[MMTK] Relaxation times, relation to W_S and W_V (Kneller et al.)?

Andreas Kring akring1729 at gmail.com
Tue Feb 19 11:56:12 UTC 2008


Konrad Hinsen skrev:
> On Feb 19, 2008, at 9:54, Andreas Kring wrote:
> 
>> How are the relaxation times for the NoseThermostat and the
>> AndersenBarostat related to the parameters W_S and W_V (Kneller et al.,
>> Phys. Rev. E, 54, 6825 (1996)), respectively?
>>
>> Are the related in the following way?
>>
>> Thermostat:
>> W_S = 3*N*k_B*Temp*(rexalation time)**2,
>> where N is the number of atoms in the system.
>>
>> Barostat:
>> W_V = [P_ext*(relaxation time)**2]/(some reference length)**3
> 
> In the MMTK source code (MMTK_dynamics.c), the parameter W_S is called 
> t_mass, and W_V is called b_mass. For the thermostat, lines 502 and 503 
> define:
> 
>     t_energy = df*t_temp*kB;
>     t_mass = t_energy*t_tau*t_tau;
> 
> where df is the number of degrees of freedom (3N minus the number of 
> constraints), t_temp is the temperature of the thermostat, and t_tau is 
> the thermostat relaxation time. kB is just the Boltzmann constant. The 
> formula that you give is thus correct in the absence of constraints.
> 
> For the barostat, line 639 defines
> 
>     b_mass = df*t_temp*kB*b_tau*b_tau/(volume*volume);
> 
> where b_tau is the barostat relaxation time and volume is the 
> (instantaneous) volume. If there is no thermostat, the kinetic energy of 
> the thermostat (df*t_temp*kB) is replaced by the instantaneous kinetic 
> energy of the system (line 692):
> 
>       b_mass = 0.5*k_energy*b_tau*b_tau/(volume*volume);
> 

Ok, great - that clarified things. Thanks for your quick answer :-)

Kind regards
Andreas



More information about the mmtk mailing list