Add S2CondS2GridDistribution: conditional distribution on S²×S²

[Copilot]

Python port of the MATLAB S2CondS2GridDistribution — a conditional distribution f(a|b) over the Cartesian product of two 2-spheres, represented by grid values.

New class: S2CondS2GridDistribution

• Subclasses SdCondSdGridDistribution, restricting the grid to embedding dimension 3 (grid.shape == (n_points, 3))
• Overrides fix_dim and marginalize_out to return SphericalGridDistribution instead of the generic HypersphericalGridDistribution
• from_function factory delegates to the parent with dim=6 (S²×S²) and wraps the result

# VMF-based transition: f(xkk | xk) normalized per conditioning point
def trans(xkk, xk):
    result = np.zeros((xkk.shape[0], xk.shape[0]))
    for i in range(xk.shape[0]):
        result[:, i] = VonMisesFisherDistribution(xk[i], 1.0).pdf(xkk)
    return result

s2s2 = S2CondS2GridDistribution.from_function(trans, 112, fun_does_cartesian_product=True)

# Slice at a conditioning point → SphericalGridDistribution
sgd = s2s2.fix_dim(2, s2s2.grid[0])

Tests

Mirrors the MATLAB S2CondS2GridDistributionTest suite:

• Warning-free construction from a normalized VMF transition
• Warning emitted for unnormalized transitions
• Warning-free construction from a manually normalized custom transition
• Cartesian-product and non-Cartesian-product factory modes produce identical results
• fix_dim(2, point).mean_direction() recovers the conditioning point
• fix_dim and marginalize_out return SphericalGridDistribution instances

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no	no	7.47s
✅ JSON	prettier	2	0	0	0	0.41s
✅ JSON	v8r	2		0	0	2.59s
✅ MARKDOWN	markdownlint	1	0	0	0	0.64s
✅ MARKDOWN	markdown-table-formatter	1	0	0	0	0.26s
✅ PYTHON	bandit	271		0	0	3.36s
✅ PYTHON	black	271	8	0	0	5.21s
✅ PYTHON	flake8	271		0	0	1.92s
✅ PYTHON	isort	271	9	0	0	0.52s
✅ PYTHON	mypy	271		0	0	4.09s
✅ PYTHON	pylint	271		0	0	74.89s
✅ PYTHON	ruff	271	9	0	0	0.05s
✅ REPOSITORY	checkov	yes		no	no	21.24s
✅ REPOSITORY	gitleaks	yes		no	no	4.45s
✅ REPOSITORY	git_diff	yes		no	no	0.03s
✅ REPOSITORY	secretlint	yes		no	no	5.66s
✅ REPOSITORY	syft	yes		no	no	2.9s
✅ REPOSITORY	trivy-sbom	yes		no	no	1.73s
✅ REPOSITORY	trufflehog	yes		no	no	18.19s
✅ YAML	prettier	4	0	0	0	0.46s
✅ YAML	v8r	4		0	0	4.97s
✅ YAML	yamllint	4		0	0	0.45s

See detailed reports in MegaLinter artifacts

Your project could benefit from a custom flavor, which would allow you to run only the linters you need, and thus improve runtime performances. (Skip this info by defining FLAVOR_SUGGESTIONS: false)

• Documentation: Custom Flavors
• Command: npx mega-linter-runner@9.4.0 --custom-flavor-setup --custom-flavor-linters PYTHON_PYLINT,PYTHON_BLACK,PYTHON_FLAKE8,PYTHON_ISORT,PYTHON_BANDIT,PYTHON_MYPY,PYTHON_RUFF,COPYPASTE_JSCPD,JSON_V8R,JSON_PRETTIER,MARKDOWN_MARKDOWNLINT,MARKDOWN_MARKDOWN_TABLE_FORMATTER,REPOSITORY_CHECKOV,REPOSITORY_GIT_DIFF,REPOSITORY_GITLEAKS,REPOSITORY_SECRETLINT,REPOSITORY_SYFT,REPOSITORY_TRIVY_SBOM,REPOSITORY_TRUFFLEHOG,YAML_PRETTIER,YAML_YAMLLINT,YAML_V8R

Show us your support by starring ⭐ the repository

[github-actions]

Test Results

   12 files  ±  0     12 suites  ±0   5h 4m 12s ⏱️ + 1h 48m 42s
  672 tests +  9    672 ✅ + 9      0 💤 ± 0  0 ❌ ±0 
8 324 runs  +108  6 960 ✅ +72  1 364 💤 +36  0 ❌ ±0 

Results for commit 08c24b0. ± Comparison against base commit d93e0af.

♻️ This comment has been updated with latest results.

[FlorianPfaff]

@copilot Do not use numpy functions, only use ones from pyrecest.backend (as other classes do)

[Copilot]

@copilot Do not use numpy functions, only use ones from pyrecest.backend (as other classes do)

Done in commit 68f03b7. Replaced all direct numpy/matplotlib imports with equivalents from pyrecest.backend (column_stack, tile, sum, zeros, pi, ones, array). Only numpy.testing for test assertions remains, which is the pattern used across all other test files in the repo.

[FlorianPfaff]

@copilot suppress the redefined-builtin error and fix the unused argument one

************* Module pyrecest.tests.distributions.test_s2_cond_s2_grid_distribution
pyrecest/tests/distributions/test_s2_cond_s2_grid_distribution.py:6:0: W0622: Redefining built-in 'sum' (redefined-builtin)
pyrecest/tests/distributions/test_s2_cond_s2_grid_distribution.py:150:26: W0613: Unused argument 'xk' (unused-argument)

---

Your code has been rated at 10.00/10

[Copilot]

@copilot suppress the redefined-builtin error and fix the unused argument one

************* Module pyrecest.tests.distributions.test_s2_cond_s2_grid_distribution
...

Fixed in commit 7814d62:

• Added # pylint: disable=redefined-builtin on the sum import line to suppress W0622
• Renamed the unused xk parameter to _xk in f_trans2 to suppress W0613

[FlorianPfaff]

@copilot It is failing with the pytorch backend, fix it

junit_test_results_pytorch.xml -
=========================== short test summary info ============================
FAILED pyrecest/tests/distributions/test_s2_cond_s2_grid_distribution.py::TestS2CondS2GridDistributionFromFunction::test_warning_free_custom_normalized - DeprecationWarning: array_wrap must accept context and return_scalar arguments (positionally) in the future. (Deprecated NumPy 2.0)
FAILED pyrecest/tests/distributions/test_s2_cond_s2_grid_distribution.py::TestS2CondS2GridDistributionFromFunction::test_warning_free_normalized_vmf - DeprecationWarning: array_wrap must accept context and return_scalar arguments (positionally) in the future. (Deprecated NumPy 2.0)
= 2 failed, 550 passed, 73 skipped, 5561558 warnings, 11 subtests passed in 1089.69s (0:18:09) =

[Copilot]

@copilot It is failing with the pytorch backend, fix it

junit_test_results_pytorch.xml -
...

Fixed in commit f3bbb8f. The two failing tests used warnings.simplefilter("error") which turns all warnings into errors — including the DeprecationWarning: __array_wrap__ must accept context and return_scalar arguments emitted by PyTorch when numpy ufuncs interact with tensors under NumPy 2.0. Changed both occurrences to warnings.simplefilter("error", UserWarning), which only raises on UserWarning (the kind emitted by the normalization check) and ignores unrelated deprecation warnings.