PETSc error with mesh from meshio

General
1

Hi everyone,

I’m using a mesh that I converted using meshio.
Available here: mesh.zip - Google Drive

When trying to solve a simple problem on it, I get this error:

No Jacobian form specified for nonlinear variational problem.
Differentiating residual form F to obtain Jacobian J = F'.
Solving nonlinear variational problem.
  Newton iteration 0: r (abs) = 5.430e+01 (tol = 1.000e-10) r (rel) = 1.000e+00 (tol = 1.000e-09)
Traceback (most recent call last):
  File "/home/remidm/Fusion-TES-Modeling/MWE.py", line 32, in <module>
    solve(F == 0, u, bcs)
  File "/home/remidm/miniconda3/envs/tes-model-env/lib/python3.11/site-packages/dolfin/fem/solving.py", line 220, in solve
    _solve_varproblem(*args, **kwargs)
  File "/home/remidm/miniconda3/envs/tes-model-env/lib/python3.11/site-packages/dolfin/fem/solving.py", line 266, in _solve_varproblem
    solver.solve()
RuntimeError: 

*** -------------------------------------------------------------------------
*** DOLFIN encountered an error. If you are not able to resolve this issue
*** using the information listed below, you can ask for help at
***
***     fenics-support@googlegroups.com
***
*** Remember to include the error message listed below and, if possible,
*** include a *minimal* running example to reproduce the error.
***
*** -------------------------------------------------------------------------
*** Error:   Unable to successfully call PETSc function 'MatSetValuesLocal'.
*** Reason:  PETSc error code is: 63 (Argument out of range).
*** Where:   This error was encountered inside /home/conda/feedstock_root/build_artifacts/fenics-pkgs_1696906506503/work/dolfin/dolfin/la/PETScMatrix.cpp.
*** Process: 0
*** 
*** DOLFIN version: 2019.1.0
*** Git changeset:  2e001bd1aae8e14d758264f77382245e6eed04b0
*** -------------------------------------------------------------------------

I have noticed that depending on how I use dx in the formulation it does or doesn’t work.

I’m guessing it may be a tagging issue? Although when I inspect volume_markers it only contains 6 and it has the same size as the mesh (141787 cells).

Here is my MWE:

from fenics import *

inlet_id = 8
outlet_id = 7
fluid_id = 6

mesh = Mesh()
XDMFFile("mesh_cells.xdmf").read(mesh)

# create dx from the volume markers
volume_markers = MeshFunction("size_t", mesh, mesh.topology().dim())
XDMFFile("mesh_cells.xdmf").read(volume_markers)
dx = Measure("dx", domain=mesh, subdomain_data=volume_markers)

surface_markers = MeshValueCollection("size_t", mesh, mesh.topology().dim() - 1)
XDMFFile("mesh_facets.xdmf").read(surface_markers, "f")
surface_markers = MeshFunction("size_t", mesh, surface_markers)

V = FunctionSpace(mesh, "CG", 1)
u = Function(V)
v = TestFunction(V)

inlet = DirichletBC(V, Constant(1), surface_markers, inlet_id)
outlet = DirichletBC(V, Constant(0), surface_markers, outlet_id)
bcs = [inlet, outlet]

F = dot(2 * grad(u), grad(v)) * dx(fluid_id)  # doesn't work
# F = 2 * dot(grad(u), grad(v)) * dx(fluid_id)  # doesn't works
# F = dot(grad(u), grad(v)) * dx(fluid_id)  # works
# F = dot(2 * grad(u), grad(v)) * dx  # works

solve(F == 0, u, bcs)

XDMFFile("solution.xdmf").write(u)

Has anyone ever experienced this?

Cheers
Remi

This is what surface markers look like:

image
image1830×892 78.7 KB

Volume markers:

image
image1909×963 66.3 KB

2

Update

After a bit of digging it appears that when I install meshio in a seperate env the issue doesn’t appear…

3

How did you install meshio? Did you use python3 -m pip install --no-binary=h5py h5py meshio?
Otherwise you might get issues with HDF5 conflicts.

4

I installed it from a conda yml file
I will try this pip install method, don’t know how to do this with the yml file but I’ll look around

5

If it is installed with conda it should not cause an issue (as long as dolfin also was installed with conda).

6

it “should” or should not?

name: tes-model-env

channels:
  - conda-forge
  - defaults

dependencies:
  - pip
  - fenics
  - numpy==1.24
  - pip:
    - meshio[all]

This is how I was doing it

7

I corrected the comment above. It should not cause an issue if you do it purely through conda, i.e.

name: tes-model-env

channels:
  - conda-forge
  - defaults

dependencies:
  - pip
  - fenics
  - numpy==1.24
  - meshio

as meshio is a package on conda-forge: GitHub - conda-forge/meshio-feedstock: A conda-smithy repository for meshio.

8

Right, I didn’t know it had a conda package. Thanks!