Kinetics of liquid-solid phase transition in large nickel clusters
Frankfurt Institute for Advanced Studies, Ruth-Moufang-Str. 1, 60438 Frankfurt am Main, Germany
arXiv:1210.3559 [physics.comp-ph] (12 Oct 2012)
@article{2012arXiv1210.3559Y,
author={Yakubovich}, A.~V. and {Sushko}, G. and {Schramm}, S. and {Solov’yov}, A.~V.},
title={"{Kinetics of liquid-solid phase transition in large nickel clusters}"},
journal={ArXiv e-prints},
archivePrefix={"arXiv"},
eprint={1210.3559},
primaryClass={"physics.comp-ph"},
keywords={Physics – Computational Physics, Condensed Matter – Mesoscale and Nanoscale Physics, Physics – Atomic and Molecular Clusters},
year={2012},
month={oct},
adsurl={http://adsabs.harvard.edu/abs/2012arXiv1210.3559Y},
adsnote={Provided by the SAO/NASA Astrophysics Data System}
}
In this paper we have explored computationally the solidification process of large nickel clusters. This process has the characteristic features of the first order phase transition occurring in a finite system. The focus of our research is placed on the elucidation of correlated dynamics of a large ensemble of particles in the course of the nanoscale liquid-solid phase transition through the computation and analysis of the results of molecular dynamics (MD) simulations with the corresponding theoretical model. This problem is of significant interest and importance, because the controlled dynamics of systems on the nanoscale is one of the central topics in the development of modern nanotechnologies. MD simulations in large molecular systems are rather computer power demanding. Therefore, in order to advance with MD simulations we have used modern computational methods based on the graphics processing units (GPU). The advantages of the use of GPUs for MD simulations in comparison with the CPUs are demonstrated and benchmarked. The reported speedup reaches factors greater than 400. This work opens a path towards exploration with the use of MD of a larger number of scientific problems inaccessible earlier with the CPU based computational technology.
October 15, 2012 by hgpu
