Return to the full study

KEDDEH SYSTEMS · E-QUANTUM-REVIEW · Technical editorial E1

KEDDEH SYSTEMS | RESEARCH AND FORMAL PUBLISHING | ENGINEER · AUTOMATE · SCALE

P15 Quantum-State Module: Nominal Replay and Admission-Boundary Review

KEDDEH-TECH-E-QUANTUM-REVIEW-001 | Technical editorial E1 | 18 October 2026
Aboudy Keddeh — Systems Architect
Keddeh Systems | Internal research · Owner review | Editorial draft

Authorship, rights and review standing

1. Abstract and review decision

The P15/A15 quantum-state module reproduces all nine nominal self-tests and the finite-input Bell reference calculation. Three adversarial input cases also expose missing admission checks: a nonfinite matrix can pass the unitary predicate and produce a nonfinite successor; a negative measurement draw and a draw at the excluded upper bound can return zero-probability outcomes.

The appropriate conclusion is a bounded functional result with explicit admission gaps. The finite Bell path remains reproduced. The API is not qualified for arbitrary inputs merely because 9/9 nominal tests pass. No source patch, native quantum derivation, physical-device experiment or production release is part of this review.

Source coverage and new replay

The original historical review script and JSON receipt were inspected earlier but their raw files were not recovered for this package. Their expected hashes are retained as historical evidence gaps. The exact P15 source and quantum mapping are included. A new editorial reproduce.mjs reproduces 9/9 nominal checks and all three adverse-input findings on Node v24.19.0; reproduction-result.json is explicitly new evidence. This edition's new replay verifies the module digest without repeating the whole-carrier comparison; the combined E-LINE/E-QUANTUM edition performs that additional comparison.

2. Reviewed source and prerequisites

The reviewed source is the extracted p15-a15-runtime.js from the retained Native Host Adapter R14_CONVERGED. Its SHA-256 is 7c00d68f35135504e8a0348611c877ff47208e6d579e0d02160e5cfaf8a2eeca, with 6,166 bytes. The whole adapter retained in the companion E-LINE/E-QUANTUM package is 21,626,453 bytes with digest f5cde46ae784fff5d99d0e7555c7da23b5c94afa93dd48b10f40892ba20e4914; it is not included in this edition's source-files.

The historical review script read the retained adapter as text, extracted its p15-a15-runtime script and required exact equality with the supplied extracted module before evaluation. The current package-relative replay checks the extracted module's digest; its carrier field is null because the whole adapter is not supplied here. The source investigation separately found that this P15 extraction matches the original Drive carrier's P15 script, while the whole local and Drive carriers differ. This report therefore identifies the exact retained artifact and does not assert whole-file equivalence.

The retained execution environment is Node v24.19.0. The review uses fs to read local sources, crypto to calculate hashes and vm to evaluate only the extracted pure module. No document, browser, I/O API, resident invoker or process hook is supplied to the evaluated module. The full HTML and S29 runtime are not executed.

The historical review identifies independent-quantum-review.mjs and its JSON receipt; their raw files are not included in this package. That unavailable script assumed its original source layout and wrote a result receipt. A package-relative reproduction harness must keep its own output separate from source-files so dated evidence is not overwritten.

3. API being tested and nominal lifecycle

P15 exports KEXQuantumResearch with KQuantumState, built-in matrices G, basis, bell, unitary, separable2 and selfTest. Its optional DOM rendering is outside this review. The representation maps internal K_NEG and K_POS labels to external |0⟩ and |1⟩. Complex vector coefficients may contain numeric zero; the all-zero vector is rejected. These are conventional representation values, not native origin definitions.

A constructor copies and normalizes amplitudes, subject to tolerance. apply checks square dimension, multiplies a matrix by the state and constructs a successor. Tensor composition multiplies component pairs. Measurement uses squared magnitudes and a supplied/default draw, returning an outcome, selected probability and normalized one-hot successor. separable2 checks a four-amplitude pure state's coefficient determinant.

The nominal lifecycle is basis preparation → Hadamard transformation → tensor product with a second basis state → CNOT → Bell probabilities/non-separability → deterministic measurement probes. The module's selfTest is a nine-check convenience suite rather than complete input validation.

4. Independent method and reproduction

The historical review first compared embedded and extracted bytes. The current replay instead verifies the packaged module digest and records runtime identity. It creates a fresh VM, evaluates the exact extracted module, calls selfTest, computes bell(), records labels/probabilities and separable2, and evaluates measurement at 0.25 and 0.75.

For boundary probes, complex numbers are ordinary objects with re and im fields. The nonfinite matrix is [[NaN+0i, 0+0i], [0+0i, 1+0i]]. The review separately asks unitary(matrix) and whether basis('K_NEG').apply(matrix).norm() is finite. It then calls basis('K_POS').measure(-0.1) and basis('K_NEG').measure(1). These inputs test admission behavior; they are not valid physical measurement inputs.

Run python verify_sources.py in this edition folder to verify packaged source custody. Run node reproduce.mjs for the package-relative pure-module replay. Its expected output keeps nominal tests, boundary cases and predicate observations separate. If the complete retained carrier is included, the harness should additionally enforce module/extracted-script equality; otherwise its report must state that the already-pinned module is being replayed without a new carrier comparison.

