[davrens101]

The capacitance postprocessors document NumPy matrix input, but both call .values and reject actual ndarrays. The single-qubit helper also crashes on documented optional configurations: omitted qubit frequency, or a four-conductor qubit/readout circuit without a feedline.

• Convert supplied matrices with np.asarray in both one- and two-qubit entry points, retaining DataFrame/indexed-CSV support.
• Read resonator properties independently of the optional qubit-frequency calculation.
• Do not format unavailable linewidth/Purcell values as floats.
• Return f_q_GHz=None when no frequency was supplied, and document these cases.

No circuit formula, optimizer, public signature or existing numeric-zero placeholder changes.

Verified before/after with real imports and temporary Palace-format CSVs, without mocks:

• 18 single-qubit calls: 3/4/5-conductor circuits, with and without qubit frequency, each using ndarray, DataFrame and indexed CSV. All complete after the fix. Effective capacitance and charging energy match independent circuit equations; resonator outputs match the real resonator object.
• Three two-qubit calls using the same input forms and a real quarter-wave resonator agree numerically, including EC1 = 0.3283630743796598 GHz and qubit-qubit coupling = 0.8867184116435455 MHz.
• Changed Python compiles. Independent read-only AI review found no blocker. No new permanent tests; no full-suite or external EM-solver claim. CPU-only.

Review and assistance

I supervised and checked the AI-assisted work as it progressed. AI assistance was used for investigation, implementation and verification. No maintainer review or approval is claimed. Verification was CPU-only; no GPU validation is claimed.

[davrens101]

I cherry-picked all seven SQDMetal fixes (#91-97) onto their shared upstream base and they apply cleanly together. The existing focused tests give 14 passed, 21 subtests passed (19 source/dependency/SciPy warnings, not suppressed). I also ran the 15 ordered three-mode EPR selections, 18 single-qubit capacitance configurations, three two-qubit input forms, single- and multi-port linewidths, and absent optional files directly, and they all work. No new permanent tests for #95-97, and I did not run the full suite or an external EM solver.

[PP501]

Somewhat pointless even moving it here; either way, keep the comment # res-feedline mutual capacitance.

[davrens101]

Restored the # res-feedline mutual capacitance comment on C_c = C_rf in de3af06. The failing deploy check is the Sphinx docs deploy: it gets a 403 pushing to gh-pages from the fork PR, so it isn't caused by this change.

[PP501]

I can approve, once the small change has been made.

[davrens101]

Removed the redundant parentheses around res is not None. No behaviour change.