Re: [ESPResSo-users] DPD

[Dr. Jens Smiatek]

Dear all,

just as a remark:
@Stefan: You are right. Lower densities will result in the Boltzmann
regime where the standard Marsh, Backx, Ernst-Theory will not be
applicable anymore.
Anyway, we have found that the tunable-slip boundaries can be
effectively tuned and analytically described for number densities
between 3 - 10 (in agreement with the DPD method itself).
Otherwise, the collision numbers per particle in one timestep are too
low and a description within a mean-field approach fails.
Therefore I suggest that using a number density of 3 for DPD is more or
less mandatory for (more or less) analytical control.
You should always remember the fact that the DPD method produces a very
low Schmidt number which corresponds more to a gas than a fluid.
With regard to this, lower densities are also in terms of a physical DPD
interpretation as a solvent doubtful.
And last but not least...
C'mon guys, everything below 100K particles is a small simulation ;-)

Best,
Jens


On 09/11/2014 04:20 PM, Stefan Kesselheim wrote:
> Dear Dusan,
>
> On Sep 11, 2014, at 4:07 PM, Dudo <address@hidden> wrote:
>
>> Dear Jens,
>>
>> thank you very much for the lead.
>>
>> Well, so I've compiled in a feature for "TUNABLE_SLIP"
>> and I have set up interactions: inter $cid $idpolym tunable_slip 
>> $temperature $gamma_L $r_cut_L $timestep .....
>> I have turned the flag to 2: constraint cylinder center $cx $cy $cz axis 
>> $cnx $cny $cnz radius $crad length $clength direction $cdirection type $cid 
>> reflecting 2 .....
>> as Stefan suggested, and I'm trying with the parameters from your paper on 
>> polyelectrolytes J. Phys Chem B 2010..
>> Now I see you have used explicit solvent with density 3.75, while modelling 
>> a chain of 20 beads.
>> In my case this would mean zillions of solvent particles..
>> Well at the begining Chris asked me, why would I do add explicit particles - 
>> so.. is there a way around?
> The particle numbers that you will need are necessary are considerable but 
> OK. For a chain of 20 beads, the radius of gyration is probably around 5, and 
> thus you need a box size of 10, and 20 is still doable. With a density of 1, 
> you'll have 1k to 8k particles. That should be enough to perform the 
> measurements, especially as you probably don't need too many steps to get a 
> meaningful mobility. 
>
> In principle you can use even lower densities, but at some point the mean 
> free path of the particles becomes comparable to the box size, and then you 
> are in a region where you should be sure what you are doing.
>
> The strength of the hydrodynamic interaction, anyways, depends in the 
> "hydrodynamic radius" of the particles. This is the size of a sphere with the 
> same diffusion coefficient as an isolated particle with the same viscosity. 
> If this quantity is too small (~1/(6 pi visc D), then hydrodynamics will be 
> weak.
>
> Cheers and good luck
> Stefan


-- 
================
Dr. Jens Smiatek

Institute for Computational Physics
University of Stuttgart
Allmandring 3
70569 Stuttgart
Germany

Office: 1.032
Phone: +49-(0)711/685 63757
E-Mail: address@hidden