[gmx-users] interrupting mdrun

[Berk Hess]

>On Tue, 2004-03-23 at 20:10, Martina Bertsch, PhD wrote:
> > I am running gmx 3.1.4. Can I stil kill mdrun with -USR1?
> > yes but your continuation run will not be identical to the one that will
>be killed. This has been fixed in 3.2.1. The differences arev ery small
>though, and (in my opinion) only relevant in special cases where you run
>without Temp and Pres coupling.

If you run without temperature and pressure coupling (and free energy)
you already got exact continutation in all previous versions.
As of 3.2.0 temperature coupling information is added to the energy file
and exact restarts are now also possible with temperature and pressure 
coupling.
However in most cases you don't care about exact continuation, since MD
simulations are chaotic and the error in the continuation is much smaller 
than
the accuracy of the simulation.
The only important case I use it for is when I want to redo part of a 
simulation
with more frequent trajectory output. In this case you want this run to be 
exactly
identical to the original one.

Berk.

_________________________________________________________________
MSN Search, for accurate results! http://search.msn.nl