search

URDME / urdme

Stochastic reaction-diffusion simulation on unstructured meshes.
www.urdme.org
GNU General Public License v3.0
28 stars 12 forks source link

polarization.mph fails in CMP 4.3 #19

Closed ahellander closed 11 years ago

ahellander commented 11 years ago

The file loads, but assembly segfaults.

ahellander commented 11 years ago

For some reason this 3.5 file loads fine in CMP 4.x, but when loaded in the Livelink, it is not possible to assemble matrices from it. I don't understand the CMP4.x object well enough to fix this.

The neuron had a similar problem, but there I managed to make it work somehow by saving a new CMP 4.x file. That file is then not compatible with 3.5a....

Do we need to maintain two files for the example? But then why is coli.mph working?

pavolbauer commented 11 years ago

I see you fixed it, the reason was that the old polarization.mph was done in CMP < 3.5a. Update: I see now, the issue is the size of the matrix. Compare with the version in SVN.

ahellander commented 11 years ago

Thank's Pavol. I'll try it now. I did change the file yesterday, but only after I encountered the problem. What I did was to try to remove an old 3.5a solution structure. I had to do that for the neuron. I'll check if Brian altered it, otherwise I'll copy over the SVN version again.

It is odd that it works in 3.5a though, or is this what you expect will happen?

ahellander commented 11 years ago

Ok, after trying everything else I could think of, I tried to resart Comsol... It worked. I will dismiss this issue as a CMP 4.3 instability, alternatively some Java problem. Sorry for taking up your time.