Bad Global Exclusion Count Errors in (Relatively) Simple System

From: Jason Richard Mick (dw2413_at_wayne.edu)
Date: Tue Nov 16 2010 - 18:32:49 CST

Hi All,

I've seen lots of posts on this topic on the NAMD mailing list, but none of the suggestions really seemed to explain why my system is encountering this issue. I've also read this:
http://www.ks.uiuc.edu/Research/namd/wiki/index.cgi?NamdTroubleshooting

Basically, from what I understand, you may encounter this error if:
1. You have hydrogen bonds that overlap the PBCs.
2. You have too small a cutoff when minimizing.
3. You have coordinates at 0,0,0
4. A residue is in multiple chains.

I have 500 methyl acetate molecules inside a simulation box, and want to run an NPT simulation which I expect to go to the gas phase.

I've checked my *.pdb file to ensure that #3 doesn't apply. #1 doesn't apply as I don't have hydrogen bonding groups. For #3 I've tried a number of box sizes (50x50x50, 80x80x80, etc.) and bumped my cutoff up very high (30,50, etc.) and I still get this error. #4 isn't the case as I only have one chain ID (A) in my system.

I would expect my system to be in the gas phase once heated, so I would assume I want to start with a large box size (e.g. 80x80x80) so as to ensure a lower density. I would assume that when heated, the system might initially contract, but then would expand to the gas phase as the heating continues.

What seems to happen is that if I run a sufficiently large time step (1 or 2 fs), I immediately (and consistently) get:

OPENING EXTENDED SYSTEM TRAJECTORY FILE
FATAL ERROR: Bad global exclusion count!

at timestep 0.

If I bump the timestep size down to 0.5 fs, this error goes away, but my system gets stuck in a meta-stable state, I believe, as it stays at a liquid-like density, when it should be transitioning to vapor (target temp and pressure is 475 K, 25 atm).

Any ideas on how I can run a larger timestep without getting this error besides those outlined in the troubleshooting guide???

Has anyone run into this issue??

Cheers,
Jason
-------------------
Jason R. Mick
Potoff Group
Wayne State University

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:23:23 CST