problems running multiphase models

[nlooije]

Hi,

first of all thank you for your work on this software!

Im able to run the singlephase examples from the repository without any problems.

However, when i run the multiphase examples the simulations are unstable.

In the case of SC models, the visualization is completely black with max v: nan for all examples.

In the case of binary liquid, there is chaotic phase separation throughout the domain for all examples.

I read in some other posts that some others had problems with max v: nan but no solutions were posted.

Any ideas how i can solve this issue?

thx in advance,

Niels

[Tundrashark]

Hi,

I am having the exact same problem.  I built it on ubuntu 11.04 64 bit.  I put numpy 1.6.1 on and had to change
the geo.py to use numpy.unique1d and #pragma OPENCL EXTENSION cl_amd_fp64 : enable in backend_opencl.py
to use my radeon hd 6970 card.

I am hoping to do some multi-phase simulation. Just getting started.

Thanks
Patrick

[Michal Januszewski]

Hi,

First of all, I'm sorry for my extremely late response..

I believe the problems you were experiencing were due to a problem in the force implementation that for some models caused us to use a spurious density factor.  This should finally be resolved in both the master and multigpu branches on https://github.com/sailfish-team.  Please make sure you are synced to the latest commit and let me know if this is still an issue for anyone.

Thanks,

Michal

--
Michal Januszewski
http://people.gentoo.org/spock