[gmx-users] output frequency modification

[Mark Abraham]

Stanislaw Bobritsky wrote:
> Hello all!
> I have partially computed trajectory (several ns), but I need to modify 
> one simulation option at the current time: the output frequency. And to 
> continue the simulation with new output frequency from that point.
> May I "decompile" a .cpt checkpoint file in one way or another to get 
> .tpr with my modified option?
> Or how can I do this in another way?

I believe you can use the original .mdp, change the nxxxout settings, 
and regenerate the .tpr with grompp, and use that with your current .cpt 
as input to mdrun. The .cpt has only the current state information 
needed by the various algorithms, and not time-independent quantities 
(like reference temperatures for T-coupling or output frequencies). The 
original (and regenerated) .tpr hold both sorts of information, but the 
time-dependent information is over-ridden if a .cpt is supplied to 
mdrun. It will be trivial to verify with "mdrun -reprod" that runs with 
the old and new .tpr produce the same simulation with different rates of 
output.

Mark