AMBER Archive (2007)

Subject: Re: AMBER: large RMS fluctuations in turning off VDW interactions by TI

From: Thomas Steinbrecher (steinbrt_at_scripps.edu)
Date: Fri Jul 06 2007 - 12:55:32 CDT


Hi everybody,

Holly, the numbers you sent for rms look very much as if you used
klambda=1, if youre system behaves this way at klambda=4 or 6 I would be
very interested to try it with the new softcore potentials. If all you
want is a relatively solvation free energy, then setting things up should
be easy, even if you only have the amber8 input files. TI format changed a
bit over time, and softcore potentials as implemented now unfortunately
need still different input compared to amber9 (but now the format is as
simple as I could make it...). Would you rather like to get your hands on
an amber10 account (Dave would need to help there) or shall we use your
input files to run things here at Scripps?

I'll be in Germany until the week after next, but from now on I should
check mail more often, so keep me updated on what you want to do.

And regarding the error vs rms debate, David is very right here. It might
be a good idea to print out dvdl every step or so for a short run and
compute the autocorrelation time to get an estimate of how many
uncorrelated samples you really have. But even then you might run into
trouble if some of your degrees of freedom (dihedral angles for example)
are sampled very badly, like transititng only once or twice between minima
during your run.

Kind Regards,

Thomas

Dr. Thomas Steinbrecher
The Scripps Research Institute
10550 N. Torrey Pines Rd.
San Diego CA 92037, USA

On Thu, 5 Jul 2007, David Mobley wrote:
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu