# Suspense-Account Resolution — final local validation

25 October 2026 · D30-F03-A04 · private local package. Root owns the shared visitor synchronization and final iframe/public-route receipt.

## Definition and arithmetic

The root-frozen exact API and independently prescribed **78 complete transition/state cases** were written before the algorithm. Eleven cases are accepted; the remainder cover rejected, indeterminate and replay routes. No implementation was imported to generate the independent expected journals or active totals. Retained pinned Exact Reference Match inputs provide **105 complete independent source-result cases**.

Canonical independently calculated sequence: origin 700; classify 400 -> classified 400 / remaining 300; classify 300 -> 700 / 0; reverse and replace the original 400 -> active R-2(300) + R-3(400) = 700 / 0. A correction of the replacement remains 700 / 0. Origin cash delta stays -700; no classification or correction leg names cash. Reversal and replacement are balanced separately and modeled as one returned group. All original records remain unchanged.

Origin recording 20 October 12:00 precedes source observation 21 October 08:00. Equality is accepted; recording one millisecond after the observer rejects binding chronology. New classification effective time cannot precede origin recording; correction effective time cannot precede the target. Approval IDs cannot bind two different accepted commands.

## Executed checks

- `python -B tests/test_algorithm.py`: four suites pass; complete independent outputs, detached state/result isolation, every accepted journal balanced/no cash leg, exact conservation and object-order-independent replay.
- `python tests/run_parity.py`: strict TypeScript compilation and **78 transition cases / 11 accepted + 105 source cases**; entire Python/TypeScript output/state equality, not selected metrics. See `parity-results.json`.
- Capacity: independently prescribed 128-transition chain validates, exact accepted replay succeeds at the limit and a new identity rejects `STATE_CAPACITY`.
- `node tests/build_lab.cjs`: exact canonical TypeScript and pinned selector compiled into the offline HTML. Reference SHA-256 `474d5f4d429f0579a8e7257ceea14428ad564288b37b7a539f0911992ee600fa`.
- `node tests/browser_qa.cjs`: Chrome rendered five scenarios at 390, 768 and 1440 px; full canonical 12-attempt lifecycle, final version 4 / classified 700 / remaining 0 / origin cash -700; no document horizontal overflow, controls at least 44 px, no console warnings/errors. Audit table has an explicitly labeled local horizontal scroll on narrow screens.
- Browser interaction: Step/Back/Play/Pause/Reset, running speed change, keyboard arrows, scenario timer cancellation and reduced-motion one-step Play without timer. Incomplete source repaired under the same unaccepted identity; pending/self-approval rejection, exact maximum 30-digit money and reversed observer chronology inspected.
- Both standalone SVGs parsed as XML and decoded/rendered by Chrome. Both workflow and inline-article Mermaid sources parsed/rendered; screenshots retained under `visuals/qa/`. Actual mobile lifecycle, both static diagrams and workflow pixels inspected; readable spacing adjusted before final browser rerun.
- Final full topic-package validator: 11 Markdown files, zero errors and zero warnings. Playground validator: one checked, one passed. Standalone Python usage example reproduces all four canonical accepted states.

## Primary evidence and editorial checks

ACCA article, Federal Reserve May 2025 FAM4.50 and OCC handbook version 1.0/2025 cover annotation were checked online. OCC printed page 27 was independently downloaded as a document, rendered and visually inspected; suspense placement does not automatically justify reporting-asset classification. Federal Reserve relevant page text and the official screenshot endpoint were checked; no local PDF download is needed for its bounded contextual claim. Source facts, author controls and synthetic data remain separated in REFERENCES and CLAIM-LEDGER. No real-bank historical fixture or legally authenticated approval is fabricated.

Article has descriptive YAML title, inline SVGs/full-size fallbacks, inline Mermaid, actual code/fixture links and downloadable examples receipts. Related tutorial names verified against catalog; next tutorial is Balance-Proof Algorithm. Video is a scene-based script/asset map with exact canonical outro instructions, not encoded media. Public downloads live in examples; canonical research/test links remain local where appropriate.

## Scope of proof

This is a sequential accepted-snapshot model. `committed` does not mean durable database commit. Trusted source provenance, access control, reporting treatment, global origin allocation, global journal/approval uniqueness, non-overlapping ownership and actual transactional item-version/identity/journal/link writes are host obligations. No SQL server, production service, remote publication or new infrastructure was touched for this topic. External withdrawal remains evidence history; zero suspense does not prove statement confirmation or balance reconciliation.