[Top][All Lists]

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

Re: [ESPResSo-devel] Nsquare cellsystem and folded positions

From: Axel Arnold
Subject: Re: [ESPResSo-devel] Nsquare cellsystem and folded positions
Date: Mon, 13 Aug 2007 15:35:34 +0200
User-agent: KMail/1.8.2

On Monday 13 August 2007 15:23, Olaf Lenz wrote:
> > In case of the nsquare cellsystem, that means the absolute position,
> > because it doesn't fold at all. Wouldn't make much sense.
> Why would it not make sense? For me, I always interpreted "folded
> position" as the position in the central simulation box when using
> periodic boundary conditions.

In theory, yes, in practice, no. For the nsquare algorithm, it would simply be 
costly to fold the positions, and there is no need for it. For domain 
decomposition, it would be even worse: folding also means to shift particles 
across cells. If we fold every time step, we communicate every time step. 
Moreover, a particle performin Brownian motion along a processor boundary 
would hit the performance so much you could probably better use only one cpu.
That is why there are skins...

So, in short there are good reasons for internal folded positions to be only 
"folded" in a certain sense. The question is, whether the part print 
folded_position should maybe just print the correctly folded position, rather 
than the internal one.


Dr. Axel Arnold
FOM Institute for Atomic and Molecular Physics
Kruislaan 407                                   Phone: +31 20 6081 275
1098 SJ Amsterdam, The Netherlands              E-mail: address@hidden

reply via email to

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