[FabianHofmann]

Closes # (if applicable).

Changes proposed in this Pull Request

Note

The following content was generated by AI.

LinearExpression.is_constant checked nterm == 0. That misreads an expression without variables that still has term slots:

• a sparse store keeps at least one slot, so from_constant in a sparse model (since fix(spec): a sparse model built from a spec stays sparse #1145) is not constant;
• a dense sum can leave empty slots, e.g. ((x * 0) @ w).is_constant is False on both backends.

Each store now answers is_constant from its terms: csr.nnz == 0 (O(1)) for CSR, all labels -1 for dense. A zero-cell expression keeps the old answer, so x * x over an empty dimension stays quadratic.

Found via PyPSA's sparse CI on linopy master (PyPSA/PyPSA#2129): from_constant(m, 0) reported size == 1 on sparse vs 0 on dense.

Why not drop the one-slot floor in csr_nterm

Dropping it makes term-less results report 0 slots everywhere, also in dense models (reductions route through CSR). PyPSA then skips zero-cost components in opex(), which breaks released PyPSA 1.4.0 on dense models.

Checklist

• AI-generated content is marked (see AGENTS.md).
• Code changes are sufficiently documented; i.e. new functions contain docstrings and further explanations may be given in doc.
• Unit tests for new features were added (if applicable).
• A note for the release notes doc/release-notes.md of the upcoming release is included.
• I consent to the release of this PR's code under the MIT license.

---

Generated by Claude Code

[codspeed]

Merging this PR will not alter performance

✅ 181 untouched benchmarks
⏩ 181 skipped benchmarks¹

---

_{Comparing claude/tender-noether-ht5foe (514d7dc) with master (810d296)}

Footnotes

1. 181 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