Move all of pyMOR's PDE toolkit to discretizers.builtin #847
Conversation
|
I get an error when running |
| discretize_stationary_fv, discretize_instationary_fv, | ||
| OnedGrid, TriaGrid, RectGrid, load_gmsh) | ||
| from pymor.discretizers.builtin.domaindiscretizers.default import discretize_domain_default | ||
| from pymor.discretizers.builtin.grids.boundaryinfos import EmptyBoundaryInfo, GenericBoundaryInfo, AllDirichletBoundaryInfo |
There was a problem hiding this comment.
we could even drop this and add some pymor.discretizers.basic which would result in
from pymor.basic import *
from pymor.discretizers.basic import *Its only one additional line in interactive sessions but would further document the splitting.
There was a problem hiding this comment.
Or just
from pymor.basic import *
from pymor.discretizers.builtin import *
I can't reproduce this. Could you try with a fresh build environment? |
Codecov Report
|
This moves all components of pyMOR's builtin discretization toolkit to
pymor.discretizers.builtin. The rationale:pymor.domaindescriptionsintopymor.analyticalproblems.discretizers.fenicsanddiscretizers.ngsolvemodules with discretizers that take the sameanalyticalproblemsas input as the builtin toolkit. For that we will need to makeFunctionsinteroperable between different PDE solvers, probably using sympy expressions which are converted to the appropriate format.There is still
pymor.discretizers.diskwhich seems misplaced. (It does not discretize anything.) I would like to completly ditch the current implementation and have a YAML-based input file format that is shared by allModelsthat can be read from disk. I will discuss this in another issue, however.I want to merge this PR soon in order to have time for letting things settle before the next release.