style: add type hinting to variational distributions

[leahaeusel]

Description and Context:
What and Why?

This PR adds type hinting to our variational distributions.

I have introduced new type aliases to indicate the shapes of numpy arrays, since the docstrings provided this information for some of the variables. In some cases, the shapes were outdated, though, as I realized when I checked the shapes with our tests. In my opinion, it would be really helpful to eventually replace all or most of the np.ndarrays in our code base with these shape-indicating aliases.

FYI: I had to adapt FullRankNormal.reconstruct_distribution_parameters() because mypy doesn't like functions with a varying number of returned variables.

Related Issues and Pull Requests

• Closes
• Related to Explore Adding Type Annotations to QUEENS #74

Interested Parties

@gilrrei

Note: More information on the merge request procedure in QUEENS can be found in the Submit a pull request section in the CONTRIBUTING.md file.

[Copilot]

Pull request overview

This PR adds comprehensive type hinting to the variational distributions module in QUEENS, introducing two new type aliases (Array1xN and ArrayNxM) for better shape indication in numpy arrays. The PR also refactors FullRankNormal.reconstruct_distribution_parameters() to satisfy mypy's requirement for consistent return types by creating a separate method for cases where the Cholesky decomposition is needed, and enables mypy checking for the variational distributions module by removing it from exclusion lists.

Key Changes

• Introduced new type aliases Array1xN and ArrayNxM in src/queens/utils/type_hinting.py to indicate array shapes
• Added type hints throughout the variational distributions hierarchy including the base class and all concrete implementations
• Split FullRankNormal.reconstruct_distribution_parameters() into two methods to resolve mypy's varying return count issue

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 27 comments.

Show a summary per file

File	Description
src/queens/utils/type_hinting.py	New file introducing Array1xN and ArrayNxM type aliases for shaped numpy arrays
src/queens/variational_distributions/_variational_distribution.py	Added type hints to base class including Array1xN for variational parameters and proper return types
src/queens/variational_distributions/particle.py	Added type hints to all methods, updated docstrings, and modified initialization to pass n_parameters
src/queens/variational_distributions/mixture_model.py	Added type hints, improved variable naming (logpdf_lst, parameters), and initialized FIM properly
src/queens/variational_distributions/mean_field_normal.py	Added type hints to all methods and updated docstrings for consistency
src/queens/variational_distributions/full_rank_normal.py	Added type hints, split reconstruct_distribution_parameters() into two methods for mypy compatibility
src/queens/variational_distributions/joint.py	Added type hints, fixed logpdf initialization to use proper numpy array instead of scalar
src/queens/variational_distributions/__init__.py	Added type hints to __getattr__ function and imported base class for type checking
pyproject.toml	Removed variational_distributions from mypy exclusion list to enable type checking
.pre-commit-config.yaml	Removed variational_distributions from mypy exclusion pattern in pre-commit hooks

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 5 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[gilrrei]

I think the return statements could be more precise :)

[Copilot]

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated 23 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[danielwolff1]

Thanks a lot for providing the additional type hints in the variational distribution classes, that really must have a been quite some effort and sorry for taking so long to review it. I left a few comments to mark points where I'm unsure about because I'm not too familiar with variational distributions 🙈

[danielwolff1]

since I am not familiar with the math here: Is this something that is super specific to a joint distribution or would it make sense to move this to _variational_distribution as well?

[danielwolff1]

Is the remark here about the third dimension still true? 🤔
Just wondering because then the typing annotation would technically not be correct, right, since we define ArrayNSamplesXNDims: TypeAlias = np.ndarray[tuple[NSamples, NDims], np.dtype[np.floating]], i.e., as an array with exactly two dimensions.

[danielwolff1]

Same question as below / above (I never know in which order GitHub will display my comments 🙈): Is the remark here about the third dimension still true?

[danielwolff1]

Is this from the meaning fundamentally different from the NDimsComponent type variable defined in joint.py? Or could they be unified and moved to _variational_distribution.py?