NRMC – A GPU code for [formula omitted]-Reverse Monte Carlo modeling of fluids in confined media

Abstract

NRMC is a parallel code for performing N-Reverse Monte Carlo modeling of fluids in confined media [V. Sánchez-Gil, E.G. Noya, E. Lomba, J. Chem. Phys. 140 (2014) 024504]. This method is an extension of the usual Reverse Monte Carlo method to obtain structural models of confined fluids compatible with experimental diffraction patterns, specifically designed to overcome the problem of slow diffusion that can appear under conditions of tight confinement. Most of the computational time in N-Reverse Monte Carlo modeling is spent in the evaluation of the structure factor for each trial configuration, a calculation that can be easily parallelized. Implementation of the structure factor evaluation in NVIDIA® CUDA so that the code can be run on GPUs leads to a speed up of up to two orders of magnitude. Program summaryProgram Title: NRMC_gpuProgram Files doi:http://dx.doi.org/10.17632/kbbgbkn68m.1Licensing provisions: GNU General Public License 3 (GPL)Programming language: FORTRAN, C and NVIDIA® CUDASupplementary material: An example calculation is providedExternal routines/libraries: LAPACK (for gfortran) or Intel® MathKernel for Intel® Fortran (included in Intel’s distribution)Nature of problem: Determination of structural models of confined fluids compatible with experimental diffractogramsSolution method:N-Reverse Monte Carlo simulations using GPUs