Sutton-Chen Potential (parameters for gold)

Fig. 1. Example of an LPMD control file. The components are loaded (use...enduse) and

We will denote by structural property, any quantity *AS* which depends on the instant *t* only

LPMD allows the calculation of several structural properties (either as instantaneous values or as averages), including the radial distribution function *g*(*r*) (using the gdr plug-in) and common neighbor analysis (through the cna plug-in), both of which can be used to measure

*AS*(*t*) = *AS*(*r*1(*t*),...,*rN*(*t*)), (6)

www.lpmd.cl.

cell cubic 28.56

steps 5000 #Integrator

enduse

enduse

enduse

then applied.

integrator vv

dt 1.0

#CellManager

use linkedcell mode auto cutoff 7.5

use suttonchen as sc e 0.013 n 10 a 4.08 m 8 c 34.408 cutoff 7.5

#- Applying Plugins -#

cellmanager linkedcell potential sc Au Au

**3.1 Structural properties with LPMD**

through the atomic coordinates,

with *N* the number of particles.

a degree of deviation from an ideal crystal structure.

periodic false true true

at Nanoscopic Level: A Molecular Dynamics Study

use velocityverlet as vv

Most codes cannot handle in an easy way the requirements of some setups, such as non-periodic boundary conditions, non-negligible variations of density inside a sample, or initial states prepared far from equilibrium. It might be possible to modify these codes to lift some of the limitations, but it could be cumbersome and error-prone. For these cases, a more flexible MD code is needed, even though some performance could be sacrificed.

We could say that the early way of doing MD was to implement a tailor-made computer program with precisely the chosen algorithms for numerical integration of the equations of motion and computation of the interatomic potentials and forces. Thus, one different computer code for each system to be simulated.

The next stage in MD computer codes is the ability to choose the interatomic potential at runtime (i.e., every time the program is executed, without the need to recompile for every change) along with all the other options such as the time step used for integration, total simulation time, initial conditions of pressure and temperature and so on. This has led to general purpose MD codes such as Moldy (Refson, 2000) and DL\_POLY (Smith & Forester, 1996) among many others. While the ability to choose the potential function is commonplace nowadays, very few computer codes offer the choice of changing the integration algorithm at runtime, although several have the choice at compile-time (i.e., during the compilation stage).

From a general point of view, the MD procedure consist of four main stages, namely: (a) the initialization of the sample, (b) the calculation of interatomic forces, (c) the integration of the equations of motion, and (d), collecting statistics and the computation of properties. It work quite well in several different cases, like equilibrium conditions or even for metastable system, like glasses (see, for example (Gutiérrez et al., 2010)). But also MD procedure can be applied to more extreme conditions.

When the MD simulation that we intend to perform is not standard, for example in the case of simulations far away from thermodynamic equilibrium (shockwaves (Loyola et al., 2010), high velocity impacts, ) or non-standard potential functions and forces (for example friction forces or external fields) one can clearly see the need for an hybrid approach between the tailor-made MD code (containing exactly the algorithms we need for a given simulation) and the general purpose MD code (with several choices available at run-time and compile-time). We would want to replace pieces of the program at will, including (but not limited to) integration methods, potential functions and other algorithms, such as the one responsible for computing interatomic distances or the thermostat algorithms used to control the applied temperature or pressure in an isothermal-isobaric (NPT) MD simulation. Here the general purpose approach is not general enough, only allowing some limited choices.

Our motivation for writing yet another MD code, *Las Palmeras Molecular Dynamics* (LPMD) (Davis et al., 2010) is to fill this practical void. LPMD is designed as a completely modular MD code, consisting of a set of interchangeable pieces or *plug-ins* which can be linked together in different ways to accommodate the needs of a non-standard MD simulation. Beyond that, the user can also perform post-simulation analysis, convert between input/output formats, prepare samples with ease and visualize simulations in real time. LPMD's modular design also improves efficiency in some cases. It also allows the user to add new pieces (integration methods, interatomic potentials, properties, file formats, and many others) without the need for learning the complete code architecture. LPMD is open source software written in standard C++ language, and released under the General Public License

4 Will-be-set-by-IN-TECH

said optimization it performs poorly on a different kind of system. This, in practice, only

Most codes cannot handle in an easy way the requirements of some setups, such as non-periodic boundary conditions, non-negligible variations of density inside a sample, or initial states prepared far from equilibrium. It might be possible to modify these codes to lift some of the limitations, but it could be cumbersome and error-prone. For these cases, a more

We could say that the early way of doing MD was to implement a tailor-made computer program with precisely the chosen algorithms for numerical integration of the equations of motion and computation of the interatomic potentials and forces. Thus, one different

The next stage in MD computer codes is the ability to choose the interatomic potential at runtime (i.e., every time the program is executed, without the need to recompile for every change) along with all the other options such as the time step used for integration, total simulation time, initial conditions of pressure and temperature and so on. This has led to general purpose MD codes such as Moldy (Refson, 2000) and DL\_POLY (Smith & Forester, 1996) among many others. While the ability to choose the potential function is commonplace nowadays, very few computer codes offer the choice of changing the integration algorithm at runtime, although several have the choice at compile-time (i.e., during the compilation stage). From a general point of view, the MD procedure consist of four main stages, namely: (a) the initialization of the sample, (b) the calculation of interatomic forces, (c) the integration of the equations of motion, and (d), collecting statistics and the computation of properties. It work quite well in several different cases, like equilibrium conditions or even for metastable system, like glasses (see, for example (Gutiérrez et al., 2010)). But also MD procedure can be applied

When the MD simulation that we intend to perform is not standard, for example in the case of simulations far away from thermodynamic equilibrium (shockwaves (Loyola et al., 2010), high velocity impacts, ) or non-standard potential functions and forces (for example friction forces or external fields) one can clearly see the need for an hybrid approach between the tailor-made MD code (containing exactly the algorithms we need for a given simulation) and the general purpose MD code (with several choices available at run-time and compile-time). We would want to replace pieces of the program at will, including (but not limited to) integration methods, potential functions and other algorithms, such as the one responsible for computing interatomic distances or the thermostat algorithms used to control the applied temperature or pressure in an isothermal-isobaric (NPT) MD simulation. Here the general

Our motivation for writing yet another MD code, *Las Palmeras Molecular Dynamics* (LPMD) (Davis et al., 2010) is to fill this practical void. LPMD is designed as a completely modular MD code, consisting of a set of interchangeable pieces or *plug-ins* which can be linked together in different ways to accommodate the needs of a non-standard MD simulation. Beyond that, the user can also perform post-simulation analysis, convert between input/output formats, prepare samples with ease and visualize simulations in real time. LPMD's modular design also improves efficiency in some cases. It also allows the user to add new pieces (integration methods, interatomic potentials, properties, file formats, and many others) without the need for learning the complete code architecture. LPMD is open source software written in standard C++ language, and released under the General Public License

purpose approach is not general enough, only allowing some limited choices.

flexible MD code is needed, even though some performance could be sacrificed.

allows the study of certain systems and conditions.

computer code for each system to be simulated.

to more extreme conditions.

(GPL) version 3. Figure 1 displays an example of the control file. For more information, visit www.lpmd.cl.

```
#This is a comment. Comments are used usually as a title:
#########################################