Twenty-five runnable research utilities

The previous twenty-three utilities now include exact fixed-margin law/event auditing and source-oriented bounded-flow reduction/diagnostics. Thirty-two saved component reports contain 3,748 passed checks: 2,651 earlier plus 617 law/event, 460 flow, 13 existing-flow-solver and seven documentary-contract controls. Source proofs, unrestricted counting/sampling backends, real customer workflows and profitability remain unverified.

Exact finite movement comparison

kserver_reference.py accepts a rational metric, initial labeled tuple and fixed request word. It computes an attained offline optimum by conventional DP, replays a witness, and evaluates fixed-label, nearest-label and independent uniform-label policies exactly. It implements no source competitive policy. It rejects nonmetrics and caps eight points, four servers, 64 requests and three million state-label transitions. Budget exhaustion leaves the complete comparison unknown. Dimensions bound frontier/backpointer storage; work units exclude input validation and do not represent bit operations or wall time.

python3 research/tools/kserver_reference.py research/fixtures/movement-startup-2026-10-09-v1.json

137 checks compare complete short label-string enumeration with DP and policy expectations, then check witness replay, horizon distinctions, metrics and budgets. The saved 30-request startup fixture has fixed-label cost 30 and offline optimum 2. This is a small finite control for the source's activation/additive-cost distinction, not performance evidence on a customer workload.

Exact transport sharpness geometry

transport_sharpness_reference.py reproduces the source's uniform-square three-atom example. It clips rational affine halfplanes, integrates nine map-cell intersections, and compares masses and map norm with analytic formulas. Target W1/W2 use the explicit matching and second-coordinate lower witness; no general transport optimizer runs. It accepts a 256-bit rational parameter 0<a<1/2 and caps 10,000 polygon-vertex examinations. Partial work returns unknown. It proves no arbitrary transport-map or solver guarantee.

python3 research/tools/transport_sharpness_reference.py research/fixtures/transport-three-atom-2026-10-09-v1.json

131 checks cover rational geometry, target metrics, exact half-exponent growth and input/budget boundaries. Saved examples include movement comparisons, explicitly conditional activation schedules and transport controls. No source constructor time is measured.

Documentary and semantic records

The thirteen-contract registry adds language-class, uniform k-server and Brenier-map contracts, with 12 checks. Complete documentation still grants no applicability/independent-verification certificate and does not verify supplied evidence contents. The new joined report flags eleven gaps in one self-authored startup claim. It archives all six exact report inputs. Earlier bundles also have separately recovered input archives whose bytes match the original recorded hashes; their original manifests/reports remain preserved. 31 archive checks verify those bytes and historical registry counts without executing the archived tools.

65 selected definition observations have no detected intended-model mismatch in this limited review. All ten interfaces in this finite static queue are reviewed at source level; whole imports, proofs and programs remain pending. This is source inspection, not a trusted checker or whole-proof result. Operational limits.

Fourier and algebra references

audit_fourier_aliasing.py compares exact sparse |u|^(p-1)u with centered modulo-grid folding. It uses physical rational coefficients rather than weighted Sobolev coordinates. 110 controls include closed cubic formulas, independent direct expansions, complex single-wave formulas, a twelve-dimensional axis fixture and failure limits. Saved grid-three report.

python3 research/tools/audit_fourier_aliasing.py research/fixtures/fourier-aliasing-12d-2026-10-09-v1.json

chromatic_basis_reference.py enumerates proper colorings and reconstructs the entire exact finite polynomial after elementary-basis conversion for at most six vertices. An optional supplied nondescent-permutation witness is compared coefficient by coefficient. 106 controls cover closed formulas, all eligible h through four vertices, paper/Sage examples, witness mutation and budgets. This does not implement the source packet construction or certify its all-r witness. Algebra dossier.

python3 research/tools/chromatic_basis_reference.py research/fixtures/chromatic-path-three-witness-2026-10-09-v1.json

Prepared ten-claim experiment

The review packet contains ten self-authored scenarios, eight intentional scope errors and two synthetic documentary-completeness controls. It preserves source-only/report-assisted cards, a separate self-authored answer key and blank expert results. 25 controls check expected code boundaries and exact file pointers for all thirteen contracts. This is a prepared experiment; expert benefit, accuracy and buyer value remain unmeasured. Enriched registry.

Queryable permutation reference

queryable_permutation_reference.py has three modes: explicit shared-bit point/inverse replay; exact tiny full permutation laws; and deterministic short-seed support bounds. Missing bits and exhausted law budgets are unknown. 176 switching/law controls use independent physical all-coin enumeration and materialized/cycle tests. 21 support controls distinguish statistical full-law guarantees from seeded application goals. Four-card calibration does not select a universal sweep count.

python3 research/tools/queryable_permutation_reference.py research/fixtures/coordinate-sparse-63bit-2026-10-09-v1.json
python3 research/tools/queryable_permutation_reference.py research/fixtures/coordinate-law-d3-L1-m8-2026-10-09-v1.json
python3 research/tools/queryable_permutation_reference.py research/fixtures/permutation-entropy-d30-b256-2026-10-09-v1.json

The fourteen-contract enriched registry adds source trace-smoothing and shared-randomness obligations, with six controls. A synthetic complete record still verifies no evidence or mathematics. The new joined report flags twelve documentary gaps in a self-authored four-sweep seeded claim and archives its six exact inputs. New application dossier, SDK specification.

