbf{g})\rangle,\tag{\text{III.4}}$$

But this is equal to that ĝInt(G),

$$\det(\text{I}\text{- }\text{Ad}(\hat{\mathbf{g}})|\!|\!\perp) = 0,\tag{\text{III. 5}}$$

Then

$$\text{Int}(\mathbf{G})^\perp = \langle \mathbf{g} \in \mathbf{G} \mid \det(\mathbf{I} \cdot \text{Ad}(\hat{\mathbf{g}})|\!\perp) = 0 \rangle,\tag{\text{III. 6}}$$

Since ghg-1 = Ad(h)h gG, then GH' Int(G) .

Now, of the proof of lemma of the Weyl's formula is deduced that

$$\int\_{\mathbf{C}[\mathbf{H}^{\mathsf{H}}]} \mathbf{f}(\mathbf{g})d\mathbf{g} = (1/\mathbf{w}) \int\_{\mathbb{H}} \mathbb{H} |\det(\mathbf{I} - \mathbf{A}\mathbf{d}(\mathbf{h})\! \! \! \! \! \! \mathbf{L})\Big| \, \text{c}\_{\mathsf{H}} \mathbf{f}(\mathbf{g}\!\! \! \! \! \mathbf{g}\!\! \! \! \! \! \! \mathbf{H}\!\! \! \! \! \! \mathbf{L}\!\! \! \! \! \! \mathbf{H}\!\! \! \! \! \! \! \mathbf{L}\!\! \! \! \! \! \mathbf{L}\!\! \! \! \! \! \mathbf{H}\!\! \! \! \! \! \! \! \! \! \mathbf{H}\!\!\! \!$$

where dg is an invariant measure on G, we fix an invariant measure on H, and we take to dgH, like the invariant quotient measure on G/H.

$$\text{Let } \mathbf{w} = \mathbf{o}(\mathbf{N}(\mathbf{H})/\mathbf{H}), \text{ be where } \mathbf{N}(\mathbf{H}) = \langle \mathbf{g} \in \mathbf{G} \mid \mathbf{g} \text{h} \mathbf{g}^{-1} = \mathbf{H} \rangle. \text{ Be } \mathfrak{h} = \mathfrak{t} \oplus \mathfrak{a}, \text{ and } \mathfrak{g} = \mathfrak{t} \oplus \mathfrak{p}. \text{ Let } \mathfrak{t}$$

$$\tau = (\mathbf{H} \in \mathfrak{h}) \Big|\ \det(\mathbf{ad}(\mathbf{H})|\_{\mathfrak{h}}) \neq 0\rangle,\tag{\text{III.8}}$$

which is equal to t, a subalgebra of Cartan of g, that consider to T, also as

$$\tau = \langle \mathbf{t} \in \Gamma \mid \det(\mathbf{Ad}(\mathbf{t}) | \boldsymbol{\perp}) \neq 0 \rangle,\tag{\text{III.9}}$$

of where K'' .

The previous integral formula applied to K, and G, implies that,

**Lemma III. 1** Exist a positive constant C, such that

$$\int\_{\mathbb{C}\times\mathbb{V}^{-}} \mathbf{f}(\mathbf{g})d\mathbf{g} = \mathbf{c} \Big| \kappa |\det(\mathbf{I} - \operatorname{Ad}(\mathbf{k})| \mathbf{\tilde{\boldsymbol{\sigma}}})| \operatorname{cf}(\mathbf{g} \,\mathrm{kg}^{-1}) \,\mathrm{dg} \,\mathrm{d}\mathbf{k},\tag{11.10}$$

*Proof.* Indeed, consider you

$$\mathbf{G}[\mathbf{K}^{\prime\prime}] = \langle \mathbf{g} \mathbf{k} \mathbf{g}^{\prime} \mathbf{e} \mathbf{H} \mid \mathbf{k} \in \mathbf{K}^{\prime\prime}, \mathbf{g} \in \mathbf{G} \rangle,\tag{\text{III. 11}}$$

with

$$\mathbf{K}^{\prime\prime} = \langle \mathbf{k} \in \mathbf{K} \mid \det(\mathbf{I} - \text{ad}(\mathbf{k})|\mathbf{\hat{\boldsymbol{\nu}}}) \neq \mathbf{0} \rangle,\tag{\text{III. 12}}$$

Since K, is compact then N(K) GK''. Then gKg1 G gG. Therefore,

$$|\det(\mathbf{I} - \mathbf{A}\mathbf{d}(\mathbf{h})\mathbb{L})| \le c|\det(\mathbf{I} - \mathbf{A}\mathbf{d}(\mathbf{k})\mathbb{L})|\tag{11.13}$$

Then by the integral formula of <sup>G</sup>K''f(g)dg, is had that:

$$\int \mathbf{c}\_{\mathbf{f}^{[k^\*]}} \mathbf{f}(\mathbf{g}) d\mathbf{g} = \int \kappa |\text{det}(\mathbf{I} - \text{Ad}(\mathbf{H}') \mathbb{I} \bot)| \text{c.v.f}(\mathbf{g} \text{kg}^{-1}) \text{ dg} \mathbf{K} \text{ d}\mathbf{k},\tag{\text{III. 14}}$$

and since gKg1 G, if and only if G/K G, then gK = g, of where is obtained the conclusion of lemma.

If > 0, then we have

$$\mathbf{G}\_{\mathbf{e},\varepsilon} = \langle \mathbf{g} \mathbf{t} \mathbf{g}^{-1} \vert \det(\mathbf{A} \mathbf{d}(\mathbf{t}) - \mathbf{I}) \vert \boldsymbol{\omega} \rangle \rangle \tag{11.15}$$

and

$$\mathbf{K}\_x = \langle \mathbf{k} \in \mathbb{K} \mid \det(\mathbf{Ad(k)} - \mathbf{I}) | \mathbf{\color{red}{\mathbf{\color{red}{\cdot \cdot \cdot}}}} \rangle \text{ (} \mathbf{\color{red}{\cdot \cdot \cdot}}\text{)}\tag{\text{III. 16}}$$

and (G'')e, = GK.

In fact, the defined subgroups explicitly and previously are the spherical forms of the orbits GK'', and K''.

Fixing a norm ,, on G, the which we assume given like the corresponding norm operator to a representation (, F), of G, on the Hilbert space of finite dimension F. Also we assume that (g)\* = (g), and that det(g) = 1, gG.

**Lemma III. 2** Let 0 << 1. Then

22 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

The previous integral formula applied to K, and G, implies that,

**Lemma III. 1** Exist a positive constant C, such that

Then by the integral formula of <sup>G</sup>K''f(g)dg, is had that:

of where K'' .

with

of lemma.

and

If > 0, then we have

and (G'')e, = GK.

GK'', and K''.

(g)\* = (g), and that det(g) = 1, gG.

*Proof.* Indeed, consider you

which is equal to t, a subalgebra of Cartan of g, that consider to T, also as

Since K, is compact then N(K) GK''. Then gKg1 G gG. Therefore,

T = tT det(Ad(t)t) 0, (III. 9)

GK'' = gkg-1H kK'', gG, (III. 11)

and since gKg1 G, if and only if G/K G, then gK = g, of where is obtained the conclusion

Ge, = gtg1 det(Ad(t) I)t)>, (III. 15)

In fact, the defined subgroups explicitly and previously are the spherical forms of the orbits

Fixing a norm ,, on G, the which we assume given like the corresponding norm operator to a representation (, F), of G, on the Hilbert space of finite dimension F. Also we assume that

<sup>G</sup>K''f(g)dg = cKdet((I Ad(k)p)Gf(gkg) dg dk, (III. 10)

K'' = kK det(I ad(k)p) 0, (III. 12)

det((I Ad(h)h) cdet((I Ad(k)p), (III. 13)

K = kK det(Ad(k) I)p)>, (III. 16)

<sup>G</sup>K''f(g)dg = Kdet((I Ad(H')h)G/Kf(gkg) dgK dk, (III. 14)

$$\left| \left( \mathbf{c}^{\circ} \right)\_{\circ} \mathbf{(1} + \log \| \mathbf{g} \|\right)^{\circ} \mathsf{E} \mathsf{(g)} \mathbf{dg} \le \mathsf{C} \mathsf{a}^{\circ \text{-d} \mathsf{2}} \mathsf{I}\_{\mathsf{A}} \mathsf{v} \mathsf{(a)} \left( \mathbf{1} + \log \| \mathbf{a} \|\right)^{\circ} \mathsf{E} \mathsf{(a)} \mathbf{da} \,\mathsf{A} \,\mathsf{B} \,\mathsf{v} \,\mathsf{E} \mathsf{(b)} \,\mathsf{A} \,\mathsf{B} \,\mathsf{V} \,\mathsf{A} \,\mathsf{B} \,\mathsf{V} \,\mathsf{A} \,\mathsf{B} \,\mathsf{V} \,\mathsf{A} \,\mathsf{V} \,\mathsf{A} \,\mathsf{B} \,\mathsf{V} \,\mathsf{A} \,\mathsf{V} \,\mathsf{B} \,\mathsf{V} \,\mathsf{A} \,\mathsf{V} \,\mathsf{B} \,\mathsf{V} \,\mathsf{A} \,\mathsf{V} \,\mathsf{B} \,\mathsf{V} \,\mathsf{A} \,\mathsf{V} \,\mathsf{B} \,\mathsf{V} \,\mathsf{V} \,\mathsf{A} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V} \,\mathsf{V$$

where is such that aA,

$$\gamma(\mathbf{a}) = \prod\_{aw} \text{an serh (u(H))}.\tag{11.18}$$

*Proof*. The lemma. III. 1, affirms that under the constants of normalization, if f, is integrable on G, then

1. Gf(g)dg = <sup>K</sup> A+ Kf(k1ak2)(a) dk1da dk2, (III. 19)

This implies that > 0,

$$\int\_{(\mathbf{G'})^{s,\mathfrak{c}}} (1 + \log \|\mathbf{g}\|)^{-\mathfrak{g}} \Xi(\mathbf{g}) \mathrm{d}\mathbf{g} = 0$$

 =Ke det((Ad(k) I)p)<sup>K</sup> A+ (1 + logau1kua1) d( au1kua1)du da dk, (III. 20)

Since u1ku = k, uK, with then

$$\begin{aligned} \left| \mathbb{\kappa} \left( \text{Ad}(\text{k}(\text{k}) - \text{I}) \mathbb{\left|} \right| \mathbb{\kappa}\_{\times^{\text{A+}}} \left( 1 + \log \| \text{au}^{-1} \text{kua}^{-1} \| \right) \right)^{-\text{d}} \Xi (\text{au}^{-1} \text{kua}^{-1}) \text{du da dk} = \\\\ = \int\_{\mathbb{\kappa}} \left| \text{det}((\text{I} - \text{Ad}(\text{k})) \| \cdot) \right| \left| \mathbb{\kappa}\_{\text{A+}} \left( 1 + \log \| \text{alka}^{-1} \| \right) \right|^{-\text{d}} \Xi (\text{ak} \text{a}^{-1}) \text{da dk}, \end{aligned} \tag{\text{III.21}}$$

If Xp, then (X), is self adjunct. Indeed, if Xp, then

$$(\mathbf{I} - \operatorname{Ad}(\mathbf{k})|\_{\mathbb{H}}) \chi = 0,\tag{11.22}$$

of where (adX) = (X). Thus (X), is self adjunct. Then if aA, then a = expH, with Ha. Thus,

$$\|\mathbf{l}\mathbf{k}\mathbf{a}^{-1}\| = \|\mathbf{l}\mathbf{k}\mathbf{a}^{-1}\mathbf{k}^{-1}\| = \|\mathbf{expH}\exp(-\mathbf{Ad}(\mathbf{k})\mathbf{H})\|\,\tag{\text{III. 23}}$$

Since e(I Ad(k))H = e(H) eAd(k)H. But by Apendix A, if (X) is self adjunct then tr(X) = 0, and dim F > 1, then tr(I Ad(k))H = 0, and due to that dim F > 1, then

$$\left\|\left|\mathbf{e}\_{\pi(\mathsf{H})}^{\mathsf{A}}\mathbf{e}\_{\mathsf{A}}\mathbf{v}(\mathsf{k})\right\|\right\| \geq \mathsf{G}\_{\mathsf{A}}(\mathsf{l}\_{1}\cdot\mathsf{A}\mathsf{v}(\mathsf{k})\mathsf{H}) \|\cdot\|\text{(prime}\_{\mathsf{L}}\cdot\mathsf{1})\tag{\mathsf{III}.24}$$

reason for which we see that

$$\log \| \text{aka}^{-1} \| \ge \| [\pi(\text{H} - \text{Ad}(\text{k})) \text{H}] \| / (\text{dim F} - 1) \text{.} \tag{\text{III. 25}}$$

We define kK, the minimal value (k) like

$$\mu(\mathbf{k}) \equiv \min \| (\text{Ad}(\mathbf{k}) - \mathbf{I}) \| \,\, \| \,\, \tag{\text{III. 26}}$$

Then we have demonstrated that

	- 2. a positive constant C, such that

$$\log \| \text{aka}^{-1} \| \ge \text{C\mu(k)} \text{log} \| \text{a} \| \text{.} \tag{\text{III.27}}$$

Let 1, , 2q, be the eigenvalues of (I Ad(k))p, with numerable multiplicity. If we assume that kK'', then (I Ad(k))p 0, and since (III. 26) then j = (k), j = 1, 2. Then is clear that j 2, i, since

$$\mu|\mu| = \mu(\mathbf{k}) \le \text{tr } \pi(\mathbf{X}) \text{ } \pi(\mathbf{X})^{\*} + \text{tr } \pi(\mathbf{X})^{-1} \pi(\mathbf{X})^{-1} \text{\*}= 1 + 1 = 2,\tag{11.78}$$

since det (X) = 1, xp. Thus, if kK, then

$$\varepsilon \ll |\mu\_1, \dots, \mu\_{2q}| \le \mu(\mathbf{k})^2 \mathcal{D}^{2q^{-2}},\tag{11.29}$$

If C= 2q + 1, then we have demonstrated that

3. If kK, then(k) C1/2.

This combined with (2) and the realized calculus to the begining implies that

$$\begin{aligned} \left| \mathbf{c}^{\square \gamma, \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \scriptstyle \end \end{aligned} \right] $$

If fC(G), then we define the function f ,as the map

$$\mathbf{f} \colon \mathbf{G} \to \mathbf{N}(\mathbf{G}) \tag{11.31}$$

with rule of corresponding

$$\mathbf{g} \mid \rightarrow \mathbf{f}^{\circ}(\mathbf{g}) = \left\lfloor \text{\textmu\textprime\textprime\textprime} \right\rfloor \text{\textmu\textprime\textprime},\tag{\text{III.32}}$$

If fCc (G), then define the map Qf, like

$$\mathbf{Q} \colon \mathbf{K} \to \mathbf{N(K)}\tag{11.33}$$

with rule of corresponding

$$\mathbf{k} \mid \rightarrow \mathbf{Q} \text{(k)}=\int \text{cf}(\mathbf{g} \, \text{kg}^{-1}) \text{dg},\tag{\text{III.34}}$$

with domain on the space of the kK, to the which the integral converge absolutely.

**Lemma III. 3** If fCc (G), then the domain of Qf, it contains K''. In consequence QfC(K'').

*Proof*. Let h(g) = f (g). Then

$$\text{[ch(gtg^{-1})dg = ]c|f^{\ast}(\text{gtg^{-1}})|dg} \tag{\text{III. 35}}$$

But f (gtg1)(1 + loga) d(gtg1)1, 1, r(f), with 1, 1, r(f) = 1, since fCc (G). But by lemma III. 2, and considering (a) = (1 + loga) d, with d > 0, we have that

$$\int \mathrm{c}(1+\log|\mathrm{a}|)^{-d} \Xi(\mathrm{g}\mathrm{tg}^{-}) \mathrm{dg} = \int\_{A+\chi^{\mathrm{K}}} \chi(\mathrm{a}) \, \Xi(\mathrm{autu}^{-}\mathrm{la}^{-}) \mathrm{da} \, \mathrm{d}u,\tag{\mathrm{III. 36}}$$

Then

24 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

logaka1 C(k)loga, (III. 27)

Let 1, , 2q, be the eigenvalues of (I Ad(k))p, with numerable multiplicity. If we assume that kK'', then (I Ad(k))p 0, and since (III. 26) then j = (k), j = 1, 2. Then is clear that

This combined with (2) and the realized calculus to the begining implies that

(aka1)da dk CdC1(d/2)A+ (a)(1 + loga)

j = (k) tr (X) (X)\* + tr (X)(X)\* = 1 + 1 = 2, (III. 28)

d(g)dg Cd1/2Kdet((Ad(k) I)p)A(a)(1 + loga)

,as the map

Qf : K N(K) (III. 33)

k ∣Qf(k) = Gf(gkg1)dg, (III. 34)

(G), then the domain of Qf, it contains K''. In consequence QfC(K'').

with domain on the space of the kK, to the which the integral converge absolutely.

Gh(gtg1)dg = Gf

<1, , 2q(k)222q 2, (III. 29)

: G N(G) (III. 31)

(g) = K f(kgk1)dk, (III. 32)

(gtg1)dg, (III. 35)

d

dK(aka1)da dk, (III. 30)

2. a positive constant C, such that

since det (X) = 1, xp. Thus, if kK, then

If C= 2q + 1, then we have demonstrated that

(G'')e, (1 + logg)

If fC(G), then we define the function f

f

g ∣ f

(G), then define the map Qf, like

3. If kK, then(k) C1/2.

with rule of corresponding

with rule of corresponding

**Lemma III. 3** If fCc

(g). Then

*Proof*. Let h(g) = f

If fCc

j 2, i, since

$$\frac{1}{c} \text{|f'(gtg^{-1})| dg} = \int\_{\Lambda \vdash \chi} \chi(\mathbf{a}) \mathbf{h}(\text{autu}^{-1}\mathbf{a}^{-1}) d\mathbf{a} \text{ du},\tag{\text{III.37}}$$

Thus

$$\text{lcf}(\text{gtg}^{-})\text{dg} = \bigvee\_{\mathbb{A}\hookrightarrow\mathbb{A}}\gamma\_{\mathbb{A}}\gamma(\text{a})\text{h}(\text{autu}^{-1}\text{a}^{-1})\text{da du},\tag{\text{III.38}}$$

Then the argument of the demonstration in the lemma III. 2 (2), demonstrates that

4. If kK,and if aA, then logata11/2Cloga,which implies that if uCc (G) and if supp u B r(G), where B r(G), is the full ball with r > 0,

$$\mathbf{a} \cdot \mathbf{(G)} = \langle \mathbf{g} \in \mathbf{G} \mid \log \|\mathbf{g}\| \le \mathbf{r} \rangle,\tag{11.39}$$

Then exist c > 0, independient of u, such that u(autu1a1) = 1, only into of the region

B r(G). Then the integral

$$\int \mathbf{A} \star \! \! \times \! \! \mathbf{y} \text{(a)} \mathbf{h} \text{(autu}^{-1} \mathbf{a}^{-1}\text{)} \text{da du} \text{<} \! \! \! \! \! \times \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \! \!$$

kK'', that is to say; h(Ge, ) B r(G), thus K K''. Then the integral converge kK''.

Now we demonstrates that QfC(K''). Indeed, let X1, , Xn, be a base of g. If Yt, then

$$\text{Ad}(\mathbf{g})\mathbf{Y} = \Sigma \mathbf{C}\_{\mathbf{i}}(\mathbf{g})\mathbf{X}\_{\mathbf{i}} \tag{\text{III. 41}}$$

with each Cj, a matrix coefficient of the representation (Ad, g), of where

$$\mathbf{d}^{\mathbb{k}} / \mathbf{d}\mathbf{t}^{\mathbb{k}} |\_{\mathbb{t}^\*} \mathbf{d}\mathbf{f} (\mathbf{g}\mathbf{k} \exp \mathbf{t} \mathbf{Y} \mathbf{g}^{-1}) = (\mathbf{A} \mathbf{d} (\mathbf{g}) \mathbf{Y})^{\mathbb{k}} \mathbf{f} (\mathbf{g} \mathbf{k} \mathbf{g}^{-1}) = (\sum \mathbf{C} (\mathbf{g}) \mathbf{X})^{\mathbb{k}} \mathbf{f} (\mathbf{g} \mathbf{k} \mathbf{g}^{-1}), \tag{\text{III. 42}}$$

Thus exist constants D > 0, and u 0, such that Cj(g) Dgu, of where

$$|\mathbf{d}^{\mathbf{k}} / \mathbf{d}\mathbf{t}^{\mathbf{k}}|\_{\mathbf{t}=\mathbf{0}} \mathbf{f} (\mathbf{g} \mathbf{k} \, \exp \mathbf{Y} \mathbf{g}^{-1}) | \le \mathbf{C} \| |\mathbf{g}|^{\mathbf{k}\mathbf{u}} \Sigma |\mathbf{u} \mathbf{f} (\mathbf{g} \mathbf{k} \mathbf{g}^{-1}) | \,\tag{\text{III. 43}}$$

where uj, is a base of Uk(g**C**). Newly this implies that if fC(G), r > 0, such that

$$|\mathbf{f}(\mathbf{g}\mathbf{k}\mathbf{g}^{-1})| \le r,\tag{11.44}$$

gG, and kK``, since ujf(gkg1)ujf(gkg1) = ujr. Then QfC(K'').

We fix to , on (g**C**, t**C**). We make that , is T-integral (which is possible due to that always is feasible to give a covering group of G).

Let W = W(g**C**, t**C**), be then

$$
\Lambda(\mathbf{t}) = \mathbf{t}^{\rho} \Pi\_{w \times \mathbf{0}} + (\mathbf{1} - \mathbf{t}^{-\alpha}) = \Sigma\_{\mathbf{s}} \mathbb{w} \det(\mathbf{s}) \mathbf{t}^{\rho}, \tag{11.45}
$$

 tT. Let T'' = T K''. If fCc (G), then

5. Ff T(t) = (t)G f(gtg1)dg, tT''.

Indeed, since Ad(G) G, and gtg1gTg1 gG. Then Ad(G)t gTg1N(T). Thus gG, Ad(g)t = t, tT, since the space of actions on the torus T, is

$$\text{Ad(G)t} = \langle \mathbf{g} \in \mathbf{G} \mid \text{Ad(g)t} \subset \mathbf{T} \,\,\forall \,\,\mathbf{t} \in \mathbf{T} \mid \rangle,\tag{\text{III. 46}}$$

Then

$$\text{lcf}(\text{Ad}(\text{g})\text{t})\text{dg} = \text{lcf}(\text{g})\text{dg} \tag{\text{III. 47}}$$

since the left member of (III. 46) absolutely converge tT'', and fCc (G). Also that,

$$\text{[cf(g)dg = π(t)]cf(gtg^{-1})dg} \tag{\text{III. 48}}$$

But for the lemma lemma III. 1, and since

$$\mathbf{T}^{\prime\prime} = \{ \mathbf{t} \in \mathbf{T} \mid \det(\mathbf{A}(\mathbf{t})|\_{\mathbb{P}}) \neq 0 \},\tag{\text{III. 49}}$$

we have that

$$\int \mathrm{cf}(\mathrm{g})d\mathrm{g} = \left[\mathrm{r}^{\circ}|\mathrm{det}(\mathrm{Ad}(\mathrm{t})|\mathrm{\mathfrak{h}})\right] \mathrm{cf}(\mathrm{gtg}^{-1})\mathrm{d}\mathrm{g}\mathrm{dt},\tag{\mathrm{III.50}}$$

But in W = W(g**C**, t**C**), is satisfied that

$$\left|r^{\circ}\det(\text{Ad}(\mathbf{t})|\_{\mathsf{P}})\right|\mathsf{d}\mathbf{t}=\sum\_{\mathsf{c}}\mathsf{c}\_{\mathsf{c}}\mathsf{wdet}(\mathbf{s})\mathsf{t}^{\circ\mathsf{P}}=\Delta(\mathsf{t}),\tag{\text{III. 51}}$$

of where

$$\mathrm{Fr}^{\mathrm{T}}(\mathbf{t}) = \Delta(\mathbf{t}) \Big| \mathrm{cf}(\mathbf{g} \mathrm{tg}^{-1}) \mathrm{dg} \Big| \tag{11.52}$$

tT''.

Likewise, the lemma III. 1, implies that Ff T(t)C(T''). Consider

$$\mathbf{T}\mathbf{c} = \mathbf{G}\_{\mathbf{c},\varepsilon} \cap \mathbf{T} = \{\mathbf{t} \in \mathbf{T} \mid |\Delta(\mathbf{t})|^2 \gtrsim \mathbf{c}\},\tag{\text{III. 53}}$$

which is equaldet(Ad(t)p) = n(t) 2, tT. But (t) 2>, tT, then det(Ad(t)p) 0, tT. Then TT''.

**Lemma. III. 4** Let d, be such that

26 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

(1 t

Indeed, since Ad(G) G, and gtg1gTg1 gG. Then Ad(G)t gTg1N(T). Thus gG,

(G), then

since the left member of (III. 46) absolutely converge tT'', and fCc

T''det(Ad(t)p)dt = <sup>s</sup>Wdet(s)ts

Ad(g)t = t, tT, since the space of actions on the torus T, is

is feasible to give a covering group of G).

(t) = t

T(t) = (t)G f(gtg1)dg, tT''.

But for the lemma lemma III. 1, and since

But in W = W(g**C**, t**C**), is satisfied that

Ff

Likewise, the lemma III. 1, implies that Ff

T = Ge, T = tT (t)

tT. Let T'' = T K''. If fCc

, on (g**C**, t**C**). We make that , is T-integral (which is possible due to that always

) = <sup>s</sup>W det(s)ts

Ad(G)t = gG Ad(g)t T tT , (III. 46)

G f(Ad(g)t)dg = G f(g)dg, (III. 47)

G f(g)dg = (t)G f(gtg1)dg, (III. 48)

T'' = tT det(A(t)p) 0, (III. 49)

T(t) = (t)G f(gtg1)dg, (III.52)

G f(g)dg = T''det(Ad(t)p)G f(gtg1)dgdt, (III. 50)

T(t)C(T''). Consider

, (III. 45)

(G). Also that,

= (t), (III. 51)

2>, (III. 53)

We fix to

5. Ff

Then

we have that

of where

tT''.

Let W = W(g**C**, t**C**), be then

$$\int\_{\mathbb{A}} \mathbf{\hat{z}} \cdot (1 + \log \|\mathbf{a}\|)^{-d} \Xi(\mathbf{a})^2 \mathbf{\hat{y}}(\mathbf{a}) d\mathbf{a} < \infty,\tag{11.54}$$

Then exist positive constants C, and u, such that if fCc (G), then

$$\int \mathrm{Tr} |\mathrm{F}i^{\mathrm{T}}(\mathrm{t})| \, \mathrm{d}\mathbf{t} \le \mathrm{C}e^{\mathrm{u}}\sigma\mathbf{1}, \, \mathrm{d}\_{\prime} \tag{\mathrm{III. 55}}$$

*Proof*. Let= 1, 1, d,be and let

$$\int\_{\mathbb{C}\times\mathbb{C}\_{\approx\gamma}} \mathbf{r}|\mathbf{f}(\mathbf{g})| \, \mathrm{d}\mathbf{g} \leq \int\_{\mathbb{R}} \mathbf{r}|\mathbf{F}\mathbf{i}^{\mathrm{T}}(\mathbf{t})| \, \mathrm{d}\mathbf{t} \leq \int\_{\mathbb{T}} \mathbf{r}|\Delta(\mathbf{t})|^{2} \|\mathbf{c}\| \, \mathrm{|f(\mathbf{g}\,\mathrm{tg}^{-1})| \, \mathrm{dg} \,\mathrm{d}\mathbf{t},\tag{\text{III. 56}}.$$

Now (t) C1/2, tT. Thus

$$\int\_{\mathbb{T}\_{\delta}} |\mathbf{F}\mathbf{i}^{\mathrm{T}}(\mathbf{t})| \, \mathrm{d}\mathbf{t} \leq \mathrm{C} \varepsilon^{-1/2} \|\mathbf{r}\_{\mathrm{i}}[\mathbf{A}(\mathbf{t})] \| \, \mathrm{c} | \, \mathrm{f}(\mathbf{g}\,\mathrm{tg}^{-1}) | \mathrm{dg} \, \mathrm{d}\mathbf{t}$$

$$\leq \mathrm{C} \varepsilon^{-1/2} \,\, \sigma(\mathbf{f}) \Big[ \mathrm{r}\_{\mathrm{i}} |\mathbf{A}(\mathbf{t})|^{2} \Big[ \mathrm{c} (1 + \log \|\mathbf{g}\,\mathrm{tg}^{-1}\|)^{\mathrm{-d}} \Xi(\mathbf{g}\,\mathrm{tg}^{-1})^{2} \gamma(\mathbf{g}\,\mathrm{tg}^{-1}) \mathrm{d}\mathbf{t} \, \mathrm{dg} \,\|\mathbf{g}\,\|\right]$$

$$= \mathrm{C} \varepsilon^{-1/2} \,\, \sigma(\mathbf{f}) \Big[ \mathrm{c} \varepsilon\_{\mathrm{d}} (1 + \log \|\mathbf{g}\,\|\|)^{-\mathrm{d}} \Xi(\mathbf{g}) \mathrm{d}\mathbf{g}$$

$$\leq \mathrm{C} \varepsilon \sigma(\mathbf{f}) \varepsilon^{-(\mathrm{d}+1)/2} \Big[ \mathrm{A} + \gamma(\mathbf{a}\,\|\mathbf{a}\,\|)^{-\mathrm{d}} \Xi(\mathbf{a})^{2} \mathrm{da} \,\|\mathbf{g}\,\|\}$$

$$\leq \mathrm{C} \varepsilon \sigma(\mathbf{f}) \varepsilon^{-(\mathrm{d}+1)/2} \Big[ \mathrm{A} + \gamma(\mathbf{a}) (1 + \log \|\mathbf{a}\,\|)^{-\mathrm{d}} \Xi(\mathbf{a})^{2} \mathrm{da} \,\,\mathrm{d}\mathbf{g} \,\|\}$$

By the lemma III. 1, if we take u= (d + 1)/2, then is followed (III. 54), since

$$|\text{Fe}^{\text{T}}(\text{t})| \leq \text{or} \text{, } \text{a}(\text{f})\Xi(\text{a})(1+\log||\text{a}||)^{\text{-d}} \text{ }\text{ }\tag{\text{III. 57}}$$

Now well,

$$\text{j}\,\text{s}\,\Xi \,\text{(aka}^{-1})\text{dk}=\Xi\text{(a)}\Xi\text{(a}^{-1})=\Xi\text{(a)}^{2},\tag{\text{III. 58}}$$

Then using the precedent inequalities we have that

$$\int\_{(G',\mathbb{C})} (1+\log\|\mathbf{g}\|)^{-\mathsf{d}} \,\mathsf{E}\,\mathsf{(g)}\mathrm{dg} = \int\_{A} +\gamma(\mathsf{a})(1+\log\|\mathbf{a}\|)^{-\mathsf{d}} \mathsf{E}\,\mathsf{(a)}^{2}\mathrm{da},\tag{\text{III. 59}}$$

If fC(G), then we define the function f , as the map

$$\text{If } \mathbf{f} \colon \mathbf{G} \to \mathbf{N}(\mathbf{G}), \tag{\text{III.60}}$$

with rule of correspondence

$$\mathbf{g} \mid \rightarrow \mathbf{f}^{\circ}(\mathbf{g}) = \left| \times \mathbf{f}(\mathbf{k} \,\mathrm{g} \,\mathrm{k}^{-}) \mathrm{d} \mathbf{k} \right| \tag{\text{III.61}}$$

If fCc <sup>∞</sup>(G), then we define the map Qf, as

$$\mathbb{Q}\colon \mathbb{K}\to\mathbb{N}(\mathbb{K})\tag{\text{III.62}}$$

with rule of correspondence

$$\mathbf{Q}: \mathbf{k} \mid \rightarrow \mathbf{Q}(\mathbf{k}) = \bigvee \mathbf{c} \mathbf{f}(\mathbf{g} \mathbf{k} \mathbf{g}^{-1}) \mathbf{d} \mathbf{g} \tag{\text{III.63}}$$

With domain on the space of the kK, to the which, the integral converges absolutely.

**Lemma. III. 5** If fCc <sup>∞</sup>(G), then the domain of Qf, includes K''. In consequence QfC∞(K'').

*Proof*. Let h(g) = f (g). Then

$$\int \mathrm{ch}(\mathrm{gtg}^{-1})\mathrm{dg} = \int \mathrm{cl}\dagger^\*(\mathrm{gtg}^{-1}) \mathrm{dg} \,\tag{\text{III. 64}}$$

but f (gtg1) ≤ (1 + loga) d(gtg1)1, 1, r(f), by lemma III. 2, with 1, 1, r(f) = 1, since fCc <sup>∞</sup>(G).

If fC(G), then we write (III. 60) and if fCc <sup>∞</sup>(G), then (III. 62). But by lemma. 3, and considering (a) = (1 + loga) d, with d > 0, we have that

$$\frac{1}{2}\left[\mathbf{c}(1+\log\|\mathbf{a}\|)\right]^{-1}\boldsymbol{\Xi}\text{ (gt}\mathbf{g}^{-1}\text{)}\mathbf{dg} = \left[\boldsymbol{\Lambda}+\boldsymbol{\tau}\cdot\mathbf{x}\,\gamma\text{(a)}\boldsymbol{\Xi}\text{(autu}^{-1}\text{a}^{-1}\text{)}\text{dadu},\tag{\text{III.65}}\right]$$

Then

$$\times \text{|}\text{\{f\{gtg^{-1}\}|d\mathbf{g}=\}} \times \text{+} \times \text{>} \text{(a)} \text{h\{autu^{-1}a^{-1}\}dadu,} \tag{\text{III.66}}$$

then

$$\text{[ch(gtg}^{-1})\text{ dg} = \text{l}\_{\text{l}} + \text{ }\_{\text{x}} \text{y} \text{(a)} \text{h(autu}^{-1} \text{a}^{-1}) \text{dadu}, \tag{\text{III.67}}$$

Therefore the argument of the demonstration in the lemma III. 2 (2), demonstrates that

6. If kK, and if aA, then logata11/2Cloga,which implies that if uCc (G), and if supp u B 1, 1, r(G), where B 1, 1, r(G), (where B 1, 1, r(G) = B r(G)) is the full ball with r > 0.

Then exist C > 0, independent of u, such that u(autu1a1) = 0, to kK'', actually;

h(Ge, ) B r(G), thus K K''. Thus the integral converge kK''.

Now we demonstrate that QfC∞(K''). In effect, be X1, X2, , Xn, a base of g. If Yt, then Ad(g)Y = Cj(g)Xj, with each Cj,a matrix coefficient of the representation (Ad, g), of where

$$(\mathbf{d}^k / \mathbf{d}\mathbf{t}^k |\_{\mathbf{t} = 0} \mathbf{f} (\mathbf{g} \mathbf{k} \, \exp \mathbf{t} \mathbf{Y} \mathbf{g}^\perp) = (\mathbf{A} \mathbf{d} (\mathbf{g}) \mathbf{Y})^k \mathbf{f} (\mathbf{g} \mathbf{k} \mathbf{g}^{-1}) = (\Sigma \mathbf{C} (\mathbf{g}) \mathbf{X})^k \mathbf{f} (\mathbf{g} \mathbf{k} \mathbf{g}^{-1}),$$

Exist constants D > 0, and u 0, such that Cj(g) ≤ Dgu.

Note: Ad(g) = dIg, gG and Ad(g)Y = d/dtt = 0gtK,

thus

28 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Qf : k ∣Qf(k) = Gf(gkg1)dg, (III. 63)

<sup>∞</sup>(G), then the domain of Qf, includes K''. In consequence QfC∞(K'').

d(gtg1)1, 1, r(f), by lemma III. 2, with 1, 1, r(f) = 1, since fCc

d (gtg1)dg = A+ × K (a)(autu1a1)dadu, (III. 65)

(gtg1)dg =A+ × K (a)h(autu1a1)dadu, (III. 66)

Gh(gtg1) dg =A+ × K (a)h(autu1a1)dadu, (III. 67)

With domain on the space of the kK, to the which, the integral converges absolutely.

Gh(gtg1)dg = Gf

d, with d > 0, we have that

Therefore the argument of the demonstration in the lemma III. 2 (2), demonstrates that

supp u B 1, 1, r(G), where B 1, 1, r(G), (where B 1, 1, r(G) = B r(G)) is the full ball with r > 0.

Now we demonstrate that QfC∞(K''). In effect, be X1, X2, , Xn, a base of g. If Yt, then Ad(g)Y = Cj(g)Xj, with each Cj,a matrix coefficient of the representation (Ad, g), of where

dk/dtkt = 0f(gk exptYg-1) = (Ad(g)Y)kf(gkg1) = (Cj(g)Xj)kf(gkg1),

6. If kK, and if aA, then logata11/2Cloga,which implies that if uCc

Then exist C > 0, independent of u, such that u(autu1a1) = 0, to kK'', actually;

h(Ge, ) B r(G), thus K K''. Thus the integral converge kK''.

Exist constants D > 0, and u 0, such that Cj(g) ≤ Dgu.

Qf : K N(K), (III. 62)

(gtg1)dg, (III. 64)

<sup>∞</sup>(G), then (III. 62). But by lemma. 3, and

<sup>∞</sup>(G).

(G), and if

<sup>∞</sup>(G), then we define the map Qf, as

with rule of correspondence

(gtg1) ≤ (1 + loga)

considering (a) = (1 + loga)

(g). Then

If fC(G), then we write (III. 60) and if fCc

G(1 + loga)

Gf 

**Lemma. III. 5** If fCc

*Proof*. Let h(g) = f

but f 

Then

then

If fCc

$$|\mathbf{d}^{\mathbf{k}} / \mathbf{d}\mathbf{t}^{\mathbf{k}}|\_{\mathbf{t}=0} \text{ f}(\mathbf{g}\mathbf{k}\,\,\text{espt}\,\mathbf{Yg}^{-1}) \mathbf{f} \le \mathbf{C} ||\mathbf{g}||^{\text{ku}} \boldsymbol{\Sigma} |\mathbf{u} \mathbf{f}(\mathbf{g}\,\mathbf{k}\,\mathbf{g}^{-1}) |\tag{\text{III.68}}$$

where uj, is a base of Uk(g**C**). Newly, this implies that if fCc (G), r > 0, such that

$$|\mathbf{f}(\mathbf{g} \mathbf{k} \mathbf{g}^{\top})| \le \mathbf{r},$$

gG, and kK'', since

$$\Sigma|\mathbf{u}\mathbf{f}(\mathbf{g}\mathbf{k}\mathbf{g}^{-1})| \le \Sigma|\mathbf{u}||\mathbf{f}(\mathbf{g}\mathbf{k}\mathbf{g}^{-1})| \le \Sigma|\mathbf{u}||\mathbf{r}, \mathbf{r}$$

Thus QfC∞(K'').

We fix to on(g**C**, t**C**). We assume that , is T-integral (which is possible due to that always is easy to give a covering group of G).

Let W = W(g**C**, t**C**), be then

$$
\Lambda(\mathbf{t}) = \mathbf{t}^{\rho} \prod\_{a: a \nmid \ast} (\mathbf{1} - \mathbf{t}^{-a}) = \sum\_{\ast} \mathbb{1}\_{a} \mathbb{W} \det(\mathbf{s}) \mathbf{t}^{\rho}, \tag{\text{III. 69}}
$$

tT. Let T'' = T K''. If fCc (G), then

1. Ff T(t) = (t)G f(gtg1)dg, tT''.

Indeed, since Ad(G) G and gtg1gTg1, gG. Then

$$\operatorname{Ad}(\mathcal{G}) \\
\mathfrak{t} \left( \Box \mathcal{g} \mathcal{T} \mathcal{g}^{-1} \right) \in \operatorname{N}(\mathcal{T}).$$

We can demonstrate that:

$$\begin{pmatrix} 1' \end{pmatrix} \ \mathbf{F}\_{\mathbf{z}\mathbf{f}}^{\mathsf{T}}(\mathbf{t}) = \boldsymbol{\gamma}(\mathbf{z}) \mathbf{F}\_{\mathbf{f}}^{\mathsf{T}}(\mathbf{t}) \boldsymbol{\omega}$$

 fCc (G), tT', and zZ(g). In effect, for definition

$$\mathbf{F} \mathbf{t}^{\mathrm{T}}(\mathbf{t}) = \Delta(\mathbf{t}) \Big| \mathbf{c} \,\mathrm{f}(\mathbf{g} \,\mathrm{tg}^{-1}) \mathrm{dg}\_{\mathrm{V}} \,\mathrm{h}$$

 fCc (G), tT''. If zZ(g**C**), then (z), is a homomorphism of Harish-Chandra defined as

$$
\gamma: \mathbf{Z(çc)} \to \mathbf{U(0c)},
\tag{\text{III.70}}
$$

with rule of correspondence

$$\mathbf{z} \mid \rightarrow \Delta (\mathbb{S} \langle \mathbf{z} \rangle) \boldsymbol{\Delta}^{-1} = \mathbf{P}\_{\mathsf{h}} \tag{11.71}$$

hH, such that (z) = ((z)). Then (t)(z) = (z)(t). Thus

$$\mathbf{F}\_{\rm tr} \mathbf{\hat{r}} (\mathbf{t}) = \boldsymbol{\Delta} (\mathbf{t}) \mathbf{\hat{c}} \cdot \mathbf{z} \mathbf{f} (\mathbf{g} \,\mathbf{t} \,\mathbf{g}^{-1}) \mathbf{d} \mathbf{g} = \boldsymbol{\Delta} (\mathbf{t}) \mathbf{\hat{c}} \,\boldsymbol{\delta} (\mathbf{z}) \mathbf{f} (\mathbf{g} \,\mathbf{t} \,\mathbf{g}^{-1}) \mathbf{d} \mathbf{g} \,\tag{11.72}$$

since to , a character of a representation of finite dimension of G, is

$$\mathbf{z}\boldsymbol{\sigma}\_{\mu}|\_{\mathbb{H}} = \delta(\mathbf{z})(\boldsymbol{\sigma}\_{\mu}|\_{\mathbb{H}})\prime$$

which is the same that

$$\text{zf}(\text{gtg}^{-1}) = \delta(\text{z})\text{f}(\text{gtg}^{-1})\dots$$

Then (III. 71), take the form

$$
\langle \delta(\mathbf{z}) \Delta(\mathbf{t}) \Big| \mathbf{c} \,\mathbf{f}(\mathbf{g} \,\mathbf{t} \mathbf{g}^{-1}) \mathbf{d} \mathbf{g} = \gamma(\mathbf{z}) \mathbf{F}^{\mathrm{T}}(\mathbf{t}).
$$

 fCc (G), tT', and zZ(g**C**).

Let 1,2,,n, be elements in . Be T' = tT ti 1, I = 1, 2, , r. If > 0, then

$$\mathbf{T'} = \{ \mathbf{t} \in \mathbf{T} \mid |\mathbf{1} - \mathbf{t}^{\mathbf{a}}| \succeq \mathbf{e}\_{\prime} \; \forall \; \mathbf{i} = \mathbf{1}, \; \mathbf{2}, \; \dots, \mathbf{r} \}, \tag{\text{III. 73}}$$

Let

$$\mathbf{B(T') = \{f \in C^\circ(T') \mid \varpi(f) < \infty, \forall \ P \in \mathbf{U(tc)}\}}, \tag{11.74}$$

Then is necessary show the following basic Harish-Chandra theorem:

### **Theorem III. 1.**

1. If fCc (G), then Ff <sup>T</sup>B(T').

2. The map f ∣Ff T, of Cc (G) in B(T'), extends to a continuous map of C(G), in B(T'). *Proof.* Let V = C(G), and W = Cc (G), s(w) = Fw, A = Z(g**C**), = , and = C. Then s(f)B(T'), fCc (G), and s, extends to a continuous map of C(G), in B(T').

**Theorem III. 2.** Let , be a continuous semi-norm on B(T'). Then exist d 0, and a continuous semi-norm , on C(G), such that

$$\int \mathbb{K} \,\mu(\mathbb{F}\_{\mathbb{R}}^{\mathbb{T}}(\eta \mathbb{g}) \mathbf{f}) \, \mathrm{d}\mathbf{k} \le (1 + \log \|\mathbf{g}\|)^{d} \Xi(\mathbf{g}) \sigma(\mathbf{f}).\tag{11.75}$$

to fC(G), and gG.

For the previous theorem III. 1, exist a continuous semi-norm , of C(G), and q, d > 0, such that if fC(G), then

$$\int\_{\mathbb{T}} \mathbb{1}\_{\mathsf{f}} |\mathsf{F}\_{\mathsf{f}} \mathsf{F}\_{\mathsf{f} \mathsf{g} \mathsf{f}}(\mathsf{f})| \mathsf{d}\mathsf{t} \,\mathsf{d}\mathsf{k} \leq \varepsilon^{-\rho} \mathsf{a}(\mathsf{f}) \,(1 + \log \|\mathsf{g}\|)^{\rho} \Xi(\mathsf{g}), \quad \forall \varepsilon > 0,\tag{\text{III.76}}$$

For the lemma. III. 5,

30 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

hH, such that (z) = ((z)). Then (t)(z) = (z)(t). Thus

which is the same that

fCc

Let

**Theorem III. 1.** 

2. The map f ∣Ff

to fC(G), and gG.

that if fC(G), then

1. If fCc

fCc

Then (III. 71), take the form

Let 1,2,,n, be elements in

(G), then Ff

*Proof.* Let V = C(G), and W = Cc

(G), tT', and zZ(g**C**).

since to , a character of a representation of finite dimension of G, is

T' = tT1 - t<sup>i</sup>

Then is necessary show the following basic Harish-Chandra theorem:

(G), and s, extends to a continuous map of C(G), in B(T').

<sup>T</sup>B(T').

T, of Cc

continuous semi-norm , on C(G), such that

FzfT(t) = (t)G zf(gtg)dg = (t)G (z)f(gtg)dg, (III. 72)

zH = (z)(H),

zf(gtg) = (z)f(gtg),

(z)(t)G f(gtg)dg = (z)Ff

T(t).

B(T') = fC(T')P(f) <, PU(t**C**), (III. 74)

(G) in B(T'), extends to a continuous map of C(G), in B(T').

**Theorem III. 2.** Let , be a continuous semi-norm on B(T'). Then exist d 0, and a

For the previous theorem III. 1, exist a continuous semi-norm , of C(G), and q, d > 0, such

(G), s(w) = Fw, A = Z(g**C**), = , and = C. Then s(f)B(T'),

K (FRT(kg)f(f))dk (1 + logg)d(g)(f), (III. 75)

KTFRT(kg)f(f))dtdk q(f) (1 + logg)d(g), > 0, (III. 76)

>, i = 1, 2, , r, (III. 73)

. Be T' = tT ti 1, I = 1, 2, , r. If > 0, then

$$\int\_{\mathbb{K}} \int\_{\mathbb{K}} |\mathsf{F}\_{\mathbb{K}} \mathsf{F}\_{(\mathsf{k}\mathsf{g})} \mathsf{f}(\mathsf{f})| \mathsf{d}\mathsf{t} \,\mathrm{d}\mathsf{k} = \int\_{\mathbb{K}} \int\_{\mathbb{K}} |\mathsf{T}\_{\mathsf{e}} |\Delta(\mathsf{T})| \left| \mathsf{c} \,\middle|\mathsf{f}(\mathsf{x} \mathsf{t} \mathsf{x}^{-1} \mathsf{k} \mathsf{g}) \right| \,\mathrm{d}\mathsf{x} \,\mathrm{d}\mathsf{t} \,\mathrm{d}\mathsf{k},$$

of where , on C(G), and > 0, since (t) C1/2, tT

KT(T))Gf(xtxkg)) dxdtdk (f)1/2KT(T)) 2(1 + logxtxkg)d(xtxkg)dxdtdk, with = 1, 1, d.

Note that the norm in G, satisfies x, yG, that

$$||\mathbf{x}\mathbf{y}|| \ge ||\mathbf{x}|| ||\mathbf{y}^{-1}||^{-1} = ||\mathbf{y}||^{-1} ||\mathbf{x}||\_{\nu}$$

Thus logxy + logy logx.Of where

(1 + logxy)(1 + logy) 1 + logx,

Thus the expression (III. 75), take the form

$$\int \mathbf{k} \mathbf{r}\_t |\mathbf{F} \mathbf{k}^T \mathbf{u}\_{\partial \Omega} \mathbf{t}(\mathbf{f})| \, \text{d}\mathbf{t} \, \text{d}\mathbf{k} \leq \mathbf{z}^{-1/2} \boldsymbol{\sigma}(\mathbf{f}) \left(1 + \log \|\mathbf{g}\|\right) \mathbf{t} \, \text{r}\_{\times} \, \text{r}\_t |\Delta(\mathbf{t})|^2 \mathbf{k} \, \text{(1 + \log \|\mathbf{x}\mathbf{t}^{-1} \mathbf{k} \mathbf{g}\|)^{-\frac{1}{4}} \boldsymbol{\sigma} \Xi(\mathbf{x} \mathbf{t}^{-1} \mathbf{k} \mathbf{g}) \mathbf{d} \mathbf{x} \, \text{d}\mathbf{t} \, \text{k} \leq \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}^{-1/2} \mathbf{z}$$

$$\leq \varepsilon^{-1/2} \sigma(\mathbf{f}) \Xi(\mathbf{g}) (1 + \log ||\mathbf{g}||)^{\mathsf{d}} \|\_{C^{\mathsf{a}, \mathsf{c}}} (1 + \log ||\mathbf{x}||)^{-\mathsf{d}} \Xi(\mathbf{x}) \mathsf{d} \mathbf{g} \|\_{\mathsf{c}}$$

Then 1/2(f) (1 + logg)dT(t) G(1 + logxtx) d(xtxkg)dxdtdk, but

$$
\Xi(\text{xt}\text{x}^{-1}\text{kg}) = \Xi(\text{xt}\text{x}^{-1}) \cdot \Xi(\text{kg}) = \Xi(\text{xt}\text{x}^{-1}) \cdot \Xi(\text{g}).
$$

then

$$\varepsilon^{-1/2}\sigma(\mathbf{f})\left(1+\log\|\mathbf{g}\|\right)^{\mathsf{d}}\mathrm{r}\_{\mathrm{c}}|\Lambda(\mathbf{t})|^{2}\Big\{\mathrm{c}\left(1+\log\|\mathsf{x}\mathbf{t}\mathbf{x}^{-1}\|\right)^{-\mathsf{d}}\Xi\left(\mathsf{x}\mathbf{t}\mathbf{x}^{-1}\mathbf{kg}\right)\mathrm{d}\mathsf{x}\mathrm{d}\mathbf{t}\mathrm{d}\mathbf{k}=\mathsf{0}\right\}$$

$$=\varepsilon^{-1/2}\sigma(\mathbf{f})\left(1+\log\|\mathbf{g}\|\right)^{\mathrm{d}}\mathrm{Tr}\_{\varepsilon}|\Delta(\mathbf{f})|^{2}\mathrm{d}\left(1+\log\|\mathsf{x}\mathbf{x}^{-1}\|\right)\Xi(\mathsf{x}\mathbf{x}^{-1})\,\Xi(\mathbf{g})\mathrm{f}\mathbf{x}\mathrm{d}\mathbf{x},$$

Considering G T = Ge, , (> 0), and given that

$$\mathbf{dx} \mathbf{d} \mathbf{t} \mathbf{d} \mathbf{k} = \mu(\mathbf{F}^{\mathrm{T}\_{\mathrm{R}(\mathbf{k} \oplus \mathbf{f})}}) \mathbf{d} \mathbf{k},$$

fC(G), and gG, then

$$\mathbb{E}^{-1/2}\mathsf{o}(\mathsf{f})\left(1+\log\|\mathsf{g}\|\|\right)^{\mathsf{d}}\mathsf{r}\_{\mathsf{e}}|\Delta(\mathsf{t})|^{2}\mathsf{c}\left(1+\log\|\mathsf{x}\mathsf{t}\mathsf{x}^{-1}\|\right)\Xi(\mathsf{x}\mathsf{t}\mathsf{x}^{-1})\,\Xi(\mathsf{g})\Big{[}\mathsf{x}\mathsf{d}\mathsf{x}\leq\mathsf{t}\mathsf{x}$$

$$\leq \varepsilon^{-1/2} \sigma(\mathbf{f}) \Xi(\mathbf{g}) (1 + \log \|\mathbf{g}\|)^{\mathrm{d}} \Big|\_{\mathrm{C}^{\mathbf{e}} \times \mathbb{E}} (1 + \log \|\mathrm{xt}^{-1}\|)^{-\mathrm{d}} \Xi(\mathbf{x}) \mathrm{dg} \,\mathsf{d}$$

Then

$$\int \mathsf{k}\mathfrak{\mu}(\mathsf{F}^{\mathsf{T}\_{\mathsf{R}(\mathsf{k}\mathfrak{g})\mathfrak{f}})d\mathtt{k} \leq (1+\log||\mathsf{g}||)^{\mathsf{d}}\mathfrak{\sigma}(\mathsf{f})\Xi(\mathsf{g}).\qed$$

Suppose that every subgroup of Cartan, , is integral. Let h, be a subalgebra of Cartan of g, and let H = exp(h). Let () () h h (gC, hC), hh, (here *X* ,is the conjugate complex of *X*gC, relative to g). Let , be a positive root system to (gC, hC), such that if , and , then . Let

$$
\Sigma = \{ \alpha \in \Phi^+ \: \Big|\ \; \overline{\alpha} \neq -\alpha \},
\tag{11.77}
$$

Let

$$\Delta\_{\rm H}(\mathbf{h}) = \mathbf{h}^{\rho} \prod\_{\alpha \in \Phi^{+} - \Sigma} (\mathbf{l} - \mathbf{h}^{-\alpha}) \left| \prod\_{\alpha \in \Sigma} (\mathbf{l} - \mathbf{h}^{-\alpha}) \right| \tag{11.78}$$

Clearly hhhH. Indeed, for definition

$$\Delta(\mathbf{h}) = \mathbf{h}^{\rho} \prod\_{\alpha:\alpha \blacktriangleright} (\mathbf{l} - \mathbf{h}^{-\alpha})\_{\gamma}$$

hH. Since () = then

$$\Delta\_{\mathbf{H}}(\mathbf{h}) = \mathbf{h}^{\rhd} \left| \prod\_{\alpha \equiv \Phi^{+}} (1 - \mathbf{h}^{-\alpha}) \right|\_{\mathbf{h}^{\alpha}}$$

which is equivalent to that

$$\left| \prod\_{\alpha \in \Phi^{+}} (\mathbf{l} - \mathbf{h}^{-\alpha}) \right| = \begin{cases} -\prod\_{\alpha \in \Phi^{+}} (\mathbf{l} - \mathbf{h}^{-\alpha})\_{\prime} & \text{if } \quad \alpha \in \Sigma\_{\prime} \\ +\prod\_{\alpha \in \Phi^{+}} (\mathbf{l} - \mathbf{h}^{-\alpha})\_{\prime} & \text{if } \quad \alpha \in \Phi^{+} - \Sigma\_{\prime} \end{cases}$$

Thus

$$\Delta\_{\mathrm{H}}(\mathrm{h}) = \mathrm{h}^{\rho} \langle \pm \prod\_{\alpha \in \Phi^{+}} (\mathrm{l} - \mathrm{h}^{-\alpha}) \rangle = \pm \mathrm{h}^{\rho} \prod\_{\alpha \in \Phi^{+}} (\mathrm{l} - \mathrm{h}^{-\rho}) = \pm \Delta(\mathrm{h})\_{\prime \mu}$$

Now well, if fC(G), then

$$\mathbf{F}\_{\rm f}^{\rm H}(\mathbf{h}) = \Delta\_{\rm H}(\mathbf{h}) \int \limits\_{\rm G/H} \mathbf{f}(\mathbf{g} \mathbf{h} \mathbf{g}^{-1}) \mathbf{dg} \mathbf{H}\_{\rm f}$$

The measure on G/H, is elected like in the theorem III. 1, and the domain is the set of all the hH, such that the integral converges absolutely.

We assume that H = HF = TFAF, with (PF, AF), a parabolic cuspidal canonical pair. Assume that, like in theorem III. 1, we have

$$\mathbf{F}\_{\mathbf{f}}^{\mathrm{H}}(\mathbf{h}) = \Delta\_{\mathrm{H}}(\mathbf{h}) \int\_{\mathbf{K} \times \mathbf{0}^{0} \, \mathrm{M} \times \mathrm{N}} f(\mathbf{k} \mathrm{m} \mathrm{n} \mathrm{h} \mathrm{n}^{-1} \mathrm{m}^{-1} \mathrm{k}^{-1}) \mathrm{dk} \mathrm{d} \mathrm{m} \mathrm{d} \mathrm{n} \mathrm{d} \mathrm{x}$$

3.

, and

32 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

*X*gC, relative to g). Let

hH. Since () =

which is equivalent to that

Now well, if fC(G), then

. Let

Clearly hhhH. Indeed, for definition

then

1 h

hH, such that the integral converges absolutely.

that, like in theorem III. 1, we have

( )

, then

Let

Thus

2.

Suppose that every subgroup of Cartan, , is integral. Let h, be a subalgebra of Cartan of g, and let H = exp(h). Let () () h h (gC, hC), hh, (here *X* ,is the conjugate complex of

<sup>H</sup> (h) h 1 h 1 h ( ) ( ),

( ( ), h) h 1 h 

<sup>H</sup> (h) h 1 h ( ), 

 

1 h

<sup>H</sup> (h) h 1 h h h h ( ( )) (1 ) ( ), 

H 1

G/H F (h h f(ghg )dgH ) () ,

The measure on G/H, is elected like in the theorem III. 1, and the domain is the set of all the

We assume that H = HF = TFAF, with (PF, AF), a parabolic cuspidal canonical pair. Assume

 

f H

1 h

( ), ,

*if*

( ), ,

*if*

 

 

, be a positive root system to (gC, hC), such that if

{ α (III. 77)

(III. 78)

We define to hH, h(n) = h1nhn1, nNF.

Likewise, is had the following lemma:

### **Lemma III. 5.**

1. h(NF) NF.

2. If det((Ad(h1) I)nF) 0, then h, is a diffeomorphism of NF, in NF. In consequence, if f, is integrable on NF, then

$$\left| \det(\text{Ad}(\text{h}^{-1}) - \text{I})|\_{\text{F}} \right| \left| \text{N} \text{rf}(\text{h}^{-1} \text{nh} \text{n}^{-1}) \text{dn} = \right|\_{\text{N}} \text{rf}(\text{n}) \text{dn}, \tag{\text{III.79}}$$

The demonstration of this result is essentially the same that the realized by corresponding to the lemma IV. 2, that in the next chapter will be seen.

If fC(G), let f, as (III. 32)-(III. 34). We note that if hH, then

$$|\det((\mathrm{Ad}(\mathbf{h}^{-1}) - \mathbf{I})\_{\mathrm{LF}})| = |\Pi\_{aa\Sigma}(\mathbf{1} - \mathbf{h}^{-\alpha})|\,\tag{\text{III.80}}$$

Then we can define the cusp forms:

$$\mathsf{Fr}^{\mathsf{H}}(\mathsf{h}) = \mathsf{C}\mathsf{rh}^{\mathsf{p}}\Delta\mathsf{u}(\mathsf{h})\mathsf{l}\mathsf{w}\_{\times}\mathsf{w}\mathsf{f}(\mathsf{mh}\mathsf{m}^{-1}\mathsf{n})\mathsf{d}\mathsf{m}\mathsf{d}\mathsf{n} = \mathsf{C}\mathsf{rh}^{-\mathsf{p}}\Delta\mathsf{u}(\mathsf{h})\mathsf{l}\mathsf{w}\_{\times}\mathsf{w}\mathsf{f}(\mathsf{mh}\mathsf{m}^{-1})\mathsf{d}\mathsf{m}\mathsf{d}\mathsf{n},\text{ (II.81)}$$

Where M, is the "", for (mC, hC). In the next chapter (III. 81) will appears again as other version adequated to Lie algebras.

## **Orbital Integrals on Reductive Lie Algebras**

## **IV. 1. Orbital integrals on reductive Lie algebras**

Let t'' = t T'', be where

$$\mathfrak{U}^{\prime\prime} = \langle \mathfrak{X} \in \mathfrak{U} \mid \det(\text{ad}(\mathfrak{X})|\_{\mathfrak{Y}}) \neq 0 \rangle,\tag{\text{IV. 1}}$$

We fix a system of positive roots ,in(g**C**, t**C**). Let

$$
\pi = \prod\_{\alpha \in \Phi^+} \alpha\_\prime \tag{1V.2}
$$

Let

$$\clubsuit\_{\mathsf{h}} = \langle \alpha \in \spadesuit (\mathsf{foc}, \mathsf{tc}) \mid (\mathsf{foc})^{\alpha}\_{\alpha} \square \mathsf{ptc} \rangle,\tag{\mathsf{IV}.\,3}$$

and let

$$
\pi\_{\rm av} = \prod\_{\alpha w \preccurlyeq\_{\prec} \preccurlyeq\_{\rm av}} \alpha\_{\star} \tag{1V. 4}
$$

If Ht, then det(ad(X)p) = n2. Let T, be the subgroup of Cartan of G, corresponding to t. Let

$$\mathfrak{Sp}(\mathfrak{g}) = \langle \mathfrak{f} \in \mathcal{C}^{\omega}(\mathcal{G}) \mid \mu\_{\mathfrak{e}, \mathfrak{f}}(\mathfrak{f}) \le 0 \rangle,\tag{1V.5}$$

where

$$\mu\_{\mathbb{A}, \mathsf{d}}(\mathbf{f}) = \sup \chi\_{\mathbb{A}, \mathsf{d}} \| |\mathsf{X}| |\mathsf{T}| \mathsf{d} \, |\mathsf{C}^{\|} / \partial \mathsf{x} \mathsf{d} \mathsf{f}(\mathsf{X}) | \mathsf{<} \infty \,\tag{\text{IV. 6}}$$

where the topology of the space S(g), is the constructed by the semi-norms s, t. Note that the semi-norms qs, t, are continuous on the space S(g).

If fS(g), and Ht'', then we consider

$$\Phi \iota^{\mathfrak{r}}(\mathcal{H}) = \mathfrak{n}(\mathcal{H}) \big| \text{cf}(\text{Ad}(\mathcal{g})\mathcal{H}) \text{dg} \tag{\text{IV. 7}}$$

then <sup>f</sup> <sup>T</sup>C∞(t'').

To demonstrate this, we elect H1, , Hr, a orthonormal base of t, and let t1, , tr, be the corresponding coordinates in the algebra t.

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Then is necessary proving that:

**Lemma IV. 1**Exist a constant u, such that if I, is a rth-mult-index then exist a continuous semi-norm I, on S(g), such that

$$|\hat{\boldsymbol{\phi}}^{\mathrm{|l|}}/\hat{\boldsymbol{\phi}}\mathbf{x}^{\mathrm{|l|}}\,\mathsf{@}\mathbf{\hat{r}}^{\mathrm{|l|}}(\mathbf{H})|\leq|\boldsymbol{\pi}\mathbf{u}(\mathbf{H})|^{-n}\boldsymbol{\mu}(\mathbf{f}).\tag{1V.8}$$

fS(g), Ht''.

*Proof*. If pS(g**C**), let p , be the image of isomorphism

$$\mathbf{S(ptc)} \to \mathbf{S(tc)},\tag{1V.9}$$

with rule of correspondence

$$\mathbf{p} \mid \begin{matrix} \rightarrow \mathbf{p}^{\sim} \end{matrix} \tag{1V.10}$$

let the space

$$\mathbf{I} = \langle \mathbf{p}^{\sim} \mid \mathbf{p} \in \mathbb{S}(\mathfrak{g}\mathbf{c})^{\mathbb{C}}\rangle,\tag{1V.11}$$

Then by the theorem of the isomorphism of Harish-Chandra exist p S(g**C**), such that p = p<sup>t</sup>**C**, whose linear isomorphism pjS(g**C**)G, is defined by

$$\mathbf{p}^{\gamma} \text{= } \text{Res}\,\mathbf{s}\_{(w \wedge h)}(\mathbf{p}),\tag{1V.12}$$

with j = 1, 2, , d. Then S(t**C**), is invariant under the module class I, that is to say, is a finitely generated module as I-module. Then pjS(g**C**), then

$$\mathbf{S(tc)} = \boldsymbol{\Sigma}\_{\boldsymbol{\Lambda}} \mathbf{I} \mathbf{p}\_{\boldsymbol{\Lambda}} \tag{\text{IV. 13}}$$

with j = 1, 2, , d. We consider pjS(t**C**), j = 1, 2, , d, such that (IV. 13). Be

$$\mathbf{t}' = \{\mathbf{H} \in \mathbf{t} \mid \pi(\mathbf{H}) \neq 0\}. \tag{\text{IV. 14}}$$

Be Ht' and be W = Ut'(H), the open neighborhood of H, in t'. Be U = Ad(G)W. If fS(g), then

$$\log(\lambda) = \left| \text{cf}(\text{Ad}(\text{g})\lambda)\text{dg} \right.\tag{1V.15}$$

Then gC∞(U).

We consider YAd(G)W = U, then we define

$$\log(\text{Ad}(\mathbf{x})\mathbf{Y}) = \left| \text{cf}[\text{Ad}(\mathbf{x})(\text{Ad}(\mathbf{x})\mathbf{Y})] \text{dx} = \left| \text{cf}(\text{Ad}(\mathbf{x})\mathbf{Y}) \text{dx} = \mathbf{g}(\mathbf{Y}) \right| \tag{1V. 16}$$

since Y = Ad(x)Y, xG. Then g(Ad(x)Y) = g(Y), YU. From theorem III. 1, we have that pS(g**C**)G,

$$\mathbf{p}\mathbf{g}|\mathbf{w} = \pi^{-1}\mathbf{p}'\pi\mathbf{g}|\mathbf{w},\tag{1V.17}$$

This implies that

36 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

/x<sup>f</sup>

, be the image of isomorphism

I = p

Then by the theorem of the isomorphism of Harish-Chandra exist p

with j = 1, 2, , d. We consider pjS(t**C**), j = 1, 2, , d, such that (IV. 13). Be

= p<sup>t</sup>**C**, whose linear isomorphism pjS(g**C**)G, is defined by

**Lemma IV. 1**Exist a constant u, such that if I, is a rth-mult-index then exist a continuous

T(H) ≤ n(H)

S(g**C**) S(t**C**), (IV. 9)

with j = 1, 2, , d. Then S(t**C**), is invariant under the module class I, that is to say, is a finitely

S(t**C**) = jIpj, (IV. 13)

t' = Ht(H) 0. (IV. 14)

Be Ht' and be W = Ut'(H), the open neighborhood of H, in t'. Be U = Ad(G)W. If fS(g),

uI(f). (IV. 8)

, (IV. 10)

S(g**C**), such that

pS(g**C**)G, (IV. 11)

j = ResS(g**C/**t**C**)(pj), (IV. 12)

g(X) = Gf(Ad(g)X)dg, (IV. 15)

g(Ad(x)Y) = GfAd(x)(Ad(x)Y))dx = Gf(Ad(x)Y)dx = g(Y) , (IV. 16)

Then is necessary proving that:

semi-norm I, on S(g), such that

fS(g), Ht''.

let the space

p

then

Then gC∞(U).

*Proof*. If pS(g**C**), let p

with rule of correspondence

p ∣p

p

We consider YAd(G)W = U, then we define

generated module as I-module. Then pjS(g**C**), then

$$\mathbf{p}^{\square}\mathbf{\hat{O}}^{\square}\mathbf{(H)}=\mathbf{\hat{O}}^{\square}\mathbf{(H)},\tag{1V.18}$$

fS(g), and Ht'. In effect, ):

$$\begin{aligned} \mathsf{Op}\_{\mathsf{H}^{\mathsf{I}}}(\mathsf{H}) &= \pi(\mathsf{H}) \Big\| \mathsf{c} \mathsf{p} \mathsf{f}(\mathsf{Ad}(\mathsf{g})\mathsf{H}) \mathsf{d}\mathsf{g} \\\\ &= \pi(\mathsf{H}) \Big\| \mathsf{c} \pi^{-1} \mathsf{p}^{\mathsf{T}} \mathsf{n} \mathsf{f}(\mathsf{Ad}(\mathsf{g})\mathsf{H}) \mathsf{d}\mathsf{g} \\\\ &= \pi(\mathsf{H}) \Big\| \mathsf{c} \mathsf{p}^{\mathsf{T}} \mathsf{f}(\mathsf{Ad}(\mathsf{g})\mathsf{H}) \mathsf{d}\mathsf{g} \\\\ &= \mathsf{p}^{\mathsf{T}} \langle \pi(\mathsf{H}) \Big\| \mathsf{c} \mathsf{f}(\mathsf{Ad}(\mathsf{g})\mathsf{H}) \mathsf{d}\mathsf{g} \rangle = \mathsf{p}^{\mathsf{T}} \mathsf{d}\mathsf{f}^{\mathsf{T}}(\mathsf{H}) \,\,\|\mathsf{c} \mathsf{d}\mathsf{g} \| \end{aligned}$$

Ht', and fS(G). ):

$$\mathbf{p}^{\prime}\Phi\mathbf{r}^{\prime}(\mathbf{H}) = \mathbf{p}^{\prime}(\pi(\mathbf{H})\int\_{\Gamma} \text{cf}(\mathbf{Ad}(\mathbf{g})\mathbf{H})d\mathbf{g})$$

$$= \pi(\mathbf{H})\int\_{\Gamma} \text{cf}^{-1}\mathbf{p}^{\prime}\pi\mathbf{f}(\mathbf{Ad}(\mathbf{g})\mathbf{H})d\mathbf{g}$$

$$= \pi(\mathbf{H})\int\_{\Gamma} \text{cf}(\mathbf{Ad}(\mathbf{g})\mathbf{H})d\mathbf{g}$$

$$= \pi(\mathbf{H})\int\_{\Gamma} \text{cf}(\mathbf{Ad}(\mathbf{g})\mathbf{H})d\mathbf{g} = \Phi\_{\mathbb{R}^{\mathbb{L}}}(\mathbf{H})\_{\prime}$$

 Ht', and fS(G). If we consider as partial derivative of multi-index order, the product of the canonical components of the field Ht', we have

$$\mathbf{H} \mathbf{r}^{\downarrow} \,, \,, \,, \mathbf{H} \mathbf{r}^{\downarrow} = \hat{\mathcal{O}}^{\parallel \mathbf{l}} / \hat{\mathcal{O}} \mathbf{t}^{\downarrow} = \sum \mathbf{J} \mathbf{u}^{\uparrow} \mathbf{p} \mathbf{p} \,, \tag{\text{IV. 19}}$$

ujS(g**C**)G. Then pjS(t**C**),

$$|\mathbf{p}\otimes\mathbf{h}^{\mathrm{T}}(\mathbf{H})| \leq \mathrm{c}|\pi\_{\mathrm{T}}(\mathbf{H})|^{-\mathrm{r}}\mu(\mathbf{f}),\tag{1V.20}$$

with j, a continuous semi-norm on S(g), and rj, depends only on the degree of pj. Realizing an adequate election of j, we can to replace rj, for r, the maximum of the rj. Thus

$$|\mathbf{p}\,\mathsf{@}\mathbb{A}^{\mathsf{T}}(\mathbf{H})| \le \mathsf{c}|\mathsf{π}\,\mathsf{(H)}|^{-\mathsf{r}}\Sigma\mu(\mathbf{f}),\tag{\mathsf{IV. 21}}$$

Ht'. Thus

$$|\hat{c}^{\|\|} / \hat{c} \mathfrak{k}^{\mathsf{I}} \,\Phi \mathfrak{r}^{\mathsf{I}}(\mathsf{H})| \le \mathsf{c} |\pi\_{\mathsf{H}}(\mathsf{H})|^{-} \Sigma \mu(\mathsf{u} \mathsf{f}).\tag{\mathsf{IV.22}}$$

 Ht'. Since both sides of the previous inequality are continuous on t'', then is followed (IV. 8).

If U, is an open of t, then we define to the space

$$\mathfrak{S}(\mathsf{U}) = \langle \mathsf{f} \in \mathsf{C}^{\times}(\mathsf{U}) \,|\,\mathsf{u}\mathsf{u}\,\mathsf{\mathsf{f}}\,\,\mathsf{s}\,\mathsf{f}\,\mathsf{f} \rangle = \mathsf{s}\mathsf{u}\mathsf{p}\times\_{\mathsf{c}}\mathsf{U} \|\mathsf{X}\|^{\mathsf{c}}\Sigma\_{\mathsf{c}|\mathsf{c}\mathsf{s}}|\widehat{\mathsf{c}}^{\mathsf{U}}/\mathsf{C}\mathsf{f}\,\mathsf{f}\,\mathsf{f}\,\mathsf{(X})\|\mathsf{<}\mathsf{c}\mathsf{s}\,\mathsf{s}\rangle,\tag{\mathsf{IV.7.21}}$$

Said space is a Frèchet space.

**Theorem IV. 1**(Theorem of Harish-Chandra). If fS(g), then <sup>f</sup> <sup>T</sup>S(t''). Furthermore, the map f ∣<sup>f</sup> T, from S(g), to S(t'') is continuous.

