[gmx-users] Re: Re: gmx-users digest, Vol 1 #653 - 10 msgs

[Michel Cuendet]

Hi everyone,

A few precisions about thermostats:

- The Berendsen thermostat is a brute force temperature regulation, but 
it does not reproduce the right phase space distributions.

- The Nose-Hoover thermostat reproduces a correct NVT (canonical) 
distribution, ONLY IF the only conservation law for the system is the 
physical Hamiltonian. If there is more than one conservation law, like 
the center of mass motion for example, the simple Nose-Hoover thermostat 
fails.

- The Nose-Hoover CHAIN thermostat produces a right NVT ensemble in the 
particles positions, even if there are several (identified) conservation 
laws. The momentum distribution is also correctly reproduced, apart from 
the angular components of the center of mass momentum. Provided that the 
Hamiltonian does generally not depend on these degrees of freedom, the 
distributions will nevertheless be correct.

- It is not clear to me what kind of phase space distribution the 
Parrinello-Rahman barostat gives. Does someone have a hint ?

- Another example is the Martyna-Tobias-Klein algorithm, which is proved 
to generate the correct momentum and configuration space distributions 
as well as a correct volume distribution of the NPT ensemble.  Maybe one 
should consider this one for future implementation ?

An excellent reference for the thermostat question is
Tuckerman et al., J. Chem. Phys. 115(4) : 1678 (2001)

Cheers,

Michel


Michael Shirts wrote:

>Actually, this might be considered more a problem with Berendsen temperature
>scaling methods, which additionally simulate an ensemble that is not actually
>NPT; a temperature control mechanism such as an Andersen thermostat (periodic
>stochastic collisions with particles in the box with a time constant
>sufficiently not to mess up the dynamics much) will not exhibit such partition
>of energy into seperate modes.  I'm not sure about Nose-Hoover temperature
>control methods - anybody know?
>
>Cheers,
>Michael Shirts
>
>>Date: Tue, 04 Mar 2003 14:13:08 -0600
>>To: gmx-users at gromacs.org
>>From: Eric Jakobsson <jake at ncsa.uiuc.edu>
>>Subject: Re: [gmx-users] Re: gmx-users digest, Vol 1 #653 - 10 msgs
>>Reply-To: gmx-users at gromacs.org
>>
>>This is a particularly egregious problem with PME.  We published on this in
>>J. Comp. Chem. a couple of years ago.  It is the "flying ice cube"
>>effect--as the translational motion builds up the heat bath keeps v squared
>>the same so the temperature goes down, and if you let it go you have an ice
>>cube hurtling through space very fast.
>>
>>As Erik says, the solution is to keep resetting the center-of-mass motion
>>to zero.
>>
>>Eric
>>
>
>
>Cheers,
>Michael
>
>
>_______________________________________________
>gmx-users mailing list
>gmx-users at gromacs.org
>http://www.gromacs.org/mailman/listinfo/gmx-users
>Please don't post (un)subscribe requests to the list. Use the 
>www interface or send it to gmx-users-request at gromacs.org.
>