Efficient configurational-bias Monte-Carlo simulations of chain molecules with 'swarms' of trial configurations

摘要

The pruned-enriched Rosenbluth method (PERM) is a popular and powerful Monte-Carlo technique for sampling flexible chain polymers of substantial length. In its original form, however, the method cannot be applied in Markov-chain Monte-Carlo schemes, which has rendered PERM unsuited for systems that consist of many chains. The current work builds on the configurational-bias Monte-Carlo (CBMC) method. The growth of a large set of trial configurations in each move is governed by simultaneous pruning and enrichment events, which tend to replace configurations with a low statistical weight by clones of stronger configurations. In simulations of dense brushes of flexible chains, a gain in efficiency of at least three orders of magnitude is observed with respect to CBMC and one order of magnitude with respect to recoil-growth approaches. Moreover, meaningful statistics can be collected from all trial configurations through the so-called "waste-recycling" Monte Carlo scheme. Published by AIP Publishing.