Page 2 of 2 < 1 2
Topic Options
#36572 - 08/27/17 05:58 PM Re: water model for permeation study [Re: rmv]
stimme Offline
Forum Member

Registered: 12/03/07
Posts: 8
Dear Rmv,

I did the following:
- using charmm41, I ran the first script (no efie) five times - simulations were identical, and seed values were identical; when running a second script (with efie) I got different seed values in comparison to the first script, and again, running the second script five times led to identical seed values and identical dynamics;
-using charmm41, I changed the seed myself for all simulations - I get identical dynamics results for all five runs of the first script and identical simulations for all five runs of the second script;
- using charmm37, the results were the same, I even made five identical scripts and run them and got identical results.
In all the scripts I read the same .crd and restart file written after a short equilibration dynamics. Is this the explanation, that velocities are red from the .rst and are the same for all runs?
Thank you!

Top
#36573 - 08/27/17 10:49 PM Re: water model for permeation study [Re: stimme]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8079
Loc: 39 03 48 N, 77 06 54 W
With DYNA RESTART, the velocities are read from the restart file; if you want divergent simulations from the same initial model, you should use DYNA START.

Sometime after c37, the code was changed to auto-seed from a unique integer based on the time with DYNA START; previously, one had to set the seed manually to get divergent simulations.
_________________________
Rick Venable
computational chemist


Top
#36579 - 09/09/17 06:23 PM Re: water model for permeation study [Re: rmv]
stimme Offline
Forum Member

Registered: 12/03/07
Posts: 8
Thank you very much! I redid calculations with dynamics start and everything worked well. Indeed, all results on water permeation were random, regardless of efie value and direction.

Top
Page 2 of 2 < 1 2

Moderator:  BRBrooks, John Legato