mindspore.ops.slogdet rejects 1x1 matrix while paddle.linalg.slogdet succeeds

[tingPetty]

mindspore.ops.slogdet errors on a valid 1x1 float32 matrix with row size must be greater than or equal to 2, while paddle.linalg.slogdet succeeds. This appears to be an unnecessary shape restriction.

Environment

Hardware Environment(Ascend/GPU/CPU):

/device cpu

Software Environment:

• MindSpore version (source or binary): 2.8.0 (binary)
• Python version (e.g., Python 3.7.5): 3.10.18
• OS platform and distribution (e.g., Linux Ubuntu 16.04): Windows 10 (10.0.26200)
• GCC/Compiler version (if compiled from source): N/A

Describe the current behavior

For input shape (1, 1), mindspore.ops.slogdet raises:
ValueError: For primitive[LogMatrixDeterminant], the row size must be greater than or equal to 2, but got 1.
Paddle reference succeeds.

Describe the expected behavior

slogdet should support 1x1 square matrices (standard linear algebra behavior), matching mainstream framework behavior.

Steps to reproduce the issue

1. Run pd_ms_test_1/simple_tests/ms_error_only_candidate_209_linalg_slogdet.py.
2. Observe [PD] success and [OTHER] error.
3. Verify MindSpore throws row-size restriction error for 1x1 matrix.

Related log / screenshot

ValueError: For primitive[LogMatrixDeterminant], the row size must be greater than or equal to 2, but got 1.
- C++ Call Stack:
mindspore/ops/infer/ops_func_impl//log_matrix_determinant.cc:50

Special notes for this issue

• Paddle version in repro: 3.2.0