Replica exchange - Temperatures and number of replicas

From: Siri Søndergaard (
Date: Sun Nov 25 2012 - 17:06:44 CST


I'm trying to run replica exchange simulations with metadynamics and the
issue is to determine the most suitable temperatures and number of
replicas. My system contains app. 20.000 atoms and I want around 20% of the
swaps to be accepted. I tried several combinations of temperatures and
replicas but only one resulted in an actual swap: T1 = 300, T2 = 350 and
the number of replicas was 12. This gave a probability of accepted swaps of
8%. I think 12 replicas is a quite high number for a small temperature
difference but if the difference is increased with as little as 50K I get
no swaps at all. Does anybody have experience setting up these parameters?

Siri Sondergaard

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:22:17 CST