# Quantum v4.21–v4.36 integration This integration brings the verified local quantum chain into the Leanbook main line based on public commit `5276529c6c1010722cc96c39217892b9e874f6a8`. The candidate chain ends at `7981d0f6e1ec9dfda342eb32a8ad312a53b6ef30`. It contains fifteen commits adding twenty formal modules. v4.25 is a historical prototype, not a separate commit. Existing theorem statements and proofs, toolchain, dependency manifest and verification mechanisms are unchanged. Integration adds documentation and restores the missing license header of QuantumAdaptiveWernerCascade, as required by the hosted style linter; its mathematical body is byte-for-byte unchanged. ## Formal coverage - v4.21–v4.24: correlated system/environment inputs, decoupled reduction, independent stages, post-analyzer coupling and CPTP post-processing. - v4.26–v4.28: two-round instruments, Werner reduction, adaptive delivered-message feedback, faithful detector loss, weighted conditional states and trace balance. - v4.29–v4.31: constructive twelve-operation Clifford twirling, operator BBPSSW, ideal orbit limits, binary resource-tree probabilities, heterogeneous inputs and the operator bridge between swapping and distillation. - v4.32–v4.33: finite repeater trees, nested purification depth, resource counts, one-round visibility decay and finite-scale quality preservation after a gain. - v4.34–v4.36: post-depolarized BBPSSW, fixed points and complete physical orbit classification for constant quality 0<=gamma<1. Ideal gamma=1 is covered by v4.30. These temporal fidelity iterates are not spatial repeater-map iterates. The last three modules have 90 named theorems and 102 named exports. Counts are code metrics, not independent scientific results. Each stage retains its source inventory, precise assumptions, regression tests and original scoped report. ## Noisy orbit classification The model applies a depolarizing channel to the entire retained pair after ideal BBPSSW and twirling. The success probability remains P(F), and R_gamma(F)=(1-gamma)/4+gamma*R(F). It does not model arbitrary errors inside CNOT. Let gamma_c=3*(sqrt(5)-1)/4 and F_star=(3*sqrt(5)-1)/8. For 0<=F0<=1: - 0<=gamma=1 equals 1/4; F0=1/4 is already stationary at index zero. - gamma=gamma_c: starts below F_star converge to 1/4; starts at or above F_star converge to F_star. The double root attracts from above, not from below. - gamma_c