[gmx-users] tau_t and tc_grps for v-rescale (2)

[Justin Lemkul]

On 9/28/12 3:18 AM, Felipe Pineda, PhD wrote:
> Hi again,
>
> I posted yesterday the query below, but have not received any feedback up to
> now. I would like to put my question in another way: in Justin's "CAP-15 in
> DPPC" tutorial he uses during NVT equilibration Bussi's thermostat (V-rescale)
> together with three separate coupling groups:
>
> ; Temperature coupling is on
> tcoupl          = V-rescale                 ; modified Berendsen thermostat
> tc-grps         = Protein DPPC SOL_CL   ; three coupling groups - more accurate
> tau_t           = 0.1   0.1     0.1             ; time constant, in ps
> ref_t           = 323   323     323             ; reference temperature, one for
> each group, in K),
>
> Is this separate coupling needed at all taking into account that this
> temperature coupling method produces the correct coupling? Is it probably better
> to use a single coupling constant (system) and so avoid problems using this
> artificial correction to deal with the "hot solvent / cold solute" artifact?
>

Technically the correct way to do it is to use one global thermostat rather than 
individual ones, but multiple thermostats are used for the reason described.  I 
think it is actually more to do with electrostatics approximations than the 
thermostat itself (i.e. plain cutoffs leading to poor energy conservation and 
thus heating of the solvent, which diffuses rapidly) than anything else, so 
using PME is more robust.  Someone please correct me if I'm remembering wrong. 
I don't recall having seen a test of an inhomogeneous system with a single 
thermostat, so multiple thermostats remain common practice.  There are numerous, 
extensive discussions in the list archive about these topics so I would 
encourage you to search around a bit.

> Could somebody kindly tell me how large could the coupling constant, tau_t, be?
> Is it OK to switch, e.g., from 0.1 to 0.3 ps after T equilibration is reached,
> i.e., during the production phase?
>

This issue is discussed in the reference for the V-rescale thermostat (Bussi et al).

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================