Error: L5 : elliptic PDE (PETSc) solver diverged. Reason: -8.

Question

Hello

When trying to run a bidomain simulation, I get the error mentioned in the title. This is probably connected to the warning that I also get:

L1 : Elliptic lhs matrix is singular!

The weird thing is that on the 1st of October, I was able to finish a bidomain simulation without it crashing and giving me this error. When I redid it on the 3rd of November it crashed (however in a later stadium of the simulation). I checked and the parameter file is almost exactly the same as well as the version of git:

CARP GIT commit hash: c7cd7bb24b74bd94e5e2a0a2adeb644416d89fef .

With almost, I mean that in one case I used 10.0 instead of 10 (int). The monodomain version of the simulation works well. I get this error on another mesh as well so it seems to be not only the problem of one specific mesh.

I tried to look in the source code and figure out what the source of this error is and how to solve it, but got a bit lost in the code.

The question: What causes this error and how can I solve it?

Thanks a lot!

Lore

Answer 1

Hi!

The singular lhs warning means that the extracellular potential solution is not unique. This is because you do not have a extracellular ground defined. The solver usually deals with this on its own (by removing the constant component from the solution which leads to a zero-average phie), but maybe you should define a ground electrode and see if this helps.

Cheers, Aurel