Re: multiple MD simulations

From: Marc Q. Ma (
Date: Tue Mar 29 2005 - 07:42:46 CST

Hi Aaron,

I have not done this type of experiments myself, however, I have seen
other papers that support the same claims as Caves et al.

The practice is that you run multiple MD simulations with different
initial random seeds -- so that each simulation starts with a different
initial velocity profile. This randomness will contribute to barrier
crossing and thus improve sampling efficiency.

Another trick is how to define "short" and "long." People may use 300
ps simulation as a "short" trajectory -- which in some applications is
already "long."

On Mar 28, 2005, at 7:29 PM, Aaron Oakley wrote:

> Greetings NAMD users,
> Have any of you seen this paper?
> Caves et al., (1998) Locally accessible conformations of
> of proteins: Multiple MD simulations of crambin.
> Protein Sci. 7:649-666
> In it the authors argue that several short MD simulations
> give a more representative ensemble that one long simulation.
> Has anyone had experience with this?
> Thanks,
> aaron++

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:40:37 CST