HDF TypeError

Question

Greetings,

I'm trying to transition to parallel solution, I had previously used xml as the storage media for the mesh but to parallelise we need to use hdf5.

Following code elsewhere in the support forum I have now

# Import micro mesh
macromesh = Mesh()
hdf = HDF5File(mpi_comm_world(),'./meshes/channel_10k.h5', "r")
hdf.read(mesh, '/mesh', False)
macroboundaries = FacetFunction("size_t", macromesh)
hdf.read(boundaries, "/boundaries")

Looking at the h5 file with h5ls I have

/                        Group
/boundaries              Group
/boundaries/coordinates  Dataset {5436, 2}
/boundaries/topology     Dataset {15921, 2}
/boundaries/values       Dataset {15921}
/mesh                    Group
/mesh/cell_indices       Dataset {10486}
/mesh/coordinates        Dataset {5436, 2}
/mesh/topology           Dataset {10486, 3}

Running the code gives me

Traceback (most recent call last): File "upscaled_ch.py", line 354,
in
main() File "upscaled_ch.py", line 267, in main
hdf.read(mesh, "/mesh", False) TypeError: in method 'HDF5File_read', argument 2 of type 'dolfin::GenericVector &' Aborted
(core dumped)

And I cannot see what is causing the typeerror, is it a problem with the h5 file, or the arguments I'm passing (none are a vector, nor should be that I can see?)

Answer 1

I think you may need to define mesh, e.g. mesh=Mesh() before reading it in. Otherwise, python can't know what type of object you are trying to read. Or maybe you meant to put macromesh instead of mesh.

Comments

Many thanks, I got fixated on the error message, he problem was with the first argument, not the second as the message says