Re: Periodic cell too small

From: Brian Bennion (brian_at_youkai.llnl.gov)
Date: Tue May 25 2004 - 17:39:45 CDT

Hello,
This is my brief understanding of the issues at work here. Which can be
superceded by a developer if I say something wrong.

The way NAMD decomposes the system can lead to the errors talked about
below.

There are two methods used, one assigns individual atoms to "patches" in
grid that resembles the system dimensions. In constant pressure
situations the system can be compressed to a point where patches in the
grid become "empty". And these empty patches are the cause of the
problems. This will hinder performance, which is why I think you get the
same system to work in older versions of NAMD, ie NAMD2.5 is more
optimized and picky about the contents of patches in order to maximize
performance.

I hope that doesn't sound to circular. Since this question does come up
and the documentation is vague as to the direct cause of this particular
error, maybe a final say by a developer is in order. The response can
then be posted on the NAMD wiki page for future reference...if it hasn't
already.

brian

On Tue, 25 May 2004, Ekta Khurana wrote:

>
> HI everyone,
> I am running constant surface tension simulation with a constant
> normal pressure.If I use NAMD 2.5 I get the error
>
> FATAL ERROR: Periodic cell has become too small for original patch grid!
> Possible solutions are to restart from a recent checkpoint,
> increase margin, or disable useFlexibleCell for liquid simulation.
>
> Sometimes I am able to fix the problem using the suggested
> solutions BUT the same system runs fine with the earlier version of NAMD. The same
> question has been asked on the list earlier also but there was no clear
> answer.Can someone explain this error a little bit in detail.And also why
> it appears only with the newer version.
>
> Thankyou,
> Ekta
>

*****************************************************************
**Brian Bennion, Ph.D. **
**Computational and Systems Biology Division **
**Biology and Biotechnology Research Program **
**Lawrence Livermore National Laboratory **
**P.O. Box 808, L-448 bennion1_at_llnl.gov **
**7000 East Avenue phone: (925) 422-5722 **
**Livermore, CA 94550 fax: (925) 424-6605 **
*****************************************************************

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:37:38 CST