On Sun, Feb 12, 2012 at 1:23 AM, Georg Rempfer
<address@hidden> wrote:
Hello,
I don't think it was fine since our GPU code didn't show this behaviour. Owen commited Wolfgangs fix but I don't know whether Olaf pulled it into the master, yet.
Parameters for which we sawthe effect were everything set to one, except for the viscosity, that should be 3 or more.
The fix also made the discrepancy in the particle velocities disappear that we saw during the coding week.
Greets,
Georg
Am 11.02.2012 20:48, schrieb Stefan Kesselheim:
Hi,
just a quick question.
Am 09.02.2012 um 14:28 schrieb Wolfgang Riefler:
Implementing this line, I get a much better result, with only a little offset, which is probably due to the discrete boundary.
Why would this change the offset situation? It should only change the magnitude of the force, right? For forces that are not incredibly large, everything is linear in the applied force, and everything should scale with the force (also the transient behaviour).
Could you give a few more details on your parameters? Viscosity, timestep, density, temperature, number of timesteps that you integrate. I did careful checks on that, and I do believe that everything should be working fine.
Cheers
Stefan
Thanks again for your help,
Wolfgang
Stefan Kesselheim
Institute for Computational Physics
University of Stuttgart
address@hiddenstuttgart.de