OpenMPI FEP Yields SegFaults and/or infinite dG

From: Andrew Miglino (
Date: Wed Apr 10 2013 - 12:51:05 CDT

Hi all,

I'm relatively new to NAMD, but I've gone through both the FEP
tutorial and the ABF tutorial in preparation for running my own FEP
calculations to derive octanol/water partition coefficients (i.e.
deltaG_hydration - deltaG_solvation). I've been able to compute simple
deltaG_hydration values for methane as in the ABF tutorial on a single
core on my host computer. However, when I then apply the same method
to an OpenMPI implementation of NAMD v2.9 I get signal 11 errors and a
rapidly expanding box volume. This occurs regardless of solvent/solute
pair. I've seen the error "corrupted double-linked list", which I see
has an explanation but no clear solution.

In runs that do perform FEP calculations before showing a signal 11,
the .fepout file shows dG to be +99999999. Again, there is some
explanation for this in the troubleshooting guide, but since I've
tried to use the tutorial files in this setup to no avail, I strongly
doubt the error to be similar to "atoms moving too fast".

Are there suggestions as to how I can proceed from here?

Thank you for your time

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:08 CST