AW: patch becoming too small

From: Norman Geist (
Date: Tue Aug 09 2011 - 02:41:19 CDT


constant area tells namd to only let changes to the dimensions of z happen,
this only would help you, if you do constant pressure with a system that may
not get smaller within x and y, for example if you have fixed atoms near the
boundary of x or y. Then the simulation will always crash with constant
pressure, because namd tries to make the box smaller, but can't move the
fixed atoms back into the box.

Generally the "pbc box has become too small for original patch grid" error
comes due to a shrinking box due to pressure, which gets smaller then the
PME patch grid, which was generated at the begin of the simulation. You can
force namd to initialize this patch grid by just restarting the simulation.
So instead of running 100000 fs of constant pressure, just do 10*10000 steps
and look if that helps.

You can also try to let the box not shrink to fast, by setting higher values
for damping and period in langevin pressure.

If you don't use constant pressure, you maybe just set wrong box dimensions.

When does your simululation crashes (which step), if it is initially, you
have maybe wrong dimensions or fixed atoms near boundary, if it crashes far
later, its to constant pressure and restarting should solve it.

Best wishes

Norman Geist.

-----Ursprüngliche Nachricht-----
Von: [] Im Auftrag
von Jorgen Simonsen
Gesendet: Dienstag, 9. August 2011 08:15
Betreff: namd-l: patch becoming too small

Hi all,

I am doing a membrane simulation with a protein and I get the following

FATAL ERROR: Periodic cell has become too small for original patch grid!
Possible solutions are to restart from a recent checkpoint,

so I believe I have equilibrated the system rather well - the change
in volume had a gradient close to zero and the total energy has also
been equilibrated it is oscillating around a certain value. When I
restart the system the simulation continues and the observables such
as total energy and volume are still oscillating around the same
values as before(first part of simulation 9 ns and second part 9 ns).

I am using periodic boundary conditions but I have not set the
constant area to true

useConstantArea no ;

could this resolve the problem?


This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:57:33 CST