[gmx-users] stochastic dynamics , langevin

[Matthew Hoopes]

Along these same lines...
(a)is there an thermostat setting in GROMACS that comes close to the DPD
thermostat as described in (T. Soddemann, B. D¨unweg, and K. Kremer.
Dissipative particle dynamics: A useful thermostat for equilibrium and
nonequilibrium molecular dynamics simulations. Phys. Rev. E, 68:046702,
2003.)
(b) if nothing approaches DPD, how would a one go about starting to write a
new thermostat in GROMACS?

-Matt

---------------------------------------------
Matthew Hoopes
Biophysics Graduate Group
University of California, Davis

mihoopes at ucdavis.edu
530-752-6452
--------------------------------------------- 


Hi,

Brownina dynamics is a Langevin equation for the coordinates (no inertia).
Stochastic dynamics is a Langevin equation for the velocities (with
inertia).

Everything depends on what you want to do, which you do not tell in detail.
If you want to leave out the solvent, but you want to simulate a system in
solvent, SD is not going to help you, since there is no implicit solvent
potential, so your potential and therefore your sampling is nonsense.

tau_t has no effect on the distribution, only on the dynamics. If you want
the correct dynamics, you will have fit tau_t to reproduce some kinetic
quantity that you are interested in. For different quantities tau_t can be
different.

Berk


Date: Thu, 23 Oct 2008 15:31:16 +0200
From: omermar at gmail.com
To: gmx-users at gromacs.org
Subject: [gmx-users] stochastic dynamics , langevin

Dear All,
I am trying to perform Langevin dynamics of large peptides / proteins. After
reading the manual & going over some old mails in this list, I have two
points I hope you could clear for me:

[Gromacs version 3.3.3]

1) I am a bit confused with Brownian vs. Langevin dynamics. Is this the
proper keywords I should use for Langevin dynamics: integrator = sd ;

bd-fric    = 0 ;
tau_t      = 10 ;
ref_t      = 300 ;
With bd-fric=0, the friction is taken as the inverse tau_t.


2) From your experience, what are good values of tau_t (or 1/tau_t) for
simulating a protein? In 2006 list, David has commented that choosing tau_t
is very important (
http://www.gromacs.org/pipermail/gmx-users/2006-July/023089.html ).


Your help is appreciated. Omer.  
Koby Levy research group,
Weizmann Institute of Science. 
http://www.weizmann.ac.il/sb/faculty_pages/Levy/