Changelog — SCPN-Quantum-Control

v1.1.0 Latest

29 July 2026

Added exact finite parity-sector derivatives, topology-constrained proposal custody, multiscale synchronisation observables, and deterministic evidence replay.
Simulation-only forecasting, reservoir, sensing, and co-design lanes retain explicit refusals for unsupported hardware, deployment, efficacy, and advantage claims.
Current generated inventory: 697 Python source modules, 1,373 public classes, 177 Rust bindings, 1,155 Python test files, 841 public API exports, 109 notebooks, 38 domain families, and 37 examples.

17 July 2026 · amendments through 18 July

First stable public-API commitment across the hardware runner, control surfaces, analysis stack, and bounded differentiable-programming routes.
Release inventory: 643 Python source modules, 1,238 public classes, 177 Rust bindings, 1,095 Python test files, and 841 public API exports.
Committed tensor-network baselines state the classical-simulability boundary; no broad quantum advantage is claimed at current system sizes.
CHSH significance corrected per pair: 7.54σ for S=2.165 and 8.94σ for S=2.188.
Matched-basis re-run published with raw mismatch rates of 1.39%–3.00%; no QKD security, key-rate, or viability claim follows.
IBM v2 raw counts and job identifiers recovered publicly; the former hardware coherence-protection interpretation is retired, while the data remains available.

26 June 2026

SCPN-STUDIO federation surface — a scpn_quantum_control.studio capability manifest on the SCPN-STUDIO platform contract, declaring eight verbs (compile / simulate / analyse / validate / benchmark / replay plus mitigate / execute), nine studio.*.v1 evidence schemas mapped from the five-class hardware ledger, a content-addressed digest, and an additive architecture-map extension block for Hub ingestion.
Uncertainty quantification — zero-noise extrapolation now carries a propagated standard error and coverage interval (weighted-least-squares when per-scale shot-noise errors are supplied), and synchronisation metrics gain analytic delta-method and distribution-free bootstrap intervals. Mitigated and hardware-derived results now carry defensible error bars instead of bare point estimates.
Architecture documentation — an Architecture Map (per-stage IO contracts, backend/dispatch wiring, cross-repo integration points, honest scope boundaries) and an Architecture Decision Records page covering six major design decisions.
Capability inventory — 524 Python source modules, 919 public classes, 171 Rust PyO3 bindings, 853 test files, 100 notebooks, across 32 subpackages.

v0.9.6 – v0.9.12 · April–June 2026

Publications (25 May 2026) — Seven open-access preprints posted to Zenodo with citable DOIs: DLA parity hardware observation, Kuramoto–XY on 156 qubits, monitored feedback vs open-loop, reduced-Pauli entanglement checks, Fisher-information feedback, Rust/VQE methods, and the software-framework note.
v0.9.12 (15 Jun) — Upgraded the Rust PyO3/NumPy binding stack; refreshed public release metadata and documentation surfaces.
v0.9.11 (5 Jun) — Promoted the native compiler-backed whole-program automatic-differentiation determinant path.
v0.9.10 (4 Jun) — Polished the differentiable-programming documentation; documented the enterprise-readiness boundary.
v0.9.9 (2 Jun) — Added a public differentiable-programming documentation route; refreshed onboarding and quickstart.
v0.9.8 (1 Jun) — Added a public onboarding page; aligned the release-readiness audit coverage threshold.
v0.9.7 (19 May) — First GOTM-SCPN Paper 0 K_nm preregistered replay; Paper 0 methodology-paper outline.
v0.9.6 (17 Apr) — Security: bumped rand 0.9.2→0.9.4 for an upstream advisory; removed an IBM Cloud CRN prefix log in the job-retrieval script.

April 2026

DLA parity asymmetry confirmed on ibm_kingston (Heron r2, 156 qubits) — first publishable hardware result
342 circuits, 4–8 qubits, up to 21 repetitions per point, ~700,000 shots total
Mean asymmetry for depths ≥ 4: +10.8% (odd sector more robust than even sector)
Strongest signal at Trotter depth 6: +17.48% (21 reps, binomial p = 0.035)
7 of 8 depth points positive — statistically significant at p < 0.05
Consistent with the apriori classical simulator prediction (4.5–9.6%)

GUESS symmetry decay ZNE — physics-informed error mitigation using the XY Hamiltonian's conserved total magnetisation, Rust fit + batch extrapolation (Oliva del Moral et al., arXiv:2603.13060)
DynQ qubit mapper — topology-agnostic placement via Louvain community detection on calibration-weighted QPU graphs, Rust quality scoring (Liu et al., arXiv:2601.19635)
ICI pulse sequences — Pontryagin Maximum Principle optimal three-segment STIREP protocol with Lindblad decay simulation (Liu et al., 2023)
(α,β)-hypergeometric pulse shaping — unified adiabatic pulse family via Gauss ₂F₁, subsumes Allen-Eberly, STIRAP, and Demkov-Kunike (Ventura Meinersen et al., arXiv:2504.08031)

36 Rust exported functions (up from 22), all with PyResult<T> boundary validation
44× Rust speedup for hypergeometric envelope via custom ₂F₁ series expansion and rayon parallelisation
New Rust modules: validation.rs (16 unit tests), symmetry_decay.rs, community.rs, pulse_shaping.rs
Inner/outer function pattern for testability without Python context
Release profile: LTO fat, opt-level 3, codegen-units 1, strip debuginfo

March 2026

IBM Heron r2 hardware validation campaign (33 jobs, 176K+ shots)
Bell inequality violation: S = 2.165 (>8σ)
QKD QBER was reported as 5.5%; amended 17 July 2026 because the legacy value is not independently derivable and supports no QKD-security claim.
VQE 0.05% ground state error
3 novel synchronisation witnesses
DLA parity theorem: dim = 2^2N−1 − 2
FIM investigation series (34 notebooks)
Cross-repo bridges (sc-neurocore, SSGF)

February 2026

January 2026

Rust engine (PyO3): the former 5,401× cold-start figure is retired; current public wording uses the committed warmed declared-hardware baseline.
Error mitigation suite: ZNE, PEC, DD, Z₂
Identity subpackage: VQE attractor, coherence budget
Cryptography: BB84, Bell tests, topology-authenticated QKD
QEC: toric code, surface code, repetition code UPDE

December 2025 – January 2026