My three cent's of opinions worth for today.
I insist that we stop talking about equilibrium, when we clearly mean
'relaxation'.
Equilibrium, or equilibrated simulations are of course impossible to
achieve. They would have to run for exactly: forever.
What we might to know is, if we keep running will the answer change ?
Again, tricky, tricky question. Of course it will change!
Current ideas tells you that you should run multiple simulations for
quite a bit of time, and compare those answers.
Now, for relaxation runs, one should NOT even look a the answers until
the important properties have 'stabilized', by which I mean plateaued.
Look at the density if using NPT, NOT at the pressure (which fluctuates
a lot).
The fluctuations in the total energy should be roughly the same as those
for potential energy.
The RMS vs the original structure should be more or less becoming
constant. This is not as trivial as it looks. If you start from a
structure very different from your final result, the RMSD will indeed
look constant but that is a simple artifact of the behavior of RMSD when
structures are very different from each other.
If teh RMSD becomes 'stable' at values smaller than 4 or 5, them it is
ok to think of the system as relaxed.
Cheers
Adrian
Bill Ross wrote:
>> I want to know the criterion of equilibrium of molecular dynamics, when
>> analysize the summary.ETOT, what can be called the system has equilibrated,
>> in other words, how much the energy can diviate from the average?
>
> I think the thing to watch is not so much the short-term fluctuations,
> but rather the long-term trends of various quantities, to see if they
> have leveled off.
>
> Bill
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Dr. Adrian E. Roitberg
Associate Professor
Quantum Theory Project
Department of Chemistry
Senior Editor. Journal of Physical Chemistry
American Chemical Society
University of Florida PHONE 352 392-6972
P.O. Box 118435 FAX 352 392-8722
Gainesville, FL 32611-8435 Email adrian.qtp.ufl.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Oct 24 2009 - 11:30:02 PDT
