return b vector at the end of newtonsolver

[keiyamamo]

Fix #78

[keiyamamo]

Hi @jorgensd

I update the pytest and the failure comes from the segmentation fault error

subprocess.CalledProcessError: Command 'turtleFSI --problem TF_cfd -dt 0.01 -T 0.05 --verbose True --folder tmp --sub-folder 1' died with <Signals.SIGSEGV: 11>.

This is the same error I keep getting in the new fsi repository that we are working on and it seems to happen only on Mac computer. I tracked down the problem a little bit and think that the segmentation fault occurs when FEniCS tries to solve the linear system https://github.com/KVSlab/turtleFSI/blob/34925bed03008e463ce6ddc82b875287fc866177/turtleFSI/modules/newtonsolver.py#L92

Do you have any suggestions how to fix the problem? Here is the output of conda list for fenics part

fenics                    2019.1.0        py310h9116bd8_37    conda-forge
fenics-dijitso            2019.1.0        py310h2ec42d9_37    conda-forge
fenics-dolfin             2019.1.0        py310h5349f8f_37    conda-forge
fenics-ffc                2019.1.0        py310h2ec42d9_37    conda-forge
fenics-fiat               2019.1.0        py310h2ec42d9_37    conda-forge
fenics-libdolfin          2019.1.0            h6cc1711_37    conda-forge
fenics-ufl                2019.1.0        py310h2ec42d9_37    conda-forge

while the conda environment that works outputs

fenics-dijitso            2019.1.0        py310h2ec42d9_35    conda-forge
fenics-dolfin             2019.1.0        py310h42dc3f7_35    conda-forge
fenics-ffc                2019.1.0        py310h2ec42d9_35    conda-forge
fenics-fiat               2019.1.0        py310h2ec42d9_35    conda-forge
fenics-libdolfin          2019.1.0            h2bb9e8c_35    conda-forge
fenics-ufl                2019.1.0        py310h2ec42d9_35    conda-forge

The build number is different between these two but not sure if that’s the problem.

[keiyamamo]

I tried with different linear solver and turtleFSI works fine with umfpack, but not with MUMPS. Tried to get more information from MUMPS why it fails but I only managed to get this kind of information

Entering DMUMPS 5.2.1 from C interface with JOB, N, NNZ =   1       83012        6240608
      executing #MPI =      1, without OMP

 =================================================
 MUMPS compiled with option -Dparmetis
 MUMPS compiled with option -Dptscotch
 This MUMPS version includes code for SAVE_RESTORE
 =================================================
L U Solver for unsymmetric matrices
Type of parallelism: Working host

 ****** ANALYSIS STEP ********

Entering analysis phase with ...
                N        NNZ         LIW       INFO(1)
              83012    6240608    12564229             0
Matrix entries:    IRN()   ICN()
           1      1           1      2           1      3
           1      4           1      5           1      6
           1      7           1      8           1      9
           1     10
 ... Structural symmetry (in percent)=  100
 Average density of rows/columns =   74
 Average density of rows/columns =   74
 ... No column permutation
 Ordering based on METIS
Segmentation fault: 11