*Proof.* Let C, the convex component of t'', that is to say, the space

$$\mathbf{C} = \langle \mathbf{H} \boxplus \mathbf{t}^{\prime\prime} \vert \pi\_{\mathbf{n}}(\mathbf{H}) = \det(\text{ad}(\mathbf{H}) \vert\_{\mathbb{P}}) \rangle,\tag{\text{IV. 22}}$$

If Hit, then (H)R, , = (g**C**,t**C**). Thus, if *n*, then i> 0, or i< 0, on the space C. Then we can to define the rule of correspondence

$$\begin{aligned} \text{i.e., if } a \ge 0, \\ \mid a \mid\_{\mathbb{C}} &= \\ -\text{i.e., if } a \ge 0, \end{aligned} \tag{\text{IV. 23}}$$

Thus

$$|\mathfrak{m}(\mathbf{H})| = |\prod\_{a:a\otimes^+ \wedge \otimes^\times} a| \subset \prod\_{a:a\otimes^+ \wedge \otimes^\times} |a| \operatorname{c(H)}\tag{1V.24}$$

Hit. Let xCl(C), and we fix x0C. Then

$$|\alpha|\mathbf{c}(\mathbf{x} + \mathbf{t}\mathbf{x}) = |\alpha|\mathbf{c}(\mathbf{x}) + \mathbf{t}|\alpha|\mathbf{c}(\mathbf{x}) \ge \mathbf{t}|\alpha|\mathbf{c}(\mathbf{x}),\tag{1V.25}$$

if t ≥ 0. Let fS(g). The previous result implies that if F = f, and if q = n, then

$$|\mathsf{DF}(\mathsf{x} + \mathsf{t}\mathsf{x}\mathsf{o})| \le \mathsf{t}^{-\mu} \mathsf{r}\mathsf{h}^{\mathsf{b}}(\mathsf{f})\tag{\mathsf{IV}.26}$$

t > 0, and pS(t). p is a continuous semi-normon S(g). In effect,

/t f(F) ≤ n(x) uI(f),

But

$$|\hat{\boldsymbol{\beta}}^{\|} / \partial \mathbf{t}^{\|} \, \Phi \mathbf{t}(\mathbf{F})| \le |\mathbf{d} / \mathbf{d} \mathbf{t} \, \mathbf{p} \mathbf{F} \text{(\$\mathbf{x} + \mathbf{t} \mathbf{x} \boldsymbol{\beta}\$)} \le |\pi\_{\mathbf{n}} \text{(\$\mathbf{x}]} \, \mathbf{\hat{\boldsymbol{\beta}}}^{\|} \boldsymbol{\mu}\_{\mathbb{P}} \text{(\$\mathbf{f}\$)},\tag{\text{IV. 27}}$$

xCl(C). Then

Chapter IV – Orbital Integrals on Reductive Lie Algebras 39

$$|\mathbf{pF(x+txo)}| \le |\pi \mathbf{u(x)}|^{-n} \mu(\mathbf{f}) \le \mathbf{f}^{-n\eta} \mu^{\eta}(\mathbf{f}) \tag{1V. 28}$$

 xCl(C), or pF(x + tx0) ≤ tuqp(f), q = n.

Now, to the kth-derivative we have that

38 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

/t <sup>f</sup>

If U, is an open of t, then we define to the space

Said space is a Frèchet space.

S(U) = fC∞(U)U, r, s(f) = supX<sup>U</sup>X<sup>r</sup>

**Theorem IV. 1**(Theorem of Harish-Chandra). If fS(g), then <sup>f</sup>

*Proof.* Let C, the convex component of t'', that is to say, the space

T, from S(g), to S(t'') is continuous.

C. Then we can to define the rule of correspondence

Hit. Let xCl(C), and we fix x0C. Then

(IV. 8).

map f ∣<sup>f</sup>

Thus

But

xCl(C). Then

T(H) ≤ cn(H)

Ht'. Since both sides of the previous inequality are continuous on t'', then is followed

If Hit, then (H)R, , = (g**C**,t**C**). Thus, if *n*, then i> 0, or i< 0, on the space

i , if 0,

 

 

f(F) ≤ n(x)

f(F) ≤ d/dt pF(x + tx0) ≤ n(x)

i , if 0,


if t ≥ 0. Let fS(g). The previous result implies that if F = f, and if q =

t > 0, and pS(t). p is a continuous semi-normon S(g). In effect,

/t  /t  r

 I <sup>≤</sup>s/t 

C = Ht''n(H) = det(ad(H)p), (IV. 22)

n(H) = n**C** ≤ n**C**(H), (IV. 24)

**C**(x + tx0) = **C**(x) + t**C**(x0) ≥ t**C**(x0), (IV. 25)

uI(f),

pF(x + tx0) ≤ tuqp(f), (IV. 26)

**<sup>C</sup>** (IV. 23)

j(ujf). (IV. 22)

f(X)< ∞, (IV. 21)

<sup>T</sup>S(t''). Furthermore, the

n, then

up(f), (IV. 27)

$$\mathbf{d}^{k} / \mathbf{d}\mathbf{t}^{k} \,\mathrm{pF}(\mathbf{x} + \mathbf{t}\mathbf{x}) = (\mathbf{x}u)^{k} \,\mathrm{pF}(\mathbf{x} + \mathbf{t}\mathbf{x}),\tag{\text{IV. }\mathcal{D}9}$$

xCl(C), where is necessary have always all the time that:

$$\mathbf{S(t) = (p)} \in \mathbf{S(ptc)} \mid \mathbf{p = \Sigma \downarrow \mathbf{p}}\text{)}\,,\tag{1V. 30}$$

where I = p pS(g**C**)G. S(t), is identified with the space of differential operators of constants coefficients on the algebra t. This means that if u(t) = pF(x + tx0), then

$$\mathbf{u}^{(k)}(\mathbf{t}) = (\mathbf{x}\mathbf{o})^{k} \mathbf{p} \mathbf{f}(\mathbf{x} + \mathbf{t}\mathbf{x}\mathbf{o}) \tag{1V.31}$$

of where the value of the kth-derivative of u(t), comes bounded as

u(k)(t) ≤ tuqxp(f)n(x0) u, (IV. 32)

**Schollium:** Let uC∞(0, 1), be and suppose that

$$|\mathfrak{u}^{(k)}(\mathfrak{t})| \le \mathfrak{t}^{-m} \mathfrak{a}\mathfrak{a}\_{\prime} \tag{1V.33}$$

to 0 < t ≤ 1, and rank k = 0, 1, 2, . Then

$$|\mathbf{u}(\mathbf{t})| \le \mathbf{C} (\mathbf{a} \mathbf{o} + \mathbf{a} \dots + \mathbf{a} \mathbf{a} + \mathbf{t}) \tag{\text{IV. 34}}$$

to 0 < t ≤ 1. Here C, depends only on m. Indeed, we assume that m ≥ 1. If m > 1, then we write to u(k)(t), as:

$$\mathbf{u}^{(k)}(\mathbf{t}) = \left[\mathbf{v}^{1}\mathbf{u}^{(k+1)}(\mathbf{s})\mathbf{ds} - \left[\mathbf{i}^{1}\mathbf{u}^{(k+1)}(\mathbf{s})\mathbf{ds} = \mathbf{u}^{(k)}(\mathbf{1}) - \left[\mathbf{i}^{1}\mathbf{u}^{(k+1)}(\mathbf{s})\mathbf{ds}\right] \right.\tag{IV.35}$$

Having that

$$|\mathfrak{u}^{(k)}(\mathfrak{t})| \le -\left[\mathfrak{t}^{1}\mathfrak{t}^{-\mathsf{m}}\mathfrak{a}\mathfrak{a}\ast\mathfrak{t}\mathfrak{d}\mathfrak{t} + \int \mathfrak{t}^{1}\mathfrak{t}^{-\mathsf{m}}\mathfrak{a}\,\,\mathsf{x}\,\ast\mathfrak{d}\mathfrak{t}\mathfrak{t}\mathfrak{d}\mathfrak{t}\right] \tag{\text{IV. 36}}$$

Then

$$|\mathfrak{u}^{(k)}(\mathfrak{t})| \le \mathfrak{a}\mathfrak{s} + \mathfrak{t}^{\top}\mathfrak{r}^{m+1}/(\mathfrak{m}-1) + \mathfrak{a}\mathfrak{s}^{\bot} + \mathfrak{a}\mathfrak{s} \tag{1V.37}$$

0 < t ≤ 1, where we have that

$$|\mathbf{f}^\*| |\mathbf{u}^{(k)}(\mathbf{t})| \le \left| \mathcal{T}^{\mathbf{m}+1} \langle \mathbf{a}\_{k+1} + \mathbf{a} \rangle \right| \tag{\text{IV. 38}}$$

0 < t ≤ 1. Then using (\*), we have

$$|\mathfrak{l}^{\ast\ast}\rangle|\mathfrak{u}^{(k)}(\mathbf{t})| \le \mathcal{D}^{\mathbf{m}^{-1}}\mathfrak{l}^{-1}\langle\mathfrak{a}\mathfrak{k}+\dots+\mathfrak{a}\mathfrak{a}+\mathfrak{n}\rangle,\tag{\text{IV.39}}$$

0 < t ≤ 1. Applying (\*\*), to the case k = 1, we find that

$$|\mathbf{u}^{(1)}(\mathbf{t})| \le 2^m \log(1/\mathbf{t}) (\mathbf{a} + \dots + \mathbf{a}\_{m+1}) \tag{IV. 40}$$

$$= \mathbf{u}^{(1)}(\mathbf{1}) - \begin{bmatrix} \mathbf{i}^1 2^m \ \mathbf{i}^{-1} \end{bmatrix} \mathbf{\dot{t}}^{-1} (\mathbf{a} + \dots + \mathbf{a}\_{k+m}) \mathbf{d} \mathbf{t}, \quad \text{ (k=1)}$$

to 0 < t ≤ 1.

If we integrate both members of the before inequality, we obtain the estimation that affirms the lemma to u(0) = u. In effect, only we consider the integral

$$\mathbf{u}^{(0)} = \left[v^{\dagger}\mathbf{u}^{(1)}(\mathbf{t})\mathbf{d}\mathbf{t} = \right] \mathbf{t}^{\prime} \mathbf{2}^{m} \log(\mathbf{1}/\mathbf{t}) (\mathbf{a}\mathbf{1} + \dots + \mathbf{a}\mathbf{m} + \mathbf{1}) \mathbf{d}\mathbf{t} \tag{\text{IV. 41}}$$

and we arrive to that

u(0)(t)= u(t)≤ C(a0 + + am+1).

Now, the conclusion of the schollium with the identity

$$\text{(\text{\textbullet}\text{)}\text{ p}^{\text{\textbullet}}\text{\textbullet}^{\text{\textbullet}}\text{H}\text{)}=\text{\textbullet}^{\text{\textbullet}^{\text{\textbullet}}\text{\textbullet}^{\text{\textbullet}}\text{H}}$$

fS(g) and Ht', will implies that if XC, then

$$|\mathbf{p}\Phi\mathbf{\varGamma}^{\mathsf{T}}(\mathbf{X})| \le \mathbb{E} |\pi\_{\mathbf{n}}(\mathbf{x}\mathbf{o})|^{-\mathsf{u}} \Sigma^{\mathsf{u}\_{\neq \mathbf{t}\_{k} = 0}} \varPi^{\mathsf{u}\_{\neq \mathbf{t}\_{k}}} \mathsf{u}\_{\mathbf{n}}(\mathbf{y}) \,\tag{\text{IV. 42}}$$

with E, a independent constant of f. Seeing that if p(X) = B(X, X), then p(X) = X2, Xt, which implies that

$$\mathbf{M}^{\mathrm{T}} \mkern-1.140mu \mathbf{(X} \mathbf{(X}) = ||X||^{2\mathbf{k}} \mkern-1.141mu \mathbf{(X} \mathbf{(X}) \text{)}\tag{\text{IV. 43}}$$

Then due to that the integral Gf(Ad(g)Y)dg, converges absolutely Yt'', and fC(g) (that defines a smooth function, g(Y) gt''), we have that the image <sup>f</sup> <sup>T</sup>S(t''). Then the map

$$(g) \rightarrow \oplus (t^{\prime\prime}),$$

is continuous.

Now we study the orbital integrals to other subalgebras of Cartan.

Let (P0, A0), be a minimal p-pair to G. Let h, be a subalgebra of Cartan of g. Then by the proposition that says that if H, is a Cartan subgroup of G, then there exists a standard cuspidal p-pair [Nolan Wallach], which implies that exist a canonical cuspidal p-pair (PF, AF), and xG0, such that

$$\mathbf{h} = \mathbf{A} \mathbf{d}(\mathbf{x}) \mathbf{l} \mathbf{f} \mathbf{r} \tag{\text{IV. 44}}$$

Let H and HF, be the corresponding subgroups of Cartan of G corresponding to the algebras h, and hF.

For definition of Subalgebra of Cartan

40 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

u(1)(t) ≤ 2mlog(1/t)(a1+ + am+1), (IV. 40)

2m log(1/t)(a1 + + am+1)dt, (IV. 41)

uuq+1k= 0 x0(k)p(f), (IV. 42)

T(X), (IV. 43)

h =Ad(x)hF, (IV. 44)

<sup>T</sup>S(t''). Then the map

12m 1 t1(ak + + ak+m)dt, (k = 1),

If we integrate both members of the before inequality, we obtain the estimation that affirms

u(0)(t)= u(t)≤ C(a0 + + am+1).

with E, a independent constant of f. Seeing that if p(X) = B(X, X), then p(X) = X2, Xt,

Then due to that the integral Gf(Ad(g)Y)dg, converges absolutely Yt'', and fC(g) (that

S(g)S(t''),

Let (P0, A0), be a minimal p-pair to G. Let h, be a subalgebra of Cartan of g. Then by the proposition that says that if H, is a Cartan subgroup of G, then there exists a standard cuspidal p-pair [Nolan Wallach], which implies that exist a canonical cuspidal p-pair (PF,

Let H and HF, be the corresponding subgroups of Cartan of G corresponding to the algebras

Tpkf(X) = X2k<sup>f</sup>

T(H) = pfT(H),

0 < t ≤ 1. Applying (\*\*), to the case k = 1, we find that

= u(1)(1)<sup>t</sup>

the lemma to u(0) = u. In effect, only we consider the integral

Now, the conclusion of the schollium with the identity

p<sup>f</sup>

defines a smooth function, g(Y) gt''), we have that the image <sup>f</sup>

Now we study the orbital integrals to other subalgebras of Cartan.

fS(g) and Ht', will implies that if XC, then

u(1)(t)dt = 0t

(\*\*\*) p

f

T(X) ≤ E n(x0)

to 0 < t ≤ 1.

u(0) = 0t

and we arrive to that

which implies that

is continuous.

h, and hF.

AF), and xG0, such that

$$\mathfrak{h}^{\text{new }}\mathfrak{h}\mathfrak{r}\oplus\mathfrak{h}\mathfrak{r}\_{\mathfrak{t}}\tag{1V.45}$$

where tF, is maximum Abelian in m T.

Let TF, be the subgroup of Cartan of 0MF, corresponding to tF. Then is easy to see that

$$\mathbf{H} \mathbf{\bar{r}} = \mathbf{T} \mathbf{\bar{r}} \mathbf{A} \mathbf{\bar{r}} = \mathbf{x} \mathbf{H} \mathbf{x}^{-1} \tag{\text{IV. 46}}$$

xG0. Indeed, we consider the space

$$^0\text{M} \equiv \langle \text{m} \in \text{M} \models \middle| \text{Ad} \langle \text{m} \rangle = \text{I}, \forall \text{ Ad} \in \text{End} \langle \text{M} \models \rangle,\tag{\text{IV. 47}}$$

and let TF 0MF, be the corresponding subgroup of Cartan of 0MF, defined explicitly

$$\mathbf{T}^{\mathrm{pr}} \text{ (}\mathbf{m} \in {}^{0}\text{M} \mathbf{r} \text{ }\big|\text{ Ad(m)}\big|\_{\varnothing} = \mathbf{I}\text{)},\tag{\text{IV. 48}}$$

and such that TF, = exp(tF), with tF= hF T. Then m0MF, and xG0,

$$\operatorname{Ad}(\mathfrak{m})|\_{\mathbb{V}} \ge \operatorname{Ad}(\mathfrak{h})\mathfrak{x} = \mathfrak{h}\mathfrak{x} = \mathfrak{x}\mathfrak{h} = \mathfrak{x}\mathfrak{y}$$

since that furthermore 0MF, is Abelian. Then xh = x, xG0, which is similar to

$$\text{xhx}^{-1} = 1\_{0i\_{\ell}}$$

since that furthermore tF= hF T, and HF, is the subgroup of Cartan of H. Then hH and xG0,

xHxF,

Then by the corresponding exp(tH) = TF, and exp(tFaF) = TFAF = HF = exp(hF), tF= hF T, we have finally

$$\mathbf{H} \mathbf{F} = \mathbf{T}\_{\mathbf{F}} \mathbf{A} \mathbf{F} = \mathbf{x} \mathbf{H} \mathbf{x}^{-1} \mathbf{y}$$

xG0.

On HF, we have the invariant measure dtFdaF, where dtF, is the normalized invariant measure on TF, and daF, is the corresponding Lebesgue measure to an ortonormal base of aF. Easy is to see that this measure is independent of the elections done in their definition.

We fix an invariant measure of G, and we have the quotient measure dgH on G/H. Let

$$\boldsymbol{\Phi}^{+} = (\boldsymbol{\alpha} \in \boldsymbol{\Phi}(\mathfrak{g}c, \mathfrak{h}c) \mid \mathrm{Re}\boldsymbol{\alpha} > 0),\tag{\text{IV. 49}}$$

with = . Be <sup>R</sup> = Re( ). Let

$$\mathfrak{h}' = \langle \mathbf{h} \in \mathfrak{h} \mid \mathfrak{a}(\mathbf{h}) \neq 0, \,\forall \alpha \in \mathfrak{\Phi} \rangle,\tag{1V.50}$$

If hh', then we define the correspondence

$$\mathbf{a(h) = sgn(\prod\_{a:a \oplus r \nmid r} a(h))},\tag{1V.51}$$

If fS(g), we can to define the operator on h',

$$\Phi\mathbf{v}\_{\rm li}(\mathbf{h}) = \mathbf{z}(\mathbf{h})\,\,\mathbf{z}(\mathbf{h})\,\mathrm{[c\,vdf(\mathbf{A}d(\mathbf{g})\mathbf{h})\,\mathrm{d}\,\mathrm{(g\,H)}]\,\tag{1V.52}$$

with the domain of <sup>f</sup> H, equal to the set of all the hh', to which the integral converge absolutely.

Note that <sup>f</sup> H, depends on the election of but only in sign. If in (g**C**, (hF)), we elect like space of positive roots to the set

$$\langle \alpha \circ \text{Ad}(\mathbf{x})^{-1} \mid \alpha \in \Phi^{+} \rangle,\tag{1V.53}$$

then

 <sup>f</sup> H(h) = <sup>f</sup> HF(Ad(x)h), (IV. 54)

 xG. Indeed, o Ad(x)= , ,if only if, = Ad(x), , since Ad(x)h = h, xG. Then

$$\Phi^{\operatorname{I}}\!\_{\operatorname{I}^{\operatorname{I}}}(\operatorname{I}) = \Phi^{\operatorname{I}\operatorname{Ad}(\operatorname{I})^{\operatorname{I}}}(\operatorname{Ad}(\operatorname{x})\operatorname{h}) = \Phi^{\operatorname{I}\operatorname{h}}(\operatorname{Ad}(\operatorname{x})\operatorname{h}),$$

 xG, since Ad(x)hHF, xG. We can consider the particular case when H = HF, to these integral formulas.

We consider on G/AF, the corresponding quotient measure to our election of invariant measures on AF. Then is clear that

$$\Phi\eta\_{l}(\mathbf{h}) = \mathfrak{z}(\mathbf{h})\,\,\mathfrak{z}(\mathbf{h})\Big|\_{\mathbb{C}\vee\mathbb{H}} (\operatorname{Ad}(\mathbf{g})\mathbf{h})\mathrm{d}(\mathbf{g}A\mathbf{v}),\tag{\text{IV. 55}}$$

Now well, the lemma A. 1, implies that the invariant measures on K, 0MF and NF, can be normalized such that

 <sup>f</sup> H(h) = (h) (h)<sup>K</sup>0M Nf(Ad(knm)h)dkdmdn, (IV. 56)

Let hh. If nNF, then we define the map

$$\text{T} \& \text{N} \vdash \text{Ad}(\text{N} \& ) \varnothing \mid \varnothing \rangle \tag{\text{IV. 57}}$$

with rule of correspondence

$$\mathbf{n} \mid \rightarrow \text{Ad}(\mathbf{n})\mathfrak{h} - \mathfrak{h},\tag{1V.58}$$

If nNF, then n = expX, XnF. If we develop the exponential series to

$$\operatorname{Ad}(\mathbf{n}) = \mathbf{e}^{\operatorname{ad}\chi} \tag{\text{IV. 59}}$$

Then Th(n)nF.

Indeed, considering that

$$\mathfrak{m} = \langle \mathsf{X} \in \mathfrak{n} \mid \langle \mathsf{ad}\mathsf{X} - \mathsf{I} \rangle \mathsf{v} = 0, \forall \ \mathsf{v} \in \mathsf{F} \text{ and } \mathsf{k} \in \mathbb{Z}^{+} \rangle,\tag{\text{IV. 60}}$$

and

42 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

If hh', then we define the correspondence

If fS(g), we can to define the operator on h',

H, depends on the election of

f

H(h) = <sup>f</sup>

<sup>f</sup>

space of positive roots to the set

<sup>f</sup>

xG. Indeed, o Ad(x)= ,

measures on AF. Then is clear that

<sup>f</sup>

with the domain of <sup>f</sup>

absolutely.

Note that <sup>f</sup>

then

xG. Then

integral formulas.

normalized such that

<sup>f</sup>

with rule of correspondence

Let hh. If nNF, then we define the map

h' = hh(h) ≠ 0, , (IV. 50)

(h) = sgn(R(h)), (IV. 51)

o Ad(x)

H(h) = <sup>f</sup>

Ad(H) h

H(h) = (h) (h)G/Hf(Ad(g)h)d(gH), (IV. 52)

but only in sign. If in (g**C**, (hF)), we elect like

HF(Ad(x)h), (IV. 54)

, (IV. 53)

, since Ad(x)h = h,

H, equal to the set of all the hh', to which the integral converge

,if only if, = Ad(x),

HF(Ad(x)h),

H(h) = (h) (h)G/AFf(Ad(g)h)d(gAF), (IV. 55)

H(h) = (h) (h)<sup>K</sup>0M Nf(Ad(knm)h)dkdmdn, (IV. 56)

(Ad(x)h) = <sup>f</sup>

xG, since Ad(x)hHF, xG. We can consider the particular case when H = HF, to these

We consider on G/AF, the corresponding quotient measure to our election of invariant

Now well, the lemma A. 1, implies that the invariant measures on K, 0MF and NF, can be

Th: NF Ad(NF)h\h, (IV. 57)

n ∣Ad(n)h h, (IV. 58)

$$\text{Ad}\,(\text{n}) = \exp\big(\text{ad}\,\text{X}\big) = \sum\_{k=0}^{\infty} \text{(1/k!)} \left(\text{ad}\,\text{X}\right)^{k},\tag{\text{IV. 61}}$$

Then

$$\mathrm{T\_h(n) = Ad(n)h\text{-}h\text{-}h = \sum\_{\mathbf{k=0}}^{\omega\_0} (1/\mathbf{k}!)(\text{ad}\,\mathrm{X}\cdot\mathrm{I})^\mathbf{k}\mathrm{h} = \sum\_{\mathbf{k=0}}^{\omega\_0} (1/\mathbf{k}!)(\text{ad}\,\mathrm{Xh\text{-}I\text{h}})^\mathbf{k} = \sum\_{\mathbf{k=0}}^{\omega\_0} \mathrm{T\_h}^\mathbf{k} \left(\text{m}\right)\text{-N}$$

 nNF. But (adXh Ih)k = (adX I)kh = 0, then Thk(n) = 0, kZ+. Thus Th(n)nF. An obvious evaluation give

$$(\mathbf{d}\mathbf{T}\mathbf{h})\_{\mathbf{n}}(\mathbf{X}) = \mathbf{A}\mathbf{d}(\mathbf{n})[\mathbf{X}, \mathbf{h}]\_{\mathbf{n}} \tag{\text{IV. 62}}$$

In effect, for one side

$$\text{adT}\_{\mathbb{H}}(\mathbf{n})\mathbf{X} = \text{d}(\text{Ad}(\mathbf{n})\mathbf{h})\mathbf{X} = (\text{d}\text{Ad}(\mathbf{n})\mathbf{X} - \mathbf{X})\mathbf{h} = [(\text{ad}\mathbf{X} - \mathbf{X})\mathbf{h}]\_{\mathbb{H}}\mathbf{h}$$

For other side

$$\text{Ad}(\mathbf{n})[\mathbf{X}, \mathbf{h}] = [\text{Ad}(\mathbf{n})\mathbf{X}, \text{Ad}(\mathbf{n})\mathbf{h}] = \text{Ad}(\mathbf{n})(\mathbf{X}\mathbf{h} - \mathbf{h}\mathbf{X}) = \text{Ad}(\mathbf{n})\mathbf{X}\mathbf{h} - \text{Ad}(\mathbf{n})\mathbf{h}\mathbf{X}, \mathbf{h}$$

But Ad(n)X = X, nNF, and XnF. Then

$$\text{Ad}(\mathbf{n})\mathbf{X}\mathbf{h} - \text{Ad}(\mathbf{n})\mathbf{h}\mathbf{X} = \text{Ad}(\mathbf{n})\mathbf{X}\mathbf{h} - \mathbf{X}\mathbf{h} = (\text{Ad}(\mathbf{n})\mathbf{X} - \mathbf{X})\mathbf{h} = \text{d}\mathbf{T}\_{\mathbb{H}}(\mathbf{n})(\mathbf{X})\_{\mathbb{H}}$$

Thus (IV. 62), is verified nNF, and XnF. This implies that

1. If *det*(ad(h)n) ≠ 0 then Th, is regular everywhere. *Note*: (adX)(h) = dTh(X) 0, then Th, is regular everywhere.

**Lemma IV. 2.** If *det*(ad(h)n) ≠ 0 then Th, is a diffeomorphism of NF, in nF, such that

$$\det(\text{ad}(\mathbf{h})|\_{\mathbb{P}}) \Big| \bowtie \text{f}(\text{Ad}(\mathbf{n})\mathbf{h} - \mathbf{h}) \text{dn} = \big|\_{\text{u}} \text{f}(\text{X})\text{dX},\tag{\text{IV.63}}$$

to f, a rapidly decreasing function on nF.

If we demonstrate that Th, is a diffeomorphism of NF, in nF, then the integration formula can follows of the formula

$$\text{V}\Phi\text{(h)} = \text{z(h)}\,\text{\pi(h)}\,\text{c}\wedge\text{v(Ad(g)h)d(gNr)}\,\text{\AA}\tag{1V.64}$$

to the differential of Th. Let h0aF, be such that (h0) > 0, (PF, AF). Let at = exp(th0), be then Th(atnat) = Ad(at)Th(n). Since Th, is in particular, regular in 1NF, exist an open neighborhood U0, of 0 = 1nF, in nF, such that Th, of U1, in U0, is a diffeomorphism.

In effect,

Th(atnat) = Ad(atnat)h = Ad(at)Ad(n)Ad(at) = Ad(at)Ad(at)Ad(n)h = Ad(at)Th(n),

Then

$$\mathbf{U} \smile\_{\mathbb{R}^{3}} \mathbf{Ad}(\mathbf{a}) \mathbf{U} \mathbf{0} = \mathfrak{m}\_{\mathsf{F}} \tag{\text{IV. 65}}$$

of which the equi-variance Th(atnat) = Ad(at)Th(n), implies that Th, is suprajective, since

$$\bigcup\_{\mathbf{t}\geq 0} \mathbf{Ad(a)} \mathbf{U}\_{0} = \mathbf{T\_{h}(\mathbf{a}\mathbf{n}\mathbf{a}\mathbf{\_{t}})} = \mathfrak{m}\_{\mathbf{F}\_{0}}$$

Now only is necessary to demonstrate the injectivity of Th. To it, suppose that Th(n1) = Th(n2). Let t, be such that atna<sup>t</sup>U1, j = 1, 2. Then

$$\mathrm{Tr}(\mathsf{a}\mathsf{n}\mathsf{n}\mathsf{a}\_{\mathrm{-}}) = \mathrm{Ad}(\mathsf{a}\mathsf{l})\mathrm{T}\_{\mathsf{h}}(\mathsf{n}\mathsf{l}) = \mathrm{Ad}(\mathsf{a})\mathrm{T}\_{\mathsf{h}}(\mathsf{n}\mathsf{z}) = \mathrm{T}\_{\mathsf{h}}(\mathsf{a}\mathsf{n}\mathsf{z}\mathsf{a}\_{\mathrm{-}}),$$

Thus atn1at= atn2at. Thus n1 = n2. Then Th, is injective. Then Th, is bijective. Thus Th, is a diffeomorphism of NF, in nF.

Now we elect , such that if and if a 0, then a(PF, AF). Let , be the set of all the whose restriction to aF, is not vanishing, to know;

$$
\Sigma = \{ \alpha \in \Phi^+ \mid \alpha|\_{\mathbb{F}} \neq 0 \},
\tag{1V.66}
$$

If hh, then det(adhnF) = (h) = (h)(h).

This combined with the lemma IV. 2., implies that

2. <sup>f</sup> H(h) = CF(h) (h)<sup>K</sup> M N f(Ad(k)Ad(m)(h + X)) dk dm dn, If f is a soft function on g, then

Chapter IV – Orbital Integrals on Reductive Lie Algebras 45

$$\mathbf{f}^{\ast}(\mathbf{X}) = \left| \ltimes \mathbf{f}(\operatorname{Ad}(\mathbf{k})\mathbf{X}) \operatorname{d} \mathbf{k} \right| \text{ d} \mathbf{k}, \tag{\text{IV. 67}}$$

Since Ad(0M), preserve dX, on nF, we have with the notation upward that:

3. <sup>f</sup> H(h) = CF(h)0M nf (Ad(m)(h + X)) dm dX, If fS(g), and Q = PF, then to ZnF,

4. f(Q)(Z) = nF f(Z + X) dX, If fS(g), and if hh, then we write h = h, haF, and httF, then

$$\mathbf{u}(\mathbf{Z}) = \mathbf{u}(\mathbf{f}, \mathbf{h}\_{-})(\mathbf{Z}) = \mathbf{f}^{\prime(\mathbf{Q})}(\mathbf{h}\_{-}\mathbf{t}\mathbf{Z}),\tag{1V.68}$$

Z0nF, of where is demonstrated that

$$\Phi\mathsf{h}^{\mathsf{H}}(\mathsf{h}) = \mathsf{C}\mathsf{v}\Phi\mathsf{u}^{\mathsf{T}}(\mathsf{h}+),\tag{\mathsf{IV}.69}$$

Then the evaluations in (1), (2), (3), (4), and (5), imply the following theorem of Harish-Chandra:

### **Theorem IV. 2.**

44 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

f

Let t, be such that atna<sup>t</sup>U1, j = 1, 2. Then

, such that if

If hh, then det(adhnF) = (h) = (h)(h).

This combined with the lemma IV. 2., implies that

whose restriction to aF, is not vanishing, to know;

=

H(h) = CF(h) (h)<sup>K</sup> M N f(Ad(k)Ad(m)(h + X)) dk dm dn,

diffeomorphism of NF, in nF.

If f is a soft function on g, then

Now we elect

the

2. <sup>f</sup>

follows of the formula

In effect,

Then

If we demonstrate that Th, is a diffeomorphism of NF, in nF, then the integration formula can

to the differential of Th. Let h0aF, be such that (h0) > 0, (PF, AF). Let at = exp(th0), be then Th(atnat) = Ad(at)Th(n). Since Th, is in particular, regular in 1NF, exist an open

Th(atnat) = Ad(atnat)h

= Ad(at)Ad(n)Ad(at)

= Ad(at)Ad(at)Ad(n)h

= Ad(at)Th(n),

of which the equi-variance Th(atnat) = Ad(at)Th(n), implies that Th, is suprajective, since

<sup>t</sup>0Ad(at)U0 = Th(atnat) = nF,

Now only is necessary to demonstrate the injectivity of Th. To it, suppose that Th(n1) = Th(n2).

Th(atn1at) = Ad(at)Th(n1) = Ad(at)Th(n2) =Th(atn2at), Thus atn1at= atn2at. Thus n1 = n2. Then Th, is injective. Then Th, is bijective. Thus Th, is a

neighborhood U0, of 0 = 1nF, in nF, such that Th, of U1, in U0, is a diffeomorphism.

H(h) = (h) (h)G/NFf(Ad(g)h)d(gNF), (IV. 64)

<sup>t</sup>0Ad(at)U0 = nF, (IV. 65)

and if a 0, then a(PF, AF). Let , be the set of all

a<sup>F</sup> 0, (IV. 66)


The theorem IV. 1., is followed of the behalf formulas.

If Xg, then

$$\det(\text{ad}\mathcal{X} - \text{t}\mathcal{I}) = \sum \text{t}\mathcal{D}\_{\!\!\!/} (\mathcal{X})\_{\!\!\!/} \tag{\text{IV.70}}$$

with dim g = n. Let D = Dr(X), with dim h = r. Then from theorem of Harish-Chandra (theorem IV. 2), is followed:

**Corollary IV. 1.**D 1/2, is locally integrable on g.

*Proof*. Let the integral formula of Weyl on g,

$$\int \mathfrak{gl}\,\mathfrak{f}(\mathsf{X})d\mathsf{X} = \sum \mathsf{C}\_{\mathsf{f}}|\_{\mathfrak{h}}|\mathfrak{w}(\mathsf{h})|^{2}\Big|\_{\mathrm{C}\models\mathsf{H}}\mathfrak{f}(\mathrm{Ad}(\mathfrak{g})\mathsf{h})\mathrm{d}(\mathfrak{g}\mathsf{H})\mathrm{d}\mathsf{h}\Big|\tag{\mathsf{IV}.71}$$

In the sense of that the right side converges if the left side that does. By appendix A,

D(h)=j(h) 2, hhj.

Let fCc(g), be non-negative. Then by the theorem IV. 2., fS(hj), j = 1, 2, , r. Thus

Cjh<sup>j</sup>f(h)dh = Cjh'j(h)G/Hjf(Ad(g)h)dgHj dhj <sup>=</sup>Cjh'j(h)G/Hjj(h)D(Ad(g)h) -1/2f(Ad(g)h)dgHj dhj <sup>=</sup>Cjh'j(h) <sup>2</sup>G/HjD(Ad(g)h) -1/2f(Ad(g)h)dgHj dhj <sup>=</sup>gD(X) -1/2f(X)dX.

## **Integral Transforms on Lie Groups and Lie Algebras**

## **V. 1. Harish-Chandra function**

Harish-Chandra construct a spherical function, that have the finality of shape an estimation criteria to the rapidly decreasing of the nFmodules V , and the bound of the matrix coefficients under the asymptotic behavior that they apply. Such criteria come established under the range of definition of the function , of Harish-Chandra.

Let V a admissible (g, K)-modules. Let

$$\mathbf{V}^{\bullet} = \langle \mu \in \mathbf{V}^{\bullet} | \mathbf{K} \mu \text{, generates a subspace of finite dimension of } \mathbf{V}^{\bullet} \rangle , \tag{\text{V. 1.1}}$$

V , is also an admissible (g, K)-module. Give a characterization of vector space of the admissible (g, K)-module V .

**Lemma V. 1.1.** Let W, a (g, K)-module. Suppose that exist a complex bilinear map

$$\mathbf{b}: \mathbf{V} \times \mathbf{W} \to \mathbf{C} \tag{\text{V.1.2}}$$

such that

46 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

<sup>=</sup>Cjh'j(h)

<sup>=</sup>Cjh'j(h)G/Hjj(h)D(Ad(g)h)

Cjh<sup>j</sup>f(h)dh = Cjh'j(h)G/Hjf(Ad(g)h)dgHj dhj


<sup>2</sup>G/HjD(Ad(g)h)

<sup>=</sup>gD(X)




*Proof.* The idea is demonstrate that exist a g, K-invariant between the subspaces W, and V , of the homomorphism spaces Homg, K(V, W) and *Hom*g, K(V\*, V ). We define

$$\text{T:}\,\text{Hom}\_{\mathbb{R}}\,\text{x}(\text{V},\text{W})\to\text{Hom}\_{\mathbb{R}}\,\text{x}(\text{V}^{\*},\text{V}^{\*}),\tag{\text{V}.1.3}$$

whose rule of explicit correspondence wW, is T(w)(v) = b(v, w). Then, in principle T, is a belonging homomorphism to the space *Hom*g, K(W, V ). Indeed, consider to the composition

$$\mathbf{T} = \mathbf{T}^{\prime} \mathbf{o} \ \mathbf{T}^{\prime} \text{ \AA} \tag{\text{V}.14}$$

where T *Hom*g, K(V, W), and T *Hom*g, K(V\*, V ). Since T , and T , are homomorphism, these satisfies that T (V) W, and T (V\*) V . Then

$$\Gamma(\mathsf{W}) = \Gamma^{\circ} \circ \Gamma^{\prime}(\mathsf{W}) \subseteq \mathsf{V}^{\prime},\tag{\mathsf{V}.1.5}$$

Since b(v, w) 0, vV, wW, if and only if *ker*T = 0, if and only if T, is injective. Let K , and V(), be their corresponding representation in the (g, K)-module V. We define to \* = V (), such that T*Hom*g, K(W, V ), T(V()) W(\*). But by the non-degenerate of b, given that V 0, and the K-invariance of b,

$$\dim \mathcal{V}(\mathfrak{\chi}) = \dim \text{ (}\mathcal{W}(\mathfrak{\chi}^\*)\text{)},\tag{V.1.6}$$

then T(V()) = W(\*). Thus T, is suprajective.

Then T*Hom*g, K(W, V ), is a (g, K)-isomorphism of W, a V .

Let P = 0MAN, be the minimal parabolic subgroup of G, with G, real and reductive. Let (,H) = IndPG(), be with , the representation of P, given for

$$
\sigma\_{\mu}(\text{man}) = \mathbf{a}^{\mu+\rho} \sigma(\text{m}),
\tag{V.1.7}
$$

 m0M, aA, and nN, (a**C**)\*. Let , be the dual representation of .

**Lemma V. 1.2.** (H)K = (H)K.

*Proof.* Let f(H)K, and g(H)K, be then we realize

$$\text{} = \left[ \text{\(\kappa\), g(k)>} \text{\(k\)} \text{\(\kappa\)}\right] \tag{V. 1.8}$$

We can verify the properties of the bilinear form as inner product in the (g, K)-module W (exercise).

Indeed, Xg, and considering to , a linear conjugate isomorphism, to know;

$$
\sigma\_{\mu}(\mathbf{f})(\mathbf{g}) = \preccurlyeq \mathbf{f}, \mathbf{g} > \tag{V.1.9}
$$

that is to say, the isomorphism such that

$$(\mathbf{H}^{\nu})^\prime \mathbf{x} = \sigma(\mathbf{H}\mathbf{x}) = (\mathbf{H}\mathbf{x})^\prime,\tag{V.1.10}$$

we have then that,

$$
\sigma\_{\mu}(\mathbf{f})(\mathbf{g}) = \text{<\X\mu\text{f}, \text{g}>}
$$

$$
= \int\_{\mathbf{k}} \text{<\X\mu\text{f}(\mathbf{k}), \text{g}(\mathbf{k})>} \text{dk}
$$

$$
= \int\_{\mathbf{k}} \text{<\f}(\mathbf{k}), -\text{>\X\mu\text{g}(\mathbf{k})>} \text{dk}
$$

$$
= -\int\_{\mathbf{k}} \text{<\f}(\mathbf{k}), \text{<\X\mu\text{g}(\mathbf{k})>} \text{dk}
$$

$$
= -\text{<\f}, \text{<\X\mu\text{g}>} = \sigma\_{-\mu}(\mathbf{g})(\mathbf{f}) = \sigma\_{-\mu}^{\prime}(\mathbf{f})(\mathbf{g}).
$$

For other side (V. 1. 9), is non-degenerate, since (HK)(g) = <HK, g> = 0, if g = 0, and (f)(g) = <f, HK> = 0, if f = 0. Thus , is a isomorphism between the spaces (H)K, and (H, )K.

We define a functional . Consider (a**C**)\*, kK and gG. Then the functional , that through of the representation (g), is realized in the class space (H )K() = **C**I, come defined as:

$$\Xi\_{\mu}(\mathbf{g}) = \lnot \pi\_{\mu}(\mathbf{g}) \mathbf{1}\_{\mu} \ \mathbf{1}\_{\mu} \ \hspace{0.1.11} \tag{V.1.11}$$

where is clear that

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

(), such that T*Hom*g, K(W, V

given that V 0, and the K-invariance of b,

then T(V()) = W(\*). Thus T, is suprajective.

m0M, aA, and nN, (a**C**)\*. Let

that is to say, the isomorphism such that

*Proof.* Let f(H)K, and g(H)K, be then we realize

(H)'K = (HK) = (HK)

**Lemma V. 1.2.** (H)K = (H)K.

48

K

\* = V

(exercise).

we have then that,

Then T*Hom*g, K(W, V

.

, be the dual representation of .

<f, g> = K<f(k), g(k)> dk, (V. 1.8)

, (V. 1.10)

), T(V()) W(\*). But by the non-degenerate of b,

Since b(v, w) 0, vV, wW, if and only if *ker*T = 0, if and only if T, is injective. Let

 *dim* V() = dim (W(\*)), (V. 1.6)

Let P = 0MAN, be the minimal parabolic subgroup of G, with G, real and reductive. Let

(man) = a(m), (V. 1.7)

We can verify the properties of the bilinear form as inner product in the (g, K)-module W

(f)(g) = <f, g>, (V. 1.9)

(f)(g) = <Xf, g>

<sup>=</sup>K<Xf(k), g(k)> dk

<sup>=</sup>K<f(k), Xg(k)> dk

<sup>=</sup>K<f(k), Xg(k)> dk

= <f, Xg> = (g)(f) =

(f)(g),

Indeed, Xg, and considering to , a linear conjugate isomorphism, to know;

), is a (g, K)-isomorphism of W, a V

(,H) = IndPG(), be with , the representation of P, given for

, and V(), be their corresponding representation in the (g, K)-module V. We define to

$$\pi\_{\mu}(\mathbf{g})(\mathbf{1}\_{\mu}) = \lhd \pi\_{\mu}(\mathbf{g})\mathbf{1}\_{\mu} \quad \mathbf{1}\_{\mu} > -\int\_{\mathbb{X}} \mathbf{a}(\mathbf{k}\mathbf{g})^{\mu+\rho} \, \mathrm{d}k,$$

 gG, where **1**(g) = a(g), and **1**(g) = 1, gG, kK. We call = 0. Then we define the function , of Harish-Chandra as follows:

**Def. V. 1.1.** Let g0MAN. We define to the function (a), like a solution to the integral equation

$$\int \mathbf{x} \, \mathbf{f}(\mathbf{n} \mathbf{a} \mathbf{k}) \mathbf{a}(\mathbf{k})^{\mu+\rho} \, \mathrm{d}\mathbf{k} = \mathbf{f}(\mathbf{n}) \mathbf{f}(\mathbf{a}),\tag{V.1.12}$$

n, a0MAN, where explicitly whose function have a range defined by

Cta(1 + logad)(a) a, (V. 1.13)

**Theorem V. 1.1.** Exists positive constants C, and d > 0, such that

$$\mathbf{a}^{-\rho} \leq \Xi \text{(a)} \leq \mathbf{C} \mathbf{a}^{-\rho} \text{(1} + \log \|\mathbf{a}\| \text{\textquotedbl{}}\text{)}\tag{V.1.14} \text{\textquotedbl{}}\tag{V.1.14}$$

aCl(A+).

*Proof.* Let (, H), be denoted by (0, H0), under the functional < , >, on HK = (HK) . To this only is required remember that we define the functional isomorphism

$$
\boldsymbol{\sigma} . \mathbf{H} \boldsymbol{\omega} \times \mathbf{H} \mathbf{k} \to \begin{pmatrix} \mathbf{H} \mathbf{k} \end{pmatrix} \text{\textquotedbl{}}\tag{\text{V.1.15}}
$$

with rule of correspondence

$$\vdash (\mathbf{v}, \mathbf{w}) \, \mathbf{I} \rightarrow \sigma(\mathbf{v}) \\
\mathbf{w} = \operatorname{\!\!\sim} \mathbf{v}, \; \mathbf{w} \rhd \,, \tag{\text{V. 1.16}}$$

 v, wHK. Likewise, by the lemma V. 1. 2, extending said product to all the compact subgroup K, it goes over to that (HK) = (HK)' = (HK). Thus HK = (HK) , under the functional (.)(,) = < , >.

Let 10, be the identity of the class space HK(0) = **C**I. Let 10HK (0). Then

$$
\Xi(\mathbf{g}) = \text{<\pi(g)1} \boldsymbol{\upiota}, \; 1 \boldsymbol{\upiota} \succeq \; \tag{V.1.17}
$$

Let V, be a (g, K)-submodule of HK, generated by 10, that is to say

$$\mathbf{V} = \sum \pi(\mathbf{K}) \mathbf{1}\_{0} \tag{\text{V. 1.18}}$$

Then under < , >, V = V. Indeed,

$$\Xi(\mathbf{g}) = \lnot\pi(\mathbf{g}) 1\_{\lozenge} 1\_{\lozenge} = \int\_{\mathbb{K}} \lnot\pi(\mathbf{g}) 1 \bullet \mathbf{(k)} \,\, 1 \bullet \mathbf{(k)} \rhd \mathbf{k} \,\, \mathbf{k}$$

$$= \int\_{\mathbb{K}} \lnot\lnot\mathbf{(k)} \,\, \pi(\mathbf{g}) 1 \bullet \mathbf{(k)} \rhd \mathbf{k}$$

$$= < 1 \imath \,\, \pi(\mathbf{g}) 1 \oslash \mathbf{0}$$

Thus (g)V = V = V . Suppose that , is a weight of a, on V/nV, that is to say, (V/nV), is a generalized eigenspace (a**C**)\*, and such that

$$\bigoplus\_{\mu,\,\rho\,\simeq} \text{(}\mathsf{v}\text{\textquotedblleft}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblleft}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblright}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblright}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedbl}\text{\textquright}\text{\textquright}\text{\textquright}\text{\textquright}\end{\text{\textquotedblleft}\text{\textquotedblleft}\text{\textquotedbl}}\text{\textquright}\text{\textqu0}\text{\textqu0}\text{\textqu$$

Let , be a 0M-type of the weight space (V/nV) = (H) = H , with a, acting for ()I. Then exist an element non-null in *Hom*g, K(V, (H )K). The reciprocity theorem of Frobenius implies that , can be of type 0M. Then this implies that = 0. Then, due to that of the class space (HK )(0) = **C**I, we can deduce directly of the isotopic class 0, that

$$
\gamma\!(\mathbb{C}(\mathfrak{g})^{\mathbb{K}}) \subset \mathbb{C}(\mathfrak{a})^{\mathbb{N}}\,\tag{V.1.20}
$$

with W = W(g, a), if and only if = <sup>s</sup>, sW, and (a**C**)\*. In particular, if = 0, then = s = 0, sW(G, A). Thus = 0, where V = .

Then

$$\Xi(\mathbf{a}) \le \mathbf{C} \mathbf{a}^{\Lambda \mathbf{v}} (1 + \log \|\mathbf{a}\|^{\mathbf{d}}) \le \mathbf{C} \mathbf{a}^{-\mathbf{p}} (1 + \log \|\mathbf{a}\|^{\mathbf{d}}),\tag{V.1.21}$$

Now we demonstrate the inferior inequality. By the formula

$$\Xi\_{\mu}(\mathbf{a}) = \mathbf{a}^{\mu+\rho} \Big|\_{\mathbf{T}} \mathbf{\underline{a}(\underline{\mathbf{n}})^{-\mu+\rho}} \mathbf{a}(\mathbf{a}^{-1}\underline{\mathbf{n}}\mathbf{a})^{\mu+\rho} d\underline{\mathbf{n}},\tag{V.1.22}$$

then

$$\Xi(\mathbf{a}) = \mathbf{a}^{\rho} \Big|\_{\Delta \mathbf{a}(\underline{\mathbf{n}})} ^{\rho} \mathbf{a}(\mathbf{a}^{-1} \underline{\mathbf{n}} \mathbf{a})^{\rho} d\underline{\mathbf{n}} \tag{V.1.23}$$

Realizing the variable change described by the correspondence

$$
\underline{\mathbf{n}} \mid \begin{array}{c} \rightarrow \underline{\mathbf{a}} \underline{\mathbf{a}}^{-1} \end{array} \tag{V.1.24}
$$

we have

Chapter V – Integral Transforms on Lie Groups and Lie Algebras 51

$$\Xi(\mathbf{a}) = \mathbf{a}^{-\rho} \Big|\_{\Xi \mathbf{a}} \mathbf{a} (\mathbf{a} \underline{\mathbf{a}} \mathbf{a}^{-1} \mathbf{a} (\underline{\mathbf{n}})^{-1})^{\rho} \mathbf{a} (\underline{\mathbf{n}})^{\underline{\mathbf{n}} \cdot} \mathbf{d} \underline{\mathbf{n}} \tag{\text{V. 1.25}}$$

Knowing that in G is defined a norm , exists (a**C**)\*, with (Hj) > 0, j, and C1, C2> 0, (constants), such that

 C1aaC2a , (V. 1.26)

aCl(A+). Of this inequality is deduced immediately the inferior inequality.

Now we will demonstrate as Harish-Chandra has used the before result to prove the convergence of two important integrals.

**Theorem V. 1.2.** Let dR+. If and if F0, then

$$\mathbf{a}^{-\rho} \Big|\_{\underline{\mathbf{u}} \neq \underline{\mathbf{u}}} (\underline{\mathbf{u}})^{\rho} (1 - \rho(\log \underline{\mathbf{u}}))^{-\underline{\mathbf{u}} - \underline{\mathbf{u}}} \mathrm{d}\mathbf{r} \lhd \mathbf{a},\tag{\text{V. 1.27}}$$

that is to say, said integral is convergent to a invariant measure on NF.

*Proof*. Consider hCl(a+), and at = *exp*th. Then the theorem V. 1. 1., implies that exist a positive constant such that

$$(\mathbf{a})^0 \Xi(\mathbf{a}) \le \mathbf{C} (1+\mathbf{t})^d,\tag{V.1.28}$$

but by the theorem V. 1. 1., and the theorem D. 1., (Apendix D), is had that

$$\Xi(\mathbf{a}) \le \text{(a)}^\rho \Xi(\mathbf{a}) = \mathbf{a}^{-\rho} \Big\| \Xi(\underline{\mathbf{n}})^\rho \mathbf{a} (\mathbf{a} \underline{\mathbf{n}} \mathbf{a}^{-1})^\rho \mathbf{d} \underline{\mathbf{n}} \tag{\text{V. 1.29}}$$

We elect a hCl(a+), such that

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Let V, be a (g, K)-submodule of HK, generated by 10, that is to say

= V. Indeed,

is a generalized eigenspace (a**C**)\*, and such that

Then exist an element non-null in *Hom*g, K(V, (H

s = 0, sW(G, A). Thus = 0, where V = .

(a) = a

Now we demonstrate the inferior inequality. By the formula

Realizing the variable change described by the correspondence

Let , be a 0M-type of the weight space (V/nV) = (H) = H

50

Then under < , >, V

Thus (g)V = V = V

space (HK

Then

then

we have

. Suppose that , is a weight of a, on V/nV, that is to say, (V/nV),

(a**<sup>C</sup>**)\*(V/nV) = V/nV, (V. 1.19)

)K). The reciprocity theorem of Frobenius

dn, (V. 1.23)

, with a, acting for ()I.

V = (K)10, (V. 1.18)

(g) = <(g)10, 10>, (V. 1.17)

(g) = <(g)10, 10> = K<(g)10(k), 10(k)> dk

<sup>=</sup>K<10(k), (g)10(k)> dk

= <10, (g)10>,

implies that , can be of type 0M. Then this implies that = 0. Then, due to that of the class

0(U(g)K) U(a)W, (V. 1.20)

with W = W(g, a), if and only if = <sup>s</sup>, sW, and (a**C**)\*. In particular, if = 0, then =

(a)CaV(1 + logad) Ca(1 + logad), (V. 1.21)

(a) = aNa(n)a(a1na)dn, (V. 1.22)

n ∣ana1, (V. 1.24)

a(a1na)

Na(n)

)(0) = **C**I, we can deduce directly of the isotopic class 0, that

$$\begin{cases} \alpha(h) = 0, \forall \alpha \in F, \\ \alpha(h) = 1, \forall \alpha \in \Lambda\_0 - F, \end{cases} \tag{V.1.30}$$

then mF = Cg(h). Indeed, for other side

$$\mathbf{C}\_{\mathsf{f}}(\mathsf{h}) = \langle \mathbf{h} \in \mathfrak{a}^{\star} | \mathrm{Ad}(\mathsf{g}) \mathbf{h} = \mathbf{h}, \,\forall \, \mathbf{g} \in \mathbf{G} \rangle,\tag{V.1.31}$$

Then gG and ha,

$$\mathbf{M} = \{ \mathbf{g} \in \mathbf{G} | \mathbf{Ad} | \!= \mathbf{I} \}, \tag{\text{V. 1.32}}$$

But M = exp(m**C**) then

$$\mathfrak{m} = \langle \mathbf{h} \in \mathfrak{a} | \mathrm{Ad}(\mathfrak{g}) \mathbf{h} = \mathbf{h}, \text{ if and only if } [\mathbf{h}, \mathfrak{g}] = 0 \rangle,\tag{V.1.33}$$

Then mF = aF [g, g], of where

$$\mathfrak{m} = \langle \mathfrak{h} \in \mathfrak{a}^\* | \mathrm{Ad}(\mathbf{g}) \mathfrak{h} = \mathfrak{h}, \,\forall \, \mathbf{g} \in \mathbf{G} \rangle,\tag{V.1.34}$$

Thus mF = Cg(h).

Let \*aF = mF nF. Then nF= \*mF nF. Due to a lemma on the normalization of all invariant measure on the subgroup \*NF, such that

$$\int\_{\Delta \mathsf{F}} \mathsf{f}(^{\ast}\underline{\mathsf{n}})^{2} \mathsf{d}\underline{\mathsf{n}}\mathsf{F} \equiv 1,\tag{\mathsf{V}.1.35}$$

By properties that are deduced of the normalization, can normalize the invariant measure on NF, such that if fCc(N), then

Nf(n)dn=\*NFNFf(\*nFnF) d\*nFd\*nF, (V. 1.36)

of which we affirm is that

$$\begin{array}{c} \mbox{(a)}^{\rho}\Xi(\mbox{\bf a}) = \int\_{\mbox{N}\mathsf{V}} \mbox{a}(\mbox{\bf n})^{\rho}\mbox{a}(\mbox{a}\mbox{\bf a}\mbox{\bf a}\mbox{\bf a}^{-1})^{\rho}\mbox{d}\mbox{n} \end{array} \tag{\bf V. 1.37}$$

Indeed, denoting by J(t), the right side of (V. 1. 37), we have that if at xat = x, x\*NF, and xNa(x)k(x), is had that the transformation of (V. 1. 29), using (V. 1. 36), we give

<sup>J</sup>(t) = \*NFNFa(\*nF)2 a(k(\*nF)nF) a(k(\*nF)atnat 1) d\*nFd\*nF, (V. 1.38)

Indeed, we do resource of that analytic expression of the function of Harish-Chandra

$$\Xi(\mathbf{a}) = \mathbf{a}^{-\rho} \Big|\_{\Delta\mathbf{\hat{a}}} (\mathbf{n})^{2\rho} (\mathbf{a} (\mathbf{a}\underline{\mathbf{a}}\,\mathbf{a}^{-1}) \mathbf{a}(\underline{\mathbf{n}})^{-1})^{\rho} \mathrm{d}\underline{\mathbf{n}},\tag{V.1.39}$$

which is a simple consequence of change of variable contemplated in (V. 1. 24), and effectuated in the function (V. 1. 23). If at\*nFat= \*nF, \*nF\*NF, and \*nFNFa(\*nF)k(\*nF), then

(at) (at) = Na(n)2 (a(ana1)a(n)1) dn <sup>=</sup>\*NFNFa(a(a\*nFnFa1)a(\*ntn)1) a(\*nFnF) d\*nFdnF <sup>=</sup>\*NFNFa(\*nFnF) a(a\*nFnFa1) a(\*nFnF) d\*nFdnF,

But k(\*nF) = nF, since if \*nFNFa(\*nF)k(\*nF), and k(\*NF) KF, then

$$\mathbf{n} = \overline{\mathbf{n}} \mathbf{a} (^{\ast}\mathbf{n}) \mathbf{k} (^{\ast}\mathbf{n} \overline{\mathbf{n}}) , \tag{V.140}$$

then

(at) (at) = J(t), <sup>=</sup>\*NFNFa(\*nF)2 a(k(\*nF)nF) a(k(nF)atnat) d\*nFdnF,

having that a(katnat) = a(atknk1at), kKF, tR, and nNF. In KF, compact dknFk1 = dnF, on NF, kKF, then

$$\mathfrak{A}(\mathfrak{t}) = \int \underline{\text{\raisebox{-0.5ex}{ $\cong$ }}} \underline{\text{\raisebox{-0.5ex}{ $\cong$ }}} \mathbf{\underline{\text{\raisebox{-0.5ex}{ $\cong$ }}}} (\underline{\text{\raisebox{-0.5ex}{ $\cong$ }}})^{\mathfrak{\color{0.5ex}{ $\cong$ }}} \mathbf{\underline{\text{\color{0.5ex}{ $\cong$ }}}} (\underline{\text{\underline{\text{\color{0.5ex}{ $\cong$ }}}})^{\mathfrak{\color{0.5ex}{ $\cong$ }}}} \mathbf{\underline{\text{\color{0.5ex}{ $\cong$ }}}} \mathbf{\underline{\text{\color{0.5ex}{ $\cong$ }}}} \mathbf{\underline{\text{\color{0.5ex}{ $\cong$ }}}} \mathbf{\underline{\text{\color{0.5ex}{ $\cong$ }}}} $$

where the \*NFa(\*nF)d\*nF = 1, which will implies the integral expression of (at) (at). In particular (V. 1.37), implies that

$$\int \mathsf{A} \mathsf{z} \mathsf{a} (\underline{\mathsf{n}} \mathsf{r})^{\mathsf{o}} \mathsf{a} (\mathsf{a} \mathsf{n} \mathsf{F} \mathsf{a} \mathsf{n} \mathsf{r})^{\mathsf{o}} \mathsf{d} \mathsf{n} \mathsf{r} \mathsf{d} \mathsf{\*} \mathsf{n} \leq \mathsf{C} (1 + \mathsf{t})^{\mathsf{d}},\tag{\mathsf{V}. 1.41}$$

If F = , then to any t 0,

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

<sup>J</sup>(t) = \*NFNFa(\*nF)2

(at) 

<sup>=</sup>\*NFNFa(\*nFnF)

<sup>=</sup>\*NFNFa(\*nF)2

But k(\*nF) = nF, since if \*nFNFa(\*nF)k(\*nF), and k(\*NF) KF, then

(a) = aNa(n)2

(at) = NF a(n)

xNa(x)k(x), is had that the transformation of (V. 1. 29), using (V. 1. 36), we give

a(k(\*nF)nF)

Indeed, we do resource of that analytic expression of the function of Harish-Chandra

(at) = Na(n)2

(at) 

a(k(\*nF)nF)

<sup>=</sup>\*NFNFa(a(a\*nFnFa1)a(\*ntn)1)

52

Thus mF = Cg(h).

measure on the subgroup \*NF, such that

on NF, such that if fCc(N), then

(at)

of which we affirm is that

then

then

NFf(\*nF)2dnF= 1, (V. 1.35)

dn, (V. 1. 37)

d\*nFd\*nF, (V. 1.38)

dn, (V. 1.39)

Nf(n)dn=\*NFNFf(\*nFnF) d\*nFd\*nF, (V. 1.36)

1) 

 dn

d\*nFdnF

d\*nFdnF,

d\*nFdnF,

Let \*aF = mF nF. Then nF= \*mF nF. Due to a lemma on the normalization of all invariant

By properties that are deduced of the normalization, can normalize the invariant measure

Indeed, denoting by J(t), the right side of (V. 1. 37), we have that if at xat = x, x\*NF, and

which is a simple consequence of change of variable contemplated in (V. 1. 24), and effectuated in the function (V. 1. 23). If at\*nFat= \*nF, \*nF\*NF, and \*nFNFa(\*nF)k(\*nF),

a(a\*nFnFa1)

nF = nFa(\*nF)k(\*nF), (V. 1.40)

(at) = J(t),

a(k(nF)atnat)

a(atnat 1) 

a(k(\*nF)atnat

(a(ana1)a(n)1)

(a(ana1)a(n)1)

 a(\*nFnF)

 a(\*nFnF)

$$\mathbf{a}(\mathbf{a}\underline{\mathbf{n}}\mathbf{a}...)^{-2} = \|\mathbf{a}(\mathbf{a}\underline{\mathbf{n}}\mathbf{a}...)^{-1}\mathbf{v}\mathbf{v}\|\,\tag{V.1.42}$$

is the polynomial in a subspace W 0, to XnF. Developing (atnFat) 1v02,

$$\begin{aligned} \|\mathsf{f}(\mathsf{a}\underline{\mathsf{n}}\mathsf{n}\mathsf{a}\mathsf{n}\mathsf{-})^{-1}\mathsf{v}\_{0}\|^{2} &= \|\mathsf{v}\_{0} + \Sigma\_{\mathsf{i}\mathsf{0}}\mathsf{e}^{-\mathsf{j}\mathsf{t}}(\mathsf{o}\mathsf{(}\underline{\mathsf{n}}\mathsf{F}\mathsf{)}^{-1}\mathsf{v}\_{0})\|\|^{2} \leq \\ &\leq 1 + \|\mathsf{e}^{-\mathsf{j}\mathsf{t}}(\mathsf{o}\mathsf{(}\underline{\mathsf{n}}\mathsf{E}\mathsf{)}^{-1}\mathsf{v}\_{0})\|\|^{2} \\ &= 1 + \mathsf{e}^{-\mathsf{j}\mathsf{t}}\mathsf{a}(\underline{\mathsf{n}}\mathsf{F})^{-4\mathsf{d}}\leq \\ &\leq (1 + \mathsf{e}^{-\mathsf{t}}\mathsf{a}(\underline{\mathsf{n}}\mathsf{E})^{-2\mathsf{d}})^{2}, \end{aligned}$$

where it has been demonstrated that

$$\mathbf{a}\langle \underline{\mathbf{a}} \underline{\mathbf{n}} \underline{\mathbf{a}}\_{-} \rangle \ge (1 + \mathbf{e}^{-\mathfrak{p}} \mathbf{a} \langle \underline{\mathbf{n}} \underline{\mathbf{e}} \rangle^{-2\mathfrak{p}})^{-1/2},\tag{V.1.43}$$

 t 0, and nFNF. If r > 0, then we define (NF)r = nNF a(n) r. By a lemma on bounden subgroups of a real reductive group, is had that (NF)r, is a bounded subgroup to any r > 0. For which said subgroup is compact r >0. In (V. 1. 43), we take t = 2*log* r, 0 < r < 1. Then (V. 1. 41), implies that if n(NF)r, then

$$\mathbf{a}(\mathbf{a}\underline{\mathbf{n}}\underline{\mathbf{a}}\mathbf{a}\dots) \ge \mathcal{D}^{-1/2}\mathbf{a}$$

For which is necessary find that

$$\mathbb{C}(1+\mathsf{t})^{\mathsf{d}} \geq \int\_{(\mathsf{i}\mathbb{B})^{\mathsf{r}}} \mathsf{a}(\underline{\mathsf{n}})^{\mathsf{o}} \mathsf{a}(\underline{\mathsf{a}}\underline{\mathsf{n}}\underline{\mathsf{a}}\ldots)^{\mathsf{o}} \, d\underline{\mathsf{n}} \geq 2^{-1/2} \int\_{(\mathsf{i}\mathbb{B})^{\mathsf{r}}} \mathsf{a}(\mathsf{n})^{\mathsf{o}} d\mathsf{n} \,\tag{\mathsf{V}. 1.44} \geq 1$$

which implies

$$\int\_{(\underline{\mathbf{N}})^{\mathfrak{r}}} \mathsf{a}(\underline{\mathbf{n}})^{\mathfrak{r}} d\underline{\mathbf{n}} \leq \mathsf{C}'(1 - \mathfrak{Z} \log \mathbf{r})^{\mathfrak{d}},\tag{\mathsf{V}.1.45}$$

Now we have rp = exp(2 ), p = 0, 1, , with the notation of (\*X), is had that

$$\int\_{(\Delta\mathbb{P})^{\mathrm{p}}} \mathbf{a}(\underline{\mathbf{n}})^{\mathrm{p}} d\underline{\mathbf{n}} \le \mathbf{C}'(1+\mathcal{D}^{\mathrm{p+1}})^{\mathrm{d}} \le \mathbf{C}(1/\mathcal{D})^{\mathrm{pd}},\tag{\text{V. 1.46}}$$

If n(NF)rp+1 (NF)r, then rp a(n) rp+1. For which, on this same inequality we have

$$1 + 2^{\underline{p}} \le (1 - \rho(\log \text{ a}(\underline{\text{n}})) \le 1 + 2^{\underline{p}+1},\tag{V.1.47}$$

This implies that if > 0, then

$$\int \underline{\operatorname{\mathfrak{L}}} \operatorname{\mathfrak{p}} \underline{\operatorname{\mathfrak{p}}} \text{\- } \underline{\operatorname{\mathfrak{L}}} \operatorname{\mathfrak{p}} \underline{\operatorname{\mathfrak{q}}} (1 - \operatorname{\mathfrak{p}} (\log \operatorname{\mathfrak{a}} \underline{\operatorname{\mathfrak{m}}}))^{\operatorname{\mathfrak{q}} - \operatorname{\mathfrak{c}}} \operatorname{\mathfrak{d}} \underline{\operatorname{\mathfrak{m}}} \leq \operatorname{C}' (1 + 2\operatorname{\mathfrak{p}})^{\operatorname{\mathfrak{q}} - \operatorname{\mathfrak{c}}} 2^{\operatorname{\mathfrak{q}}} \leq \operatorname{C} (1/2)^{\operatorname{\mathfrak{q}}} \leq \operatorname{C}^{\operatorname{m}} 2^{-\operatorname{\mathfrak{c}}} \;/\ \ \text{ (V. 1.48)}$$

If we add on p > 0, then

$$\int\_{(\Delta \mathbb{F})^{-} (\Delta \mathbb{F})^{\mathbb{P}}} (\underline{\mathbf{n}})^{\mathbb{P}} (1 - \rho(\log \mathbf{a}(\underline{\mathbf{n}}))^{-\mathbb{L} - \mathbb{E}} \underline{\mathbf{d}} \underline{\mathbf{n}} \leq \mathbf{C}^{\mathbf{m}} \underline{\sum} 2^{-\mathbb{E}\_{\mathbf{p}}} < \infty)$$

This implies that

$$\int\_{\Delta \mathbb{F}} \mathbf{a}(\mathbf{n})^{\mathbb{P}} (1 - \rho(\log \mathbf{a}(\underline{\mathbf{n}}))^{-\mathbb{d}-\varepsilon} \mathrm{d}\mathbf{n} \lhd \infty)$$

with (NF)r, compact.

**Theorem V. 1.3.** If r > 0, and if q > d + r, then

$$\int\_{\underline{\mathsf{T}}} \underline{\mathsf{a}}(\mathsf{n})^{\diamond} \Xi^{\mathsf{r}}(\underline{\mathsf{n}}) (1 + \mathsf{b}(\log||\mathsf{m} \,\mathsf{r}(\underline{\mathsf{n}})||)^{\mathsf{q}} (1 - \mathsf{b}(\log|\mathsf{a}(\underline{\mathsf{n}})|)^{\mathsf{q}} \mathsf{d}\underline{\mathsf{n}} \lhd \infty) \tag{\mathsf{V. 1.49}})$$

*Proof*. To prove such affirmation, first is necessary to prove that exist a constant C > 0, such that

$$|1 + \log ||\mathbf{m} \text{( $\underline{n}$ )}|| \le \mathbf{C} (1 + \log ||\mathbf{n}|| - \rho \log \mathbf{a}(\underline{n})) . \tag{V. 1.50}$$

nNF.

We assume:

i. g = (g), with (, F), a representation of finite dimension of G, and (g), a Hilbertian norm of (g), relative to < , >, of F, such that

$$
\sigma(\mathbf{g})^\* = \sigma(\theta(\mathbf{g}))^{\cdot 1},
$$

ii. We elect a base uj F, such that the block of the representation of MF, let be a diagonal block:

$$
\begin{bmatrix}
A\_1, \dots, \quad \dots & \dots & 0 \\
\dots A\_2, \dots, \quad \dots & \dots & \dots \\
\dots, \dots, \dots & \dots & \dots & \dots \\
0, \dots, \dots & \dots & \dots & A\_d
\end{bmatrix}
$$

and the elements of NF, conform the block of the superior triangular form:


then

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

(NF) (NF)r0 a(n)

NF a(n)

(1 (*log* a(n))ddn C'(1 2p)

If n(NF)rp+1 (NF)r, then rp a(n)

This implies that if > 0, then

If we add on p > 0, then

This implies that

with (NF)r, compact.

that

nNF.

We assume:

block:

(NF)rp+1(NF)r a(n)

**Theorem V. 1.3.** If r > 0, and if q > d + r, then

54

1 + 2p (1 (*log* a(n)) 1 + 2p+1, (V. 1.47)

d2pd C(1/2)pd Cm2p, (V. 1.48)

rp+1. For which, on this same inequality we have

(1 (*log* a(n))ddn Cm2p <,

NF a(n)FFn(1 (*log*mF(n))d(1 (*log* a(n))qdn<, (V. 1.49)

0

,

*A*

1 + *log*mF(n) C(1 + *log*n*log* a(n)), (V. 1.50)

(1 (*log* a(n))ddn<,

*Proof*. To prove such affirmation, first is necessary to prove that exist a constant C > 0, such

i. g = (g), with (, F), a representation of finite dimension of G, and (g), a

1 2

*A A*

and the elements of NF, conform the block of the superior triangular form:

(g)\* = ((g))-1, ii. We elect a base uj F, such that the block of the representation of MF, let be a diagonal

0 *<sup>d</sup>*

 

Hilbertian norm of (g), relative to < , >, of F, such that

$$||\mathfrak{a}(\overline{\mathfrak{n}})\mathfrak{a}(\overline{\mathfrak{n}})\_{\*}|| = ||\mathfrak{a}(\mathfrak{n}\circ(\overline{\mathfrak{n}})\mathfrak{a}(\mathfrak{m}\circ(\overline{\mathfrak{n}}))\mathfrak{a}(\mathfrak{m}\circ(\overline{\mathfrak{n}}))^{\*}\mathfrak{a}(\mathfrak{n}\circ(\overline{\mathfrak{n}}))^{\*}\mathfrak{a}(\mathfrak{m}(\overline{\mathfrak{n}}))^{\*}||$$

where

$$||\mathfrak{a}(\mathbf{m}\mathfrak{r}(\underline{\mathbf{n}}))\mathfrak{a}(\mathbf{m}\mathfrak{r}(\underline{\mathbf{n}}))^{\*}\mathfrak{a}(\mathbf{a}\mathfrak{r}(\underline{\mathbf{n}}))^{2}||\overline{\varsigma}||\mathfrak{a}(\underline{\mathbf{n}})\mathfrak{a}(\underline{\mathbf{n}})^{\*}||$$

which implies that

$$||\mathfrak{a}(\mathbf{m}\mathfrak{r}(\underline{\mathbf{n}}))\mathfrak{a}(\mathbf{m}\mathfrak{r}(\underline{\mathbf{n}}))^{\*}|| \le ||\mathfrak{a}(\mathbf{a}\mathfrak{r}(\underline{\mathbf{n}}))||^{2}||\underline{\mathbf{n}}||^{2} \tag{V.1.51}$$

,

If we apply the referent lemmas to the properties of boundess of a subgroup of a real reductive group, the consideration of the logarithm of the last inequality must be put yet, we have (V. 1. 50), nNF.

Now well, this last inequality, to the light of the properties of invariant integration on subgroups of a real reductive group, implies that is sufficient to demonstrate that the following integral is finite, q > d + r, that is to say;

$$\mathfrak{J}(\mathfrak{t}) = \int \overline{\mu} \mathbf{z} \mathbf{a}(\mathfrak{n})^{\otimes} \Xi \overline{\mathfrak{z}} \overline{\mathfrak{z}} (\overline{\mathfrak{n}}) (1 - \rho(\log \mathbf{a}(\mathfrak{n}))^{\mathbb{L}\_{-\mathbb{Q}}} \mathrm{d}\underline{\mathfrak{n}} \lhd \underline{\mathfrak{c}}) \tag{V.1.52}$$

 q > d + r. But for definition of the function F, actually, the function , to 0MF, we have that said integral take the form

$$\mathbf{\tilde{g}(t) = \int \mathsf{T}\mathsf{F}\mathsf{a}(\mathsf{n})^{\mathsf{d}t} \int \mathsf{x}\mathsf{a}(\mathsf{k}\mathsf{m}\mathsf{r}(\mathsf{n}))^{\mathsf{d}} \,\mathrm{d}\mathsf{k} (1 - \mathsf{p}(\log \mathsf{a}(\mathsf{n}))^{\mathsf{r} - \mathsf{q}} \,\mathrm{d}\mathsf{n}) \tag{V.1.53}$$

If kKF, and nNF, then

$$\mathbf{k}\,\underline{\text{kp}} = \mathbf{k}\,\mathrm{nm}\,\underline{\text{r}}(\underline{\mathbf{n}})\mathrm{a}\,\underline{\text{r}}(\underline{\mathbf{n}})\mathrm{k}\,\underline{\text{r}}(\underline{\mathbf{n}}) = \mathbf{k}\,\mathrm{nk}^{-1}\mathbf{k}\,\mathrm{m}\,\underline{\text{r}}(\underline{\mathbf{n}})\mathrm{k}^{-1}\mathbf{a}\,\underline{\text{r}}(\underline{\mathbf{n}})\mathrm{k}\,\underline{\text{r}}(\underline{\mathbf{n}}),\tag{V.1.54}$$

with nNF. Thus kmF(n) = nF(knk1), and aF(kn) = aF(n). This will implies that

$$\begin{split} \mathfrak{F}(\mathfrak{f}) &= \int\_{\Delta \mathbb{F} \times \mathbb{K} \text{Pa}(\underline{\mathbf{k}} \underline{\mathbf{n}} \, \underline{\mathbf{k}}^{-1})^{\mathbb{P}\_{\overline{\mathbf{f}}}} \mathbf{a} (\underline{\mathbf{m}} \underline{\mathbf{k}} \underline{\mathbf{n}} \, \underline{\mathbf{k}}^{-1}) (1 - \mathfrak{p} (\log \underline{\mathbf{a}} \underline{\mathbf{n}}))^{\mathbb{P}\_{-\overline{\mathbf{f}}}} \, \underline{\mathbf{d}} \, \underline{\mathbf{d}} \, \underline{\mathbf{k}} \\ &= \int\_{\Delta \mathbb{F} \times \mathbb{K} \text{Pa}(\underline{\mathbf{n}})^{\mathbb{P}\_{\overline{\mathbf{f}}}}} (1 - \mathfrak{p} (\log \underline{\mathbf{a}} (\underline{\mathbf{k}} \underline{\mathbf{n}} \underline{\mathbf{k}}^{-1}))^{\mathbb{P}\_{-\overline{\mathbf{f}}}} \, \underline{\mathbf{d}} \, \underline{\mathbf{d}} \, \underline{\mathbf{k}} \, \underline{\mathbf{k}} \\ &\leq \int\_{\Delta \mathbb{F} \mathbf{a}(\underline{\mathbf{n}})^{\mathbb{P}\_{\overline{\mathbf{f}}}}} (1 - \mathfrak{p} (\log \underline{\mathbf{a}} (\underline{\mathbf{n}}))^{\mathbb{P}\_{-\overline{\mathbf{d}}}} \, \underline{\mathbf{d}} \, \underline{\mathbf{d}} \, \underline{\mathbf{k}} \, \underline{\mathbf{k}} \, \end{split}$$

which is obvious by the theorem that establishes the existence of positive constant to continuous semi-norm (appendix C), in the study of asymptotic behavior of the matrix coefficients.

## **V. 2. Twistor transform to ladder representations**

Let P+, and P+, be open orbits of SU(2, 2), on **M** = Gr2(**C**4), and P = CP3 = P(C4), respectively. The dual projective twistor space is P\* = P((C4)\*). The sheaves of holomorphic functions and of holomorphic 3-forms (that is to say, belonging to space 3(S3))), are denoted by *O*, and 3, respectively.

The Penrose transform to this case result completely natural and, in particular this isomorphism with cohomology result be SU(2, 2)-equivariant.

Is clear that twistor description of this representation is one of the most clear than the version of Maxwell fields to effects of induction of G-modules. The action of SU(2, 2), is automatic. Now well, is natural question how the scalar product of the Hermitian form arises in this twistor correspondence.

Indeed, motivated by the classical construction is well acquaintance that this can be feasible by an integral intertwining operator. The cohomology H1(P+, 3), (whereP +, represent the closure of the open orbit P +) give an ascent directly to fields on P +, considering that the description "potential module gauge", of the same fields finds their interpretation on the dual twistor space like H1(P \* ,*O*).

Note that P +, is simply connect (contractible really), and have base neighborhood of Stain such that the original description of field and the description "potential modulo gauge", meet.

Thus, the integral intertwining operator, which establish the crucial ingredient in the classical construction is interpreted in the twistor description like a integral intertwining twistor operator, being this the twistor transform

$$\mathcal{T}: \mathbf{H^1(]P^+|\ \mathcal{Q}^\clubsuit) \to \mathbf{H^1(]P^{\*\top}|\ \mathcal{Q})} \,, \tag{\text{V. 2.1}}$$

Considering this operator is easy describe the scalar product in a path in the which is necessary the SU(2, 2)-invariance. If H1(P\* , *O*), then H1(P, *O*), and to obtain the invariance under conformal transforms of such product, is necessary the multi-linear pairing of G-modules H1(P+, 3), and H1(P, *O*),given by

$$\mathbf{H}^{\mathfrak{l}}(\mathbb{P}^{\star}|\mathcal{Q}^{\triangleright}) \not\stackrel{\mathfrak{o}}{\oplus} \mathbf{H}^{\mathfrak{l}}(\mathbb{P}^{\neg}|\mathcal{Q}) \to \mathbb{C}.\tag{V.2.2}$$

Such pairing can be obtained taking representatives forms , onP +,of type (3, 1), and , on P,of type (0, 1), and integrating the 5-form , onP0. This is equivalent in the language of the homomorphism to

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

**V. 2. Twistor transform to ladder representations** 

isomorphism with cohomology result be SU(2, 2)-equivariant.

,*O*).

twistor operator, being this the twistor transform

necessary the SU(2, 2)-invariance. If H1(P\*

of G-modules H1(P+, 3), and H1(P, *O*),given by

arises in this twistor correspondence.

dual twistor space like H1(P \*

meet.

56

coefficients.

respectively.

which is obvious by the theorem that establishes the existence of positive constant to continuous semi-norm (appendix C), in the study of asymptotic behavior of the matrix

Let P+, and P+, be open orbits of SU(2, 2), on **M** = Gr2(**C**4), and P = CP3 = P(C4), respectively. The dual projective twistor space is P\* = P((C4)\*). The sheaves of holomorphic functions and of holomorphic 3-forms (that is to say, belonging to space 3(S3))), are denoted by *O*, and 3,

The Penrose transform to this case result completely natural and, in particular this

Is clear that twistor description of this representation is one of the most clear than the version of Maxwell fields to effects of induction of G-modules. The action of SU(2, 2), is automatic. Now well, is natural question how the scalar product of the Hermitian form

Indeed, motivated by the classical construction is well acquaintance that this can be feasible by an integral intertwining operator. The cohomology H1(P+, 3), (whereP +, represent the closure of the open orbit P +) give an ascent directly to fields on P +, considering that the description "potential module gauge", of the same fields finds their interpretation on the

Note that P +, is simply connect (contractible really), and have base neighborhood of Stain such that the original description of field and the description "potential modulo gauge",

Thus, the integral intertwining operator, which establish the crucial ingredient in the classical construction is interpreted in the twistor description like a integral intertwining

Considering this operator is easy describe the scalar product in a path in the which is

invariance under conformal transforms of such product, is necessary the multi-linear pairing

H1(P +, 3) H1(P, *O*) C (V. 2.2)

, *O*), (V. 2.1)

, *O*), then H1(P, *O*), and to obtain the

*T* : H1(P+, 3) H1(P\*

$$\mathrm{H^{!}}(\mathbb{P}^{+}\downarrow\Omega^{\flat})\,\,\,\Theta\,\mathrm{H^{!}}(\mathbb{P}^{-}\downarrow\mathcal{O})\to\mathrm{H^{2}}(\mathbb{P}^{0},\Omega^{\flat})\to\mathrm{H^{!}}(\mathbb{P},\Omega^{\flat})\to\mathrm{C},\tag{V.2.3}$$

and by duality in twistor theory, can be deduced the scalar product wanted. Thus *F, G*H1(P+, 3), we define

$$\lhd F, G \rhd = F.\underline{\mathfrak{T}G} \tag{V.2.4}$$

As exercise, consulting [Eastwood and Ginsberg, 1981, 177-196], is possible demonstrate the coincidence of this twistor construction with the classical construction on the massless fields.

The twistor construction has much advantage on the classical construction, being of major relevance the possibility of to be generalized to an ambit of spaces of major dimension, or of structures that are defined under topologies more weakly. Many important representations of reductive Lie groups occurring naturally as a cohomology on a homogeneous space, such is the case, for example, of some induced representations on generalized orbital spaces. In this point, one can have the possibility of to know these representations are unitarizables. To this goal, result useful the work with cohomology, eluding of this manage arguments of the space-time that can hinder the study of the unitarization of the representations.

Nevertheless, we have not that lose of view that we want obtain G-modules that can be classified like £-modules with the goal of obtain all the unitary representations of the integral operators that acts in the general solution of the Maxwell equations.

Thus want understand and generalize so much as be possible the twistor transform *T* : H1(P+, 3) H1(P\* , *O*), like integral operator in electrodynamics of type intertwining to the obtaining of solutions in the space-time modulo the Maxwell fields into of the context of the unitary representations.

To can use the formulation of the twistor transform like a ensemble of classical intertwining integral operators and give their corresponding representations is necessary to check that the Hermitian form originated of the twistor transform let be symmetric. Finally one can verify this on the L-types. This have been did by example to the ladder representations of SU(p, q).

Consider a vector complex space T, of dimension (N + 1), with Hermitian form , with signature (p, q), p + q = N + 1. Consider by simplicity that 2 p q (if p = 1 (or pair if p = 0), the general process is valid but the conclusions are subject to certain little modifications [Eastwood, 1983]). Be G = SU(p, q), the subgroup of SL(N + 1, C) the which preserve . The projective space PN(C), is sliced in, P+, Py P0, under the action of G. Here P+(respectively P ), is the space of lines x **T**, such that x , is positive (respectively negative) defined.

The space P0,sometimes denoted by *PN*, consists of those lines on which , is null. Thus P0, is a real hyper-surface in PN(C). Note that P+, and P , are indicating open subsets like complex manifolds. The fact of that these spaces are the unique open G-orbits do the construction of the appearing simpler.

Let **M**, be the p-dimensional Grassmanian subspaces of **T**. These open and *min*(p, q) + 1, G-orbits in **M**, corresponds to the possible ways in the which , can be reduced like a nondegenerated form on a space of dimension p. We are interested in the orbit **M**+, on whose points is reduced to be positive defined. Note that this orbit **M**+, is a Stein manifold.

As it is demonstrated originally in [Eastwood, 1983] the composition of two Penrose transforms establishes the isomorphism given by the twistor transform to SU(p, q), to know

$$\mathcal{T}: \mathsf{H}^{\neg}(\mathsf{P}^{\ast}, \mathcal{O}(\cdot \mathsf{n} \cdot \mathsf{p})) \to \mathsf{H}^{\neg}(\mathsf{P}^{\ast}, \mathcal{O}(\mathsf{n} \cdot \mathsf{p})),\tag{V.2.5}$$

The sheaf (-k), correspond to the kth-power of the tautological bundle of lines on PN(C). Thus when p = n = 2, (-4) = 3, this meet with the map (V. 2. 1). The demonstration of that (V. 2. 1), and (V. 2. 5), are isomorphisms. For example, when *n* 1, one demonstrate that the left side is isomorphic to fields on **M**+ , satisfying the differential equations analogous to the massless equations of helicity *n*/2, on the Minkowski space. Similarly, the right side is naturally isomorphic to potentials module gauge to the same fields on **M**\* . But **M**\* , is the same like in **M**+, in the isomorphism. Newly one can to extend the isomorphism to the closures of P**+**, and P , given the theorem:

**Theorem V. 2.1.** Exist a map SU(p, q)-equivariant

$$\mathcal{T} \colon \mathbf{H}^{\mathsf{p}^{-}}(\mathbb{P}^{\mathsf{p}} \backslash \mathcal{O}(-\mathsf{u} - \mathsf{p})) \to \mathsf{H}^{\mathsf{q}^{-}}(\mathbb{P}^{\mathsf{p}^{-}} \backslash \mathcal{O}(\mathsf{u} - \mathsf{q})) ,\tag{V.2.6}$$

which is isomorphism.

Proof. [10].

## **V. 3. Radon-Schmid transform to D-modules like induced representations**

We consider the Penrose transformation, [4] through of the correspondence

$$
\bigcup\_{\mathbf{v}} \bigcup\_{\mathbf{w}}^{\mathbf{v}} \bigcup\_{\mathbf{w}}^{\mathbf{w}} \tag{V.3.1}
$$

where F = F1, 2(V), is the manifold of flags of dimension one and two, associated to 4-dimensional complex vector space V. Be P = F1(V), such that F1(V) P3(C), (complex lines inC4), and be M = F2(V), such thatF2(V) G2, 4(C), (Grassmannian manifold of 2-dimensional complex subspaces), with M R<sup>4</sup>RC, where

$$\mathbf{M} = \{ \underline{\mathbf{z}} \in \mathbb{C}^4 \mid \underline{\mathbf{z}} = \langle \mathbf{z} \mathbf{i}, \mathbf{z} \mathbf{z} \rangle, \mathbf{z} \mathbf{i} \}, \\ \forall \ \mathbf{z} = \mathbf{x} + \mathbf{j} \mathbf{y}\_t \ \forall \ \mathbf{x}\_t \ \mathbf{y} \in \mathbb{R} \}, \tag{\text{V. 3.2}}$$

is the 4-dimensional complex compactified Minkowski space [4]. The projections of F, are given for:

$$\mathbf{v}(\mathbf{L}, \mathbf{\!}, \mathbf{L} \mathbf{\!}) = \mathbf{L} \mathbf{\!}, \tag{\text{V}. 3.3}$$

and

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

is a real hyper-surface in PN(C). Note that P+, and P

 *T* : Hp1(P+,*O*(- *n* - p)) Hq1(P\*

construction of the appearing simpler.

closures of P**+**, and P

which is isomorphism.

**representations** 

Proof. [10].

58

, are indicating open subsets like

, *O*(*n* - p)), (V. 2.5)

, *n*q)), (V. 2.6)

(V. 3.1)

. But **M**\*

 , is the

The space P0,sometimes denoted by *PN*, consists of those lines on which , is null. Thus P0,

complex manifolds. The fact of that these spaces are the unique open G-orbits do the

Let **M**, be the p-dimensional Grassmanian subspaces of **T**. These open and *min*(p, q) + 1, G-orbits in **M**, corresponds to the possible ways in the which , can be reduced like a nondegenerated form on a space of dimension p. We are interested in the orbit **M**+, on whose points is reduced to be positive defined. Note that this orbit **M**+, is a Stein manifold.

As it is demonstrated originally in [Eastwood, 1983] the composition of two Penrose transforms establishes the isomorphism given by the twistor transform to SU(p, q), to know

The sheaf (-k), correspond to the kth-power of the tautological bundle of lines on PN(C). Thus when p = n = 2, (-4) = 3, this meet with the map (V. 2. 1). The demonstration of that (V. 2. 1), and (V. 2. 5), are isomorphisms. For example, when *n* 1, one demonstrate that the left side is isomorphic to fields on **M**+ , satisfying the differential equations analogous to the massless equations of helicity *n*/2, on the Minkowski space. Similarly, the right side is

same like in **M**+, in the isomorphism. Newly one can to extend the isomorphism to the

naturally isomorphic to potentials module gauge to the same fields on **M**\*

**V. 3. Radon-Schmid transform to D-modules like induced** 

We consider the Penrose transformation, [4] through of the correspondence

, given the theorem:

 *T* : Hp1(P+,*n*p Hq1(P\*

**Theorem V. 2.1.** Exist a map SU(p, q)-equivariant

$$
\pi(\mathbf{L}, \mathbf{L}, \mathbf{L}) = \mathbf{L}.\tag{V.3.4}
$$

where L1 L2 V, are complex subspaces of dimension one and two, respectively, defined a element (L1, L2), of F, to know

$$\mathbb{F}^{\mathsf{m}} \left\langle \begin{pmatrix} \mathsf{L} \mathsf{\downarrow} , \mathsf{L} \mathsf{\downarrow} \end{pmatrix} \in \mathbb{V} \times \mathsf{V} \ \mid \ \mathsf{L} \mathsf{\downarrow} \subset \mathsf{L} \mathsf{\downarrow} \subset \mathsf{V} \ \mathsf{\downarrow} \ \mathsf{\downarrow} \ \mathsf{L} \ \mathsf{\downarrow} = \mathsf{L} \mathsf{\downarrow} \ \mathsf{\downarrow} \ \mathsf{\uparrow} \ \mathsf{L} \ \mathsf{\downarrow} \ \mathsf{\downarrow} = \mathsf{L} \ \mathsf{\downarrow} \ \rangle \,\tag{V} \ \mathsf{3.5} \right\} $$

If M, is a compactified Minkowski space [4] then

*set of equations of massless fields*dF = 0, dF\* = j, W º = 0, Rij = 0, RijgijR = 0, ,

$$\{\text{UPVM}\}\tag{V.3.6}$$

that is to say, is required the *spectral resolution of complex sheaves* [17], of seated certain class in the *projective space* P, to give solution to the field equations modulus a flat conformally connection [10]

$$O^{\mathbb{P}}\text{.(h)}\to\dots\to O^{\mathbb{M}}\text{.(h)}\to\mathbb{O}^{+1}\text{.(h)}\to\dots\to\mathbb{O}\tag{V.3.7}$$

Let *P*, be the *Penrose transform* [29] *associated to the double fibration* in (1), used to represent the holomorphic solutions of the *generalized wave equation* [4],with parameter of *helicity* h [29]:

$$
\Box \vdash \Phi = 0 \,,\tag{V.3.8}
$$

on some open subsets U M, in terms of *cohomological classes of bundles of lines* [4], on U = ((U)) P(P, is the *super-projective space*).

Is necessary to mention that *these cohomological classes are the conformal classes* that are wanted determine to solve *the phenomenology of the space-time* to diverse interactions studied in *gauge theory* [19], and can construct a general solution of the general cohomological problem of the *space-time*.

With major precision we sign that bundle of lines on P, are given kZ, by the kth-tensor power *O*P(k), of the tautological bundle [19], (is the bundle that serve to explain in the context of the bundles of lines on P, the general bundle of lines of M).

Let h(k) = (1 + k/2), be xP, and letx = (x). Then a result that establish the *equivalences* on the cohomological classes of the bundle of lines on U, and the family of solutions of the equations (V. 3. 8), (equations of the massless fields family on the Minkowski space M, with helicity h) is the given by:

**Theorem V. 3. 1.***(with classic Penrose transform)* Be U M, an open subset such that U x, is connect and simply connect xU. Then k < 0, *the associated morphism to the twistor correspondence* (1); the which map a 1-form on U, to the integral to along of the fibers of, of their inverse image for, induce an isomorphism of cohomological classes:

$$\mathbf{H^1(\underline{U}, O\_\mathbb{P}(k))} \cong \ker(\mathbf{U}, \Box h\_{\mathbb{R}} u\_{\mathbb{R}}),\tag{V.3.9}$$

Part of the object of our research is centred in *the extension of the space of equivalences of the type* (V. 3. 8)*,* under a more general context given through of the language of the *D-modules,*  that is to say, we want *extend our classification of differential operators of the field equations* to the context of the homogeneous bundle of lines and obtain a complete classification of all the differential operators on *the curved analogous of the Minkowski space of* M*.* Thus our moduli space will be those of *the equivalences of the conformal classes given in* (V. 3. 8), in the language of the *algebraic objects D-modules with coefficients in a coherent sheaf.* 

We consider a correspondence

$$
\left\langle \begin{array}{c} \text{v} \\ \text{f} \\ \text{g} \end{array} \right\rangle^{\pi} \qquad \tag{V.3.10}
$$

where all the manifolds are analytic and complex, , and , are proper and (), induces a closed embedding *S* X Y [22]. Be dS = dimC*S*, with d*S*/Y = d*S*dY.

We define the transform of a sheaf *F* on X, (more generally, of a object of the derived category of sheaves) like

$$\mathbf{0} \circ \mathbf{F} = \mathbf{R} \pi \mathbf{v}^{-1} F[\mathbf{d} \mathbf{s} \circ \mathbf{l}] \,\tag{\text{V. 3.11}}$$

and we define the transform of a *D*X-module *M*, like *SM* = D *M*, where , and denotes the direct and inverse images of andrespectively, in the sense of the *D*-modules1 [22], and we consider also

$$\Phi \underline{\mathbf{s}} G = \mathbf{R} \mathbf{v} \pi^{-1} G \mathbf{l} \mathbf{d} \boldsymbol{s} \boldsymbol{\chi} \text{.}\tag{\text{V. 3.12}}$$

To a sheaf *G*, on Y. Then is had the formula

$$\mathfrak{ospR}\,\mathrm{Homno}\_{\mathbb{V}}(\mathcal{M},\mathrm{O}\chi) = \mathrm{R}\,\mathrm{Homno}\_{\mathbb{V}}(\underline{\mathrm{@s}}\mathcal{M},\mathrm{O}\chi) \tag{V.3.13}$$

of the which it is deduce the formula, to *G*, the sheaf on Y, (coherent sheaf):

R(X; RHomDX(*MSG* , *O*X)[dX] R(Y; RHomDY (*sM G*, *O*Y)[dY], (V. 3. 14)

*This define a categoric equivalence of the transformation* (V. 3. 1), in the context of the *right derived D*-*modules*, Db(*M*Rqc(D )), because is necessary to give a equivalence with a sub-category of the *right D*-modules that have support in Y, to of that form guarantee the inverse image of *P*, and with it to obtain a image of closed range of the operator *P*, conformed their uniquely [21, 22, 26].

*Theorem of Kashiwara* **V. 3.2.** Let i : Y � X, be of the correspondence (V. 3. 10), a closed immersion. Then the direct image functor i+, is an equivalence of MRqc(Di ), with full sub-category of MRqc(D), consisting of modules with support in Y.

*Proof*: [46].

(V. 3.10)

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

context of the bundles of lines on P, the general bundle of lines of M).

their inverse image for, induce an isomorphism of cohomological classes:

of the *algebraic objects D-modules with coefficients in a coherent sheaf.* 

We consider a correspondence

category of sheaves) like

Minkowski space M, with helicity h) is the given by:

60

With major precision we sign that bundle of lines on P, are given kZ, by the kth-tensor power *O*P(k), of the tautological bundle [19], (is the bundle that serve to explain in the

Let h(k) = (1 + k/2), be xP, and letx = (x). Then a result that establish the *equivalences* on the cohomological classes of the bundle of lines on U, and the family of solutions of the equations (V. 3. 8), (equations of the massless fields family on the

**Theorem V. 3. 1.***(with classic Penrose transform)* Be U M, an open subset such that U x, is connect and simply connect xU. Then k < 0, *the associated morphism to the twistor correspondence* (1); the which map a 1-form on U, to the integral to along of the fibers of, of

H1(U, *O*P(k)) ker(U, h(k)), (V. 3.9)

Part of the object of our research is centred in *the extension of the space of equivalences of the type* (V. 3. 8)*,* under a more general context given through of the language of the *D-modules,*  that is to say, we want *extend our classification of differential operators of the field equations* to the context of the homogeneous bundle of lines and obtain a complete classification of all the differential operators on *the curved analogous of the Minkowski space of* M*.* Thus our moduli space will be those of *the equivalences of the conformal classes given in* (V. 3. 8), in the language

where all the manifolds are analytic and complex, , and , are proper and (), induces a

We define the transform of a sheaf *F* on X, (more generally, of a object of the derived

*SF* = R*F*[d*S*/Y], (V. 3.11)

closed embedding *S* X Y [22]. Be dS = dimC*S*, with d*S*/Y = d*S*dY.

This equivalence preserves *coherency* and *holonomicity* [17, 19]. Then preserve *conformability*  in M[19].

Let's formulate in the language of the *D*-Modules and its sheaves, like was given in a resolution (V. 3. 7); the correspondence between the space of coherent *D*-Modules and the massless field equations space of the Penrose transform.

$$\operatorname{Rv}\_{+}(\mathcal{V}) = \operatorname{Rv}\_{\operatorname{V}}(\mathcal{V} \otimes^{\operatorname{n}} \operatorname{v} \operatorname{D}\_{Y \to^{\mathcal{X}}}),$$

where *V*,is the *characteristic manifold* and R, is the *right derived functor* following:

$$
\mathbb{R} : \mathcal{M}^{\kappa}{}\_{\mathbb{R}}(\mathbb{D}^{\circ}) \to \mathcal{M}\_{\mathbb{R}}(\mathbb{D}).
$$

<sup>1</sup> To define the images of direct functors to *D*-modules is necessary use *derived categories*. For it, is simpler defining them for *right D-modules*. Let Db[MRqc(D�)], be the derived category bounded for right quasi-coherent *D* -modules of the form

Also r : Y X, and DY X = \*(D) = *O*Y(*OX*)D. Then DY X, is a right*D*- module to the right multiplication in the second functor.

But for it, is necessary to include a result that establishes the regularity in the analytical sense of the Riemannian manifold, which shapes the space - time, and that allows the application of the involutive distribution theorem on integral submanifolds as solutions of the corresponding massless field equations on G2, 4(C).

By the *Kashiwara* (*theorem V. 3. 1*), and some results of Oshima on involutive manifolds [46], we can characterize certain spaces to the regularity of the images of *P*, in *D*-modules. These spaces are the induced representations that we want exhibite, such as Schmid demonstrate it [47].

### **V. 4. Penrose transform to locally compact representations**

The roll of the Borel sub-algebras and the corresponding Cartan subalgebras are equivalents in different context, the first uses in complex analytic manifolds to characterize the points of a coherent sheaf of differential operators of the corresponding flag manifolds of a holomorphic complex bundle (bundle of Borel subalgebras), with cohomology group of dimension cero (theorem of Borel-Weil). The global sections in this coherent sheaf denotated for

$$
\Gamma(\mathcal{X}, \mathfrak{o}(\lambda)) = \mathsf{H}^{\mathfrak{p}}(\mathcal{X}, \mathfrak{o}(\lambda)),
\tag{V.4.1}
$$

conform an irreducible G-module of finite dimension with maximal weight , and Flag manifolds X, (which is a projective manifold).

For other side, the Cartan subalgebras are the unique global constant sections of the holomorphic complex bundle modulus their nilpotent radical bundle of Borel subalgebras, which are points of a Lie algebra of the general linear group subjacent in the reductive Lie group G. In others words the holomorphic bundle modulus a radical bundle result be a trivial vector fibred bundle on X, and is a subbundle of the bundle of Borel subalgebras, thus result natural to think that using certain structure more fine of the flag manifolds like for example, the given for open orbits of flag manifold and the continuous homomorphism between said open orbits can obtain an extension of the classification of the irreducible representations that exist in the space H0(X, o()), and that under the association of irreducible minimal K-types is obtained a class more wide of irreducible unitary representations classified for the theory of Langlands.

One of the important theorems to the obtaining of induced representations (with canonical P-pairs (P, A), with Levi decomposition P = MN = 0MAN, P = MN, and N = (N), (P = (P)), like (m, KM)-subquotients of V/nV, is the theorem of Casselman, the which can be generalized to quotients of the form

Chapter V – Integral Transforms on Lie Groups and Lie Algebras 63

$$\mathbf{V}\_{u0} = \mathbf{V}\_{\boldsymbol{\eta}} / ((\mathbf{V}\_{\boldsymbol{\eta}})(\mathbf{\tilde{u}})) ,\tag{\text{V}. 4.2}$$

on Frèchet spaces. Then in the structure of (g, K)-module of V, there is a structure of (g, K) module of V, there is a structure of subquotient (m0, K M)-module of Vm0, where in turn, it there is an analytic structure (C), of Frèchet space of K-finite vectors such that V, is finitely generated and admissible like (g, K)-module. Then exist a continuous homomorphism between V, and the Langlands subrepresentation of data (P, , ), belonging to space

$$\operatorname{Hom}\_{\mathbb{L}}\kappa(\operatorname{V}, \operatorname{I}\_{\mathbb{V}\circ\operatorname{\mathcal{V}}\circ\operatorname{V}}) \neq \operatorname{0},\tag{\operatorname{V}.4.3}$$

Using a cohomology with respect to the representations of principal series we can characterize an isomorphism between 0M-modules like the obtained for the Casselman subrepresentations2 and given for the cohomology group

$$\mathbf{H}(\mathfrak{l}\mathfrak{n}\_{\mathsf{L}},\mathbf{E}\_{\mathsf{k}}) = \bigoplus\_{\mathsf{c}\_{\mathsf{L}}:\mathsf{l}\mathfrak{l}\mathfrak{P},\ \mathbb{l}(\mathsf{s})=\mathsf{l}} \mathbf{E}\_{\mathsf{s}(\mathsf{c}\_{\mathsf{L}}+\mathsf{l}\mathfrak{l})-\mathsf{l}\mathsf{s}} \tag{\mathsf{V}.4.4}$$

Where the weights s() , are all distinct and dominating. Given that s, changes in all the range of WP, then the decomposition of H\*(n, F), like MC-module, is of free multiplicity.

If we consider a nilpotent radical Lie algebra u, of a -stable parabolic subalgebra q, such that also to a subalgebra

$$\mathfrak{l} = \langle \mathbb{X} \in \mathfrak{g} \mid [\mathbb{X}, \mathbb{H}] = 0 \rangle,\tag{\text{V. 4.5}}$$

it is satisfy that

$$\mathfrak{q} = \mathfrak{l} \oplus \mathfrak{u},\tag{V.4.6}$$

with

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

**V. 4. Penrose transform to locally compact representations** 

the corresponding massless field equations on G2, 4(C).

manifolds X, (which is a projective manifold).

representations classified for the theory of Langlands.

generalized to quotients of the form

62

demonstrate it [47].

denotated for

But for it, is necessary to include a result that establishes the regularity in the analytical sense of the Riemannian manifold, which shapes the space - time, and that allows the application of the involutive distribution theorem on integral submanifolds as solutions of

By the *Kashiwara* (*theorem V. 3. 1*), and some results of Oshima on involutive manifolds [46], we can characterize certain spaces to the regularity of the images of *P*, in *D*-modules. These spaces are the induced representations that we want exhibite, such as Schmid

The roll of the Borel sub-algebras and the corresponding Cartan subalgebras are equivalents in different context, the first uses in complex analytic manifolds to characterize the points of a coherent sheaf of differential operators of the corresponding flag manifolds of a holomorphic complex bundle (bundle of Borel subalgebras), with cohomology group of dimension cero (theorem of Borel-Weil). The global sections in this coherent sheaf

(X, o()) = H0(X, o()), (V. 4.1)

conform an irreducible G-module of finite dimension with maximal weight , and Flag

For other side, the Cartan subalgebras are the unique global constant sections of the holomorphic complex bundle modulus their nilpotent radical bundle of Borel subalgebras, which are points of a Lie algebra of the general linear group subjacent in the reductive Lie group G. In others words the holomorphic bundle modulus a radical bundle result be a trivial vector fibred bundle on X, and is a subbundle of the bundle of Borel subalgebras, thus result natural to think that using certain structure more fine of the flag manifolds like for example, the given for open orbits of flag manifold and the continuous homomorphism between said open orbits can obtain an extension of the classification of the irreducible representations that exist in the space H0(X, o()), and that under the association of irreducible minimal K-types is obtained a class more wide of irreducible unitary

One of the important theorems to the obtaining of induced representations (with canonical P-pairs (P, A), with Levi decomposition P = MN = 0MAN, P = MN, and N = (N), (P = (P)), like (m, KM)-subquotients of V/nV, is the theorem of Casselman, the which can be

$$\mathfrak{g} = \bigoplus\_{\alpha \in \Lambda(\mathfrak{b} \cdot \mathfrak{d}) \times \alpha(\mathfrak{h}) < 0} \mathfrak{g}^{\alpha} \quad \underline{\mathfrak{h}} = \bigoplus\_{\alpha \in \Lambda(\mathfrak{b} \cdot \mathfrak{d}) \times \alpha(\mathfrak{h}) > 0} \mathfrak{g}^{\alpha} \tag{\text{V. 4.7}}$$

is possible to built to a connect subgroup an u-cohomology of Dolbeault to the complex context of the space of the generalized D-modules on q, whose class space G/Q, induce their structures on the open G0-orbits D, then the generalized flag manifold given for G0/L0, where L0, is the centralizator of a compact torus T, (the which is reductive and connect like compact subgroup L), is the space of Borel subalgebras, which will include all the representations of

H\*(g, K; I Ls) = H\*(m, K M; H Ls) a\*<sup>C</sup>

<sup>2</sup> For the cohomology theorem of the principal series exist a representation of finite dimension L, of M, such that

with a variation grade of *l*(s). Then the first factor of M, have concentrate cohomology in the interval [q0(0 M), q0(0M) + *l*0(0M)]. Ls, is a representation of a connect reductive group.

elliptic type3 in G, (the Casimir operator act for 0 on the (L, l)-module of finite dimension Es() ), and extensions of the fundamental series.

This can pre-write it intrinsically through of the sheaf of holomorphic functions o. Using the u-specialization of the Dolbeault cohomology4 and considering L a connect component of the group G, corresponding to Lie algebra l, we can define the relative complex cohomology of Lie algebras given to the particular case L0, like the space Hom(( u, C(G0) V)L0, always with the defined differentials for

$$\begin{split} df(\mathbf{X}\_{1}\wedge...\wedge\mathbf{x}\_{n}) &= \sum\_{k} (-1)^{k+1} \gamma(\mathbf{X}\_{k}) \Big\{ f(\mathbf{X}\_{1}\wedge\mathbf{X}\_{2}\wedge...\wedge\overline{\mathbf{X}}\_{k}\wedge...\mathbf{X}\_{n}) \Big\} + \\ &+ \sum\_{r$$

where Xku, k = 1, , , and fHom(( u, C(G0) V))L0, is the action of u, on C(G0) V, through of the right translation r . Extending this cohomology on all the component L then is necessary consider a space of infinite dimension representation, which must use the cohomological space of representations Hj (G/L, o(E)), to obtain the classification on the opened L- orbit on the flag manifold of Borel subalgebras of l = q<sup>r</sup> , and the corresponding location through the Zuckerman functor Aj (G, L, q; ) of the fundamental series (recovery the classic series or of Harish-Chandra) inside of the space of induced irreducible unitary representations. Then from it, we can to propose a theorem using the -cohomology of Schmid and the possibility of generalize the D-modules on an extension of continuous homomorphism between open orbits of a complex holomorphic bundle G/L E, and by this way to give an u-implicit classification with the corresponding Langlands data for the irreducible representations. Nevertheless the following problems arise: we can give a clear definition of the topology of the Dolbeault complex for this case and the problem of closed range of the operator -cohomology with the same infinitesimal character also of a vanishing theorem to the anti-dominating case. After we could give an analogous result to the case of infinite dimension considering representations on measurable orbits of flag manifolds.

Newly, considering the Casselman subrepresentations and let (, H), be a differentiable admissible Fréchet 0M-module with infinitesimal character , and let a\***C**. Then the induced representation (P, , IP, ), is the representation defined by the right translations on the defined (g, K)-module as the space

A (V) = HomL0 K0( u, C(G0) V)

where V is a complex g-module.

<sup>3</sup> All the canonical representations of L, whose character have support on an elliptic space in G. This subspace is a fundamental series representation class. The elliptic set in G, is the union of the G-conjugated of T G', where H A, is a fundamental Cartan subgroup, T = H G, and G', a regular set.

<sup>4</sup> **Def. V. 1.** A u-specialization of the Dolbeault cohomology is the relative complex Lie algebras given for the space

$$\mathbf{I} \cdot\_{\sigma, \mathbf{v}} = \langle \mathbf{f} \in \mathbf{C}^n(\mathbf{G}, \mathbf{H}\_\sigma) \mathbf{f} \rangle \mathbf{(mang)} = \mathbf{a}^{\otimes p + \mathbf{v}} \sigma(\mathbf{m}) \mathbf{f}(\mathbf{g}), \forall \; \mathbf{a} \in \mathbf{A}, \; \mathbf{g} \in \mathbf{G}, \; \mathbf{n} \in \mathbf{N} \text{ and } \mathbf{m} \in \mathbf{M} \rangle, \tag{V.4.9}$$

where using the Casselman notation

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

of Lie algebras given to the particular case L0, like the space Hom((

*k*

opened L- orbit on the flag manifold of Borel subalgebras of l = q<sup>r</sup>

1 1 2

*k*

1 2

Es() ), and extensions of the fundamental series.

*r s*

with the defined differentials for

*r s*

manifolds.

where Xku, k = 1, , , and fHom((

cohomological space of representations Hj

location through the Zuckerman functor Aj

on the defined (g, K)-module as the space

where V is a complex g-module.

is a fundamental Cartan subgroup, T = H G, and G', a regular set.

A

64

<sup>1</sup>

(V. 4.8)

u, C(G0) V))L0, is the action of u, on C(G0) V,

(G/L, o(E)), to obtain the classification on the

(G, L, q; ) of the fundamental series (recovery

u, C(G0) V)L0, always

, and the corresponding

elliptic type3 in G, (the Casimir operator act for 0 on the (L, l)-module of finite dimension

This can pre-write it intrinsically through of the sheaf of holomorphic functions o. Using the u-specialization of the Dolbeault cohomology4 and considering L a connect component of the group G, corresponding to Lie algebra l, we can define the relative complex cohomology

( ) ( 1) ( ) ( )

*df X x X fX X X X*

*n k k n*

*r s rsn*

( 1) ([ , ] )

through of the right translation r . Extending this cohomology on all the component L then is necessary consider a space of infinite dimension representation, which must use the

the classic series or of Harish-Chandra) inside of the space of induced irreducible unitary representations. Then from it, we can to propose a theorem using the -cohomology of Schmid and the possibility of generalize the D-modules on an extension of continuous homomorphism between open orbits of a complex holomorphic bundle G/L E, and by this way to give an u-implicit classification with the corresponding Langlands data for the irreducible representations. Nevertheless the following problems arise: we can give a clear definition of the topology of the Dolbeault complex for this case and the problem of closed range of the operator -cohomology with the same infinitesimal character also of a vanishing theorem to the anti-dominating case. After we could give an analogous result to the case of infinite dimension considering representations on measurable orbits of flag

Newly, considering the Casselman subrepresentations and let (, H), be a differentiable admissible Fréchet 0M-module with infinitesimal character , and let a\***C**. Then the induced representation (P, , IP, ), is the representation defined by the right translations

3 All the canonical representations of L, whose character have support on an elliptic space in G. This subspace is a fundamental series representation class. The elliptic set in G, is the union of the G-conjugated of T G', where H A,

<sup>4</sup> **Def. V. 1.** A u-specialization of the Dolbeault cohomology is the relative complex Lie algebras given for the space

u, C(G0) V)

(V) = HomL0 K0(

*fXX X X X X X*

$$\mathbf{I}\_{\mathbf{P},\sigma,\mathbf{v}} = \mathbf{Ind}\nu^{\mathbf{G}}(\mathbf{H}\_{\sigma} \otimes \mathbf{C}\_{\mathbf{p}^{\mathbf{P}}+\mathbf{v}}),\tag{\text{V. 4.10}}$$

Then by the cohomology theorem of the principal series [], is clear that h\***C**, a dominate weight and F, a simple G**C**-module with parameter , some of the induced representations of some tempered representations are the given by the cohomological space of dimension qN,

$$\text{H}\mathfrak{q}(\mathfrak{g}, \mathbf{K}; \mathbf{I} \otimes \mathbf{F}\_{\lambda}) = 0,\tag{V.4.11}$$

 q[q0(0M) + l(s), q0(0M) + l(s) + l0(0M) + dimA]. Then using the defined isomorphism by the cohomology of the 0M-modules (like the obtained by the Casselman subrepresentations given in (5. 4. 4)) and their corresponding u-specialization we have the u-specialization given by the space

$$\text{HI}(\mathbf{u}, \mathbf{F}) = \bigoplus\_{\mathbf{s}\_{\mathbf{u}} \in \mathrm{WT}, \, l(\mathbf{s}) = 1} \mathbf{E}\_{\mathbf{s}(\mathbf{s} + \lambda) - \mathbf{p} \prime} \tag{V. 4.12}$$

where F, is a irreducible g-module of finite dimension with maximum weight , and prescribed like l-module under this orthogonal composition. Then a possible generalization of the induced representations for (V. 4. 12), (in the sense of classify representations more fine inside of a global context ((g, K)-*cohomology to the classification of a major class of fine representations*)) can be obtained using the Osborne lemma and the theorem of Casselman subrepresentation given by the identification

$$\mathbf{U(u^{-})} \mathbf{F} \mathbf{U(g)}^{\mathrm{rel}} \mathbf{U(u)} = \mathbf{U(gx)},\tag{V.4.13}$$

where if j = n = *dim*u, then in particular is had that

$$\mathbf{H}^{\rm n}(\mathbf{u}, \mathbf{F}) = \mathsf{P}\mathbf{u}^{\rm n} \otimes \mathbf{F}/\mathsf{u}\mathbf{F},\tag{V.4.14}$$

Now, the advantage of construct an extension of the space F, to the case of infinite dimension like in (V. 4. 13), reduce the evaluations followed topass of the Vogan-Zuckerman scheme to the D-modules scheme inside of the orbital context on all L. Also result clear through of this extension the followed way to prove the theorem of (g, K)-cohomology (Vogan-Zuckerman), through of the induction of the u-cohomology to all the reductive algebra g. Likewise, if q = b, and considering the equivalence between the categories of U(g) modules with infinitesimal character with the category of the D-modules on the flag manifolds of g, is possible deduce an analogous extension on the coherent sheaves space of the form

$$\mathcal{U}l(\mathfrak{n})\mathcal{F}\_{\lambda}\mathcal{Z}(\mathfrak{g})^{\mathfrak{n}}\mathcal{U}l(\mathfrak{h})=\mathcal{U}l(\mathfrak{g}),\tag{V.4.15}$$

where F, is the irreducible G-module of finite dimension belonging to an associative sheaf *D*, with maximum weight .

As establish Vogan in the study of the unipotent representations and cohomological induction of these representations, we can consider an irreducible unitary representation class of homogeneous space G/L, with Levi subgroup L, and determine through of a functorial process on (l, L K)-modules of finite length, the induction to (g, L K)-modules of the form

$$\mathbf{W} \widetilde{\mathbf{u}} \equiv \operatorname{Ind}\_{\mathbf{d}} \mathbb{L}\_{\square} \kappa \mathbf{G} / \mathbb{L}\_{\succ} \tag{V.4.16}$$

and in all this process see the preservation of the Hermitian form of the Zuckerman functors on Wg. But this method is an orbital version of the construct of representations through of parabolic induction (which use n-cohomology to the obtaining of unitarizable irreducible modules) considering hypothesis on the polarization q. Nevertheless, this restriction not establish a criteria of classification to the very infinitesimal numbering characters that appears and their relation with the co-adjunct nilpotent orbits, thus is necessary the application of the minimal K-types of Salamanca-Vogan. This via can be the algebraic way to the solution of the problem of classification of all the irreducible unitary representations. Nevertheless only can get us partial solutions, stayed several aisled representations in this big class of representations.

Is clear that when G, is non-compact and L, is compact the representations are the identified by the Langlands conjecture, which was proved by Schmid. To L, non-compact the analytic situation is no clear and to it, the via that is used is consider only the cocycles C0, s(K/(K L)) (*Space of the* (0, s)-*forms where* K/K L, *are the* s-*dimensional complex compact submanifolds of* G/L), of the space H0, s(K/(K L), C2(u)), that is an u-specialization of the cohomological space of the Borel-Weil theorem, H0, q(G/L, V), and we determine the Penrose transform on said cocycles. But the identification of irreducible unitary representations with , appropriate result very difficult, since as is told above of this exposition is not result clear that the co-boundary operator , on the images of the Dolbeault complexes under the Penrose transform belonging to space *P*(H0, s(K/(K L), C2(u))) be of closed range.

## **Chapter VI**

## **Intertwinning Operators**

### **VI. 1.**

Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

66

*D*, with maximum weight .

big class of representations.

of the form

where F, is the irreducible G-module of finite dimension belonging to an associative sheaf

As establish Vogan in the study of the unipotent representations and cohomological induction of these representations, we can consider an irreducible unitary representation class of homogeneous space G/L, with Levi subgroup L, and determine through of a functorial process on (l, L K)-modules of finite length, the induction to (g, L K)-modules

Wg Indg, L KG/L, (V. 4.16)

and in all this process see the preservation of the Hermitian form of the Zuckerman functors on Wg. But this method is an orbital version of the construct of representations through of parabolic induction (which use n-cohomology to the obtaining of unitarizable irreducible modules) considering hypothesis on the polarization q. Nevertheless, this restriction not establish a criteria of classification to the very infinitesimal numbering characters that appears and their relation with the co-adjunct nilpotent orbits, thus is necessary the application of the minimal K-types of Salamanca-Vogan. This via can be the algebraic way to the solution of the problem of classification of all the irreducible unitary representations. Nevertheless only can get us partial solutions, stayed several aisled representations in this

Is clear that when G, is non-compact and L, is compact the representations are the identified by the Langlands conjecture, which was proved by Schmid. To L, non-compact the analytic situation is no clear and to it, the via that is used is consider only the cocycles C0, s(K/(K L)) (*Space of the* (0, s)-*forms where* K/K L, *are the* s-*dimensional complex compact submanifolds of* G/L), of the space H0, s(K/(K L), C2(u)), that is an u-specialization of the cohomological space of the Borel-Weil theorem, H0, q(G/L, V), and we determine the Penrose transform on said cocycles. But the identification of irreducible unitary representations with , appropriate result very difficult, since as is told above of this exposition is not result clear that the co-boundary operator , on the images of the Dolbeault complexes under the

Penrose transform belonging to space *P*(H0, s(K/(K L), C2(u))) be of closed range.

In the following exposition will establishes some consequent properties of the behavior of the corresponding principal series to the temperate representation theory. Thus of it, will studies some integral identities such that

$$\lim\_{\mathsf{h}\to\mathsf{x}}\mathsf{a}^{\mathsf{u}+\mathsf{p}(\mathsf{h})}\mathsf{c}\mathsf{x}(\mathsf{exp}(\mathsf{th})\mathsf{f},\mathsf{g}) \rhd = \int\_{\mathsf{\underline{\mathsf{NF}}}\mathsf{f}(\mathsf{\underline{\mathsf{n}}})} \mathsf{g}(1)\mathsf{x}\mathsf{d}\mathsf{\underline{n}}\,\,\,\forall\mathsf{\underline{\mathsf{n}}}\mathsf{e}\,\,\mathsf{\underline{\mathsf{N}}}\mathsf{x}$$

that will use to establish an important criteria to the Langlands classification of the tempered (g, K)-modules belonging to the class L2(G), on a real reductive group G.

Consider F 0, and let (PF, AF), be corresponding canonical parabolic pair. We fix to (, H), like a representation of 0MF, such that satisfies the weak inequality. Let NF = (NF), KF = K MF, and we define explicitly to the root system

$$\Phi(\text{P}\mathbb{F}\_{\prime}, \text{A}\mathbb{F}) = \{\alpha \in \Phi(\text{P}, \text{A}) \mid \alpha|\mu = 0\},\tag{\text{VI. 1}}$$

that is equivalent to say that the root system (PF, AF), denotes the root set of aF, on nF.

**Lemma VI. 1.** Let aFC\*, such that Re() > 0, to any PF, AF

1. If f(H), and if w(H)K, then

$$\mathsf{h}\_{\mathsf{ZF}} \mathsf{|}\mathsf{cf}(\underline{\mathsf{n}}) \text{, } \mathsf{w} \succeq \mathsf{|}\mathsf{dq} \mathsf{x} \mathsf{c} \mathsf{o} \tag{\mathsf{VI}.2}$$

with major reason the map

$$\text{if} \quad | \rightarrow \rangle\_{\underline{\text{NF}}} \mid \forall \langle \underline{\text{n}} \rangle \text{ w} \succeq | \mathsf{d} \underline{\text{n}} \tag{\text{VI.3}}$$

is continuous on (H).

2. If w(H)K, is not null then exist fIPF, , such thatNF<f(n), w>dn, is not null. *Proof*. Consider nNF, then f(n) = f(nmF(n)aF(n)kF(n)). Then

$$\mathbf{f}(\mathbf{\underline{n}}\mathbf{m}\mathbf{\underline{r}}(\underline{n})\mathbf{a}\mathbf{\underline{r}}(\underline{n})\mathbf{k}\mathbf{\underline{r}}(\underline{n})) = \mathbf{a}\mathbf{\underline{r}}(\underline{n})^{\mu+\rho\underline{r}}\mathbf{a}(\mathbf{m}\mathbf{\underline{r}}(\underline{n})\mathbf{f}(\mathbf{k}\mathbf{\underline{r}}(\underline{n})) = \mathbf{f}(\underline{n})\mathbf{\underline{r}}$$

Then

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

$$\int\_{\square \overline{\mathbb{F}} \mid \leqslant \mathbf{f}(\underline{\mathbf{n}})} \, \le \, \mathsf{w} \, \mathsf{|}\mathsf{d}\underline{\mathbf{n}} = \int\_{\square \overline{\mathbb{F}} \, \mathsf{a} \mathsf{F}(\underline{\mathbf{n}})} \mathsf{a}^{\boxtimes \mathsf{a} + \mathsf{p}} |\mathsf{o}\langle \mathsf{m} \mathsf{r}(\underline{\mathbf{n}}) \mathsf{f}(\mathsf{k} \mathsf{r}(\underline{\mathbf{n}})), \mathsf{w} \rangle |\mathsf{d}\underline{\mathbf{n}} \rangle$$

But

$$\leq \varpi(\mathsf{m}\models(\underline{\mathsf{m}})\mathsf{f}(\mathsf{k}\models(\underline{\mathsf{n}})),\mathsf{w}\succeq\mathsf{z})\mathsf{f}(\mathsf{k}\models(\underline{\mathsf{n}})\mathsf{f}(1+\log||\mathsf{m}\models(\underline{\mathsf{n}})||)^{\mathsf{x}}\mathsf{Xi}\mathsf{r}(\mathsf{m}\models\underline{\mathsf{n}})) $$

then

$$\int\_{\Delta \mathsf{F}} |\mathsf{xf}(\underline{\mathsf{n}})| \, \mathsf{w} \rhd |\mathsf{d}\underline{\mathsf{n}} \leq \mathsf{l}\_{\mathsf{F}} \, \mathsf{j}(\mathsf{f}(\mathsf{k}\mathsf{r}(\underline{\mathsf{n}})) \mathsf{a}\mathsf{r}(\underline{\mathsf{n}})^{\mu+\rho} (1+\log \|\mathsf{m}\mathsf{r}(\underline{\mathsf{n}})\|)^{\nu} \mathsf{E}\mathsf{r}(\mathsf{m}\mathsf{r}(\underline{\mathsf{n}})) \, \mathsf{d}\underline{\mathsf{n}} \leq \mathsf{l}\_{\mathsf{F}} \, \mathsf{j}(\mathsf{l}\mathsf{r}(\underline{\mathsf{n}})) \, \mathsf{d}\underline{\mathsf{n}} \leq \mathsf{l}\_{\mathsf{F}} \, \mathsf{j}(\mathsf{l}\mathsf{r}(\underline{\mathsf{n}})) \, \mathsf{d}\underline{\mathsf{n}} \leq \mathsf{l}\_{\mathsf{F}} \, \mathsf{j}(\mathsf{l}\mathsf{r}(\underline{\mathsf{n}})) \, \mathsf{d}\underline{\mathsf{n}} \leq \mathsf{l}\_{\mathsf{F}} \, \mathsf{j}(\mathsf{l}\mathsf{r}(\underline{\mathsf{n}})) \, \mathsf{d}\underline{\mathsf{n}} \leq \mathsf{l}\_{\mathsf{F}} \, \mathsf{j}(\mathsf{l}\mathsf{r}(\underline{\mathsf{n}})) \, \mathsf{d}\underline{\mathsf{n}} \leq \mathsf{l}\_{\mathsf{F}} \, \mathsf{j}(\mathsf{l}\mathsf{r}(\underline{\mathsf{n}})) \, \mathsf{d}\underline{\mathsf{n}} \leq \mathsf{j}(\mathsf{l}) \, \mathsf{d}\underline{\mathsf{n}} \, \mathsf{$$

with , a continuous semi-norm on the space (H) , where has used the weak inequality. We define (f) = supk<sup>K</sup>(f(k)). Now, by a lemma of asymptotic boundess (consult [Bulnes, F, 2002]), we have

$$\text{au(\underline{n})}\_{\text{R}^{\text{R}}\mathbb{H}^{\text{s}}\text{-}\mathbb{L}} \leq \text{C}^{\text{q}}\_{\text{l}} (1 - \text{b}(\text{log}\text{ar}(\underline{n})) \text{-}^{\text{q}} \not\models \text{ d} \not\geq 0) \tag{\text{V. 4}}$$

Then

$$
\mathbb{P}(\mathsf{f}(\mathsf{kF}(\mathsf{\underline{n}})\mathsf{a}\mathsf{\underline{r}}(\mathsf{\underline{n}})\mathsf{a}\mathsf{\underline{r}}(\mathsf{\underline{n}})^{\mathsf{n}+\mathsf{p}}(1+\log||\mathsf{m}\mathsf{\underline{r}}(\mathsf{\underline{n}})||)^{\mathsf{c}}\mathsf{\underline{r}}\mathsf{F}(\mathsf{m}\mathsf{\underline{r}}(\mathsf{\underline{n}})) < \epsilon
$$

$$
< \mathsf{C}\_{\mathsf{q}}\mathsf{\underline{r}}(\mathsf{f})\mathsf{a}\mathsf{\underline{r}}(\mathsf{\underline{n}})\mathsf{\underline{r}}\mathsf{F}(\mathsf{m}\mathsf{\underline{r}}(\mathsf{\underline{n}}))(1-\mathsf{p}\mathsf{log}\mathsf{a}\mathsf{\underline{r}}(\mathsf{\underline{n}}))^{\mathsf{d}}.
$$

q > 0. Then given that

$$\int\_{\underline{\lambda}\underline{\mathsf{F}}} \mathsf{C}\_{\mathsf{q}} \underline{\chi}(\underline{\mathsf{f}}) \mathsf{a} \mathsf{F}(\underline{\mathsf{n}}) \Xi \mathsf{F}(\mathsf{m} \mathsf{F}(\underline{\mathsf{n}})) (1 - \mathsf{p} \mathsf{log} \mathsf{a} \mathsf{r}(\underline{\mathsf{n}}))^{-\mathsf{q}} d \mathsf{n} \leq \infty,$$

Then with major reason NF<f(n), w>dn<. Then stay demonstrated (1). To demonstrate that the map defined by the correspondences

$$\mathfrak{f} \mid \rightarrow \begin{cases} \\ \mathsf{u} \mathsf{F} \mid \leq \mathsf{f}(\underline{\mathsf{n}}) \text{, } \mathsf{w} \rhd \mathsf{|d} \underline{\mathsf{n}} \end{cases}$$

is continuous on the space (H), is only necessary see that NF<f(n), w>dnNF(f(kF(n))aF(n)(1 + logmF(n))r F(mF(n))dn, where

$$\mathbf{f(\underline{n}) = ar(\underline{n})^{\mu\_+ + \rho\_\mathbf{F}} \sigma(\mathbf{m} \mathbf{\underline{r}(\underline{n})}) f(\mathbf{k} \mathbf{\underline{r}(\underline{n})})},$$

that is to say, the topology of f(n), is the induced for the semi-norms (mF(n)), on 0MF, in the decomposition of Langlands 0MAN.

In particular, all p-canonical pair has an infinitesimal equivalent to (, (H) ) induced to a representation on (H). Then the induced topology is the semi-norms topology in (H). Thus the map given by (VI. 3), is continuous.

We consider now a partition of the unity in NF, that is to say, a family of functions {Hi}i<sup>I</sup> Cc(NF), such that

$$\mathbf{H(\underline{n})} = \begin{vmatrix} 1, & \text{if } & \underline{n} \in \mathbb{N}\_{\rm F} \\ 0, & \text{if } & \underline{n} \notin \mathbb{N}\_{\rm F} \end{vmatrix} \tag{\text{VI. 5}}$$

H(n)Cc(NF). Since

68 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

with , a continuous semi-norm on the space (H)

But

then

Then

2002]), we have

q > 0. Then given that

the map defined by the correspondences

w>dnNF(f(kF(n))aF(n)(1 + logmF(n))r

Thus the map given by (VI. 3), is continuous.

decomposition of Langlands 0MAN.

NF<f(n), w>dn = NFaF(n)Re(mF(n)f(kF(n)), w>dn,

define (f) = supk<sup>K</sup>(f(k)). Now, by a lemma of asymptotic boundess (consult [Bulnes, F,

aF(n)Re Cq(1 (logaF(n))q, q > 0, (VI. 4)

< Cq(f)aF(n)F(mF(n))(1 logaF(n))q,

NF Cq(f)aF(n)F(mF(n))(1 logaF(n))qdn<,

Then with major reason NF<f(n), w>dn<. Then stay demonstrated (1). To demonstrate that

<sup>f</sup>∣NF<f(n), w>dn,

is continuous on the space (H), is only necessary see that NF<f(n),

f(n) = aF(n)F(mF(n))f(kF(n)), that is to say, the topology of f(n), is the induced for the semi-norms (mF(n)), on 0MF, in the

representation on (H). Then the induced topology is the semi-norms topology in (H).

In particular, all p-canonical pair has an infinitesimal equivalent to (, (H)

F(mF(n))dn, where

F(mF(n)),

F(mF(n))dn,

) induced to a

, where has used the weak inequality. We

F(mF(n)) <

<mF(n)f(kF(n)), w>(f(kF(n))(1 + logmF(n))r

NF<f(n), w>dnNF(f(kF(n))aF(n)(1 + logmF(n))r

(f(kF(n)aF(n)aF(n)(1 + logmF(n))r

$$\mathbf{f}(\texttt{r}\texttt{r}\texttt{n}\texttt{n}\texttt{p}) = \mathbf{a}^{\mu+\rho}\texttt{\sigma}(\texttt{m})\mathbf{H}(\texttt{\underline{n}})\mathbf{w} = \texttt{\sphericalangle}(\texttt{m})\mathbf{H}(\texttt{\underline{n}})\; \mathbf{w} \rhd \; \tag{\text{VI.6}}$$

nNF, m0MF, aAF, and nNF, and due to that the topology with such partition of the unity is equivalent to the induced for the semi-norms in (H), since f(nman) = <(m)H(n), w>, is continuous in the same sense that

$$\text{f(nm\underline{r}(\underline{n})a\underline{r}(\underline{n})k\underline{r}(\underline{n})) = a\underline{r}(\underline{n})^{\mu+\rho\underline{r}}\sigma(m\underline{r}(\underline{n}))\mathbf{f}(k\underline{r}(\underline{n})) \text{.}$$

in HPF, (that is to say, f(n) = f(nmF(n)aF(n)kF(n)), nNF), then f(H), and

$$\int\_{\underline{\Delta}^{\underline{\mathbf{F}}}} \mathsf{c}\underline{\mathbf{f}}(\underline{\mathbf{n}}) , \le \mathsf{w} \, |\, \underline{\mathbf{dn}} = \mathsf{c}\mathbf{w} \, \mathsf{w} \, \mathsf{w} \gg 0,$$

Being IPF, , dense in (H) then fIPF, NF<f(n), w>dn, is not null.

**Lemma VI. 2.** Let fIPF, , be then there is a subspace of finite dimension V(f), of (H)K, such that

$$\int\_{\Delta \mathbf{F} \le \mathbf{f}(\underline{\mathbf{n}})} \mathbf{w} \rhd \underline{\mathbf{n}} = \mathbf{0}\_r$$

w(H)K, orthogonal to V(f).

*Proof*. Let w(H)K. If kKF, then

NF<f(n), w>dn = NF<(k)f(knk), w>dn = NF<(k)f(n), w>dn <sup>=</sup>NF<f(n), (k)w>dn,

where has been used the invariance of dn, on NF, under conjugation for kF. Let

$$\mathbf{S} = \{ \boldsymbol{\gamma} \}\_{\boldsymbol{\gamma} \in \mathbb{K} \times \boldsymbol{\gamma}} \tag{\text{VI.7}}$$

and we consider

$$\text{V(f)} = \sum\_{\mathbb{S}\&\mathbb{S}\text{-}\&\{\text{yl}\}\subset\mathbb{K}\times\text{H}\_{\mathfrak{a}}(\mathbb{S})} \kappa \cdot \text{H}\_{\mathfrak{a}}(\mathbb{S})\,\tag{\text{VI.8}}$$

since S, is finite S, then

$$\dim \mathcal{H}\_{\sigma}(\delta) \mathbf{f} \ll \alpha\_{\prime} \,\forall \, \mathbf{f} \in \mathcal{V}(\mathbf{f}),\tag{\text{V. 9}}$$

then dim V(f) <. Then since

$$\int\_{\underline{\Delta}^{\mathsf{F}}} \mathsf{cf}(\underline{\mathsf{n}}) , \le \mathsf{w} \rhd \underline{\mathsf{dn}} = \mathsf{cV}(\mathsf{f}) , \le \mathsf{w} \rhd \mathsf{w}$$

 f(H) . Then

$$\int\_{\underline{\Delta}\mathbb{F}\triangleleft(\underline{n}\right)} \text{w>d}\underline{n} = 0$$

If and only if (H)0K V(f), with (H)0K = {w} (H)K.

Now well, since

$$\bigcup\_{\underline{\text{NP}} \le \mathbf{f}(\underline{\mathbf{n}})} \mathbf{w} \rhd \underline{\mathbf{d}} \underline{\mathbf{n}} = \operatorname{\mathbf{V}}(\mathbf{f}) \text{, w} \rhd \mathbf{w}$$

Then exist a map

$$\mathfrak{glv}\_{"\sigma,\mu"} \text{:} \text{Iv}\_{"\sigma,\mu} \to (\text{H}\_{\sigma}) \text{x}\_{"\prime} \tag{\text{VI.10}}$$

whose correspondence rule is

$$\mathbf{f} \mapsto \mathsf{R}^{\mathsf{pr}\_{\mathsf{F},\mathsf{a}}} \mathfrak{a}, \mathfrak{h} \mathsf{f} \mathsf{w} = \mathsf{J}\_{\mathsf{NP}} \mathsf{$$

Since kKF, and nNF,

$$\begin{aligned} \mathsf{BPF}\_{\mathsf{c},\mathsf{o},\mathsf{h}}(\mathsf{f}(\underline{\mathsf{n}}\mathsf{k})) \mathsf{w} &= \mathsf{BPF}\_{\mathsf{c},\mathsf{o},\mathsf{h}}(\mathsf{c}(\mathsf{k})\mathsf{f}(\mathsf{k}^{-1}\underline{\mathsf{n}}\mathsf{k})) \mathsf{w} \\ &= \mathsf{BPF}\_{\mathsf{c},\mathsf{o},\mathsf{h}}(\mathsf{c}(\mathsf{k})\mathsf{f}(\underline{\mathsf{n}})) \mathsf{w} \\ &= \mathsf{BPF}\_{\mathsf{c},\mathsf{o},\mathsf{h}}(\mathsf{f}(\underline{\mathsf{n}})\mathsf{c}(\mathsf{k}^{-1})) \mathsf{w}\_{\mathsf{c}} \; \forall \; \mathsf{w} \in (\mathsf{H}\_{\mathsf{c}}) \mathsf{x}\_{\mathsf{c}} \end{aligned}$$

Then PF, , is an invariant homomorphism under the conjugation of KF. Being the (0MF, K) modules IPF, , and (H)K, KF- modules, the homomorphism PF, , is a homomorphism of KF-modules.

**Lemma VI. 3.**Let PF, (nF, IPF, ) = 0. Let PF, , be the corresponding linear map

$$\mathsf{hP}\_{"\sigma,\mu"} \colon \mathsf{I}\_{\mathsf{IP},\sigma,\mu} / \underline{\mathsf{u}} \mathsf{I}\_{\mathsf{IP},\sigma,\mu} \to (\mathsf{H}\_{\sigma}) \mathsf{k}\_{\prime} \tag{\mathsf{VI}.12}$$

whose rule of correspondence is

$$\mathbf{f} \mapsto \alpha \mathbf{p} \mathbf{e}\_{\cdot} \sigma, \mu (\mathbf{f}) \mathbf{w} = \int\_{\underline{\mathbf{N}}} \underline{\mathbf{v}} \xi \mathbf{f}(\underline{\mathbf{n}}) \, \mathbf{w} \rhd \underline{\mathbf{d}} \underline{\mathbf{n}}$$

then PF, HommF, K(IPF, /nFIPF, , (H)K).

*Proof*. Indeed, consider fIPF, , and let XnF. If w(H)K, then we can designate

70 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

If and only if (H)0K V(f), with (H)0K = {w} (H)K.

then dim V(f) <. Then since

. Then

whose correspondence rule is

Since kKF, and nNF,

whose rule of correspondence is

then PF, HommF, K(IPF, /nFIPF, , (H)K).

f(H)

Now well, since

Then exist a map

KF-modules.

dimH()f <, fV(f), (VI. 9)

NF<f(n), w>dn = <V(f), w>,

NF<f(n), w>dn = 0

NF<f(n), w>dn = <V(f), w>,

PF, : IPF, (H)K, (VI. 10)

<sup>f</sup> PF, (f)w = NF<f(n), w>dn, (VI. 11)

PF, (f(nk))w = PF, ((k)f(knk))w

= PF, ((k)f(n))w

= PF, (f(n)(k))w, w(H)K, Then PF, , is an invariant homomorphism under the conjugation of KF. Being the (0MF, K) modules IPF, , and (H)K, KF- modules, the homomorphism PF, , is a homomorphism of

<sup>f</sup> PF, (f)w = NF<f(n), w>dn,

PF, : IPF, /nFIPF, (H)K, (VI. 12)

**Lemma VI. 3.**Let PF, (nF, IPF, ) = 0. Let PF, , be the corresponding linear map

$$\gamma\_{\mathbb{W}}(\mathbf{f}) = \begin{cases} \underline{\text{NF}} \lhd \mathbf{f}(\underline{\mathbf{n}}) \text{, w} \rhd \underline{\text{d}\underline{\mathbf{n}}},\\ \underline{\text{0}} \text{, w} \urcorner \text{, w} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urcorner \text{,} \urleft( \text{,} \urcorner \text{,} \urcorner \text{,} \urleft( \text{,} \urcorner \text{,} \urleft$$

Since w, is a defined functional in IPF, , and this is a dense space in (H), are the induced by the same numerable space of semi-norms, then w, is a continuous functional. Then XnF,

$$\gamma\_{\mathbb{W}}(\mathsf{Xf}) = \mathsf{d}/\mathsf{df} \; \gamma\_{\mathbb{W}}(\pi\_{\sigma,\mu}(\exp(\mathsf{tX})\mathsf{f})) \; \_{1=0} = 0,\tag{\mathsf{V}I. 13}$$

Which is true for the right invariance of dn, on NF, (then PF, (nF, IPF, ) = 0). If X0mF, then for the invariance under the conjugation of elements 0MF, is had that:

$$\gamma\_{\mathbb{W}}(\mathsf{X}\mathsf{f}) = \mathsf{d}/\mathsf{d}t \,\gamma\_{\mathbb{W}}(\mathsf{x}\_{\mathsf{f},\mathsf{u}}(\mathsf{exp}(\mathsf{tX})\mathsf{f})\,\vert\_{\mathsf{t}=0} = (\mathsf{d}/\mathsf{d}t)\mathsf{l})\_{\mathsf{i}\mathbb{R}^{d}} \,\mathsf{f}(\underline{\mathsf{e}}\mathsf{exp}(\mathsf{tX})), \,\mathsf{w}\rhd\mathsf{d}\underline{\mathsf{n}}\,\vert\_{\mathsf{t}=0}$$
 
$$= (\mathsf{d}/\mathsf{d}t)\Big[\_{\mathsf{i}\mathbb{R}^{d}}\mathsf{c}\sigma(\exp(\mathsf{tX}))\mathsf{f}(\exp(-\mathsf{tX}))\underline{\mathsf{n}}\exp(\mathsf{tX})\rangle, \,\mathsf{w}\rhd\mathsf{d}\underline{\mathsf{n}}\,\vert\_{\mathsf{t}=0}$$
 
$$= (\mathsf{d}/\mathsf{d}t)\Big[\_{\mathsf{i}\mathbb{R}^{d}}\mathsf{c}\sigma(\exp(\mathsf{tX}))\underline{\mathsf{f}}(\underline{\mathsf{n}}), \,\mathsf{w}\rhd\mathsf{d}\underline{\mathsf{n}}$$

where the last equation is followed of the invariance of the measure dn, on NF, under conjugation by elements of 0MF. Since there is a subspace of finite dimension such that w(H)K, {w}, is orthogonal to said subspace of IPF, , then PF, (nF, IPF, ) = 0, and is 0mFinvariant, that is to say,

$$
\beta \nu \mathbb{5}\_{,\sigma,\mu}(\mathsf{Xf}) = \mathsf{X} \beta \nu \mathbb{5}\_{,\sigma,\mu}(\mathsf{f}), \quad \forall \quad \mathsf{X} \in {}^{0}\mathsf{m}\_{,}\tag{\mathsf{VI.14}}
$$

If haF, then is necessary consider in the KF- homomorphism PF, , the following NFmeasure

$$\mathbf{d(a\underline{n}a^{-1}) = a^{-2\rho\_s}d\underline{n}}\tag{V1.15}$$

and find that under conjugation of KF, satisfies that

$$\begin{aligned} \mathsf{h}\_{\mathsf{PF},\mathsf{\{\{\mu\}\}}} &= (\mathsf{\mu} - \mathsf{\{\mu\}}) \mathsf{\{\mu\}} \mathsf{\{\mu\}}\_{\mathsf{\{\text{PF},\mathsf{\{\mu\}}\}}} \mathsf{\{\{\mathsf{f}\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\{\mathsf{f}\}}\_{\mathsf{\{\mu\}}}) \mathsf{\{\{\mathsf{f}\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}}) \mathsf{\{\}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}}) \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\{\mu\}}} \mathsf{\{\}}\_{\mathsf{\$$

*The lemma VI. 3.,* jointly with the lemma on the fast decreasing of a temperate irreducible representation of G, of integrable square (Conference Vol. 1 temperate (g, K)-modules [28]) implies that exist a homomorphism of (g, K)-modules

$$\mathsf{j}\mathsf{v}\_{",\mathsf{c},\mathsf{u}"}; \mathsf{ I}\mathsf{v}\_{",\mathsf{c},\mathsf{u}"} \to \mathsf{I}\mathsf{v}\_{",\mathsf{c},\mathsf{u}"} \tag{16.16}$$

whose rule of correspondence is

$$\mathbf{f} \mapsto \alpha \mathbf{v} \cdot\_{\sigma, \mathfrak{u}} (\mathbf{f}) = \mathbf{j} \cdot\_{\sigma, \mathfrak{u}} (\mathbf{f}) (1), \tag{V1.17}$$

and since IPF, , is a (g, 0M)-module non-vanishing then jPF, , is non-vanishing by the lemma on fast decreasing of a temperate irreducible representation belonging to the space L2(G). This last will be an important question in the after applications of the representation theory.

**Theorem VI. 1.** Maintaining the previous affirmations, let f(H), and gIPF, . Let haF, be such that (h) > 0, (PF, AF), then

$$\lim\_{\mathsf{Hilm}\to\mathsf{t}\to\mathsf{t}} \mathsf{c}^{\mathsf{t}(\mathsf{P}-\mathsf{h})\mathsf{h}} \approx \mathsf{\pi}(\mathsf{exp}(\mathsf{th})) \mathsf{f}, \mathsf{g} \textbf{ \underline{\mathsf{s}}} = \int \mathsf{\Delta}\mathsf{v} \cdot \mathsf{f}(\underline{\mathsf{n}}) \cdot \mathsf{g}(1) \mathtt{\underline{\mathsf{d}}} \,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{}\,\mathsf{d}\underline{\mathsf{s}}\,\mathsf{d}\underline{\mathsf{n}}\,\mathsf{$$

where = PF, .

*Proof.* For one side g, is K-finite and by the lemma on the finitely generated [5, 28, 40], the generated by g(K), is of finite dimension. Since f(H), then f(g)(H) , gG. Consider at = exp(th), haF, then

$$<\text{\textbullet}(\text{exp(th)})\text{f, g>} = <\text{\textbullet}(\text{exp(an)})\text{f, g>} = \left[ \times \text{\textbullet}(\text{ka}), \text{ g(k)>} \text{dk}, \quad \forall \text{ k} \in \text{K} \tag{\textbullet} \quad \text{(VI.19)} \right]$$

But, since the measure on G (K-invariant) can be normalized then by the theory of integration on parabolic pairs [Wallach] is had that

$$\int\_{\mathbb{X}} \lhd \mathsf{f}(\mathsf{k}\mathsf{a}), \operatorname{g}(\mathsf{k}) \rhd \mathsf{d}\mathsf{k} = \int\_{\mathsf{i}\mathsf{F}\mathsf{a}\mathsf{F}\mathsf{a}\mathsf{F}\mathsf{(n}} \mathsf{j}^{\mathscr{D}} \lhd \mathsf{f}(\mathsf{k}\mathsf{(n}\mathsf{)}\mathsf{a}), \operatorname{g}(\mathsf{k}\mathsf{(n}\mathsf{)}) \rhd \mathsf{d}\mathsf{n} = \int\_{\mathsf{i}\mathsf{F}\mathsf{j}} \operatorname{g}(\mathsf{k}\mathsf{a}) \operatorname{g}(\mathsf{k}\mathsf{n}) \rhd \operatorname{g}(\mathsf{k}\mathsf{n}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{b}) \operatorname{g}(\mathsf{k}\mathsf{$$

 k(n)atKF, k(n)K, and nNF. Now well, n = nmF(n)aF(n)kF(n) nNF. Consider kF(n)NF(mF(n)aF(n))-1n. This circumstance will imply to the measure dg that:

$$\begin{aligned} \mathsf{r}\mathsf{s}\mathsf{r}(\mathsf{exp}(\mathsf{a}))\mathsf{f}, \; \mathsf{g}\mathsf{s} &= \int\_{\mathsf{NFAF}} (\mathsf{n})^{2\mathsf{p}} \mathsf{ar}(\mathsf{n})^{-\mathsf{p}-\mathsf{a}} \mathsf{c}\mathsf{o}(\mathsf{mr}(\mathsf{n}))^{-1} \mathsf{f}(\mathsf{na}), \; \mathsf{g}(\mathsf{k}(\mathsf{n})) \mathsf{z}\mathsf{d}\mathsf{n} \\\\ &= \int\_{\mathsf{NFAF}} (\mathsf{a}\mathsf{na}\mathsf{a}\cdot\mathsf{i})^{2\mathsf{p}} \mathsf{ar}(\mathsf{a}\mathsf{na}\mathsf{a}\cdot\mathsf{i})^{-\mathsf{p}-\mathsf{a}} \mathsf{c}\mathsf{o}(\mathsf{mr}(\mathsf{n}))^{-1} \mathsf{f}(\mathsf{a}\mathsf{na}\mathsf{a}\cdot\mathsf{i}), \; \mathsf{g}(\mathsf{k}(\mathsf{n})) \mathsf{z}\mathsf{d}\mathsf{n} \\\\ &= \int\_{\mathsf{NFAF}} \mathsf{a}^{\mathsf{p}+\mathsf{n}} \mathsf{ar}(\mathsf{n})^{\mathsf{p}-\mathsf{n}} \mathsf{c}\mathsf{o}(\mathsf{mr}(\mathsf{n}))^{-1} \mathsf{f}(\mathsf{a}\mathsf{na}\mathsf{a}\cdot\mathsf{i}), \; \mathsf{g}(\mathsf{k}(\mathsf{n})) \mathsf{z}\mathsf{d}\mathsf{n} \\\\ &= \mathsf{a}^{\mathsf{p}+\mathsf{n}} \mathsf{f}\_{\mathsf{NFAF}}(\mathsf{n})^{\mathsf{p}-\mathsf{n}} \mathsf{c}\mathsf{o}(\mathsf{mr}(\mathsf{n}))^{-1} \mathsf{f}(\mathsf{a}\mathsf{na}\mathsf{a}\mathsf{a}\mathsf{}), \; \mathsf{g}(\mathsf{k}(\mathsf{n})) \mathsf{z}\mathsf{d}\mathsf{n} \end{aligned}$$

where has been considered that

$$\mathbf{a}^{\rho^{+} + \mu} \Xi(\mathbf{a}) = \int\_{\Delta \mathbf{F} \mathbf{a} \mathbf{a}} \mathbf{\underline{\mathbf{u}}} \mathbf{\underline{\mathbf{u}}}^{\rho - \mu} \mathbf{a} (\mathbf{a} \mathbf{a} \mathbf{\underline{a}} \cdot \mathbf{\underline{\mathbf{v}}})^{\rho} d \mathbf{\underline{\mathbf{u}}} \tag{\text{VI. 20}}$$

where a(atna-t) = (mF(n))f(atna-t), nNF. Then

$$<\text{π(exp(th))f},\text{g}> = \text{α}^{\text{p}+\text{μ}}\text{]} \times \text{πə} \text{ε(}\underline{\text{n}}\text{)}^{\text{p}-\text{μ}} \text{ç} \text{σ(}\text{m}\text{-(}\underline{\text{n}}\text{)}^{\text{-}1}\text{f} \text{(}\underline{\text{n}}\text{)},\text{g} \text{(}k \text{(ənəzə.)}\text{)} \text{ç} \text{d}\underline{\text{n}}\text{ }\tag{\text{VI. 21}}$$

If is possible to exchange the limit for the integral, the result will follows trivially. For this goal, we consider a measurable subset E, of N, to know;

$$\mathbf{I}\_{\mathbf{f}}(\mathbf{E}) = \left[ \mathsf{zar}(\mathbf{a}\underline{\mathbf{a}}\underline{\mathbf{a}}\cdot\mathbf{\iota})^{\rho-\mu} \mathsf{c}\sigma(\mathbf{m}\mathbf{\iota}(\mathbf{a}\underline{\mathbf{a}}\underline{\mathbf{a}}\cdot\mathbf{\iota}))^{-1} \mathbf{f}(\underline{\mathbf{n}}), \; \mathbf{g}(\mathbf{k}\langle \mathbf{a}\underline{\mathbf{a}}\underline{\mathbf{a}}\cdot\mathbf{\iota} \rangle) \succeq \mathsf{d}\underline{\mathbf{n}}.\tag{V1.22}$$

that is to say, we will elect a measurable subset E, of N, whose integral is defined under said rule of correspondence and under the normalization of their measure.

