Round 2 verification audit for C021

Model: gpt-5

Claim: Fig. 2a: applying up to 16 cycles to a range of initial states drives the measured fidelity toward |Psi^->, with fidelity differences between input states decaying exponentially but still resolvable after 16 cycles.

Source alignment: source/main.tex:127-132 (Fig. 2a)

Prior official verdict: not_verified with failure_reason data_unavailable.

Executable evidence: run.py. Sandbox rerun logs: run.log.

Independent audit: I checked the paper inventory and source context. This claim depends on measured/raw experimental data or device-specific calibration values, and inventory.json records no paper-provided dataset/code artifact. The correct action is not to synthesize data; the data_unavailable verdict is preserved.

Decision: Round 2 verdict is not_verified with failure_reason data_unavailable and limitations ['data_unavailable']. Notes: Fig 2a is MEASURED fidelity vs cycles for several initial states. No raw experimental data or data/code link exists (inventory.json), so the measured convergence curves cannot be reproduced. figure.png is an illustrative simulated overlay only. Claim concerns measured data -> not_verified / data_unavailable.