Round 2 verification audit for C026

Model: gpt-5

Claim: Correlated phase-flip errors (I_e Z_e + Z_e I_e) leave the fidelity unaffected because |Psi^-> resides in a decoherence-free subspace.

Source alignment: source/main.tex:145-145 (Theoretical error analysis)

Prior official verdict: verified with failure_reason None.

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

Independent audit: I scanned the copied script for imports/shared helper dependencies and reran it through the sandbox. The code is self-contained in this attempt directory and targets the claim strategy: Confirm |Psi^-> is invariant (up to global phase) under collective sigma_z rotations and hence is a +1 eigenstate insensitive to correlated dephasing.. I checked the relevant family model rather than relying only on exit status; the rerun is treated as one reproducibility input.

Decision: Round 2 verdict is verified with failure_reason None and limitations []. Notes: Claim: correlated phase-flip (I_eZ_e+Z_eI_e) leaves fidelity unaffected because |Psi^-> is in a DFS. (1) Structural: U_z(eps)|Psi^->=e^{i phi}|Psi^-> for all eps; ||=1.0 and residual ~1.6e-16 for eps in {0.1..2.3}. |Psi^-> lives in the {down,up} sector, sigma_z,de gives a common collective phase to both terms => global phase only. (2) Dynamical: per-cycle U_z(sqrt(2p)) leaves steady-state F=1 to machine precision (max 1-F=1.1e-14) for p in {0..0.1}. Clean mathematical fact (collective-sigma_z invariance of the singlet), not a reproduced measurement. verified.