By this way, one can to demonstrate that exist an integrable function u, on NF, such that

$$\text{Li(E)} \lesssim \text{\(\mu\)nd} \underline{\text{n}} \text{d} \underline{\text{n}} , \text{ } \forall \text{ } \forall \text{ } \text{\(\text{V}\text{.} \, 23\text{)}$$

In this way, the justification of the interchange of the limit for the integral is a mere consequence of the Vitali theorem. Likewise, the rule of transformation of f implies that

$$\mathbf{H}(\mathbf{E}) = \int \mathbf{E} \mathbf{a} \mathbf{r} (\mathbf{a} \mathbf{\overline{a}} \mathbf{a} \cdot \mathbf{l})^{\rho - \mu} \mathbf{a} \mathbf{r} (\underline{\mathbf{n}})^{\rho + \mu} \operatorname{c} \sigma (\mathbf{m} \mathbf{r} (\mathbf{a} \mathbf{\overline{a}} \mathbf{a} \cdot \mathbf{l}))^{-1} \operatorname{m} \sigma (\underline{\mathbf{n}}) \mathbf{f} (\underline{\mathbf{n}}) \operatorname{g} (\underline{\mathbf{k}} (\mathbf{a} \mathbf{\overline{a}} \mathbf{a} \cdot \mathbf{l})) \operatorname{c} \mathbf{d} \underline{\mathbf{n}} \,\underline{\mathbf{n}} $$

where

72 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

be such that (h) > 0, (PF, AF), then

integration on parabolic pairs [Wallach] is had that

<sup>&</sup>lt;(exp(at))f, g> = NFaF(n)2

<sup>=</sup>NFaF(atna-t)2

<sup>=</sup>NFat

= at

where has been considered that

at

<(exp(th))f, g> = at

where a(atna-t)

where = PF, .

at = exp(th), haF, then

and since IPF, , is a (g, 0M)-module non-vanishing then jPF, , is non-vanishing by the lemma on fast decreasing of a temperate irreducible representation belonging to the space L2(G). This

**Theorem VI. 1.** Maintaining the previous affirmations, let f(H), and gIPF, . Let haF,

limt+et()(h)<(exp(th))f, g> = NF<f(n), g(1)>dn, (VI. 18)

*Proof.* For one side g, is K-finite and by the lemma on the finitely generated [5, 28, 40], the

<sup>&</sup>lt;(exp(th))f, g> = <(exp(at))f, g> = K<f(kat), g(k)>dk, kK (VI. 19)

But, since the measure on G (K-invariant) can be normalized then by the theory of

k(n)atKF, k(n)K, and nNF. Now well, n = nmF(n)aF(n)kF(n) nNF. Consider

aF(n)<(mF(n))f(atna-t), g(k(n))>dn

NFaF(n)<(mF(n))f(atna-t), g(k(n))>dn,

(at) = NFaF(n)a(atna-t)

If is possible to exchange the limit for the integral, the result will follows trivially. For this

<f(k(n)at), g(k(n))>dn,

aF(n)<(mF(n))f(nat), g(k(n))>dn

<(mF(n))f(atna-t), g(k(n))>dn

NFaF(n)<(mF(n))f(n), g(k(atna-t))>dn, (VI. 21)

dn, (VI. 20)

, gG. Consider

generated by g(K), is of finite dimension. Since f(H), then f(g)(H)

K<f(kat), g(k)>dk = NFaF(n)2

kF(n)NF(mF(n)aF(n))-1n. This circumstance will imply to the measure dg that:

aF(atna-t)

= (mF(n))f(atna-t), nNF. Then

goal, we consider a measurable subset E, of N, to know;

last will be an important question in the after applications of the representation theory.

aF(atna-t) aF(n)<(mF(atna-t))mF(n)f(n), g(k(atna-t))> aF(atna-t) Re F(mF(atna-t))mF(n)aF(n)Re (1 + logmF(n))d,

Thus the expression of the limit followed of the integral It(E), result be finite and bounded by the integral (VI. 23). Then analyzing the expression of the function F, evaluated inmF(atna-t))mF(n)0MF, is applied the property to this case, following:

$$\Xi\_{\mathbf{F}} \left( \mathbf{x}^{-1} \mathbf{y} \right) = \int\_{\mathbf{k} \mathbf{F}} \mathbf{a} (\mathbf{k} \mathbf{x})^{\rho} \mathbf{a} (\mathbf{k} \mathbf{y})^{\rho} \mathbf{d} \mathbf{k}, \ \forall \ \mathbf{x}, \ \mathbf{y} \in {}^{0}\mathbf{M}\_{\mathbf{F}} \tag{I}$$

signing

$$\begin{aligned} \boldsymbol{\Xi}\_{\mathsf{F}} \left( \mathbf{x}^{-1} \mathbf{y} \right) &= \int\_{\mathsf{KF}} \mathbf{a} (\mathbf{k} \mathbf{x}^{-1} \mathbf{y}) \mathrm{d}\mathbf{k} = \int\_{\mathsf{KF}} \mathbf{a} (\mathbf{k} (\mathbf{k} \mathbf{x}) \mathbf{x}^{-1} \mathbf{y}) \mathbf{a} (\mathbf{k} \mathbf{x})^{2} \mathrm{d}\mathbf{k} \\\\ &= \int\_{\mathsf{KF}} \mathbf{a} (\mathbf{k} \mathbf{y})^{\mathsf{P}} \mathbf{a} (\mathbf{k} \mathbf{x})^{\mathsf{P}} \mathrm{d}\mathbf{k}, \quad \forall \ \mathbf{x}, \ \mathbf{y} \in {}^{\mathsf{0}} \mathrm{M} \mathbf{x} \end{aligned}$$

where kx = n(kx)a(kx)k(kx). Then

$$\int\_{\mathbb{X}\mathbb{F}} \mathbf{a}(\mathbf{k}\mathbf{y})^\mathsf{o}(\mathbf{k}\mathbf{x})^\mathsf{o}\mathbf{d}\mathbf{k} = \int\_{\mathbb{X}\mathbb{F}} \mathbf{a}(\mathbf{k}(\underline{\mathbf{n}})\mathbf{x})^\mathsf{o}\mathbf{a}(\mathbf{k}(\underline{\mathbf{n}})\mathbf{y})^\mathsf{o}\mathbf{a}(\mathbf{n})^{2\mathsf{p}} d\underline{\mathbf{n}}$$

The previous orbital integral imply to the inequality It(E) Eu(n)dn, t> 0, that

$$\mathbf{I}(\mathbf{E}) \preceq\_{\mathbb{E}\_{\times}} \underline{\text{\underline{\mathbf{\underline{\mathbf{\underline{\mathbf{\boldsymbol{\beta}}}}}}}} \mathbf{\underline{\mathbf{a}}}} \mathbf{\underline{\mathbf{a}}} \mathbf{\underline{\mathbf{a}}} \mathbf{\underline{a}} \mathbf{\underline{a}} \mathbf{\underline{\mathbf{a}}} \mathbf{\underline{\mathbf{a}}} \mathbf{\underline{\mathbf{a}}} (\mathbf{\underline{\mathbf{a}}} \mathbf{\underline{\mathbf{m}}} \mathbf{\underline{\mathbf{n}}} \mathbf{\underline{\mathbf{n}}}) \mathbf{\underline{\mathbf{a}}} \mathbf{\underline{\mathbf{n}}} (\underline{\mathbf{n}})^{\diamond} \mathbf{\underline{\mathbf{v}}} \mathbf{\underline{\mathbf{a}}} (\underline{\mathbf{n}})^{\diamond \mathbf{a}} \mathbf{\underline{\mathbf{r}}} \mathbf{\underline{\mathbf{n}}} \mathbf{\underline{\mathbf{n}}} \mathbf{\underline{\mathbf{n}}} \mathbf{\underline{\mathbf{n}}} \mathbf{\underline{\mathbf{v}}} \mathbf{\underline{\mathbf{a}}}$$

with v(n) = (1 + logmF(n))d. Now, considering a(mF(g))aF(g) = a(g), and

$$\mathsf{r}^\* \underline{\mathbf{n}} \mathsf{m} \mathsf{r}(\mathsf{a} \underline{\mathbf{n}} \mathsf{a} \cdot \mathsf{t}) = \mathsf{m} \mathsf{r}(\mathsf{a}^\* \underline{\mathsf{n}} \mathsf{a} \cdot \mathsf{t}),$$

then

$$\left| \mathbf{I} \mathbf{(E)} \right| \leq \left| \mathbf{E}\_{\times} \text{ :} \underline{\text{ν}} \mathbf{a} (\! \underline{\text{'}} \underline{\text{n}} \underline{\text{n}} \underline{\text{n}} \underline{\text{n}} \cdot \mathbf{i})^{\rhd} \mathbf{a} (\! \underline{\text{''}} \underline{\text{n}} \underline{\text{n}} \underline{\text{v}})^{\rhd} \mathbf{\underline{\text{v}}} (\! \underline{\text{n}} \underline{\text{v}} \underline{\text{d}} \underline{\text{n}} \underline{\text{d}} \underline{\text{n}} \underline{\text{v}}) \right| $$

But due to the K-invariance of the measure dg, normalized in NF, then

$$\int\_{\mathbb{E}\_{\times} \times \mathfrak{Z} \text{ar}(\text{"nana} \cdot \text{"})^{\text{\textquotedblleft}} \mathbb{R}^{\text{\textquotedblright}} \text{a}(\text{"nna})^{\text{\textquotedblleft}} \text{v}(\text{n}) \mathrm{d}\mathbf{n} \,\text{dn}^{\text{\textquotedblleft}}$$

$$= \int\_{\mathbb{E}\_{\times} \times \mathfrak{Z} \text{a}(\text{k}(\text{"n}) \text{a} \underline{\text{n}} \cdot \text{i})^{\text{\textquotedblleft}} \text{a}(\text{k}(\text{"n}) \text{n})^{\text{\textquotedblleft}} \text{v}(\text{n}) \mathrm{a}(\text{"n}) \text{a}(\text{"n})^{\text{\textquotedblright}} \text{d}\mathbf{n} \,\text{d}\underline{\text{n}}^{\text{\textquotedblleft}}$$

$$= \int\_{\mathbb{E}\_{\times} \times \mathfrak{Z} \text{a}(\text{ak}(\text{"n}) \text{n} \text{k}(\text{"n}) \text{a} \cdot \text{i})^{\text{\textquotedblleft}} \text{e} \text{a}(\text{k}(\text{"n}) \text{n} \text{k}(\text{"n}))^{\text{\textquotedblleft}} \text{b}^{\text{\textquotedblleft}} \text{d}\mathbf{n} \text{}^{\textquotedblleft} \text{d}\mathbf{n} \text{}^{\textquotedblright}$$

$$= \int\_{\mathbb{E}\_{\times} \times \mathfrak{Z} \text{a}(\text{"n})^{\textquotedblright}} \text{a}(\text{a} \text{a} \text{a}\_{\text{}} \cdot \text{i})^{\text{\textquotedblleft}} \text{e} \text{a}(\text{!})^{\textquotedblleft}} \text{1} + \log \| \,\text{m} \text{\textquoted$$

Let 0 << 1, be, such that ReF, > 0, (PF, AF). Then the kernel a(atna-t) Re a(n)Re , takes the form

$$\mathbf{a}(\mathbf{a}\overline{\mathbf{a}}\mathbf{a}\cdot)^{\rho\_{\mathbb{P}}\mathbf{a}\mu}\mathbf{a}(\overline{\mathbf{n}})^{\rho\_{\mathbb{P}}\mathbf{a}\mu} = \mathbf{a}(\mathbf{a}\overline{\mathbf{a}}\mathbf{a}\cdot)^{\rho\_{\mathbb{P}}\mathbf{a}\mu}\mathbf{a}(\mathbf{a}\overline{\mathbf{a}}\mathbf{a}\cdot)^{(\rho\_{\mathbb{R}}\mu-\mathbf{a}\mu\_{\mathbb{E}})}\mathbf{a}(\overline{\mathbf{n}})^{\rho\_{\mathbb{R}}\mu-\mathbf{a}\mu\_{\mathbb{E}}}\mathbf{a}(\overline{\mathbf{n}})^{\rho\_{\mathbb{P}}\mathbf{a}\mu} \leq \mathbf{a}(\overline{\mathbf{n}})^{\rho\_{\mathbb{P}}\mu\mu}, \text{[5.28]}, \quad \mathbf{a}(\overline{\mathbf{n}}) \leq \mathbf{a}(\overline{\mathbf{n}})^{\rho\_{\mathbb{P}}\mu}$$

Considering the appendix C, is natural suppose that to any q > 0, there is *C*q> 0 such that

$$\mathsf{a(\underline{n})}^{p+\varepsilon\mathsf{op}} \leq \mathsf{C}\_{\mathsf{q}} \mathsf{a(\underline{n})}^{p} (1 - \log \mathsf{l} \ \mathsf{a(\underline{n})} \mathsf{l} \ )^{\mathsf{I}} \text{.}\tag{\mathsf{VI.24}}$$

Considering to u(n) = a(n)F(1 logn)d, is had that u, is integrable on N, and then E measurable subset of N, one satisfies that (VI. 23). Therefore the integral (VI. 19) exist, and since the limit exist then the Haar measure on NF, is reduced to the integral

$$\mathbf{I} = \int\_{\Delta \mathbb{F}} \mathsf{L}\mathsf{F}\mathsf{f}(\mathsf{n}), \,\mathsf{g}(1) \rhd \mathsf{d}\underline{\mathfrak{n}}, \,\forall \underline{\mathfrak{n}} \in \underline{\mathbf{N}}\mathsf{F}. \,\mathsf{T}$$

## **Some Examples of Orbital Integrals into Representation Theory on Field Theory and Integral Geometry**

### **VII. 1.**

74 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

It(E) <sup>E</sup> \*NFa(\*natna-t)

<sup>E</sup> \*NFa(\*natna-t)

<sup>=</sup><sup>E</sup> \*NFa(k(\*n)atna-t)

<sup>=</sup><sup>E</sup> \*NFa(atk(\*n)nk(\*n)a-t)

<sup>=</sup><sup>E</sup> \*NFa(\*n)a(atna-t)

, takes the form

= a(atna-t)

a(n)F*C*qa(n)

<sup>=</sup>Ea(atna-t)

But due to the K-invariance of the measure dg, normalized in NF, then

Re

Re

Re

Re

Re

a(n)Re

Considering the appendix C, is natural suppose that to any q > 0, there is *C*q> 0 such that

Considering to u(n) = a(n)F(1 logn)d, is had that u, is integrable on N, and then E measurable subset of N, one satisfies that (VI. 23). Therefore the integral (VI. 19) exist, and

I = NF<f(n), g(1)>dn, nNF.

Re

Fa(atna-t)

since the limit exist then the Haar measure on NF, is reduced to the integral

a(\*nn)Re

a(\*nn)Re

a(k(\*n)n)Re

a(k(\*n)nk(\*n)))

a(n)Re

Let 0 << 1, be, such that ReF, > 0, (PF, AF). Then the kernel

v(n)dndn\*,

v(n)a(\*n)dndn\*

v(n)a(\*n)dn\*dn

q, (VI. 24)

v(n)dndn\*

Re

(1 + logmF(n))ddn,

(1 loga(n))

(1 + logmF(n))ddn\*dn,

ReF)a(n)ReFa(n)F a(n)F, [5, 28],

then

a(atna-t)

a(atna-t)

Re

Re

a(n)Re

a(n) Re

The development of orbital integrals into of the mathematical physics, conclave to use of integral representations that are realizations of certain unitary representations of Lie groups such as U(n), SU(2), SU(n), SU(2, 2), SU(p, q),and U(p, q), and that helps to obtain general solutions of partial differential equations in context of an algebra, like for example a quaternion algebra. In the case of establish equivalences through of a moduli space that can be descript by algebraic relations among geometrical identities, as the of quaternion analysis, carry us to establish a structural equivalence among the spaces R4 C2, that which conclave to the re-definition in a context of hyper-complex analysis of vector objects in R4. Other example that we find in the integrals of the integral geometry is the given to G2, 4(C), through of the space P3(C), which establish a twistor geometry.

But the relations of isomorphism in vector homogeneous bundles are similar to the follows in integral developmental operator cohomology in vector tomography. Now, the intertwining integral operators among cohomological classes of both context (respective cycles), result be equivalent in context of bundles of lines (with equivalent cocycles). Then some conjectures that can be possible of prove to the light of some integral operators that establish this isomorphism is:

**Conjecture VII. 1.**"The Penrose transform is a vector Radon transform on sections of homogeneous vector bundles in **M**', being **M**, a complex Riemannian manifold.

And considering a reconstruct of a Riemannian manifold through of cycles, we can consider the following conjecture.

**Conjecture VII. 2.** "The twistor transform is the generalization of the Radon transform n Bundles of lines and the Penrose transform is a specialization of the twistor transform in S4"

Of fact, the twistor transform can determines by the pair of Penrose transform on the Gorbits P**+**, y P , [32, 33]. This is analogue to determine the Fourier transform in R*n*, like the combination of the Radon transform with the unidimensional Fourier transform calculated these also on Euclidean G-orbits[32, 33, 37]**,** (hyperplanes and hyperlines).

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

In a more general sense, the vector bundles to the that it is does allusion are the seated in homogeneous spaces GC/L, with homogeneous subspaces GC/Q, where persist a complex holomorphic manifold with Q-orbits (parabolic orbits), of GC/L, which have a Haar induced measure in every close submanifold given for the flags of the corresponding holomorphic vector G-bundles, and whose orbital integrals are evaluated on said orbits obtaining applicable invariants to all the homogeneous space in question, GC/L. This process called of orbitalization of the homogeneous space GC/L, is the base in major part in the inclusion of homogeneous spaces obtained for reduction of GC/L, through their holonomy.

Likewise, calculating the orbital integrals on the corresponding orbits to the reductive homogeneous spaces obtained by reduction of GC/L, that is to say; on the subjacent orbits in each one of the components of the sequence of inclusions [39]:

We obtain a constant operator in the applications of a homogeneous space in other, given by the complex integral operator on cohomological classes Hs (D, ()), where D, is the set of certain positive lines in P*n*, and whose image is the group of cohomology Hs (G/H, *n*).*<sup>n</sup>* represent the corresponding homogeneous bundle on the orbit **M**+, and *s*, is the complex dimension of **M**. The lines in P*n*, is identified in every case like the corresponding trajectories in **M**, to know,

Of this way, a singularity in the G-structure of a differentiable manifold **M**, is a zero in the *D*h-structure of **M**. This is precisely one of the developmental aspects by Penrose in their twistor theory through of the integral geometry to the manipulation of the infinites. Dh, is

the space of lines in P*n*, whose image under certain integral transform is the space of points P*n*\*, (the dual projective space to P*n*), and whose dual natural pairing of homogeneous coordinates is vanished in the context of the complex bundle of lines on *L*.

The spaces Hs (G/H, *n*), result be *fine representations*, of G/L, through of the bundle of flags corresponding to complex holomorphic G-bundle seated in G/L. The realizations of these unitary representations are the orbital integrals on the flags that are the K-orbits of the corresponding vector G-bundle. If we consider the corresponding Lie algebras to whose homogeneous spaces, in the frame of homogeneous space G/P, the representation spaces can be obtained and classified in the via of the *L2-cohomología*, on parabolic orbits and through the corresponding orbital integrals.

This is precisely used in unitary representations, that is to say; the use of representations of G, (a complex Lie group), constructed on spaces of holomorphic sections of vector bundles and generalizations. The problem that here arises is the validation and verify of the *Hermitian forms* calculated through of the diverse integral realizations with the corresponding representations electing the *differentiable and adequate minimal globalization*. To it, we must use a general version of complex cohomology called *topological cohomology of Dolbeault*that is a *Serre generalization*, of the Dolbeault cohomology. The intertwining integrals in this case include *Szëgo operators* and others operators on *quotients of Langlands.*

A particular applied case to the context of the compact groups, and thus of the L2-cohomology more simple (a subquotient of the L2-cohomology on G/L), is that the holomorphic vector Gbundle of G/L **M**, is isomorphic to vector bundle of G/K, with L = ZG(T), with T, a maximal torus of G. The integral operators in this case results be Fourier transforms in the context of the complex holomorphic bundles. The lines in this case, result be the circles S1, in C*n*.

The obtaining of an analytic function of a space to other, is realized of agree to the integral cohomology, (*n* – 1 q)--cohomology H*n*1, q(C*n*/D, V), where D, is a linearly concave domain in C*n*. Likewise for example, if we consider the space C2, the integral operator that obtain a harmonic function in R4, through of to calculate functions of C(C3),on lines S1,in C2, comes given for

$$\phi(\mathbf{w}, \mathbf{x}, \mathbf{y}, \mathbf{z}) = \left| \operatorname{s1} \left\{ (\mathbf{w} + \mathbf{x}) + (\mathbf{y} + \mathbf{i}\mathbf{z})\zeta\_{\nu} \left( \mathbf{y} - \mathbf{i}\mathbf{z} \right) + (\mathbf{w} - \mathbf{x})\zeta\_{\nu}\zeta\_{\mathbf{z}} \right\} \operatorname{d}\zeta,\tag{\text{VII. 3}}$$

which have their geometrical re-interpretation like a Radon transform on lines of a flag manifold F = LL R4, 27, 31, 33

$$
\mathcal{R} \colon \mathbf{C}^r(\mathbf{C}^\circ) \to \mathbf{C}^\circ(\mathbb{R}^\circ),
\tag{VII.4}
$$

with rule of correspondence

76 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

each one of the components of the sequence of inclusions [39]:

the complex integral operator on cohomological classes Hs

trajectories in **M**, to know,

In a more general sense, the vector bundles to the that it is does allusion are the seated in homogeneous spaces GC/L, with homogeneous subspaces GC/Q, where persist a complex holomorphic manifold with Q-orbits (parabolic orbits), of GC/L, which have a Haar induced measure in every close submanifold given for the flags of the corresponding holomorphic vector G-bundles, and whose orbital integrals are evaluated on said orbits obtaining applicable invariants to all the homogeneous space in question, GC/L. This process called of orbitalization of the homogeneous space GC/L, is the base in major part in the inclusion of

Likewise, calculating the orbital integrals on the corresponding orbits to the reductive homogeneous spaces obtained by reduction of GC/L, that is to say; on the subjacent orbits in

We obtain a constant operator in the applications of a homogeneous space in other, given by

represent the corresponding homogeneous bundle on the orbit **M**+, and *s*, is the complex dimension of **M**. The lines in P*n*, is identified in every case like the corresponding

Of this way, a singularity in the G-structure of a differentiable manifold **M**, is a zero in the *D*h-structure of **M**. This is precisely one of the developmental aspects by Penrose in their twistor theory through of the integral geometry to the manipulation of the infinites. Dh, is

certain positive lines in P*n*, and whose image is the group of cohomology Hs

(VII. 1)

(G/H, *n*).*<sup>n</sup>*

(VII. 2)

(D, ()), where D, is the set of

homogeneous spaces obtained for reduction of GC/L, through their holonomy.

$$\text{if(p, q, \zeta)} \mid \rightarrow \Big\{ \mathsf{L}\_{\mathsf{L} \gets \mathsf{R}} \mathsf{1} \,\mathsf{f}(\mathsf{p, q, \zeta}) \,\mathsf{d}\zeta,\tag{\text{VII.5}}$$

which is an integral on a line of C2, of the bundle of lines seated in F10, 27.

Which is the behavior of these integrals on a complex Riemannian manifold?

Why is the special behavior of these integrals on a complex Riemannian manifold? Is possible establish criteria to the study of certain geometrical properties (such as curvature or torsion) of a complex Riemannian manifold **M**, based in integrability on a holomorphic complex bundle in a manifold through of their differentiable projections? Such criteria raise certain class of integrals calculated in sections of holomorphic bundle of M, in an analogous cohomology to the given by the space H*n*1, q(C*n*/D, V)?

Is feasible in this sense to obtain a generalization of cohomology on G-invariant orbits, belonging to holomorphic bundles seated in reductive homogeneous spaces of such form that are obtained integrals of generalized functional to the geometrical observables of the space **M**, and their orbits? Will establish this cohomology some theory of integration on certain complex spaces, parallel to the obtained by the integral cohomology on homogeneous bundles evaluating integrals in a direct and global via on the vector flux that conform them and define them?

The answer to first question save, clear this, the securing of harmonic functions , (it which no always is possible through of this path) is positive to the case of analytic functions and have that see with the structure a lo más Hermitiana of **M**, that can be establish constructing a fibration of **M** .

Then the resulting integrals are calculated on orbits of sections of the bundle TJ(**M**), the which is usually inherit the G-structure of J(**M**), the which in turn have the G-structure of **M**.

To the second question, is feasible to obtain such criteria by means of the use of tomography where the construction of the generalized functional required in the determination of geometrical invariants such as the curvature, requires the use of the Hermitian G-structure of the corresponding closed submanifolds used to the determination of the observable of the space in question. If we consider m = 1, and the Radon transform of the Dolbeault cohomology on functions fD(**M**), (that is to say, with coefficients in a sheaf of functions in D(**M**)) then we can to induce anything commutative algebraic diagram of a vector bundle TD'(**M**), in a commutative algebraic diagram to a complex bundle TD'(**M** RC). The determination for example, of the curvature to this case is not trivial, and to it is required the use of certain compact components of a reductive homogeneous space whose orbital classes are holomorphic equivariant embeddings in the image of the endomorphism J, of the Hermitian structure of **M**. This it bears to the nullity of the anti-symmetric part of the tensor of curvature Wij, (tensor de Weyl5), which is included with help of the Fröbenius theorem, since we consider the real analytic distributions of the bundle TJ(**M**). This establish a condition of integrability (VII. 1), that is a subtle pronouncement on the existence of a general integral to the equations of curvature of the Riemannian tensor to **M**, and that only depends of the Hermitian structure of the manifold **M**. Of this form all the differential geometry it could find their solution in an integral geometry, whose invariants could be

<sup>5</sup> tensor of Riemann = tensor of Ricci + tensor of Weyl. The tensor of Weyl represent the anti-symmetrical part of the tensor of curvature

fundament in Hermitian structures of a big collection of complex holomorphic manifolds. In this part result subtle the pronouncement of hypercomplex analysis on the fact of that hyperholomorphicity of several complex variables permits complex analyticity and real complex Riemannian manifolds, at least complex and to more Hermitians.

Some concrete examples we have in the identification of H, withC4, which induce an application of complex lines that can be quaternionized. Indeed, be H = x = x0+ ix1 + jx2 + kx3 xmR, the algebra of the quaternions. The elements xH, are explained of unique way as

$$\mathbf{x} = \mathbf{z}\mathbf{1} + \mathbf{i}\mathbf{z}\mathbf{2},\tag{\text{VII.6}}$$

considering the correspondences or equations of transformations

$$\mathbf{z}\mathbf{u} = \mathbf{x}\mathbf{i} + \mathbf{i}\mathbf{x}\mathbf{2}, \quad \mathbf{z}\mathbf{z} = \mathbf{x}\mathbf{s} + \mathbf{i}\mathbf{x}\mathbf{4} \tag{\text{VII.7}}$$

which permit identify to H, with C2. Also we can identify to H, with C4, considering the space EC4, that is to say; considering the map

$$\mathbf{E} \rightharpoonup \mathbf{H}^2,\tag{\text{VII.8}}$$

with rule of correspondence

78 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

cohomology to the given by the space H*n*1, q(C*n*/D, V)?

conform them and define them?

a fibration of **M** .

tensor of curvature

Which is the behavior of these integrals on a complex Riemannian manifold?

Why is the special behavior of these integrals on a complex Riemannian manifold? Is possible establish criteria to the study of certain geometrical properties (such as curvature or torsion) of a complex Riemannian manifold **M**, based in integrability on a holomorphic complex bundle in a manifold through of their differentiable projections? Such criteria raise certain class of integrals calculated in sections of holomorphic bundle of M, in an analogous

Is feasible in this sense to obtain a generalization of cohomology on G-invariant orbits, belonging to holomorphic bundles seated in reductive homogeneous spaces of such form that are obtained integrals of generalized functional to the geometrical observables of the space **M**, and their orbits? Will establish this cohomology some theory of integration on certain complex spaces, parallel to the obtained by the integral cohomology on homogeneous bundles evaluating integrals in a direct and global via on the vector flux that

The answer to first question save, clear this, the securing of harmonic functions , (it which no always is possible through of this path) is positive to the case of analytic functions and have that see with the structure a lo más Hermitiana of **M**, that can be establish constructing

Then the resulting integrals are calculated on orbits of sections of the bundle TJ(**M**), the which is usually inherit the G-structure of J(**M**), the which in turn have the G-structure of **M**. To the second question, is feasible to obtain such criteria by means of the use of tomography where the construction of the generalized functional required in the determination of geometrical invariants such as the curvature, requires the use of the Hermitian G-structure of the corresponding closed submanifolds used to the determination of the observable of the space in question. If we consider m = 1, and the Radon transform of the Dolbeault cohomology on functions fD(**M**), (that is to say, with coefficients in a sheaf of functions in D(**M**)) then we can to induce anything commutative algebraic diagram of a vector bundle TD'(**M**), in a commutative algebraic diagram to a complex bundle TD'(**M** RC). The determination for example, of the curvature to this case is not trivial, and to it is required the use of certain compact components of a reductive homogeneous space whose orbital classes are holomorphic equivariant embeddings in the image of the endomorphism J, of the Hermitian structure of **M**. This it bears to the nullity of the anti-symmetric part of the tensor of curvature Wij, (tensor de Weyl5), which is included with help of the Fröbenius theorem, since we consider the real analytic distributions of the bundle TJ(**M**). This establish a condition of integrability (VII. 1), that is a subtle pronouncement on the existence of a general integral to the equations of curvature of the Riemannian tensor to **M**, and that only depends of the Hermitian structure of the manifold **M**. Of this form all the differential geometry it could find their solution in an integral geometry, whose invariants could be

5 tensor of Riemann = tensor of Ricci + tensor of Weyl. The tensor of Weyl represent the anti-symmetrical part of the

$$\mathbf{z} \mid \rightarrow (\mathbf{w}\mathbf{o}, \mathbf{w}\mathbf{i}) := (\mathbf{z}\mathbf{o} + \mathbf{j}\mathbf{z}\mathbf{i}, \mathbf{z}\mathbf{z} + \mathbf{j}\mathbf{z}\mathbf{i}),\tag{\text{VII.9}}$$

that induce the application

$$\mathbf{P}^{3}\text{(C)}\rightarrow\mathbf{P}\text{(H}^{2}\text{)}\tag{\text{VII.10}}$$

that is to say, the corresponding quaternionic projective space of P3(C), is P(H2). This space also is identified in twistor geometry like the corresponding twistor space of H. Of this manage, the application (VII. 10), define a bundle with fiber P(C).

Thus exist a projection : P3(C) S4, (the sphere of dimension four (which is a Riemannian manifold)), whose fibers are the real lines of P3(C), that conforms the set of parameters of these straight lines in S4. Observing that P(H2), have a recovering by charts each one is diffeomorphic to R4, that is that P(H2)S4. Then considering open orbits in S4, and the homogeneous holomorphic bundle of lines (k),on P3(C), we have that the cohomological class H1((**x**), (-2)) **x**S4, define a homogeneous bundle of lines on S4. Then the subjacent orbital integrals in S4, in the integral operator defined on this cohomology it is becomes in integrals of line on sections of the complex holomorphic bundle of lines *L*, with cohomology H1(P(H2), *L*), subjacent in the Penrose integral transform. The Hermitian structure in the corresponding manifold is invariant under these transformations, Where this the tomography? The tomography is realized on sections of the complex holomorphic bundle of

lines of S4, through of the lines of P3(C). Is a form of re-write the vector Radon transform. If we consider the double fibration

$$\begin{array}{c} \text{(F)}\\ \mu \text{\(\mu\)} \\\\ \text{P}^{\circ} \text{(C)} \quad \text{G}\_{2,4} \text{(C)} \\\\ \pi \text{\(\mu\)} \\\\ \text{S}^{4} \end{array} \tag{\text{VII.11}}$$

The vector Radon transform that can be re-written be on the lines of S4, as the integral operator

$$\mathcal{R}(\mathbf{u}, \mathbf{L}) = \int \mathbb{L}\_{\leq -S} \mathbb{t} \mathbf{u} \, \mathsf{d} \mathbf{s} \, \mathsf{d} \mathbf{s}, \tag{\text{VII.12}}$$

finds their expression considering the flag manifold F, given by the space (L, P)L P C4, with the X-ray/John integral

$$\phi(\mathbf{P}) = \int \mathbb{L}\_{\subset \mathbf{P}} \mathfrak{Z}\_{|\mathbf{C}|} \mathbf{f},\tag{\text{VII.13}}$$

This integral comes of the application of the X-ray transform on lines of the projective bundle of lines in P3(C), followed of the John transform. These integrals correspond to the orbital integrals of the Penrose transform that recovers harmonic functions of an open of S4. Then in this sense, the cohomology H1((**x**), (-2)), not result very distinct of the cohomology H*n*1, q(C*n*/D, V), save the fact of that the Hermitian structure deduced of the structure at least complex of **M**, results a relevant characteristic to the re-interpretation of the Radon integrals in complex holomorphic bundles of the manifold **M**. In this part is suggested the response to third question.

The fourth question is a difficult problem of decode and describe, and is a problem that can be resolve with help of the solution to the problem of the irreducible unitary representations in Lie groups to the space G/L, since is necessary the construction of a adequate correspondence between invariant bilinear forms and unitary representations of an analytic cohomology to some cases of finite dimension non-covered through this analysis.

In this sense, is necessary develop a way to modify the Dolbeault cohomology to produce minimal globalizations in major grade that the maximal globalizations calculated by Wong to the case of finite dimension. This establishes a relation of duality between the maximal globalizations calculated through of the Dolbeault cohomology and the minimal calculated for certain cohomology to define. How can identify the dual topological space of a cohomological space of Dolbeault on the non-compact complex manifold?

To respect results of great useful the holomorphic G-invariant vector bundle and their corresponding bundles of lines associated to the (*n*, 0)-forms on vector topological spaces. Then is obtained a version of the Dolbeault cohomology called topological cohomology of Dolbeault and we construct representations of real reductive Lie groups G, beginning with the measurable complex flag manifold X = G/L, and using G-equivariant holomorphic bundle of lines on X, (conservation of the G-structure in submanifolds of X).

80 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

we consider the double fibration

C4, with the X-ray/John integral

suggested the response to third question.

(P) = <sup>L</sup>P3(

operator

lines of S4, through of the lines of P3(C). Is a form of re-write the vector Radon transform. If

The vector Radon transform that can be re-written be on the lines of S4, as the integral

*<sup>R</sup>*(**u**, L) = <sup>L</sup> S4**u** d**s**, (VII. 12) finds their expression considering the flag manifold F, given by the space (L, P)L P

This integral comes of the application of the X-ray transform on lines of the projective bundle of lines in P3(C), followed of the John transform. These integrals correspond to the orbital integrals of the Penrose transform that recovers harmonic functions of an open of S4. Then in this sense, the cohomology H1((**x**), (-2)), not result very distinct of the cohomology H*n*1, q(C*n*/D, V), save the fact of that the Hermitian structure deduced of the structure at least complex of **M**, results a relevant characteristic to the re-interpretation of the Radon integrals in complex holomorphic bundles of the manifold **M**. In this part is

The fourth question is a difficult problem of decode and describe, and is a problem that can be resolve with help of the solution to the problem of the irreducible unitary representations in Lie groups to the space G/L, since is necessary the construction of a adequate correspondence between invariant bilinear forms and unitary representations of an analytic

In this sense, is necessary develop a way to modify the Dolbeault cohomology to produce minimal globalizations in major grade that the maximal globalizations calculated by Wong to the case of finite dimension. This establishes a relation of duality between the maximal globalizations calculated through of the Dolbeault cohomology and the minimal calculated for certain cohomology to define. How can identify the dual topological space of a

To respect results of great useful the holomorphic G-invariant vector bundle and their corresponding bundles of lines associated to the (*n*, 0)-forms on vector topological spaces. Then is obtained a version of the Dolbeault cohomology called topological cohomology of

cohomology to some cases of finite dimension non-covered through this analysis.

cohomological space of Dolbeault on the non-compact complex manifold?

(VII. 11)

C) f, (VII. 13)

The hypercomplex analysis results useful only in the description of the unitary representations through of their realization and integral descriptions. The techniques of quaternions used describe some fields of some Lie algebras that are relevant and suggest the extension of those techniques to all the Lie algebras even of infinite dimension (Kac-Moody algebras, for example). Nevertheless, How it might be solved the problem of G-invariance on cycles and co-cycles of a complex Riemannian manifold of such lucky that it is not see affected the G-invariance of Hermitian structure to the symmetrical and non-symmetrical part of a complex Riemannian manifold, and be possible of be applying the methods of the hypercomplex analysis? Is possible establish a L2-cohomology of integral operators based exclusively in the subjacent Hermitian structure of a complex manifold or at least complex?

To this last question, we consider the realization of holomorphic sections of certain homogeneous holomorphic bundles on **M+**, through an integral operator belonging to an integral L2-cohomology. Consider the representation on L2-holomorphic4-forms, where L2,is defined with respect to invariant inner product

$$\ll \Theta, \kappa \gg = \left[ \mathfrak{M} + \Theta \wedge \kappa \right] \tag{VII.14}$$

seated in **M**, is the Flag bundle in C4,given by. The pre-image of **M+** , under the natural projection is ++meaning that the Hermitian form have the indicated signatures (+, ++, ++ y ++), when has been restricted to each part of the flag. This is one of six orbits of SU(2, 2), on . Like is well knower, each orbit correspond to a different type of discrete series. This correspondence is realized making the cohomology of the appropriate bundle of lines on the several orbits. This process is analogous to discussed with anteriority to exhibit the equivalences of cohomological spaces on fibered in S4,and P3(C), but now is to the spaces **M**, and . Then to the orbit ++, we note that the discrete series are H2(++, 6) (**M+**, 4). In this part result relevant the contribution of the generalized conform structure implicit on G, where maximal submanifolds of G, are horo-spheres, that is to say, bi-lateral components of maximal nilpotent subgroups. The transformation of a function to their integrals on horospheres play an important role in representation theory, such as is mentioned in this example. Their inversion is essential in the derivation of the Plancherel formula. Then, Is possible that SU(2, 2), can be represented on aL2-cohomology based in integral operators on each one of their orbits? The answer is positive, and is possible through the twistor transform that is a SU(2, 2)-equivariant isomorphism, to know

$$\mathcal{T} \colon \mathcal{H}^2(\mathbb{S}^{++-}|, \mathcal{Q}^\diamond) \to \mathcal{H}^3(|\mathfrak{I} - \mathfrak{I}^{++-}|, \mathcal{Q}),\tag{VII.15}$$

In the best of the cases, we could find an integral transform to each one of their orbits and for this way to find the complete discrete series. The true is that such isomorphism it is given in the level of minimal K-types of the representations [48].

This situation explain the importance of the G-structure of the manifold **M**, in the role on the determination of properties and invariants that can be appreciated on their orbits when this are K-invariants. But this can be studied for the via of the complex geometry of the vector Gbundles and the orbitalization of the reductive homogeneous spaces that are obtained in the process of reduction of G/H, which is not more than a consequence of the reduction of the group of holonomy of **M**.

We consider a K-invariant connection of reductive homogeneous space G/G0, corresponding to a Stein manifold MD, subjacent to the complex Riemannian manifold **M**, (that is to say, we consider a close and compact orbit of **M**). Then G0 = K.

To it, we consider the cohomology of De Rham of the exterior subalgebras (V),and (V\*), of the vector bundle E **M**, and we construct the K-invariant connection on the vector bundle P **M**, (the vector G-bundle) that be an affine connection of **M**(2). For orbitalization

$$\text{G/D } \bigotimes \text{G/P} \bigotimes \text{G/G}\_{\emptyset} \text{ G}\_{\emptyset} = \text{K}, \tag{\text{VII.16}}$$

we have that can to built a smooth embedding in J(**M**), of flag submanifolds F, being G0 orbits (is to say, K-orbits) in J(**M**). But this always is possible for the reduction of the holonomy group of M, and that M, be a connect and inner locally symmetric complex Riemannian manifold, that is to say, that the tensor of Nijenhuis satisfies on the corresponding sub-bundle J(**M**), of **M**, that R*I* M(j) = 0. This establishes the conditions of integrability to the symmetric part. Thus our sub-bundle have integrable structure and as the space considered is a manifold at least complex, one can to find complex submanifolds of J(**M**), to which the K-orbits are flag manifolds of the G-structure of **M**. From the point of view very particular of one study of the integral curvature [16, 17], this allow us to affirm that the contribution to the curvature on K-orbits only succeed in symmetric spaces (null Weyl tensor and null Ricci tensor), since the unique integrable complex submanifolds that can be realized like K-orbits in symmetric inner simply connect **M**, and of compact type are the flags G-orbits. In effect, for the arguments given to the second question, we observe that the use of the structure at least complex of the holomorphic bundle TJ(**M**), to the determination of fundamental 2-forms that evaluated in certain spaces result nulls, is a sufficient criteria to the integrability in **M**.

This condition establishes a criteria of *integrability* of the equations of curvature on the Riemannian manifold at least complex **M**. The class is of integrals that determine the functional that resolves the curvature equations belonging to integral operators L2 cohomologyon submanifolds of **M**.

This geometrically is re-interpreted: the component of flat space in a Riemannian manifold at least complex result be the component with structure at least Hermitian, and given the local structure of Hilbert space that is subjacent in all Riemannian manifold, which can extend it to all the distributions, and in particular to the horizontal of the bundle TJ(**M**), result by the Fröbenius theorem, that the corresponding curvature tensor to the structure at least complex I, given by TJ(**M**) is vanished in said component, and not only that, also the Hermitian forms of the local structure of Hilbert space induces a symmetrical structure, and then in all component of flat space of **M**, are satisfied these conditions, transmitting the Gstructure on the corresponding flag spaces (submanifolds of the corresponding complex holomorphic bundle TJ(**M**)), that in a adequate physical interpretation result be appropriates to the model and study of the phenomena to Max Planch lengths, writing and developing an integral operator cohomology based in integral operators whose integrals of contour are cohomological functional of energy states (cocycles).

82 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

consider a close and compact orbit of **M**). Then G0 = K.

corresponding sub-bundle J(**M**), of **M**, that R

sufficient criteria to the integrability in **M**.

cohomologyon submanifolds of **M**.

group of holonomy of **M**.

This situation explain the importance of the G-structure of the manifold **M**, in the role on the determination of properties and invariants that can be appreciated on their orbits when this are K-invariants. But this can be studied for the via of the complex geometry of the vector Gbundles and the orbitalization of the reductive homogeneous spaces that are obtained in the process of reduction of G/H, which is not more than a consequence of the reduction of the

We consider a K-invariant connection of reductive homogeneous space G/G0, corresponding to a Stein manifold MD, subjacent to the complex Riemannian manifold **M**, (that is to say, we

To it, we consider the cohomology of De Rham of the exterior subalgebras (V),and (V\*), of the vector bundle E **M**, and we construct the K-invariant connection on the vector bundle P **M**, (the vector G-bundle) that be an affine connection of **M**(2). For orbitalization

we have that can to built a smooth embedding in J(**M**), of flag submanifolds F, being G0 orbits (is to say, K-orbits) in J(**M**). But this always is possible for the reduction of the holonomy group of M, and that M, be a connect and inner locally symmetric complex Riemannian manifold, that is to say, that the tensor of Nijenhuis satisfies on the

*I*

integrability to the symmetric part. Thus our sub-bundle have integrable structure and as the space considered is a manifold at least complex, one can to find complex submanifolds of J(**M**), to which the K-orbits are flag manifolds of the G-structure of **M**. From the point of view very particular of one study of the integral curvature [16, 17], this allow us to affirm that the contribution to the curvature on K-orbits only succeed in symmetric spaces (null Weyl tensor and null Ricci tensor), since the unique integrable complex submanifolds that can be realized like K-orbits in symmetric inner simply connect **M**, and of compact type are the flags G-orbits. In effect, for the arguments given to the second question, we observe that the use of the structure at least complex of the holomorphic bundle TJ(**M**), to the determination of fundamental 2-forms that evaluated in certain spaces result nulls, is a

This condition establishes a criteria of *integrability* of the equations of curvature on the Riemannian manifold at least complex **M**. The class is of integrals that determine the functional that resolves the curvature equations belonging to integral operators L2-

This geometrically is re-interpreted: the component of flat space in a Riemannian manifold at least complex result be the component with structure at least Hermitian, and given the local structure of Hilbert space that is subjacent in all Riemannian manifold, which can

G/D G/P G/G0, G0 = K, (VII. 16)

M(j) = 0. This establishes the conditions of

Now well, Can these induce it the properties of the complex integral operators determined in the frame of a Hermitian structure of an Universe given for a complex Riemannian manifold to integral operators on a "microscopic" structure to the study of complex submanifolds more little that the flag manifolds, for example to null surfaces and null curves? Can be extended these operators to an integral operators cohomology isomorphic to the given by the resolution of an integral transform defined on homogeneous holomorphic bundles?

In this last point (on physic of particles) is induced to that the integral operators cohomology can be induced to a cohomology on diagrams of Feynman type. Then to this case arises the natural question of the cohomological classes to differential forms given , in the study on resolution of equations of vector fields ¿How re-emplace the integrals of contour

$$\downarrow\_{\kappa} \text{oj},\tag{\text{VII.17}}$$

by a conformal scheme (given, for example by MD G/K)) that helps to calculate these integrals of contour to these phenomena in the spaces of F, and that not requires the evaluation of the vector field in orbital submanifolds like null curves, curves, minimal surfaces or null quadric?

For other side, we consider the problem of the integrability on complex Riemannian manifolds and their relation with the curvature tensor Rjk, to the conformal component of **M**.

Let G/L, be with L, a flag manifold. We consider the Radon transform on hyperplanes of arbitrary co-dimension of the Grassmannian Gn,m. Then to this particular case the integral operators cohomology on the complex flags is isomorphic to an integral operators cohomology on submanifolds of a complex maximal torus. Then the case of a symmetric connection to the space-time stay completely results, and can be computed the corresponding integrals, of this manage as study the space-time through of gauge fields on the which we can to calculate integrals on geodesics (for example, light geodesics to the determination of curvature of the space-time [18]). This defines an integral operators L2 cohomology on orbits of the complex torus T. The corresponding unitary representations are determined representations for quotients of the L2-cohomology

$$\bigoplus\_{\mu:\mathbb{C}\to\mathrm{Homc}(\mathrm{H}\_{\mu}\text{ L}^{2}(\mathrm{G}/\mathrm{L}))} \mathsf{GrH}(\mathrm{G}/\mathrm{L};\mathfrak{o}(\mathfrak{O}\_{\mu})),\tag{\text{VII. 18}}$$

If we consider L = ZG(T), the holonomic bundle that reduces the orbits of the homogeneous space G/K, to the orbits of the maximal torus in g/t,the which is a subspace of the Lie algebra of G/K, as can to see in the "orbitalization" of G/H

$$\begin{array}{l} \mathsf{J}(\mathsf{R}^{2n}) \\ \triangleq \\ \mathsf{g} \land \mathsf{t} \in \mathsf{g} \land \mathsf{p} \in \mathsf{g} \land \mathsf{s} \bullet (2\mathsf{n}) \subset \mathsf{g} \land \mathsf{l}, \\ \end{array} \tag{\text{VII. 19}}$$

then the quotiens are exhibited.

Here a flag manifold comes given like F = G/C(T), where C(T), is a centralizator of the complex torus T. Then the center ZG(T), induce a structure in F, such that ZG = R **<sup>M</sup>**(j)End(2T\*(**M**) ) R **<sup>M</sup>**(j) = 0, and this consist of a finite number of connect components on each a of the G that act transitively.

Also any flag G-manifold6 is realized like such orbit to some **M**. The requirement of the transitive action of G on the orbits is indispensable to the hypothesis of special isotropy that always is considered in the studies of the universe in the construction of an integral operators cohomology to the curvature of the space-time, since the integrals of curvature are defined on G-invariant orbits and will must be calculated by reduction of the corresponding holonomy group on K-invariant orbits.

The condition that is established with R **<sup>M</sup>**(j) = 0,is the condition of integrability7 of the symmetric connection that define the field equations in this case.

For filtration it can be possible to pass of the calculation of integrals on geodesic of the space-time to Feynman integral or integrals on strings considering the orbit of the gfiltration corresponding to the reductive homogeneous space of **M**. The process of reduction of the holonomy group of the structure of the vector G-bundle of **M** helps us to obtain reductive homogeneous spaces of G/H with inherit orbits the G-structure of the big space and therefore of **M**.

<sup>6</sup> Flag G-manifolds of **M**. G act transitively on such complex submanifolds.

<sup>7</sup> **Proposition (Burstall).** Be jJ(**M**). Then G j, is a submanifold at least complex of J(**M**) on the which , is integrable if and only if j, fall into of the null set of the Nijenhuis tensor R **M**(j).

**Theorem VII. 1.** (F. Bulnes). The K-invariance given by the G-structure SG(**M**) of **M** complex and holomorphic is induced to each closed submanifold given by the flag manifolds of the corresponding vector holomorphic G-bundle. The integral operators cohomology given on such complex submanifolds is also equivalent to the integral cohomology on submanifolds of a complex maximum torus.

84 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

are determined representations for quotients of the L2-cohomology

2n

)J 

R

(

of G/K, as can to see in the "orbitalization" of G/H

components on each a of the G that act transitively.

then the quotiens are exhibited.

**<sup>M</sup>**(j)End(2T\*(**M**) ) R

holonomy group on K-invariant orbits.

and therefore of **M**.

The condition that is established with R

symmetric connection that define the field equations in this case.

6 Flag G-manifolds of **M**. G act transitively on such complex submanifolds.

and only if j, fall into of the null set of the Nijenhuis tensor R

<sup>7</sup> **Proposition (Burstall).** Be jJ(**M**). Then G j, is a submanifold at least complex of J(**M**) on the which

R

G HomG(H, L2(G/L)) Hs

determination of curvature of the space-time [18]). This defines an integral operators L2 cohomology on orbits of the complex torus T. The corresponding unitary representations

If we consider L = ZG(T), the holonomic bundle that reduces the orbits of the homogeneous space G/K, to the orbits of the maximal torus in g/t,the which is a subspace of the Lie algebra

g / t g / p g / so g / h,

Here a flag manifold comes given like F = G/C(T), where C(T), is a centralizator of the complex torus T. Then the center ZG(T), induce a structure in F, such that ZG =

Also any flag G-manifold6 is realized like such orbit to some **M**. The requirement of the transitive action of G on the orbits is indispensable to the hypothesis of special isotropy that always is considered in the studies of the universe in the construction of an integral operators cohomology to the curvature of the space-time, since the integrals of curvature are defined on G-invariant orbits and will must be calculated by reduction of the corresponding

For filtration it can be possible to pass of the calculation of integrals on geodesic of the space-time to Feynman integral or integrals on strings considering the orbit of the gfiltration corresponding to the reductive homogeneous space of **M**. The process of reduction of the holonomy group of the structure of the vector G-bundle of **M** helps us to obtain reductive homogeneous spaces of G/H with inherit orbits the G-structure of the big space

**<sup>M</sup>**(j).

n

**<sup>M</sup>**(j) = 0, and this consist of a finite number of connect

**<sup>M</sup>**(j) = 0,is the condition of integrability7 of the

, is integrable if

2

(G/L; o()), (VII. 18)

(VII. 19)

To demonstrate this result it is necessary to demonstrate some fundamental previous facts. For we consider it the following theorem due to (Burstall and Rawnsley, 1989), that is;

**Theorem VII. 2.** Let **M** = G/K, be a symmetrical internal simply related connect Riemanniana manifold and of compact type. Then

$$Z = \langle \mathcal{R}\_j^M(\mathbf{j}) \in \text{End}(\mathcal{G}^2 \mathcal{T}^\*(\mathbf{M}) \otimes \cdot \, : \, \, \rangle \mid \mathcal{R}\_j^M(\mathbf{j}) = 0 \rangle,\tag{VII.20}$$

which consists of a related finite number of components on each one of those which G acts transitively. Any flag G-manifold is also carried out as such orbit for some **M**.

The requirement of the transitive action of G, on the orbits is for example, more important for the hypothesis of isotropy space in the construction of an integral operators cohomology for the curvature of the space-time, the integral of curvature should be determined since on G-invariants orbits and they will be calculated by reduction of the group of corresponding holonomy on K-invariants orbits.

Let us consider the space of classes G/C(T), which admits Kählerian complex invariant structures, that is to say; we can consider the G-structure K-invariant SG(**M**), of **M**. To fix ideas we use the G-structure exactly as complex Kählerian of G/C(T). The complexified of the group G, given by G**C**, acts transitively on G/C(T), for bi-holomorphism with parabolic groups as stabilizers. Reciprocally if P G**C**, is then a parabolic subgroup the action of G, on G**C**/P, it has more than enough it is transitive and G P, is the centralizing of a torus in G. For the infinitesimal situation (VII. 19) let F = G/C(T), be a flag manifold and let o the origin in F. Considers the decomposition of the Lie algebra g**C**,

$$
\mathfrak{g}\mathfrak{c} = \mathfrak{h}\oplus\mathfrak{m},\tag{\text{VII.21}}
$$

with mTo(F), and h, the Lie algebra of the stabilizer of o, in G. Then the complex invariant structure of F induces a decomposition adh-invariant of m**C**, in those (1, 0), and (0, 1), spaces

$$\mathfrak{mc} = \mathfrak{m}^\* \oplus \mathfrak{m}^-,\tag{\text{VII.22}}$$

with m+, m+m+, for integrability. It can demonstrate himself that m+, and m are nilpotent subalgebras g**C**, and in fact h**C**m is a parabolic algebra of g **C**, with radical nilpotente m . If P, is then the corresponding parabolic subgroup of G**C**/P, it is the stabilizer of o, and we obtain a bi-holomorphism among the complex space of classes G**C**/P, and the flag manifold F.

Reciprocally, be P G**C**, a parabolic subgroup with Lie algebra p, and be n, the one conjugated of the radical nilpotent of p, (with regard to the real form g). Then H = G P, is the centralizing of a torus and we have the orthogonal decomposition (with regard to the Killing form, that is to say, the *ad*H-decomposition Hh):

$$
\mathfrak{p} = \mathfrak{h}c \oplus \overline{\mathfrak{u}} \quad \mathfrak{k} := \mathfrak{h}c \oplus \mathfrak{w} \oplus \overline{\mathfrak{u}} \tag{VII.23}
$$

which defines a structure complex invariante G/H, has more than enough carrying out the bi-holomorphism with G**C**/P. Considering an orbitalization of G/H, through the n-filtration of flag manifolds in G**C**/P,

$$\mathfrak{h} = \mathfrak{m}\_{k+1} \subset \mathfrak{U} \mathfrak{h} \subset \dots \subset \mathfrak{U}^{\mathfrak{l}} = \mathfrak{n}\_{\mathfrak{l}} \tag{\text{VII. 24}}$$

of n, defined for

$$\mathfrak{m} = \{\mathfrak{n}, \mathfrak{m}\\_1\},\tag{\text{VII.25}}$$

We realize orthogonalization this filtration using the form of Killing and we have that

$$\mathfrak{gl} = \mathfrak{ll}^{\perp} \mathfrak{l} + \mathfrak{n} \cap \mathfrak{W} \tag{\text{VII.26}}$$

to i 1, and extended to a g**C**, decomposition making correspond g0 = h**C** = (gp)**C**, and gi = gi, for i 1. Then

$$
\mathfrak{gc} = \sum \mathfrak{g},\tag{\text{VII.27}}
$$

it is an orthogonal decomposition with

$$\mathfrak{p} = \sum\_{l \le \cdot \le 0} \mathfrak{p}\mathfrak{p}\_{\prime\prime} \mathfrak{n} = \sum\_{\succ \ge 0} \mathfrak{p}\mathfrak{p}\_{\prime\prime} \tag{\text{VII.28}}$$

The crucial property of this decomposition is that

$$\{\mathfrak{a}\land\mathfrak{a}\dagger\}\Box\mathfrak{a}\dagger\}\tag{\text{VII. 29}}$$

which can be proven demonstrating the existence of an element h, with the property that each iad, has eigenvalue(-1)I, has more than enough gi. This element (necessarily only since g, is semi-simple) it was demonstrated that it exists in (Burstall and Rawnsley, 1989), which called it canonical element of p. Since *ad* has eigenvalues in (-1)・, Ad exp, is an involution of g, the one which exponencing to obtain an internal involution of G, and therefore an internal symmetrical space G/K, where K = (G)0. Clearly, K, has the Lie algebra given for

$$
\mathbf{k} = \mathbf{u} \cap \sum \mathbf{p}\_t \tag{VII.30}
$$

and that H contains, of where we obtain the homogeneous fibration

Chapter VII – Some Examples of Orbital Integrals Into Representation Theory on Field Theory ... 87

$$\text{G/H} \rightarrow \text{G/K} \tag{\text{VII.31}}$$

of our flag manifold on our symmetrical space. This will be important to establish an isomorphism among the cohomology of the integrals on G/H, and the corresponding integral submanifolds that are had in the ker-equations, that like it has been said, they only exist as integral of those -equations in **M**, (**M**, integrable) if R*<sup>I</sup>* M(j) = 0. The filtration is also essentially only. The only ambiguity in the prescription is that the distinct points in the symmetrical space can have the same stabilizer K (that is to say, antipodal points in a sphere). Nevertheless the number of such points is finite and we can give a finite number of such fibrations. These fibrations will call them canonical of F. Then it is necessary to enunciate the following result (Burstall and Rawnsley, 1989):

**Theorem VII. 3.** Let F = GC/P, be a flag manifold. Then an only internal symmetrical Gspace always exists **M**, associated to F, with a number of homogeneous fibrations F M.

For the demonstration of this result, let us consider a conjugation class p, of g**C**, and the election in a real way of g. Now what it is necessary to see is that each flag manifold is a fiber on each internal symmetrical space. Reciprocally, this it is the road to demonstrate that each space symmetrical intern is the objective of the canonical fibrations of at least a flag manifold. This finally will give an idea of how the geometry of the complex holomorphic bundle is J(**M**).

Be

86 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Killing form, that is to say, the *ad*H-decomposition Hh):

of flag manifolds in G**C**/P,

of n, defined for

gi = gi, for i 1. Then

given for

it is an orthogonal decomposition with

The crucial property of this decomposition is that

Reciprocally, be P G**C**, a parabolic subgroup with Lie algebra p, and be n, the one conjugated of the radical nilpotent of p, (with regard to the real form g). Then H = G P, is the centralizing of a torus and we have the orthogonal decomposition (with regard to the

p = h**C**n, g・ = h**C**nn, (VII. 23)

which defines a structure complex invariante G/H, has more than enough carrying out the bi-holomorphism with G**C**/P. Considering an orbitalization of G/H, through the n-filtration

n1 = n, n<sup>i</sup><sup>1</sup>, (VII. 25)

gi = ni + 1ni, (VII. 26)

to i 1, and extended to a g**C**, decomposition making correspond g0 = h**C** = (gp)**C**, and

<sup>g</sup>**C**<sup>=</sup><sup>g</sup>i, (VII. 27)

<sup>p</sup> = <sup>i</sup><sup>0</sup>gi, n = i> 0gi, (VII. 28)

which can be proven demonstrating the existence of an element h, with the property that each iad, has eigenvalue(-1)I, has more than enough gi. This element (necessarily only since g, is semi-simple) it was demonstrated that it exists in (Burstall and Rawnsley, 1989), which called it canonical element of p. Since *ad* has eigenvalues in (-1)・, Ad exp, is an involution of g, the one which exponencing to obtain an internal involution of G, and therefore an internal symmetrical space G/K, where K = (G)0. Clearly, K, has the Lie algebra

k = ggi, (VII. 30)

and that H contains, of where we obtain the homogeneous fibration

We realize orthogonalization this filtration using the form of Killing and we have that

0 = nk + 1nkn1 = n, (VII. 24)

gi, gjgi + j, (VII. 29)

p: F **M**,

a canonical fibration. For construction, the fibers of p are complex submanifolds of F, and this facilitates us to define a fiber map

$$\mathbf{i}\_{\mathbb{P}} \colon \mathbf{F} \to \mathbf{J}(\mathbf{M}),\tag{\text{VII.32}}$$

as it continues: In fF, has the orthogonal decomposition of TfF, in horizontal and vertical subspaces, both of which are then low invariants the complex structure of F. Then dp restrict giving an isomorphism of the horizontal part with Tp(f)**M**, and therefore it induces a structure at most Hermitiana Tp(f)**M**, it has more than enough, this is; ip(f)(f)Jp(f)**M**. Such a construction is possible provided we have a Riemannian submersion of a Hermitian manifold with submanifolds like fibers. For this case we consider:

**Proposition VII. 1.** ip: F J(**M**), it is a fitting holomorphic G-equivariant.

It proves: (Burstall, 1987).

This implies that ip(F), it is at most a complex submanifold of J(**M**), on which **I**, is integrable. For a corollary of the Burstall theorem of the proposition II. 1., we have:

**Corollary VII. 1.** ip(F) it is a G-orbit in Z J(**M**). It proves in, (Burstall, 1987).

In particular this guarantees that the tensor space of Z is I don't empty. The reciprocal of this corollary one is also true.

The following result due to Rawnsley determines which the fitting is given by the proposition II. 1., and the theorem II. 3:

**Theorem VII. 4.** If jZ J(**M**) then G j, is canonically a flag fibreded manifold **M**. It has more than enough in fact, G j = ip(F) for some fibration p: F **M** of a flag manifold F.

Their demonstration is continued of the observation that in (j), we have the symmetrical decomposition

$$
\mathfrak{g} = \mathfrak{k} \oplus \mathfrak{g} \tag{VII.33}
$$

with qTp(f)**M**. Ifq,is the (0, 1)-space for j, then

$$[\mathfrak{q}^-, \mathfrak{q}^-] \oplus \mathfrak{q}^- \tag{\text{VII. 34}}$$

it is the nilpotent radical of the parabolic subalgebra p. Then it is demonstrated that G ∙ j, is equivariantly bi-holomorphic to the corresponding flag manifold such G**C**/P, and like it is described in the theorem. II. 4. For more details see (Burstall and Rawnsley, 1989).

Now complete the demonstration of the theorem. II. 2. We have to see that each canonical fibration of a flag manifold will give to a G-orbit in Z, for some internal symmetrical G-space **M**, and that all such orbits come from the same procedure. But fixed G, stops, it exist alone a finite number of flag manifolds of bi-holomorphism of this type. These are in biyective correspondence with the conjugated classes of parabolic subalgebras of g**C,** and each flag manifold admits a finite number of canonical fibrations. Then Z, is made up of a finite number of G-orbits all of which are closed and the theorem II. 2., it is continued. Then since each one of these G-orbits exists like a K-orbit of the space of classes G/K, with tensor of curvature of null Nijenhuis then each flag submanifold is a K-orbit of the vector and holomorphic G-bundle of the 2n-dimensional irreducible symmetrical Riemanniana manifold J(**M**), reason why the theorem II. 1., is continued.

## **Orbital Integrals on Cuspidal Forms**

## **VIII. 1. Introduction**

88 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

corollary one is also true.

decomposition

proposition II. 1., and the theorem II. 3:

with qTp(f)**M**. Ifq,is the (0, 1)-space for j, then

manifold J(**M**), reason why the theorem II. 1., is continued.

**Corollary VII. 1.** ip(F) it is a G-orbit in Z J(**M**). It proves in, (Burstall, 1987).

In particular this guarantees that the tensor space of Z is I don't empty. The reciprocal of this

The following result due to Rawnsley determines which the fitting is given by the

**Theorem VII. 4.** If jZ J(**M**) then G j, is canonically a flag fibreded manifold **M**. It has more than enough in fact, G j = ip(F) for some fibration p: F **M** of a flag manifold F.

Their demonstration is continued of the observation that in (j), we have the symmetrical

g = k q, (VII. 33)

it is the nilpotent radical of the parabolic subalgebra p. Then it is demonstrated that G ∙ j, is equivariantly bi-holomorphic to the corresponding flag manifold such G**C**/P, and like it is

Now complete the demonstration of the theorem. II. 2. We have to see that each canonical fibration of a flag manifold will give to a G-orbit in Z, for some internal symmetrical G-space **M**, and that all such orbits come from the same procedure. But fixed G, stops, it exist alone a finite number of flag manifolds of bi-holomorphism of this type. These are in biyective correspondence with the conjugated classes of parabolic subalgebras of g**C,** and each flag manifold admits a finite number of canonical fibrations. Then Z, is made up of a finite number of G-orbits all of which are closed and the theorem II. 2., it is continued. Then since each one of these G-orbits exists like a K-orbit of the space of classes G/K, with tensor of curvature of null Nijenhuis then each flag submanifold is a K-orbit of the vector and holomorphic G-bundle of the 2n-dimensional irreducible symmetrical Riemanniana

, (VII. 34)

q , q q

described in the theorem. II. 4. For more details see (Burstall and Rawnsley, 1989).

To can to study the cuspidal forms and determine their orbital integrals on whose cuspidal forms is necessary to do the use of the orbital classes of the group SL(2, R), and of their corresponding Lie algebra sl(2, R). To them, we identify to a Lie group L, locally isomorphic to SL(2, R), and whose Lie algebra corresponding it is identify isomorphically with sl(2, R). The orbital integrals as obtained will be orbital integrals of the minimal parabolic subgroup in G. Applying the Harish-Chandra to the obtained functions of such orbital integrals we will obtain the orbital integrals on cuspidal forms.

We begin this section with some computes on the group SL(2, R). All the results it is will must to Harish-Chandra.

Let L, be a locally connect Lie group isomorphic to SL(2, R). We identify the corresponding Lie algebra of L, with the Lie algebra sl(2, R).

