[Top][All Lists]

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

Re: [ESPResSo-users] why does Espresso calculate interactions between ne

From: Owen Hickey
Subject: Re: [ESPResSo-users] why does Espresso calculate interactions between neutral particles in debye huckel?
Date: Fri, 19 Oct 2012 15:47:49 +0200


I think his point is that there is a surprising large slowdown of the simulations when he turns on the Debye-Hueckel interaction even though he has relatively few charged particles.  It is true the underlying code calculates the DH interaction between all particles within the cutoff.  You could try adding an if (p1->q*p2->q!=0) before the call is made but you might still find it much slower.  This is because that even for the short ranged interactions between neutral particles the cutoff for the DH interaction would be used as their is only one neighbor list in Espresso.

I would recommend first trying to use P3M as there is a good chance it provides acceptable performance.  Adding an if to the c code might also help though I suspect that it might still be slower than switching to P3M.


On Fri, Oct 19, 2012 at 3:23 PM, Olaf Lenz <address@hidden> wrote:

On 10/18/2012 04:13 PM, amir khosravani wrote:
I have many simulations that these simulations have neutral and
charged particles, and the number of neutral particles are more than
charged particles. I used debye huckel interaction; and my programs
are very slow. I have tested and understood that Espreeso calculates
the interactions between neutral particles and charged-neutral
particles, whereas these interactions are zero. Are you have any idea
about the solution of this problem?

I do not completely understand your mailing. You have charged and
neutral particles. The charged particles interact via the Debye-Hueckel
potential, to model the Coulomb interaction. However, the neutral
particles so also interact via some kind of LJ potential or the like,
don't they? Therefore, they also take some time. If the neutral
particles would not interact at all, why would you include them into the
simulation in the first place?


Dr. rer. nat. Olaf Lenz
Institut für Computerphysik, Allmandring 3, D-70569 Stuttgart
Phone: +49-711-685-63607

reply via email to

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