AMBER Archive (2005)

Subject: Re: AMBER: REMD error!!!

From: Carlos Simmerling (carlos_at_ilion.bio.sunysb.edu)
Date: Thu Oct 20 2005 - 11:14:25 CDT


Guanglei Cui wrote:

> Hi, Anthony
>
> Your trajectory may have been corrupted. Perhaps just a few frames in
> the middle. It's not straigtforward to fix it. But restarting can be
> easier. Since REM sampling is essentially Monte Carlo, you can just
> take a snapshot of the replica that's restrt missing, and equilibrate
> it by a short MD simulation to just get the velocities. Then you can
> get REM going again. Perhaps before you restart, you may want to check
> if the sampling you've already had may have converged for the
> properties you're interested in.

that's true, but getting the snapshots for that replica is not completely
straightforward, especially if the trajectory has corrupted frames.
If the structures for eacch replica were very different then you might
be able to visually see which one was the missing one by looking at
all of the other restarts and then checking one or more trajectory
files that were not corrupted.

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu