[gmx-users] nose-hoover implementation

[Michel Cuendet]

Hi guys,

I have submitted this message already, but forgot to edit the [gm-users] 
subject. As nobody has replied, I post it agin, with the right subject this time!

Is there a justification for not using a nose-hoover chain thermostat in
gromacs, or is it just not yet implemented?

 From what I know, the single nose-hoover thermostat does not in theory
reproduce a canonical ensemble, if there are other constants of the
motion for the unperturbed system than just the energy (eg. the
momentum, or center of mass velocity). See Tuckerman et al, JCP
115(4):1678,2001. It seems that for big (sufficiently mixing) systems
however this issue is not too severe.

There is another undesired effect : the Toda Demon, see Holian et al
Phys Rev 52(3) : 2338, 1995. If the coupling constant is not set
properly, the demon hidden in the Nose-Hoover dynamics will generate
slow modulations of the temperature, superimposed to the normal fast temperature
fluctuations. So the temperature variations will differ from those related to a 
canonical ensemble. A chain should cure this.

Also, shouldn't the nose-hoover extended variable be integrated together
with the particle positions and momenta, with a integration scheme more 
accurate than just xi[i] += dt*Qinv*(T - To) ?

Thanks,

Michel

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Michel Cuendet, Ph.D. student
Laboratory of Computational Biochemistry and Chemistry
Swiss Federal Institute of Technology in Lausanne (EPFL)
CH-1015 Lausanne						
Switzerland                            Phone : +41 21 693 0324
lcbcpc21.epfl.ch/~mitch	                       +41  1 633 4194
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++