Complex polynomial eigenvalue problem with Morley elements using SLEPc

General Discussion
1

Dear developers,
I am trying to use SLEPc to solve a complex polynomial eigenvalue problem with a sequential program (I don’t need to parallelize it yet) and the command PEPSolve doesn’t seem to be implemented, at least for me. I can run all the examples using EPSSolve as well as the only example using NEPSolve but when I try to use the polynomial eigenproblem solver this error occurs: “The Identifier PEPSolve does not exist”

Part of the reason I don’t want a parallel program is because I use Morley elements, I am working on a plate bending problem. And the script “macro_ddm.idp” called in these examples doesn’t seem to be adapted to this element (the size of the eigenvectors is 3).

I also tried to solve my problem with the nonlinear solver NEPSolve using the example “nonlinear-2d-SLEPc-complex.edp”. I attempted to turn this parallel program into a sequential program, in vain. I am new to parallel computation and unlike the linear examples this one doesn’t perform any geometric domain decomposition so I am a bit lost.

Do you have any idea to solve my problem?
Thank you.
Antoni

2

Dear Antoni,

  1. I’ve put a new parallel example in the distribution that uses Morley elements (as a general rule of thumb, for your own safety, it’s best not to look at macro_ddm.idp)
  2. PEPSolve is not interfaced yet, because no one asked me about it. If you need it, I can definitely hook it up

I’ll reach out to you in private to see how to move forward (NEPSolve should not be used if your problem is indeed polynomial).