Re: [ESPResSo-users] DPD interaction

[Ulf Schiller]

On 11/19/2013 09:20 PM, Dr. Jens Smiatek wrote:
> On 11/19/2013 08:02 PM, Vincent Ustach wrote:
> > Hi everyone,
> >
> > I want to understand DPD in ESPResSo a little better. If I wanted to
> > obtain a diffusion coefficient for one colloid (say for a protein) in
> > a liquid on the order of 10^-7 what knobs would I turn? Is this
> > possible? Since DPD imposes friction on DIFFERENCES in velocities for
> > particles within the cut off radius, am I correct in thinking you need
> > a relatively crowded system to obtain that damping? For T = 1.0 and
> > volume fraction < 0.15 I am seeing diffusivity values higher than gas
> > particles.
> >
> > p.s. You may recall I am working with lattice Boltzmann to get
> > hydrodynamics. As such I am only looking for understanding.
> >
> >
> >  inter 1 1 lennard-jones 1.0 0.30 0.33673 0.25 0
> >  setmd time_step 0.001
> >  setmd skin      0.4
> >  thermostat dpd 1.0 1.0 0.16837
> >
> > etc...
> >
> >
> > Best Regards,
> >
> > --Vincent Ustach
>
> Dear Vincent,
>
> whatever you do with DPD, it mostly behaves as a gas. This becomes
> specifically obvious by regarding the Schmidt number.
[...]
> For all of these reasons, DPD would not be the method of choice for the
> considered system if you are really interested in correct diffusion
> constants.

Just a small addendum: There is a difference to be made between 
self-diffusion and diffusion of solute molecules. As long as one does 
not seek to reproduce a specific equation of state, a gas-like solvent 
is not in principle a problem. In fact, most mesoscopic methods (MPC, 
LB) have an ideal gas equation of state and can still produce correct 
diffusion coefficients. The challenge is more in modelling the colloid 
and its interactions with the solvent, and a simple point-like colloid 
is often not sufficient (for the reasons Jens explained). For DPD, the 
Schmidt number and transport coefficients have indeed a limited range, 
which does especially affect non-equilibrium properties of solute 
molecules, however, I'd expect the latter effect to be of less 
importance for colloids (w/o internal degrees of freedom) than for, 
e.g., polymers. In any case, Schmidt numbers up to 10^5 can easily be 
achieved using the Lowe-Anderson thermostat for DPD. Certainly, in a 
bulk system SD/BD would be more efficient, but isn't easily generalized 
to confined systems where HI become anisotropic. Of course, things are 
different in a melt/concentrated suspension when HI are screened and 
instead lubrication corrections may be needed. In my view, the Skolnick 
method does not address this situation satisfactorily, and I'd recommend 
checking results carefully against other methods, in case you are 
inclined to try.

Cheers,
Ulf

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

smime.p7s
Description: S/MIME Cryptographic Signature