[gmx-users] restarting a replica exchange simulations

[Mark Abraham]

sarbani chattopadhyay wrote:
>   Hi,
>     I had previously ran a replica exchange molecular dynamics 
> simulation for 10ns. There
> were 17 ".tpr" input files for 17 replicas at 17 different temperatures.
>             However, the run had stopped and now I need to restart it. I 
> had generated 17 ".tpr"
> files for the rest of the simulation left, using the  command "tpbconv", 
> that is usually used for
> generating ".tpr" file to restart a "md" run.
>             Then again I gave the command for "mdrun" using "replex" 
> option such that the input
> files are the 17 new restart ".tpr " files.
> 
> Is this the right way to restart the replica exchange molecular dynamics 
> simulation? Will the
> trajectories of the 17 replicas start from the point inteneed and 
> perform replica exchange as
> per the command?

Under 3.x you need to apply a suitable, consistent tpbconv to each 
independent simulation - see 
http://wiki.gromacs.org/index.php/Doing_Restarts. Under 4.x the restart 
mechanism has changed, and the checkpoint file is now the main thing you 
have to take care of. I'll update the wiki once I've needed to work out 
how that works.

Mark