An Open Dossier · 001 · audit trail · 12 June 2026

Temporal-Imaging-Enhanced Dispersive-Optics Quantum Key Distribution

Irfan Ali-Khan — Independent Researcher, Saratoga, California · companion verification record to the manuscript
claims 16 / 16 pass citations 36 verified · 1 flagged red team 1 resolved · 7 open human residue 3 items
Explore the result Self-explaining edition Read the PDF Audit trail Source & issues
01 · The claims, executable

Run the paper

Every number in the manuscript — Schmidt limits, magnification gains, break-even lens efficiencies, Table 1, the 108 b/s headline — is recomputed below from its stated assumptions, in your browser, when you press the button. This is the same arithmetic as the repository's verify_numbers.py; view this page's source to audit it.

verification console · 16 checks · model: Eqs. (2)–(7)
IDLEAwaiting run. No claim on this page is asking to be trusted; each is asking to be executed.
During this audit, recomputation caught the original Table 1 overstating photon information efficiency by 1.5–2.5 bits (lens point-spread silently dropped). The manuscript was corrected; the 10⁸ b/s headline survived. That catch is the argument for this page.
02 · Adversarial review

Red-team findings

A hostile pass over the finished draft, published with the paper rather than hidden from it. Resolved findings were fixed; open findings are stated in the manuscript as caveats and listed here ranked by referee risk.

RESOLVEDTable 1 and abstract overstated PIE by 1.5–2.5 bits
Recomputed values from the caption's own parameters: 13.07 and 14.42 bits where 14–15 and 16–17 were claimed. Cause: prose silently assumed negligible lens point-spread. Abstract, Table 1, and two Section 4 figures corrected; headline rate claim verified to survive.
OPEN · HIGHEST RISKThe nonlocal-imaging claim has no derivation in the paper
Section 5 asserts the distributed imaging condition at the Gaussian level, citing the Tsang–Psaltis/Patera formalism, but shows no propagated two-photon amplitude and no explicit sign conventions. This is the one original derivation, and the most likely location of a real physics error. Author disposition (v1.0): hand derivation declined by design — manuscript softened to "is expected to yield," claim C11 labeled OPEN-UNVERIFIED in the ledger, and the derivation posted as an open challenge: the first correct treatment, confirming or refuting, is credited in v1.1.
OPENLens field of view vs. cw frames — the two key assumptions are in tension
Headline PIE assumes 64 ns cw frames; the enabling lens covers a far smaller temporal aperture. The paper names the tension but does not quantify the duty-cycle penalty of a gated lens or the reduced Schmidt number of a pulsed/Talbot pump. One honest paragraph would close it. Sharpened by author reality-check (12 Jun, issue #1): the penalty is ~1/M of accepted input time per channel; claim C07 and all rate projections are now labeled architecture-dependent pending a quantified aperture architecture.
OPENSaturation-rate argument needs one inequality
"Pump harder to compensate lens loss" is bounded only verbally by multi-pair statistics; the mean-pairs-per-frame ceiling and the value of the coincidence-to-singles ratio κ are unquantified.
OPENNovelty claims carry a literature-recall ceiling
No prior lens-in-QKD integration was found, but conference abstracts and in-review work are poorly indexed, and U.S. Patent 8,744,086 covers adjacent territory. Strongest mitigation: preprint quickly rather than polish long.
OPEN"Parity with record" has a short shelf life
115.8 Mb/s (Li et al. 2023) is the verified DV record, but Gb/s-class CV-QKD preprints exist. The text hedges correctly; expect to refresh this paragraph at proof stage, and never let outreach copy say "fastest ever."
OPENTwo security statements to tighten
The bypass audit is a calibration check, not a proof element — say so plainly. And the finite-key proof's "carry-over" through the magnifier's modified POVM should read "expected to carry over" unless verified against the proof's assumptions.
OPEN · MINOROne bibliography page number from secondary sources
Grünenfelder et al. page 422 triangulated, not fetched from the publisher. One-minute check before submission.
03 · Sources

Citation audit

All 37 bibliography entries checked against primary sources or triangulated through three or more independent citing works. 36 verified; 1 flagged. The full table:

Show all 37 entries
#ReferenceStatus
04 · The method, reusable

The verified open research pipeline

The architecture that produced this dossier, written so the next project gets the same treatment. The numbering is the actual execution order. Governing principle: decorrelate the verifier from the generator — every claim is checked by code, by a fresh adversarial session, or by the human, and the human's hours go only where human judgment is irreplaceable.

  1. Conception & prior-art reconnaissanceExtract candidate novelty claims; search each claim, its components, and their combinations; build the established / adjacent / open table. Kill or pivot if the center is taken.
  2. The claim ledgerEvery claim typed — CITE, NUM, DERIV, NOVEL, EST — and every type routed to its mandated verifier.
  3. DraftingManuscript written against the open ledger; reproducibility section and AI-assistance disclosure are mandatory boilerplate.
  4. Mechanical verificationExecutable recomputation of every number (fail = fix the paper, never the tolerance); full citation re-audit at final draft.
  5. Adversarial reviewFresh AI sessions — ideally different models — briefed as Referee 2 with instructions to reject. Findings published, RESOLVED or OPEN.
  6. Human-judgment residueA short, named list: original derivations by hand, novelty claims against domain memory, the boldest abstract sentences. Signed item by item.
  7. Publication & community reviewarXiv first for citable priority; repository with CI as the verification layer; this page as the reading layer; preprint emailed to the groups it builds on — the fastest deep review on earth.

Want to publish like this?

Open Dossier is a free template — executable claims, a self-explaining edition, a DOI and blockchain timestamp on every release. From idea to live publication in an afternoon, with an AI agent doing the heavy lifting. No LaTeX, no gatekeepers, no fees.

Publish your own →