AMBER Archive (2009)

Subject: Re: [AMBER] Averaging simulations

From: Carlos Simmerling (carlos.simmerling_at_gmail.com)
Date: Thu Apr 02 2009 - 10:24:28 CDT


well you can't calculate the values for each run separately and then average
the results, since they my change order for example, but you might combine
the trajectories and do the analysis on the combined data. i would be very
careful about saying "boltzmann sampling" for multiple short runs. it's
clearly not that simple, unless you are exploring a very small phase space.

On Thu, Apr 2, 2009 at 11:21 AM, Naser Alijabbari <na3m_at_virginia.edu> wrote:

> When running multiple short MD simulation (using sander and AMBER10) using
> different initial velocities (ntx=1, irest=0, ig=-1), does it make sense to
> average the eigenvalues and eigenvectors from individual runs, as a means
> of
> determining average properties? The idea would that by by somehow combining
> individual trajectories one can also achieve Boltzmann sampling. But the
> question is how does one combine individual trajectories.
> Naser
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber