From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Thu Aug 30 2012 - 02:16:40 CDT
But only for the NPT run as a higher margin will slow down the simulation.
The error you see comes due the parallelization of namd in combination with
variable volume=NPT. The box will initially be departed to compute them
separated on multiple cpus. Than the constant pressure will shrink the box
to reach the target density. If the initial density of the solvent was too
low, the box will shrink that small, that some of the box parts will get
empty. This will cause a crash as the departing of the box cannot be done
again without restarting the simulation.
So, usually a restart of the simulation from the latest restart files(of
course with the xsc file) will solve it, as the box will mostly not shrink
so much again and the departing will be done again at simulation start. To
prevent it from crashing already, you can increase the margin parameter, but
as already mentioned, this will cause a dramatic slow down of the
simulation. You can decide what to use, or how to deal with that. I usually
let the simulations run for some days and don't want to check if a
simulation as broke down, so I use the higher margin. If you nevertheless
wait for your simulation to finish, it's no problem for you to just restart
it when it crashed. Remember, this error will usually only happen one time
for a system, during the 1st NPT run.
Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag
von snoze pa
Gesendet: Mittwoch, 29. August 2012 22:34
Betreff: Re: namd-l: question and NAMD operation
Increase margin, say 4.
On Wed, Aug 29, 2012 at 3:19 PM, Patrick Chiu <patrick817us_at_yahoo.com>
I have the following error in my NAMD operation:
Reason: FATAL ERROR: Periodic cell has become too small for original patch
grid!Possible solutions are to restart from a recent checkpoint,
increase margin, or disable useFlexibleCell for liquid simulation.
My question is how can I prevent this problem?
See Hong Chiu
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:22:28 CST