Cannot turn rotating convection example into an IVP

[Galen Neil Wilcox]

Hello everyone,

I have encountered another example where meridional NCCs cause problems for IVPs. Attached is my attempt to turn the rotating convection example into an IVP. The matrices are quite poorly conditioned (~1e13), even when running in serial. In this case even small noise in the initial conditions makes the solution blow up immediately unless the time step is very, very small.

In parallel, the matrices above MPI rank 0 are very nearly singular (cond ~ 1e88), and the code does not run at all, with the specific error message

On entry to ZGEMV  parameter number  2 had an illegal value

The posts linked here exhibit similar issues:

https://groups.google.com/g/dedalus-users/c/6FBIx5hvJcE/m/lCQnQ4QeCwAJ

https://groups.google.com/g/dedalus-users/c/34kf4gcgS4s

I wonder if anyone else would have success turning this example into an IVP, and if not, whether the handling of meridional NCCs could have a bug.

Thank you for any insight!

Galen