[gmx-users] Can a Serial MD simulation be continued in Parallel?

[Mark Abraham]

Sridhar Acharya wrote:
> Dear GMX-users
> 
> I want to know whether a MD simulation that was run as a serial job 
> could be continued in parallel using tpbconv?
> If so, how to give the number of processors for the command?

There's no reason you couldn't use grompp with the .trr and .edr files 
as inputs to generate a .tpr for a different number of processors. There 
must be a numerical perturbation to your model physics in doing so - 
you're adding up lists of numbers in different orders - but the effect 
should be negligible within the MM approximation.

Mark