Re: AMBER: langevin, GB and simulated annealing - 2

From: David A. Case <case.scripps.edu>
Date: Tue, 19 Apr 2005 17:04:56 -0700

On Tue, Apr 19, 2005, pascal.baillod.epfl.ch wrote:
>
> But SA doesn't work for the same input file, in which I replace the Berendsen
> thermostat (NTT=1) by Langevin dynamics (NTT=3) with the collision frequency
> (GAMMA_LN) set to 91. The temperature jumps from 0 to 300 K in 50 steps and
> doesn't change for the rest of the simulation.

Generally, Langevin dynamics takes the system to the target temperature very
quickly, even when gamma_ln is quite small. This makes it different from the
Berendsen thermostat, where the rate of temperature change is controlled by
the tautp variable.

To use Langevin dynamics, you could set a series of target temperatures,
(say 600, 500, 400, 300, 200, 100, 10, etc.) to take effect at different times
during the simulation. This could have the effect of fairly slow cooling.
But the Berendsen approach seems to me to be a more natural way of carrying
out simulated annealing.

.....dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Apr 20 2005 - 01:53:00 PDT
Custom Search