Datasets:
claim_id stringlengths 15 47 | evidence stringlengths 55 183 | limitation stringlengths 40 110 | next_step stringlengths 32 96 | stage stringclasses 4
values | statement stringlengths 43 154 | status stringclasses 4
values |
|---|---|---|---|---|---|---|
predecessor-v14 | cjc0013/laplacian-matching-theorem-frontier-v14@453f9cdbfb552e385d4d2b980b7c0142a6cde33a | The predecessor does not contain the complete v15 Hall-pair partition. | Control the complete Schur image. | prior_dataset | The original v14 release is the immutable predecessor of this expanded frontier. | proved |
m-matrix-harmonic-comparison | data/certificates/block_schur_dirichlet_to_neumann_certificate.json | This comparison does not alone prove the graph-local lower envelope is nonnegative. | Prove the remaining core/interface inequality. | v15 | For lambda=max_j b_j/(D1)_j, the residual harmonic extension satisfies D^(-1)b <= lambda 1. | proved |
bounded-monotone-envelope | 199728 applicable sign partitions; zero failures; minimum 17/9. | Finite exact verification is not a dimension-free proof. | Extract a graph-local symbolic proof. | v15 | The monotone lower envelope is nonnegative on every applicable frontier coordinate through order eleven. | verified |
order-degree-envelope-split | data/certificates/bounded_replay.json; ten groups; zero baseline, identity, coupling-sign, or envelope failures. | The grouping is finite and does not establish a dimension-free inequality. | Generalize the cell constants to symbolic functions of order and maximum degree. | v15 | Each of the ten observed order/degree cells admits an exact scalar decoupling of the graph-local envelope. | verified |
sharp-unattached-hall-pair-family | data/certificates/block_schur_dirichlet_to_neumann_certificate.json; exact formulas, three positive cubic cases, and 29991 zero-failure symbolic checks. | The proof does not cover attached Hall neighbours, nonuniform residual signs, or larger critical Hall cores. | Extend the symbolic decomposition to the remaining attachment and sign classes. | v15 | The graph-local envelope is positive for every unattached two-vertex critical Hall core with uniform core signs and uniform opposite residual signs. | proved |
mixed-residual-sign-hall-pair-family | data/certificates/block_schur_dirichlet_to_neumann_certificate.json; lambda<2, core ratio at least 2, 119964 zero-failure symbolic cases, and 49932 exact blocks. | The proof does not cover the remaining uniform-residual-sign attachment classes or larger canonical cores. | Close the attached uniform-opposite-sign classes and then treat mixed core signs dimension-free. | v15 | The graph-local envelope is positive for every uniform-sign critical Hall pair with mixed residual signs and any Hall-neighbour residual attachment count. | proved |
attached-uniform-opposite-sign-hall-pair-family | data/certificates/block_schur_dirichlet_to_neumann_certificate.json; generic lambda<2 comparison, two exact boundary bounds, 59982 zero-failure symbolic cases, and 16662 exact blocks. | This is one component of the two-core/two-residual Hall-pair topology, not a proof for larger canonical cores. | Combine it with the baseline sign classes into an exhaustive Hall-pair partition. | v15 | The graph-local envelope is positive for every attached uniform-sign critical Hall pair with uniform opposite residual signs. | proved |
complete-two-by-two-hall-pair-sign-partition | 199728 exact blocks assigned once; all 25704 lambda>1 blocks absorbed; 329901 symbolic cases and zero failures. | The partition does not cover canonical Hall cores larger than the critical pair topology. | Lift the sign-local arguments to larger canonical Hall cores. | v15 | Four disjoint dimension-free families prove the graph-local envelope for every two-core/two-residual critical Hall-pair sign and attachment pattern. | proved |
harmonic-unit-cap | 3096 exact failures through order eleven; maximum 14/9. | This refutes a shortcut, not the surviving comparison. | Use lambda rather than a unit cap. | v13 | The residual harmonic extension is always coordinatewise at most one. | refuted |
rank-one-interface | data/certificates/core_residual_rank_one_compression_obstruction.json | The arbitrary-rank Schur route survives. | Preserve full coupling topology. | v13 | Every core-residual interface has rank one. | refuted |
trace-cut-shortcut-coverage | 4 exact-trace failures, 4 weighted-cut failures, and 32 triangle-corrected failures. | The full M-matrix and monotone-envelope certificates still pass; these are not theorem counterexamples. | Use the order/degree-conditioned envelope rather than the weaker shortcuts. | v15 | The exact-trace, weighted-cut, and triangle-corrected sufficient bounds cover every order-eleven signed matrix. | refuted |
graph-local-envelope | Exact order-eleven frontier support plus a complete dimension-free certificate for the two-core/two-residual Hall-pair topology; no certificate for larger canonical cores yet. | This remains the branch obligation only beyond the completed Hall-pair topology. | Prove or falsify the attachment-local inequality for larger canonical Hall cores. | next | Every admissible core coordinate satisfies A1_i >= lambda(-B1)_i. | open |
vc3-theorem-alternate-route | cjc0013/vc3-laplacian-matching-theorem@4494842b0bd0bdb7d935b68a153f8581a6c84a03 | The later proof does not establish the v14 graph-local envelope inequality. | Retain the v14 inequality as an independent branch-level open problem. | v15 | For every connected simple G with tau(G)<=3, mu_1(G)<=Delta(G)+nu(G). | proved |
Laplacian Matching Theorem Frontier v15
Produced by the Ouroboros AI Research System, under human direction.
