Re: [ESPResSo] Rectangutar Box

[Axel Arnold]

On Wednesday 12 August 2009 09:04, Olaf Lenz wrote:
> Hi!
>
> Ahmad A. J. Agung wrote:
> > I need to do a simulation in a rectangular box with electrostatics
> > interaction.
> > Could anyone help me giving an assistance or at least a clue which code
> > to be modified, to make Espresso works in a rectangular box with P3M
> > activated?
>
> At this time, ESPResSo only supports P3M with a cubic box.
> It is possible to extend P3M to use rectangular or even general
> triclinic boxes.
> In general, so far we found it simpler to extend our systems to fit into
> a cubic box than to implement P3M for generic rectangular systems.
>
> If you really need a rectangular box, the work is probably not too hard,
> but it is pretty tedious.
> This article will give you a good introduction on P3M (better readable
> than the original publication):
> http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JCPSA6
>000109000018007678000001&idtype=cvips&gifs=yes

Hi!

As far as I know, Christian Holm's group is or will be soon working on an 
anisotropic P3M. The modifications are however not so simple, since an 
anisotropic P3M also requires an anisotropic mesh, which in the current state 
is not well supported by the FFT backend. Therefore, you need to modify not 
only the P3M, which should be relatively easy following the reference Olaf 
gave, but mainly you need to rewrite the FFT parallelization, which will be 
more tedious.

Axel

-- 
Dr. Axel Arnold 
Fraunhofer SCAI
Schloss Birlinghoven, 53754 Sankt Augustin, Germany
Tel: +49 2241 14 2575