[gmx-users] Fw: energy minimization problem

[Justin Lemkul]

On 11/4/13 8:55 AM, kiana moghaddam wrote:
> Dear Justin
>
> It is obvious that emtol value can not be zero or negative. but you wrote in
> your email " For more sensitive calculations like free energy simulations and
> normal modes, you will want to minimize much more thoroughly (for NM, emtol <
> 1) and in double precision". what did you mean by saying emtol<1? Anyway, my

I mean exactly what I said.  You should set emtol to some value of 1 or less as 
a target for NMA.

> question was: for free energy calculation in double precision, what should be
> the best value for emtol?
>

I have not worked with systems that have required such strenuous minimization, 
but I believe the general recommendation (if necessary for whatever you're 
working with, and I'm not prepared to try to guess about what you're doing) is 
to minimize as far as you possibly can.  Set some very low value of emtol and 
some very high value of nsteps and let the EM algorithms churn away.  You will 
usually need several rounds of EM using different methods to achieve this. 
Published literature and previous posts to this list (in the archive) should be 
sufficient to guide you.

-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

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