[gmx-users] energy minimization problem

[Justin Lemkul]

On 5/16/15 8:30 AM, Ming Tang wrote:
> Dear Justin,
>
> I tried to use small time step. First,  set dt=0.005ps, after running 20ps, I found one chain of the triple helix just went to another water box totally. I guess there is something wrong with the martini ff because of the small time step.
> Then, I increased it to 0.01ps, after running 40ps, I checked the .gro, and found all the three chains stay together, which is reasonable. However, even use dt=0.01, the simulation can only run about 1ns. And, after simulating 100 ps using dt=0.01ps, the simulation can just run thousands of steps when dt is increased to 0.02ps. I tried many times and many different time steps, but still could not see the possibility for it to run hundreds of  nanoseconds, which is quite normal when using martini force. If the system can be further minimized, then it can run longer maybe.
>

This seems entirely random and suggests instead that there is something simply 
physically unstable with the system or problematic in the topology.  I don't use 
MARTINI (or CG models in general) so there's little else I can suggest.  Maybe 
someone more experienced with such systems will chime in.

Also note that a chain "moving into another box" is probably just a PBC effect 
and not a true dissociation.  If it is a dissociation, it's not simply because 
of the time step.

-Justin

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

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

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