AMBER Archive (2009)

Subject: RE: [AMBER] why is TAUTP scaled by "e" ?

From: Ross Walker (
Date: Thu Oct 15 2009 - 17:08:52 CDT

> >From my simulations, it seems that the time evolution of the
> temperature
> upon heating the system from 300K to 550K with TAUTP=5000psec (very
> slow
> heating) in the input file results in the system being heated with an
> effective TAUTP=13570=e*5000
> Does anybody know why this re-scaling of the TAUTP factor?

Are you sure of this? There should not be any scaling going on. Note that
tautp is just the scaling for the thermostat, it does not account for the
response of your system. There will always be a lag in the way your system
responds to the thermostat so it is reasonable to think that it takes longer
to reach the target temperature. Where did you come up with the value of
13570? Also note that you could be losing energy during the simulations so
this would make it take longer for the thermostat to heat your system. Try
setting the shake tolerance an order of magnitude tighter. Also set dsum_tol
an order of magnitude smaller.

Also note that ntt=1 (Berendsen) does not effectively equilibrate the
system. It does nothing to ensure that there is thermal equilibrium in the
system. Try heating with Langevin instead (ntt=3) it will do a much better
job of equilibration.

All the best

AMBER mailing list