Fix Tensor.nonzero as_tuple behavior

[fallenmi]

Type

• Bug fix (non-breaking change which fixes an issue): Fixes inconsistent behavior of as_tuple in nonzero method between open3d and pytorch #7166
• New feature (non-breaking change which adds functionality). Resolves #
• Breaking change (fix or feature that would cause existing functionality to not work as expected) Resolves #

Motivation and Context

Tensor.nonzero() called its two existing core methods from the opposite as_tuple branches. As a result, the default and explicit False returned a per-dimension list, while True returned a single tensor.

This restores the branch mapping that existed before the pure-C++ pybind rewrite. It intentionally preserves Open3D's existing [num_dims, num_non_zeros] tensor orientation and the historical per-dimension list behavior. Changing the orientation or introducing a literal Python tuple is outside this bug fix.

Checklist:

• I have run python util/check_style.py --apply to apply Open3D code style
  to my code.
• This PR changes Open3D behavior or adds new functionality.
  • Both C++ (Doxygen) and Python (Sphinx / Google style) documentation is
    updated accordingly.
  • I have added or updated C++ and / or Python unit tests OR included test
    results (e.g. screenshots or numbers) here.
• I will follow up and update the code if CI fails.
• For fork PRs, I have selected Allow edits from maintainers.

Description

Plan summary:

1. Lock the documented historical branch contract.
2. Add a regression for the default, explicit False, and True paths.
3. Swap only the existing NonZero and NonZeroNumpy calls.
4. Align the injected Python documentation and tutorial output.
5. Validate with a clean CPU/Python build and core tests.

Validation on macOS arm64 with AppleClang 21 and Python 3.14:

• RED on main: the focused regression failed because the default path returned a list.
• Release CPU/Python package build: 722/722 build steps completed.
• Focused regression: 1 passed.
• python/test/core/test_core.py: 279 passed.
• Broader core test set: 359 passed, 1 skipped.
• python util/check_style.py --apply --no_parallel: passed.
• git diff --check: passed.

AI assistance disclosure: OpenAI Codex helped investigate and draft this change. I reproduced the regression, reviewed every changed line, and ran the validation above.

[update-docs]

Thanks for submitting this pull request! The maintainers of this repository would appreciate if you could update the CHANGELOG.md based on your changes.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB