From: Jeffrey Potoff (jpotoffx_at_gmail.com)
Date: Wed Jun 26 2013 - 13:10:26 CDT
On 6/26/2013 1:15 PM, Jon Kusner wrote:
> Hello,
>
> We have been trying to create an acetonitrile solvent box. We have
> been following the instructions of other troubleshooters and looking
> on other forums for guidance but have still not been successful.
> Currently we have a psf of our final solvent box, a 12x12x12 array of
> acetonitrile molecules, that we are trying to equilibrate. Whenever we
> try to run an equilibration we get through 3000 step of minimization
> but as soon as the program begins molecular dynamics we receive the error:
> Fatal error on PE 2> 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.
> We have tried each of these possible solutions and received the same
> result. If there is any guidance we could receive it would be much
> appreciated. Thank you for all of your help.
>
> Best,
> Jon
If you search the mailing list for "patch grid error", you will turn up
many posts on this, including some of mine.
http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l.2009-2010/2898.html
The issue is that your initial configuration is at a density that is far
from the equilibrium density. As the system equilibrates, which in this
case means shrinking, eventually system becomes small enough that the
original patch grid calculated by NAMD doesn't work any more. The
solution is to start from a density that is closer to the correct
value. If you don't want to do that, or don't know what the correct
density should be, the alternative is to run a series of short
simulations. This will allow the simulation to run for a while and drop
a few configurations that you can reload into NAMD to continue the
equilibration process after you get "patch grid error".
A frequent problem with getting around patch grid error is the
simulation will spit a patch grid error before it can drop anything to
the DCD file. You can get around this by shrinking your timestep,
increasing (temporarily) the langevin dampening parameters, and
increasing the frequency at which you dump the DCD and/or restart
files. Sometimes if you increase the "margin" parameter (again,
temporarily) you can get around this. I prefer not to mess with
"margin" because it's easily forgotten once things are running, and has
the potential to reduce the performance of the simulation.
Regards,
Jeff
-- ====================================================================== Jeffrey J. Potoff jpotoff_at_wayne.edu Associate Professor and Director of Early Engineering Programs Department of Chemical Engineering and Materials Science Wayne State University 5050 Anthony Wayne Dr Detroit, MI 48202 http://potoff1.eng.wayne.edu ======================================================================
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:21 CST