Closed angela-ko closed 6 months ago
@akaszynski Hi Alex! I see you've already noticed the PR but I just linked the relevant issue it's associated with + fixed a minor pydoc linter issue. Would you mind taking a look at it again?
Also, we added support for both passing in a bgmeshfilename and the bgmesh directly through a pyvista mesh but we weren't certain whether or not the bgmeshfilename feature would be necessary. Let me know if there's any changes we should make!
cc: @dionhaefner
Will do. Will be reviewed by this evening.
Will do. Will be reviewed by this evening.
Apologizes for the delay. Build was failing and bumping cibuildwheel
fixed that, but discovered a few other minor issues while updating the build in #63, including the need for Python 3.12 wheels for pymeshfix
.
Made a handful of changes and commented out what appears to be debugging code and plotter setup, which reduced the test time.
One note, it's a bit less verbose to generate a subgrid with:
cell_center = grid.cell_centers().points
subgrid = grid.extract_cells(cell_center[:, 2] < 0)
extract_cells supports boolean masks, and using cell_centers means you don't have to access the cells
attribute and manually compute cell centers.
Another note:
We can use cell_connectivity, avoiding the need for a loop and a temporary list within _process_bgmesh
since we guarentee that the UnstructuredGrid is composed of only tetrahedra. This greatly improves performance. Implemented in 69b064cf63413472f2d1a39af8d030ef1dd556b2.
LGTM! Thanks for the speedy review / edits.
Glad to learn that cell connectivity could be used to get the tets instead as well.
Awesome job!
Wow, that was fast - thanks everyone!
Adds the support for sizing func as mentioned in issue #59