Hybrid Monte Carlo simulation of polymer chains

Abstract

We develop the hybrid Monte Carlo method for simulations of single off-lattice polymer chains. We discuss implementation and choice of simulation parameters in some detail. The performance of the algorithm is tested on models for homopolymers with short- or long-range self-repulsion, using chains with $16\le N\le 512$ monomers. Without excessive fine tuning, we find that the computational cost grows as $N^{2+z^{\prime }}$ with $0.64<z^{\prime }<0.84$. In addition, we report results for the scaling of the end-to-end distance, $r_{1N}\sim N^{\nu }(\ln N{)}^{-\alpha }$.

Comments: 13 pages, single postscript file, uuencoded form, Lund Preprint LU-TP 93-20