You could take the final snapshot and do a minimization on it. Such a large
term will almost certainly give you the largest gradient. This will point
you to one of the two atoms, and visualization may help you narrow it in
from there.
Perusing the "bond" subroutine in ene.f (AMBERHOME/src/sander) leads me to
believe that what you're looking for is not supported. You can always use
"idecomp" to do a residue-wise decomposition, and this should help you
narrow down where it's occurring. You could always try and add a couple
print statements in that subroutine that is triggered by a large energy, but
expect to see a performance hit for having an additional conditional
embedded in a potentially large loop (but it's only for debugging, after
all).
All that said, I've always found that such occurrences can usually be
spotted pretty easily just visualizing the system, so you may be best served
by loading up your favorite visualizer and poking around a little.
Note that debugging through sander may require you to print out details more
often so you can see exactly what's happening (ntwx, ntpr, ntwr, etc.)
Good luck!
Jason
On Wed, Oct 13, 2010 at 1:25 PM, Dian Jiao <oscarjiao.gmail.com> wrote:
> Hi all,
>
> Is there way to output pairwise interaction energies between atoms in the
> simulation mdout file? I got huge bond energy and that killed the run. Just
> wanted to pinpoint which bond(s) is causing that. Thanks.
>
> ~ DJ
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Oct 13 2010 - 11:00:02 PDT
