[gmx-users] overlapping trajectories in multiple MD runs

[Justin A. Lemkul]

Lizhe Zhu wrote:
> Dear all,
> 
> I have run multiple runs of a system using the following command:
> 
> mdrun_mpi -deffnm md -multi 8 >& main.log
> 
> in which the md.tpr is obtained from grompp with md.mdp file containting 
> the following parameters to get different initial velocities for each run:
> 
> ###########################
> gen_vel                  = yes
> gen_temp               = 300.0
> gen_seed                = 173529
> ###########################
> 
> But still out of these 8 parallel runs, there are always certain number 
> of trajectories overlapping each other.
> 
> Anyone has a clue?
> 

If you are using the same .mdp file (the same gen_seed), then theoretically the 
trajectories should actually be identical.  Given the inherent chaos in MD and 
other optimization factors (like dynamic load balancing), you will probably see 
some divergence.

But if you want different initial velocities, use a different seed for each.

-Justin

> Many thanks in advance.
> 
> Lizhe
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================