GEOPY-2880: Allow EM receiver orientations to be defined as 2D array (per location)

[domfournier]

GEOPY-2880 - Allow EM receiver orientations to be defined as 2D array (per location)

[Copilot]

Pull request overview

Enables EM receiver orientations to be provided per-location by extending the validate_direction utility to accept 2D arrays and normalize them row-wise.

Changes:

• Expand validate_direction to accept either a single direction vector or a per-location matrix of direction vectors.
• Normalize 2D orientation inputs along the last axis.

Suppressed comments (2)

simpeg/utils/code_utils.py:1225

• The validate_direction docstring still describes obj/return value as a single (dim,) vector, but the implementation now accepts and returns an (n, dim) array (per-location/per-receiver orientations). Updating the docstring will prevent API confusion.

    obj = validate_ndarray_with_shape(

simpeg/utils/code_utils.py:1234

• This change alters validate_direction semantics (2D inputs are now valid and should be normalized row-wise). The existing unit tests currently assert that a 2D input like [[1, 2, 3]] raises, and there are no tests covering the new (n, dim) behavior (including normalization and error cases like wrong second dimension). Please update/add tests accordingly to lock in the new contract.

    obj = validate_ndarray_with_shape(
        property_name, obj, shape=[(dim,), ("*", 3)], dtype=float
    )

    # Normalize the orientation
    # do this to make a copy of the input
    if obj.ndim > 1:
        return obj / np.linalg.norm(obj, axis=-1)[:, None]
    else:
        return obj / np.linalg.norm(obj)

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[gmcga]

looks like this caused a regression - test_direction_validation now fails as [[1, 2, 3]] no longer raises ValueError

[domfournier]

We have long abandonned maintaining the tests on simpeg side.