AMBER Archive (2009)
Subject: Re: [AMBER] Equilibration Scaling to System Size
From: Carlos Simmerling (carlos.simmerling_at_gmail.com)
Date: Tue Feb 24 2009 - 16:49:10 CST
the relaxation to a decent local minimum probably depends more on your
starting structure and how close it is to being "equilibrated" than it
does on system size.
my experience is that different sizes don't take significantly longer
if it is a very good structure and all you need is the surface solvent
to relax and to fix any crystal contacts, etc.
BUT if by equilibrated you mean how long to generate an equilibrium
ensemble, then likely it depends a lot on system size- but that's
probably not what you mean.
On Tue, Feb 24, 2009 at 5:31 PM, Hopkins, Robert <hopkins_at_uhcl.edu> wrote:
> AMBER Users,
> I have looked for information about the time required for proper
> equilibration vs. system size in a regular MD run, but haven't found
> anything very close in the reflector archive.
> So, here is my question: from practical experience, *roughly* how does
> the system time (not clock time) required for appropriate equilibration
> (prior to MD production) scale with total number of atoms (N) in the
> system? It seems that a time proportional to somewhere between N^3 and
> N! might be on the right track, but surely it also depends on how many
> TIP3P waters with limited degrees of freedom are in the system, etc.
> So, for example, if a 45,000 atom system needs on the order of 20 ns of
> equilibration time, can one estimate how much time a 15,000 atom system
> might need?
> I would greatly appreciate hearing about anyone's experiences or 'rules
> of thumb' relating to this question. Thanks in advance.
> Robert Hopkins
> AMBER mailing list
AMBER mailing list