[gmx-users] restarting MD after error in the middle of trajectory

[Justin Lemkul]

On 1/7/14, 11:34 AM, didymos wrote:
> Hi,
>
> I have strange situation.
> I am running MD with 4 replicas and when I recently checked the trajectory
> with gmxcheck in one replica I got this error:
>
> Fatal error:
> Magic Number Error in XTC file (read 0, should be 1995)
>

That usually means the trajectory was corrupted.

>
> It is arround 50ns of the simulation, and right now I have 120ns.
> I would like to restart my simulation from the part before 50ns but I am not
> sure how to do this because my cpt file are written after 120ns...
> Can I generate new cpt file based on the part of the simulation before 50ns?

Unless you have saved all of them (mdrun -cpnum), which is not the default 
behavior, the previous states are lost as checkpoint files are recycled.  You 
could, in theory, extract matching time frames from an .edr and .trr file and do 
an old-style restart from those frames, i.e. 
http://www.gromacs.org/Documentation/How-tos/Extending_Simulations#Version_3.3.3_and_Before. 
  You will not get exact continuation, but the resulting trajectory should be 
reasonable.

-Justin

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

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441

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