Fix two bugs in decompose and add test coverage for Stoner.analysis.functions

[Copilot]

Data.decompose is defined in Stoner/analysis/functions.py and bound onto Data via the @class_modifier decorator in Stoner/core/data.py. Two bugs were present in the implementation.

Bugs fixed

• Wrong column header for asymmetric output — when asym= was explicitly provided, the added column was labelled "Symmetric Data" instead of "Asymmetric Data".

• Shape broadcast error with explicit sym=/asym= columns — ycol is always promoted to a list internally, causing interp1d to return (n, 1) arrays. add_column does np.atleast_2d(data).T, which transposes (n, 1) → (1, n), producing a ValueError on assignment. Fixed by applying np.squeeze() to symd/asymd before use.

# Previously raised ValueError: could not broadcast shape (40401,1) into (201,201)
x = np.linspace(-5, 5, 201)
d = Data(np.column_stack([x, x**2, np.zeros_like(x), np.zeros_like(x)]),
         setas="xy..", column_headers=["X", "Y", "Sym", "Asym"])
d.decompose(sym="Sym", asym="Asym", replace=True)  # now works correctly

Tests added

New tests/Stoner/analysis/test_functions.py covering:

• decompose is importable from Stoner.analysis.functions and accessible as Data.decompose
• Pure symmetric (even) and antisymmetric (odd) component extraction
• Mixed even+odd decomposition
• Explicit output column writing with correct headers

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.