Full utility directory

| Fourier aliasing reference | Exact sparse odd-power convolution/circular-grid comparison | | Chromatic basis reference | Exact finite elementary expansion and supplied-witness audit |

| Queryable permutation reference | Shared-bit point/inverse, tiny laws and seeded-support bounds |

Dated validation drivers/output files preserve their earlier editions. Existing successful output paths cannot be reused. Finite checks, syntax/hash consistency and an elapsed work window do not establish broad algorithm superiority, buyer demand or profitability.

Shared-bit provider reference

permutation_provider_reference.py implements exact decimal-string point/inverse requests under an immutable plan. A 256-bit HMAC mapping provides reproducible seeded values, with no source statistical or security guarantee. A separate local POSIX append-only store assigns an OS-entropy bit once per switch, locks a whole bounded request batch, flushes/fsyncs every new row and reuses rows after restart. Its actual fairness, network filesystem behavior and power-loss recovery are unverified. A visibly separate test-tape kind supports explicit controls.

62 controls compare materialized swaps, query order, inverse/cycle behavior, high indices, local processes, bounded lock waits, capacity exhaustion, wrong epochs and preserved damaged files. 11 hardening controls check immutable plan isolation and the corrected small-store loader. Canonical/hash-linked records are not an authenticated history; tail removal and suitably rewritten hashes cannot be detected without an external anchor. Damaged stores are refused and never repaired by truncation.

python3 research/tools/permutation_provider_reference.py research/fixtures/permutation-provider-seeded-63bit-2026-10-09-v1.json --seed-hex 0000000000000000000000000000000000000000000000000000000000000000

The initial workload exposed a needless 8 MB read allocation. The preserved repeated run used the bounded actual file size: 16 queries on 256 slots retained 249 switches/35,972 store bytes, then full traversal reached 2,048 switches/293,486 bytes. A packed int64 permutation array for that domain is 2,048 bytes; this is a representation baseline, not measured Python list storage. Traced Python sparse peak was 43,318 bytes after the fix. No PyTorch, production network service, shard batching or real dataset task ran. The materialized reference is a simple Python oracle, not an optimized performance baseline; no latency speedup is inferred.

Initial and repeated workload controls each passed 14 checks. Timing phases share one provider assignment and are dependent, and the OS bits differ between runs. Tool bytes are preserved; the second run's provider source was explicitly reconstructed by reversing its later plan-isolation patch. Current production source, probability law and profitability remain unverified. Constant audit, SDK specification.

Exact palindrome calibration utility

palindrome_minorant_reference.py enumerates only two/four slots under the source-oriented recursive butterfly/palindrome definitions. It produces a full law, explicit exponent-zero minorant, exact rational centered regular matrix, scaled-projection identity and attaining eigenvector. 50 independent controls compare actual-card swaps, inverse sweep law and separately reconstructed transition columns, then verify conservative finite sweep bounds and typed caps.

python3 research/tools/palindrome_minorant_reference.py 2

The four-slot report has full support with 256 coin strings, minimum mass 1/32, explicit minorant mass 3/4/gap bound 5/8, and centered operator norm exactly 1/4 through B^2=(1/4)B. The conservative norm and minorant bounds reach the four-slot target at 13 and 37 sweeps respectively; direct exact TV already passes at six. These are distinct finite statements. The two-slot report is uniform. No universal numerical P or independent proof/program/randomness certificate follows. Source orientation and exact reviewed extents.

Public fixed-margin event-law audit

audit_contingency_law.py computes uniform and conditional-independence masses on a fully enumerated bounded support, holds one declared event fixed and reports its probability/threshold classification under each model. Bounds require an explicit conditioned-independence label. It caps dimension six, total 128, 2,000 tables and four million enumeration work units. Exhaustion never returns a partial law or test decision. 617 controls, public report.

python3 research/tools/audit_contingency_law.py research/fixtures/contingency-scipy-fisher-threshold-2026-10-09-v1.json

Bounded flow reduction and existing-solver distinction

bounded_flow_reference.py preserves the source private-vertex flow-to-table bijection, signed lower bounds, loops and parallel arcs. Its conventional exact engine caps table dimension six/total 128, with 2,000-vector diagnostics. Large valid reductions remain count-unknown; exact completed counts can survive a later diagnostic budget failure. 460 controls.

python3 research/tools/bounded_flow_reference.py research/fixtures/two-route-unit-flow-2026-10-09-v1.json
python3 research/tools/bounded_flow_reference.py research/fixtures/public-ortools-capacity-flow-2026-10-09-v1.json

13 baseline controls exercise NetworkX 3.4.2 on generated cases and a public model, with separate integer feasibility checks. It supplies one proposal, not a count or uniform sampler. The public model exceeds the exact reference at dimension 14/total 3,422; a valid solver proposal does not fill the unknown count. OR-Tools was not run, and no independent optimum, timing advantage or buyer value was measured.

The fifteen-contract registry adds the paper-level bounded-flow corollary with seven documentary controls. The new joined report archives six exact inputs and flags twenty gaps in a self-authored scope-error claim. The selected table Comparator does not separately select its flow bridge/sampler. New dossier.