## **VIII. 2. Exposition**

Let T, be A0, and N, connect subgroups of G, whose corresponding Lie algebra are the spaces

$$\mathbf{R}\mathbf{h} = \langle \mathbf{h}\mathbf{e}\mathbf{t} \mid \mathbf{h} = \begin{bmatrix} 0 & 1 \\ -1 & 0 \end{bmatrix} \rangle, \quad \mathbf{R}\mathbf{H} = \langle \mathbf{H}\mathbf{e}\mathbf{a}^{0} \mid \mathbf{H} = \begin{bmatrix} 1 & 0 \\ 0 & -1 \end{bmatrix} \rangle,\tag{\text{VIII. 1}}$$

and

$$\mathbb{R}X = \{ \mathbf{X} \in \mathfrak{M} \mid \mathbf{X} = \begin{bmatrix} 0 & 1 \\ 0 & 0 \end{bmatrix} \},\tag{\text{VIII.2}}$$

Let A, be a Cartan subgroup of G, corresponding to Lie algebra a = a0 t. Be R, the map

$$\mathbf{t} \colon \mathbb{R} \to \mathbb{T}, \tag{\text{VII. 3}}$$

whose rule of correspondence is

$$\begin{array}{cccc} \Theta \mid \rightarrow \exp(\pi \theta \mathbf{h}) , & \forall \ \mathbf{h} \in \mathsf{t} \; \; \; \Theta \in \mathbb{R} \end{array} \tag{VIII. 4}$$

whose rule explicit is

$$\mathbf{t}(\theta) = \exp(\pi \theta \mathbf{h}),\tag{VIII.5}$$

If fC(G), then Ff T(t()) = Ff T(). In effect, by definition

$$\mathrm{Fr}^{\mathrm{r}}(\mathrm{t}(\theta)) = \Delta(\mathrm{t}(\theta)) \Big| \mathrm{cf}(\mathrm{gt}(\theta)\mathrm{g}^{-1}) \mathrm{dg} = \Lambda(\exp(\pi \theta \mathrm{h})) \Big| \mathrm{cf}(\mathrm{gexp}(\pi \theta \mathrm{h}) \mathrm{g}^{-1}) \mathrm{dg} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d} \Big| \mathrm{d}$$

But

$$\mathbf{t}(\theta) = \exp \pi \theta \begin{bmatrix} 0 & 1 \\ -1 & 0 \end{bmatrix} = \exp \begin{bmatrix} 0 & \pi \theta \\ -\pi \theta & 0 \end{bmatrix} = \begin{bmatrix} 0 & e^{\pi \theta} \\ -e^{-\pi \theta} & 0 \end{bmatrix} \tag{\text{VIII. 7}}$$

Then

$$\begin{aligned} \, \_\Delta \text{Tr} \begin{bmatrix} 0 & e^{\pi \theta} \\ -e^{-\pi \theta} & 0 \end{bmatrix} \int\_{\mathbb{C}} \text{ cf} \mathbf{g} \begin{bmatrix} 0 & e^{\pi \theta} \\ -e^{-\pi \theta} & 0 \end{bmatrix} \text{g}^{-1} \text{)dg} & \quad ^{\text{(c)}} \text{f} (\text{gg}^{-1}) \text{dg} \\\\ = 2 \text{sen } \pi \theta \int\_{\mathbb{C}} \text{cf} \mathbf{g} \pi \begin{bmatrix} 0 & e^{\theta} \\ -e^{-\theta} & 0 \end{bmatrix}, \text{ g}^{-1} \text{)dg} = \text{F} (\theta) \end{aligned}$$

Note that T'' = T' = t()R\Z. In effect, for one side T'' = T K'', with

$$\mathcal{K}^{\prime\prime} = \{ \mathbf{k} \in \mathbb{K} \mid (\mathbf{I} - \operatorname{Ad}(\mathbf{k})|\_{\mathbb{V}}) \neq 0 \},$$

Since K'', is maximal in G''[K], of G, we have that T K''= tT(I Ad(t)t'') 0. But since to each K'' exist k1Z, such that (I ad(H)t'')k1= 0, and given that g = t p a, then exist k2R\Z, such that (I ad(H)p)k2= 0. Thus, and in particular to the classes Rh, in t,

$$\mathbf{T}^{\prime\prime} = \mathbf{T}^{\prime} = \{ \mathbf{t}(\boldsymbol{\theta}) \,|\, \boldsymbol{\theta} \in \mathbb{R} \,\backslash \boldsymbol{\Sigma} \}. \tag{\text{VIII.8}}$$

A direct computing, using the formulas in 7. 4. 3 and 7. 4. 4., give

$$1. \qquad \mathsf{Fr}(\theta) = 2 \text{isen } \pi \theta \stackrel{\Rightarrow}{\int} \text{senh}(\text{2t}) \text{f}(\exp(\theta \pi \begin{bmatrix} 0 & \mathsf{e}^{2t} \\ -\text{e}^{-2t} & 0 \end{bmatrix}) \text{d}\mathfrak{t}, \text{ etc.}$$

Let u = cosh (2t), be. Then we have to

$$\text{2.}\qquad\text{Fr}(\theta) = 2\text{i}(\text{sen }\pi\theta \,|\,\text{π}\theta \,|) \int\_{\pi\theta}^{\infty} \text{f}(\exp(\text{sgn}\theta \begin{bmatrix} 0 & \text{z}(\theta,\text{u}) \\ \text{z}(\theta,-\text{u}) & 0 \end{bmatrix}) \,\text{d}\mathbf{u},$$

with z(, u) = u + (u2 ()2)1/2. The two values (mod 2Z), to which we have singularities of type "jump" are = 0, or 1. The before formula show that this is not a singularity "jump" to = 1. We concentrate to case = 0. Likewise, (2) implies that:

$$\text{f(3)} \underset{\theta \longrightarrow 0^{+}}{\text{lim}} \text{F}\_{\text{f}}(\theta) = 2 \text{i} \int\_{0}^{\eta} \text{f}(\exp(2\text{Xu})) \text{du}, \qquad \text{and} \qquad \lim\_{\theta \longrightarrow -\text{0}^{-}} \text{F}\_{\text{f}}(\theta) = 2 \text{i} \int\_{0}^{\eta} \text{f}(\exp(-2\text{Xu})) \text{du}.$$

This implies that:

90 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

T(t()) = Ff

t() = exp 0 1

<sup>0</sup>

*e*

 

t() = exp(h), (VIII. 5)

0

0 *e*

<sup>0</sup> , <sup>0</sup> *e*

 

K'' = kK(I Ad(k)p) 0, Since K'', is maximal in G''[K], of G, we have that T K''= tT(I Ad(t)t'') 0. But since to each K'' exist k1Z, such that (I ad(H)t'')k1= 0, and given that g = t p a, then exist

2t

with z(, u) = u + (u2 ()2)1/2. The two values (mod 2Z), to which we have singularities of type "jump" are = 0, or 1. The before formula show that this is not a singularity "jump" to

2t

0 z(θ, u) f(exp(sgn<sup>θ</sup> )) z(θ, u) 0 du,

 

0 e

e 0

 

T(t()) = (t())G f(gt()g1)dg = (exp(h))G f(gexp(h)g1)dg, (VIII. 6)

0   <sup>0</sup> , <sup>0</sup> *e*

 

(VIII. 7)

*e*

g)dg G f(gg)dg

T'' = T' = t()R\Z. (VIII. 8)

g)dg = Ff(),

T(). In effect, by definition

= exp

G f(g <sup>0</sup>

*e*

*e*

 

1 0 

0 *e*

= 2sen G f(g

A direct computing, using the formulas in 7. 4. 3 and 7. 4. 4., give

senh(2t)f(exp(θπ ))

dt,

Note that T'' = T' = t()R\Z. In effect, for one side T'' = T K'', with

k2R\Z, such that (I ad(H)p)k2= 0. Thus, and in particular to the classes Rh, in t,

whose rule explicit is

If fC(G), then Ff

Ff

1. Ff() = 2isen

2. Ff() = 2i(sen

0

Let u = cosh (2t), be. Then we have to

πθ

= 1. We concentrate to case = 0. Likewise, (2) implies that:

But

Then

$$\text{f(4)} \mathop{\text{lim}}\_{\boldsymbol{\Theta} \xrightarrow{\text{prod}} \boldsymbol{\mathcal{F}\_{\boldsymbol{\Gamma}}}} \operatorname{F}\_{\boldsymbol{\Gamma}}(\boldsymbol{\Theta}) \text{ - } \mathop{\text{lim}}\_{\boldsymbol{\Theta} \xrightarrow{\text{prod}} \boldsymbol{\mathcal{F}\_{\boldsymbol{\Gamma}}}} \operatorname{F}\_{\boldsymbol{\Gamma}}(\boldsymbol{\Theta}) = \mathop{\text{i}} \int\_{-\infty}^{\infty} \mathbf{f}(\exp(\text{Xu})) \text{du} \dots$$

If we derive the formula (2), we have

$$\mathrm{E}\left(\frac{\mathrm{d}}{\mathrm{d}\theta}\right)\mathrm{F}\_{\mathrm{f}}(\theta) = \frac{\theta(\pi\theta\cos\pi\theta - \mathrm{sen}\,\pi\theta)}{\pi\theta^{2}\mathrm{sen}\,\pi\theta}\mathrm{F}\_{\mathrm{f}}(\theta) - 2\pi\mathrm{i}(\frac{\mathrm{sen}\,\pi\theta}{\pi\theta})\mathrm{f}(\mathrm{exp}\pi\theta\mathrm{h}) + \mathrm{E}(\theta)) \tag{\text{VIII. 9}}$$

with limE() = 0. Thus it is have that

$$(\mathcal{S}) \mathop{\rm lim}\_{\mathbf{0} \xrightarrow{\mathbf{0}} \mathbf{0}} \left( \frac{\mathbf{d}}{\mathbf{d}\theta} \right) \mathbf{F}\_{\mathbf{f}}(\theta) = 2\pi \mathbf{if}(\mathbf{1}).$$

Now re-define (4), and (5), in terms of integrals on the subgroup of Cartan A of G. To it, you consider the endomorphism Hf(t) = Ff A(exp(tH)). Then by (III.79) and (III. 80), we affirm to this case that:

$$\text{(6)}\,\text{H}\_{\text{f}}\,\text{(t)} = \text{e}^{\text{t}} \int\_{-\infty}^{\infty} \text{f}(\text{expptHexNx})\text{(dx.)}$$

of where thus it is conclude that

$$\text{(\text{\textquotesingle}{})}\mathop{\text{lim\textquotesingle}}\limits\_{\theta\text{---}\to 0^{\text{\textquotesingle}{}}}\mathop{\text{F}}\_{\text{f}}(\theta) - \mathop{\text{lim\textquotesingle}{}}\mathop{\text{F}}\_{\text{f}}(\theta) = \mathop{\text{i}}\mathop{\text{lim\textquotesingle}{}}\limits\_{\text{t} \xrightarrow{} \text{0}} \text{H}\_{\text{f}}(\text{t}).$$

But the definition of Ff, implies that Ff(a) = Ff(a1). Thus

$$\lim\_{\mathbf{t}\to\mathbf{t}\to\mathbf{0}} (\mathbf{d}/\mathbf{dt}) \mathbf{H}\_{\mathbf{f}}(\mathbf{t}) = \mathbf{0},\tag{\text{VIII.10}}$$

then

$$\text{I}\_{\text{f}}(\text{8}) \mathop{\lim}\_{\text{0} \longrightarrow \text{0}^{+}} \left( \frac{\text{d}}{\text{d}\theta} \right) \text{F}\_{\text{f}}(\theta) - \mathop{\lim}\_{\text{0} \longrightarrow \text{0}^{-}} \left( \frac{\text{d}}{\text{d}\theta} \right) \text{F}\_{\text{f}}(\theta) = 0 = \mathop{\lim}\_{\text{t} \longrightarrow \text{0}} \left( \frac{\text{d}}{\text{d}\theta} \right) \text{H}\_{\text{f}}(\text{t}).$$

Be CZ(LC), such that T(C) = (d2/d ), and A(C) = d2/dt . Here J, is the isomorphism of Harish- Chandra associated with the subgroup of Cartan J. C, (is save a scalar multiple and subtraction of a scalar) the Casimir operator of the algebra L. The Casimir operator, and their analogous to FA, combined with (7), and (8), imply that

$$\text{I}\_{\bullet}(\theta) \mathop{\rm lim}\_{\bullet \longrightarrow 0^{+}} \left( \frac{\mathbf{d}}{\mathbf{d}\theta} \right)^{\mathbf{k}} \mathop{\rm F}\_{\mathbf{f}}(\theta) - \mathop{\rm lim}\_{\bullet \longrightarrow 0^{-}} \left( \frac{\mathbf{d}}{\mathbf{d}\theta} \right)^{\mathbf{k}} \mathop{\rm F}\_{\mathbf{f}}(\theta) = 0 = \mathbf{i}^{\mathbf{k}+1} \mathop{\rm lim}\_{\bullet \longrightarrow 0} \left( \frac{\mathbf{d}}{\mathbf{d}\mathbf{t}} \right)^{\mathbf{k}} \mathop{\rm H}\_{\mathbf{f}}(\mathbf{t}).$$

We assume that G = G0, where

$$\mathbf{^0G} = \{ \mathbf{g} \in \mathbf{G} \mid \mathbf{Ad(g)}^k \mathbf{X} = \mathbf{0}, \forall \ \mathbf{X} \in \mathbf{X}(\mathbf{G}) \},$$

where Ж(G) = Hom(G, R\*)is continuous. We assume that G is of inner type (Ad(G) G). Be A H. Then

$$\mathrm{Fr^{H}(h) = \Delta(h)[c \, f(\mathrm{ghg^{-1}}) \mathrm{dg} \, h]}$$

If fC(G), and if aA, then R(a)f0G = uC(0G). In consequence, Ff H(ha) = FuH0G(h) to hH 0G.

This will be of utility to transfer the results of the case G = 0G, to a more general situation.

**Theorem VIII. 2. 1.** Be fC(G), and we assume that Ff H = 0, to each subgroup of Cartan of G, that is not fundamental. If H, is fundamental then Ff H, extends to a smooth function on H.

*Proof*.In the demonstration of the theorem is necessary to use orbital integrals on the different real reductive groups. If L, is a reductive group and if J, is a subgroup of Cartan of L, then we define

$$\mathbf{^LF} = \mathbf{F}\mathbf{\dot{\epsilon}} \tag{\text{VIII.11}}$$

This notation will can to help to identify to real reductive group on which the integration have happen.

That's right; we demonstrate the result by induction on the dimension of G.

i. If dimG = 0, or 1, then G = H, is the unique subgroup of Cartan. Then, if fC(G),

$$\text{F}\_{\text{f}}^{\text{H}}(\text{h}) = \Delta\_{\text{H}}(\text{h}) \int\limits\_{\text{H}/\text{H}} \text{f}(\text{h}) \text{dh} \text{H} = \Delta\_{\text{H}}(\text{h}) \text{f}(\text{k}), \tag{\text{VIII. 12}}$$

But *H*(*h*) = (*h*), hH. If H = G, then *H*(*h*) = (*h*) = 1, of where

$$\mathbf{F} \mathbf{f}^H(\mathbf{h}) = \mathbf{f}(\mathbf{h}), \quad \forall \ \mathbf{h} \in \mathbf{H}.$$

ii. Suppose that the result is valid to the cuspidal form LFf J , with J, a subgroup of Cartan of the real reductive group L, with 0 dim L dim G. Is necessary demonstrate the result on a cuspidal form 0GFf J ,of 0G, a real reductive group whose subgroup of Cartan is J, fC (G), such that

$$0 \le \dim \mathcal{L} \le \dim \, \mathbb{G} \subseteq \dim \mathcal{G} \,\tag{711.13}$$

In effect, if G 0G, then dim 0G< dim G, for be 0G, a connect subgroup of G (identity component of G). Be A, a splitting component of G. If J, is a subalgebra of Cartan of 0G, then JA, is a subalgebra of Cartan of G, and each subalgebra of Cartan of G, is of this form. That's right, only is necessary consider the cuspidal form

$$\mathbf{F}\mathbf{f}^{\mathrm{FA}}(\mathbf{j}\mathbf{a}) = {}^{\mathrm{0C}}\mathbf{F}\mathbf{f}^{\mathrm{J}}(\mathbf{j}) = \mathbf{F}\mathbf{f}^{\mathrm{J}}(\mathbf{j})\mathbf{a}$$

and to use the Harish-Chandra transform fP(ja) = ua(j), to realize a right translation of fC (G), and of that manage to obtain a extension of Ff J , on L, like smooth function in L, (If fC (G), and if aA, then R(a)f0G = uC (G)). Thus

$$\mathsf{F}^{\mathsf{F}\mathsf{A}}(\mathsf{ja}) = \mathsf{F}^{\mathsf{u}} \mathsf{I}^{\top \otimes \mathsf{C}}(\mathsf{j}) = \mathsf{F} \mathsf{d}(\mathsf{j}), \quad \forall \ \mathsf{j} \in \mathsf{I} \cap \mathsf{V} \tag{\mathsf{VIII.14}}$$

Then (VIII. 13), follows. Thus Ff H, it is can extend like smooth function to H, a subgroup of Cartan of G.

Now we assume that G = 0G. Also we assume in G, that H, is a non-compact fundamental subgroup of Cartan of G. Then we can assume that H = HF, and PF, is proper in G. We consider Q = PF, L = 0MF, and T = TF. If J, is a subgroup of Cartan of L, then JA, is a subgroup of Cartan of G, and if fC (G),

$$\mathbf{F}^{\mathsf{l}\mathsf{A}}(\mathsf{j}\mathsf{a}) = {}^{\mathsf{L}}F^{\mathsf{l}}\_{\mathsf{R}(\mathsf{a})\mathsf{f}\mathsf{Q}}(\mathsf{j}), \ \forall \ \mathsf{j} \in \mathsf{J}, \ \mathsf{a} \in \mathsf{A}\mathsf{p}.$$

Thus the induction hypothesis to this non-compact case also prevalence.

If T is a compact subgroup of Cartan of G, then

$$\mathbf{F} \mathbf{f}^{\mathsf{T}}(\mathbf{t}) = \mathbf{F} \mathbf{f}(\mathbf{t}), \quad \forall \ \mathbf{t} \in \mathbf{T}, \ \mathbf{f} \in \mathbb{S}(\mathbf{G}).$$

Let n = (gC)pC. Let X = X= *conj*X, XgC, respect to g. Then (gC)= (gC). Let n, Z(gC), and W = Z = *conj*Z = Z. If Z 0, then Z + W 0, in p (just not pC) since Z + Z= 2Xp.

Consider the map

92 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Ff

**Theorem VIII. 2. 1.** Be fC(G), and we assume that Ff

that is not fundamental. If H, is fundamental then Ff

LFf

If fC(G), and if aA, then R(a)f0G = uC(0G). In consequence, Ff

We assume that G = G0, where

(Ad(G) G). Be A H. Then

hH 0G.

L, then we define

have happen.

kk k

0G = gGAd(g)kX = 0, XЖ(G), where Ж(G) = Hom(G, R\*)is continuous. We assume that G is of inner type

H(h) = (h)G f(ghg1)dg,

This will be of utility to transfer the results of the case G = 0G, to a more general situation.

*Proof*.In the demonstration of the theorem is necessary to use orbital integrals on the different real reductive groups. If L, is a reductive group and if J, is a subgroup of Cartan of

This notation will can to help to identify to real reductive group on which the integration

That's right; we demonstrate the result by induction on the dimension of G.

H

ii. Suppose that the result is valid to the cuspidal form LFf

J

on a cuspidal form 0GFf

(G), such that

But *H*(*h*) = (*h*), hH. If H = G, then *H*(*h*) = (*h*) = 1, of where

Ff

i. If dimG = 0, or 1, then G = H, is the unique subgroup of Cartan. Then, if fC(G),

f H H H/H

H(h) = f(h), hH.

the real reductive group L, with 0 dim L dim G. Is necessary demonstrate the result

0 dim L dim 0G dim G, (VIII. 13)

θ 0 θ 0 t 0 dd d (9) lim F ( ) lim F ( ) 0 i lim H (t). <sup>d</sup><sup>θ</sup> <sup>d</sup><sup>θ</sup> dt

ff f

k 1

H(ha) = FuH0G(h) to

H = 0, to each subgroup of Cartan of G,

, with J, a subgroup of Cartan of

H, extends to a smooth function on H.

J = Ff, (VIII. 11)

F (h) <sup>Δ</sup> (h) f(h)dhH <sup>Δ</sup> (h)f(k), (VIII. 12)

J

,of 0G, a real reductive group whose subgroup of Cartan is J, fC

$$[\ ] : (\mathfrak{g}\circ)^{a} \times (\mathfrak{g}\circ)^{-a} \to [(\mathfrak{g}\circ)^{a} \ (\mathfrak{g}\circ)^{-a}] \tag{VIII. 15}$$

whose rule of correspondence be

$$(\mathbb{Z}, \mathbb{W}) \mid \to [\mathbb{Z}, \mathbb{W}]\_{\mathbb{V}} \tag{\text{VIII.16}}$$

followed of the map

$$\alpha\colon[(\mathfrak{g}\mathfrak{c})\_{\alpha}\ (\mathfrak{g}\mathfrak{c})\_{-\alpha}] \to \mathbb{R}\tag{\text{VII.17}}$$

whose rule of correspondence is:

$$\text{[Z\_2 W]} \mid \rightarrow \text{a([Z\_2 W])},\tag{\text{VIII.18}}$$

we have that ([Z, W]) = (ZW WZ), and give the normalization Z/Z2, we have

$$\frac{\alpha(Z\mathcal{W}-\mathcal{W}Z)}{\left\|Z\right\|^2} = \frac{\alpha Z\mathcal{W}-\alpha\mathcal{W}Z}{\left\|Z\right\|^2} = \frac{(\alpha\left\|Z\right\|^2 - \alpha\left\|\underline{Z}\right\|^2)}{\left\|Z\right\|^2} = \frac{(\left\|Z\right\|^2 + \overline{\alpha}\left\|Z\right\|^2)}{\left\|Z\right\|^2} = \frac{2\left\|Z\right\|^2}{\left\|Z\right\|^2} = 2\sqrt{\left\|\boldsymbol{\nabla}\right\|^2}$$

given that Z2 = Z2. Thus ([Z, W]) = 2. Calling H = Z + W, and h = i[Z, W], and X = (1/2)([Z, W] + i(Z W)). It is can verify that H, h, and X, complies with the same relations of commutation like the given to the elements of sl(2, R).

Let l = RH + RX + t. Then [l ,l ] sl(2, R). Let the orbits of T, T= tT t= 1, and

T ' = tT t1, , then the space T 'exp(RH), is open in T. That' right, be L , the connect subgroup of G, with Lie subalgebra l . Then T, is the center of L . Let k() = exph. If t = uk()T'', and if fCC(G), then

$$\begin{aligned} \text{iii} \qquad \int\limits\_{\mathbf{G}} \mathbf{f}(\mathbf{x}\mathbf{x}\mathbf{x}^{-1}) &= \int\limits\_{\mathbf{G}/\mathbf{L}^{\mathbf{g}}} \int\limits\_{\mathbf{L}^{\mathbf{g}}} \mathbf{f}(\mathbf{g}\mathbf{x}\mathbf{x}\mathbf{k}\_{\alpha}(\theta)\mathbf{x}^{-1}\mathbf{g}^{-1}) \mathbf{dx} \mathbf{d}g\_{\alpha} \end{aligned}$$

Considering (t) = t/2(1 t ), then

$$\text{iv} \qquad \qquad \qquad \Lambda (\text{uk}\_a(\theta)) = 2 \text{i} \Lambda\_a (\text{uk}\_a(\theta)) \text{sen } (\pi \theta),$$

Let the cuspidal form on L ,

$$\mathcal{R}\_{\mathsf{f}}(\mathsf{g},\mathsf{u},\theta) = \Delta\_{a}(\mathsf{uk}\_{a}(\theta)) \text{sen}(\mathsf{π}\theta) \int\_{\mathsf{G}/\mathsf{L}^{a}} \mathsf{R}\_{\mathsf{f}}(\mathsf{g},\mathsf{u},\theta) \text{dg} \mathsf{L}^{a}\mathsf{g}$$

Then on the classes in G/L , we can consider it, cuspidal form

$$\text{Fr}(\text{uk}\_a(\theta)) = \left[ \text{c}\_{\text{'} \text{L} \alpha} \text{ Rf}(\text{g}, \text{u}, \theta) \text{dg} \text{L}^a \right]$$

Let fCC (G). Let uT'. We fix pU(tC). Note that if , with , enough little and such that 0, then uk()T'. Furthermore, writing for the derivation rule of the cuspidal form Ff(uk2()), that

$$\begin{aligned} \text{ph}^k \mathbf{F}\_{\mathbf{f}}(\text{uk}\_a(\theta)) &= \\ \mathbf{F} &= \Sigma \begin{pmatrix} \mathbf{k} \\ \mathbf{j} \end{pmatrix} \Big| \left( \mathbf{i} \frac{\mathbf{d}}{\mathbf{d}\theta} \right)^{\mathbf{k}-\mathbf{j}} \mathbf{p} \Delta\_a(\text{uk}\_a(\theta)) \Big( \mathbf{i} \frac{\mathbf{d}}{\mathbf{d}\theta} \Big)^{\mathbf{j}} \text{sen} \mathbf{r} \theta \int \mathbf{R}\_{\mathbf{f}}(\mathbf{g}, \mathbf{u}, \theta) \text{d}\mathbf{g} \mathbf{L}^a \Big) \end{aligned}$$

Let J, be the centralizator of G, of Texp(RH). Then using (9), save a constant, we have

$$\begin{aligned} &\lim\_{\boldsymbol{\Theta}\to\boldsymbol{0}^{+}}\mathrm{ph}^{\mathrm{k}}\mathrm{F}\_{\mathrm{f}}^{\mathrm{T}}(\mathrm{uk}\_{\boldsymbol{\alpha}}(\boldsymbol{\Theta}))-\lim\_{\boldsymbol{\Theta}\to\boldsymbol{0}^{-}}\mathrm{ph}^{\mathrm{k}}\mathrm{F}\_{\mathrm{f}}^{\mathrm{T}}(\mathrm{uk}\_{\boldsymbol{\alpha}}(\boldsymbol{\Theta})) \\ &=\mathrm{p}\Sigma\begin{pmatrix}\mathrm{k}\\\mathrm{j}\end{pmatrix}\mathrm{(i}\frac{\mathrm{d}}{\mathrm{d}\boldsymbol{\theta}}\end{aligned}\mathrm{(i}\begin{aligned} &\mathrm{k}^{\mathrm{k}}\mathrm{F}\_{\mathrm{f}}^{\mathrm{T}}(\mathrm{uk}\_{\boldsymbol{\alpha}}(\boldsymbol{\Theta})) \\ &\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\mathrm{(i}\mathrm{d}\mathrm{k}\mathrm{(k}\mathrm{(i}^{\mathrm{T}}\mathrm{k}\mathrm{(k}\mathrm{)})\mathrm{A}\_{\mathrm{a}}(\mathrm{u})^{-1}\lim\mathrm{Im}\left(\frac{\mathrm{d}}{\mathrm{d}\mathrm{t}}\right)^{\mathrm{j}}\mathrm{F}\_{\mathrm{f}}^{\mathrm{T}}(\mathrm{u}\mathrm{exp}\mathrm{H}),\end{aligned}$$

Since the both sides of (iv), are continuous on C (G), it is fC (G). This is the condition of jump mentioned to begin of the demonstration of theorem. Then the formula (v), implies that if Ff H = 0, the non-fundamental subgroups of Cartan H, of G, then Ff, is smooth in a neighborhood of each tT', n+, is to say, to all point in the space T'expRH. Now well, suppose that k+. Be

$$\mathbf{l}^{a} = \mathbf{g} \cap \langle \mathbf{tc} + \langle \mathbf{gc} \rangle\_{a} + \langle \mathbf{gc} \rangle\_{-a} \rangle. \tag{VIII. 19}$$

Let L , be the connect subgroup of G, corresponding to l . Then L , is compact. By the cuspidal lemma, the differences given in (v), are not jumps, are cusped in this case.

Then it is have demonstrate that: If Ff H = 0, to all the subgroups of Cartan non-fundamentals of G, then the cuspidal form Ff H, is smooth in a neighborhood of each tT, such that t = 1, to at least a Then

$$\mathrm{T}'\_a \mathrm{exp}(\mathbb{R}\mathrm{h}) \cap \mathrm{T}\_a \mathrm{exp}(\mathbb{R}\mathrm{H}) = \mathrm{T}'\_a \mathrm{exp}(\mathbb{R}\mathrm{h}),$$

is open in T, and since pf, it is extends like a smooth map on T', pU(tC), then f, it is extends like a smooth function on T.

One direct application is:

**Corollary VIII. 2. 1.**Be fC (G), a cuspidal form. If H, is a subgroup of Cartan of G, whose compact module is not the centre of G, then Ff H = 0. If H, is compact module that is centre of G, then Ff H, it is extend to a smooth function on H.

Proof. See [1].

94 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

 

of commutation like the given to the elements of sl(2, R).

 ,l 

the connect subgroup of G, with Lie subalgebra l

k() = exph. If t = uk()T'', and if fCC(G), then

= RH + RX + t. Then [l

Considering (t) = t/2(1 t

k f α

ph F (uk (θ))

 

Let the cuspidal form on L

Then on the classes in G/L

Let fCC

Ff(uk2()), that

Let l

iii

we have that ([Z, W]) = (ZW WZ), and give the normalization Z/Z2, we have

2 2 2 22 ( ) ( )( )2 2, *ZW WZ ZW WZ ZZZZ Z Z Z Z ZZ*

given that Z2 = Z2. Thus ([Z, W]) = 2. Calling H = Z + W, and h = i[Z, W], and X = (1/2)([Z, W] + i(Z W)). It is can verify that H, h, and X, complies with the same relations

T ' = tT t1, , then the space T 'exp(RH), is open in T. That' right, be L

1 1 1 α

f(xtx ) f(guxk (θ)x g )dxdg,

α α

), then

f α α f

k j j

Let J, be the centralizator of G, of Texp(RH). Then using (9), save a constant, we have

R (g, u,θ) <sup>Δ</sup> (uk (θ))sen(πθ) R (g, u,θ)dgL ,

, we can consider it, cuspidal form

Ff(uk()) = G/L Rf(g, u, )dgL

that 0, then uk()T'. Furthermore, writing for the derivation rule of the cuspidal form

G G/L L

iv (uk()) = 2i(uk())sen (),

,

] sl(2, R). Let the orbits of T, T= tT t= 1, and

α

G/L

(G). Let uT'. We fix pU(tC). Note that if , with , enough little and such

α α f

<sup>k</sup> d d <sup>Σ</sup> i p<sup>Δ</sup> (uk (θ)) i senπθ R (g, u,θ)dgL <sup>j</sup> <sup>d</sup><sup>θ</sup> <sup>d</sup><sup>θ</sup> ,

[Z, W] ∣([Z, W]), (VIII. 18)

22 22 2

 

. Then T, is the center of L

α

α

,

G/L

,

. Let

**Chapter IX** 

## **Some Applications in the Study of Harmonic Analysis**

### **IX. 1. Introduction**

We remember that a cuspidal form cans to be defined to be defined by means of the Harish-Chandra transform as those function fC (G), such that (L(x)R(x)f)P = 0, with P a proper parabolic subgroup of G, and x, yG.

But due to that the space of the cuspidal forms Ef, can to be identified like the closure space Cl(C (G)), of the matrix coefficients K-finites of the discrete series then by the theorem on closure of the space C (G) [39], can to be assumed that

$$\mathbf{E} = \{ \mathbf{f} \in \mathbb{R}(\mathbf{G}) \mid \dim \mathbf{Zc}(\mathfrak{g}) \lhd \infty \} \cong^{0} \mathbb{R}(\mathbf{G}), \tag{\text{D.1.1}}$$

where

$$^0\boldsymbol{\varepsilon}^0(\mathbf{G}) = \{ \mathbf{f} \in \boldsymbol{\varepsilon}^\circ(\mathbf{G}) | (\mathbf{L}(\mathbf{x}) \mathbf{R}(\mathbf{x}) \mathbf{f})^\mathbf{P} = \mathbf{0}, \forall \text{ x, y} \in \mathbf{G} \text{ and } \mathbf{P} = {}^0\mathbf{M} \mathbf{N} \subset \mathbf{G} \}, \tag{\text{IX. 1.2}}$$

Likewise, if we consider all the compact Cartan subgroups T, of G, result some interesant applicable properties of harmonic type to cuspidal forms space 0C (G).

Considering that all the Cartan subgroups of G, were unidimensionals, we can identify to Lie algebra g, of G, as the corresponding algebras of the subgroups SL(2, R), and SU(2). For it is enough to consider the integrals of cuspidal forms on the orbits T', and T, of T, k.

For the case of compact Cartan subgroups of G, whose dimension to be major that 1, it is possible to apply the Peter- Weyl theorem and the orthogonality relations of Schür and obtain an orbital algebra analogous to space

$$\mathbf{l}^{a} = \mathbf{g} \cap (\mathbf{t}\_{\mathbb{C}} + (\mathbf{g}\_{\mathbb{C}})\_{a} + (\mathbf{g}\_{\mathbb{C}})\_{-a})\_{\prime} \tag{\text{D'}.1.3}$$

We can thus to assum that the Cartan subgroups of G, could be at least 2-dimensional.

### **IX. 2. Harmonic analysis of cuspidal forms**

Let G, be a real reductive group such that Ad(G) G, and G = 0G. Let's visualize to symmetric algebra S(gC), as the algebra of differential operators with constant coefficients on

g. Let R, be such that t() = exp (h), hRH. If fC (G), then Ff T(f())= Ff(). Let q = dim t, and p = dim p.

Let 0C (G), like was defined in (IX. 1. 2). If G, have a compact Cartan subgroup T, and if +, is a positive roots system to (gC, tC), then

$$\overline{\alpha} = \prod\_{\alpha \in \Phi^+} \mathcal{H}\_{\alpha} \in \mathcal{U}(\mathfrak{t}\_{\mathbb{C}}),\tag{13.2.1}$$

**Theorem IX. 2.1.** If G, have non-compact Cartan subgroups then 0C (G) = {0}. If T, is compact Cartan subgroup of G, then there is a non-nule constant *C*G, such that if f0C(G), then

$$\overline{\text{co}}\,\mathrm{F}\_{\text{f}}^{\mathrm{T}}(1) = \mathbb{C}\_{\text{G}}\,\mathrm{f}(1),\tag{18.2.2}$$

The result is a special case of a more general theorem of Harish-Chandra, which establish a similar formula fC (G), replacing T, for fundamental Cartan subgroups. Since, we come to generalize on the compact Cartan subgroups of G, then we demonstrate first that <sup>0</sup>C (G) = {0}, if G, have non-compact Cartan subgroups. For it, we have that consider some important facts on the formulas of the theorem 6.7.1, to this concrete case and two lemmas that were demonstrated in their time.

Let X1, , Xn, be a pseudo-base of g, relative to B (that is to say B(Xj, Xk) = j,k) such that g Rn. Of fact, through the diagram

(IX. 2.3)

we can deduce the composition map

$$\mathbb{B} = \emptyset \text{ o } P: \mathfrak{g} \times \mathfrak{g} \to \mathfrak{g} \to \mathbb{R},\tag{1\%.2.4}$$

whose rule of correspondence is

$$\mathcal{B}: (\mathcal{X}, \mathcal{X}) \mid \to P(\mathcal{X}), \tag{1\%.2.5}$$

Then X Yj j , with Xj, Yj{Xk}, of g, relative to B. Since n 2, then n = p + q, p, qZ+, and given that g Rn, then

$$
\mathbb{R}^n \equiv \mathbb{R} \mathbb{V} \oplus \mathbb{R} \mathbb{V} = \mathfrak{p} \oplus \mathfrak{t},\tag{\mathbb{C}X.2.6}
$$

Then we define to p, q 1, the cuspidal form Fp, q = Fg = F, such if Fp, q, is integrable then there is a not null constant Cp, q, such that

$$\mathfrak{a} \uparrow \downarrow \uparrow \mathfrak{b} \tag{13.2.3}$$

Chapter IX – Some Applications in the Study of Harmonic Analysis 99

$$\int\_{\mathfrak{g}\_{\varepsilon}} \mathbf{F}\_{\mathbf{p},\mathbf{q}}(\mathbf{x}) \overline{\mathbf{o}}^{\mathbb{I}^{\mathbf{n}/2}\mathbb{I}} \mathbf{f}(\mathbf{x}) d\mathbf{x} = \mathbb{C}\_{\mathbf{p},\mathbf{q}} \mathbf{f}(\mathbf{0}),\tag{1\%.2.7}$$

Since that F = Fp, q, p, q 1, is G-invariant then F(Ad(g)X) = F(X), gG, and Xg. If we consider a system of roots j relative to gC, (hj)Cwe can define

$$\pi\_j(\mathbf{h}) = \prod\_{a \in \Phi\_j} a(\mathbf{h}), \tag{\text{IX. 2.8}}$$

 hhj. Then D(h) = j(h) 2, hhj. Thus there are constants Cj, (j = 1, 2, , n) and normalized measures on g, and hj, such that

$$\begin{aligned} \mathbf{f}(\mathbf{g}) &= \int\_{\mathbf{0}} \mathbf{f}(\mathbf{X}) \mathrm{d}\mathbf{X} = \sum\_{j} \mathbf{C}\_{j} \int\_{\mathbf{0}\_{j}} \left| \mathrm{D}(\mathbf{h}) \right| \left( \int\_{\mathbf{G}/\mathbf{H}\_{j}} \mathbf{f}(\mathbf{A} \mathbf{d}(\mathbf{g}) \mathbf{h}) \mathrm{d}(\mathbf{g} \mathbf{H}\_{j}) \mathrm{d}\mathbf{h}\_{j} \right) \right| \\ &= \sum\_{j} \mathbf{C}\_{j} \int\_{\mathbf{0}\_{j}} \left| \Phi\_{\mathbf{f}}(\mathbf{h}) \right| \mathrm{d}\mathbf{h}\_{j'} \end{aligned}$$

But ge Rn, then

98 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

q = dim t, and p = dim p.

is a positive roots system to (gC, tC), then

that were demonstrated in their time.

g Rn. Of fact, through the diagram

we can deduce the composition map

whose rule of correspondence is

and given that g Rn, then

is a not null constant Cp, q, such that

g. Let R, be such that t() = exp (h), hRH. If fC (G), then Ff

Let 0C (G), like was defined in (IX. 1. 2). If G, have a compact Cartan subgroup T, and if +,

**Theorem IX. 2.1.** If G, have non-compact Cartan subgroups then 0C (G) = {0}. If T, is compact

The result is a special case of a more general theorem of Harish-Chandra, which establish a similar formula fC (G), replacing T, for fundamental Cartan subgroups. Since, we come to generalize on the compact Cartan subgroups of G, then we demonstrate first that <sup>0</sup>C (G) = {0}, if G, have non-compact Cartan subgroups. For it, we have that consider some important facts on the formulas of the theorem 6.7.1, to this concrete case and two lemmas

Let X1, , Xn, be a pseudo-base of g, relative to B (that is to say B(Xj, Xk) = j,k) such that

Then X Yj j , with Xj, Yj{Xk}, of g, relative to B. Since n 2, then n = p + q, p, qZ+,

Rn = Rp Rq = p t, (IX. 2.6)

Then we define to p, q 1, the cuspidal form Fp, q = Fg = F, such if Fp, q, is integrable then there

 <sup>t</sup>

Cartan subgroup of G, then there is a non-nule constant *C*G, such that if f0C(G), then

T

U( ),

T(f())= Ff(). Let

(IX. 2.3)

C (IX. 2.1)

f G F f(1) (1) , *C* (IX. 2.2)

B = o *P* : g g g R, (IX. 2.4)

B : (X, X) ∣*P*(X), (IX. 2.5)

$$\mathbf{f}(\mathbf{0}) = \sum\_{\mathbf{j}} \mathbf{C}\_{\mathbf{j}} \int\_{\mathbf{j}} \left| \mathbf{D}(\mathbf{h}) \right| \left( \int\_{\mathbf{G}/\mathbf{H}\_{\mathbf{j}}} \left| \mathbf{D}(\mathbf{Ad}(\mathbf{g})\mathbf{h}) \right|^{1/2} \mathbf{f}(\mathbf{Ad}(\mathbf{g})\mathbf{h}) \mathbf{d}(\mathbf{g}\mathbf{H}\_{\mathbf{j}}) \mathrm{d}\mathbf{h}\_{\mathbf{j}} \right) \rho$$

But D(Ad(g)h)-1/2 = Fp, q(h)[n/2] = F(h)[n/2], then

$$\int\_{\mathbf{G}/\mathcal{H}\_j} \left| \mathbf{D}(\mathbf{Ad}(\mathbf{g})\mathbf{h}) \right|^{-1/2} \mathbf{f}(\mathbf{Ad}(\mathbf{g})\mathbf{h}) \mathrm{d}\mathbf{g}\mathbf{H} = \mathbf{F}(\mathbf{h}) \boldsymbol{\varpi}^{\{\mathbf{n}/2\}} \int\_{\mathbf{G}/\mathcal{H}\_j} \mathbf{f}(\mathbf{Ad}(\mathbf{g})\mathbf{h}) \mathrm{d}\mathbf{g}\mathbf{H}\_j$$

Of fact, D 1/2, is locally integrable on g, in the general case.

Then (I) can be written as:

$$\text{f}(\mathbf{0}) = \sum\_{\mathbf{j}} \mathbf{C}\_{\mathbf{j}} \int\_{\mathbf{b}\_{\mathbf{j}}} \left| \pi\_{\mathbf{j}}(\mathbf{h}) \right|^{2} \int\_{\mathbf{G}/\mathbf{H}\_{\mathbf{j}}} \text{f}(\text{Ad}(\mathbf{g})\mathbf{h}) \text{dg} \mathbf{H} \text{dh}\_{\mathbf{j}}$$

Then considering the cuspidal form of [n/2]f, relative to the Cartan subgroup Hj, we have

$$\boldsymbol{\Phi}\_{\varpi^{\{n/2\}}\!}^{\operatorname{H}\_{j}}(\mathbf{h}) = \boldsymbol{\Sigma}\_{\mathbf{j}} \boldsymbol{\varpi}^{\{n/2\}} \int\_{\mathbf{G}/\mathcal{H}\_{j}} \mathbf{f}(\operatorname{Ad}(\mathbf{g})\mathbf{h}) \operatorname{dgH}\_{j}$$

The incise (i) take the form:

$$\mathbf{f}(\mathbf{0}) = \sum\_{\mathbf{j}} \mathbf{C}\_{\mathbf{j}} \int\_{\mathbf{b}\_{\mathbf{j}}} \left| \pi\_{\mathbf{j}}(\mathbf{h}) \right| \varepsilon\_{\mathbf{j}}(\mathbf{h}) \mathbf{F}(\mathbf{h}) \boldsymbol{\Phi}\_{\mathbf{w}^{\{\mathbf{w}\}} \mathbf{f}}^{\mathbf{H}\_{\mathbf{j}}}(\mathbf{h}) \mathbf{dh}\_{\mathbf{w}^{\{\mathbf{w}\}}} $$

But<sup>j</sup> [n/2]<sup>f</sup> Hj (h) = [*n*/2] <sup>f</sup> Hj (h), jS(hj). Then by the identity in appendix in [Wallach],

$$\text{if (0)} = \sum\_{\mathbf{j}} \mathbf{C}\_{\mathbf{j}} \int\_{\mathbf{b}\_{\mathbf{j}}} \left| \pi\_{\mathbf{j}}(\mathbf{h}) \right| \varepsilon\_{\mathbf{j}}(\mathbf{h}) \mathbf{F}(\mathbf{h}) \varpi\_{\mathbf{j}}^{\{\mathbf{n}/2\}} \boldsymbol{\Phi}^{\mathbf{H}\_{\mathbf{j}}} \mathbf{h} \,^{\mathbf{H}\_{\mathbf{j}}} \mathbf{h} \,^{\mathbf{n}} \mathbf{h} \,^{\mathbf{n}}$$

Let Dj = Dr(X),be with r = dim hj, Xg. Let r = *range*X, (assuming that r 2)and let *n* = dim g. To every t > 0, we define the space

$$\Omega = \{ \mathbf{X} \in \mathfrak{g} \mid \left| \mathbf{D}(\mathbf{X}) \right| \le \mathfrak{t}, \; \mathbf{r} \le \mathfrak{j} \,\,\forall n \}, \tag{\mathbb{X}. 2.9}$$

we consider the transcendent number = 3.14Then we consider the lemma:

**Lemma IX. 2.1** Suppose that G, is semisimple. If 0 < t <, then expt, is a diffeomorphism.

*Proof*. Since to every XgC, there is a neighborhood t, defined as the space

$$\Omega = \{ \mathbf{X} \in \text{End}(\mathfrak{g}\_{\mathbb{C}}) \mid \left| \mathbf{D}(\mathbf{X}) \right| \le \mathfrak{t}, \mathfrak{t} \ge 0 \}, \tag{\text{IX 2.10}}$$

If j <1, then the map

$$
\mathfrak{g}\_{\mathbb{C}} \xrightarrow{} \exp(\mathfrak{g})\tag{13.2.11}
$$

is a diffeomorphism of t, in a open subset of I = [g, g], in int(g). Then given that dAde, is the homomorphic map

$$\text{dA} \\ \mathfrak{a} \colon \mathfrak{g} \to \mathfrak{gl}(\mathfrak{g}) . \tag{18.2.12}$$

whose rule of explicit correspondence is

$$\mathbf{d}(\operatorname{Ad}(\mathbf{X})) = \operatorname{ad}\mathbf{X}, \quad \forall \quad \mathbf{X} \in \mathfrak{g} \tag{\text{I} \mathbf{X}.\text{ } \mathbf{1}\mathfrak{d} \mathfrak{g}}$$

the map exp, restricted to t, result be a covering homomorphism that maps open sets in open sets of the space gC, in End(gC), where these endomorphisms are differential operators. Therefore expt, is a diffeomorphism.

Let W, be an open neighborhood of O, in *z*(g), such that expW, is a diffeomorphism. Let t, be an open set in the ideal [g, g]. Let Wt = W t, be. Then

1. If 0 < t <, then exp, is a diffeomorphism of Wt, in an open neighborhood Vt, of 1, in G.

Let uCc (R), 0 u(s) 1, be such that u(s) = 1, to s ()/2, and u(s) = 0, to s > 2()/3. Let OY Cl(Y) and Cl(Y) W, with open Y, and Cl(Y), compact. Let hCc (W), be with h(X) = 1, to XCl(Y). Then we define the function C(*z*), such that

$$
\beta: \mathcal{W}\_{\pi^{-1}} \to \mathbb{C}^\*(\mathcal{W}),
\tag{18.2.14}
$$

with rule of correspondence

$$\mathcal{X} = \mathbb{Z} + \mathbb{T} \mid \to \mathbb{B}(\mathbb{X}), \tag{1\mathbb{X}.2.15}$$

where explicitly

100 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

j

j

we consider the transcendent number = 3.14Then we consider the lemma:

*Proof*. Since to every XgC, there is a neighborhood t, defined as the space

h

(h), jS(hj). Then by the identity in appendix in [Wallach],

jj j j f

f 0 C h hFh h dh [ /2] () () ()() () , *<sup>n</sup>*

Let Dj = Dr(X),be with r = dim hj, Xg. Let r = *range*X, (assuming that r 2)and let *n* = dim

t = {XgDj(X)< t, r j <*n*}, (IX. 2.9)

**Lemma IX. 2.1** Suppose that G, is semisimple. If 0 < t <, then expt, is a

t = {XEnd(gC)Dj(X)< t, t > 0}, (IX. 2.10)

gC exp(g), (IX. 2.11)

is a diffeomorphism of t, in a open subset of I = [g, g], in int(g). Then given that dAde, is the

dAde : g gl(g), (IX. 2.12)

 d(Ade(X)) = ad X, Xg, (IX. 2.13) the map exp, restricted to t, result be a covering homomorphism that maps open sets in open sets of the space gC, in End(gC), where these endomorphisms are differential operators.

Let W, be an open neighborhood of O, in *z*(g), such that expW, is a diffeomorphism. Let t,

1. If 0 < t <, then exp, is a diffeomorphism of Wt, in an open neighborhood Vt, of 1, in

Let OY Cl(Y) and Cl(Y) W, with open Y, and Cl(Y), compact. Let hCc

h(X) = 1, to XCl(Y). Then we define the function C(*z*), such that

(R), 0 u(s) 1, be such that u(s) = 1, to s ()/2, and u(s) = 0, to s > 2()/3.

WC(W), (IX. 2.14)

(W), be with

j

But<sup>j</sup>

[n/2]<sup>f</sup> Hj

diffeomorphism.

If j <1, then the map

homomorphic map

G. Let uCc

whose rule of explicit correspondence is

Therefore expt, is a diffeomorphism.

be an open set in the ideal [g, g]. Let Wt = W t, be. Then

(h) = [*n*/2] <sup>f</sup>

g. To every t > 0, we define the space

Hj

$$\mathfrak{f}(\mathsf{X}) = \mathsf{h}(\mathsf{Z}) \prod \mathsf{T}\_{\mathsf{L} \leq \downarrow \leq n-1} \mathsf{u}(\mathsf{D}(\mathsf{T})) , \tag{\mathsf{X}. 2.16}$$

Outside W, = 0. Then

2. C(g), supp W. , is Ad(G)-invariant, since

3. (Ad(g)X) = (X), Xg, and gG.

4. If h, is a Cartan subalgebra of g, then supp h, is compact.

Indeed, the partitions of the unity u and h, have supports included in W, and , respectively. Given that W= W , then supp W . Then huDjC(g).

The Ad(G)-invariance is followed of the existence of the diffeomorphism exp<sup>W</sup>, which is a covering homomorphism of int(G). The last affirmation is followed [Chevalley]. We will introduce a function , on G, defined as the map

$$a: \mathcal{G} \to \mathbb{C}^\*(\mathcal{W}\_{\pi^{-1}}),\tag{18.2.17}$$

with rule of correspondence

$$\mathfrak{g} \mid \rightarrow \mathfrak{a}(\mathfrak{g}) . \tag{1\%.2.18}$$

and such that applying the diffeomorphism exp<sup>W</sup>, to the image (g), is had that:

$$\alpha(\exp \mathcal{X}) = \beta(\mathcal{X}), \quad \forall \quad \mathcal{X} \in \mathfrak{g}.$$

inside W, and = 0, outside W.

Such map is smooth on G, since (g)C(G), and (gxg-1) = (x), x, yG. Then, if fC(G), then C(W)f = f , where f (X) = (X)f(expX), (X)C(W), and Xg. Clearly f (0) = (0)f(1). But (0) = 1, s <, since u(s) = 1, s ()/2, and h(s) = 1 in all Cl(W). For other side f (0) = f(1).

Consider H, a Cartan subgroup of G. Then we can write to H(h)f(h), as (using (h))

$$\Phi\_{\vec{\mathbf{f}}}^{\rm H}(\mathbf{H}) = \frac{\varepsilon\_{\rm H}(\mathbf{h})}{\Delta\_{\rm H}(\mathbf{h})} \beta(\mathbf{h}) \mathbf{F}\_{\rm f}^{\rm H}(\exp \mathbf{h}),\tag{18.2.19}$$

 fCc (G), and hh''.

Since H(exp h)/(h) 0, hWh, then (h)/H(exp h), define a smooth family on Wh, and since the map given by f ∣ Ff H, is extended to a continuous map of C(G) in C(H''), we have that

5. The map f ∣fH, is extended to a continuous map of C(G) in Cc (h''). Then

$$\mathbf{f}(\mathbf{0}) = \sum\_{\mathbf{j}} \mathbf{c}\_{\mathbf{j}} \int\_{\mathbf{h}'} \left| \pi\_{\mathbf{j}}(\mathbf{h}) \right| \varepsilon\_{\mathbf{j}}(\mathbf{h}) \mathbf{F}(\mathbf{h}) \varpi\_{\mathbf{j}}^{\{\mathbf{n}/2\}} \boldsymbol{\Phi}\_{\mathbf{f}'}^{\mathbf{H}\_{\mathbf{j}}}(\mathbf{h}) d\mathbf{h} \,\tag{\text{IX. 2.20}}$$

implies that if fC(G), then

$$\begin{split} \text{f}(\mathbf{1}) &= \sum\_{j} \mathbf{c}\_{j} \int\_{\mathbf{h}^{\*}} \pi\_{j}(\mathbf{h}) \mathbf{F}(\mathbf{h}) \varpi\_{j}^{[n/2]} \boldsymbol{\Phi}\_{\hat{\mathbf{f}}}^{\mathrm{H}\_{j}}(\mathbf{h}) d\mathbf{h}, \\ &= \sum\_{j} \mathbf{c}\_{j} \int\_{\mathbf{h}^{\*}} \pi\_{j}(\mathbf{h}) \mathbf{F}(\mathbf{h}) \varpi\_{j}^{[n/2]} \frac{\pi\_{j}(\mathbf{h})}{\Delta H\_{j}(\exp \mathbf{h})} \boldsymbol{\beta}(\mathbf{h}) \mathbf{F}\_{\hat{\mathbf{f}}}^{\mathrm{H}}(\exp \mathbf{h}) d\mathbf{h}, \end{split} \tag{1X.2.21}$$

Likewise (IX. 2. 21) implies that if f0C(G), and if G, include non-compact subgroups (where G = 0G) then f(1) = 0. Now if f0C(G), then R(g)f0C(G). Therefore, if G include non-compact subgroups then to 0C(G), is necessary the null space {0}. This last demonstrate the first part of the theorem.

Now we demonstrate that: If T is a Cartan compact subgroup of G then there is a constant cG 0, such that f0C(G),

$$\mathfrak{wF}^{\sf T}(1) = \text{ccf}(1)\_{\sf n}$$

FfC(T).

We assume that Hj = T, and f0C(G), then (IX. 2. 21) takes the form

$$\mathbf{f}(\mathbf{l}) = \mathbf{c}\_{\mathrm{j}} \int\_{\mathbf{t}} \pi\_{\mathrm{j}}(\mathbf{h}) \mathbf{F}(\mathbf{h}) \mathbf{w}^{\mathrm{[n/2]}} \frac{\pi(\mathbf{h})}{\Delta\_{\mathrm{t}}(\exp(\mathbf{h}))} \alpha(\mathbf{h}) \mathbf{F}\_{\mathrm{f}}^{\mathrm{T}}(\exp \mathbf{h}) \mathrm{d}\mathbf{h} \tag{\text{IX. 2.22}}$$

FfC(T).

We consider the following lemma that will be essential in the demonstration of the second affirmation of the theorem IX. 2. 1.

**Lemma IX. 2.2.** There is a constant Mg 0, such that if gA(t), then

$$\int\_{\mathbf{h}} \pi\_j(\mathbf{h}) \mathbf{F}(\mathbf{h}) \varpi^{[n/2]} \mathbf{g}(\mathbf{h}) d\mathbf{h} = \mathbf{M}\_{\mathbf{g}} \mathbf{g}(\mathbf{0}),\tag{1X.2.23}$$

Note: Mg, will be calculated in the course of the demonstration.

Let pP(t <sup>C</sup>), be such that p = D0(X), Xt. Let X, YEnd(V), with V, a vector space. Then are valid the following commutation identities:

$$\mathbf{i} \quad \text{ } \quad \mathbf{[X^{k}, Y]} = \sum\_{j=0}^{\mathbf{k} \cdot \mathbf{l}} \binom{\mathbf{k}}{j} (-1)^{\mathbf{k} - j + 1} X^{j} \text{(} \text{(ad} \mathbf{X} \text{)}^{\mathbf{k} - j} \text{Y)} = \sum\_{j=0}^{\mathbf{k} \cdot \mathbf{l}} \binom{\mathbf{k}}{j} (\text{(ad} \mathbf{X} \text{)}^{j} \text{Y}) \mathbf{X}^{\mathbf{k} - j} \text{,} \tag{\text{IX. 2.24}}$$

If we consider [[*<sup>n</sup>*/2], ] = [*n*/2][*<sup>n</sup>*/2], then sustituying (IX. 2. 23), we have

$$\begin{cases} \mathbf{F}(\mathbf{h})\pi(\mathbf{h})\varpi^{\lceil n/2 \rceil}\mathbf{g}(\mathbf{h})\mathbf{dh} = \int \mathbf{F}(\mathbf{h})\varpi^{\lceil n/2 \rceil}\pi(\mathbf{h})\mathbf{g}(\mathbf{h})\mathbf{dh} - \int \mathbf{F}(\mathbf{h})[\varpi^{\lceil n/2 \rceil}, \pi]\mathbf{g}(\mathbf{h})\mathbf{dh} \\ \mathbf{t} = \mathbf{I} - \mathbf{I}\mathbf{I}, \end{cases}$$

Then if T, is a Cartan subgroup then p = dimp, and q = dimt, are pairs (q = dimT). By the theorem 7.A 5. 8, there is a not null constantBg 0, such that

$$\mathbf{T} = \mathbf{B}\_8(\varpi^{\{u/2\}^-\{r,2\}}(\pi \mathbf{g})(\mathbf{0}),\tag{18.2.25}$$

Then is necessary calculate the integral *II*. Note that *n r*, is even then [*n*/2] [*r*/2] = (*n*r)/2. Therefore by i):

$$[\![\mathfrak{w}^{\left[n/2\right]}\!\!/,\pi] = \sum\_{j=0}^{\left[n/2\right]-1} \binom{\left[n/2\right]}{j} (-1)^{\left[n/2\right]-j+1} \int\_{\mathbf{t}} \mathbf{F}(\mathbf{h}) \varpi^{\mathbf{j}}(\{\operatorname{ad}\varpi^{\left[n/2\right]}\}^{\mathbf{j}}\mathbf{\pi}) \mathbf{g}(\mathbf{h}) d\mathbf{h} \,,\tag{\mathbf{I}\mathbf{X}\ 2.26}$$

Therefore (ad)[*<sup>n</sup>*/2] <sup>j</sup> = 0, if j < [*r*/2]. Then newly we apply the theorem 7. A. 5. 8, and we find that *II*, is expressed as:

$$\begin{aligned} \mathbf{J}\mathbf{J} &= \int \mathbf{F}(\mathbf{h}) [\varpi^{\{n/2\}}, \pi] \mathbf{g}(\mathbf{h}) \mathbf{d} \mathbf{h} = \sum\_{j=0}^{\lfloor n/2 \rfloor \cdot 1} \binom{\lceil n/2 \rceil}{j} \\ &\quad (-1)^{\lfloor n/2 \rfloor - j + 1} \mathbf{B}\_{\mathbf{g}}(\{\varpi^{\mathbf{j} - \lceil r/2 \rceil} (\mathbf{ad} \varpi)^{\lceil n/2 \rceil - j}, \pi)) \mathbf{g}(\mathbf{0}), \end{aligned} \tag{1X.2.27}$$

Appliying the second identity in (IX. 2. 21) in terms of the adjunct map "ad" and observing that the coefficients of ad, are annulled to j > [*r*/2] (Escolium 7. A. 2. 9) is had to j [*r*/2] that

$$\mathfrak{w}^{\mathfrak{h}^-\left[n/2\right]}(\operatorname{ad}\mathfrak{w}^{\{n/2\}^-\left(\pi\right)}\mathfrak{x})\mathfrak{g}(\mathbf{0}) = (\operatorname{ad}\mathfrak{w}^{\{n/2\}^-\left[r/2\right]}\mathfrak{x})\mathfrak{g}(\mathbf{0}),$$

Then

102 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

5. The map f ∣fH, is extended to a continuous map of C(G) in Cc

j '

j '

j

h

j

h

j h

j ' j

jj j f

f c h F(h) h dh

Ff

We assume that Hj = T, and f0C(G), then (IX. 2. 21) takes the form

**Lemma IX. 2.2.** There is a constant Mg 0, such that if gA(t), then

t

t

(1) () ,

jj j f

Likewise (IX. 2. 21) implies that if f0C(G), and if G, include non-compact subgroups (where G = 0G) then f(1) = 0. Now if f0C(G), then R(g)f0C(G). Therefore, if G include non-compact subgroups then to 0C(G), is necessary the null space {0}. This last demonstrate the first part

Now we demonstrate that: If T is a Cartan compact subgroup of G then there is a constant

T(1) = cGf(1),

j j f t <sup>h</sup> f c h F(h) (h)F (exp h)dh <sup>h</sup> (1) , (exp( ))

We consider the following lemma that will be essential in the demonstration of the second

/2 j g h F(h) g h dh M *<sup>g</sup>*(0),

 

/2 T

*<sup>n</sup>* (IX. 2.22)

*<sup>n</sup>* (IX. 2.23)

Wh, and since the map given by f ∣ Ff

C(H''), we have that

implies that if fC(G), then

of the theorem.

FfC(T).

FfC(T).

cG 0, such that f0C(G),

affirmation of the theorem IX. 2. 1.

Then

Since H(exp h)/(h) 0, hWh, then (h)/H(exp h), define a smooth family on

j j j jf

[n/2] Hj

[n/2] j H

h c h F(h) (h)F (exp h)dh H h

f c h h F(h) h dh (0) () ,

H, is extended to a continuous map of C(G) in

[n/2] Hj

, (exp )

(h'').

(IX. 2.20)

(IX. 2.21)

$$\begin{cases} \mathbf{F}(\mathbf{h})\pi(\mathbf{h})\varpi^{\lceil n/2 \rceil}\mathbf{g}(\mathbf{h})\mathbf{dh} = \mathbf{B}\_{\underline{\mathbf{g}}}(\varpi^{\lceil n/2 \rceil - \lceil r/2 \rceil})(\pi\mathbf{g})(\mathbf{0}) - \mathbf{B}\_{\underline{\mathbf{g}}}(\textup{ad}\varpi^{\lceil n/2 \rceil - \lceil r/2 \rceil}\pi)\mathbf{g}(\mathbf{0})\\ \mathbf{f} = \mathbf{C}((\textup{ad}\varpi^{\lceil n/2 \rceil - \lceil r/2 \rceil}\pi)\mathbf{g})(\mathbf{0}), \end{cases}$$

where

$$\mathbf{C} = \mathbf{B}\_{\mathbf{g}} \left( \sum\_{\mathbf{j}=\lceil r/2 \rceil}^{\lceil n/2 \rceil} \binom{\lceil n/2 \rceil}{\mathbf{j}} (-1)^{\lceil n/2 \rceil - \mathbf{j}} \right) \tag{\text{IX} . 2.28}$$

But the scolium 7. A. 2. 9, implies that

ad[*n*/2] [*r*/2]2[*<sup>n</sup>*/2] [*<sup>r</sup>*/2]([*n*/2] [*r*/2])H,

From the second idendity in (IX. 2. 21) and for the identity

$$\sum\_{j=p}^{k} (-1)^{k-j} \binom{k}{j} = (-1)^{k-p} \binom{k-1}{j-1} \neq 0,\tag{18.2.29}$$

if k p > 0, we have that

$$\begin{aligned} &\int\_{\mathbf{h}} \mathbf{F}(\mathbf{h}) \pi(\mathbf{h}) \varpi^{\lceil n/2 \rceil} \mathbf{g}(\mathbf{h}) \mathbf{dh} = \left[ \sum\_{j=p}^{\lceil n/2 \rceil} (-1)^{k-p} \binom{\mathbf{k}-1}{\mathbf{p}-1} 2^{\lceil n/2 \rceil - \lceil r/2 \rceil} (\lceil n/2 \rceil - \lceil r/2 \rceil)! \prod\_{a \in \Phi^{+}} \mathbf{H}\_{j} \right] \mathbf{g}(\mathbf{0}) \\ &= \mathbf{M}\_{\mathbf{g}} \mathbf{g}(\mathbf{0}), \end{aligned}$$

Also we use the estimation of tF(h)(h)[*<sup>n</sup>*/2]g(h)dh, through of a smooth function u, that is W-invariant defined on a W-invariant neighborhood of the 0.

**Lemma IX. 2.3.** Let W = W(gC, t <sup>C</sup>). Let uCW(UeW). Then

$$\sum\_{\mathbf{F}\in\mathfrak{G}^{+}} \left( \left( \prod\_{a\notin\mathbf{F}} \mathbf{H}\_{a} \right) \mathbf{u} \right) \mathbf{(0)} \prod\_{a\in\mathbf{F}} \mathbf{H}\_{a} = \mathbf{u}(\mathbf{0}) \prod\_{a\in\mathbf{F}} \mathbf{H}\_{a'} \tag{\text{IX. 2.30}}$$

*Proof.* Let , be a simple root in +. If F +, then we define the map :

$$
\wp: \mathcal{F} \to \mathcal{F}^{\phi},
\tag{\mathbb{X}.2.31}
$$

with rule of correspondence

$$\mathfrak{a} \mid \lnot \mathfrak{a}^{\wp}, \tag{\text{I} \mathfrak{X}. 2.32}$$

and where

$$\mathbf{F}^{\rho} = \{ \mathbf{s}\_{\mathbf{u}} \mathbf{F} \text{ if } \mathbf{u} \notin \mathbf{F} \text{ (}\mathbf{a} \in \mathbf{F}^{\rho} \text{) and } \ \langle \mathbf{s}\_{\mathbf{u}} | \mathbf{F} - \{ \mathbf{a} \} \rangle \cup \langle \mathbf{a} \rangle \text{, if } \mathbf{a} \in \mathbf{F} \text{)} \tag{\text{IX 2.33}}$$

Then the map , is a bijection of P(+). Indeed, if we consider that the dense space or kernel is the set

$$\mathbf{N}(\not\circ\rho) = \{\alpha \in \mathbf{F} \mid \not\rho \cdot \alpha = \alpha^{\not\circ} = 0\},\tag{\text{IX 2.34}}$$

and since , is simple root then 0. Then ker {0}. Therefore , is injective. Also is suprajective, since, let

$$p = \int\_{\mathfrak{h}} \mathbf{F}(\mathbf{h}) \pi(\mathbf{h}) \varpi^{\lceil n/2 \rceil} \mathbf{g}(\mathbf{h}) d\mathbf{h} \,\,\omega$$

Then s*p*, comes given as:

104 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

From the second idendity in (IX. 2. 21) and for the identity

k

j p

[ /2] [ /2] k - p

j p

W-invariant defined on a W-invariant neighborhood of the 0.

*Proof.* Let , be a simple root in +. If F +, then we define the map :

*<sup>n</sup> <sup>n</sup>*

But the scolium 7. A. 2. 9, implies that

if k p > 0, we have that

g

*g*

(0),

**Lemma IX. 2.3.** Let W = W(gC, t

with rule of correspondence

and where

is the set

M

t

[ /2]

*r*

*n*

]

j

ad[*n*/2] [*r*/2]2[*<sup>n</sup>*/2] [*<sup>r</sup>*/2]([*n*/2] [*r*/2])H,

k k - 1 (-1) <sup>j</sup> j - 1 ( 1) 0,

 

k j k p

k - 1 F h (h) g h dh (-1) H g p - 1

[ /2] [ /2] () () 2 ([ / 2] [ / 2])! (0)

Also we use the estimation of tF(h)(h)[*<sup>n</sup>*/2]g(h)dh, through of a smooth function u, that is

<sup>C</sup>). Let uCW(UeW). Then

F F FF

F = {sF, if F (F\F) and (s(F – {}) {}, if F, (IX. 2.33)

Then the map , is a bijection of P(+). Indeed, if we consider that the dense space or kernel

N() = {F= = 0}, (IX. 2.34)

H u H u(0) H (0) ,

(IX. 2.30)

: F F, (IX. 2.31)

∣, (IX. 2.32)

 

g j [ /2 [ / 2] C B

j

(IX. 2.29)

*n r n r*

 

*<sup>n</sup>* (IX. 2.28)

j

[ /2]

( 1) *<sup>n</sup>*

$$\mathbf{s}\_{\alpha}p = p^{\phi} = \left[ \sum\_{\mathbf{F} \in \Phi^{+}} \left( \left( \prod\_{\alpha \notin \mathbf{F}} \mathbf{H}\_{\alpha} \right) \mathbf{u} \right) \mathbf{(0)} \prod\_{\alpha \in \mathbf{F}} \mathbf{H}\_{\alpha} \right]^{\phi} \tag{\text{IX. 2.35}}$$

Given that s*p* = *p* = *p*, *p*P(t <sup>C</sup>), then (IX. 2. 34) takes the form

$$p^{\circlearrowright} = -\sum\_{\mathbf{F} \in \Phi^{+}} \left( \left( \prod\_{\mathfrak{F} \notin \mathbf{F}^{\nu}} \mathbf{H}\_{\mathfrak{F}} \right) \mathbf{u} \right) \mathbf{(0)} \prod\_{\mathfrak{F} \in \mathbf{F}^{\nu}} \mathbf{H}\_{\mathfrak{F}} = -p\_{\prime} \tag{\text{IX. 2.36}}$$

Where is had used strongly the property xu(0) = (sx)u(0). xS(t <sup>C</sup>). Then s*p* = det(s)*p*, sW, which implies that *p* = *q* H, *q*S(t <sup>C</sup>). Given that

$$p = \mathbf{u}(0) \prod\_{a \in \Phi^+} \mathbf{H}\_a = q \prod\_{a \in \Phi^+} \mathbf{H}\_{a'} \quad \forall \quad q \in \mathbf{S}(\mathbf{t\_C}) \tag{18.2.37}$$

then *q* = u(0). Therefore

$$\sum\_{\mathbf{F} \in \mathfrak{G}^+} \left( \left( \prod\_{\alpha \notin \mathbf{F}} \mathbf{H}\_{\alpha} \right) \mathbf{u} \right) \mathbf{(0)} \prod\_{\alpha \in \mathbf{F}} \mathbf{H}\_{\alpha} = \mathbf{u}(\mathbf{0}) \prod\_{\alpha \in \Phi^+} \mathbf{H}\_{\alpha} \dots$$

Applying IX. 2.1 (1) and the lemma IX. 2. 2, we have

$$\mathbf{f}(\mathbf{l}) = \mathbf{c}\_{\mathbf{j}} \mathbf{M}\_{\mathbf{g}} \left( \prod\_{\alpha \in \Phi^{+}} \mathbf{H}\_{\alpha} \right) \frac{\pi(\mathbf{h})}{\Delta(\exp(\mathbf{th})) \Big|\_{\mathbf{l} = \mathbf{l}}} \mathbf{f}(\mathbf{h}) \mathbf{F}\_{\mathbf{f}}^{\top}(\exp(\mathbf{th})) \Big|\_{\mathbf{h} = 0, \mathbf{t} = \mathbf{l}, \mathbf{\tilde{\mathcal{R}}} = \mathbf{l}'} \tag{\text{IX. 2.38}}$$

Considering u(h) = (h)/((exp(h)), and given that C(g), with u = 1, in Ue = W. Then uC(W)W, and u(0) = 1. Then

$$\mathbf{f}(1) = \mathbf{c} \mathbf{M}\_{\\$} \boldsymbol{\mathfrak{au}}(0) \boldsymbol{\mathfrak{B}}(0) \mathbf{F}^{\mathrm{T}}(1),$$

or equivalently Ff T(1) = CGf(1).

### **IX. 3. Integral transforms in algebraic analysis**

In this section, we state some of the foundational results we develop in order to study Hecke categories. A key idea of algebraic analysis is to replace the functions and distributions of harmonic analysis by the algebraic systems of differential equations that they satisfy. One can view this as a form of categorification where the resulting *D*-modules play the role of the original functions and distributions, and categories of *D*-modules play the role of generalized function spaces. For example, the exponential functionf(x) = eλx, is characterized by the algebraic equation (∂x − λ)f(x) = 0,and hence is a solution of the *D*-module *D*A1/*D*A1(∂x − λ). Similarly, the delta distribution δλ, is

characterized by the algebraic equation (x − λ)f(x) = 0, and hence is a solution of the DmoduleDA1/DA1(x − λ).Natural operations in harmonic analysis are given by integral transforms acting on function spaces

$$\mathbf{f}(\mathbf{x}) \mapsto (\mathbf{K} \ast \mathbf{f})(\mathbf{y}) = \int \mathbf{f}(\mathbf{x}) \mathbf{K}(\mathbf{x}, \mathbf{y}) d\mathbf{x},\tag{\Pi \mathbf{x}. 3.1}$$

where K(x, y), is an integral kernel. For example, (one normalization of) the Fourier transform on the real line is given by the integral kernel K(x, y) = e−2πixy. In continued analogy, natural operations in algebraic analysis are given by integral transforms acting as functors between categories of *D*-modules.In this context, derived versions of tensor product and push forward replace multiplication and integration respectively. To be more precise, given varieties X, Y, and a *D*-module *K*, on the product X × Y, one defines a functor on derived categories of *D*-modules

$$\mathcal{D}(\mathbf{x}) \to \mathcal{D}(\mathbf{Y}),\tag{\text{D.3.2}}$$

with the rule of correspondence

Y\* X ( ), *<sup>F</sup> F K* (IX. 3.3)

by pulling back from X, to the product X × Y, tensoring with the integral kernel *K*, and then pushing forward to Y, via the natural diagram

$$\mathbf{X} \xleftarrow{\pi\_{\mathbf{X}}} \mathbf{X} \times \mathbf{Y} \xrightarrow{\pi\_{\mathbf{Y}}} \mathbf{Y}\_{\prime} \tag{\mathbb{R}} \\ \tag{\mathbb{R}} \\ \tag{\mathbb{X}} . \mathbf{3.4} \tag{\mathbb{R}}$$

For example, the geometric Fourier transform of Malgrange, an autoequivalence of *D*modules on A1, is given by the integral kernel *K*= *D*Ax1×Ay1/*D*Ax1×Ay1(∂x− iy), with solution K(x, y) = eixy. The classical Fourier transform of a solution of a *D*-module *F*, is a solution of the geometric Fourier transform of *F*.

For another example, given a correspondence of varieties

$$\mathbf{X} \xleftarrow{f} \mathbf{Z} \xrightarrow{f} \mathbf{Z} \xrightarrow{g} \mathbf{Y}, \tag{\mathbb{X}.3.5}$$

one defines a functor on derived categories of *D*-modules by the similar formula from

*D*(X) *D*(Y): \* g f ), *F F* (IX. 3.6)

By the projection formula, this functor coincides with the integral transform given by the integral kernel *K*= (f × g)\**O*Z on the product X × Y. In general, integral transforms can be interpreted as operations on systems of differential equations, transforming solutions to one system into solutions of a new (and potentially more accessible)system. The theory of integral transforms for *D*-modules has been developed and applied to a host of problems in integral geometry and analysis, in particular to the study of the Radon, Laplace and Penrose transforms, starting with the influential paper of Brylinski [Br] and continuing in the beautiful work of Goncharov, Kashiwara, Schapira, D'Agnolo [22, 41] and others. The *D*module approach allows one to separate the algebraic and geometric aspects underlying a system of differential equations from the analytic problems involving solvability in different function spaces, allowing one to obtain powerful general results.

### **IX. 4. Revisited integrable square representations**

Consider to G, a real reductive group of inner type and such that 0G = G. Consider the space E2(G), of equivalence classes of irreducible square integrable representations of G, that is to say, the space

$$\mathsf{E}\mathfrak{e}(\mathsf{G}) = \langle \mathsf{o}\in[\mathsf{o}] \mid \left| \pi\_{\mathfrak{o}}(\mathsf{g}) \right| \, 2 \lhd \alpha, \forall \ \mathsf{g} \in \mathsf{G}, \text{ and } (\pi\_{\mathfrak{o}}, \mathsf{H}\_{\mathfrak{o}}) \in \mathsf{o} \rangle,\tag{\mathsf{D}X.4.1}$$

Where

106 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

*D*A1/*D*A1(∂x − λ). Similarly, the delta distribution δλ, is

transforms acting on function spaces

on derived categories of *D*-modules

with the rule of correspondence

pushing forward to Y, via the natural diagram

the geometric Fourier transform of *F*.

For another example, given a correspondence of varieties

harmonic analysis by the algebraic systems of differential equations that they satisfy. One can view this as a form of categorification where the resulting *D*-modules play the role of the original functions and distributions, and categories of *D*-modules play the role of generalized function spaces. For example, the exponential functionf(x) = eλx, is characterized by the algebraic equation (∂x − λ)f(x) = 0,and hence is a solution of the *D*-module

characterized by the algebraic equation (x − λ)f(x) = 0, and hence is a solution of the DmoduleDA1/DA1(x − λ).Natural operations in harmonic analysis are given by integral

where K(x, y), is an integral kernel. For example, (one normalization of) the Fourier transform on the real line is given by the integral kernel K(x, y) = e−2πixy. In continued analogy, natural operations in algebraic analysis are given by integral transforms acting as functors between categories of *D*-modules.In this context, derived versions of tensor product and push forward replace multiplication and integration respectively. To be more precise, given varieties X, Y, and a *D*-module *K*, on the product X × Y, one defines a functor

 *D*(X) *D*(Y), (IX. 3.2)

by pulling back from X, to the product X × Y, tensoring with the integral kernel *K*, and then

For example, the geometric Fourier transform of Malgrange, an autoequivalence of *D*modules on A1, is given by the integral kernel *K*= *D*Ax1×Ay1/*D*Ax1×Ay1(∂x− iy), with solution K(x, y) = eixy. The classical Fourier transform of a solution of a *D*-module *F*, is a solution of

one defines a functor on derived categories of *D*-modules by the similar formula from

f x K f y f x K(x, y)dx ( ) ( )( ) ( ) , (IX. 3.1)

Y\* X ( ), *<sup>F</sup> F K* (IX. 3.3)

X X Y Y X Y , (IX. 3.4)

f g X Y *Z* , (IX. 3.5)

$$||\pi\_{\sigma}(\mathbf{g})||^{2} = \left[c|\pi\_{\sigma}(\mathbf{g})|^{2}\mathbf{dg} \right. \\ \left. < \infty, \text{ } \forall \ \mathbf{g} \in \mathbf{G},$$

If v, w(H)K, then the coefficients to the classes [], comes given by the map

$$\circ\_{\text{V},\text{w}}\colon\mathbf{G}\to\ell^{\prime}(\mathbf{G}),\tag{\text{D}}$$

whose rule of correspondence is:

$$\mathbf{g} \mapsto \pi\_{\mathfrak{g}}(\mathbf{g})\mathbf{v}, \mathbf{w} >\_{\prime} \quad \forall \quad \mathbf{v}, \mathbf{w} \in (\mathbf{H}\_{\sigma})\_{\mathbf{K}} \tag{\text{D. 4.3}}$$

where cv, w(g)C (G), since (H)K, is rapidly increasing which is also a finite Z(g)- module (dim ZG(g)cv, w <). Then cv, w, is a cuspidal form on G. Therefore cv, w(g)0C (G). Then is had the theorem:

**Theorem IX. 4.1.**E2(G), is non-vanishing if and only if G, have a compact subgroup.

*Proof.* First we will demonstrate the implication ). If E2(G) , then exist a integrable square representation (H), such that

$$||\mathbf{c}\_{\mathbf{v},\mathbf{w}}(\mathbf{g})||^{2} = \int\_{\mathbf{C}} |\Leftarrow \pi\_{\sigma}(\mathbf{g})\mathbf{v}\_{\prime}\ \mathbf{w}\succsim\rangle^{2} \mathbf{dg} \prec \infty,\ \forall\ \mathbf{C}\_{\mathbf{v},\mathbf{w}}\in^{0}\in\langle\mathbf{G}\rangle,$$

Then 0C (G) {0}. Therefore G, not include non-compact subgroups of Cartan. Then G, have only compact subgroups of Cartan.

For other side, if G, have Cartan subgroups of G, then by the fundamental theorem of Harish-Chandra8,E2(G) . Then we assume that T, is one of such Cartan compact subgroups of G. We write to T, as T = ZT0.

