Re: AMBER: GB with Langevin dynamics - solution

From: Peter Varnai <pv232.cam.ac.uk>
Date: Tue, 12 Apr 2005 18:39:54 +0100 (BST)

Dear David Case,

> I'd still like to know how many atoms were in the system you examined.
> And, the explanation you give doesn't really say why setting nscm to its
> default value of 1000 was not enough.

My test system had 188 atoms, but I have seen periodicity with systems
including 447 or 1009 atoms. When mentioning the overall translation, I
considered that over 5 A/ns movement is quite substantial considering the
fact that every 1000 steps the COM movement was eliminated. I am happy to
send you the test input files if you want to check the behaviour yourself.
It only takes a couple of hours on a single CPU. Actually, I had a look at
CHARMM doing Langevin dynamics and it seems to have the same seeding
problem.

What still intrigues me is the fact that periodic behaviour is seen not
only in the global motion of the molecule but in the fitted rmsd values as
well. I think this clearly shows that the periodic "random" forces
dominate the dynamics of the system. I can also add that this behaviour is
more difficult to spot if gamma < 5.

Best wishes,
Peter
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Tue Apr 12 2005 - 18:53:00 PDT
Custom Search