Re: [AMBER] SMD average work

From: Adrian Roitberg <>
Date: Tue, 03 Mar 2015 12:04:06 -0500

Hi Mahmood

Please read some of the papers by Chris Jarzinsky and some from my group
and others on this issue.

You can of course compute the regular work average, but that is not very
useful, unless your speed for smd was truly VERY slow.

Otherwise you need to do the exponential average with the proper
weights. Again, please look at the literature on this.

If all works out, the final result should be related to the free energy
of the process. As always, drive carefully! That road is full of holes
and problems, so be very careful with the interpretations.


On 3/3/15 10:58 AM, Mahmood Jasim wrote:
> Hi AMBER users,
> I performing a steered MD calculations with 20 simulations on 20 starting
> structures of the same system. I am following the formation of a covalent
> bond with QM treatment for the region involving the formation of the bond.
> My reaction coordinate is an LCOD.
> My questions may sound stupid so forgive me. How should I calculate the
> average work across all the simualtions? Is it a simple numerical average
> or should it be expontential?
> Another questions; should I average the LCOD values as well? Although they
> are very similar between the simulations, they are not identical. The
> starting structures came from a conventional MD run in which the reaction
> coordinate was constrained.
> Regards,
> Mahmood Jasim
> _______________________________________________
> AMBER mailing list

Dr. Adrian E. Roitberg
Department of Chemistry
University of Florida
AMBER mailing list
Received on Tue Mar 03 2015 - 09:30:02 PST
Custom Search