Fix/numpy2 compat

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS

[RS-ins]

Problem: GEMAct crashes on NumPy ≥ 2.0.
Two removals bite: np.infty (gone in 2.0) and calling nonzero on 0-d arrays (now an error), the latter hitting scalar pmf/ppf calls on the zero-truncated/zero-modified distributions and Exponential.ppf. Currently masked by the numpy<2.0.0 pin, which forces a downgrade in any modern environment.
Changes: 5× np.infty → np.inf; 6× wrap the np.where condition in np.atleast_1d (as NumPy's own error message recommends), leaving the existing x.shape == () branches untouched; relax the pin.
Verification: full test suite passes on NumPy 2.5.1 (31 passed).
Before the second fix, 6 tests failed; before the first, hard crash.
Also verified on 1.26.4 — results bit-identical, so no behavior change for existing users.

[EdoLu]

@RS-ins thanks for the work!

[RS-ins]

Dear professor, I would like to notify you that I have sent a follow up PR to fix the matplotlib version dependency that came up with the PR #17.
Best regards
RS