Let the character , as the map or homomorphism

$$
\zeta\_{\mu} \colon \mathbb{Z}(\mathfrak{g}) \to \mathbb{C} \tag{\mathbb{D}X.4.4}
$$

with rule of correspondence

$$\mathbf{z} \mid \to \ulcorner\_{\mathfrak{v}\mu} (\mathbf{z}) = \mu \diamond \mathfrak{y} (\mathbf{z}),\tag{\mathbb{D}X.4.5}$$

where (t), is the Harish-Chandra isomorphism associated with T. Likewise, if we consider T0, and given that T0 = d()(), and = d(), then

$$
\zeta\_{\mu}(\mathbf{z}) = \Lambda(\mu) \diamond \gamma(\mathbf{z}) = \Lambda \mu \text{ o } \mathcal{F}^{\Gamma} \pi(\mu), \tag{\mathbb{D}X.4.6}
$$

If f0C (G), then FfC(T). Endeed, given that Ff T(t), is the continuous map:

$$\text{Fi} \colon ^{0}\ell(\text{G}) \to \text{C}^{\bullet}(\text{T}),\tag{\text{D}X.4.7}$$

with rule of correspondence

$$\mathbf{f} \mid \rightarrow \mathbf{F} \\ \mathbf{f} \mathbf{\bar{r}} (\mathbf{t}) = \Lambda \mathbf{r} (\mathbf{t}) [\mathbf{r} \mathbf{f} (\mathbf{g} \mathbf{t} \mathbf{g}^{-1}) \mathbf{d} \mathbf{t}] \tag{\text{D'} 4.8}$$

with T', an orbit of T, and since FTzf(t) = (t) o Ff T(t), then Ff T(t)C(T). Then the Peter-Weyl theorem implies that T ,

i. Ff = T(Ff) ()

If in particular we consider a hC(T), T-central then

$$\mathbf{h}^{\circ}(\mu) = \left| \operatorname{tr} \mathbf{h}(\mathbf{t}) \operatorname{conv} \mathbf{j}(\zeta\_{\mu}(\mathbf{t})) \mathbf{d}\mathbf{t}, \right| $$

Then the second part of the theorem IX. 2. 1, implies that exist CG 0, (constant) such that

$$\mathfrak{wF}^{\mathsf{T}}(1) = \mathsf{C}\mathfrak{c}\mathfrak{f}(1)\_{\mathsf{V}}$$

or equivalently

<sup>8</sup> **Theorem.** If G, contains a compact Cartan subgroup then G, has irreducible square integrable representations

$$\mathbf{f}(\mathbf{l}) = \tilde{\mathbf{C}}\_{\mathbf{G}}(\boldsymbol{\varpi} \mathbf{F}\_{\mathbf{f}}^{\mathrm{T}})(\mathbf{l}) = \tilde{\mathbf{C}}\_{\mathbf{G}} \sum\_{\mu \in \mathbf{T}^{\wedge}} \left( \prod\_{\alpha \in \Phi^{+}} (\Lambda(\mu), \alpha) \right) (\mathbf{F}\_{\mathbf{f}}^{\wedge})(\mu) \lambda$$

Then, if z*Z*(gC), then we have that

ii. Fzf = (z)Ff,

on T', and therefore on T, or their corresponding dual:

$$(\mathbf{F}\_{\mathbf{z}\mathbf{f}})^\wedge(\mu) = \Lambda(\mu)(\gamma(\mathbf{z}))(\mathbf{F}\mathbf{f}^\wedge(\mu)) = \Lambda(\mu)(\mathbf{F}\_{\mathbf{z}\mathbf{f}}\mathbf{f}(\mu)),$$

Which is correct, since (), is a character T0 = {tT Ad(t)Ff T = Ff T, Ff <sup>T</sup>C(T)}. Therefore finally and considering the map

$$
\chi\_{\sigma} \colon \mathbb{Z}(\mathfrak{g}\_{\mathbb{C}}) \to \subset \tag{\mathbb{C} \,\, 4.9}
$$

that extends as infinitesimal character of , E2(G), to , from T', to T, then

$$\mathbf{Z(g\_{C})f = f\_{\prime} \quad \forall \quad f \in \mathbb{F}^{0}} \in \text{(G)},\tag{13.4.10}$$

of where

108 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

only compact subgroups of Cartan.

with rule of correspondence

subgroups of G. We write to T, as T = ZT0.

Let the character , as the map or homomorphism

T0, and given that T0 = d()(), and = d(), then

If f0C (G), then FfC(T). Endeed, given that Ff

with T', an orbit of T, and since FTzf(t) = (t) o Ff

,

h

If in particular we consider a hC(T), T-central then

Ff

f ∣Ff

with rule of correspondence

theorem implies that T

 ()

i. Ff = T(Ff)

or equivalently

cv, w(g)2 = G<(g)v, w>

Then 0C (G) {0}. Therefore G, not include non-compact subgroups of Cartan. Then G, have

For other side, if G, have Cartan subgroups of G, then by the fundamental theorem of Harish-Chandra8,E2(G) . Then we assume that T, is one of such Cartan compact

*Z*(g) C, (IX. 4.4)

z ∣(z) = (z), (IX. 4.5)

where (t), is the Harish-Chandra isomorphism associated with T. Likewise, if we consider

(z) = () (z) = FTzf(), (IX. 4.6)

() = T h(t)conj((t))dt,

T(1) = CGf(1),

Then the second part of the theorem IX. 2. 1, implies that exist CG 0, (constant) such that

<sup>8</sup> **Theorem.** If G, contains a compact Cartan subgroup then G, has irreducible square integrable representations

Ff

T(t), is the continuous map:

T(t), then Ff

T: 0C (G) C(T), (IX. 4.7)

T(t) = T(t)T'f(gtg)dt, (IX. 4.8)

T(t)C(T). Then the Peter-Weyl

2dg <, Cv, w0C (G),

$$\mathbf{z}\mathbf{f} = \chi\_{\sigma}(\mathbf{z})\mathbf{f},\tag{\text{I\'X}}\tag{\text{I\'X}}\tag{\text{1\'X}}\tag{\text{1\'X}}$$

Thus has been demonstrated that:

**Theorem IX. 4.2.**Let E2(G), be, then there is T , such that ((), ) 0, (gC, t <sup>C</sup>), and such that the infinitesimal character of , is X().

From the before theorem is deduced the following corollary:

**Corollary IX. 4.1.**Let K , be then number of classes E2(G), such that (H)K() 0, is finite.

*Proof*. Consider the Casimir operator corresponding to B. Let CK = Ct, to K, corresponding to B. Let X1, , X*n*, be an orthogonal base of p, realative to B. Let

$$\mathbf{C} = \Sigma(\mathbf{X})^2,\tag{\text{I}\mathbf{X}.4.12},\tag{\text{I}\mathbf{X}.4.12}$$

Then C = CK + Cp. Fixing K , and considering , an eigenvalue of CK, on any representant of the class , we note that

a. If (, H), is an unitary representation of G, with C, acting for cI, and if HK() 0, then c .

Endeed, if vHK, such that v = 1, then

$$\mathbf{v} \cdot \mathbf{c} = \mathbf{c} \cdot \mathbf{v}, \mathbf{v} \cdot \mathbf{b} = \mathbf{c} \cdot \mathbf{C} \mathbf{v}, \mathbf{v} \cdot \mathbf{b} = \mathbf{c} \cdot \mathbf{C} \mathbf{x} \mathbf{v}, \mathbf{v} \cdot \mathbf{b} = \mathbf{c} \cdot \mathbf{C} \mathbf{v}, \mathbf{v} \cdot \mathbf{b} = \mu\_{\gamma} - \Sigma \mathbf{c} \mathbf{X} \mathbf{v}, \mathbf{X} \mathbf{y} \cdot \mathbf{x} \le \mu\_{\gamma}$$

If E2(G), then , is the character of (T0) , such that (C).

Let = ½ , be, then

$$\chi\_{\mathfrak{a}}(\mathbf{C}) = \|\Lambda\_{\mathfrak{a}}\| - \|\mathfrak{p}\|\,\tag{\text{TX 4.13}}$$

For

$$\mathbf{c} = \mu\_{\uparrow} - \Sigma \lhd \mathbf{X} \mathbf{v} , \lambda \mathbf{v} \rhd \succ \nu \text{ } \forall \text{ } \mathbf{v} \in \text{Hx}(\mathfrak{y}), \tag{18.4.14}$$

such that v = 1. Thus

$$||\Lambda\_{\sigma}||^2 \le ||\mathfrak{p}||^2 + \mathfrak{p}\_{\mathcal{V}} \tag{\text{I\"\X 4.15\" }}$$

Then the unique number of possibilities to obtain an infinitesimal character of integrable square whose isotopic component HK() 0, is c , that is to say, whose infinitesimal character (C), let be such that (C).

Since there is an equivalence between the finite number of irreducible (g, K)-module with infinitesimal character , in the class , then the number of E2(G), such that HK() 0, is finite.

## **Cohomological Induction and Securing of Generalized G-Modules by G(w)-Orbits to Infinite Dimensional Representations of Lie Groups**

(Talk given in Ivano-Frankivsk, Ukraine, 2009 [15])

### **X. 1. Vogan program**

110 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

(C) = , (IX. 4.13)

Then the unique number of possibilities to obtain an infinitesimal character of integrable square whose isotopic component HK() 0, is c , that is to say, whose infinitesimal

Since there is an equivalence between the finite number of irreducible (g, K)-module with infinitesimal character , in the class , then the number of E2(G), such that HK() 0, is

, such that (C).

c = <Xjv, Xjv>, vHK(), (IX. 4.14)

22 + , (IX. 4.15)

If E2(G), then , is the character of (T0)

character (C), let be such that (C).

Let = ½ , be, then

such that v = 1. Thus

For

finite.

In the study by Vogan, is to establish the following problem of the representation theory (PTR). Given an irreducible representation (, V), it is possible to give a structure of Hilbert space on V, to have to , like an unitary representation.

Strictly speaking this bears the questions:

i. We can make that V, takes a bilinear hermitian G*-* invariant form < , >?

ii. If whose form exist, ¿Is the form < , >, positive define?

The goal of the Vogan program is to analyze some difficulties that arise when one tries to study this program accurately this program. The difficulties have their origin exactly in the flexibility of the definition of a representation (, V), of G, a topological group.

Typically wants to want realize a representation of G, on a space of functions. If G, acts on the set X,then G, it acts on functions on X, for

$$\left[\mathbb{E}\mathfrak{(g)f}\right](\mathbf{x}) = \mathbf{f}(\mathbf{g}^{-1}\mathbf{x}), \ \forall \mathbf{x} \in \mathbf{X} \text{ and } \mathbf{g} \in \mathbf{G},\tag{\text{\textquotedblleft}X} \tag{\text{\textquotedblleft}Y}$$

The difficulties arise when we try to decide exactly which space of functions on X, it is necessary to consider. Since if G, is the Lie group acting smoothly on a manifold X, then one can consider C(X) , C (X) <sup>c</sup> , C X <sup>c</sup> ( ) or C (X) .

Harish - Chandra establish that: Each class of infinitesimal equivalence of admissible irreducible representations contains at most a class of equivalence of irreducible unitary representations of G, this is:

G G <sup>u</sup> , (X. 1.2)

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

with G, a class of infinitesimal equivalence of classes of admissible representations of G. This result in some way establishes a solution or general answer to the problem (PTR) of the Vogan program.

However, requires to specific the type in a positive defined way that will guarantee the Hermitian structure of the representation algebra of VK, that induces to < , >, K, like a Hermitian G-invariant form and thus endows to VK, like unitary representation in this class of infinitesimal equivalence, that is to say, the existence will have been guaranteed of at least an irreducible unitary representation in the class of unitary equivalence that is representation of G.

### **X. 2. Dificulties**

The existence in a G-invariant continuous Hermitian form < , >, on V, implies the existence of < , >, K, on VK, but the reciprocal is not true. Since VK is dense in V, exists a continuous extension at most of < , >, K, to V, but the extension cannot exist.

An important observation that one deduces from some concrete examples is that the Hermitian form can be defined only on "representations appropriately thin or small", at least it is what Vogan puts in evidence inside its program for each class of infinitesimal equivalence.

In the different spaces to consider C(X) , C (X) <sup>c</sup> , C X <sup>c</sup> ( ) ó C X( ) , the space C X <sup>c</sup> ( ), offers as a suitable candidate and appropriately small and it can be endowed with a invariant Hermitian form. The space is generally more "fat" like space to admit a invariant Hermitian form.

We define the space V\*, of continuous linear functionals on V, endowed of the strong topology, that is

$$\mathbf{V^{\*}} = \{\mathbf{\tilde{\xi}} \in \mathbf{C(V)} \, \middle| \, \tilde{\xi} : \mathbf{V} \to \mathbf{C}, \, \mathbf{\tau(\xi)} : \mathbf{W\_{\varepsilon}(B)}\}$$

$$= \mathbf{\mathfrak{T}(V, \mathbf{C})}, \tag{\text{X.2.1}}$$

 ( ), is the defined strong topology on neighborhood base in the origin consisting of groups W B( ), defined by the spaces

$$\operatorname{SW}\_{\mathfrak{c}}(\mathcal{B}) = \{ \xi \in \mathbb{E} \, \ast \, \middle| \, \operatorname{sup}\_{\mathbf{e} \in \mathcal{B}} \left| \xi(\mathbf{e}) \right| \le \varepsilon \} \subset \mathbb{E}^\*,$$

with B V , bounded.

**Theorem X. 2.1.** (Casselman, Wallach y Schmid) [46-48].

Suppose that (, V), is an irreducible admissible representation of Lie groups G, on a Banach space V. We Define

$$\begin{aligned} (\pi^{\circ \circ}, \mathbf{V}^{\circ \circ}) &= \textit{analtic.} \textit{vector.in.V.} \\ (\pi^{\circ \circ}, \mathbf{V}^{\circ \circ}) &= \textit{differential.vector.in.V.} \\ (\pi^{\circ \circ}, \mathbf{V}^{\circ \circ}) &= \textit{distribution.} \textit{vector.in.V} \\ &= \textit{dual.de.(V')}^{\circ \circ} \\ (\pi^{\circ \circ}, \mathbf{V}^{\circ \circ}) &= \textit{vectors.of.hyperfunctions.in.V.} \\ &= \textit{dual.of.(V')}^{\circ \circ}. \end{aligned}$$

Each one of these four representations is a soft representation of G, in the class of infinitesimal equivalence of , and each one only depends on that equivalence class.

The inclusions

112 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

extension at most of < , >, K, to V, but the extension cannot exist.

In the different spaces to consider C(X) , C (X) <sup>c</sup> , C X <sup>c</sup> ( ) ó C X( )

Vogan program.

representation of G.

**X. 2. Dificulties** 

equivalence.

topology, that is

W B( ), defined by the spaces

**Theorem X. 2.1.** (Casselman, Wallach y Schmid) [46-48].

with B V , bounded.

form.

with G, a class of infinitesimal equivalence of classes of admissible representations of G. This result in some way establishes a solution or general answer to the problem (PTR) of the

However, requires to specific the type in a positive defined way that will guarantee the Hermitian structure of the representation algebra of VK, that induces to < , >, K, like a Hermitian G-invariant form and thus endows to VK, like unitary representation in this class of infinitesimal equivalence, that is to say, the existence will have been guaranteed of at least an irreducible unitary representation in the class of unitary equivalence that is

The existence in a G-invariant continuous Hermitian form < , >, on V, implies the existence of < , >, K, on VK, but the reciprocal is not true. Since VK is dense in V, exists a continuous

An important observation that one deduces from some concrete examples is that the Hermitian form can be defined only on "representations appropriately thin or small", at least it is what Vogan puts in evidence inside its program for each class of infinitesimal

as a suitable candidate and appropriately small and it can be endowed with a invariant Hermitian form. The space is generally more "fat" like space to admit a invariant Hermitian

We define the space V\*, of continuous linear functionals on V, endowed of the strong

V\* { C(V) V C, W B : ( ) : ( )}

( ), is the defined strong topology on neighborhood base in the origin consisting of groups

W B E \* e E\*, e B ( ) { sup ( ) }

(V,C), (X. 2.1)

, the space C X <sup>c</sup> ( ), offers

$$\mathbf{V}^{\mathrm{\cir}} \subset \mathbf{V}^{\mathrm{\cir}} \subset \mathbf{V} \subset \mathbf{V}^{-\mathrm{\cir}} \subset \mathbf{V}^{-\mathrm{\cir}},\tag{8.2.2}$$

are continuous, with dense image. Anyone Hermitian form < , >K, on VK, expands uniquely to G-invariants continuous Hermitian forms < , >, and < , >, on V , and V.

The four representations VVV , , and V , are called minimal, smooth, distribution and maximal globalizations respectively.

Except for , be , a representation of finite dimension (such that all the spaces in the theorem X. 1, they are the same one). The Hermitian form cannot extend continually to the maximal distribution or globalization V , and V . For what could be necessary the use of representations of G, built on spaces of holomorphic sections of vector bundles and generalizations. Is in this part where later on inside this work it will be to induce and to generalize the G-modules of Harish-Chandra [5, 21] and [37], to be able to be related with the globalizations of Wong.

Now then, since through this way are obtained unitary representations, is necessary to specify a similar way or analogous to the followed to the obtaining of minimal and differentiable globalizations with the certainty of that the Hermitian forms can be defined on the representations.

### **X. 3. Minimal and maximal globalizations**

We consider a result of maximal globalizations of Harish-Chandra:

**Theorem X. 3.1.** (Wong) [36]. We assume that the admissible representation V, is the maximal globalization of the (q, LK)-module underlying. Be the G-invariant holomorphic vector bundle on X = G/L, corresponding of ( V X ) . Then the operator , for the Dolbeault cohomology has closed range and such that each one of the spaces p,q H (X, ) takes a soft representation of G. Each one of these admissible representations is the maximal globalization of their underlying module of Harish-Chandra.

*Proof***:** [17].

**Def. III.1.**Suppose G, is real and reductive. q,is a parabolic subalgebra of the complexified Lie algebra gC, and L, is the Levi factor ofq. A (q, L)-representation (, V) is to say be admissible if the representation , in L, is admissible. In this case the method of Harish-Chandra of V, is the (q, LK)-module VLK, of vectors LK-finites in V.

The theorem of Wong [36], establish that the Dolbeault cohomology let to the maximal globalizations in great generality. This means that there is not possibility to find Hermitian invariant forms on these representations of Dolbeault cohomology except in the case of finite dimension. That is to say, spaces are obtained too "fat" to be able to identify and to classify the infinitesimal equivalence classes of representations of Lie groups and to identify the unique classes of unitary representations corresponding to each one of the mentioned infinitesimal equivalence classes.

For it is necessary to develop a way to modify the Dolbeault cohomology to produce minimal globalizations in more grade that the maximal. Essentially we can follow the ideas of Serre, which are based on the realization of representations of minimal globalization obtained about generalized flag manifolds achieved first by Bratten. Of the duality used to define the maximal globalization the question it does arise, How can you identify the dual topological space of a cohomological space of Dolbeault on a complex compact neighborhood?

The question is interesting in the simplest case: Suppose an X C , is open set and H(X), is the space of holomorphic functions X, in a topological vector space X, using the topology of uniform convergence of all the derivatives on compact sets.

For what would H(X), be natural alone to be questioned, what is the dual space?

This last question has a simple answer. Be C X <sup>c</sup> ( , ), *densidades* the space of supported distributions compactly on X. We can think in this like the space of complex supported 2-forms compactly (or (1, 1)-forms) on X, with coefficients of generalized functions. For what, with more generality we write:

(p,q), A (X) (p,q) <sup>c</sup> Supported forms compactly on X, with coefficients of generalized functions,

The differential operator of Dolbeault , map (p, q)-forms to (p, q + 1)-forms and it preserves the support,

$$\overline{\mathcal{O}} : \mathcal{A}\_{\mathfrak{c}}^{\ (1,0), -\circ}(\mathcal{X}) \to \mathcal{A}\_{\mathfrak{c}}^{\ (1,1), -\circ}(\mathcal{X}) = \mathcal{C}\_{\mathfrak{c}}^{-\circ}(\mathcal{X}, denotes) \tag{\mathcal{X}.3.1}$$

Then

114 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

globalization of their underlying module of Harish-Chandra.

uniform convergence of all the derivatives on compact sets.

Chandra of V, is the (q, LK)-module VLK, of vectors LK-finites in V.

*Proof***:** [17].

infinitesimal equivalence classes.

what, with more generality we write:

neighborhood?

(p,q), A (X) (p,q) <sup>c</sup>

functions,

vector bundle on X = G/L, corresponding of ( V X ) . Then the operator , for the Dolbeault cohomology has closed range and such that each one of the spaces p,q H (X, ) takes a soft representation of G. Each one of these admissible representations is the maximal

**Def. III.1.**Suppose G, is real and reductive. q,is a parabolic subalgebra of the complexified Lie algebra gC, and L, is the Levi factor ofq. A (q, L)-representation (, V) is to say be admissible if the representation , in L, is admissible. In this case the method of Harish-

The theorem of Wong [36], establish that the Dolbeault cohomology let to the maximal globalizations in great generality. This means that there is not possibility to find Hermitian invariant forms on these representations of Dolbeault cohomology except in the case of finite dimension. That is to say, spaces are obtained too "fat" to be able to identify and to classify the infinitesimal equivalence classes of representations of Lie groups and to identify the unique classes of unitary representations corresponding to each one of the mentioned

For it is necessary to develop a way to modify the Dolbeault cohomology to produce minimal globalizations in more grade that the maximal. Essentially we can follow the ideas of Serre, which are based on the realization of representations of minimal globalization obtained about generalized flag manifolds achieved first by Bratten. Of the duality used to define the maximal globalization the question it does arise, How can you identify the dual topological space of a cohomological space of Dolbeault on a complex compact

The question is interesting in the simplest case: Suppose an X C , is open set and H(X), is the space of holomorphic functions X, in a topological vector space X, using the topology of

This last question has a simple answer. Be C X <sup>c</sup> ( , ), *densidades* the space of supported distributions compactly on X. We can think in this like the space of complex supported 2-forms compactly (or (1, 1)-forms) on X, with coefficients of generalized functions. For

Supported forms compactly on X, with coefficients of generalized

For what would H(X), be natural alone to be questioned, what is the dual space?

$$\mathbf{H(X)} \equiv \mathbf{A\_c}^{(1,1), -\alpha}(\mathbf{X}) / \overline{\overline{\partial}} \overline{\mathbf{A\_c}^{1,0}(\mathbf{X})},\tag{\text{\textquotedblleft}{}} \tag{\text{\textquotedblleft}{}} \tag{\text{\textquotedblleft}{}} \overline{\mathbf{X}}^{1,0}(\mathbf{X})\tag{\text{\textquotedblleft}{}}$$

Here the line **\_\_\_\_\_\_\_\_**on 1,0 A (X) <sup>c</sup> , it denotes closing of the space 1,0 A (X) <sup>c</sup> .

To open X, in C, the image of , is automatically closed, for that the line**\_\_\_\_\_\_\_\_** is not necessary. However this formulation has an immediate extension for any complex manifolds X (replacing 1, and 0, for the dimensions *n*, and *n*1).

Let us enunciate the generalization of Serre [13]:

**Theorem X. 3.2.** (Serre). Suppose that X, a complex manifold of dimension *n*, a vector holomorphic bundle on X, and , is the canonical bundle of lines (of (n, 0)-forms have more than enough X). We Define

$$\mathbf{A}^{0,p}(\mathbf{X}, \mathbf{v}) = \text{Space of the Soff } \boldsymbol{\nu}\text{-values } (0, p)\text{-forms on } \mathbf{X}\_{\boldsymbol{\nu}} \tag{\mathbf{X}.3.3}$$

and

$$\mathbf{A}\_{\mathbf{c}}^{\ \ (0,p), -\alpha}(\mathbf{X}, \mathbf{v}) = \text{Space of } \mathbf{v}\text{-values compactly supported}$$
 
$$(0, p)\text{-forms with coefficients of generalized functions}\tag{\text{X.3.4}}$$

Proof. [14].

We define the topological cohomology of Dolbeault de X, with values in , as

$$\mathbf{H}^{0,\mathsf{p}}\_{\mathrm{top}}(\mathbf{X},\mathbf{v}) = [\ker \overline{\mathcal{C}}](\mathbf{A}^{0,\mathsf{p}}(\mathbf{X},\mathbf{v}) / \overline{\widetilde{\mathcal{C}}} \overline{\mathbf{A}^{\mathsf{p}-1,0}(\mathbf{X},\mathbf{v})},\tag{\mathsf{X}.3.5}$$

Indeed, a quotient of the usual Dolbeault cohomology and carry a locally convex topology also usual. Similarly we define

$$\mathcal{H}^{0,p}\_{\text{top}}(\mathbf{X},\mathbf{v}) = [\ker \overline{\mathcal{C}}](\mathcal{A}^{(0,p),-\circ}(\mathbf{X},\mathbf{v}) / \overline{\overline{\mathcal{C}}} \overline{\mathcal{A}\_{\text{c}}^{(p-1,0),-\circ}(\mathbf{X},\mathbf{v})},\tag{\text{X. 3.6}}$$

the topological cohomology of Dolbeault with compact supports. Then a natural identification exists

$$\mathcal{H}^{0,\mathsf{p}}\_{\mathrm{top}}(\mathcal{X},\mathfrak{I})^{\ast} \equiv \mathcal{H}^{0,\mathsf{n}-\mathsf{p}}\_{\mathrm{C},\mathsf{top}}(\mathcal{X},\mathfrak{Q}\otimes\mathfrak{I}^{\ast}),\tag{\mathsf{X}.\ 3.7}$$

Here \*, is the vector holomorphic bundle dual of .

We consider the following case. When X, is compact then the sub index c, is not added more, and the operators , automatically have closed range.

The central idea in this part of the program of Vogan is the desire to build representations of real reductive groups G, beginning with a measurable complex flag manifold X = G/L, and using G-equivariants holomorphic bundles of lines X. Indeed, if we have X = K/T, with r = 0, in 0,r H (X, <sup>c</sup> <sup>ν</sup>), then 0,r H (X, <sup>c</sup> <sup>ν</sup>), in the irreducible G-module on the corresponding coherent sheaf of O( ), global sections of the complex holomorphic bundle , and the relationship among - cohomology and the sheaf is simple and it is given by the space

$$
\Gamma(\mathbf{X}, \mathbf{O}(\lambda)) = \mathbf{H}^0(\mathbf{X}, \mathbf{O}(\lambda)),
$$

For the infinite case, is necessary to use a finer structure of the flag manifolds like for example, the given by open orbits of flag manifolds and the continuous homomorphisms among whose open orbits to induce a classification of the irreducible representations that reside in the space <sup>0</sup> H (X,O( )), and that under the association of irreducible minimum K-types suggested by Vogan [38], the widest class in classifiable irreducible unitary representations will be obtained by the theory of Langlands.

However we will establish a special formalization of the - cohomology to be able to use Ginvariant holomorphic bundle of lines on X.

For compact G, the theorem of Borel-Weil says that all the irreducible representations of G, arise in this way as spaces of holomorphic sections of holomorphic bundles of lines.

**Def. III. 2.** Suppose that X, is a complex manifold of dimension *n*, and is a holomorphic vector bundle on X. The cohomology of (p, q)-Dolbeault compactly supported of X, with coefficients in , is for definition

$$\mathbf{H}\_{\mathbf{c}}^{0,\mathfrak{p}}(\mathbf{X},\mathbf{v})^{\mathfrak{s}} = (\ker(\overline{\mathcal{C}})(\mathbf{A}\_{\mathbf{c}}^{(\mathfrak{p},\mathfrak{q}),-\mathfrak{v}}(\mathbf{v})) / (\operatorname{Im}(\overline{\mathcal{C}}(\mathbf{A}^{(\mathfrak{p},\mathfrak{q}-1),-\mathfrak{v}}(\mathbf{v})),\tag{\mathsf{X}.3.8}$$

If , is of finite dimension then the cohomology p,q H (X, <sup>c</sup> <sup>ν</sup>), is a cohomology of Čech with compact supports of X*,* with coefficients in rhe sheaf *<sup>p</sup> O* of holomorphic p-forms with values in.

Exactly a topology natural quotient exists on this cohomology, and we can define:

$$\mathbf{H}\_{\mathbf{c},\text{top}}^{\mathrm{p,q}}(\mathbf{X},\mathbf{v}) = \mathbf{H}ausdorff\,\text{maximal}\,\,\mathrm{Quotient}\,\,\mathrm{of}$$

$$\mathbf{H}\_{\mathbf{c}}^{\mathrm{p,q}}(\mathbf{X},\mathbf{v}) = \mathrm{Ker}(\overline{\overline{\boldsymbol{c}}}) / \overline{\mathrm{Im}(\overline{\overline{\boldsymbol{c}}})},\tag{\text{\textquotedbl{}Im}(\overline{\overline{\boldsymbol{c}}})},\tag{\text{\textquotedbl{}}}$$

Then p,q H (X, top <sup>ν</sup>), it takes soft representations of G (for translation of forms).

Then clear consequence of the theorems of Serre, theorem X. 3. 2, and the theorem of Wong, theorem. X. 3. 1, using the definitions previous is have the corollary:

**Corollary X. 3.1.** (Bratten) [37]. Suppose that X, is a complex manifold G/L, and assume that admissible representation V is the minimal globalization of the (q, LK)-module. Be p,q A (X, <sup>c</sup> <sup>ν</sup>),the Dolbeault complex to with coefficients of generalized functions of compact support. Then the operator , have a closed range such that each one of the corresponding cohomological spaces p,q H (X, <sup>c</sup> <sup>ν</sup>),takes soft representation of G (on the dual of a nuclear Fréchet space). Everyone of this representations of G, is admissible and is a minimal globalization of their underlying module de Harish-Chandra.

*Proof.* (Vogan, 2000, Bratten, 2002).

116 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Here \*, is the vector holomorphic bundle dual of .

and the operators , automatically have closed range.

0,p 0,n p H (X, )\* H (X, top C,top <sup>Ω</sup> \*), (X. 3.7)

We consider the following case. When X, is compact then the sub index c, is not added more,

The central idea in this part of the program of Vogan is the desire to build representations of real reductive groups G, beginning with a measurable complex flag manifold X = G/L, and using G-equivariants holomorphic bundles of lines X. Indeed, if we have X = K/T, with r = 0, in 0,r H (X, <sup>c</sup> <sup>ν</sup>), then 0,r H (X, <sup>c</sup> <sup>ν</sup>), in the irreducible G-module on the corresponding coherent sheaf of O( ), global sections of the complex holomorphic bundle , and the relationship

<sup>0</sup> ( ( )) ( )), X,O H (X,O

For the infinite case, is necessary to use a finer structure of the flag manifolds like for example, the given by open orbits of flag manifolds and the continuous homomorphisms among whose open orbits to induce a classification of the irreducible representations that reside in the space <sup>0</sup> H (X,O( )), and that under the association of irreducible minimum K-types suggested by Vogan [38], the widest class in classifiable irreducible unitary

However we will establish a special formalization of the - cohomology to be able to use G-

For compact G, the theorem of Borel-Weil says that all the irreducible representations of G,

**Def. III. 2.** Suppose that X, is a complex manifold of dimension *n*, and is a holomorphic vector bundle on X. The cohomology of (p, q)-Dolbeault compactly supported of X, with

If , is of finite dimension then the cohomology p,q H (X, <sup>c</sup> <sup>ν</sup>), is a cohomology of Čech with

Exactly a topology natural quotient exists on this cohomology, and we can define:

0,p (p,q), (p,q 1), H (X, c c <sup>ν</sup>)\* (ker( )(A ( )) / (Im( (A ( )), (X. 3.8)

*O* 

of holomorphic p-forms with

arise in this way as spaces of holomorphic sections of holomorphic bundles of lines.

among - cohomology and the sheaf is simple and it is given by the space

representations will be obtained by the theory of Langlands.

compact supports of X*,* with coefficients in rhe sheaf *<sup>p</sup>*

invariant holomorphic bundle of lines on X.

coefficients in , is for definition

values in.

The fundamental relation of duality of minimal and maximal globalizations is given by the corollary. X. 3.1, that makes allusion to the conjecture of Serre.

The theorem demonstrated by Bratten (Bratten, 2002), is different:

He defines a sheaf of germs of global sections A(X, ν), and it demonstrates a parallel result for the cohomology of sheaves with compact support on X, with coefficients in A(X, ν).

When V, is of finite dimension, the two results are exactly the same one, being this easy to verify for that Dolbeault cohomology (with coefficients of generalized functions of compact support) and it calculates the cohomology of sheaves in that case.

The case of infinite dimension of V, comparing the corollary one enunciated previously with results of Bratten is more difficult of establish. The development of Vogan (Complex Analysis and Unitary Representations, Springer, 2003) in all the exhibitions only speech of the Dolbeault cohomology and not of the cohomology of sheaves, foreseeing that the relationship between sheaves and the Dolbeault cohomology for bundles of infinite dimension is complicated and it bears bigger difficulties that those foreseen by the own theory of representations.

For the theory of characters, they are been able to determine the infinitesimal characters of the Dolbeault cohomology of representations and it is applicable the theorem of Vogan for representations of infinite dimension [37], [38].

The same way to the Dolbeault cohomology with compact support. The weightL(u) that appears in the following corollary is therefore the infinitesimal character of the representation 0,r H (X, <sup>c</sup> <sup>ν</sup>), with , having defined by *X* = *G*/*L*.

**Corollary X. 3.2.** (Vogan). Be Z K/L K a complex compact s-dimensional submanifold of the complex n-dimensional manifold X = G/L. Be *r* = *ns*, the codimension of Z, in X. Assume that V is a (q, L)-module of infinitesimal character Lh\*, and that V, is the minimal globalization of the (q, LK)-module. Assume that L (u), is weakly anti-dominant to u, this is that L (u), is weakly dominant. Then:


*Proof*. (Complex Analysis and Unitary Representations, Springer, 2003) and of the corollary X. 3. 1.

### **X. 4.** u **— Cohomology**

Be g, a reductive Lie algebra on C. Be h, a cartan subalgebra of g, and be l, the system of positive roots to (g, h). Be b = b(P) = h P g. Be q, the subalgebra of b, enclosed b.

Be l ={(g+g)q} and correspondingP l , with

$$\mathfrak{l} = \mathfrak{l} \oplus \bigoplus\_{\alpha \in \Phi^1} \mathfrak{g}\_{\alpha \nu} \tag{\mathcal{X}.4.1}$$

and

$$\mathfrak{u} = \bigoplus\_{\alpha \in \Sigma} \mathfrak{g}\_{\alpha \nu} \tag{\mathcal{X}.4.2}$$

Then q =l u and [l,u] u. Be u =h g,and q =l u . Then q =u l u . By the theorem PBW (Poicaré-Bott-Weil),

$$\mathbf{U(d)} = \mathbf{U(l)} \oplus (\boldsymbol{\eta}\mathbf{U(d)} \oplus \mathbf{U(d)}\boldsymbol{\eta}^{-}),\tag{X.4.3}$$

to U(g), enveloping algebra of Lie algebra g.

Be V, a g-module with action. Then

$$\mathbf{C}(\mathfrak{u}, \mathbf{V}) = \operatorname{Hom}\_{\mathbb{C}}(\wedge \mathfrak{u}, \mathbf{V}),\tag{\text{\textquotedblleft}}\tag{\text{\textquotedblleft}}\mathfrak{u}, \mathbf{4}.4)$$

is a l-module under the action (X)(Y) = X((Y (adX(Y))), Xl, and Yu. Also d(X) = Xd.

The module given for (17) are the complexes of the u-cohomology.

118 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

representation 0,r H (X, <sup>c</sup> <sup>ν</sup>), with , having defined by *X* = *G*/*L*.

this is that L (u), is weakly dominant. Then:

i. 0,q H (X, <sup>c</sup> <sup>ν</sup>) 0 , under *q = r*.

0, H (X, <sup>c</sup> <sup>ν</sup>), *<sup>r</sup>* is unitary.

**X. 4.** u **— Cohomology**

Then q =l u and [l,u] u. Be u

to U(g), enveloping algebra of Lie algebra g.

Be V, a g-module with action. Then

PBW (Poicaré-Bott-Weil),

cero.

X. 3. 1.

and

The same way to the Dolbeault cohomology with compact support. The weightL(u) that appears in the following corollary is therefore the infinitesimal character of the

**Corollary X. 3.2.** (Vogan). Be Z K/L K a complex compact s-dimensional submanifold of the complex n-dimensional manifold X = G/L. Be *r* = *ns*, the codimension of Z, in X. Assume that V is a (q, L)-module of infinitesimal character Lh\*, and that V, is the minimal globalization of the (q, LK)-module. Assume that L (u), is weakly anti-dominant to u,

ii. Si max L L and V, is a irreducible representation of L, then 0, H (X, <sup>c</sup> <sup>ν</sup>), *<sup>r</sup>* is irreducible or

iii. If the module of Harish-Chandra of V, admits a Hermitian and invariant form, then

iv. If the module of Harish-Chandra of V, is unitary then the module of Harish-Chandra of

*Proof*. (Complex Analysis and Unitary Representations, Springer, 2003) and of the corollary

Be g, a reductive Lie algebra on C. Be h, a cartan subalgebra of g, and be l, the system of

u=g, (X. 4.2)

=l u

l = h lg, (X. 4.1)

. Then q =u l u

. By the theorem

), (X. 4.3)

positive roots to (g, h). Be b = b(P) = h P g. Be q, the subalgebra of b, enclosed b.

=h g,and q

Be l ={(g+g)q} and correspondingP l , with

U(g) = U(l) (uU(g) U(g)u

the module of Harish-Chandra of 0, H (X, <sup>c</sup> <sup>ν</sup>), *<sup>r</sup>* admits a Hermitian form.

### **X. 5. Generalized G-modules and ond theorem to representations of infinite dimension**

G-modules of Fréchet are induced and irreducible G-modules of infinite dimension are built whose differentiable cohomology is a cohomology of representations of applicable infinite dimension to the Langlands classification and some geometric theorems as the theorem of Borel-Weil.

**Def. X. 5.1.** For a topological G-module or simply a G-module (, V) will understand a topological vector space on which G acts via a continuous representation. Ag-module is the corresponding pre-image of a G-module of the corresponding exponential homomorphism [37, 38, 46-48].

An extension of a G-module is a open G-orbit of a holomorphic bundle of flags on Fréchet spaces [46-48].

One generalization of the extension of a G-module <sup>q</sup> H (G(w),Ο<sup>q</sup> <sup>η</sup> (E )) 0 , q s, is the case

when , is of infinite dimension. For this case is necessary to build a version of extension of G-module whose cohomology is the corresponding to a cohomology of representations of infinite dimension. s is the complex dimension of a compact maximal submanifold Z(w), of G(w), such that

$$\mathbf{Z}(\mathbf{w}) \equiv \mathbf{K} \,/\,\mathbf{K} \,\cap\,\mathbf{L},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\},\tag{\text{\textquotedblleft}},\tag{\text{\textquotedblleft}},\}\right)$$

and EG(w), the homogeneous complex holomorphic vector bundle corresponding to the open G-orbit G(w) η L, ˆ and maximum weigh .

If L = T, the extension of the G-module is reduce to set of global sections of the sheaf Oq(), of the complex holomorphic bundle of flag manifolds with maximum weigh . This is an irreducible G-module of finite dimension with maximum weigh (Theorem of Borel-Bott-Weil).

A version of extension of G-module whose cohomology is that of representations of infinite dimension can be built starting from the induction of G-modules on a differentiable cohomology defined as follows:

**Def. X. 5.2.** A generalized open G-orbit is the extension of G-module (open L-module on a differentiable cohomology) induced in the differentiable category given by the space

$$\text{Ind}^{\text{c}}(\mathcal{G}(\mathbf{w})) = \text{Ind}\_{\text{L}}^{\text{c}}\mathcal{G}(\mathbf{w}) \text{ [11]}.\tag{\text{\textquotedbl{}}}$$

**Def. IV.3.** (Bulnes, F.) [37, 38, 46-48]. A generalized G-module is the induced G-module by a differentiable cohomology of representations of infinite dimension (E,η), defined on generalized orbits of the complex homogeneous bundle

$$\mathcal{E}\_{\eta} \to \mathcal{G} / \mathcal{L} \text{ [8]}\tag{8.5.3}$$

where E, is a Fréchet space.

Using u-cohomology, continuous cohomology and the generalization of the topology on complexes of fibered holomorphic bundles of Frèchet is having that:

**Proposition X. 5.1.** (Bulnes, F.) [37, 38].

$$\mathbf{H}^{\bullet}\_{cl}(\mathfrak{u}, \mathbf{I}^{\bullet}(\mathfrak{\eta})) = \mathbf{H}^{\bullet}\_{cl}(\mathbf{G}/\mathbf{L}, \mathbf{O}(\mathbf{E}\_{\mathfrak{\eta}})),\tag{X.5.4}$$

**Proof.** [38].

Of these generalities in hand, we get immediately a description of the topological dual of Dolbeault cohomology.

Using the conjecture of Vogan on the possible extension of representations of L to modules of Harish-Chandra to G taking care that the co-border operators of the Dolbeault cohomology have all closed range, we apply this conjecture on the extension of the induced G-modules proposed to modules of Harish-Chandra of infinite dimension, obtaining the conjecture:

**Conjecture X. 5.1.** (Bulnes, F.) [38]. Suppose that Hqct(g, LK; ALE\*) is a (l, LK)-module of finite longitude, E,their corresponding representation of L, and, the make holomorphic associated bundle to G/L.Then the co-border operators of the Dolbeault cohomology are all of closed range and to the case of infinite dimension the range of the co-border operators of the Dolbeault cohomologyare closed provided certain intertwining operators 39, applied to the corresponding induced *G*-modules are modules of Harish-Chandra (which must satisfy the theorem of Vogan-Zuckerman on irreducible unitary representations of infinite dimension under character of Vogan).

Using certain technical lemmas [10], to a decomposition of the algebra u in their parts extensive y classified of the radical nilpotent part of the complex holomorphic vector bundle module a radical nilpotent bundle of Borel subalgebras [8, 12], and with pertinent generalizations of the G-modules, is have a theorem of classification of representations of infinite dimension [8]:

**Theorem X. 5.1.** (Bulnes, F.)[21, 37, 38, 41, 46, 47], and [48]. Be , denotes for Ht ct(L(w), Oq(E)), a representation of infinite dimension of (q, L) and be <sup>η</sup> E G(w), the associated homogeneous vector bundle. Then the operator , to the complex de Dolbeault A(G(w), E(q+1u)\*)L,is of closed range. Then the cohomologies Hq(G(w), Oq(E)) = 0, q s are admissible G-modules de Frèchet composition of series (These form admissible representations of finite longitude). Their modules of Harish-Chandra underlying are functors of Zuckerman [21, 48] As+t(G, M ,b, ) = A(G, L, q, ). Ehave Infinitesimal Character L, , and trivial action u (to it the generalized G-modules admit the infinitesimal G-characterG,u).

*Proof*. To that , be close is necessary that be regular in whole their domain (Wong's globalizations). Then , on Hq(G(w), Oq(*E*)), is a representative K-finite cohomology of strong harmonic L2-form. By the intertwining operator 38, 39, it is map fundamental series in harmonic forms of I(w), of Hq(G/L,*L*), with *L*, the bundle of lines and , is the unitary character of L G.Then , is regular in Aq(G, L, q, ). Then by differentiable cohomology, r*S*(F0). Then by differentiable cohomology

$$
\rho \mathcal{G}(\Phi\_0) \neq 0,\tag{\text{\textquotedblleft}pmatrix}
\tag{\text{\textquotedblleft}pmatrix}
\mathbb{C} \ \mathbb{C} \ \mathbb{S} \ \mathbb{C}
$$

 F0 Q(F) with *K*-type (0, F0) in C(*LG*, qu **C**), and *S*(F0), represent the class of cohomology non vanishing on K L. Using results of u – cohomology (lemmas of Vogan and Kostant of u – cohomology) to g-modules. This class of cohomology is the of the (g, K) – modules isomorphic to space Hq(G(w), E), when E= **C**. But **C**=**C**(u)+, and due to that the induced representations on generalized G-modules like the defined in the Def. X. 5. 1, and Def. X. 5. 2, can be identified under the Szegö intertwining operator

$$\text{S: } \text{Ind}\_{\mu} \mathsf{L}^{\perp} (\mathsf{E}\_{\sigma \otimes} \mathsf{pr}\_{\otimes 1}) = \mathsf{E}\_{\gamma \mathbf{v}} \to \mathsf{H} \mathsf{q} \mathsf{(G/L, E\_{\gamma \mathbf{v}})},\tag{X.5.6}$$

where

120 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

generalized orbits of the complex homogeneous bundle

complexes of fibered holomorphic bundles of Frèchet is having that:

ct(u, I()) = H

where E, is a Fréchet space.

**Proposition X. 5.1.** (Bulnes, F.) [37, 38].

dimension under character of Vogan).

infinite dimension [8]:

H

**Proof.** [38].

Dolbeault cohomology.

**Def. X. 5.2.** A generalized open G-orbit is the extension of G-module (open L-module on a

**Def. IV.3.** (Bulnes, F.) [37, 38, 46-48]. A generalized G-module is the induced G-module by a differentiable cohomology of representations of infinite dimension (E,η), defined on

Using u-cohomology, continuous cohomology and the generalization of the topology on

Of these generalities in hand, we get immediately a description of the topological dual of

Using the conjecture of Vogan on the possible extension of representations of L to modules of Harish-Chandra to G taking care that the co-border operators of the Dolbeault cohomology have all closed range, we apply this conjecture on the extension of the induced G-modules proposed to modules of Harish-Chandra of infinite dimension, obtaining the conjecture:

**Conjecture X. 5.1.** (Bulnes, F.) [38]. Suppose that Hqct(g, LK; ALE\*) is a (l, LK)-module of finite longitude, E,their corresponding representation of L, and, the make holomorphic associated bundle to G/L.Then the co-border operators of the Dolbeault cohomology are all of closed range and to the case of infinite dimension the range of the co-border operators of the Dolbeault cohomologyare closed provided certain intertwining operators 39, applied to the corresponding induced *G*-modules are modules of Harish-Chandra (which must satisfy the theorem of Vogan-Zuckerman on irreducible unitary representations of infinite

Using certain technical lemmas [10], to a decomposition of the algebra u in their parts extensive y classified of the radical nilpotent part of the complex holomorphic vector bundle module a radical nilpotent bundle of Borel subalgebras [8, 12], and with pertinent generalizations of the G-modules, is have a theorem of classification of representations of

G <sup>L</sup> I (G(w)) Ind G(w)

[11], (X. 5.2)

<sup>η</sup> E G/L [8] (X. 5.3)

ct(G/L, Oq(E)), (X. 5. 4)

differentiable cohomology) induced in the differentiable category given by the space

$$\text{H}^{\text{g}}(\text{G/L}, \text{E}\_{\text{v}}) \bigtimes \sum \text{Indv} \, \text{c} (\text{E}\_{\text{v}} \sim \otimes (\wedge\_{\text{d}} \text{u})^{\ast})\rangle\tag{X.5.7}$$

and IndLG(E (q+1u)\*)) = E,where

$$\mathbf{E}\_{\eta} \to \mathbf{H}^{q}(\mathbf{G}(\mathbf{w}), \mathbf{O}(\mathbf{E}\_{\eta})),\tag{X.5.8}$$

Not there is to lose of see that it is wants to carry on the classical representations Ind L(E<sup>L</sup>1), with discrete series , built up by the Vogan's algorithm of minimal K-types of Aq() to the canonical temperate representations IndLG(E(q+1u)\*)), where the restriction fiber of Ehave Dolbeault operator L(w) . In particular and using the generalization of the Borel-Bott-Weil theorem (to L, locally compact),

$$\mathbf{H}^{\mathbb{I}\mathbb{Q}}(\mathbb{K}/\mathbb{K}\cap\mathcal{L}), \mathbf{C}\_{\mathbb{P}^{(\mathbb{I})}}\text{.}\mathbb{C}\_{\mathbb{P}^{(\mathbb{I})}}\text{.}\mathbb{K}\text{.}\mathbb{Q}(\mathbb{K}\cap\mathcal{L}), \mathbf{C}\_{\mathbb{Q}(\mathbb{I})^{+}}\text{.}\mathbb{Q}\text{.}\mathbb{R}^{\mathbb{Q}(\mathbb{I})^{+}}\text{.}\text{.}\tag{\text{X.5.9}}$$

We can to use these representations like complexes of Dolbeault to a globalization of Wong of type C(L(w), E (\*LL(w))\*).If we inducing certain representations K(w) K/(K L), of G(w), to L(w) G/L, of G(w). But this is possible due to the construction of the generalized G-modules. Then , on whose complexes is close.

Then using likeu = dimC(K L)(x), with u = t + s, with t, and s, complex dimensions of (K M)(x), and (K L)(w), and the fact of that

$$\mathbf{H}^q(\mathbf{M}(\mathbf{x}), \mathfrak{W}\circ(\mathbf{E}\_{\uparrow\uparrow\sim})) = 0,\tag{\text{X.5.10}}$$

p t, and

$$\mathbf{H}^{\emptyset}(\mathcal{L}\_{\{\mathbf{x}\}}, \mathfrak{o}(\mathcal{E}\_{\gamma\mathcal{V}})) = 0,\tag{\mathcal{X}.5.11}$$

 p u, we have the spectral succession of Leray of L(x) L(w), collapse to E2. Then the vanishing of the group of cohomology Hq(L(x), ob(E)) = 0, uestablish that

$$\mathbf{H}^{\ast\ast}(\mathcal{L}(\mathbf{x}), \mathfrak{o}(\mathcal{E}\_{\gamma\ast})) = \mathbf{H}^{\ast}(\mathcal{L}(\mathbf{x}), \mathfrak{o}(\mathbf{H}^{\ast}(\mathcal{M}(\mathbf{x}), \mathfrak{o}\cdot\mathfrak{e}(\mathcal{E}\_{\gamma\ast}))),\tag{X.5.12}$$

and due to the development of the relative lemmas of Frèchet spaces it is follows that (X. 5. 12) have structure of Frèchet space to the which the action of G, is a continuous representation (generalized G-module).

Cohomologically should be induced with finer decompositions of an nilpotent algebra of u, obtaining spaces more classified "thin"; the corresponding admissible G-modules of Frèchet. The obtained theorem can classify a great part of representations of infinite dimension although not their entirety due to the difficulty of obtaining a substantial algebra whose Dolbeault cohomology has an operator of closed range on the admissible G-modules that appear in the problem of representation of those (l, LK)-modules for an algebra q.

However, choosing an appropriate infinitesimal G-character, we could establish functors of Zuckerman corresponding to a algebra pu, with u = g/t, and canonical globalizations Xg, with g = , g = , g = , or g = ; having the property of closed range.

The election of an appropriate infinitesimal G-character causes redundancies in the admissible G - modules that turn out to be unitary representations, for what is important to choose a decomposition of u in the group of Levi.

**Appendix A** 

## **APPENDIX A: Integral Formulas on Canonical p-Pairs**

Be G, a real reductive group. We fix , a involution of Cartan G = NAK, and F, the subset of 0, with 0 = (F0)a, where F0 = a = 0, and is the system of positive simple roots of (g, a). Be (PF, AF), the canonical p-pair. If (aF)\*, and if HaF, then a= exp(H), if a = exp H. We define to F(aF)\*, by

$$
\mathfrak{\wp}(\mathbb{H}) = (1/2)\text{tr}(\text{ad }\mathbb{H}|\_{\mathbb{H}}).\tag{A.1}
$$

**Lemma. A. 1.** Let dn, da, dm, be invariant measures on NF, AF, 0MF. Let dk, be a normalized invariant measure on K. Then we can to elect an invariant measure dg, on G, such that

$$\text{l.f.f(g) dg} = \left| \_{\text{NF}\_{\times} \text{Ar}\_{\times}} 0 \text{\_{\times} \text{F} \{ \text{n} \text{m} \text{k} \}} \text{a}^{\text{-} 2 \text{pr} } \text{dn da dm dk} \right. \tag{A.2}$$

to fCc(G). Also if uC(K), then

122 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

G-modules. Then , on whose complexes is close.

(K M)(x), and (K L)(w), and the fact of that

Hs+t(L(x), ob(E)) = Hs

representation (generalized G-module).

choose a decomposition of u in the group of Levi.

p t, and

We can to use these representations like complexes of Dolbeault to a globalization of Wong of type C(L(w), E (\*LL(w))\*).If we inducing certain representations K(w) K/(K L), of G(w), to L(w) G/L, of G(w). But this is possible due to the construction of the generalized

Then using likeu = dimC(K L)(x), with u = t + s, with t, and s, complex dimensions of

Hq(M(x), ob(E)) = 0, (X. 5.10)

 Hq(L(x), ob(E)) = 0, (X. 5.11) p u, we have the spectral succession of Leray of L(x) L(w), collapse to E2. Then the

(L(x), oq(Ht

and due to the development of the relative lemmas of Frèchet spaces it is follows that (X. 5. 12) have structure of Frèchet space to the which the action of G, is a continuous

Cohomologically should be induced with finer decompositions of an nilpotent algebra of u, obtaining spaces more classified "thin"; the corresponding admissible G-modules of Frèchet. The obtained theorem can classify a great part of representations of infinite dimension although not their entirety due to the difficulty of obtaining a substantial algebra whose Dolbeault cohomology has an operator of closed range on the admissible G-modules that

However, choosing an appropriate infinitesimal G-character, we could establish functors of Zuckerman corresponding to a algebra pu, with u = g/t, and canonical globalizations Xg,

The election of an appropriate infinitesimal G-character causes redundancies in the admissible G - modules that turn out to be unitary representations, for what is important to

(M(x), ob(E))), (X. 5.12)

vanishing of the group of cohomology Hq(L(x), ob(E)) = 0, uestablish that

appear in the problem of representation of those (l, LK)-modules for an algebra q.

with g = , g = , g = , or g = ; having the property of closed range.

$$\mathbb{L}\left\|\mathbf{x}\,\mathbf{u}(\mathbf{k})\,\mathrm{d}\mathbf{k} = \mathbb{I}\_{\mathbb{K}\times\mathbb{R}^{\mathsf{T}}}\mathbf{u}(\mathbf{k}\,\mathrm{rk}(\mathbf{kg})) \mathbf{a}(\mathbf{kg})^{\mathcal{D}\mathbb{P}}\,\mathrm{d}\mathbf{k} \,\mathrm{r}\,\mathrm{d}\mathbf{k},\tag{A.7}$$

where if gG, and if g = nak, nN, aA, and kK, then a(g) = a, and k(g) = k.

*Proof*. Let dp, be a left invariant measure on PF. Then we can elect an invariant measure, dg, on G, such that

$$\int \mathrm{cf}(\mathbf{g}) \, \mathrm{dg} = \int\_{\mathrm{PF}\_{\lambda} \times \mathbf{f}} \mathrm{f}(\mathbf{pk}) \, \mathrm{d}\mathbf{p} \, \mathrm{d}\mathbf{k}, \tag{A.4}$$

By the *lemma. I. 2.2.* For which can be demonstrated that save a scalar multiple dp = a2<sup>F</sup> dndadm. The lemma on suprajective diffeomorphisms9 implies that dp = h(n, a, m)dndadm, with h, a smooth function on NF AF 0M. By left invariance h, is independient of n.

By definition of 0MF, the modular function, , of PF, is 1, on 0MF. Thus dp, is right invariant under 0MF. Of where h, is a function of only one of the components.

The Jacobian of the action n ∣ ana-1 is det(Ad(a)n) = a2F, to aAF. Thus a-2Fdndadm, is left AFinvariant.

<sup>(2).</sup> The map 0MF AF NF PF,given for m, a, n I man, is a suprajective diffeomorphism.

<sup>9</sup>**Lemma.** (1). The map MF NF PF,given by m, n I mn, is a suprajective diffeomorphism.

Now we demostrate the second afirmation of the lemma. Remembering the lemma. I. 2.1., exist a continuous function compactly supported f, on G, such that

$$\text{\upharpoonrightf(pk) d}\text{p} = \text{\upharpoonrightu(kọk) dkẹ}, \text{ \upharpoonright k} \in \text{K} \tag{A.5}$$

Thus have

$$\int \mathbf{c} \, \mathbf{f}(\mathbf{x}) \, d\mathbf{x} = \int \mathbf{\kappa} \, \mathbf{u}(\mathbf{k}) \, d\mathbf{k},\tag{A.6}$$

Now,

$$\int \mathrm{cf}(\mathbf{x}) \, d\mathbf{x} = \int \mathrm{cf}(\mathbf{x}\mathbf{g}) \, d\mathbf{x} = \int\_{\mathbb{P}^{\mathbb{F}} \times \mathbb{X}} \mathrm{f}(\mathbf{p}\mathbf{k}\mathbf{g}) \, d\mathbf{p} \, d\mathbf{k},\tag{A.7}$$

Writing kg = na(kg)k(kg), like above, dp is transformed for , under right multiplication by elements of PF. Since

$$\delta(\text{na(kg)}) = \text{a(kg)}^{2\text{P}\_{\text{F}}}$$

we have

$$\text{J} \,\text{\textbullet } \text{u(k)} \,\text{dk} = \text{J}\_{\text{Fr}\_{\times} \times \text{a(kg)}} \,\text{29r (pk(kg))} \,\text{dp} \,\text{dk} = \,\text{J}\_{\text{Kr}\_{\times} \times \text{u(k} \text{rk(kg))}} \,\text{dk} \,\text{zd}, \,\text{k}$$

that is the wanted result.

To the following formula of integration we asume that G, is of inner type. Let R, be a system of positive roots to (g, a), corresponding to the election of n. Be a+, equal to the corresponding Weyl camera to R10.

Be A+ = exp(a+). If aA, a = exp H, we do corresponding

$$\gamma(\mathbf{a}) = \prod\_{aa} \text{\_{\mathbb{R}}\text{sinch}(\mathbf{a}(\mathbf{H}))}.\tag{A.8}$$

**Lemma. A. 2.** dg, can be normalized such that

G f(g) dg = <sup>K</sup> A+ <sup>K</sup>(a)f(k1ak2) dk1da dk2. (A. 9)

<sup>10</sup> Be P. Be Xg such that <X, X> = 1. Then X, X = H. Here Ha, is defined by B(H, H) = (H) Ha. Thus if x = (2/(H))X, y = X, h = (2/(H))H, then x, y, h generate a TDS (three Dimensional Simple) Lie algebra on **IR**. Thus exist a homomorphism de Lie, , of SL(2, **IR**), in G0, such that (g\*) = ((g))-1. Be k, the image of 0 1 -1 0, under . Then, if s, is defined by sH = H B(h, H)H, to Ha, then Ad(k)H = sH. Be N(a) = uK0Ad(u)a = a.

Be W(g, a) = Ad(u)auN(a). Then sW(g, a), to all P. This it is follows of the realized observations with before into of this foot of page.

Be a', the set of all the Ha, such that (H), is not null to all . A connect component of a', is called a Weyl camera of a. If C, is a Weyl camera then the set of all the , such that , is positive on C, denoted by the space PC, is called a positive roots systems. If P, and if , can not be write like a sum of two elements of P, then , is called simple in P. One interesting proposition to respect is:

**Proposition.** (1) W(g, a) is generate by the s,to , simple in P.

<sup>(2)</sup> W(g, a) act simple and transitive on the Weyl camera of a.

*Proof*. Let

124 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Thus have

elements of PF. Since

that is the wanted result.

corresponding Weyl camera to R10.

Be A+ = exp(a+). If aA, a = exp H, we do corresponding

**Lemma. A. 2.** dg, can be normalized such that

Now,

we have

10 Be P. Be Xg

into of this foot of page.

One interesting proposition to respect is:

**Proposition.** (1) W(g, a) is generate by the s,to , simple in P. (2) W(g, a) act simple and transitive on the Weyl camera of a.

exist a continuous function compactly supported f, on G, such that

Now we demostrate the second afirmation of the lemma. Remembering the lemma. I. 2.1.,

Writing kg = na(kg)k(kg), like above, dp is transformed for , under right multiplication by

(na(kg)) = a(kg) 2F,

K u(k) dk = PF K a(kg) 2F f(pk(kg)) dp dk = KF K u(kFk(kg))dkF dk,

To the following formula of integration we asume that G, is of inner type. Let R, be a system of positive roots to (g, a), corresponding to the election of n. Be a+, equal to the

(a) = Rsinh((H)). (A. 8)

x = (2/(H))X, y = X, h = (2/(H))H, then x, y, h generate a TDS (three Dimensional Simple) Lie algebra on **IR**. Thus exist a homomorphism de Lie, , of SL(2, **IR**), in G0, such that (g\*) = ((g))-1. Be k, the image of 0 1 -1 0, under .

Be W(g, a) = Ad(u)auN(a). Then sW(g, a), to all P. This it is follows of the realized observations with before

Be a', the set of all the Ha, such that (H), is not null to all . A connect component of a', is called a Weyl camera of a. If C, is a Weyl camera then the set of all the , such that , is positive on C, denoted by the space PC, is called a positive roots systems. If P, and if , can not be write like a sum of two elements of P, then , is called simple in P.

Then, if s, is defined by sH = H B(h, H)H, to Ha, then Ad(k)H = sH. Be N(a) = uK0Ad(u)a = a.

PF f(pk) dp = KF u(kFk) dkF, kK (A. 5)

G f(x) dx = G f(xg) dx = PF K f(pkg) dp dk, (A. 7)

G f(g) dg = <sup>K</sup> A+ <sup>K</sup>(a)f(k1ak2) dk1da dk2. (A. 9)

such that <X, X> = 1. Then X, X = H. Here Ha, is defined by B(H, H) = (H) Ha. Thus if

G f(x) dx = K u(k) dk, (A. 6)

$$
\mathfrak{h} : \mathfrak{a}^\* \times \mathbb{K} / \mathbb{M}^0 \to \mathfrak{p}, \tag{A.10}
$$

defined by (H, kM0) = Ad(k)H. Be p', the rank of . Since Ad(K)a = p, Ad(K)a+ = Ad(K)a', and Ad(K)(a a'), is a finite union of submanifolds of low dimension, p', is open, dense and have a complement of measure 0, in p. Is easy demonstrate that , is a diffeomorphism in p'(see proposition (1), foot of page 2).

Let Pj, be a system of positive roots for (gC, (hj)C). Set j(H) = P(H), for Hhj. Let D, is a non-zero polynomial function on g. Then D(H) = j(H) 2. Since G, and each Hj, are unimodular, each coset space G/Hj, has a G-invariant measure dxj.

Proposition. A. 1. There exist positive constants cj, j = 1, , r, and normalizations of Lebesgue measure on g, and the hj, such that

$$\int\_{\mathbf{0}} \mathbf{f}(\mathbf{X})d\mathbf{X} = \sum\_{\mathbf{j}\_{\parallel}} \mathbf{c}\_{\mathbf{j}} \Big[ \left| \mathbf{D}(\mathbf{H}) \right| \left( \int\_{\mathbf{G}/\mathcal{H}\_{\mathbf{j}}} \mathbf{f}(\mathbf{A}d \, \mathbf{x}\mathbf{H})d\mathbf{x}\_{\mathbf{j}} \right) \text{d}\mathbf{H}\_{\mathbf{j}} \tag{\text{A.11}}$$

for fCc(g).

*Proof*. For the moment, fix j, and let hj = h, etc. Let G/H h' g, be defined by (gH, h) = Ad(g)h (here h' = g' h). We may identity the complex tangent space at 1H, to G/H, with n+ + n . Translating by the elements of G, allows us to identity the tangent space at gH, with this space. A direct calculation yields

$$\mathsf{d}\mu\_{\mathsf{H}^{\mathsf{H}},\mathsf{h}}(\mathsf{X},\mathsf{Z}) = \mathsf{Ad}(\mathsf{g})(\mathsf{ad}\,\mathsf{X}\mathsf{h} + \mathsf{Z}),\tag{\mathsf{A}.\,12}$$

for X n+ + n , Zh. This implies

i. The Jacobian of , at gH, h, is D(h), up to sign.

This implies that , is everywhere regular. The remarks preceding the statement we are proving now imply that , is a [W]-fold covering of their range. Lemma11 implies that g', is the disjoint union of the open subsets Ad(G)(hj)C. The result now follows from (i).

The above result is sometimes called the Weyl integral formula for g.

Now we derive the Weyl integral formula for G. We define real analytic functions dj, on G by

$$\det(\mathbf{t}\mathbf{I} - (\mathbf{A}\mathbf{d}(\mathbf{g}) - \mathbf{I})) = \Sigma\mathbf{t}^{\flat}\mathbf{d}(\mathbf{g})\_{\prime\prime}$$

<sup>11</sup> **Lemma.** (1) If Xg', then X, is semi-simple and Cg(X) = {Yg [X, Y] = 0}, is a Cartan subalgebra of g.

<sup>(2)</sup> If X, is a semi-simple element of g, then Cg(X), is a reductive subalgebra of g, that contains a Cartan subalgebra.

Here *n* = dimG. Set d = dj, for j = rank(gC). We set G' = {gGd(g) 0}. Then G', is open, dense with complement of measure 0, in G.

Proposition A. 2. There exist positive constants mj, so that if dg and dhj, are respectively invariant measure on G, and Hj, then

$$\int\_{\mathbf{G}} \mathbf{f}(\mathbf{g})d\mathbf{g} = \sum\_{\mathbf{H}\_j} \mathbf{m}\_j \int\_{\mathbf{H}\_j} \mathbf{d}(\mathbf{h}\_j) \left| \int\_{\mathbf{G}/\mathbf{H}\_j} \mathbf{f}(\mathbf{g}\mathbf{h}\_j \mathbf{g}^{\ast 1}) \mathbf{d}(\mathbf{g}\mathbf{H}\_j) \right| \mathbf{h} \mathbf{h}\_{j\prime} \qquad \forall \quad \mathbf{f} \in \mathbf{C}\_c(\mathbf{G})\_{\prime} \tag{A.13}$$

Proof. We fix j, and for the moment drop the index j. Let G/H H' G, be defined by (gH, h) = ghg-1(here H' = H G'). We have

$$\text{Adv}\_{\mathbb{H}^{\mathbb{H}}, \mathbb{h}}(\mathsf{X}, \mathsf{Z}) = (\text{Ad}(\mathsf{g})((\text{Ad}(\mathsf{h} \cdot \mathsf{l}) - \mathsf{I})\mathsf{X} + \mathsf{Z}))\_{\sigma(\mathsf{l}^{\|\mathsf{l}\|}, \mathsf{h})} \tag{A.14}$$

for Xn+ n , Zh. The rest of the proof is now almost identical to that of proposition A. 1, and the details can be done like a exercise.

A result that helps us to derive some integration formulas that are related to the Gelfand-Naimark decomposition is the following result.

We set VF = NF. Fix invariant measures dn, dm, da, dv, respectively on NF, 0MF, AF, and VF.

**Lemma A. 2.** The invariant measure dg, can be normalized so that

$$\int\_{\mathcal{G}} \mathbf{f}(\mathbf{g})d\mathbf{g} = \int\_{\mathbf{N}\_{\mathbb{F}} \times^{0} \mathbf{M}\_{\mathbb{F}} \times \mathbf{A}\_{\mathbb{F}} \times \mathbf{V}\_{\mathbb{F}}} \mathbf{a}^{\cdot 2\rho\_{\mathbb{F}}} \mathbf{f}(\text{nmaxv}) \text{dndmdadv},\tag{A.15}$$

for fCc(G). If uC(K), then

$$\int\_{\mathcal{K}} \mathbf{u}(\mathbf{k})d\mathbf{k} = \int\_{\mathcal{K}\_{\mathbb{R}} \times \mathcal{V}} \mathbf{a}^{\ast 2\rho\_{\mathbb{R}}} \mathbf{f}(\mathbf{n} \mathbf{n} \mathbf{a} \mathbf{v}) \mathbf{d}n \mathbf{d} \mathbf{n} \mathbf{d}d \mathbf{d} \mathbf{v}\_{\prime} \tag{A.16}$$

*Proof*. [4-6 ].

## **APPENDIX B: Regularity on a Lie Algebra**

One of the fundamental theorems that gives place to born of the ordinary D-modules is the theorem or regularity of Harish-Chandra. In this is established a discussion on involutive distributions require in a decomposition of Cartan of an open G-invariant of a reductive algebra g, that which permit to obtain germs of a sheaf of differential operators in a Fréchet space. This give place to construct a regularity theory of the (g, K)-modules through of differential operators (the classical D-modules) and their extensions on G-invariant folds of a differential manifold with subjacent reductive group G.

Let G, be a real reductive group, g0, their real Lie algebra and g = g0 **IR C**, their complexification, G0, the topological component of the unit of G, defined as the space

$$\mathbf{G}^{0} = \{ \mathbf{g} \in \mathbf{G} | \text{Ad}(\mathbf{g}) = \mathbf{I}, \forall \text{ Ad} \in \text{End}(\mathbf{G}) \}, \tag{\mathbb{B}. 1}$$

Given that G is reductive then Int(g), is the space of automorphisms

$$\text{Int}(\mathfrak{g}) = \langle \mathsf{g} \in \text{Aut}(\mathbf{G}) | \exp(\text{ad}\mathbf{X}) = \mathsf{g}\_{\prime} \,\,\forall \,\, \mathbf{X} \in \mathfrak{g} \rangle,\tag{\mathbb{B}.\,\, 2}$$

where Int(g) can indentifies like the space Ad(G0) <sup>R</sup> **C**.

If Xg, and gG, then Ad(g)X = gX. If , is an open G-invariant of g, and if fC(), then (g)f(X) = f(g1X), G, and Xg. Let be the open space or open of operator invariants

$$\mathbf{D}'(\Omega)^{\mathbb{C}} = \langle \mathbf{T} \in \mathbf{D}'(\Omega) \mid \mathbf{T}\pi(\mathbf{g}) = \mathbf{T}, \,\forall \,\mathbf{g} \in \mathbf{G} \rangle,\tag{\mathbb{B}.3}$$

Let D = Dr(X), be with r = dim h, with h g, and such that Xg', where

$$\mathfrak{g}' = \langle X \in \mathfrak{gl} \mid \mathrm{D}(X) \neq 0 \rangle,\tag{\text{B.4}}$$

We consider ' = g'. Be h1, , hs, a complete set of Cartan subalgebras non-conjugates of g. Be j' = G(' hj), (Is the corresponding open orbit of the group G corresponding to the subalgebra of Cartan hj). Then ' = j'. Indeed, be Hhj(j s), then for the G-invariance of ,

$$\text{Ad}(\mathbf{g})\mathbf{H} = \mathbf{g}\mathbf{H},\tag{\text{B.5}}$$

where gHG(' hj) = G[( g') hj] = j'. Then

$$
\Sigma\_{\restriction \leq^{\mathsf{L}}} \text{Ad}(\mathbf{G}) \mathbf{H} \mathfrak{f} = \Sigma\_{\restriction \leq^{\mathsf{L}}} \mathbf{H} \mathfrak{f} = \mathfrak{g}', \forall \ \mathbf{g} \in \mathbf{G}, \ \mathbf{H} \in \mathfrak{h}\_{\flat} \tag{\mathbb{B}}
$$

126 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

j j

**Lemma A. 2.** The invariant measure dg, can be normalized so that

G N MAV

K KV

0

G H G/H

dense with complement of measure 0, in G.

(gH, h) = ghg-1(here H' = H G'). We have

and the details can be done like a exercise.

for fCc(G). If uC(K), then

*Proof*. [4-6 ].

Naimark decomposition is the following result.

for Xn+ n

invariant measure on G, and Hj, then

Here *n* = dimG. Set d = dj, for j = rank(gC). We set G' = {gGd(g) 0}. Then G', is open,

Proposition A. 2. There exist positive constants mj, so that if dg and dhj, are respectively


Proof. We fix j, and for the moment drop the index j. Let G/H H' G, be defined by

dgH, h(X, Z) = (Ad(g)((Ad(h-1) I)X + Z))(gH, h), (A. 14)

A result that helps us to derive some integration formulas that are related to the Gelfand-

We set VF = NF. Fix invariant measures dn, dm, da, dv, respectively on NF, 0MF, AF, and VF.

F F FF

F

F

(A. 15)

(A. 16


( ) ,

f g dg a f(nmav)dndmdadv

F

u k dk a f(nmav)dndmdadv ( ) ,


(A. 13)

, Zh. The rest of the proof is now almost identical to that of proposition A. 1,

f g dg m d(h ) f(gh g )d(gH dh f C G ( ) ) , ( ), 

> © 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

which is translated in <sup>j</sup> s G(' hj) = <sup>j</sup><sup>s</sup> j' = '. If gG and if H' hj, then we define the map

$$\Psi \!\!\!\/ ; \mathbb{G} \times (\Omega' \cap \mathbb{H}) \to \!\!\/ \!\/ \_{\prime} \tag{\mathbb{B}} \!\!\/ \_{\prime} \tag{\mathbb{B}} \!\!\/ \_{\prime} \tag{\mathbb{B}} \!\!\/ \_{\prime} \tag{\mathbb{B}}$$

with rule of correspondence

$$\mathbf{H} \text{ (g\\_H)} \mapsto \mathbf{g} \mathbf{H},\tag{\text{B.8}}$$

This map is a submersion of G (' hj) in j'. That's right, by appendix A, if H' hj, with gV, and Yh, then

$$\text{Adv}(\mathcal{X}, \mathcal{Y}) = \text{Ad}(\text{g})([\mathcal{X}, \mathcal{H}] + \mathcal{Y}), \tag{\mathbb{B}. 9}$$

where dj, is suprajective gG, H'hj. Of where j, is submersion.

We fix the Lebesgue measures dX, and dH, on g, and each hj, respectively. We visual to S(g), as the algebra

$$\mathbf{S(g) = (D\chi \in \text{DC(g)}) \,\mathrm{D}\chi = \Sigma \mathrm{P}(\mathbf{X}) \hat{c}\prime\prime \,\,\forall \, \mathbf{X} \in \mathfrak{g}\prime\prime^{12},\tag{\text{B.10}}$$

and consider the differential ideal I(g) = S(g)G, this due to that in S(g)G, we can establish a formula of differential ideals with the semi-simple structure of g.

We fix Cartan subalgebra h of g. Be = (gC, hC). If , and if (h) IR (respectively (h) iIR), then we say that , is real or imaginary respectively.

Let IR, and I, be the systems of real and imaginary roots respectively. Let IR = IR, and <sup>I</sup> = iI, be and = IR + I.

Let

$$\mathfrak{h}^{\prime\prime} = \{ \mathbf{H} \in \mathfrak{h} | \alpha(\mathbf{H}) \neq 0, \,\forall \; \alpha \in \Gamma \},\tag{\mathbb{B}.11}$$

Clearly h'' h'. In effect, if Hh', such that

$$\mathfrak{h}' = \{ \mathbf{H} \in \mathfrak{g}' | \alpha(\mathbf{H}) \neq 0, \,\forall \; \alpha \in \Gamma \},\tag{\mathbb{B}.12}$$

then Hh''.

12 Sea X1, , Xn, a base of g, and be x1, , xn, the corresponding coordinates of g. If DDO(g), then

D = pI<sup>I</sup> ,

where has been used the notation of multi-indices ordinary. If I = (i1, , in), with in, a non-negative integer then I = ij, and I /xi11, , xinn. If Xg, then DX = pI(X)<sup>I</sup> . Then DX, is a constant coefficient differential operator on g. Clearly, T(I S(g**C**)) (to all homeomorphism of Lie algebras belonging to space Homg, K(l, DO(g)), where l, is the subalgebra of g corresponding to Lie group L = G g, is the algebra of all the constant coefficient differential operators on g. We can thus identify a S(g**C**), with the algebra of constant coefficient differential operators on g. Note: T, is extended to a homomorphism of algebras of U(l**C**), in DO(g).

