AMBER Archive (2005)

Subject: Re: AMBER: why the dipeptide always moving out the virtual box

From: David A. Case (case_at_scripps.edu)
Date: Mon Jun 13 2005 - 17:55:19 CDT


On Mon, Jun 13, 2005, Li Su wrote:

> I'm simulating alanine dipeptide in vaccum using sander, but everytime
> after around 350 ps, it stops and tells me the "The system has extended
> beyond the extent of the virtual box. Restarting sander will recalculat a
> new virtual box with 30 Angstroms extra on each side, if there is a restart
> file for this configuration." The I use the rst file to restart , but after
> about the same period the same problem happens.

I suspect that you are seeing translational diffusion, that moves you out of
the virtual box. You should visualize your trajectory to see if this is the
case. You should be able to fix this up by setting your cutoff to 999. (which
is probably not much different from 12 for such a small molecule), and also
to set ntnb to some value bigger than the total number of steps you are
requesting. That way, the code will never try to update the list.

Basically, the current Amber code is not well designed for long time-scale
vacuum simulations. These problems do not exist for gb or periodic
simulations, which is what most people use. But I hope that the workaround
listed above will fix your problem.

...dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu