|
From: | Xikai Jiang |
Subject: | Re: [ESPResSo-users] iccp3m question |
Date: | Mon, 6 Oct 2014 00:21:49 -0400 |
I have tested the case using the same P3M parameters on 1 core and 8 cores, but they still give different net charge in the system.
One way I find to reduce the net charge is to shift all atoms in the system in x- and y- directions by a small amount (~0.1nm), I'm thinking is this related to the wall atoms that ride on the domain decomposition boundary? Thanks. Xikai > From: address@hidden > Date: Wed, 1 Oct 2014 20:41:32 +0200 > To: address@hidden; address@hidden > CC: address@hidden > Subject: Re: [ESPResSo-users] iccp3m question > > Hi, > I very briefly checked your script and it looks OK. The error appears strange indeed. I hope there is no serious issue with ICCP3M. > However, P3M can produce different results. I strongly recommend using P3M with the same parameters (rcut, alpha, cap, mesh) on all numbers of CPUs. > I have recently quit science, so my time to check it more carefully is somewhat limited. > For better debugging, might also want to consider smaller systems, where results are obtained quicker.. > Cheers and good luck > Stefan > > On Sep 29, 2014, at 1:25 AM, Xikai Jiang <address@hidden> wrote: > > > Dear All: > > > > I'm testing iccp3m for a system that consists of a single charged particle confined between two parallel walls with zero electrical potential difference in Espresso-3.3.0. All atoms are fixed in space, and the charged particle carries a charge of 50e. My purpose is to test whether the system is electrically neutral (it should be neutral in principle). > > > > When I put the test charge 0.2nm above the bottom wall, and run Espresso for 1 step using 1 core, the net charge of system is 0.2717e, which can be considered neutral. But when I run the same simulation using 8 cores, the net charge of system becomes 6.224e, and cannot be considered neutral within numerical inaccuracy. > > > > I tried to put the test charge away from the wall (1nm away), the net charge in system becomes 4.9e using both 1 core and 8 cores. Refining the grid on wall (spacing from 0.3nm to 0.1nm) helps to reduce the net charge in system to 2.4e, about 4.8% of the amount of the test charge. > > > > Does anyone have ideas where the problem is? I have attached the test code, any help or comments are appreciated. > > > > Regards. > > > > Xikai > > <test_constant_potential_parallel_plate_nve_fixed_atoms.tcl> > |
[Prev in Thread] | Current Thread | [Next in Thread] |