Reproducing the historical independent-quantum-review.mjs procedure requires recovery of that original script, its original receipt and the exact input carrier. Those historical files are not supplied here. Keep any recovered originals unchanged and write new results separately; the portable command above runs the available new replay.

A reproducible outcome includes runtime version, module digest, original/extracted identity-check status, nine nominal results, Bell readouts, three boundary-case classifications and actual observations. A later patched module requires a new source digest and new receipt; it must not silently replace this original.

5. Actual results

5.1 Nominal suite

The new isolated replay reproduces the historical 9/9 nominal result:

• Q01: basis normalized.
• Q02: Hadamard superposition probabilities.
• Q03: selected H, X and CNOT reference gates are unitary.
• Q04: tensor product.
• Q05: Bell state and non-separability.
• Q06: normalized measurement successor.
• Q07: all-zero state rejected.
• Q08: finite non-unitary falsifier recognized by unitary.
• Q09: external |0⟩ basis label retained.

Bell labels are 00, 01, 10 and 11. Probabilities are [0.4999999999999999, 0, 0, 0.4999999999999999]. separable2 returns false. Draw 0.25 returns 00; draw 0.75 returns 11. Floating-point representation explains the displayed approximation to one half; no exact-arithmetic result is claimed.

5.2 B01: nonfinite matrix

Expected admission: reject a nonfinite matrix, and never identify it as unitary. Actual result: unitary returns true; apply produces a state whose norm is not finite. The case fails its admission expectation.

The code compares deviations with greater-than tolerances. JavaScript comparisons involving NaN do not behave as finite magnitude tests, allowing this input to bypass the intended predicate. Constructor normalization similarly lacks an explicit finite-norm gate. This is source-based diagnosis of the observed result, not a patched implementation.

5.3 B02: negative draw

Input is measure(-0.1) on K_POS, whose probabilities are [0,1]. Expected admission is rejection of a sample below zero. Actual outcome is external label 0 with reported probability 0 and a one-hot successor at that outcome. The case fails its admission expectation.

5.4 B03: excluded upper-bound draw

Input is measure(1) on K_NEG, whose probabilities are [1,0]. Expected admission is rejection because valid draws lie in [0,1). Actual outcome is external label 1 with reported probability 0 and its one-hot successor. The case fails its admission expectation.

These are three adversarial cases, zero passing admission expectations and three failures. B01 produces two observations, making four boundary observations altogether. They are not four independent cases. The nine nominal tests remain a separate denominator.

6. Interpretation, limits and next engineering steps

apply does not call unitary. Even a finite matrix diag(2,1), correctly recognized as non-unitary, is accepted and renormalized; the companion module review records Hadamard probabilities changing to approximately [0.8,0.2]. Therefore a caller cannot infer quantum admissibility from apply success.

A candidate hardening change would validate finite numeric complex components, shape and nonzero finite norm; require an admitted transition policy before apply; validate explicit draws as finite and in [0,1); and test label/state consistency. These are proposed remedies. They have not been applied to the preserved source, and this edition does not claim they solve all numerical or API risks.

A future regression suite should retain Q01–Q09 and B01–B03, add malformed shapes, infinities, custom-label inconsistencies, mutation of public state, tiny norms, numerical-tolerance boundaries and deterministic replay. The exact desired policy must distinguish arbitrary linear transformations from admitted quantum gates. Adding a finite-value check to unitary alone would not make apply enforce it.

The review does not cover S29 resident execution, embedded R4 behavior, the complete host UI, physical quantum hardware, algorithm complexity, speedup or native KEX/ZCG semantics. It does not supply a line-membership-to-amplitude derivation. P15's conventional external representation must remain separate from current literal .1 and from native constituent-retention requirements.

7. References and package contents

• E-QUANTUM: P15/A15 module in Native Host Adapter R14, exact retained artifact identified above.
• E-QUANTUM-REVIEW: independent-quantum-review.mjs and independent-quantum-review.json, 18 October 2026. The original JSON digest is 1040a6417f945c4d6f4ca162aaa1dc0973c79574d2ad41c14644d378d5cec575; original script digest is 13bf1639453947b37430b0bf8c6c0527e2756467db885afb53afad03937ce798.
• Combined E-LINE/E-QUANTUM edition: complete callable contracts, S29/P15 separation and source-progression context.
• Packaged supporting sources: p15-a15-runtime.js and quantum-model.json. The historical quantum-rerun receipt is not included; new evidence is reproduction-result.json. Source/receipt hashes are recorded in original_hashes.json; reproduction tooling is separate.
• KEDDEH-STD-PUBLISHING-001 v1.0, Publishing Standard.

8. Revision history

18 October 2026 — Technical editorial E1. Expanded the historical review account and recovered module into a method, API-boundary, result and reproduction document. Preserved all three failing admission cases beside the nine passing nominal tests. No code repair or new external approval is asserted.

© 2026 Aboudy Keddeh | Keddeh Systems | All rights reserved
KEDDEH-TECH-E-QUANTUM-REVIEW-001 | Technical editorial E1 | Internal research · Owner review | Editorial draft