AMBER: bug in reprocessing with PB?

From: David Mobley <>
Date: Thu, 10 May 2007 11:08:10 -0700


I'm doing some studies to compare implicit solvent models in AMBER,
and I believe I've found a bug of some sort.

In particular, I have a set of trajectory files and I need to evaluate
energies of the snapshots from these trajectories in different
implicit solvent models. Mostly, this works fine (using sander with
imin=5 to evaluate energies of a stored trajectory). However,
sporadically, when I use PB for reprocessing, I get segmentation
faults during reprocessing for no apparent reason. The trajectories
seems to be OK (i.e., reprocessing with GB works fine).

Sometimes, these segmentation faults are architecture-dependent (i.e.
I'm mainly on Xeons; stuff that fails there will work OK, sometimes,
on Opterons). However, I'm attaching a tarball for one case that fails
(with PB) on either architecture. Again, it reprocesses fine using GB.

I'm really at a loss here -- I desperately need this reprocessing to
work -- so I will greatly appreciate any help. Just grab the tarball
and run the "" shell script, which executes sander to do the
reprocessing that's causing the seg fault.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun May 13 2007 - 06:07:28 PDT
Custom Search