From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Thu Jan 05 2006 - 15:48:36 CST
Short answer: Nothing really, it's just considered good practice to keep
it above that so that you're using a standardized level of precision.
Long answer: Not much more than the short answer. Obviously if you have
too few grid points you're going to suffer, but slight variaions below
one point per angstrom are unlikely to cause problems. I simply suggest
that people be careful so that they can have a standardized level of
precision upon analysis and reporting of results. I usually choose a
size with small positive integral factors that is slightly larger than
my starting box size.
>What happens when the PME grid density goes under 1 point per angstrom?
>The docs only mentioned that the PMDGridSize terms must only have small
>positive integer factors (2,3,5). My initial system setup had a box size
>of 54.7 so i picked the value 54. Will there be any problems with my
>Program in Molecular Biophysics
>Johns Hopkins University
>(410) 614 4435
>On Thu, 5 Jan 2006, Peter Freddolino wrote:
>>if you are running a restart, you should not respecify the periodic
>>system; this is updated by the restart.xsc file. So you should comment
>>those lines out of the restart conf file. Also, it is generally not a
>>good idea to cut things to close to 1 point per angstrom on the PME grid
>>density. You should always allow a little extra just to be sure that you
>>won't go under 1 point per angstrom (this isn't a magical boundary, but
>>it is generally helpful to be able to specify a minimum density when
>>discussing your results). I would recommend padding your PME grid sizes
>>a bit more. These two changes may help; it is very possible that
>>respecifying the wrong periodic cell dimensions on restart caused part
>>of the system to wrap into itself.
>>bo baker wrote:
>>>I try to run the MD without any restrains on. But I got the following
>>>ERROR: Atom 1804 velocity is 12661.2 -18890.9 10218.8 (limit is 5000)
>>>ERROR: Atom 1812 velocity is -12579.2 19072.4 -10061.2 (limit is 5000)
>>>ERROR: Atoms moving too fast; simulation has become unstable.
>>>ERROR: Exiting prematurely.
>>>The coordinates were from the prievious runs ( 20 ps of
>>>equilibration). I have included the .conf and .log files as
>>>Thank you for advices.
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:41:29 CST