[gmx-users] Box dimension size errors in MARTINI soft core simulation

[Justin Lemkul]

On 8/1/13 11:35 AM, Scott Pendley wrote:
> Justin,
>
> Much appreciation for the suggestions.  I found that the large time
> constant (tau_p) on the pressure coupling was indeed a significant
> contributor.  When I dropped that value down to 2.5, the box vectors scaled
> down in a more reasonable manner maintaining the cubic simulation box.
>
> With regard to your other questions:
> Visualizing is possible but is more technical to do with MARTINI beads.  I
> am still trying to figure out how to do this on vmd.
>
> The prior npt run also used parrinello-rahman but with a tau-p of 2.0.
> This may have contributed to some of the instability seen in this system so
> I may switch it to berendsen per your suggestion.
>
> I agree, isotropic coupling would be better for this system.
>

Good plan.  An initial relaxation with the weak coupling method usually leads to 
fairly robust simulations.

> Some initial runs suggests that turning off the free energy portion does
> allow the simulations to run to completion (for short simulations of 10 ns)
> but I haven't analyzed them to determine if the sampling time is
> insufficient to reveal developing instabilities.
>
> Contrary to logic, at lambda = 0 this simulation instability also occurs.
>

There have been numerous fixes to the free energy code since 4.5.5 was released. 
  I would suggest using version 4.6.3.  If it still has issues, then it will 
receive significant attention as it would probably be a bug needing fixing.  I 
think I recall a similar problem being reported that was already fixed.

> The LJ changes are composed on contributions from both solvent-solvent
> interactions and solvent-solute.  I am seeing a ~ 3000 kj/mol different
> that grows but plateaus out before the system crashes.  I would be hesitant
> about ascribing the instability to this term as the decrease in LJ and
> total energy would be expected at lamba=0.4 and the energy reached its
> maximum prior to the system instability.
>

OK, if the same issue shows up in 4.6.3, let's explore it further.  That seems 
to be the most immediate solution.

-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

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