erasure-qubits

Simulation study: erasure postselection leaves a non-Hermitian loss-asymmetry bias in a dual-rail memory. Verified (2 eta T)^2/24 scaling; swap echo vs calibrated classical inversion under shot noise, pulse error and calibration drift, with phase maps.

Simulation study: how often should a dual-rail repetition-code memory run imperfect, costly erasure checks? Exact flagged-window space-time decoder, check-cost phase map, efficiency and erasure-rate sweeps, d=3-7, 5 seeds.