Re: [AMBER] Random seed for restart files that have velocities?

From: David A Case <>
Date: Wed, 3 Feb 2021 08:44:00 -0500

On Tue, Feb 02, 2021, Liao wrote:

>Yes I’m using the Langevin thermostat. So does that mean if I save my MD
>half way but also keep it running, when I restart another one from the
>saved restart file, the new trajectory will always be different from the
>one that kept running?
>What if I wanted to exactly reproduce the continuous trajectory from the
>half-way saved restart frame?

Amber doesn't provide any way to do this using a restart file saved part way
through the trajectory: there is no record of what random number seed was
being used at that point.

You can re-run a trajectory from the beginning, because the seed that was
actually used is printed in the output. Use that value for ig in your
re-run. But you can't use CPU parallel for this, since that code is
non-deterministic. And you should not re-use an old seed for any other

...hope this helps....dac

AMBER mailing list
Received on Wed Feb 03 2021 - 06:00:04 PST
Custom Search