[gmx-users] Walls & table-based potentials

[Brett Olsen]

Hello everyone,

I am trying to use the new walls feature in the CVS version of
gromacs.  My system is a simple box of SPC water.  When using
wall_type: 9-3
or
wall_type: 10-4
with appropriate atomtype and density parameters, I have no problems,
and the runs proceed normally.

However, when I try to use
wall_type: table
I obtain segmentation faults from the md_run program.  These faults
only occur when an atom of my simulation crosses into the table-lookup
area, within 2nm of one of the walls.  So I can postpone the fault by
adding 2nm or more of vacuum on either side of my simulation box, but
as soon as a water molecule escapes, the program crashes.  I have
tried changing the timestep, temperature, and pressure coupling with
no success.

I have also tried several different tables:  the original table with
the potential I am trying to use, the various table6-*.xvg tables from
the /top directory, appropriately renamed, as well as a blank table,
containing only zeros in all but the first, index, column.  All of
them fail in precisely the same fashion.

Have you any advice on what could be causing this problem?  I presume
it must be related to the table itself, but I am at a loss as to how
to fix it.

Thank you all kindly,
Brett Olsen