espressomd-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ESPResSo-users] Educated guess about gamma in Langevin thermostat


From: Ulf Schiller
Subject: Re: [ESPResSo-users] Educated guess about gamma in Langevin thermostat
Date: Fri, 11 Jan 2013 17:40:01 +0100
User-agent: Mozilla/5.0 (X11; Linux i686 on x86_64; rv:17.0) Gecko/20130107 Thunderbird/17.0.2

Hi Salvador,

On 12/19/2012 06:34 PM, Salvador H-V wrote:
I am doing some simulations for a two-dimensional  hard-sphere
rigig-dumbbells.
The interaction potential is the purely repulsive Lennard-Jones (WCA)
using the rigid_bond feature to constraint the bond in the dimers.

I would like to choose a value  of the the friction coefficient (gamma =
6.0*Pi*dynamic_viscosity*sphere_radius / mass  )  in the Langevin thermostat
such as is representative of the solvent experimental viscosity.

Using the experimental data, I obtain the following
M ~ 4.4x10^(-15) kg
T ~ 293.15 K
tao = sigma * ( mass / kbT)^1/2 ~ 2.08x10^-3 s
viscosity ~ 0.001002 Pa * s
sigma = 2x10^-6 m

Then, gamma_langevin = 6 * Pi * viscosity * radius / mass
and in reduced units  gamma / tao = gamma_reduced ~ 8930

If my above simple calculations are right and if I understood well,
accordingly to previous post in the mail list,  we have to use a value
of time_step
such as:  gamma_reduced * dt  / 2.0 is < 1 and preferably around  0.1.

Then, I should use a time_step <= 0.00002 that is very small and will
require very long simulations to obtain the experimental time window.

I was wondering if somebody could provide suggestions of how to reduce
the value of gamma (so, i can increase the time_step) but still
representing the solvent viscosity.

I think that depends on what physics you want to look at. If inertial effects can be neglected in your system, you could try to use an artificial mass instead of matching the experimental mass. Keep in mind that a too high value of gamma will also affect the accuracy of the integrator (the Verlet algorithm with velocity dependent forces is first order only).

Another possibility is to think whether you really need the exact value of the viscosity. It might be enough to match dimensionless quantities such as the Reynolds number, Peclet number, etc. One can then typically scale the simulation parameters to obtain reasonable values. The details depend again on what you want to simulate/measure.

Hope this helps,
Ulf

--
Dr. Ulf D. Schiller                        Building 04.16, Room 3006
Institute of Complex Systems (ICS-2)       Phone:   +49 2461 61-6144
Forschungszentrum J├╝lich, Germany          Fax:     +49 2461 61-3180


Attachment: smime.p7s
Description: S/MIME Cryptographic Signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]