M. Schoen, TAYLOR-EXPANSION MONTE-CARLO SIMULATIONS OF CLASSICAL FLUIDS IN THE CANONICAL AND GRAND-CANONICAL ENSEMBLE, Journal of computational physics, 118(1), 1995, pp. 159-171
In this article the Taylor-expansion method is introduced by which Mon
te Carlo (MC) simulations in the canonical ensemble can be speeded up
significantly. Substantial gains in computational speed of 20-40% over
conventional implementations of the MC technique are obtained over a
wide range of densities in homogeneous bulk phases. The basic philosop
hy behind the Taylor-expansion method is a division of the neighborhoo
d of each atom (or molecule) into three different spatial zones. Inter
actions between atoms belonging to each zone are treated at different
levels of computational sophistication. For example, only interactions
between atoms belonging to the primary zone immediately surrounding a
n atom are treated explicitly before and after displacement. The chang
e in the configurational energy contribution from secondary-zone inter
actions is obtained from the first-order term of a Taylor expansion of
the configurational energy in terms of the displacement vector d. Int
eractions with atoms in the tertiary zone adjacent to the secondary zo
ne are neglected throughout, The Taylor-expansion method is not restri
cted to the canonical ensemble but may be employed to enhance computat
ional efficiency of MC simulations in other ensembles as well. This is
demonstrated for grand canonical ensemble MC simulations of an inhomo
geneous fluid which can be performed essentially on a modern personal
computer. (C) 1995 Academic Press. Inc.