[davrens101]

Selecting modes_to_compare=[3,1] currently labels mode 1 as mode 3 and vice versa: the retained arrays stay in ascending order, but receive requested-order labels. Selecting fewer modes also changes their dressed frequencies because unreported modes are deleted before the Lamb-shift sum.

Calculate the Hamiltonian quantities from all simulated modes, then select the requested rows and matrix columns from the output DataFrames. Both entry-point docstrings now describe the one-based reporting selection and full-mode Lamb shifts. Public signatures and return keys are unchanged.

Actual three-mode/two-junction CSV reproduction:

• Before: mode_3 reports 4 GHz and mode_1 reports 8 GHz for selection [3,1].
• After: they correctly report 8 GHz and 4 GHz. Dressed frequencies are 7.950556875742865 and 3.905084128575915 GHz, matching the full-model results.
• All 15 ordered nonempty selections match the corresponding full-result rows/columns across all six returned tables. The full Lamb shifts also agree with an independently evaluated EPR half-sum equation. Invalid mode zero is rejected.

Verified with actual imports and temporary CSV/config files, without mocks or an external solver. No permanent tests added, per contributor preference. Independent read-only AI review found no blocker. CPU-only; no full-suite or external EM-solver validation claimed.

Review and assistance

Human review: davrens101 confirmed reviewing the work live before submission. 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.