mindspore.ops.real fails on empty complex tensor while paddle.as_real succeeds (CPU)

[tingPetty]

When input is an empty complex tensor with shape (10, 0), paddle.as_real succeeds but mindspore.ops.real fails with CPU kernel CreateKernel RuntimeError. This looks like a robustness issue in empty-tensor handling.

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

mindspore.ops.real on complex128 tensor with shape (10, 0) fails with RuntimeError and C++ CreateKernel stack. Paddle reference succeeds.

Describe the expected behavior

mindspore.ops.real should support empty complex tensors and return an empty real tensor, or at least raise a clear user-facing validation error instead of internal kernel creation failure.

Steps to reproduce the issue

1. Run pd_ms_test_1/simple_tests/ms_error_only_candidate_053_as_real.py.
2. Observe [PD] success and [OTHER] error.
3. Error includes UnaryOpCpuKernelMod::Init and CPUKernelExecutor::CreateKernel.

Related log / screenshot

[ERROR] ... For 'Real', it got empty inputs or outputs, which is invalid.
RuntimeError:
- C++ Call Stack:
mindspore\ccsrc\plugin\cpu\cpu_device_context.cc:543 mindspore::device::cpu::CPUKernelExecutor::CreateKernel

Special notes for this issue

• Paddle version in repro: 3.2.0