From owner-chemistry $#at#$ ccl.net Thu Sep 21 09:47:00 2017 From: "Joshua T Berryman josh.berryman-x-uni.lu" To: CCL Subject: CCL: MC simulation of dense hard sphere system Message-Id: <-53012-170921034558-7869-50LiyhC2xjQAspwUiadPeg(0)server.ccl.net> X-Original-From: "Joshua T Berryman" Date: Thu, 21 Sep 2017 03:45:55 -0400 Sent to CCL by: "Joshua T Berryman" [josh.berryman[a]uni.lu] >>The problem I am facing is that most of the times new configurations are >>rejected because of overlapping spheres and acceptance I get is very low. >>What should I do for getter better acceptance, I tried very small >>displacements also. Yes, converging an HS system can be surprisingly expensive. You have two choices, one is to optimize your code (see the book Allen and Tildesley, example code is on github: https://github.com/Allen-Tildesley/examples ). Useful optimisations include: Cell system, Almarza barostat, automatic setting of stepsize. Almarza barostat is particularly useful for generating a dense- packed system as the acceptance rate for volume changes does not vanish with an increasing number of particles. The second choice is to use an optimised existing code. I would recommend dynamo: http://dynamomd.org/ . Note that dynamo is EDMD, not MC. Speed is in my experience comparable or better, due to the large expense of random number generation in MC. Dr Josh Berryman Uni Luxembourg