espressomd-users
[Top][All Lists]
Advanced

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

Re: [ESPResSo-users] Sistematically decreasing of Temperature with DPD t


From: Stefan Kesselheim
Subject: Re: [ESPResSo-users] Sistematically decreasing of Temperature with DPD thermostat
Date: Wed, 15 Jan 2014 17:44:36 +0100

HI,

On Jan 15, 2014, at 5:00 PM, Salvador H-V <address@hidden> wrote:

> Hi Stefan,
> 
> In all the simulations I am using:
> 
> i) 2dim system periodic 1 1 0
> ii) blx bly blz = 2.5*diam
> iii) Generalized LJ potential to mimic hard core interaction

Ar the dumbbells constraint to the plane? Then I'm not sure if that is 
compatible with RATTLE, because this actually is another constraint that needs 
to be taken into account in the system of equations.

> First simulations I was using:
> i) rigid bond command for the dimers (dumbbells) => No virtual sites
> ii) Langevin Dynamics Thermostat

Thanks, I got (i) wrong. I thought RATTLE was broken and you'd be using VS.

> Current Simulations I will try:
> i) Harmonic Bonds with high K value (but temperature seems to be dependent of 
> such value!)
> ii) DPD thermostat
> 
> I want to reproduce (qualitatively) some experimental results. So, I am 
> really interested in performing the simulations with the correct parameters 
> and methodologies.
> 
> In your email you mentioned: "bond length constraint thing... should be a 
> RATTLE or SHAKE implementation. I'm not exactly sure how it works and I'm not 
> sure if it works..."
> 
> This means that using rigid bonds or even harmonic bonds in a 2Dimensional 
> system is not a good idea? 

I was just trying to say that I'm not sure if our implementation is working. 
There is no conceptual problem with either of these things.

> When I used rigid bond command, I was able to reproduce some structural 
> properties and the system "looks fine".  The lentgh bond is keeped constant, 
> temperature is preserved and total energy is almost just kinetic energy. 
> However I am interested in dynamical properties… 

Sounds good :-). 

Using DPD is something you should however try to evaluate from a physical point 
of view. DPD is a special form of dynamics. It conserves momentum and therefore 
creates Navier-Stokes-like behaviour on long time and length scales. But the 
dynamics you get completely depend on the parameters. With very a short range 
(much smaller than particle distance) you obtain a "gas but not really 
gas"-behaviour. If many particles interact due to large interaction radii you 
will get something honey-like. But you should be aware that DPD only produces 
correct flow fields on length scales larger than the particle distance. If you 
would like to have hydrodynamics between the particles it would be necessary to 
include further particles mediating this interaction. 

Just by using DPD you don't get "better" dynamics per se. It entirely depends 
on the physical situation for which DPD is an approximation. The "almost 
gas"-behaviour would for example good to really weakly thermalise a gaseous 
system, so that it explores enough of the phase space. 
If you want a hard dumbbell fluid, you should go for NVE, or use a very, very, 
very weak DPD-interaction, that minimally alters your dynamics, but makes sure 
your sampling is OK. 
There is no general answer to this problem, but one that needs to be argued 
carefully. 

Cheers and good luck
Stefan


reply via email to

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