[gmx-users] Seg Fault in water and alcohol run

Caleb Carlin calebcarlin at gmail.com
Tue Feb 17 16:51:40 CET 2009

I am working witha system that has two heptadecanol lattices oriented toward
each other with a slab of water between.  The alcohol uses the OPLS-AA
parameters and the water is NE6 (Nada and Eerden six-site model).  When I
ran the system with the alcohols positionally fixed and their interaction
energy excluded, there was no problems.  When I fixed the water molecules
and released the alcohols, the system runs between 100 to 300 steps before
seg faulting.  Below is a description of what I have tried to fix the
problem and I can send input and output files if requested but excluded them
here due to size.  Any help in identifying the cause of the seg fault would
be greatly appreciated.

Steps taken:
added exclusions for all intra-alcohol interactions in the [ exclusions ]
added parameters for bonds and angles in the alcohols and changed
"constraints" in md.mdb to none
The alcohols are split into 4 groups; hydrogen in headgroup, oxygen in
headgroup, non-terminal carbon and hydrogens (CE), and terminal carbon and
hydrogens (CT).
Fixed the CE and CT groups positions and removed CE and CT interaction
All runs are at Temperature = 250 K.

At this point the system with just alcohols and no water runs without error,
but seg fault reappears with addition of water.

The step.pdb files that are written near the seg fault show the atoms of one
of the water models separated and spread out over a space outside of the
original system.  It is unknown if the energy blows up at the segfault as
the output appears to be limited to outputs every .1 ps.  Using random
velocities assigned using "gen_vel" results in the seg fault occurring at
different time steps and 5% of the time not happening.

Again any help would be appreciated.

Caleb Carlin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090217/73d85c81/attachment.html>

More information about the gromacs.org_gmx-users mailing list