[gmx-developers] CVS implementation of REMD

[Berk Hess]

Ken Rotondi wrote:

> Hello all,
>
> Okay, I've been wearing another hat for the last few days. It would 
> appear that I've figured this out and wanted to post for the general 
> good.
>
> Berk:
>
> I'm running the VERSION 3.3_beta_20050202 CVS code:
>
> I was worried that trying to simulate only 2 replicas was somehow 
> bollocksing up the works, so I created 4 .tpr files named (in 
> increasing temperature) topol0.tpr, topol1.tpr, topol2.tpr, & 
> topol3.tpr using the:
>
> grompp -f 1.mdp -po 1out.mdp -c md8.gro -p p39a.top -o topol0.tpr
>
> command
>
> The
> mpirun -np 4 mdrun -multi -replex 200 -reseed 1
>
> command would appear to start a REMD run (4 sets of "starting mdrun 
> 'protein in water' tags appear).
>
So it works now?

> Next question, are the acceptance ratios for exchange part of the 
> standard output? If not how could one extract them (they are 
> important, so I assume that they would be part of the output)?
>
You haven't looked at the output yet?
grep Repl md0.log
will give you all the replica exchange information.
It shows the exchange probabilities and exchanges at every exchange step
and at the end the number of exchanges and acceptance ratios.

Berk.