qmat switch

[tlunet]

After implementing most of the core coefficients generation of pySDC into qmat, here is the associated PR finalizing the switch mentioned in #436.

Main changes :

• LagrangeApproximation and NodeGenerator classes (and their associated tests) are now in qmat
• the base CollBase class is now mostly an interface to the Collocation class from qmat
• the base Sweeper class produces implicit and explicit QDelta matrices using qmat generators
• additional base pip dependency for pySDC set to qmat>=0.1.8
• docstring / documentation update to link to qmat documentation when needed

Additional changes :

• includes complete adaptation to numpy v2, based on the first changes done by @brownbaerchen in his fork. The problematic behavior for the parallelSDC project has been patch by removing testing (and coverage) of the plotting function for preconditioner_playground_MPI. If this PR is accepted, no need to merge Some fixes for the new numpy 2.0 version. #443 anymore as it was already merged in this PR.
• Sweeper.get_qDelta_implicit and Sweeper.get_qDelta_explicit only have one argument now : qd_type. The previous coll one was simply replaced by self.coll, as no Sweeper instance in pySDC code use another Collocation class as the base underlying collocation for the sweeper. This was a weird mix between OOP and Procedural Programming that does not make sense but rather the code more complex than it should be.
• complete renaming of the core sub-modules to follow a naming consistent with contribution guidelines, using the refactoring tools from VSCode. Only one issue had to be solved manually for a private class name based attribute in a Hook class (see https://github.com/tlunet/pySDC/commit/32f64bf032119f989c9302c3c077d1cb4fc07de8, I won't even try to understand the idea behind this sort of attribute ...)

[codecov]

Codecov Report

Attention: Patch coverage is 92.12730% with 47 lines in your changes missing coverage. Please review.

Project coverage is 77.83%. Comparing base (18989d3) to head (97fc219).
Report is 22 commits behind head on master.

Files	Patch %	Lines
...ySDC/projects/Performance/controller_MPI_scorep.py	0.00%	5 Missing ⚠️
pySDC/core/sweeper.py	97.43%	3 Missing ⚠️
...ns/problem_classes/GrayScott_1D_FEniCS_implicit.py	0.00%	3 Missing ⚠️
pySDC/implementations/sweeper_classes/explicit.py	0.00%	3 Missing ⚠️
...lementations/transfer_classes/BaseTransfer_mass.py	0.00%	3 Missing ⚠️
pySDC/projects/Resilience/dahlquist.py	0.00%	3 Missing ⚠️
...ementations/problem_classes/AllenCahn_2D_FD_gpu.py	0.00%	2 Missing ⚠️
...mentations/problem_classes/AllenCahn_2D_FFT_gpu.py	0.00%	2 Missing ⚠️
...entations/problem_classes/Boussinesq_2D_FD_imex.py	0.00%	2 Missing ⚠️
...tations/problem_classes/HeatEquation_ND_FD_CuPy.py	0.00%	2 Missing ⚠️
... and 14 more

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #445      +/-   ##
==========================================
+ Coverage   77.38%   77.83%   +0.44%     
==========================================
  Files         327      332       +5     
  Lines       26085    25840     -245     
==========================================
- Hits        20187    20112      -75     
+ Misses       5898     5728     -170     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[pancetta]

Great work, thanks a lot. Taking the quadrature stuff out of pySDC is a very good idea and I appreciate your work.

[pancetta]

Not sure why you're obscuring this PR with so much renaming, though.. this makes it much harder to see what actually happens here.

[pancetta]

Is there a reason why the CI pipeline is back to 40+minutes now? Or is that a GitHub glitch? I see no obvious cause.

[tlunet]

Not sure why you're obscuring this PR with so much renaming, though.. this makes it much harder to see what actually happens here.

Well, this was a big change anyway, and the naming of the core classes has been confusing me since a long time ... so that's mostly shooting two birds with one stone.

[pancetta]

Not sure why you're obscuring this PR with so much renaming, though.. this makes it much harder to see what actually happens here.

Well, this was a big change anyway, and the naming of the core classes has been confusing me since a long time ... so that's mostly shooting two birds with one stone.

Understood. As a preference, I think smaller PRs with a clear focus are better, though.

[pancetta]

And no, this is not stopping me from merging this, of course!

[tlunet]

And no, this is not stopping me from merging this, of course!

I guessed 😅 ... but don't worry, I'll stick to smaller PR with one dedicated topic from now.

Also in my opinion, it would make sense to drop a new release after the PR is merged.

[pancetta]

For sure and already on my todo list!

[tlunet]

Is there a reason why the CI pipeline is back to 40+minutes now? Or is that a GitHub glitch? I see no obvious cause.

Probably a GitHub glitch ... but project testing still take quite some time (especially the parallelSDC, DAE and Resilience). Where are we on testing projects only for one Python version ?

[pancetta]

Unless this will significantly shorten the CI time, I'd rather have broad coverage.

[tlunet]

Unless this will significantly shorten the CI time

Probably not, since most are run in parallel ... just significantly less computation ressource used (mostly energy saving)