Model.occ.fragment vs model.occ.cut and options

mesh
1

When solving Laplace/Poisson equation over a (2D) bounded region e.g. a rectangle, which contains electrodes e.g. a set of disks, there obviously no need for the interior of the electrodes to be included. Searches online suggest that one should use fragment function to ensure the mesh conforms to the boundary, but in that case I also get back meshs over the electrode regions. When using

mesh_data = dolfinx.io.gmsh.model_to_mesh(gmsh.model, MPI.COMM_WORLD, 0, gdim=2)

it will complain if those interior points are not tagged. Is there really an issue with cut not conforming to the boundary? If so, why does cut even exist?

Also, there does not seem to be a good source on using the various functions in gmsh or using the generated mesh for the PDE solvers in FEniCSx/Dolfinx. In particular both cut and fragment have options boolean options removeTool, removeObject. Neither of these options seem to do anything - literally nothing, but the documentation would imply it is supposed to remove exactly the inner regions I refer to.

2

No , if you do not need the inner nodes, cut is completely fine for what you would like to do.

There are a variety of sources: Mesh generation with Gmsh — DOLFINx 0.10.0.post5 documentation
Electromagnetic scattering from a wire with PML — DOLFINx 0.10.0.post5 documentation
Using the GMSH Python API to generate complex meshes | Jørgen S. Dokken
Meshes from external sources — FEniCS Workshop
Electromagnetics example — FEniCSx tutorial
Test problem 2: Flow past a cylinder (DFG 2D-3 benchmark) — FEniCSx tutorial

Gmsh has a large set of tutorials, such as Making sure you're not a bot!
Making sure you're not a bot!
with a description of the operators here: Gmsh 4.15.1

1 Like