This private-review dataset advances
cjc0013/laplacian-matching-theorem-frontier-v14, pinned at
revision 453f9cdbfb552e385d4d2b980b7c0142a6cde33a. It does not claim the unrestricted Laplacian
matching theorem and does not claim that the current unmatched-vertex branch
is closed.
What changed in v15
- Exact replay rows now expose the attachment-local Schur decomposition instead of only scalar summaries.
- For the Hall residual block, with
b=-B^T 1,r=D 1,h=D^(-1)b, andlambda=max_j b_j/r_j, the dimension-free M-matrix comparisonh <= lambda 1is proved. - This reduces effective-core positivity to the graph-local inequality
A 1_i >= lambda (-B 1)_i. - The exact frontier now reaches order eleven and evaluates the lower envelope on
199,728 applicable sign partitions: zero comparison failures, zero
nonnegative-envelope failures, and unchanged sharp minimum
17/9. - A new finite certificate conditions on the ten observed
(order, Delta)cells. In each cell, thelambda <= 1branch usesA1 >= -B1; thelambda > 1branch verifiesmax lambda <= min A1/(-B1)with exact rational constants. - The sharp finite witness is now embedded in a dimension-free proved family:
a two-vertex critical Hall core, one unattached Hall neighbour, uniform core
signs, and uniform opposite residual signs. A 29,991-case symbolic replay has
zero failures and minimum cleared gap
17/90. - A second dimension-free proof closes every Hall-neighbour attachment count for
uniform core signs and mixed residual signs. It covers 49,932 exact blocks,
including 21,864
lambda > 1cases, and its 119,964-case symbolic replay has zero failures. The proof combines the strict boundlambda < 2withA1/(-B1) >= 2. - A third dimension-free proof closes the attached uniform-opposite-sign cases.
The generic comparison is again
lambda < 2 <= A1/(-B1); the sole boundary coordinates atDelta=4,d=3satisfy the sharper exact bounds10/9 < 10/7for one attachment and20/23 < 7/4for two. It covers 16,662 exact blocks, including 1,890 withlambda > 1, and 59,982 symbolic cases with zero failures. - Together with a baseline
lambda < 1proof for mixed core signs or uniform cover signs, the four disjoint Hall-pair families partition all 199,728 exact blocks and absorb all 25,704lambda > 1blocks. The combined 329,901 symbolic cases have zero failures. This completes the two-core/two-residual Hall-pair topology, while larger canonical Hall cores remain open. - The earlier false shortcut
h <= 1remains retired: 3,096 exact cases exceed one, with maximum14/9. Order eleven also exposes 4 exact-trace, 4 weighted-cut, and 32 triangle-corrected sufficient-bound failures without producing a graph-local-envelope failure.
v15 theorem-route context
A later, independently replayed packet proves the VC3 theorem through a different
Hall-classification and Schur route. The authority is pinned to
cjc0013/vc3-laplacian-matching-theorem at
revision 4494842b0bd0bdb7d935b68a153f8581a6c84a03. This does not prove the v14 graph-local
inequality; it makes that inequality an optional branch-level problem rather than
a remaining requirement for the VC3 theorem.
Current claim boundary
- Proved dimension-free: canonical Hall-core reduction, rank-agnostic block
Schur algebra, Hall-residual M-matrix positivity, and the new harmonic comparison
h <= lambda 1; also a complete four-family sign partition for every two-core/two-residual critical Hall-pair topology. - Verified exactly: the graph-local monotone lower envelope is nonnegative over the complete order-eleven frontier, including a ten-cell order/degree split certificate.
- Open: extend
A 1_i >= lambda (-B 1)_ibeyond the completed Hall-pair topology to every admissible larger canonical Hall core, or find a genuine counterexample. - Proved by a later independent route: for connected simple graphs with
tau(G) <= 3,mu_1(G) <= Delta(G) + nu(G). - Not proved: the unrestricted all-graphs theorem. The v14-specific graph-local branch also remains open and is not represented as closed.
Verification surfaces
Exact rational replay is the primary mathematical evidence. CrossHair supplied a bounded symbolic falsifier for the two-residual-coordinate attachment inequality. Temporal-model checking guarded proof-state transitions; fixed-point provenance analysis checked claim propagation; answer-set checking validated the research obligation ordering. Probabilistic model checking was intentionally not applied because no calibrated theorem-workflow probabilities exist.
See FRONTIER.md for the derivation, CHANGELOG.md for the v14-to-v15 delta,
data/claim_ledger.jsonl for machine-readable claim status, and REPRODUCE.md
for exact regeneration instructions.
Review status
The bundle is prepared for private human review. Public visibility, mathematical endorsement, authorship, licensing, and any stronger claim remain human decisions. No priority or novelty claim is made.
- Downloads last month
- 22