[gmx-users] Reproducing runs

[Mark Abraham]

Ranjani Krishnan wrote:
> Hi,
> 
>  
> 
> I have noticed that I cannot reproduce MD runs exactly -- running the 
> exact same .tpr file again gives different results. From previous posts, 
> I have seen some explanations for this. However, if I do AFM pulling 
> simulations, I am able to reproduce them exactly – running the same 
> simulation twice gives identical results.
> 
>  
> 
> Why is it possible to reproduce runs with AFM pulling, but not otherwise?

Because there's a source of randomness with your other ones that isn't 
present for the AFM ones. Without knowing your compute configuration, 
.mdp options and/or grompp command line, we're hard-pressed to say why. 
Also, check out http://wiki.gromacs.org/index.php/Reproducibility

Mark