AMBER: temperature artifacts with andersen and GB?

From: David Mobley <>
Date: Wed, 11 Apr 2007 09:50:17 -0700


I know the manual highly recommends using Langevin for temperature
control when using GB, but I insisted on trying Andersen anyway while
simulating a small molecule. I find that the average temperature (in
vacuum, or in GB solvent) is consistently too low (i.e., average of
250K over 500000 steps, when thermostatting to 300K, with vrand=500).
Looking at the instantaneous temperatures from the energy file, which
I'm writing every hundred steps, I find that immediately after every
randomization, the "temperature" spikes to about 2800K, then wobbles
around somewhere in teh 100-250K range until the next randomization.

I'm now going to Langevin, which seems to be working fine. But I don't
see any reason why the Andersen thermostat should give this behavior,
so it might be worth looking into.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Apr 15 2007 - 06:07:07 PDT
Custom Search