FSI problem with SLEPc

General Discussion
1

@prj
Thanks to your guidance, I’ve completed the initial version of my code. However, I’m facing some persistent issues with its robustness:
The script isn’t robust enough, it works with -n 4 but fails with -n 6 , and sometimes it produces different results even when run multiple times with the same parameters.

For example, if I run the code withff-mpirun -n 4 FSI-SLEPc-MWE.edp -wg multiple times, it gives:

image
image1016×944 22.4 KB

image
image1289×944 24.7 KB

image
image1289×944 30.1 KB

Apparently, the first one looks better than the others.

I’ve tried debugging it but haven’t been able to identify the root cause. I’m also uncertain if the solver configuration is optimal. I would be very grateful if you could take a look and help me diagnose the problem.
Here is my code attached:
FSI-SLEPc-MWE.edp (5.4 KB)

2

A typical case when such variations in the results occur is when the value of a variable is used before it has been affected any value.

3

Your code does not run in debug mode, which makes it difficult to… debug.

CHECK_KN: (k>=0) && ( (k <n) || !step) in file: ./../femlib/RNM.hpp, line 341
  current line = 159 mpirank 3 / 4
Assertion fail : (0)
	line :53, in file ./../femlib/RNM.hpp
 err code 6 ,  mpirank 3
4

Thank you for the helpful error report. I have corrected the identified mistake on line 159.

5

By the way, I have a question about restriction-2d-PETSc.edp, the parameter ThGlCond in line 35 is not declared beforehand; however, the example compiles and runs without issues, so does ThGlNoCond. Is this associated with macro_ddm.idp?

6

It was a mistake that the matrix is built with wrong fespace. Surprisingly, it compiled and ran, but produced incorrect results. Occasionally, it even produced “correct” results. I think there might be some non-deterministic behavior in the solver.

7

Not if the problem is well-posed.