Accelerating polymer simulation by means of tree data-structures and a parsimonious Metropolis algorithm

Abstract

We show how a Monte Carlo method for generating self-avoiding walks on lattice geometries which employs a binary-tree data-structure can be adapted for hard-sphere polymers with continuous degrees of freedom. Data suggests that the time per Monte Carlo move scales logarithmically with polymer size. Next we generalize the method to Lennard-Jones polymers with untruncated monomer–monomer interaction. To this end we propose a variant of the Metropolis algorithm and demonstrate that in combination with the tree data-structure logarithmic scaling can be preserved. We further show how the replica-exchange method can be adapted for the same purpose.