Hi,
I am trying to run a number of simulations of a fairly large (~100k atoms)
explicit solvent protein-DNA system with modified AA residues using
pmemd.MPI (compiled with ./configure -intelmpi -openmp -mic2 intel for a
KNL cluster). Several of these run with no problems, but the system
containing a 3-nitrotyrosine residue runs for ~10,000 steps before quitting
with the error "Increase lin__limit in nb_pairlist.F90!!!".
This same system ran with no problems using sander.MPI, and changing number
of MPI processes / node has no effect. Changing the cutoff from 10 to 8
seems to help, and I assume I could re-compile without -mic2 or change the
hard-coded value for lin_limit (=2000) in the source code. My questions are:
1.) Why do I get this error only in the system with the nitrotyrosine
residue, and not other nearly identical systems? Does it suggest something
is incorrect about way the system was set up or run?
2.) Is there any danger in changing lin_limit in the nb_pairlist.F90
source? And can anyone suggest values to try?
Example .out from a failed run is attached below, please let me know if any
other files would be helpful.
Thanks!
Colin
########################################
#
# Colin W. Brown
# Postdoctoral Researcher
# Jeffrey Barrick's Lab
# University of Texas at Austin
# Phone: (510)-875-8194 <(510)%20875-8194>
# Email: colin.walsh.brown.gmail.com
#
########################################
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Mar 21 2018 - 08:00:04 PDT
