Error running the FitzHugh-Nagumo model

0 votes
60 views
asked Jan 11, 2018 by J-H Plank (440 points)
edited Jan 11, 2018 by J-H Plank

Dear all,

I tried to run the FitzHugh-Nagumo model which is hardcoded in the heart component. The test I wrote does run without any errors with a different cell model, however if I use the FitzHugh-Nagumo model I get the following error:

Error: Test failed:
Chaste error: ./global/src/parallel/PetscException.cpp:159: DIVERGED_NANORINF in function 'User provided function' on line 1128 of file ./linalg/src/LinearSystem.cpp
Failed

My guess is that the error is somehow caused by problems with the linear solver from Petsc. Thus I played around with some KSP settings - but to no avail. In the internet it was suggested for similar problems to use a direct solver. Is it possible to do so in Chaste and if so, how can I do it?

Surprisingly this error does not occur if I use very small stimuli well below the threshold.

Is the FitzHugh-Nagumo model still maintained in the current version of Chaste and is part of the tests?

Edit: With my original parameters also the LuoRudy1 model showed either this error or threw the "!std::isnan(i_ionic)-exception". That's why I switched back to the standard parameters. Here the LuoRudy1 model and the Bueno model (which also worked before) are running fine. However, the FitzHugh-Nagumo model still has the same problems.

The code that is creating this error can be found here: https://etherpad.gwdg.de/p/csQYkPSDEhCQM5w

Thanks a lot!
Jan-Hendrik

commented Jan 11, 2018 by GaryM (940 points)

Do you have the test that you could upload for us to look at?

commented Jan 11, 2018 by J-H Plank (440 points)

Thanks for your quick response!
I just realized that I also get this error for another cell model. It thus might be, that I chose some bad parameters, where only the one cell model is stable, but the others aren't. I'm now checking on that. If I can exclude this error, I will upload the test.

commented Jan 11, 2018 by J-H Plank (440 points)

Hi Gary, it unfortunately still doesn't work as I wrote in my edit.

commented Jan 15, 2018 by GaryM (940 points)

Have you tried reducing time steps in the simulation?

commented Jan 15, 2018 by J-H Plank (440 points)

Yes I tried, but it didn't help...

Please log in or register to answer this question.

Welcome to Chaste Q&A, where you can ask questions and receive answers from other members of the community of Chaste developers and users.

See the Chaste wiki for installation instructions, guides and tutorials.

...