# Quantum v4.34: post-depolarized BBPSSW Local gamma/scoped candidate based on `ad83f5a23cade7b6d58ae7f2447d21a06e1c9b69` (v4.33). Public Leanbook 1.0.0 and main remain unchanged; no push or release. ## Exact operator model `noisyPurified gamma M = (depolarizing gamma).map (purified M)` composes the existing ideal BBPSSW instrument and finite Clifford twirl with a trace-preserving depolarizing channel on the entire retained pair. Gamma has type Visibility. This is post-processing noise, not a general model of bilateral CNOT errors, input noise, measurement errors or noisy networking hardware. For every joint input matrix, the composite map is CP, complex linear and preserves the trace of the accepted purified output. It preserves PSD and, together with the unchanged rejected branch, has full trace balance. On two independent Werner inputs the proved matrix identity is noisyPurified gamma (W(F) tensor W(F)) = P(F) • W(R_gamma(F)) R_gamma(F) = (1-gamma)/4 + gamma*R(F). The matrix identity and trace P(F) hold for all real F. Physical conditional density claims cover F in [0,1], including below 1/4. Dividing by P(F)>0 gives a PSD unit-trace Werner state. Pre-depolarizing both inputs instead would change the argument to F_gamma and the success probability to P(F_gamma); that is not the implemented model. ## Fixed points and working window The exact factorization is R_gamma(F)-F = (4F-1)*Delta_gamma(F)/(36*P(F)) Delta_gamma(F) = -8F^2+4*(2*gamma+1)*F-(5-gamma) d(gamma) = 4*gamma^2+6*gamma-9 Delta_gamma(F) = (d(gamma)-(4F-(2*gamma+1))^2)/2. Thus the numerator is cubic. The quadratic factor remains after separating the universal fixed point F=1/4. Set gamma_c=3*(sqrt(5)-1)/4. - For 0<=gamma1/4 strictly decreases in one step. - At gamma=gamma_c, the quadratic is -8*(F-(2*gamma_c+1)/4)^2. Fixed points are 1/4 and the double root (2*gamma_c+1)/4; there is no strict gain for F>=1/4. - For gamma_c1/4, strict improvement occurs exactly when F_minus3/4. - R_(19/20)(1)=77/80; the stationary ceiling is (29+sqrt(31))/40<77/80. - The ceiling is actually a fixed point; 3/4 lies strictly inside the window. - Operator regressions include PSD, linearity, probability preservation, complete post-depolarization, and normalization of the physical input F=0. ## Validation and boundaries One module contains 42 named theorems and 50 exports (seven definitions and one suite structure besides theorems). These are code metrics, not counts of independent scientific results. The registry field is `noisy_bbpssw`. See the [inventory](quantum-v4.34-inventory.json) and [verification report](quantum-v4.34-verification.json). Checks are strict elaboration and transitive axiom audit, affected integration build, independent compiled-project axiom audit, targeted container regressions, metadata and knowledge checks. This is not a full verify-all, full test discovery or clean build. The last historical full checkpoint remains v4.21 with its original qualifications. No scientific-priority claim is made. The model uses two independently prepared identical resources. Neither that parameterization nor conditional normalization is claimed to be a linear CP single-input cloning map. No noisy-repeater limit, resource rate, expected yield, retry model, gate threshold beyond this specified channel, or iteration limit is included.