**Lemma. B. 1.** Let C, be a connect component of h''. Then exist 1, ,q, such that:

1. 1, ,q, are linearly independents, C = Hhj(H) > 0, j = 1, 2, , q.

As consequence C h', is connecting.

*Proof*. If , then *z*(g)since

$$\mathsf{Z}(\mathfrak{g}) = \langle \mathsf{X} \in \mathfrak{gl} | \mathsf{\alpha}(\mathsf{X}) = \mathsf{0} \rangle,\tag{\mathsf{B}.13}$$

Thus we can assume unloosed of generality that g, is semi-simple. Let

$$\mathfrak{h}\_{\mathsf{i}\mathsf{R}} = \{ \mathsf{H} \in \mathsf{loc} | \mathsf{ca}(\mathsf{H}) \in \mathsf{R}, \,\forall \,\,\alpha \in \mathsf{0}^{+} \},\tag{\mathsf{B.14}}$$

Then h = hIR ihIR = (hIR h) (ihIR h). If IR, (or I, respectively) then (hIR h) (or (ihIR h) = 0).

Consider the components spaces

$$(\mathfrak{h}\_{\mathbb{R}} \cap \mathfrak{h})' = (\mathcal{H} \in (\mathfrak{h}\_{\mathbb{R}} \cap \mathfrak{h}) \mid \alpha(\mathcal{H}) \neq 0, \,\forall \,\,\alpha \in \Gamma\_{\mathbb{R}}),\tag{\mathbb{B}.15}$$

and

128 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

the map

with rule of correspondence

with gV, and Yh, then

as the algebra

iI, be and = IR + I.

Let

then Hh''.

and I

which is translated in <sup>j</sup> s G(' hj) = <sup>j</sup><sup>s</sup> j' = '. If gG and if H' hj, then we define

j : G (' hj) j', (B. 7)

This map is a submersion of G (' hj) in j'. That's right, by appendix A, if H' hj,

dj(X, Y) = Ad(g)(X, H + Y), (B. 9)

We fix the Lebesgue measures dX, and dH, on g, and each hj, respectively. We visual to S(g),

and consider the differential ideal I(g) = S(g)G, this due to that in S(g)G, we can establish a

We fix Cartan subalgebra h of g. Be = (gC, hC). If , and if (h) IR (respectively (h)

Let IR, and I, be the systems of real and imaginary roots respectively. Let IR = IR, and <sup>I</sup> =

h'' = Hh(H) 0, , (B. 11)

h' = Hg'(H) 0, , (B. 12)

D = pI<sup>I</sup> ,

. Then DX, is a constant coefficient differential operator on g. Clearly, T(I

where has been used the notation of multi-indices ordinary. If I = (i1, , in), with in, a non-negative integer then I = ij,

S(g**C**)) (to all homeomorphism of Lie algebras belonging to space Homg, K(l, DO(g)), where l, is the subalgebra of g corresponding to Lie group L = G g, is the algebra of all the constant coefficient differential operators on g. We can

12 Sea X1, , Xn, a base of g, and be x1, , xn, the corresponding coordinates of g. If DDO(g), then

thus identify a S(g**C**), with the algebra of constant coefficient differential operators on g.

Note: T, is extended to a homomorphism of algebras of U(l**C**), in DO(g).

where dj, is suprajective gG, H'hj. Of where j, is submersion.

S(g) = DXDO(g) DX = PI(X)<sup>I</sup>

iIR), then we say that , is real or imaginary respectively.

Clearly h'' h'. In effect, if Hh', such that

/xi11, , xinn. If Xg, then DX = pI(X)<sup>I</sup>

formula of differential ideals with the semi-simple structure of g.

(g, H) I gH, (B. 8)

, Xg12, (B. 10)

$$(\mathfrak{ih}\_{\mathbb{R}} \cap \mathfrak{h})' = (\mathcal{H} \in (\mathfrak{ih}\_{\mathbb{R}} \cap \mathfrak{h}) \mid \mathfrak{a}(\mathcal{H}) \neq 0, \,\,\forall \,\, \mathfrak{a} \in \Gamma \mathfrak{i}),\tag{\mathbb{B}.16}$$

Then a connect component of h'', is the space of Weyl cameras C1 C2, with C1(hIR h)', and C2(ihIR h)', connect. Given that, IR, and I, are both systems of roots and h, a Cartan subalgebra of g, then = IR I, is a system of roots corresponding to the Weyl cameras of C h'. Then 1, ,q, result linearly independents. Thus if we consider

$$
\Sigma = \Phi - (\Gamma\_{\parallel \mathbb{R}} \cap \Gamma),
\tag{\mathbb{B}.17}
$$

and if , then *Re*, and *Im*, are linearly independents. Therefore the root spaces of h, relative to , is

$$(\mathfrak{h})\_a = (\mathsf{H} \in \mathfrak{h} | \alpha(\mathsf{X}) = 0),\tag{\mathsf{B}.18}$$

is such that *codim*(hj) = 2 in hj. Thus in C (hj)', (hj), have codimension

$$
\operatorname{codim}(\mathfrak{h})\_a = \dim \mathbf{C} - \dim \cup\_{\operatorname{ac}\Sigma} (\mathfrak{h})\_{\mathfrak{w}} \tag{\mathbb{B}.19}
$$

Of which the space C (hj)' = C (hj), is connect.

Consider to Ad(G), such that

$$\text{Ad}(\mathbf{G})\mathbf{Z}(\mathfrak{g}) = \mathbf{Z}(\mathfrak{g}),\tag{\text{B.20}}$$

(Act trivially on the center of the Lie algebra of Lie g).

Let 1, ,d, be homogeneous Ad(G)-invariants polynomials on [g, g], (that is to say, j = Xj Yj Yj Xj) (j = 1, , d), such that

$$[\operatorname{Ad}(\mathbf{g})\mathbf{X}\_{\boldsymbol{\nu}}\operatorname{Ad}(\mathbf{g})\mathbf{Y}\_{\boldsymbol{\nu}}] = [\mathbf{X}\_{\boldsymbol{\nu}}\ \mathbf{Y}\_{\boldsymbol{\nu}}]\_{\boldsymbol{\nu}}\tag{\mathbb{B}.21}$$

Then r > 0,

$$\Omega(\phi\_1, \dots, \phi\_\ell, \mathbf{r}) = \langle \mathbf{X} \in [\mathfrak{g}, \mathfrak{g}] \mid |\phi(\mathbf{X})| \le \mathbf{r}, \mathbf{j} = \mathbf{1}, \dots, \mathbf{d} \rangle,\tag{\mathbf{B}.22}$$

Let U, be an open and connect subset of Z(g) = Z. Let

$$\Omega = \langle X + \mathbf{Y} \vert X \in \mathbf{U}, \text{ and } \mathbf{Y} \in \Omega(\phi\_1, \dots, \phi\_{\sf t}, \mathbf{r}) \rangle,\tag{\mathbb{B}. 23}$$

**Lemma. B. 2**. is connect. More yet, if h, is a Cartan subalgebra of g, and if C is a connect component of h', then C , is connect.

*Proof*. If XU, and Y(1, ,d, r), then X + tY, to 0 t 1. Then implies by connectivity that , is connect. To demonstrate the second affirmation is sufficient consider to g = [g, g], due to the definition of the space (1, ,d, r). Indeed, considering to B a connect neighborhood of the 0, such that Be C , and C, a connect component of h'', where

$$\mathfrak{h}' = \{ \mathbf{H} \in \mathfrak{h} \mid \alpha(\mathbf{H}) \neq 0, \,\forall \; \alpha \in \Gamma \},$$

where IR = IR, and I = I, and = IR I. By the lemma I. 1, C h', is connect. Since h'' h', then C like component of h'', satisfies that

$$\mathbf{C} \cap \mathfrak{h}' = \mathbf{C} \mid \mathfrak{h}' \tag{\text{B.24}}$$

Since t[0, 1], and X, YC, tX (1 t)YC h' = C \ h'' C, of where tX (1 t)YC, t[0, 1]. Then B C is connect. Then, if XC, then exist t > 0 such that tXB C. Thus C , is connect (also B C h).

**Theorem. B. 1.** Be = X + Y XU, Y(1, ,d, r). Let TD'('), be such that dim I(gC)T < , on '. Then exist an analytic function FT = F, on ', such that


*Proof*. Assume h = hj, then Xh, and pS(g)G I(g)

Appendix B – Regularity on a Lie Algebra 131

$$
\hbar \psi^0(\mathbf{p} \mathbf{T}) = |\mathbf{D}|^{-1/2} \overline{\mathbf{p}} |\mathbf{D}|^{-1/2} \psi^0(\mathbf{T}) \tag{\text{B.25}}
$$

Thus *dim* I(g)(D -1/2<sup>j</sup> 0(T)) < . But S(hC), is finitely generated like a I(g)-module. Then the lemma13 implies that exist a function j, on ' hj, whose restriction to each connect component is an exponential polynomial and is such that

$$
\hbar\eta^0(\mathbf{T}) = |\mathbf{D}|^{-1/2}\mathbf{T}\_{\parallel\parallel\hbar} \tag{\mathbf{B}.26}
$$

Given that

130 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Let 1, ,d, be homogeneous Ad(G)-invariants polynomials on [g, g], (that is to say, j = Xj

[Ad(g)Xj, Ad(g)Yj] = [Xj, Yj], (B. 21)

**Lemma. B. 2**. is connect. More yet, if h, is a Cartan subalgebra of g, and if C is a connect

*Proof*. If XU, and Y(1, ,d, r), then X + tY, to 0 t 1. Then implies by connectivity that , is connect. To demonstrate the second affirmation is sufficient consider to g = [g, g], due to the definition of the space (1, ,d, r). Indeed, considering to B a connect neighborhood of the 0, such that Be C , and C, a connect component of h'',

h'' = Hh (H) 0, , where IR = IR, and I = I, and = IR I. By the lemma I. 1, C h', is connect. Since h''

C h' = C \ h', (B. 24)

Since t[0, 1], and X, YC, tX (1 t)YC h' = C \ h'' C, of where tX (1 t)YC, t[0, 1]. Then B C is connect. Then, if XC, then exist t > 0 such that tXB C. Thus

**Theorem. B. 1.** Be = X + Y XU, Y(1, ,d, r). Let TD'('), be such that dim I(gC)T

 If h, is a subalgebra of Cartan of g, then exist an analytic function , on h'', which is an exponential polynomial on each connect component of h'', such that Fh' = D

more yet, if we extend F, to all , doing correspond F = 0, on ', then F, is locally


< , on '. Then exist an analytic function FT = F, on ', such that

*Proof*. Assume h = hj, then Xh, and pS(g)G I(g)

(1, ,d, r) = X[g, g] j(X) < r, j = 1, , d, (B. 22)

= X + Y XU, and Y(1, ,d, r), (B. 23)

(Act trivially on the center of the Lie algebra of Lie g).

Let U, be an open and connect subset of Z(g) = Z. Let

component of h', then C , is connect.

h', then C like component of h'', satisfies that

C , is connect (also B C h).

T = TF, on ',

integrable on .

Yj Yj Xj) (j = 1, , d), such that

Then r > 0,

where

$$\mathbf{F}|\_{\Omega \cap \ ^\circ \mathbb{R}} \times \langle \mathbf{X} \rangle = |\mathbf{D}|^{-1/2} \mathfrak{P}(\mathbf{X}) = \mathbf{F}|\_{\Omega \cap \ ^\circ \mathbb{R}} \times \mathfrak{g}(\mathbf{g} \mathbf{H}) = |\mathbf{D}|^{-1/2} \mathfrak{P}(\mathbf{g} \mathbf{H}), \tag{\mathbf{B} \ \mathcal{D}}?$$

 X'j, with X = gH, Hh'j, then (X) = j(gH). But S(hC), is finitely generated like a I(g) module, thus

$$\mathbf{I}(\mathbf{g})\mathbb{A}(\lambda) = \underline{\mathbf{I}(\mathbf{g})}\mathbf{D}|\cdot|\,^{1}\mathbb{A}(\mathbf{g}\,\mathbf{H}),\tag{\text{B.28}}$$

 PI(g). Thus j(H) = (X), in ' h'j. Then if Fh' = D -1/2, then

$$\mathsf{u}\mathfrak{h}^{0}(\mathsf{T}) = |\mathsf{D}|^{\cdot 1/2}\mathsf{T}\_{\mathbb{B}} = \mathsf{u}\mathfrak{h}^{0}(\mathsf{T}\mathsf{F}) = |\mathsf{D}|^{\cdot 1/2}\mathsf{T}\_{\mathbb{B}^{\times}}$$

Given that Thus j(H) = (X), in ' h'j, then T = T'<sup>h</sup>'j= T' = TF,, of where T = TF, on '.

Note that if we extend , to , by 0, then , is locally bounded. Indeed, is sufficient only consider = 0, on ', of where

$$|\mathbf{D}|^{\cdot 1/2} \mathbf{g} |\mathbf{D}|^{1/2} \beta(\mathbf{H}) = 0\_{\prime \prime}$$

which is equivalent to have that

$$|\mathbf{D}|^{1/2}\mathbf{g}|\mathbf{D}|^{-1/2} = \beta\_{\mathbb{H}}(\mathbf{H})\_{\mathbb{H}}$$

Since *dim* I(g)(D -1/2j(T)) < , j, then exist , such that (= (0)):

$$\alpha |\mathbf{D}|^{1/2} \mathbf{g} |\mathbf{D}|^{-1/2} \ge \beta (\mathbf{H})\_{\nu}$$

with (0) = (X), Xj' j. Thus , is locally bounded. Then by the corollary14, D -1/2, is locally integrable.

We define the open set

$$\mathbb{B}^0 = (\mathcal{X} \in \Omega' \| \| \mathbb{B}(\mathcal{X}) | = \alpha(\mathcal{O}) > \mathcal{O}),\tag{\mathbb{B}.\mathcal{D}9}$$

<sup>13</sup> **Lemma:** Let U, be a connect open subset of IRn. Let TD'(U), be such that S(IRn)T, is of finite dimension. Then exist 1, , p(Rn)**C**\*, and F**C**[x1, , xn, e, , ep] such that T = TF, on U. We have used a Lebesgue measure to IRn.

<sup>14</sup> **Corollary:** D 1/2, is locally integrable on g.

that is to say, let be a neighborhood of the 0, such that B0 = hj'. Since F', is locally integrable then FB0, is locally integrable j. Then realizing an extension of j, by 0 in B0, on , and considering that F = D 1/2, then F, is locally integrable in .

The results that now we will give are extensions of the fundamental theorem of Harish-Chandra.

Let X1, , Xn, be a base of g, and we define Xj , for

$$\mathcal{B}(\mathsf{X}\downarrow\mathsf{X}) = \mathsf{S}\_{\mathsf{i}\mathsf{j}\star}$$

Let = XiXj . Then I(g), ( is a G-invariant differential operator). By definition

$$\mathbf{I(g)} = \langle \mathbf{D} \in \mathbb{S}(\mathfrak{g})^{\mathbf{c}} \mid \mathbf{D} = \Sigma \mathbf{X} \mathbf{X} \boldsymbol{\omega} \; \forall \; \langle \mathbf{X} \rangle \text{, a base of } \mathfrak{g} \rangle^{\mathbf{1} \mathbf{\tilde{s}}} \tag{\text{B. 30}}$$

since S(g)G = S(g) T(g), where T(g), is a tensor algebra in g. Then = Di i T(g)0, with T(g)0 T(g). Thus I(g).

**Theorem. B. 2.** Let , be (B. 23). Let TD'()G, be such that

$$\dim \mathbf{C}[\square] \mathbf{T} < \infty\_{\prime}$$

on . Let F = FT, be like the given by theorem B. 1. Then T = TF.

The demonstration of this result will take the reminder of the exposition. Before we have that to give some details of the demonstration, first we will develop some results on distributions on , that will be required in U *N*16.

Note that if is a G-invariant polynomial on g, then f(X) = f(Xs), Xg. Thus

$$
\Omega \cap (\mathcal{Z} \oplus \mathcal{N}) = \mathcal{U} \oplus \mathcal{N}\_\prime \tag{\mathbb{B}.31}
$$

Indeed, by a side we know that

$$\mathsf{Z}(\mathfrak{g}) = \langle \mathsf{X} \in \mathfrak{g} | \mathrm{Ad}(\mathfrak{g}) \mathsf{X} = \mathsf{X}, \forall \, \mathsf{g} \in \mathsf{G} \rangle,\tag{\mathsf{B}.32}$$

then *z N* = XgAd(g)X + adkX, kZ. Since U *z*(g), then (*z N* ) U *N*. For other side,

$$\mathbf{U} \oplus \mathcal{N} = \langle \mathbf{X} + \mathbf{Y} | \mathbf{X} \in \mathbf{U} \text{ and } \mathbf{Y} \in \mathcal{N} \rangle,\tag{\text{B.33}}$$

<sup>16</sup> *N* = Xg I+(g) = 0.

<sup>15</sup> This is a differential ideal and it is can extend or generalize to distributions on holomorphic vector bundles extending much of the results of classical differential ideals.

By u-cohomology and extending the concept of distribution on a complex holomorphic manifold subjacent in G/L, with L, a Levi subgroup in G, it is obtain a generalized -module.

Then [X, Y] = 0, Y*N*. But Ad(g)X = X, and Y = adkX, Xg, and kZ. Since furthermore [X, Y] = 0, Xg. For other side , is a convex space thus tR+, X + tY. Thus X + Y (*z N* ). Thus U *N* (*z N* ). Then the equality among the sets (*z N* ), and U *N*, is followed.

We consider that g = [g, g]. Let *N* = *O*1 *O*2 *O*r, be with *O*j = GXj, and *O*1, open in *N*, *O*2, open in *N O*1, etc. Let

$$\mathcal{N}\_{\mathsf{P}} = \bigcup\_{l} \mathsf{I}\_{l \geq \mathsf{P}} \, O\_{l} \, \tag{B. 34}$$

Then *N*p, is closed in g. Let H, X, and Y, be a canonical base to a TDS algebra17u, in g, ([H, X] = 2X, [H, Y] = 2Y and [X, Y] = H). Like u-module under ad, g, is the direct sum

$$\mathfrak{g} = \bigoplus\_{\mathbf{m}} \mathbf{V}^{\mathbf{m}},\tag{\mathbf{B}.35}$$

such that *dim* Vm = m + 1, mZ+. Then the eigenvalues , of the endomorphic equation

$$(\text{ad}\mathbf{h} - \lambda \text{I}\mathbf{v}\mathbf{m}) = \mathbf{0}, \quad \forall \text{ h} \in \mathbf{V}^{\mathbf{m}}, \tag{\text{B. } \mathbf{3}\mathbf{6}})$$

are such that Re= m 2k, with 0 k m. The proper mspace is g<sup>Y</sup> Vm, and

$$\mathbf{X}\mathbf{V}^{\mathrm{m}} = \boldsymbol{\Sigma}\mathfrak{g}^{\mathrm{Y}} \cap \mathbf{V}^{\mathrm{m}},\tag{\mathrm{B.37}}$$

and XVj XVi = , i j. Then

132 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

, and considering that F = D

Let X1, , Xn, be a base of g, and we define Xj

Chandra.

Let = XiXj

T(g). Thus I(g).

that is to say, let be a neighborhood of the 0, such that B0 = hj'. Since F', is locally integrable then FB0, is locally integrable j. Then realizing an extension of j, by 0 in B0, on

The results that now we will give are extensions of the fundamental theorem of Harish-

, for

) = ij,

i

(*z N* ) = U *N*, (B. 31)

T(g)0, with T(g)0

. Then I(g), ( is a G-invariant differential operator). By definition

I(g) = DS(g)G D = XiXi, Xj, a base of g15, (B. 30)

*dim* C[]T < ,

The demonstration of this result will take the reminder of the exposition. Before we have that to give some details of the demonstration, first we will develop some results on

 *z*(g) = XgAd(g)X = X, gG, (B. 32)

then *z N* = XgAd(g)X + adkX, kZ. Since U *z*(g), then (*z N* ) U

U *N* = X + YXU and Y*N*, (B. 33)

15 This is a differential ideal and it is can extend or generalize to distributions on holomorphic vector bundles

By u-cohomology and extending the concept of distribution on a complex holomorphic manifold subjacent in G/L,

B(Xi, Xj

since S(g)G = S(g) T(g), where T(g), is a tensor algebra in g. Then = Di

Note that if is a G-invariant polynomial on g, then f(X) = f(Xs), Xg. Thus

**Theorem. B. 2.** Let , be (B. 23). Let TD'()G, be such that

on . Let F = FT, be like the given by theorem B. 1. Then T = TF.

distributions on , that will be required in U *N*16.

extending much of the results of classical differential ideals.

with L, a Levi subgroup in G, it is obtain a generalized -module.

Indeed, by a side we know that

*N*. For other side,

<sup>16</sup> *N* = Xg I+(g) = 0.

1/2, then F, is locally integrable in .

$$\mathfrak{g} = \mathfrak{g}^{\vee} \oplus [\mathfrak{X}, \mathfrak{g}] \tag{\text{B. 38}}$$

We consider V = gY. If gG, and if ZV, then we define (g, Z) = g(X + Z), that is to say, the map defined by

$$
\Phi: \mathbb{G} \times \mathcal{V} \to \operatorname{Ad}(\Omega) \subset \Omega,\tag{\mathbb{B}.39}
$$

with rule of correspondence

$$(\mathbf{g}, \mathbf{Z}) \mid \to \operatorname{Ad}(\mathbf{g})(\mathbf{X} + \mathbf{Y}), \tag{\mathbb{B}.40}$$

Then the differential

$$\mathbf{d}\:\mathbb{N}\_{\mathbb{V}}\,\iota(\mathbf{g}\_{\mathbb{V}}\,\mathbf{V})=\mathbf{g}(\mathbf{V}+\{\mathbf{X},\,\mathbf{g}\})=\mathfrak{g}.\tag{\mathbb{B}.41}$$

Considering that V = gY, and considering the submersion

$$
\Psi \colon \mathbb{G} \times (\mathbb{V} \oplus [\mathbb{X}, \mathfrak{gl}]) \to \mathfrak{g} \tag{\mathbb{B} \text{ 42}}
$$

with rule of correspondence

<sup>17</sup> Three-Dimensional-Split algebra.

$$\vdash (\mathbf{g}\_{\prime} \; \mathrm{V} + \underline{\mathbf{Y}}) \mid \rightarrow \mathrm{Ad}(\mathbf{g})\mathbb{X}\_{\prime} \tag{\mathbb{B}.43}$$

we have in particular in a neighborhood of the 0V,

$$
\psi = \mathbf{d} \,\Phi\_{\mathbb{R}} \,\circ(\mathbf{g}\_{\mathsf{V}} \,\mathrm{V}),
\tag{\text{B.44}}
$$

thus Ad(g)X + Ad(g)adX (G (V [X, g])). Then there is a subspace

$$\mathbf{V}^{\circ} = \langle \mathbf{X} | \mathbf{X} + \mathbf{V}^{\circ} \subset \Omega \rangle,\tag{\text{B.45}}$$

with , the defined space explicitly as

$$
\Omega = \{ \mathbf{X} + \mathbf{Y} | \mathbf{X} \in \mathbf{U} \text{ (open)}, \text{ and } \mathbf{Y} \in \Omega(\phi 1, \dots, \phi 1, \mathbf{r}) \},
$$

In particular the images Ad(g)X + Ad(g)adX X + V . Therefore

$$
\Phi|\_{\mathbf{C}\times V^{\sim}} = \Psi(\mathbf{G}\times(\mathbf{V}\oplus[\mathbf{X},\mathbf{gl}])).
$$

Then <sup>G</sup><sup>V</sup> = dg, 0(g, V) = g(V + [X, g]). Note (G V ) *N*j, is open in *N*j. Indeed, , is suprajective in G V . Then (G V ), is open in . Thus j, (G V ) *N*j, is open in *N*j.

Let W18, be a G-invariant subset of g, such that W *N*j = *O*j. For definition of V , neighborhood of the 0, in V, we can re-define it on g, as

$$\mathbf{V}^{\sim} = \langle \mathbf{X} \in \mathbf{V} | \mathbf{d} / \mathbf{d} \mathbf{t} \, \exp(\mathbf{t} \mathbf{X}) |\_{\mathbf{t} = 0} = \mathbf{X}(\mathbf{0}) \rangle\_{\times}$$

But X(0)dg, 0(g, V) = g. Also, <sup>G</sup><sup>V</sup> = dg, 0(g, V), and given that (g, V )W, then Vj, is an open neighborhood of the 0, in V . Then G Vj, is open on G V . But , is a submersion in G V , therefore exist an open set W such that

$$
\Phi(\mathbf{G} \times \mathbf{V}\_{\parallel}) = \mathbf{W}\_{\prime}
$$

( j, j, is a submersion in G Vj), thus j, (G Vj) *N*j = *O*j. If X = 0, then we consider Vj = . Then

Be *O*j , and Xj*O*j. Be

$$\mathbf{V}\_{!} = \{ \mathbf{X} = \mathbf{X}\_{!} | \Phi(\mathbf{g}\_{!}, \mathbf{X}\_{!}) \in \mathbf{W}\_{!} \,\,\forall \,\, \mathbf{g} \in \mathbf{G} \}\_{\prime}.$$

Then exist a neighborhood Uj, of 0, in Vj, such that if we consider j(g, Z) = g(X + Z), gG, and XVj, then


iii. (Xj + Uj) *O*j = Xj.

<sup>18</sup> W = XggX = X, gG

Anyone neighborhood of the 0, in Vj satisfies (i), and (ii). For which we demonstrate in Vj, that is satisfied (iii). If Xj = 0, we have Vj = Uj. Then we assume that X = Xj 0. Let X, Y, H, be a base to a subalgebra TDS u, to X. Let

$$\mathbf{W}^{\mathrm{m}} = \bigoplus\_{\Lambda \le \, \mu} \mathbf{m} \, \mathbf{ad} \mathbf{V}^{\mathrm{m}} - \Lambda \mathbf{I} \mathrm{vm}\_{\star}$$

and let W = <sup>m</sup>NWm. Then adX, is a linear isomorphism of W in [X, g]. Indeed, since

$$\mathfrak{a} = \bigoplus\_{\mathbf{m}} \mathbf{V}^{\mathbf{m}\_{\mathbf{m}}}$$

then

134 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

we have in particular in a neighborhood of the 0V,

thus Ad(g)X + Ad(g)adX

suprajective in G V

G V

Vj = . Then

and XVj, then

ii. <sup>j</sup> *N*j = *O*j, iii. (Xj + Uj) *O*j = Xj.

<sup>18</sup> W = XggX = X, gG

open neighborhood of the 0, in V

Be *O*j , and Xj*O*j. Be

V

In particular the images Ad(g)X + Ad(g)adX

Then <sup>G</sup><sup>V</sup> = dg, 0(g, V) = g(V + [X, g]). Note (G V

neighborhood of the 0, in V, we can re-define it on g, as

, therefore exist an open set W such that

. Then (G V

V

i. j, is a submersion in an open neighborhood j, of X in ,

But X(0)dg, 0(g, V) = g. Also, <sup>G</sup><sup>V</sup> = dg, 0(g, V), and given that (g, V

with , the defined space explicitly as

(g, V + Y) ∣ Ad(g)X, (B. 43)

= dg, 0(g, V), (B. 44)

= XX + V , (B. 45)

) *N*j, is open in *N*j. Indeed, , is

) *N*j, is open in *N*j.

)W, then Vj, is an

. But , is a submersion in

,

(G (V [X, g])). Then there is a subspace

= X + YXU (open), and Y(1, ,1, r),

<sup>G</sup><sup>V</sup> = (G (V [X, g])).

Let W18, be a G-invariant subset of g, such that W *N*j = *O*j. For definition of V

= XVd/dt exp(tX)t = 0 = X(0),

(G Vj) = W, ( j, j, is a submersion in G Vj), thus j, (G Vj) *N*j = *O*j. If X = 0, then we consider

Vj = X = Xj(g, Xj)W, gG, Then exist a neighborhood Uj, of 0, in Vj, such that if we consider j(g, Z) = g(X + Z), gG,

. Then G Vj, is open on G V

X + V . Therefore

), is open in . Thus j, (G V

$$\begin{aligned} \text{[[X, \,\,g] = \bigoplus\_{\mathbf{m}, \,\mathbf{A} \,\, \mathbf{v} \,\, \mathbf{m}} \text{[X, } \mathbf{V}^{\mathbf{m}}] = [\mathbf{X}, \bigoplus\_{\mathbf{m}} \mathbf{V}^{\mathbf{m}}] = [\mathbf{X}, \,\mathbf{V}] = [\mathbf{X}, \,\mathbf{H} \mathbf{W}^{\mathbf{m}}] \\ &= [\mathbf{X}, \,\mathbf{g} \,\, \mathbf{r} \,\, \mathbf{J} \,\mathbf{w}^{\mathbf{m}}] = [\mathbf{X}, \,\mathbf{W}^{\mathbf{m}}]\_{\mathbf{m}, \mathbf{N}\_{\mathbf{a}} \times\_{\mathbf{a}} \mathbf{m}\_{\mathbf{r}}} \end{aligned}$$

But this equality is satisfied if g Wm, under the G-invariance on Wm, (that is to say, Ad(G), acting by 0 in Wm). Indeed, if we consider Ad(G)Wm = Wm, then the map

$$\emptyset \text{:}\ G \times \mathcal{W}^{\text{m}} \to \mathcal{U} \text{(open)},\tag{\mathbb{B}.46}$$

with rule of correspondence

$$(\mathbf{g}\,\,\lambda)\,\vert\,\rightarrow\mathbf{g}\,\mathbf{\hat{x}}\,\,\tag{\text{B.47}}$$

is a linear isomorphism of g, in Wm, since exist W0, neighborhood of the 0, in W and a neighborhood U', of the 0, in Uj, such that

$$\Phi\text{; }\mathsf{Wo}\times\mathsf{U}'\to\mathsf{W}',\tag{\mathsf{B.48}}$$

with W W, an open neighborhood of Xg, and with rule of correspondence

$$(\mathbf{x}, \mathbf{Z}) \mid \to \Phi (\exp \mathbf{x}, \mathbf{Z}), \tag{\mathbb{B}.49}$$

Let W1, be a neighborhood of the 0, in W0, such that *exp*(adW1)X, is a neighborhood of X in *N*j. If we contract to the neighborhood W0, in U', we assume that

$$\Phi(\exp \mathsf{W}v, \mathsf{U}') \cap \mathcal{N}(\subset \exp(\text{ad}\,\mathsf{W}v)\mathsf{X}) \tag{\text{B.50}}$$

Suppose that ZU' and X + Z*O*j. Then X + Z*O*j j(*exp*W0, U'). Thus X + Z = *exp*(adv)X, vW1. Of where j(1, Z) = j(*exp*v, 0), then *exp*v = 1. This implies that v = Z = 0. Then (Xj + Uj) *O*j = Xj.

Assume that g = *z*  [g, g]. Let Uj, be like before. Consider also that j(g, Uj) = g(X + Uj). Indeed, we can to apply the before argument to the neighborhood by 0, in g, Uj, of U, and demonstrate that ZUj, then Z = 0. But this is fulfilled trivially since gY = *z*(g). If gG, and ZU = *z*(g), then XUj, g(X + Y)U Uj, where U Uj, is a neighborhood by 0, of *z*(g) [g, g]. Since j(g, Uj)(G U), is a diffeomorphism in 0, of *z*(g) U', in a neighborhood of X, in g, then j(*expz*(g), U'), is an isomorphism. Let E, be a vector field on g, defined by

$$\text{Ef}(\mathbf{x} + \mathbf{y}) = (\mathbf{d}/\mathbf{d}t)(\mathbf{f}(\mathbf{x} + \mathbf{t}\mathbf{y}))|\_{\mathbf{t} = \mathbf{1}\_{\lambda}} \tag{\mathbf{B}.51}$$

 x*z*(g), and y[g, g]. If x1, , xn, are linear coordinates on g, such that Xi<sup>i</sup> q, are linear coordinates on [g, g] and Xii > q, are coordinates on *z*(g), then

$$\mathbf{E} = \Sigma \iota\_{\leq \mathbf{q}} \mathbf{x} \boldsymbol{\hat{\alpha}} \boldsymbol{\hat{\beta}} \mathbf{x} \boldsymbol{\hat{\iota}} \tag{\mathbb{B}.52}$$

If xixi, then I, xi = i + tI, i*z*(g), and i[g, g]. Considering a restriction of *z*  [g, g], in B0, (a neighborhood of the 0) then t = 0,

$$\text{exppt}[\mathsf{Z}(\mathfrak{g}) \oplus \{\mathfrak{g}, \mathfrak{g}\}] | \mathfrak{a} \mathbf{0} = \text{exppt}(\mathsf{t}\mathfrak{g})|\_{\mathfrak{i} = \mathfrak{o}\_{\mathfrak{i}}} \tag{\mathsf{B}.53}$$

Then a canonical base of Eg, in t = 0, is /x1, , /xn, and

$$
\Sigma = \Sigma\_{\mathsf{i}\rightarrow\mathsf{q}}\mathsf{x}\_{\mathsf{i}} + \Sigma\_{\mathsf{i}\\_\mathsf{q}}\mathsf{x}\_{\mathsf{i}} \,\widehat{\mathcal{O}}/\widehat{\mathcal{O}}\mathsf{x}\_{\mathsf{i}} \,\tag{\mathsf{B}.54}
$$

But by (B. 53), Eg,

$$\mathbf{E} = \Sigma\_{\mathbf{i}\leq\mathbf{q}}(\mathbf{x}\mathbf{i}+\mathbf{t}\mathbf{y}\mathbf{i})|\mathbf{t}=\mathbf{0}\quad\text{ $\mathcal{O}/\partial\mathbf{x}=\mathbf{E}\prime+\mathbf{t}\mathbf{E}\prime$ }\tag{\mathbf{B}.55}$$

Then i > qxi = 0, of where E = <sup>i</sup> qxi /xi.

**Lemma B. 3.** Let F, be a space of all the distributions with compact support (*z N*) . If TF, then *dim*C[E]T < , and the characteristic values of E, and F, are all real and strictly more little or lows that –q/2.

We fix a j, and let *O*j . Let X*z O*j, be and let j, Uj, V, and j, be such that j(g, Z) = g(X + Z), gG, and XUj. Let

$$\mathbf{V}\_{!} = \{ Z \in \mathbf{V}^{\prime} \mid \Phi(\mathbf{g}\_{\nu} Z) \in \mathbf{W}\_{\prime} \,\forall \, \mathbf{g} \in \mathbf{G} \}\_{\prime}$$

Let V = *z*(g). Assume that *O*j 0. Let y1, , yn, be linear coordinates on V [g, g], such that yk(V Vm) = 0, if m k. If ZV, then Z = mZm, with ZmV Vm. If we consider

$$\text{adHZ}\_{\text{m}} + \mu\_{\text{m}} \text{Z}\_{\text{m}} = 0\_{\text{m}}$$

H = X + Z, then

$$\mathbf{Z} = \Sigma\_{\mathbf{m}} (1/2\mu\_{\mathbf{m}} + 1)\mathbf{Z}\_{\mathbf{m}}, \qquad \text{(1/2)}\\\mathbf{H} = \chi\_{\mathbf{m}}$$

Indeed, only is necessary consider that H = X + Z, adHZm = (m + m)Zm, of where

$$\text{adHZ}\_{\text{m}} + (\mu\_{\text{m}} + \mu\_{\text{m}}) \text{Z}\_{\text{m}} = 0\_{\text{m}}$$

of where (1/2)Z = m(1/2m + 1)Zm. Then considering the map

$$\Phi \colon \mathbf{G} \times (\mathbf{V} \cap \mathbf{V}^{\mathrm{m}}) \to \mathrm{Ad}(\mathbf{g}) \\ \Omega \subset \Omega,\tag{\text{B.56}}$$

with rule of correspondence

136 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

coordinates on [g, g] and Xii > q, are coordinates on *z*(g), then

Then a canonical base of Eg, in t = 0, is /x1, , /xn, and

g], in B0, (a neighborhood of the 0) then t = 0,

Then i > qxi = 0, of where E = <sup>i</sup> qxi /xi.

But by (B. 53), Eg,

more little or lows that –q/2.

g(X + Z), gG, and XUj. Let

H = X + Z, then

ZU = *z*(g), then XUj, g(X + Y)U Uj, where U Uj, is a neighborhood by 0, of *z*(g) [g, g]. Since j(g, Uj)(G U), is a diffeomorphism in 0, of *z*(g) U', in a neighborhood of X, in g, then j(*expz*(g), U'), is an isomorphism. Let E, be a vector field on g, defined by

x*z*(g), and y[g, g]. If x1, , xn, are linear coordinates on g, such that Xi<sup>i</sup> q, are linear

If xixi, then I, xi = i + tI, i*z*(g), and i[g, g]. Considering a restriction of *z*  [g,

 *exp*t[*z*(g) [g, g]]B0 = *exp*(tg)t = 0, (B. 53)

**Lemma B. 3.** Let F, be a space of all the distributions with compact support (*z N*) . If TF, then *dim*C[E]T < , and the characteristic values of E, and F, are all real and strictly

We fix a j, and let *O*j . Let X*z O*j, be and let j, Uj, V, and j, be such that j(g, Z) =

Let V = *z*(g). Assume that *O*j 0. Let y1, , yn, be linear coordinates on V [g, g], such that

adHZm + mZm = 0,

Z = m(1/2m + 1)Zm, (1/2)H = X,

adHZm + (m + m)Zm = 0,

Indeed, only is necessary consider that H = X + Z, adHZm = (m + m)Zm, of where

(g, Z)W, gG,

Vj = ZV

yk(V Vm) = 0, if m k. If ZV, then Z = mZm, with ZmV Vm. If we consider

Ef(x + y) = (d/dt)(f(x + ty))t = 1, (B. 51)

E = <sup>i</sup> qxi/xi, (B. 52)

E = i > qxi + <sup>i</sup> qxi /xi, (B. 54)

E = <sup>i</sup> q(xi + tyi)t = 0 /xi = Ei + tEi (B. 55)

$$(\mathbf{g}\_{\prime}\mathbf{Z})\mid\rightarrow\mathbf{g}(\boldsymbol{\chi}+\mathbf{Z}),\tag{\text{B.57}}$$

whose differential is suprajective in G (V Vm), and j, we have that

$$(\mathbf{d}\Phi)\_{\mathfrak{G}}\mathbf{z}(\mathbf{(1/2)H}\_{\prime}\Sigma\_{m}(\mathbf{1/2}\mu\_{m}+\mathbf{1})\mathbf{Z}\_{m}) = \mathbf{g}(\mathbf{X}+\mathbf{Z}) = \Phi(\mathbf{g},\mathbf{Z}),\tag{\text{B.58}}$$

 gG, and ZUj. Since j, is a submersion in Uj *N*j, we can to define distributions TD'(*N* ), and ED'(U), such that

$$
\psi^0(\mathbf{T}) = \mathbf{E}',\tag{\mathbf{B}.59}
$$

where E, is related with T. But (dj)g, Z(G (V Vm)), implies that Zg,

$$\mathbf{d}\Phi^0(\mathbf{E}\mathbf{T}) = \mathbf{d}\mathbf{g}\otimes\mathbf{E} = \mathbf{Z}\Phi^0(\mathbf{T}) = (\Sigma\_\mathbf{m}(1/2\mu\_\mathbf{m} + 1)\mathbf{y}\_\mathbf{m}\hat{c}/c\mathbf{y}\_\mathbf{m})\Phi^0(\mathbf{T}), \quad \text{(B.60)}$$

The election of the system of neighborhoods of the 0, in g, given by the open sets in , (sets foreseen in page 115) implies that if

$$\text{supp}\,(\mathbf{T}) \subset (\mathcal{Z} \oplus \mathcal{N}) \cap \Omega\_{\prime}$$

Then

$$\text{supp}\,\Phi^0(\mathcal{T}) \subset \mathcal{U} \times \{0\}^\prime$$

Indeed, by definition

$$\text{supp}(\mathbf{T}) = \langle \mathbf{T} \in \mathbf{F} \mid \mathbf{T} = \mathbf{X} + \mathbf{Z}\_{\prime} \,\forall \, \mathbf{X} + \mathbf{Z} \in \langle \mathbf{U} \oplus \mathcal{N} \rangle \rangle \cap \Omega \rangle,\tag{\mathbb{B}.61}$$

and given that j, is a submersion of G (V Vm), in Ad(g), in particular 0j, is a submersion such that

$$\mathbf{Q}\_{\parallel}|\_{\Gamma \times \{0\}}(\mathbf{T}) = \mathbf{Q}\_{\parallel}(\mathbf{E}\mathbf{T}) = \mathbf{Z}\mathbf{Q}\_{\parallel}(\mathbf{T}) = \mathbf{Z}\mathbf{E}\_{\parallel}$$

where

$$\text{supp}\mathfrak{P}^{\emptyset}(\mathcal{T}) = \langle \mathcal{E} \in \mathcal{F} | \mathcal{E} = \Sigma\_{\leq \cdot \leq \Psi} \propto \hat{\mathcal{O}} \langle \hat{\otimes} \chi, \ \forall \ \mathbf{x} \in \mathfrak{Z}(\mathfrak{g}) \rangle,\tag{\mathsf{B.62}}$$

Then supp0j(T) U 0.

Let

$$\mathsf{F} \vdash \langle \mathsf{E}^{\sim} \in \mathsf{F} | \mathsf{supp} \, \mathsf{E}^{\sim} \subset \langle \mathsf{Z} \oplus \mathsf{N} \rangle \cap \Omega \rangle,\tag{\mathsf{B}.63}$$

We prove by descendent induction that is TFj, then

$$dim \mathbf{C} \mathbf{[E]} \mathbf{T} < \infty\_r$$

and the eigenvalues m, of E, on Fj, are such that m < q/2. Indeed, by the lemma19, E, acts semi-simply (diagonalizable) on Fr, with strictly eigenvalues more little than –q < -q/2. We assume to j + 1, that this last affirmation is valid and we demonstrate this to Fj.

Let TFj. Then 0j(T), have support in U 0. Since

$$\Phi^0(\mathbf{ET}) = (\Sigma\_{\mathrm{m}}(1/2\mu\_{\mathrm{m}} + 1)Z\_{\mathrm{m}})\Phi^0(\mathbf{T}), \quad \text{(B.64)}$$

with Zm = ym/ym, then 0j(ET) = m(ai + 1)Ej, aiR+ (I = 1, 2, , s) then 0j(ET), acts semisimply on the space of tempered distributions D'(U) = F'j Fj, that is to say;

0j((E ai)T) = 0,

We call m(m + 1) = q. Such m = q d. Thus ai ½(d + q). Thus of the implication

If supp(T) (*z N*j) , then supp0j(T) U 0,

we have that if suppE = suppET Fj, then

$$\text{supp}(\Pi(\mathbf{E} - \mathbf{a})) \\
\mathbf{T} \subset \mathbf{F}\_{\mathbb{H}^{+}}$$

Thus the property is established j. Then j,

dimC[E]T < ,

and all the eigenvalues m, of E, on F, are real and more strictly little that q/2.

Let (X + Z) = B(X, X), be Z*z*, and X[g, g]. Let Xi, be a base of g, such that xi*z*(g) i > q, and B(xi, xj) = iij, with i = 1.

Let 1 = <sup>i</sup> <sup>q</sup>i2/xi 2, and 0 = 1. We see to , like a differential operator under multiplication. Be h = E + (q/2)I, x = 1/2, and y = 1. Then a direct calculus gives

$$\text{I[h, x] = 2x, [h, y] = -2y, [x, y] = h, [y]}$$

**Lemma. B. 4.** If TF, and if p, is not a null polynomial in a variable then p(1)T = 0, implies that T = 0.

*Proof*Indeeed, consider the commutative ring

<sup>19</sup> **Lemma.** Let aj, be real numbers non-negative to j = 1, , n. Let D = (aj + 1)Ej. Then D act semi-simply on the spaces D'U1(U), with real eigenvalues , such that naj.

D'U1(U), is the space of distributions on U supported on U1 0.

Appendix B – Regularity on a Lie Algebra 139

$$\mathbf{C}[\mathbf{H}] = \langle \mathbf{p} \in \mathbf{P}(\mathfrak{g})^{\mathbb{C}} \mid \mathbf{p} \langle \mathbf{D} \rangle = \mathbf{p} \Sigma\_{\leq} \lhd \mathbf{x} \acute{\boldsymbol{\alpha}} \langle \grave{\beta} \mathbf{x}, \forall \ \mathbf{x} \in \mathbf{Z}(\mathfrak{g}) \rangle,\tag{\text{B.65}}$$

and consider the torsion element TF, where T, is a u-module that satisfy the hypothesis of the corollary20 then C[H]T = 0, TF. In particular to someone 1DO(') I(g),

### p(1)T = 0,

then T = 0, to someone p(1)C[H], which is equivalent to that

138 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

*dim*C[E]T < , and the eigenvalues m, of E, on Fj, are such that m < q/2. Indeed, by the lemma19, E, acts semi-simply (diagonalizable) on Fr, with strictly eigenvalues more little than –q < -q/2. We

 0j(ET) = (m(1/2m + 1)Zm)0j(T), (B. 64) with Zm = ym/ym, then 0j(ET) = m(ai + 1)Ej, aiR+ (I = 1, 2, , s) then 0j(ET), acts

0j((E ai)T) = 0,

If supp(T) (*z N*j) , then supp0j(T) U 0,

supp((E ai))T Fj+1,

dimC[E]T < ,

Let (X + Z) = B(X, X), be Z*z*, and X[g, g]. Let Xi, be a base of g, such that xi*z*(g) i > q,

[h, x] = 2x, [h, y] = 2y, [x, y] = h, **Lemma. B. 4.** If TF, and if p, is not a null polynomial in a variable then p(1)T = 0, implies

<sup>19</sup> **Lemma.** Let aj, be real numbers non-negative to j = 1, , n. Let D = (aj + 1)Ej. Then D act semi-simply on the spaces

2, and 0 = 1. We see to , like a differential operator under

and all the eigenvalues m, of E, on F, are real and more strictly little that q/2.

multiplication. Be h = E + (q/2)I, x = 1/2, and y = 1. Then a direct calculus gives

assume to j + 1, that this last affirmation is valid and we demonstrate this to Fj.

semisimply on the space of tempered distributions D'(U) = F'j Fj, that is to say;

We call m(m + 1) = q. Such m = q d. Thus ai ½(d + q). Thus of the implication

We prove by descendent induction that is TFj, then

Let TFj. Then 0j(T), have support in U 0. Since

we have that if suppE = suppET Fj, then

Thus the property is established j. Then j,

Indeeed, consider the commutative ring

D'U1(U), with real eigenvalues , such that naj. D'U1(U), is the space of distributions on U supported on U1 0.

and B(xi, xj) = iij, with i = 1.

Let 1 = <sup>i</sup> <sup>q</sup>i2/xi

that T = 0.

*Proof*

$$\operatorname{Tor}(\mathcal{F}\_{\mathbb{H}}, \mathcal{F}\_{\mathbb{H}^{+}}) = 0.\tag{B.66}$$

Now we demonstrate the following lemma that is useful in the final of the demonstration of the regularity theorem.

**Lemma B. 5.** If SF, and if p, is a polynomial not null in a real variable such that p()S = 0, then S = 0.

*Proof*. First we demonstrate that if SF, C, if ( )S = 0, then S = 0. Let S = Sbe with (h )dS= 0, to someone dZ+. Then

$$0 = (\Box - \zeta)\mathbf{S} = \Sigma(\Box \mathbf{o} - \zeta)\mathbf{S}\_{\mu} + \Sigma\Box\mathbf{i}\mathbf{S}\_{\mu}\mathbf{s}$$

Let , be the minimal , such that S, that is to say; the minimal eigenvalue of SSince

$$(\hbar - (\mu - \mathcal{D}))^d \Box \iota S\_{\mu} = 0,\tag{B.67}$$

hDO('), and (2), the corresponding value of Sd'Z+, with d' < d, then

1S= 0,

By the corollary of the foot page 10, implies that S = 0. But this contradicts the fact of that S 0 to someone , minimal. To demonstrate the lemma B. 5, apply the hypothesis of induction to the grade of p.

If *gr*p = 0, then the result is trivial. Suppose that the result to p 0, such that *gr*p = d , and we demonstrate the validity of the affirmation of the lemma to *gr*p = d, dZ+. Indeed, if *gr*p = d, then p(t) = (t )q(t), C, and q(t), a polynomial of grade d 1. But if q()S = 0, then S = 0, for hypothesis of induction. Then

$$0 = \mathbf{P}(\square)\mathbf{S} = (\square - \zeta)(\mathbf{d}(\square)\mathbf{S}).$$

But this affirm that q()S = 0, then S = 0.

<sup>20</sup> **Corollary.** Let M, be u-module such that if mM, then dim C[H]m < , and such that the eigenvalues of H, on M are real and strictly minors than 0. Then the action of C[Y], is a free torsion on M.

**Note:** T is a element of torsion since (H q)T = 0, for someone H qIC[H], (H q) 0.

## **APPENDIX C: Some Elements of the Asymptotic Behavior of the Matrix Coefficients**

In this appendix we treat of highlight the analytic part of the real reductive group in the endomorphism algebra whose coefficients are the maximum weight of the discrete principal series and that are the coefficients of the cohomology of the exact succession of the (g, K) modules that are finitely generated like U(n)-modules, which induces the Hilbert representation of G, of finite dimension useful to the representation of a Lie group of infinite dimension.

Let G, be a real reductive group, and we will do the identification G0 = 0(G0), along of this appendix, where G0, is the identity component of G, (*remember that* 0G = gG2 = 1, X(G). 0G = 0ANK, G0, *is the identity component of* G, *and* X(G), *is the space of continuous homomorphism of* G, *in the multiplicative group* R\* = (R, )/), then 0(G0), is the identity component of 0G. Then if

$$\mathbf{G}^{0} \equiv \langle \mathbf{g} \in \mathbf{G} | \mathrm{Ad}(\mathbf{g}) = \mathrm{I}, \forall \text{ } \mathrm{Ad} \in \mathrm{End}(\mathbf{G}) \rangle\_{\vee} \tag{C.1}$$

In particular

$$\mathbb{P}^{\mathbb{Q}}(\mathbb{G}^{\mathbb{Q}}) = \{ \mathbf{g} \in \mathbb{G} | \mathbb{X}(\mathbb{g}) = 1, \forall \ \mathbb{X} \in \mathbb{X}(\mathbb{q}\mathbb{G}) \},\tag{C.2}$$

Let 0, be the set of semi-simple roots of (P, A). Let F, be a subset of 0and be (PF, AF), the corresponding canonical parabolic pair, that is to say; explicitly

$$\Phi(\text{Pr}, \text{A} \mathbb{v}) = \langle \begin{pmatrix} \mathbf{p}, \mathbf{a} \end{pmatrix} \in \text{M} \mathbb{v} \mid \text{M} \mathbb{v} = \text{A} \mathbb{v} \times^0 \text{M} \mathbb{v}, \text{ and } \text{Pr} = \text{M} \mathbb{v} \text{N} \mathbb{v} \rangle,\tag{\text{C. 3}}$$

**Lemma. C. 1.** Be V*H*. Then the module V/nFV, is an admissible module finitely generated like a (mF, PF K)-module.

Proof. See [28], and [35].

**Lemma. C. 2.** Let V1 V2 , be an increasing chain of sub-modules of V .

Proof. Let Vj = vV Vj (v) = 0. Then V1 V2 , is a decreasing chain of submodules of V. Indeed, the exact succession in *H*,

$$0 \to \mathcal{V}\_{!} \to \mathcal{V} \to \mathcal{V}/\mathcal{V}\_{!} \to 0, \quad \forall \ j \ge \mathbf{k}, \tag{C.4}$$

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

induces an exact succession in *V*, given for

$$0 \leftarrow \mathcal{V}^{\rightarrow} \leftarrow \mathcal{V}^{\rightarrow} \leftarrow \mathcal{V}/\mathcal{V}^{\rightarrow} \leftarrow 0,\ \forall \ j \ge \mathbf{k},\tag{\mathbb{C}.5}$$

which jZ+, kerTj n<sup>j</sup> V . By the theory of Jacquet modules, V is a U(n)- module, finitely generated like a g-module whose finite length is the of (g, K)-module. Then exist kZ+, minimal such that Vj = Vk, j > k. But for the third application of the Jacquet modules, having a V , like a Jacquet module of V, V, is admissible and the Jacquet module of V, is

$$\mathbf{j(V) = \{ \mu \in V^{\cdot \cdot} \mid \mu(V\_{\cdot}) = 0 \}}\tag{C.6}$$

thus Vj = Vk , j > k. Then V , is finitely generated like a U(n)-module, then by a theorem that affirm that a (g, K)-module that is finitely generated like U(n)-module is admissible, V , is admissible (V = V()\*). Then V , is a (g, K)-module finitely generated admissible, thus V *H*.

Let (, H), be a Hilbert representation of G. Let (H)' = *L*(H. C). If gG (respectively Xg), and (H)', then we define the map

$$\mathbf{G} \times (\mathbf{H}^\*)' \to \mathbf{H}^\* \tag{\mathbb{C} \text{ } }$$

whose rule of correspondence is

$$(\mathbf{g}, \mu) \mid \mapsto \mathbf{g}\mu,\tag{\text{C.8}}$$

where v(H)'; g(v) = (gv), (respectively X(v) = ((X)v, Xg), then for the elemental theory of representations, is had that

$$\mathbf{g}\chi\mathbf{\iota} = (\operatorname{Ad}(\mathbf{g})\chi)\mathbf{\iota}\mu,\tag{C.9}$$

(H)', Xg, and gG. Indeed, consider the map

$$\pi \colon \mathcal{U}(\mathfrak{g}) \otimes (\mathcal{H}^{\circ})' \to \mathcal{H}^{\circ},\tag{C.10}$$

whose rule of correspondence is

$$\mathbf{g} \otimes \mathbf{v} \mid \to \pi(\mathbf{g})\mathbf{v},\tag{C.11}$$

v(H)'. Then

gXv = (g)(X)v, (C. 12)

but for be (H)', a subspace finitely generated like a (g, K)-module of H, then

$$
\pi(\mathbf{g})\pi(\mathbf{X})\mathbf{v} = \pi(\operatorname{Ad}(\mathbf{g})\mathbf{X})\pi(\mathbf{g})\mathbf{v},\tag{C.13}
$$

thus, if v(H)', Xg, and if H, is the generated for (K)v, then

(Ad(g)X)(g)v = (Ad(g)X)gv,

(H)', Xg, and gG.

142 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

V V

j(V) = V

/Vj

, like a Jacquet module of V, V, is admissible and the Jacquet module of V, is

generated like a g-module whose finite length is the of (g, K)-module. Then exist kZ+, minimal such that Vj = Vk, j > k. But for the third application of the Jacquet modules,

that affirm that a (g, K)-module that is finitely generated like U(n)-module is admissible, V

Let (, H), be a Hilbert representation of G. Let (H)' = *L*(H. C). If gG (respectively Xg),

G (H)' H, (C. 7)

 (g, ) ∣ g, (C. 8) where v(H)'; g(v) = (gv), (respectively X(v) = ((X)v, Xg), then for the

gX= (Ad(g)X)g, (C. 9)

g v ∣ (g)v, (C. 11)

(g)(X)v = (Ad(g)X)(g)v, (C. 13)

Uj

but for be (H)', a subspace finitely generated like a (g, K)-module of H, then

thus, if v(H)', Xg, and if H, is the generated for (K)v, then

0, j > k, (C. 5)

(Vj) = 0, (C. 6)

,

. By the theory of Jacquet modules, V is a U(n)- module, finitely

, is finitely generated like a U(n)-module, then by a theorem

, is a (g, K)-module finitely generated admissible, thus

(g) (H)' H, (C. 10)

gXv = (g)(X)v, (C. 12)

induces an exact succession in *V*, given for

, j > k. Then V

is admissible (V = V()\*). Then V

and (H)', then we define the map

elemental theory of representations, is had that

(H)', Xg, and gG. Indeed, consider the map

whose rule of correspondence is

whose rule of correspondence is

v(H)'. Then

V

0 Vj

which jZ+, kerTj n<sup>j</sup>

having a V

= Vk

thus Vj

V *H*. Let (H)'K = (H)' K, generate a subspace of finite dimension, then the before identity imply that (H)'K , is a (g, K)-module. Indeed, kK, (H)', Xg

$$\mathbf{k}\lambda\mathbf{\varchi} = (\mathbf{Ad}(\mathbf{k})\lambda\mathbf{\varchi})\mathbf{k}\mu\mathbf{\varchi}$$

Then the elements (K)v = Kv, generate a subspace of finite dimension of (H)', whose actions K, are continuous. In particular, Kv, are continuous and generate a all subspace of the module (H)'K. If Yt, and v(H)', then

$$(\mathbf{d}/\mathbf{d}t)\exp(t\mathbf{Y})\mathbf{v}|\_{t=0} = \mathbf{Y}\mathbf{v}\_{\prime\prime}$$

thus (H)'K, is a (g, K)-module. Let be the map

$$
\sigma: \mathcal{H} \to \mathcal{H}',\tag{\text{C. 14}}
$$

whose rule of correspondence is

$$\mathbf{v} \mid \rightarrow \mathbf{c}(\mathbf{v}),\tag{C.15}$$

Then wH, (v)(w) = <v, w>. Then , is a linear continuous and conjugate isomorphism of H, in H'. H', is the space of the linear functional on H, that is to say; H' = *L*(H, C), is endowed like Banach space of an inner product that is continuous in all point of H. Then the map

$$\mathbf{H} \times \mathbf{H'} \to \mathbf{H'} \tag{\mathbb{C} \ 16}$$

whose rule of correspondence

$$\{\mathbf{v}, \mathbf{w}\} \mid \to \preccurlyeq \mathbf{v}, \tag{\mathsf{C}. 17}$$

is a linear bijection such that c (v) = c <v, w> = <v, w> = (w). Thus the map , is a conjugate continuous linear isomorphism of H in H'.

**Lemma. C. 3.** If (, H), is admissible then (H)'K = (v)K. Furthermore, (H)'K = (K) .

*Proof*. If (, H), is admissible then K , dim H() < , then

$$\sum\_{\gamma \in \mathcal{K}^{\triangle}} \mathcal{H}(\gamma) \upsilon = (\mathcal{H}^{\circ})^{\circ}\_{\mathcal{K}^{\prime}} \tag{C.18}$$

If wH, then

$$\sum\_{\gamma \in K^{\triangle}} \mathcal{H}(\gamma) v = \mathcal{H} \otimes \mathcal{H}\_K = \sigma(\mathcal{H}\_K)\_{\prime} \tag{C.19}$$

since to all wH, and vHK, (v)(HK). Then for (B. 18), and (B. 19), it is have that

$$(\mathbf{H}^\*)\mathbf{\hat{x}} = \mathbf{\sigma(H\mathbf{x})},\tag{\text{C. 20}}$$

Then (HK) = (HK) . Given that

$$\sigma(\mathbf{H}\mathbf{x}) = \sum\_{\gamma \le \mathbf{v}} \chi^{\sim} \mathbf{H}(\gamma) \mathbf{v} = \bigoplus\_{\gamma \le \mathbf{v}} \chi^{\sim} \mathbf{H}^\*(\gamma) = \mathbf{H}^\* \tag{\subseteq. 21}$$

Since H\*, is a K-module, and for definition

$$\text{(Hk)}^{\circ} = \langle \mathbf{v} \in \mathbf{H}^{\bullet} \vert \,\mathrm{Kv} \rangle \text{.}\text{ generates a subspace of finite dimension}\rangle\text{.}\tag{C.22}$$

Using the map

$$\mathbf{K} \times \mathbf{H}^\* \to \mathbf{H}^\* \tag{\Gamma}$$

whose rule of correspondence is

$$(\mathbf{k}, \mathbf{v}) \mid \to \mathbf{k}\mathbf{v},\tag{\text{C. 24}}$$

 vH\*, then (HK) = (HK) . Then (H)'K = (HK) .

But this is posible given that (mF, PF)-modules of V/nFV, are admissible like U(\*nF)-modules and V , with

$$\mathbf{V}^{\circ} = \langle \mu \in \mathbf{V}^{\circ} \mid \Sigma \mathbf{K} \mu = \mathbf{w} \text{, with } \mathbf{w} \in \mathbf{W}\_{\prime} \text{ and } \mathbf{W} \subset \mathbf{V}^{\circ} \rangle\_{\prime} \tag{\text{C. 25}}$$

we have that VK = (H)'K y (H)'K = (H)K, with

$$\sigma \in \Lambda(\mathcal{H}, \mathcal{H}') = isom(\mathcal{H}, \mathcal{H}'),$$

and (H)'K, is a M-module. Then considering the inequalities, using images in (H)'K , to know,

$$|\langle \mathfrak{u}(\mathfrak{m}(\mathfrak{a})) \mathbf{v} \rangle| \le \mathbf{a}^{\delta} \sigma'\_{\mu}(\mathbf{v}) \; \forall \; \mathbf{v} \in \mathbf{H}^{\circ}, \text{ and } \mathbf{a} \in \mathbf{CL}(\mathbf{A}^{\circ}),$$

and

$$|(\mu(\mathfrak{m}(\mathsf{a}))\mathbf{v})| \le \mathbf{a}^{\delta - \alpha} \gamma'\_{\mu}(\mathbf{v}) \,\,\forall \,\,\mathbf{v} \in \mathsf{H}^{\varepsilon} \text{ and } \mathbf{a} \in \mathsf{CL}(\mathbf{A}^{\*})\_{\omega}$$

we can estimate the terms of the integral:

$$\mathbf{F}\left(\mathbf{t},\mathbf{a}';\mathbf{v}\right) = \exp(\mathbf{-t}\mathbf{B})\mathbf{F}\left(\mathbf{t}\_{0'},\mathbf{a}';\mathbf{v}\right) \cdot \exp(\mathbf{-t}\mathbf{B}) \int\_{-\mathbf{0}}^{\mathbf{t}} \exp(\mathbf{s}\mathbf{B}) \cdot \mathbf{G}\left(\mathbf{s},\mathbf{a}';\mathbf{v}\right) \mathrm{d}\mathbf{s},\tag{\text{C.26}}$$

that of the inequalities, we obtain

$$\|\|\mathbf{F}(\mathbf{t}, \mathbf{a'}; \mathbf{v})\|\| \le \langle \mathbf{a'} \rangle^{\delta} \|\mathbf{b}(\mathbf{v}), \forall \ \mathbf{a'} \in \mathbf{CL}(\mathbf{A}^{\star}), \tag{\mathsf{C.27}}$$

with , a continuous seminorm on HThen we obtain

$$||\mathbf{F}(\mathbf{t}, \mathbf{a}'; \mathbf{v})|| \le \exp|(\delta(\mathbf{H}) - 1)\mathbf{t}| \text{ (a')} ^\delta \text{\(\(\mathbf{v}\)} \text{\(\forall} \text{ a'} \in \text{CL}(\mathbf{A}^\*), \tag{\text{C. 28}}$$

with ', a continuous seminorm on H. But

$$||\mathsf{e}\mathbf{x}\mathbf{p}(\mathsf{s}\mathsf{B})|| \leq \mathsf{C}(1+|\mathsf{o}|) \mathsf{Fe}^{\mathsf{s}\mathsf{Re}z} \quad \forall \quad \mathsf{s} \in \mathbb{R}, \ \mathsf{p} \leq \mathsf{d}\_{\mathsf{s}}.$$

which is immediate of (B zI)P = 0, to B = bnk. Such calculation implies that

$$\mathbb{P}\left|\left|\mathbf{F}\left(\mathbf{t},\mathbf{a}';\mathbf{v}\right)\right|\right| \leq \mathbb{C}\left(1+\mathbf{t}\right)^{\mathbb{P}}\mathbf{e}^{-\mathsf{H}\mathrm{Re}\mathbf{z}}\left(\mathbf{a}'\right)^{\delta}\beta\left(\mathbf{v}\right)\left(1+\int\_{-\mathsf{H}}^{\mathsf{t}}\mathbf{e}^{\mathsf{s}\left(\mathrm{Re}\mathbf{z}+\mathbf{d}\left(\mathbf{x}\right)-1\right)}\mathrm{ds}\right) \tag{\mathsf{C.29}}$$

to some continuous seminorm , on H, y C > 0. Observe that (1 + s)p e<sup>s</sup> , is bounded by C > 0, to > 0, and s 0. Thus also

$$||\mathbf{F}(\mathbf{t}, \mathbf{a}'; \mathbf{v})|| \le \mathcal{C}(1+\mathbf{t})\mathbb{P}e^{-\mathsf{R}\cdot\mathbf{z}}\langle\mathbf{a}'\rangle^{\delta}|\mathbf{b}(\mathbf{v}) + \mathcal{C}(1+\mathbf{t})\mathbb{P}e^{(\mathsf{R}\cdot\mathbf{z}+\frac{\delta}{\delta}\chi)}\langle\mathbf{a}'\rangle^{\delta}|\mathbf{b}(\mathbf{v})\rangle\tag{\mathsf{C}\cdot\mathcal{B}0}$$

t > 0, , a continuous seminorm on H, and C > 0.

The cases exist:

144 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

Then (HK) = (HK)

(HK)

whose rule of correspondence is

vH\*, then (HK) = (HK)

, with

V

we have that VK

Using the map

and V

know,

and

(HK) = K

Since H\*, is a K-module, and for definition

. Given that

K H\* H

we can estimate the terms of the integral:

that of the inequalities, we obtain

. Then (H)'K = (HK)

= (H)'K y (H)'K = (H)K, with

'(v)

(((a))v) a

(((a))v) a'(v)

F(t, a'; v) (a')

since to all wH, and vHK, (v)(HK). Then for (B. 18), and (B. 19), it is have that

(H)'K = (HK), (C. 20)

H()v = K

 .

But this is posible given that (mF, PF)-modules of V/nFV, are admissible like U(\*nF)-modules

(H, H') = *isom*(H, H'), and (H)'K, is a M-module. Then considering the inequalities, using images in (H)'K , to

<sup>t</sup>

= vH\* Kv, generates a subspace of finite dimension, (C. 22)

V\*jKj = w, with wW, and W V\*, (C. 25)

vH, and aCL(A+),

vH and aCL(A+),

(v), a'CL(A+), (C. 27)

<sup>0</sup> <sup>0</sup> F t, a'; v = exp -tB F t , a'; v -exp -tB exp sB G s, a'; ( ) ( ) v ds, (C. 26)

H\*() = H\*, (C. 21)

, (C. 23)

(k, v) ∣ kv, (C. 24)

**Case I.** If (X) 2/3 (X), then exist a continuous seminorm , on H, such that

$$||\mathbf{F}(\mathbf{t}, \mathbf{a}'; \mathbf{v})|| \le \mathbf{C}(1+\mathbf{t})^{\mathbb{P}} \mathbf{e}^{\mathbb{A}^{\mathbb{A}}(\mathbf{x})} \text{ (a')}^{\mathbb{S}} \mathbf{b}(\mathbf{v}) \quad \forall \ \mathbf{t} \ge \mathbf{0},$$

**Case II.** If (X) 2/3 > (X), then in (I), we replace , by (1/2), and we iterate the steps to the inequality (III). Of this way, we reduce the case II to case I.

How can refines this technique to demonstrate the asymptotic behavior of the developments of the matrix coefficients (((a))v), of the finite generated Hilbert representation and admissible required to induce representations?

Let F 0, be with 0, a roots space in the system of roots (PF, AF) corresponding to canonical parabolic pair (PF, AF).

Then the asymptotic behavior of the matrix coefficients of the induced representations to operators of infinite dimension is established while that ((exp(tX)v), be asymptotic, to know

$$\Sigma\_{\mu\underline{\boldsymbol{\alpha}}}\mathbb{0}\exp(\mathsf{t}\mu(\mathsf{X}))\Sigma\_{\mathbb{Q}\underline{\boldsymbol{\alpha}}}\mathbb{c}\mathsf{+}\exp(-\mathsf{t}\mathsf{Q}(\mathsf{X}))\mathsf{P}\_{\mu}\mathbb{q}(\mathsf{T}\mathsf{X},\mathsf{c}\mathsf{t}\mathsf{v})\_{\mathsf{t}}$$

when t .

finally, given that M, is a connect group on all element of G, the coefficients of the representations of the operators of infinite dimension in M, have that satisfy also this bounding condition since G is a covering of M

## **APPENDIX D: Coexeter Diagrams to some Integral Transforms in Representation Theory**

We realize a little digression of the irreducible roots systems.

**Def. D. 1.** , is irreducible if cannot be partitioned in the union of proper subsets such that each one of their roots in each set be orthogonal to each root of the other.

Examples: The systems of roots B2, and G2:

**Figure 1.** Root systems to two and three strings (= B2, G2).

One that is not irreducible is A1 A2.

Let , a base of . , will be irreducible if only yes , cannot be partitioned as it has been mentioned earlier. We demonstrate the implication:

**).** If , cannot be partitioned then , is irreducible.

*Proof.* We demonstrate the contraposition: If Q P. We suppose that , admits a partition in subsets or classes

$$
\Phi\mathbf{\dot{\iota}} \cup \Phi\mathbf{\dot{\omega}} = \Phi\mathbf{\dot{\phi}}\tag{\Box.1}
$$

With (1 2) = 0 (is to say, orthogonal between they), to be not that , can entirely to be included in 1 or in 2 which induce a similar partition of , but 1, then (1) = 0, or (2) = 0, since , develop or generate to E. This demonstrates that , is partitioned (1 2).

**).** If , is irreducible then , is not partitioned.

*Proof.* Let , be irreducible, but 1 2, with (1 2) = 0. But by the theorem on the action of the Weyl group on , is had that, if , W, then (), that is to say, each root is conjugated to a simple root such that 1 2, with i, the set of roots, having each ithcomponent a conjugated in 1Remembering that () = 0, and , then (since ). Thus the Weyl group W, is generated by the reflections (). But the formula to the reflections affirms clearly that each root in iis given of one in ifor addition or subtraction of elements of iThus, said element fall in the subspace Ei, of E, generated by iand we see that (1 2) = 0. This implies that 1 or <sup>2</sup> of where 1 or 2 . Thus 1 2, with i = , with i = 1, 2. Thus , it does not admit to be partitioned.

**Lemma D. 1.** Let , be irreducible. Relative to partial order , exist a unique maximal root (that be maximal means that is the root of maximal height in ), , (in particular , then *htht*, and (, ) 0, ). If K, then all the K0.

*Proof*. Let K, maximal with the relation of order , evidently If 1 = {K> 0}, and 2 = {K> 0}, then 1 2, is partition. Suppose that 2, is not vanish, that is to say 2 . Then (, ) 0, 2, (by lemma of simple roots) of where , is irreducible then at least 2, can be orthogonal to 1, of where to some ', arbitrary of 1,'< 0, of where () < 0. This implies by the lemma that give criteria through of the functional (), on the character of the roots , and , is a root, but clearly > , which contradict the maximalist of . Thus 2, cannot have elements, that is to say, 2 , of where all KThis demonstrate also that () 0, (with () > 0, to some , where , generates to E).

Now we demonstrate that in the system ( ), exist a unique maximal root . Consider other root with the mentioned property in the system of irreducible roots with partial order ( ). Let such root '. The precedent argument (of the before demonstration) we apply it to ', of the form that wrap at least a root , (with positive coefficients) to which () > 0. It follows that (') > 0, and ', is a root by the lemma on the criteria of the functional (), to elemental combinations and operations of roots. Thus '. But if ', is a root then or ', or ', which result be a contradiction, since ', is maximal. Then ', of where , is unique.

**Lemma D. 2.** Let , be irreducible. Then W, act irreducible on E. In particular, the W orbit21 of a root , generates E.

*Proof*. The development of a W-orbit of a root is a W-invariant subspace of E (non-vanishing) such that the W-orbit of a root generates to E. Let E' E, and W-invariant. Then there is a subspace E\* (E\* E' = E) W-invariant. Likewise, if , such that E' and E' P, then (E') = E', therefore E', then E\*, such that every root is found inside of a subspace or of the other. This divides to , in orthogonal subsets, forcing them to that one or other let be vanishing. Of the affirmation of that , generates to E, is concluded then that E' = E. Thus (E') E', W, and W acts irreducible on E.

Notes: In other words the subspace of the reflexions of the subspace E', is an irreducible root system and therefore all automorphism W, acts irreducible22

**Lemma D. 3.** Let be irreducible. Then at least two roots of different length happen in , and all the roots of length given are conjugated under W.

*Proof*. Let , be roots of different length, then not all (), W, can be such that ((), ) = 0, where (), generates to space E (Lemma D. 2). If () 0, (where by the lemma on root spaces theory, that says: *If* , *is a base of* , *then* () 0, , *in* , *and* , *is not a root*, we know that the possible radius of the square of the root length of , and , are 1, 2, 3, ½, 1/323).

Now, if and, have equal length to replace a of these for their W-conjugated we can assume that the roots , c = , C(), and () = c W(E, ), are different and non-orthogonal. By the same lemma mentioned in this demonstration is deduced that

$$<\mathfrak{a},\mathfrak{f}\succ\mathfrak{P},\mathfrak{a}\rhd\mathfrak{v}\dashv\mathfrak{z}\text{ 1.}\tag{\mathsf{D}.1}$$

**Lemma D. 4.** Let , be irreducible, with two different roots. Then the maximal root , of the Lemma D. 1, is the major length.

<> = 2()/() = 2{/}cos

then ()< >/2 = (), where

148 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

**).** If , is irreducible then , is not partitioned.

(1 2).

(since

admit to be partitioned.

where , generates to E).

of where , is unique.

