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

TL;DR

This paper introduces a dynamic Monte-Carlo algorithm that significantly improves the efficiency of simulating dense long-chain molecules by generating multiple trial configurations and selectively cloning successful chains.

Contribution

It extends the configurational-bias Monte-Carlo method with a swarm of trial configurations and chain cloning, achieving over three orders of magnitude efficiency gain.

Findings

Efficiency improved by at least 1000 times compared to traditional methods

Method effective for dense polymer brush simulations

Waste recycling enhances statistical accuracy

Abstract

Proposed here is a dynamic Monte-Carlo algorithm that is efficient in simulating dense systems of long flexible chain molecules. It expands on the configurational-bias Monte-Carlo method through the simultaneous generation of a large set of trial configurations. This process is directed by attempting to terminate unfinished chains with a low statistical weight, and replacing these chains with clones (enrichments) of stronger chains. The efficiency of the resulting method is explored by simulating dense polymer brushes. A gain in efficiency of at least three orders of magnitude is observed with respect to the configurational-bias approach, and almost one order of magnitude with respect to recoil-growth Monte-Carlo. Furthermore, the inclusion of `waste recycling' is observed to be a powerful method for extracting meaningful statistics from the discarded configurations.