Title :
Linear Complexity and High Scalability of a Parallel Monte Carlo Simulation Method
Author :
Amaro-Rico, Guillermo ; Botello-Rionda, Salvador ; Carrillo-Tripp, Mauricio
Author_Institution :
Dept. de Comput., Centro de Investig. en Matemdticas (CIMAT), Guanajuato, Mexico
Abstract :
The Monte Carlo method has been used extensively on computational simulations of phenomena that can be described as a system of particles. In this work we propose a programming methodology based on domain grids that avoids the calculation of the distance between all pairs of particles. As a result, faster results with linear complexity and high scalability are produced. Application of the algorithm to calculate the solvation index of liquid argon was used as validation. The system size used was 10 million atoms. A comparison is made between implementations using OpenMP or CUDA.
Keywords :
Monte Carlo methods; argon; chemistry computing; computational complexity; parallel architectures; solvation; CUDA; OpenMP; computational simulations; domain grids; linear complexity; liquid argon; parallel Monte Carlo simulation method; programming methodology; scalability; solvation index; Argon; Atomic measurements; Complexity theory; Graphics processing units; Indexes; Monte Carlo methods; Potential energy; CUDA; Computational Simulation; HPC; Monte Carlo; Parallel Computing;
Conference_Titel :
P2P, Parallel, Grid, Cloud and Internet Computing (3PGCIC), 2013 Eighth International Conference on
Conference_Location :
Compiegne
DOI :
10.1109/3PGCIC.2013.124