With (1 2) = 0 (is to say, orthogonal between they), to be not that , can entirely to be included in 1 or in 2 which induce a similar partition of , but 1, then (1) = 0, or (2) = 0, since , develop or generate to E. This demonstrates that , is partitioned

*Proof.* Let , be irreducible, but 1 2, with (1 2) = 0. But by the theorem on the action of the Weyl group on , is had that, if , W, then (), that is to say, each root is conjugated to a simple root such that 1 2, with i, the set of roots, having each ithcomponent a conjugated in 1Remembering that () = 0, and , then

 (). But the formula to the reflections affirms clearly that each root in iis given of one in ifor addition or subtraction of elements of iThus, said element fall in the subspace Ei, of E, generated by iand we see that (1 2) = 0. This implies that 1 or <sup>2</sup> of where 1 or 2 . Thus 1 2, with i = , with i = 1, 2. Thus , it does not

**Lemma D. 1.** Let , be irreducible. Relative to partial order , exist a unique maximal root (that be maximal means that is the root of maximal height in ), , (in particular , then

*Proof*. Let K, maximal with the relation of order , evidently If 1 = {K> 0}, and 2 = {K> 0}, then 1 2, is partition. Suppose that 2, is not vanish, that is to say 2 . Then (, ) 0, 2, (by lemma of simple roots) of where , is irreducible then at least 2, can be orthogonal to 1, of where to some ', arbitrary of 1,'< 0, of where () < 0. This implies by the lemma that give criteria through of the functional (), on the character of the roots , and , is a root, but clearly > , which contradict the maximalist of . Thus 2, cannot have elements, that is to say, 2 , of where all KThis demonstrate also that () 0, (with () > 0, to some ,

Now we demonstrate that in the system ( ), exist a unique maximal root . Consider other root with the mentioned property in the system of irreducible roots with partial order ( ). Let such root '. The precedent argument (of the before demonstration) we apply it to ', of the form that wrap at least a root , (with positive coefficients) to which () > 0. It follows that (') > 0, and ', is a root by the lemma on the criteria of the functional (), to elemental combinations and operations of roots. Thus '. But if ', is a root then or ', or ', which result be a contradiction, since ', is maximal. Then ',

*htht*, and (, ) 0, ). If K, then all the K0.

). Thus the Weyl group W, is generated by the reflections

{ cos}/2 = ().

<sup>21</sup> Remember that the orbit as topological concept of a class space is a lateral class, where every element of W, is the class of an element in N(T), such that W t = t

<sup>22</sup> This is equivalent to say also that W, acts canonically on GL(*n*, E)

<sup>23</sup> Only remember that from () = cos, had been deduced:

*Proof*. Indeed, let , be arbitrary. Is enough demonstrate that the criteria of comparison that takes the computing of the functional () 0, , that () (), that is to say, > , Then we substitute , for a W-conjugated root that is inside the closure of the fundamental Weyl camera (relative to ).

For other side, we know by the lemma D. 2, that Thus to any regular, that is to say, C(), 24 () 0. This fact applied to the cases and is had that

$$(\chi, \chi - \alpha) = (\beta, \beta) \ge (\chi, \chi - \beta) = (\alpha, \alpha), \tag{D.2}$$

Therefore () () (), () 0, and .

We pass now to Classification study of the root systems. Said classification will serve after as a essential base in the Lie groups classification of arbitrary dimension.

Let , a root system of range *l*, W, their Weyl group and , a base of .

**Def. D. 2.** (Cartan matrix of ). Fixing a arrangement (1, , *l*) of simple roots. The matix (<I, j>), is called a Cartan matrix of . Their enters are called Cartan integers.

To systems of 2-range, we have the following matrices:

$$A\_1 \times A\_2 = \begin{pmatrix} 2 & 0 \\ 0 & 2 \end{pmatrix} \tag{\mathbb{D}.3}$$

$$A\_2 = \begin{pmatrix} 2 & -1 \\ -1 & 2 \end{pmatrix} \tag{\text{D. 4}}$$

$$B\_2 = \begin{pmatrix} 2 & -2 \\ -1 & 2 \end{pmatrix},\tag{D.5}$$

$$G\_2 = \begin{pmatrix} 2 & -1 \\ -3 & 2 \end{pmatrix} \tag{\text{D.6}}$$

Mapping in', and satisfying that <(), ()> = <> the Cartan matrix of , determines to , by the isomorphism .

 C(), and W C(),

that is to say, WC().

<sup>24</sup> Remember that the Weyl cameras in a sense more restricted, is all relative camera to . If further more is delimited for two hyper-planes then this is a closure of the Weyl camera in the more general sense. This is called a fundamental domain or fundamental Weyl camera. All fundamental Weyl camera can extending to a open connect region of the space E. Every point of C(), is W-conjugated to a point in E, to know

The shape of the matrix will depend of the elected arrangement, although this is not relevant. Important is that the Cartan matrix is independent of the election of ; this for the part b), of the theorem on the permutation of Weyl cameras or bases of , then if ', is other base of , then (') = , W (such that W, act transitively on bases).

**Proposition D. 1.** Let ' E'be other root system with base ' = {'1, , '*l*}. If <'i, 'j> = <i, j>, to 1 i, j *l*. then the bijection i I 'I, is extended univocally to the isomorphism E E'.

*Proof*. Using the extension of the Coxeter graphs and the properties of correspondence between bases and Weyl cameras is demonstrated the result.

If then is acquaintance that <><> = 0, 1, 2 0r 3, according to the identity < ><> = 4cos2.


**Table 1.** Strings and their root spaces

150 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

closure of the fundamental Weyl camera (relative to ).

Therefore () () (), () 0, and .

To systems of 2-range, we have the following matrices:

determines to , by the isomorphism .

WC().

space E. Every point of C(), is W-conjugated to a point in E, to know

that is to say,

a essential base in the Lie groups classification of arbitrary dimension.

Let , a root system of range *l*, W, their Weyl group and , a base of .

(<I, j>), is called a Cartan matrix of . Their enters are called Cartan integers.

1 2

2

2

2

*Proof*. Indeed, let , be arbitrary. Is enough demonstrate that the criteria of comparison that takes the computing of the functional () 0, , that () (), that is to say, > , Then we substitute , for a W-conjugated root that is inside the

For other side, we know by the lemma D. 2, that Thus to any regular, that is to

() = () () = (), (D. 2)

We pass now to Classification study of the root systems. Said classification will serve after as

**Def. D. 2.** (Cartan matrix of ). Fixing a arrangement (1, , *l*) of simple roots. The matix

2 0 , 0 2 

> 2 1 , 1 2

2 2 , 1 2 

2 1 , 3 2 *<sup>G</sup>* 

C(),

Mapping in', and satisfying that <(), ()> = <> the Cartan matrix of ,

24 Remember that the Weyl cameras in a sense more restricted, is all relative camera to . If further more is delimited for two hyper-planes then this is a closure of the Weyl camera in the more general sense. This is called a fundamental domain or fundamental Weyl camera. All fundamental Weyl camera can extending to a open connect region of the

C(), and W

*A A* (D. 3)

*A* (D. 4)

*B* (D. 5)

(D. 6)

say, C(), 24 () 0. This fact applied to the cases and is had that

**Def. D. 3.** A Coxeter graph of , is the graph of *l*, vertices which the ith-vertex is joined to the jth-vertex with i j, for <i, j> = <j, i>, edges. For example, to the root systems of range *l* = 2:

The Coxeter graph determine the numbers <i, j>, in the case of that all roots have the same length. Then <i, j> = <j, i>. In a more general case, where there is more of one different length (case of the root systems of range two; *G*2, *B*2), the graph fails to the concrete case in which is had a vertices pair that correspond to a simple short root, which cans be a long root or of big length.

A relevant fact on the Coxeter graph is that these determine completely to Weyl group, essentially because these determine the orders of the direct products of generators of W.

**Proposition D. 2.** From the Table 1, of inner products and length of roots that change in the interval 0 <><> 4cos2, is had that the order of , in W, is respectively 2, 3, 4, 6, when = /2, /3 (or 2/3), /4 (or 3/4), /6 (or 5/6). [Note that, is the rotation through 2].

To the case when we have roots of short length, is ordinary to add an arrow to designate it in the Coxeter graph. This additional information help us to recover the integers of Cartan, called to resulting graph; Dynkin diagram of (as before, this depend on the number of simple roots). Example;

Somehow someone exist such component irreducibles like the system of roots shaped by a pair of roots. Let's study the component irreducible systems. Let's remember that a system , is irreducible if and only if, (or equivalently ) do not admit a division into two orthogonal proper subsets. Then by the correspondence between Coxeter graphs and root systems, is clear that , is irreducible if and only if their graph of Coxeter is connect (in the usual sense)25.

In general is possible to give a number of connect components of the Coxeter graph, likewise, let

$$
\Delta = \Lambda\_1 \cup \Lambda\_2 \cup \Lambda\_3 \cup \dots \cup \Lambda\_{l'} \tag{D.7}
$$

the corresponding partition of , in mutually orthogonal subsets. If Ei, is the generated by i, is clear that

<sup>25</sup> If this is non-connect it would be equivalent to that f, admits a partition into two orthogonal proper subsets.

$$\mathbf{E} = \mathbf{E}\_1 \oplus \mathbf{E}\_2 \oplus \dots \oplus \mathbf{E}\_\prime \tag{\mathbf{D}.8}$$

For other side, the linear Z-combinations of i, which are roots (the root system i) obviously shape a root system in Ei, whose Weyl group is the restriction to Ei, of all , , that is to say, the W-orbit of the root , if and only if , acts trivially in Ei.

**Proposition D. 3.** Let E' E. If the reflection , leave E' invariant then E', E' P, with P = {E() = 0}.

Note: Demonstrating this proposition is deduced immediately that every root fall in some of the spaces Ei, where

$$
\Phi = \Phi\_1 \cup \Phi\_2 \cup \Phi\_3 \cup \dots \cup \Phi\_{l^\vee} \tag{D.9}
$$

with 1 i *l*.

152 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

range *l* = 2:

or of big length.

simple roots). Example;

usual sense)25.

likewise, let

is clear that

**Def. D. 3.** A Coxeter graph of , is the graph of *l*, vertices which the ith-vertex is joined to the jth-vertex with i j, for <i, j> = <j, i>, edges. For example, to the root systems of

The Coxeter graph determine the numbers <i, j>, in the case of that all roots have the same length. Then <i, j> = <j, i>. In a more general case, where there is more of one different length (case of the root systems of range two; *G*2, *B*2), the graph fails to the concrete case in which is had a vertices pair that correspond to a simple short root, which cans be a long root

A relevant fact on the Coxeter graph is that these determine completely to Weyl group, essentially because these determine the orders of the direct products of generators of W.

**Proposition D. 2.** From the Table 1, of inner products and length of roots that change in the interval 0 <><> 4cos2, is had that the order of , in W, is respectively 2, 3, 4, 6, when = /2, /3 (or 2/3), /4 (or 3/4), /6 (or 5/6). [Note that, is the rotation through 2].

To the case when we have roots of short length, is ordinary to add an arrow to designate it in the Coxeter graph. This additional information help us to recover the integers of Cartan, called to resulting graph; Dynkin diagram of (as before, this depend on the number of

Somehow someone exist such component irreducibles like the system of roots shaped by a pair of roots. Let's study the component irreducible systems. Let's remember that a system , is irreducible if and only if, (or equivalently ) do not admit a division into two orthogonal proper subsets. Then by the correspondence between Coxeter graphs and root systems, is clear that , is irreducible if and only if their graph of Coxeter is connect (in the

In general is possible to give a number of connect components of the Coxeter graph,

the corresponding partition of , in mutually orthogonal subsets. If Ei, is the generated by i,

25 If this is non-connect it would be equivalent to that f, admits a partition into two orthogonal proper subsets.

<sup>123</sup> ,*<sup>l</sup>* (D. 7)

*B*2 :

G2 :

*Proof*. Let E' E. Then to any E', then E, and () = E', that is to say, (E') = E', but as E' E, then (E) = E, and(E') (E). But

$$\sigma\_a(\mathcal{E}) = \{ \emptyset \in \mathcal{E} \mid (\emptyset, \alpha) = 0 \} = \mathbb{P}\_{a\nu}$$

and since (E') = E', then

$$
\sigma\_a(\mathbf{E'}) = \mathbf{E'} \subset \sigma\_a(\mathbf{E}) = \mathbb{p}\_{a\nu}
$$

where E' P.

Then to every iEi, 1 i *l*, and i, is had that (D. 9).

**Proposition D. 4.** , is decomposed (univocally) as the union of irreducible root systems <sup>I</sup> (in subspaces Ei, of E) such that

$$\mathbf{E} = \mathbf{E}\_1 \oplus \mathbf{E}\_2 \oplus \dots \oplus \mathbf{E}\_n$$

*Proof*. One simple consequence of the before proposition, since if every Ei P(that is to say, iEi) with 1 i *l*, then to

$$
\sum\_{\gamma\_i \in \Delta\_i} < \alpha\_{i'} \beta > \gamma\_i = 0 (\Delta\_i \subset \Delta)\_{\prime} \tag{D.10}
$$

is had that <i, > = 0, thus E = E1 E*l*.

The before discussion establish a criteria of enougness to classify irreducible root systems or equively the Dynkin diagrams, this last, by the proposition that extends bijections between Euclidean root spaces to an isomorphism between corresponding root systems to said

Euclidean spaces, which are isomorphic in nature form. The isomorphism will be linearly represented by the Cartan matrix in every case.

**Theorem D. 1.** If , is a irreducible root system of range *l*, their Dynkin diagram is one of the following cases to *l*-vertices:


Note: The signed restrictions to the integer *l*, that is to say, *l n*, *n*Z, are considered only to not duplier before cases.

**Table 2.** Coxeter Groups and their Dynkin graphs.

**Appendix E** 

## **APPENDIX E: Definition of a Real Reductive Group**

Let G, be a semisimple Lie group. A real reductive group is a complex algebraic group defined on the real numbers, which is covering of an open subgroup of a group of real points (such real points are Cartan subalgebras of the Lie real reductive algebra).

Let GC, is a symmetric subgroup of GL(*n*, C), with real points and let the real connected component GR, of G<sup>C</sup> GL(*n*, C); a real reductive group G, is the finite covering of an open subgroup G0, of GR. Of this way, if p is the covering homomorphism defined for the map

$$\text{p: } G \to \text{G} \alpha \subset \mathbb{G}\_{\mathbb{R}} \tag{\mathbb{E} \ 1}$$

whose rule of correspondence is

154 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

**Group Dynkin Diagram**

represented by the Cartan matrix in every case.

the following cases to *l*-vertices:

*Al* (*l*  1)

*Bl* (*l*  2)

*Cl* (*l*  3)

*Dl* (*l*  4)

*E*<sup>6</sup>

*E*<sup>7</sup>

*E*8

*F*4

*G*<sup>2</sup>

duplier before cases.

**Table 2.** Coxeter Groups and their Dynkin graphs.

Euclidean spaces, which are isomorphic in nature form. The isomorphism will be linearly

**Theorem D. 1.** If , is a irreducible root system of range *l*, their Dynkin diagram is one of

Note: The signed restrictions to the integer *l*, that is to say, *l n*, *n*Z, are considered only to not

$$\mathbf{g} \mid \to \mathbf{p(g)}.\tag{\mathbb{E}\text{ }\mathcal{D}}$$

then explicitely a real reductive group is the space

$$\mathbf{G} = \{ \mathbf{g} = \mathbf{p}(\mathbf{g})^{\perp} \in \mathbf{G} \mid \mathbf{p}(\mathbf{g}) \in \mathbf{G} \boldsymbol{\alpha} \subset \mathbf{G}\_{\mathbb{R}} = \mathbf{G}\_{\mathbb{C}} \cap \mathbf{GL}\langle \boldsymbol{\eta}, \mathbf{C} \rangle \}. \tag{\mathbb{E}. 3)}$$

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

### **Exercises**

## **Exercises**


$$\mathbf{P}^+ = \{ \mathbf{[z]} \in \mathbf{CP} \\ \forall \; \\ \mathbf{(z, z)} = \mathbf{i} \\ \mathbf{r} \; \\ \forall \; \mathbf{r} \ge \mathbf{0} \\ \} \dots$$

This is a complex orbital submanifold of P. A compact maximal submanifold in P, is a copy of CP3, having complex dimension three. Demonstrate that the Penrose transform *P*, on the cohomological classes H3(P+, *L*), give the isomorphism of right fields:

where *L*, is a homogeneous bundle of lines in P, where k 3.


Use the diagram:

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

	- 8. Demonstrate that the aseveration of the exercise 6, is equivalent to the asseveration of that exist a only cohomological contour to the inner product and suggest a method to demonstrate that those contours are in effect, cohomological.
	- 9. Determine the Penrose transform on:
	- a. The quadric M, of dimension six and projective space P, of the 3- dimensional planes in M, with their corresponding Bernstein-Gelfand-Gelfand resolution to the canonical bundles on P.
	- b. On the flat semi-conformal complex spinorial 4-manifold .
	- c. On a complex line of the projective space P3(C), and of the bundle of lines (2).
	- d. On the co-cycles of the u-specialization of the space H0, q(G/L, V).
	- e. Whose integral cohomology of vector fields is H1(B, (4, 3)).

Here B, is the open orbit or minimal K-to the action of the group SU(1, 2), on B. This conform the space

B= L1 L2 L1, is defined positive and L2, have type ().

wth , a Hermitian form on C3, of type ().


$$\mathrm{H^{1}\$\_{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\scriptscriptstyle{\mathcal{\boldsymbol{\cdot}}}}}}}}}}}}}}}}}}}}}}}}}}}}}$$

and

$$\mathrm{H^{1}\_{\boldsymbol{\up}}(\mathrm{U}\prime\prime,\mathrm{O}(\mathrm{-4})) \to \mathrm{H^{4}}\_{\boldsymbol{\up}}(\mathrm{U}\prime\prime,\mathrm{C}) \cong \mathrm{C}\_{\prime}$$

cans be re-write using resolutions of <sup>n</sup>2\*, and <sup>n</sup>2, like

$$\operatorname{H}^{1}\_{\underline{\mathbf{z}}}(\operatorname{U}^{\prime\prime},\operatorname{O}(\operatorname{n}-\operatorname{2})) \to \operatorname{H}^{4}\_{\underline{\mathbf{z}}}(\operatorname{U}^{\prime\prime},\operatorname{\mathfrak{O}}^{\mathrm{n}-2}\operatorname{T}^{\ast}) \cong \operatorname{\mathfrak{O}}^{\mathrm{n}-2}\operatorname{T}^{\ast},$$

and

$$\operatorname{H}^{1}\_{\prescript{\cdot}{\cdot}}(\operatorname{U}^{\prime\prime},\operatorname{O}(\operatorname{n}-\operatorname{2})) \to \operatorname{H}^{4}\_{\prescript{\cdot}{\cdot}}(\operatorname{U}^{\prime\prime},\operatorname{\mathfrak{O}}^{\cdot\prime-2}\operatorname{T}) \cong \operatorname{\mathfrak{O}}^{\mathrm{n}-2}\operatorname{T}.$$


$$\text{cm} + \text{C}(\text{m} + 1, 2) + \dots + \text{C}(\text{m} + \text{p} - 1, \text{p})\dots$$


[Suggestion: Use topological arguments of connectivity and integration in chains to satisfy the Helmholtz theorem and other necessary results in the process of materialization of the universe like a fluid. Demonstrate that the universe have many degrees of circulation representing by cohomologies of vector fields isomorphic to contour cohomologies]


$$\lnot\!d\alpha, \beta \rhd \simeq \Big\{ d\alpha \land\_\* \beta \mathrel{=} \Big\} \alpha \land\_\* \delta \rhd \beta \color{=} \lnot\alpha, \delta \rhd \rhd$$


$$
\Box^2 \phi = 0,
$$

in the space RN+1.

158 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

9. Determine the Penrose transform on:

wth , a Hermitian form on C3, of type ().

11. Demonstrate that the integrals of twistor electrical field

H1 *L*

H1 *L*

cans be re-write using resolutions of <sup>n</sup>2\*, and <sup>n</sup>2, like

singularities of a complex space are poles of Cauchy.

of a submanifold of dimension m, have dimension

(U'', O(n 2)) H4

(U'', (n 2)) H4

H1 *L*

> H1 *L*

canonical bundles on P.

conform the space

by 0 C

frequency

and

and

demonstrate that those contours are in effect, cohomological.

b. On the flat semi-conformal complex spinorial 4-manifold .

d. On the co-cycles of the u-specialization of the space H0, q(G/L, V). e. Whose integral cohomology of vector fields is H1(B, (4, 3)).

8. Demonstrate that the aseveration of the exercise 6, is equivalent to the asseveration of that exist a only cohomological contour to the inner product and suggest a method to

planes in M, with their corresponding Bernstein-Gelfand-Gelfand resolution to the

a. The quadric M, of dimension six and projective space P, of the 3- dimensional

c. On a complex line of the projective space P3(C), and of the bundle of lines (2).

Here B, is the open orbit or minimal K-to the action of the group SU(1, 2), on B. This

B= L1 L2 L1, is defined positive and L2, have type ().

10. Which is the Penrose transform of the holomorphic succession of De Rham on P, given

theory: potentials module gauge and fields for both parts of positive and negative

*L*

*L*

*L*

*L*

13. Demonstrate using Radon transform of dimensions, that an osculating space of order p,

(U'', /C) H2

(U'', (4)) H4

12. Demonstrate through of classical functionals that in a contour cohomology, the

that include all the spaces of fields that are of interest in electromagnetic

(U'', C) C,

(U'', C) C,

(U'', n 2\*) n 2\*,

(U'', n 2) n 2.

	- 27. Using complex lines in the space P1(C), determine the isomorphic space to P1(C) whose realization in P5(R), is a null quadric. Whose null quadric represents a twistor hypersurface of the action of the Lie group Oo(5, 1), on P3(C).
	- 28. Demonstrate that a L2-curvature is a generalized curvature in a symmetrical homogeneous space with positive defined Hermitian form.
	- 29. Compute the Penrose transform on:
	- i. The cycles of the flag manifold of complex dimensional (2, 3).
	- ii. On holomorphic pencils of a surface of 2-dimensional Lobachevski.
	- iii. On a -curve of a analytic hypersurface of a complex Riemannian manifold of dimension 4.
	- iv. On the sheaf (T2**M**), of germs that are quadric forms in **M**.
	- v. Strings of a 10-dimensional p-brane.
	- Generalize the integral of the monopole

g(z) = (1/2i) f(z)/(z z0) dz,

to the case C2, considering circles S1, like orbits in C 2.


$$\mathbf{d} / \mathbf{d} \mathbf{t} (\mathbf{H} / \rho) = (\mathbf{H} / \rho \, \mathbf{\cdot} \, \mathbf{grad}) \, \mathbf{v}\_{\prime}$$

Use the orbits of the classes space given by G/C(T), with T, a complex torus.


$$\mathbf{H} \subset \mathbf{N} \subset \mathbf{G}.$$

Assume that G/H, and G/N, have G-invariant positive measures dgH, and dgN, respectively. Demonstrate that N/H, have a N-invariant positive measure dnH, the which (normalized adequately) satisfies

$$\int\_{\mathsf{C}\models\mathsf{H}} \mathsf{f}(\mathsf{g}\mathsf{H})\mathsf{d}\mathsf{g}\_{\mathsf{H}} = \int\_{\mathsf{C}\models\mathsf{N}} \Big(\int\_{\mathsf{N}\models\mathsf{H}} \mathsf{f}(\mathsf{g}\mathsf{n}\mathsf{H})\mathsf{d}\mathsf{n}\mathsf{n}\mathsf{}\mathsf{h}\mathsf{g}\_{\mathsf{N}}\Big)\mathsf{d}\mathsf{g}\_{\mathsf{N}}\Big)$$

fCc(G/H).

 Which are the orbital integrals to the formal integral of Feynman type given in a evaluating given by

$$\{\Gamma, \Omega\} = \sum\_{i} \int\_{\Gamma\_{\Gamma}^{0}(\nu)} \alpha(\Gamma) \rangle\_{i} = \int\_{\Gamma = \Gamma\_{1} \cup \Gamma\_{2}} \varphi = \int\_{\Gamma\_{1}} \varphi(e) + \int\_{\Gamma\_{2}} \varphi(e) = 0,$$

where j, are the jth-paths in the evaluation of the integral on Jacobi graphs space C (). , is the model of the graph used to describe the path or trajectory used by the electron *e*, with energy state (*e*).


dim Mt = Gdim Bdim(g)codim B0 (g),

where (g), is a measure of Borel positive define

$$\dim \mathbf{G} = \dim |\mathbf{B}| \aleph\_0^{\aleph\_0}$$

where 0, is the first infinite cardinal in set theory. B0 = B/**1**・, where

$$\mathcal{B} = \{ \mathcal{B}\_i \: \mid \: 0 \le \mu(\mathcal{B}i) \le 1 \},$$

41. Consider V*V*. Then

160 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

hypersurface of the action of the Lie group Oo(5, 1), on P3(C).

homogeneous space with positive defined Hermitian form.

i. The cycles of the flag manifold of complex dimensional (2, 3). ii. On holomorphic pencils of a surface of 2-dimensional Lobachevski.

iv. On the sheaf (T2**M**), of germs that are quadric forms in **M**.

29. Compute the Penrose transform on:

v. Strings of a 10-dimensional p-brane. Generalize the integral of the monopole

generalized orbits of D2.

adequately) satisfies

evaluating given by

fCc(G/H).

to the case C2, considering circles S1, like orbits in C 2.

dimension 4.

28. Demonstrate that a L2-curvature is a generalized curvature in a symmetrical

iii. On a -curve of a analytic hypersurface of a complex Riemannian manifold of

g(z) = (1/2i) f(z)/(z z0) dz,

cohomology on a real Riemannian manifold and whose tensor of curvature be the direct

materials of IIH,that satisfies /t + div(**v**) = 0. Demonstrate that the equation of the

d/dt(**H**/) = (**H**/ grad) **v**,

H N G. Assume that G/H, and G/N, have G-invariant positive measures dgH, and dgN, respectively. Demonstrate that N/H, have a N-invariant positive measure dnH, the which (normalized

G/Hf(gH)dgH = G/N(N/H f(gnH)dnH)dgN,

Which are the orbital integrals to the formal integral of Feynman type given in a

31. Give an example of a integral curvature on symmetrical spaces that conforms a L2-

vorticy currents take the form of the equation of the "frozen fields",

Use the orbits of the classes space given by G/C(T), with T, a complex torus.

34. Let G, be a Lie group; H, and N, closed subgroups such that

33. Calculate the integral on cycles of the hyperbolic 2-dimensional disc D2,that are

sum of regular images of co-cycles of the real Riemannian manifold in question. Let H = dvb + 2HIIH(**E**, ), be with b, Euclidean in M3,(is to say in R3) and with

27. Using complex lines in the space P1(C), determine the isomorphic space to P1(C)

whose realization in P5(R), is a null quadric. Whose null quadric represents a twistor

1. V = V, dim V< , (aC)\*,

2. there are q(aC)\*, such that Vj, is not vanishing and if Vis not vanishing then = j Qj, to some j q, and QL+.

Note: V{vV(H (H))kv = 0. t. s. kZ+, and Ha}, and the space L+, is the set of points:

L+ = {Set of the all integer combinations of non-negative elements of (a, g)},

Also the space *V*, is the category

*V* = {Category of finitely generated (g, 0M)-modules V}.

42. Consider the representation (, H), of integrable square such that gG, wHK, and vH, is satisfied

$$\|\mathbf{c}\| \lhd \pi(\mathbf{g}) \\ \text{v} \\ \text{w} \succ \text{\textquotedblleft d} \mathbf{g} < \infty, \text{\textquotedblleft}$$

Demonstrate that aA,

$$\|\cdot\|\!\| + \log |\,\|\,\mathbf{a}\,\|\,\|\,\mathbf{\gamma}^{\mathrm{d}-\mathrm{r}}\mathrm{d}\mathbf{a} < \infty$$

with (g)L2(G), if and only if r .

 Construct an intertwining integral operator to the representations IndMANG(1 eL 1), and ind A(P, P, 1, G), where L = /2, (, is a restrict root in L) with , such that (u p)Also

P P,

with P = MAN, and P = MANwith N =N, where , is an involution. G = G+, is the set of restrict roots of G.

44. Demonstrate that the operator of Szego type

$$\mathcal{S}: \operatorname{ind}\_{\mathsf{M}\to\mathsf{N}}\operatorname{\mathcal{C}}(\sigma\otimes\mathsf{e}^{\mathsf{q}\mathsf{L}}\otimes 1) \to \operatorname{ind}\mathcal{C}(\mathsf{V}^{\pi}) \quad \xleftarrow{\mathsf{c} \text{ind}\_{\mathsf{C}}(\mathsf{V}^{\mathsf{c}})} \quad \mathsf{ind}\_{\mathsf{C}}(\mathsf{C}\_{\mathsf{L}}^{\mathsf{c}}\otimes(\wedge\mathsf{u})\*),$$

should be given by on integral formula of the form

$$\text{L\,\text{Sf}(\text{x}) = \left| \begin{smallmatrix} \text{L} \ \text{L} \end{smallmatrix} \right| \text{L\,\text{s}}^{\text{-}} \text{L} = \text{l} \text{P} \text{[} \begin{smallmatrix} \text{I} \ \text{l} \end{smallmatrix} \text{L\,\text{s}}^{\text{-}} \text{L} = \text{l} \text{P} \text{[} \begin{smallmatrix} \text{I} \ \text{L} \end{smallmatrix} \text{L\,\text{s}}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}} \text{L}^{\text{-}}$$

With T : V0 V' C# <sup>s</sup> uan LM, map.


$$I(\mathbf{t}) = \lim\_{\underline{\mathbf{n}} \to \mathbf{r}} \mathbf{a}(\mathbf{n})^{\otimes \overline{\mathbf{r}}} \, \, \, \, \Xi \mathbf{r}(\underline{\mathbf{n}}) (1 - \mathfrak{b}(\log(\underline{\mathbf{n}}))^{\mathbf{r} \cdot \mathbf{q}} \, \underline{\mathbf{n}})$$

q > d + r, take the form

*I*(t) = NF a(n)FNF u(kmF(n)) dk(1 (loga(n))r - qdn.


Let , be a measure of positive type on H. If , is identified with their image under the canonical injection of H, in G. Show that , is of positive type as a measure on G. (For each function f*H*(G), show that

$$\iint\_{\mathbf{G}\times\mathbf{H}} \mathbf{f}(\mathbf{s}\mathbf{x}) \, \overline{\mathbf{f}(\mathbf{s})} \, \mathrm{d}\mathbf{m}\_{\mathbf{G}}(\mathbf{s}) \mathrm{d}\mu(\mathbf{x}) = \iiint\_{\mathbf{G}\times\mathbf{H}\times\mathbf{H}} \mathbf{h}(\mathbf{s}) \mathbf{f}(\mathbf{s}\mathbf{y}^{\cdot1}\mathbf{x}) \overline{\mathbf{f}(\mathbf{s}\mathbf{y}^{\cdot1})} \mathrm{d}\mathbf{m}\_{\mathbf{G}}(\mathbf{s}) \mathrm{d}\mathbf{m}\_{\mathbf{H}}(\mathbf{y}) \mathrm{d}\mu(\mathbf{x}) \, \mathbf{y}^{\cdot1}$$

where h*H*(G), is such that

162 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

A(1 + loga )d<sup>r</sup>

 Construct an intertwining integral operator to the representations IndMANG(1 eL 1), and ind A(P, P, 1, G), where L = /2, (, is a restrict root in L) with , such that (u

P P, with P = MAN, and P = MANwith N =N, where , is an involution. G = G+, is the set of

*S*f(x) = L/LMAN'(*l*)[Tf(x*l*)]d*l* = <sup>L</sup><sup>K</sup>'(*l*)[Tf(x*l*)]d*l*,

*I*(t) = NF a(n)<sup>F</sup> F(n)(1 (log(n))r - qdn,

47. Use the Harish-Chandra function to demonstrate that if (, H), is of square integral then

48. Let K, be a representation of P/0M = AN, in V/nkV/0m. Let IK : V/nkV V/nkV, be and we

49. Which are the *D*-modules that agree with the representations given by images under

51. Let G, be a (separable, metrizable) locally compact group and H, a closed subgroup of G. Suppose that G, and H, are unimodular, and let mG, and mH, denote Haar measures

50. How could be related the Radon transform on *DG/H*-equivariant modules with the

K = TK. Demonstrate that kerTK, contains to nkV.

uan LM, map.

*I*(t) = NF a(n)FNF u(kmF(n))

45. From the before problem (problem 44), What happens when M is compact? 46. Use the definition of the Harish-Chandra function , to demonstrate that the finite

da < ,

) indLG(C#

dk(1 (loga(n))r - qdn.

<sup>s</sup>

u

Demonstrate that aA,

p)Also

restrict roots of G.

With T : V0 V' C#

integral

q > d + r, take the form

consider IK(V/nkV) = I

Penrose transform?

Penrose-Ward transform?

on G, and H, respectively.

with (g)L2(G), if and only if r .

44. Demonstrate that the operator of Szego type

*S* : indMANG( eL 1) indLG(V'

should be given by on integral formula of the form

<sup>s</sup>

(, H), satisfies the strong inequality.

$$\int\_H \mathbf{h}(\text{su}) \text{dm}\_H(\mathbf{u}) = 1,$$

52. Let H*<sup>n</sup>*, , denote the space of automorphic forms f, on P, which are holomorphic on P, and such that for each , we have

$$\mathbf{f}(\mathbf{z}) = \mathbf{J}\_{\gamma}(\mathbf{z})^{\text{u}} \mathbf{f}(\gamma \bullet \mathbf{z})\_{\nu}$$

where J(z) = (cz + d)1, if

$$
\lambda = \begin{pmatrix} \mathbf{a} & \mathbf{b} \\ \mathbf{c} & \mathbf{d} \end{pmatrix}'
$$

Since z = () z, we cannot have f = 0, unless *n* = 2*k*, is even; in which case f, is said to be an automorphic form of weight *k*, relative to . Let f, be any holomorphic function on P, and F = fz0. Show that for each compact subset A, of G, there exists a compact neighborhood B, of A, in G, such that

$$\sup\_{s \in \mathcal{A}} \left| \mathcal{F}(\mathbf{s}) \right| \leq \mathcal{M} \int\_{\mathcal{B}} \left| \mathcal{F}(\mathbf{s}) \right| \mathrm{d}m\_{\mathcal{G}}(\mathbf{s}) \,\prime \,\, \mathcal{A}$$

Where M, is a constant independent of f. (Apply Cauchy's formula to f, and the projection of A, on G/K = P). Deduce that if FH1(*n*), the family (sups<sup>A</sup>f(s)), is summable (use the last that BB , is finite). Consequently the family {f(s)}, is absolutely summable for each sG, and the function

$$\mathbb{F}\_{\Gamma}(\mathbf{s}) = \sum\_{\gamma \in \Gamma} \mathcal{F}(\gamma \mathbf{s})\_{\gamma}$$

is of the form fz0, where f, is an automorphic form of weight *k*, on P, which is holomorphic in P.

53. Let G, be a (separable, metrizable) locally compact commutative group. For each continuous unitary representation U, of G, on a separable Hilbert space E, show that there exists a unique representation *L*, of the involutory Banach algebra <sup>B</sup> <sup>C</sup>(G ) (with thse usual product) on E, such that *L*(<s, >) = U(s), for all sG.

	- 54. Let (G, K), be a Gelfand pair and let g, be a function belonging to C (K\G), such that for each fH (K\G/K), there exists a complex number f, such that f \* g = fg. If g(s0) 0, for some s0G, show that the function

$$\mathbf{\color{red}{oo(s) = g(s\_0)^{-1}}} \int \mathbf{g(sts\_0)dm\_K(t)\_{/s}}$$

is a spherical function.

55. Let G, be an amenable group, let K, be a compact subset of G, and M, a real number major that 0. Show that for each > 0, there exists a function f 0, belonging to LC 1(G), such that N1(f) = 1, and such that, for each function gLC 1(G), which is zero on K, and satisfies N1(g) M, we have

$$\text{N}\_1(\text{(g }\text{\*}\text{)} - (\text{]g}(\text{x})\text{dx}\alpha|\text{t}) \le \varepsilon\_r$$

where , is Haar measure on G.


$$E\_{\rm f}(\mathbf{a}) = \mathbf{e}^{\wp(\log \mathbf{a})} \int\_{\rm N} \mathbf{f}(\mathbf{an}) d\mathbf{n} \,\prime$$

fL#(G), such that

$$\int\_{G} \Phi\_{\mathsf{V}}(\mathbf{x}) \mathbf{f}(\mathbf{x}) d\mathbf{x} = \int\_{G} \mathbf{e}^{\mathrm{iv}(\log \mathsf{a})} \mathbf{F}\_{\mathsf{f}}(\mathbf{a}) d\mathsf{a},$$

Verify the following assertations:


iii. The mapping f I Ff (fL#(G)) is one-to-one.

Note: L#(G) = {fL1(G)f are bi-invariant under K}.

 Let G, be a Lie group, H and N closed subgroup such that H N G. Assume that G/H, and G/N, have positive G-invariant measures dgH, and dgN. Show that N/H, has an N-invariant positive measure dnH, which (Suitable normalized) satisfies

$$\int\_{G/H} \mathbf{f}(\mathbf{g}H) \mathbf{dg}\_H = \int\_{G/N} \left( \int\_{N/H} \mathbf{f}(\mathbf{gn}H) \mathrm{dn}\_H \right) \mathrm{dg}\_{N'} $$

fCc(G/H).

59. Let G, be a semisimple connected compact Lie group and let T, be a maximal torus in G. Let w denote the order of the Weyl group of G, and let

$$D(\mathbf{t}) = \prod\_{\alpha \in \Lambda^{+}} 2 \sec \left( \frac{1}{2} \alpha(\mathbf{i} \mathbf{H}) \right),$$

if t = expHT. Let dt, and dg, respectively denote the invariant measures on T, and G, normalized by

$$\int\_{\mathbb{T}} \mathrm{d}\mathbf{t} = \int\_{\mathcal{G}} \mathrm{d}\mathbf{g} = 1,$$

Derive Weyl's formula

$$\int\_{\mathcal{G}} \mathbf{f}(\mathbf{g})d\mathbf{g} = \frac{1}{\mathbf{w}} \int\_{\mathbb{T}} \left| D(\mathbf{t}) \right| \mathbf{d} \mathbf{t} \int\_{\mathcal{G}} \mathbf{f}(\mathbf{g} \,\mathrm{tg}^{-1}) \mathrm{dg}, \,\, \mathbf{t} \right|$$

fC(G).

1(G),

1(G), which is zero on K, and

164 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

such that N1(f) = 1, and such that, for each function gLC

some s0G, show that the function

satisfies N1(g) M, we have

where , is Haar measure on G.

G, has finite center. Let

Verify the following assertations:

i. Ff g = Ff Fg, f, gL#(G), ii. Ff\* = (Ff)\*, fL#(G),

iii. The mapping f I Ff (fL#(G)) is one-to-one. Note: L#(G) = {fL1(G)f are bi-invariant under K}.

56. Demonstrate that H1, 0[

fL#(G), such that

fCc(G/H).

is a spherical function.

54. Let (G, K), be a Gelfand pair and let g, be a function belonging to C (K\G), such that for each fH (K\G/K), there exists a complex number f, such that f \* g = fg. If g(s0) 0, for

> -1 0 0K (s) g(s ) g(sts )dm t( ),

55. Let G, be an amenable group, let K, be a compact subset of G, and M, a real number major that 0. Show that for each > 0, there exists a function f 0, belonging to LC

N1((g \* f) – (g(x)dxf) ,

57. Let (G, K), be a Riemmanian symmetric pair of the non-compact type and assume that

log a) <sup>f</sup> <sup>N</sup> E (a) e f(an)dn,

G G

P1/ ](P1, *O*(*n*)), is not a Verma module to *<sup>n</sup>* 2.

i (log a) f

H H N

f(gH)dg f(gnH)dn dg , 

(x)f(x)dx e F (a)da,

 Let G, be a Lie group, H and N closed subgroup such that H N G. Assume that G/H, and G/N, have positive G-invariant measures dgH, and dgN. Show that N/H, has

an N-invariant positive measure dnH, which (Suitable normalized) satisfies

G/H G/N N/H

 Let V, be a n-dimensional vector space over a non-archimedean local field F, for instant the field of p-adic numbers. Let : V V, be a linear endomorphism with distinct eigenvalues in an algebraic closure of F. The centralizer I, of , is of the form

$$\mathcal{I}\_{\gamma} = E\_1^{\ast} \times \dots \times E\_{n'}^{\ast}$$

where *E*<sup>1</sup> , , *E*<sup>r</sup> , are finite separable extensions of F. This is a commutative locally compact topological group. Let *O*F, denote the ring of integers in F. We consider the set of lattices of V, that are sub-*O*F-modules *V* V, of finite type with maximal rank. We are interested in the subset *M*, of lattices *V*, of V such that (*V*) V. The group I, acts the set *M*. This set is infinite in general but the set of orbits under the action of I is finite. We fix a Haar measure dt, on the locally compact group I. We consider a set of representatives of orbits of I on *M*, and for each x in this set, let denote I,x, the compact open subgroup of I, of elements stabilizing x. Demostrate that the finite sum

$$\mathcal{I} = \sum\_{\mathbf{x} \in \mathcal{M}\_{\mathbb{Y}}/\mathbf{I}\_{\mathbb{Y}}} \frac{1}{\mathrm{Vol}(\mathbf{I}\_{\mathbb{Y}, \mathbf{x}}, \mathbf{dt})},$$

is a typical orbital integral.


$$\mathcal{N}\_{\mathbf{f}} = (\mathbf{\tilde{f}} \ast \alpha)(\mathbf{c}) = \int\_{\mathcal{G}} \mathbf{f}(\mathbf{s}) \alpha(\mathbf{s}) d\mathbf{m}\_{\mathcal{G}}(\mathbf{s}) = \mathbf{\tilde{f}}, \alpha > \mathbf{\tilde{f}}$$

	- 63. The product of two functions of positive type on G, is a continuous function of positive type.
	- 64. Let G, an unimodular locally compact group (separable and metrizable) and K, a compact subgroup of G. Let mK, be the Haar measure on K, with total mass equal to 1. If we put

$$\mathbf{f}^{\#}(\mathbf{s}) = \iint\_{\mathbf{K} \times \mathbf{K}} \mathbf{f}(\mathbf{t} \mathbf{t} \mathbf{t}') \mathrm{d}m\_{\mathbf{K}}(\mathbf{t}) \mathrm{d}m\_{\mathbf{K}}(\mathbf{t}'),$$

For all function fK (G), the mapping f ∣ f #, is a projector on the vector space C (G), onto the vector space C (K\G/K). Let prove fC (K\G/K), and gC (G), that

$$(\mathbf{f}\mathbf{g})^\circ = \mathbf{f}\mathbf{g}^\circ.$$

65. Let K R*n*, be a symmetric compact convex set having 0, as an interior point. Let V, denote the Lebesgue measure of 2K = K + K. Show that if 0, is the only point of Z*n*, which belongs to 2K, then

$$\mathfrak{D}'' = \mathbf{V} + \frac{1}{2^{2n}\mathbf{V}} \sum\_{\substack{m \in \mathbb{Z}^n \\ m \neq 0}} \left| \int\_{\mathcal{K}} \mathbf{e}^{\cdot 2\pi i(\mathbf{x})m\mathbf{i}} d\mathbf{x} \right|^2 \dots$$

(Show that Poisson's formula (with G = R*n*, and H = Z*n*) may be applied to the function f = <sup>K</sup> K). Hence give another proof of Minkowski theorem.

66. Considering the Cartan matrix

$$
\begin{bmatrix}
2 & -1 & 0 & 0 \\
0 & -1 & 2 & -1 \\
0 & 0 & -1 & 2
\end{bmatrix}'
$$

of the root system *F*4, let obtain the graph if < .

67. Let define the equivalences given by the Penrose-Ward transform of the twistor correspondence

$$
\mathbb{C}^4 \times \mathbb{P}^1
$$

$$
\mathbb{L}
$$

$$
\mathcal{P}
$$

$$
\mathbb{P}^1 \mathbb{L}
$$


70. Let a *>*0, b > 0, and let f, be the continuous real-valued function defined on R, which is equal to b, when x = 0, is zero for x a, and is linear in each of the intervals [a, 0], and [0, a]. We have

$$\begin{pmatrix} \mathsf{F} & \mathsf{f} \end{pmatrix} \mathbf{(t)} = \frac{\mathsf{b} \sin^2 \pi \mathbf{at}}{\pi^2 \mathbf{at}^2} \mathbf{\pi}$$

Deduce that if

$$\mathbf{g}(\mathbf{x}) = \sum\_{\mathbf{n}=-N}^{N} \mathbf{f}(\mathbf{x} + \mathbf{n})\mathbf{y}$$

Then

166 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

#

K). Hence give another proof of Minkowski theorem.

of the root system *F*4, let obtain the graph if < .

68. What about topological B-model and twistor strings?

<sup>3</sup> *P* <sup>2</sup>

type.

we put

which belongs to 2K, then

66. Considering the Cartan matrix

correspondence

69. What is a quadric in *P* <sup>2</sup>

63. The product of two functions of positive type on G, is a continuous function of positive

K K K K f s f(tst')dm (t)dm (t') ( ) ,

For all function fK (G), the mapping f ∣ f #, is a projector on the vector space C (G),

(fg)# = fg#.


, 4

C

2

65. Let K R*n*, be a symmetric compact convex set having 0, as an interior point. Let V, denote the Lebesgue measure of 2K = K + K. Show that if 0, is the only point of Z*n*,

> 2 K <sup>1</sup> V e dx

 

0

(Show that Poisson's formula (with G = R*n*, and H = Z*n*) may be applied to the function f = <sup>K</sup>

2 10 0

0 0 12

67. Let define the equivalences given by the Penrose-Ward transform of the twistor

*P*

12 20 , 0 12 1

14

3? Which are their orbits?

PC

Z

2 . *<sup>n</sup> m m*

2 V

*<sup>n</sup> <sup>n</sup>*

onto the vector space C (K\G/K). Let prove fC (K\G/K), and gC (G), that

compact subgroup of G. Let mK, be the Haar measure on K, with total mass equal to 1. If

64. Let G, an unimodular locally compact group (separable and metrizable) and K, a

$$(\sf F \quad \text{g})(\mathbf{t}) = \frac{\mathbf{b} \sin^2 \pi \mathbf{a} \mathbf{t}}{\pi^2 \mathbf{a} \mathbf{t}^2} \frac{\sin(2\mathbf{N} + \mathbf{l}) \pi \mathbf{t}}{\sin \pi \mathbf{t}}.$$

Let give the Dynkin diagram for a complex simple Lie algebras:

a. so(7, C),



Note: F123, is the full flag manifold of lines inside 3-dimensional subspace in C4.

73. Consider the case G = SU(2, 2), acting on the open orbit M+. The holomorphic discrete series are those which are realizable as holomorphic sections of certain homogeneous vector bundles over M+. Let obtain the representation on L2-holomorphic 4-forms space, where L2, is defined with respect to the manifolds invariant inner product

$$<\langle \alpha, \eta \rangle := \int\_{\mathbb{M}^+} \alpha \wedge \overline{\eta}.$$


Is the quadric in CP6, defined by

$$\mathbf{Q(x) = x\_0 2 + \chi\_1 2 + \dots + \chi\_6 2 = 0} \text{ .}$$

where Q(x), is a five dimensional complex projective manifold which is a complexification of the five dimensional sphere.

76. From the two isomorphisms generates by the Penrose transform

$$\begin{array}{c} \text{H}^{1}(\text{b}^{+}, \xleftarrow{\text{k}} \xrightarrow{\text{b}} \text{\raisebox{-1.2pt}{\text{k}}} \text{\raisebox{-1.2pt}{\text{k}}}) \cong \text{ker}\{\text{M}^{+}, \xleftarrow{\text{k}\cdot\text{2}} \xrightarrow{\text{b}\cdot\text{k}+1} \text{\raisebox{-1.2pt}{\text{k}}} \text{\raisebox{-1.2pt}{\text{k}}}, \\\\ \xleftarrow{\text{k}\cdot\text{b}\cdot\text{3}} \xrightarrow{\text{k}} \text{\raisebox{-1.2pt}{\text{k}}} \text{\raisebox{-1.2pt}{\text{k}}}, \end{array}$$

and

$$\begin{array}{c} \text{H}^{1}(\text{p}\text{-}\text{\*}\text{-}\text{\*} \xrightarrow{\text{p}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\\\\ \xleftarrow{\text{q}+\text{n}+1}\qquad\xleftarrow{\text{p}-\text{q}-\text{q}-\text{n}-\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-}\text{-$$

where k – 4 b + c, and n 0, let deduce an isomorphism for the left-hand sides:

*T* : H1(P+, ) H1(P\* , ),

where M\* , is the component of the space-time to left-handed fields (potentials modulo gauge). M+, is canonically isomorphic to M\* .

"The torsion of the space-time begins in the classes of the homogeneous space G/C(T), under the twistor transform *T*, which are orbits of the sided handed-fields in M."


Note: The space StC, is a ∞-category of stable presentable C-linear ∞-categories with morphisms given by continuous (colimit-preserving) functors.


168 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

76. From the two isomorphisms generates by the Penrose transform

, ) ker{M

, ) ker{M\*

*T* : H1(P+, ) H1(P\*

a triplet edges is the Coxeter graph G2, .

, and nontrivially on V+, at the fixed point.

80. For any h, with image []h/W, we have the following:

dualizable Calabi-Yau algebras in StC.

gauge). M+, is canonically isomorphic to M\*

where k – 4 b + c, and n 0, let deduce an isomorphism for the left-hand sides:

the twistor transform *T*, which are orbits of the sided handed-fields in M."

 .

"The torsion of the space-time begins in the classes of the homogeneous space G/C(T), under

77. Let demonstrate that unique connect graph , of an admissible set U, which can include

78. Let demonstrate that group G, lies in the subgroup SU(2) SO(4), which acts trivially

79. Let Z, be the twistor space of the conformally anti-self-dual manifold (X, g), and Z, is the twistor space of the orbifold (X, g). The following vanishing theorem is the key to

a. The -twisted and -monodromic Hecke categories HG, , and HG, , are self-dual, fully

the structure of these two complex spaces. Let prove H1(Z, *O*(1)) = 0.

Q(x) = x02 + x12 + + x62 = 0, where Q(x), is a five dimensional complex projective manifold which is a complexification of

+

, is the component of the space-time to left-handed fields (potentials modulo

},

},

,

,

, ),

Is the quadric in CP6, defined by

the five dimensional sphere.

H1 (P +

H1 (P\*

and

where M\*

on V


85. a). Let ∈ h\*. Let Q, be a K-orbit in X (flag variety of g), and , an irreducible Khomogeneous connection on Q, compatible with + . Then the direct image of , with respect to the inclusion Q → X, is the standard Harish-Chandra sheaf *I*(Q, ). Since , is holonomic, *I*(Q, ), is also a holonomic D-module and therefore of finite length. This implies that their cohomologies Hp(X, *I*(Q, )), p <sup>∈</sup> Z+, are Harish-Chandra modules of finite length [12]. Let calculate these cohomology modules in terms of "classical" Zuckerman functors.

b). Fix x∈Q. Denote by bx, the Borel subalgebra of g, corresponding to x, and by *S*x, the stabilizer of x, in K. Then the geometric fiber Tx(), of , at x, is an irreducible finitedimensional representation , of *S*x. We can view it as an *S*x-equivariant connection over the *S*x-orbit {x}. Therefore, we can consider the standard Harish-Chandra sheaf *I*() = *I*({x}, ). It is an *S*x-equivariant D-module. Then let demonstrate that

$$\Gamma\_{\mathbb{K},S\_{\mathfrak{s}}}^{\otimes \otimes 0}(D(I(\mathfrak{o}))) = D(I(Q,\mathfrak{r})) [-\dim \mathcal{Q}].$$

 Let ∈h\*, Q, a K-orbit in X (flag variety of g), and , an irreducible K-homogeneous connection compatible with + . Let x∈Q, and let *S*x, be the stabilizer of x, in K. Let , be the representation of *S*x, in the geometric fiber Tx(). Then we have

$$\mathbf{H}^{\mathbb{P}}(\mathsf{X}, \mathsf{I}(\mathsf{Q}, \mathsf{T})) = \mathsf{R}^{\mathbb{P} + \dim \mathsf{Q}} \mathsf{T} \mathsf{x}\_{\mathsf{Z}^{\mathbb{Q}}}(\mathsf{M}(\mathsf{o})), \quad \forall \ \mathsf{p} \in \mathbb{Z}.$$


$$\mathbf{H}(\mathbf{K}^{\mathbf{A}\mathbf{A}^\*}) = -\underset{\neq}{\text{d}} \oint \mathbf{K}^{\alpha\beta} \mathbf{o}\_{\alpha}^{\mathbf{A}} \mathbf{d} \pi\_{\beta}^{\mathbf{A}^\*} \wedge \mathbf{d} \mathbf{x}\_{\mathbf{A}\mathbf{A}^\*}$$

This expression depends on KAA', and their decomposition into spinors. Let demonstrate that using d(iA'dxAA') = d2A = RBAB, we have the original definition of Penrose integral of line given in field theory.

92. Having that K G, the compact component of G,

$$\mathcal{F}\_{\mu}(\mathbf{g}) = \int\_{\mathcal{K}} \mathbf{a}(\mathbf{K}\mathbf{g})^{\mu+\rho} \, \mathrm{d}\mathbf{k} \,\prime \,\forall \mathbf{g} \in \mathcal{G},$$

Demonstrate that F(g), is holomorphic in .

93. Let H, be the (p, 0M)-module H, with a, acting for ()I, and n, acting for 0. Let V, be the (g, K)-module. Let THomg, K(V, H), such that T (v) = T(v)(1), vV. Prove that T Homp, 0M(V/nV, H).


K, x

images on SU(2, 2)-orbits.

nonvanishing section , of KX

of SO(4, C).

the -parts of the

line given in field theory.

that T

( ( ( ))) ( ( , ))[ dim ]. <sup>Q</sup> *geo <sup>S</sup> D DQ I I*

connection compatible with + . Let x∈Q, and let *S*x, be the stabilizer of x, in K. Let ,

Hp(X, *I*(Q, )) = Rp+dimQK,*<sup>S</sup>*x(M()), pZ.

88. Which is the orbitalization process in non-concurrets integrals? Let investigate the non-

90. Let (X, ω, J), be a compact Kähler manifold, and let H be a complex hypersurface in X representing twice the anticanonical class. Then the complement of H carries a

\ H carries a special Lagrangian foliation whose lift to the Calabi-Yau double cover Y can be perturbed to a Z/2-invariant special Lagrangian torus fibration. Let construct an isomorphism between the foliations mentioned above and spaces in field theory inside

91. For Penrose's quasi-local mass construction [Penrose 1982] the quasi-killing vectors are constructed out of four linearly independent solutions of the twistor equation

= iA'dxAA'

generates deformations of H, with boundery value KAA', on *j*, is obtained by inserting

AA' A' H(K i K d dxAA' ) , *j*

This expression depends on KAA', and their decomposition into spinors. Let demonstrate that using d(iA'dxAA') = d2A = RBAB, we have the original definition of Penrose integral of

> <sup>K</sup> F g a(Kg) dk g G, ( ) ,

93. Let H, be the (p, 0M)-module H, with a, acting for ()I, and n, acting for 0. Let V,

 

*j*

*j*

2, with poles along H. Let H, For a suitable choice of H, X

, is a matrix of constants and A', are

. The value of the Hamiltonian that

(v) = T(v)(1), vV. Prove

A =

Let ∈h\*, Q, a K-orbit in X (flag variety of g), and , an irreducible K-homogeneous

87. Let find a relation between torsion (like field observable) and the twistor transform

be the representation of *S*x, in the geometric fiber Tx(). Then we have

concurrent integral concept inside the orbitalization theory. 89. What are the horo-spheres? What are the horocycles? Let explain.

(0A, , 3A). They are given by KAA'= K = K()

92. Having that K G, the compact component of G,

be the (g, K)-module. Let THomg, K(V, H), such that T

Demonstrate that F(g), is holomorphic in .

Homp, 0M(V/nV, H).

A, defined by dA

this decomposition KAA', into the Witten-Nester integral

96. The various tempered series exhaust enough of G , for a decomposition of L2(G), essentially as

$$\sum\_{\mathbf{H}\in\operatorname{Har}(\mathbf{G})} \sum\_{\boldsymbol{\upupup}\otimes\mathbf{e}^{\mathbf{v}}\in\mathbf{\tilde{T}}} \int\_{\mathbf{\tilde{A}}} \operatorname{H}\_{\pi\_{\boldsymbol{\upupup},\boldsymbol{v},\boldsymbol{\upupup}\boldsymbol{\upupup}}} \otimes \operatorname{H}^{\*}\_{\pi\_{\boldsymbol{\upupup},\boldsymbol{v},\boldsymbol{v},\boldsymbol{\upup}}} \operatorname{m}(\mathbf{H}:\boldsymbol{\upupup}:\mathbf{v}:\boldsymbol{\upupright}\mathbf{c}) \operatorname{d}\boldsymbol{\upupup}\sigma,\mathbf{v}$$

Prove it! Here m(H: )d, is the Plancherel measure on G .


$$\mathbf{g(x,y) = (gx, \sigma(g)y), \forall \ \mathbf{x}, \mathbf{y} \in \mathcal{X} \times \mathcal{X}\_{\sigma}$$

where , is the involution and gG.

	- i. The corresponding orbits to d, ', whose dimensions are d(), d('), where ', is an irreducible representation of a real form and , is the Langlands parameter26 (P( G )).

$$
\phi \colon \mathbf{w}\_k \to \, ^\vee \mathbf{G}^\mathbf{1},
$$

compatible with the maps into , and such that (C ), is formed of semi-simple elements:

$$\mathbf{w}\_{\mathbf{k}} \mathbf{-} \mathbf{C}^{\mathbf{x}} \times \{\mathbf{i}, \mathbf{j}\}\_{\mathbf{\prime}} \qquad \|\mathbf{z}\mathbf{j}\|^{-1} = \underline{\mathbf{z}} \mathbf{\prime} \qquad \mathbf{j}^{\mathbf{z}} = -1. \ \mathbf{z}$$

<sup>26</sup> **Def.** A Langlands parameter is a group homomorphism

ii. The even orbit *O* <sup>l</sup>, and a K G-orbit *O*<sup>R</sup> <sup>l</sup>, in *O* <sup>l</sup> ( s l), such that

$$\left[\mathfrak{D}^{\wedge \mathbb{R}} \lrcorner \iota \lrcorner \left[\mathfrak{D}^{\wedge \mathbb{R}} \lrcorner + \left(\ulcorner \mathfrak{n} \lrcorner \smile \mathfrak{s}\right)\right] \right] = \mathfrak{D}^{\vee \mathbb{R}} \lrcorner \iota + \left(\ulcorner \mathfrak{n} \lrcorner \smile \mathfrak{s}\right).$$

101. Whih is the orbit in the integral transform of dimensions?

$$\dim \mathbf{F} = \int\_{\mathbf{B}} \dim \mathbf{B}^{\dim \chi(\mathbf{g})} \mathbf{c} \cdot \operatorname{codim } \mathbf{B}^0 \mu \Lambda(\mathbf{g}).$$

Here B, is a hypersurface of the C2, (g), their index of representation induced from the corresponding vector bundle of lines and (g), is a special measure through of the fractal measure followed from the vector bundle measure on lines.

 The hypercohomology of E , of E (D), computes the cohomology of the local system *H*, on M, in other words

$$\mathsf{HP}(\mathsf{M}, \mathsf{M}^{\uparrow}) \cong \mathsf{HP}(\underline{\mathsf{M}}, \mathsf{\mathcal{E}}^{\bullet}(\log \mathsf{D})) \cong \mathsf{HP}(\underline{\mathsf{M}}, \mathsf{\mathcal{E}}^{\bullet}(\mathsf{c} \mathsf{D})) \amalg \mathsf{HP}(\mathsf{M}, \mathsf{\mathcal{E}}^{\bullet}) . \mathsf{L}$$

Prove it!

103. Let Sa, b(G), be the space of all fC(G), such that pa, b, x, y, r(f) < , for all x, y, r, endowed with the topology given by the above semi-norms. If K , and if fC(G), then we set

$$E\_{\gamma} \mathbf{f}(\mathbf{g}) = \mathbf{d}(\gamma) \int\_{\mathbb{K}} \chi\_{\gamma}(\mathbf{k}) \mathbf{f}(\mathbf{k}^{-1} \mathbf{g}) d\mathbf{k},$$

Demonstrate Sa, b(G), is a Fréchet space.

Note: If fC(G), then we set for r 0, x, yU(g)

$$\mathbf{p}\_{\mathbf{a},\mathbf{b},\mathbf{x},\mathbf{y},\mathbf{r}}(\mathbf{f}) = \sup \mathbf{p}\_{\mathbb{R}^{\mathbf{a}}} \mathbf{c} \mathbf{a}(\mathbf{g}) \mathbf{b}(\mathbf{g})^{-1} |\mathbf{L}(\mathbf{x})\mathbf{R}(\mathbf{y})\mathbf{f}(\mathbf{g})|.$$

**Technical Notation** 

## **Technical Notation**


172 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

<sup>l</sup>, and a

101. Whih is the orbit in the integral transform of dimensions?

measure followed from the vector bundle measure on lines.

Hp(M, *H* ) Hp(M, E

*O*R<sup>l</sup> [*O*R<sup>l</sup>+ (

K

B

, of E

with the topology given by the above semi-norms. If K

G-orbit *O*<sup>R</sup>

dim (g) 0

(logD)) Hp(M, E

103. Let Sa, b(G), be the space of all fC(G), such that pa, b, x, y, r(f) < , for all x, y, r, endowed

K E f(g) d( ) (k)f(k g)dk,

pa, b, x, y, r(f) = supgGa(g)r

u

dimF dim B codim B g .

Here B, is a hypersurface of the C2, (g), their index of representation induced from the corresponding vector bundle of lines and (g), is a special measure through of the fractal

<sup>l</sup>, in *O*

s)] = *O*R<sup>l</sup>+ (

<sup>l</sup> (

u s).

(D), computes the cohomology of the local system

(D)) Hp(M, E

b(g)L(x)R(y)f(g).

).

, and if fC(G), then we set

s

l), such that

ii. The even orbit *O*

The hypercohomology of E

*H*, on M, in other words

Demonstrate Sa, b(G), is a Fréchet space.

Note: If fC(G), then we set for r 0, x, yU(g)

Prove it!


Homg, K(V, W) – Space of (g, K)-invariant homomorphisms that go from module V to the module W.

m – Corresponding algebra of the Lie subgroup M of the Lie Group G.

m = {haFAd(g)h = h if and only if [h, g] = 0}.

n – Nilpotent algebra.

H – Cartan subgroup of the Lie group.

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

l – Levi algebra. Algebra used in the Levi decomposition of Lie algebra g, corresponding to the Lie group G.

G0 – Space which arise of the identification G0 = 0(G0), that is to say the space of points {gGAd(g) = I, AdEnd(G)}.

GR– Component of real points of the real reductive Lie group *G* (open subgroup of *G*).

G<sup>C</sup> – Analytic Lie group.

F – Simple roots subspace of 0.

C() –Weyl camera of simple roots space of .

L – Levi group.

X(G) – Space of continuous homomorphism of G in the multiplicative group R\* = (R, )/{0}.

M0 – Identity component of Lie Group M. Space of the points: {mM Ad(m) = I, AdEnd(M)}

 - Tensor product of modules belonging to a associative ring endowed of the tensor product to their elements.

M – Connected component of a Lie group G.

Reflection.

<sup>G</sup> Invariant integration on the group *G*.

N – Nilpotent component of the Lie group G. Also is the normal subgroup of G, when N is the normalizator of G.

NF - Nilpotent component of the Lie group G, restricted to the simple roots subspace F 0.

NF – Compact nilpotent component of the Lie group G, restricted to the simple roots subspace F 0.

N – Compact nilpotent component of the Lie group G. Also is the normal subgroup of G, when N is the normalizator of G.

P – Parabolic subgroup of Lie group G.

PF –Parabolic subgroup restricted to the subspace F, of the simple roots of 0

Cl() – Closure.

174 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

the Lie group G.

{gGAd(g) = I, AdEnd(G)}.

F – Simple roots subspace of 0.

C() –Weyl camera of simple roots space of .

M0 – Identity component of Lie Group M.

M – Connected component of a Lie group G.

<sup>G</sup> Invariant integration on the group *G*.

Space of the points: {mM Ad(m) = I, AdEnd(M)}

G<sup>C</sup> – Analytic Lie group.

product to their elements.

L – Levi group.

Reflection.

the normalizator of G.

when N is the normalizator of G.

P – Parabolic subgroup of Lie group G.

subspace F 0.

l – Levi algebra. Algebra used in the Levi decomposition of Lie algebra g, corresponding to

G0 – Space which arise of the identification G0 = 0(G0), that is to say the space of points

GR– Component of real points of the real reductive Lie group *G* (open subgroup of *G*).

X(G) – Space of continuous homomorphism of G in the multiplicative group R\* = (R, )/{0}.


N – Nilpotent component of the Lie group G. Also is the normal subgroup of G, when N is

NF - Nilpotent component of the Lie group G, restricted to the simple roots subspace F 0.

NF – Compact nilpotent component of the Lie group G, restricted to the simple roots

N – Compact nilpotent component of the Lie group G. Also is the normal subgroup of G,

PF –Parabolic subgroup restricted to the subspace F, of the simple roots of <sup>0</sup>

0(G0) – Identity component of the Lie group G0.

A– Abelian Subgroup of Lie group G. Also Abelian component of the Iwasawa decomposition of the real reductive group G.

AF – Abelian subgroup of the Lie group G, restricted to subspace F, of the simple roots of 0.

mFLie algebra of the subgroup M of the Lie group G, restricted to subspace of the simple roots of 0. mF = {ha<sup>+</sup> Ad(g)h = h gG}.

F – Spherical function of Harish-Chandra restricted to subspace F.

nF Compact nilpotent algebra restricted to the simple root subspace F 0.

T(g) Tensor algebra of the Lie algebra g.

0MAN – Langlands decomposition endowed of homomorphism , and space modulo the Hilbert space H.

W(g, a) Weyl group of the homomorphisms on g, modulus a.

L2(G) Space of integrable square representations of the group G.

dg – invariant measure on the group *G* (invariant under proper movements of G).

GL(*n*, E) – General linear group on the Euclidean space of dimension *n*.

A(k) – Kernel of the integral equation whose solution is a function of spherical type.

\*aF – Nilpotent algebra to the minimal parabolic subgroup of MF.

V\*[n] – g-modules belonging to the category *H*.

V – Discriminant functional of the Lie algebra \*a, of the category of modules in *H*.

E(PF, V) – Homomorphism space of the Osborne lemma applied to the decomposition of the algebra U(\*a).

(PF, AF) – Space of weights of aF, on MF, of a roots system (P, A).

P(a)K – Polynomic Lie Algebra of all the K-invariant polynomies on the algebra g.

	- t Lie algebra corresponding to maximum torus in G.
	- System of semisimple positive roots.
	- [g, g] Bilateral ideal of antisymmetric elements.
	- Base of simple roots of a semisimple root space.

dim – Dimension.


IndPG() – Induced representation , of the group P, to the group G.

0M – Connected component of the Lie subgroup M, of the parabolic subgroup P, of the real reductive group G.


*H* – Category of U\*(n)-modules.

0G – Identity component of G. Explicitly 0G = 0ANK, or through continuous homomorphisms X(G), to know, 0G = {gG2(g) = 1, X(G)}.

## **Conclusion**

## **Conclusion**

176 Orbital Integrals on Reductive Lie Groups and Their Algebras, by Francisco Bulnes

t – Lie algebra corresponding to maximum torus in G.

System of semisimple positive roots.

dim – Dimension.

reductive group G.

codim – Codimension.

W – Weyl group.

Z(g) – Centre of Lie algebra g.

J(V) – Jacquet module of the module V.

K – Compact subgroup of the Lie group G.

N(T) – Normalizator group of the torus T.

X(G), to know, 0G = {gG2(g) = 1, X(G)}.

*H* – Category of U\*(n)-modules.

[g, g] – Bilateral ideal of antisymmetric elements.

(aC)\* - Abelian dual algebra of Abelian algebra a.


IndPG() – Induced representation , of the group P, to the group G.

p – Minimal compact subalgebra of the decomposition g = t p.

ZG(m) – The center of the algebra m, explicitly defined {GAd()x = x, xm}.

0G – Identity component of G. Explicitly 0G = 0ANK, or through continuous homomorphisms

Ad – Endomorphism of the Lie group G. Adjunct Map on the group G.

ad – Endomorphism of the Lie algebra g. Adjunct map on the algebra g.

gC – Complexivity of the real reductive Lie algebra g.

0M – Connected component of the Lie subgroup M, of the parabolic subgroup P, of the real

Base of simple roots of a semisimple root space.

The systematical securing of spherical functions in representations theory, the obtaining of geometrical and physical properties of the space through their cycles and co-cycles, the construction of generalized functionals in complex cohomology with coefficients in a Lie algebra for the solution of the partial and ordinary differential equations in field theory, the development of intertwining integrals for the obtaining of principal representations and interrelation of induced representations (Barchini, 1992),the evaluation of integrals on groups and algebras of Lie, the calculation of topological dimensions and determination of characters of a unitary representation are only some examples of some applications that can be solved by a suitable theory of orbital integrals in the context of the theory of topological groups and their operators. Some of the said problems have given place to the development of a global harmonic analysis with the perspective to generalize the formula of Plancherel (Wallach, 1982).In the way of the study of the invariance of this formula, it was possible to have obtained a specialization of the above mentioned analysis for determination of representations located by cuspidals using the evaluation of orbital integrals on groups of Lie and their algebras. In particular the harmonic analysis that has been realized in complete form, using these methods is on the groups *SL*(2, C), *SL*(2, R), *SU*(2, 2), and their compact orbits of these groups. Across a parallel study continued by the group of Harvard (Schmid, 1992) a development of induced representations has been obtained using co-adjunct orbits of a sheaf of complex holomorphic bundles, which initially was chasing to determine a method of integral transforms that were establishing classes of solutions for differential equations in field theory, using the invariance and conformability of cycles of the spacetime. Nevertheless, and under a study on homogeneous spaces one manages to establish that the orbital classes are representations induced for the co-adjunct orbits determined in relative cohomology (Bulnes, 2004).Many of these induced representations could be had obtained by them through the images of integral transforms which co-cycles are *G*/*L* representations, with *G*, not compact and *L*, compact, considering certain co-adjunct orbits on au-specialization (Bulnes, 2004) that are minimal *K*-Types (Salamanca-Riba, 2004), (Vogan, 1992).A suitable theory of integral intertwining operators on *G*/*L*, can help to the calculation of these minimal K-Types up to certain level.

## **Acknowledgements**

I am grateful to Demetrio Moreno-Arcega, Master of L, General Director of TESCHA, Humberto Santiago, Eng, Academic Sub-director of Tescha and Rodolfo Morales, B.L, Financing Sub-director of Tescha, for the financing and moral support to announce this mathematical research.

© 2013 Bulnes, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Bulnes, licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

## **References**


180

1998

129-158, 1981.

[28] Bulnes, F. J., *The(*

2005.

Recillas. 2004.

(Oxford Preprint) 22 8-13, 1986.

Oxford University Press, 1989.

Institute, México, 2005.

Phys. 78 (1981), 26-30.

Lond. Math. Soc. 1 (2)(1904) 451-458.

Physics, Cambridge University Press. 1990.

g

Cambridge University Press. 1990.

Series, v.351. London, 1996.

[20] F. Bulnes, *Course of Design of Algorithms to the Master of Applied Informatics*, UCI

[21] V. Knapp, *Harish-Chandra Modules and Penrose Transforms*, AMS, pp1-17, USA, 1992. [22] P. Shapira, and A. Agnolo, *Radon-Penrose Transform for D-Modules*, Elesevier, Holland,

[23] Grothendieck, Alexander (1960/1961). "Techniques de construction en géométrie analytique. I. Description axiomatique de l'espace de Teichmüller et de ses variantes.".*Séminaire Henri Cartan 13 no. 1, Exposés No. 7 and 8*, Paris, France. [24] Atiyah, M. F., *Green's Functions for self-dual four-manifolds,* Adv. Math. Supp Studies 7A.

[25] Baston, R. J., *Local cohomology, elementary states, and evaluation twistor,* Newsletter

[26] Baston, Eastwood, M., *The Penrose transform: its interaction with representation theory,* 

[27] Bateman, H., *The solution of partial differential equations by means of definite integrals,* Proc.

[29] Eastwood, M., Penrose, R., and Wells.,*Cohomology and massless fields,* Commun, Math.

[31] Gindikin, S., *Between Integral Geometry and Twistors,* Twistors in Mathematics and

[32] Gindikin, S., *Generalized Conformal Structures,* Twistors in Mathematics and Physics,

*Mathematical Physics,* Addison-Wesley Longman Ltd., Pitman Res. Notes in Math.

Conference in Representation Theory of Reductive Lie Groups., IM/UNAM, Mexico,

[38] Bulnes, F. "Some Relations Between the Cohomological Induction of Vogan-Zuckerman and Langlands Classification," Faculty of Sciences, UNAM *phD. Dissertation*, ed., F.

[39] Bulnes, F. "Conferences of Mathematics," Vol. 2., *Institute of Mathematics, UNAM, Compilation of conferences in Mathematics*, 1st ed., F. Recillas, Ed. México: 2002. [40] Mumford, D. *Geometric invariant theory*. Springer-Verlag, Berlin Heidelberg (1965).

[35] Bulnes, F. "Conferences of Mathematics," Vol. 1., *Institute of Mathematics, UNAM, Compilation of conferences in Mathematics*, 1st ed., F. Recillas, Ed. México: 2001. [36] Wong, H. "Dolbeault Cohomological Realization of Zuckerman Modules Associated with Finite Rank Representations," J. Funct. Anal. 129, No. 2 (1993), p428-459. [37] Bulnes, F. "Extension of G-Modules and Generalized G-Modules," International

[30] Gelfand, *Generalized Functions,* Vol. 5. Academic Press, N. Y., 1952.

[33] Helgason, S., *The Radon Transform,* Prog. Math. Vol. 5. Birkhäuser 1980. [34] Kravchenko, V. V., Shapiro, M. V., *Integral Representations for Spatial Models of* 

*, K)-modules Theory,* Applied math I Sepi-National Polytechnic

(University of Informatics Sciences), Habana, Cuba, 2006.


*Authored by Francisco Bulnes*

The purpose is to present a complete course on global analysis topics and establish some orbital applications of the integration on topological groups and their algebras to harmonic analysis and induced representations in representation theory.

Orbital Integrals on Reductive Lie Groups and Their Algebras

Orbital Integrals on Reductive

Lie Groups and Their Algebras

*Authored by Francisco Bulnes*

Photo by agsandrew / iStock