The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
schema_version: int64
release_name: string
release_date: timestamp[s]
repository_target: string
repository_type: string
visibility: string
status: string
primary_result: string
historical_bound: int64
exact_maximum_claimed: int64
sharp_non_pg24_maximum: int64
rigidity_threshold: int64
reproduction: struct<python: string, dependencies: list<item: null>, command: string>
child 0, python: string
child 1, dependencies: list<item: null>
child 0, item: null
child 2, command: string
keywords: list<item: string>
child 0, item: string
caveats: list<item: string>
child 0, item: string
equivalent_problem: string
johnson_notation: string
canonical_problem: string
project: string
claims: list<item: struct<id: string, statement: string, status: string, machine_check: string>>
child 0, item: struct<id: string, statement: string, status: string, machine_check: string>
child 0, id: string
child 1, statement: string
child 2, status: string
child 3, machine_check: string
to
{'schema_version': Value('int64'), 'project': Value('string'), 'status': Value('string'), 'canonical_problem': Value('string'), 'equivalent_problem': Value('string'), 'johnson_notation': Value('string'), 'claims': List({'id': Value('string'), 'statement': Value('string'), 'status': Value('string'), 'machine_check': Value('string')}), 'caveats': List(Value('string'))}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
schema_version: int64
release_name: string
release_date: timestamp[s]
repository_target: string
repository_type: string
visibility: string
status: string
primary_result: string
historical_bound: int64
exact_maximum_claimed: int64
sharp_non_pg24_maximum: int64
rigidity_threshold: int64
reproduction: struct<python: string, dependencies: list<item: null>, command: string>
child 0, python: string
child 1, dependencies: list<item: null>
child 0, item: null
child 2, command: string
keywords: list<item: string>
child 0, item: string
caveats: list<item: string>
child 0, item: string
equivalent_problem: string
johnson_notation: string
canonical_problem: string
project: string
claims: list<item: struct<id: string, statement: string, status: string, machine_check: string>>
child 0, item: struct<id: string, statement: string, status: string, machine_check: string>
child 0, id: string
child 1, statement: string
child 2, status: string
child 3, machine_check: string
to
{'schema_version': Value('int64'), 'project': Value('string'), 'status': Value('string'), 'canonical_problem': Value('string'), 'equivalent_problem': Value('string'), 'johnson_notation': Value('string'), 'claims': List({'id': Value('string'), 'statement': Value('string'), 'status': Value('string'), 'machine_check': Value('string')}), 'caveats': List(Value('string'))}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Hsin-Heart Exact 84: Two-Intersection Families on 21 Points
Exact candidate theorem, complete extremal classification, sharp stability gap, prior-art audit, and replayable verification package.
Headline result
For a family
[ \mathcal F\subseteq {[21]\choose 13} ]
such that every two distinct sets satisfy
[ |A\cap B|\in{7,9}, ]
the release proves the candidate exact value
[ \boxed{|\mathcal F|\le 84}, ]
with equality attained. Equivalently, after complementation,
[ \boxed{A(J^{21,8},{4,6})=84}. ]
The strengthened release additionally gives:
- complete extremal rigidity: every 84-family lies in the unique self-orthogonal ([21,9,8]) code associated with (PG(2,4));
- explicit classification of every extremizer by one incident-line choice at each of the 21 projective points;
- sharp stability: every valid family of size at least 82 lies in the (PG(2,4)) core, while a non-(PG(2,4)) family of size 81 exists;
- a direct 38-core verification route, replacing the earlier 18,216-shortening route;
- a wide prior-art and novelty audit.
Scholarly status: this is a reproducible candidate theorem/research release. The included verifier passes, but the result has not yet undergone independent peer review. Classical coding-theory, design-theory, Golay-code, and finite-geometry ingredients are explicitly separated from the claimed new conclusions.
Why this repository exists
This repository is designed for four audiences at once:
- Researchers — a compact statement of the theorem, proof architecture, historical context, and references.
- Referees / verifiers — deterministic source data, certificates, hashes, and a standard-library verifier.
- Search engines — explicit parameter notation and equivalent formulations.
- AI agents — machine-readable manifest, claim inventory, file map, reproduction command, and epistemic-status labels.
Equivalent searchable formulations
The same problem may appear under different terminology. This release deliberately indexes all of them:
- two-intersection family of 13-subsets of a 21-set, intersections (7,9);
- 8-subsets of a 21-set with pair intersections (2,4);
- fixed-distance Johnson-space code (J^{21,8}) with distances (4,6);
- (A(J^{21,8},{4,6}));
- binary constant-weight-8 code with Hamming distances (8,12);
- doubly-even ([21,9]) binary-code closure;
- unique self-orthogonal ([21,9,8]) code;
- (PG(2,4)) line symmetric differences;
- Golay / octad / Petersen conflict-graph geometry;
- Delsarte upper bound 105 for the historical parameter set.
Core theorem package
| File | Purpose |
|---|---|
paper/FINAL_BREAKTHROUGH_Exact84_Rigidity_Stability.md |
Main strengthened manuscript |
verification/hsin84_final_verify.py |
Standalone deterministic verifier |
verification/core38_final_data.json |
38 direct maximal doubly-even ([21,9]) representatives and certificates |
verification/hsin84_final_certificate.json |
Frozen theorem and witness certificate |
verification/core38_invariants.csv |
Human-readable 38-core invariant table |
prior_art/Prior_Art_Audit_Exact84_21_13_7_9.md |
Detailed prior-art / novelty audit |
prior_art/Prior_Art_Source_Matrix_Exact84.csv |
Structured source matrix |
AGENTS.md |
Agent-readable orientation and epistemic rules |
CLAIMS.json |
Machine-readable claim inventory |
RELEASE.json |
Release metadata and reproduction entry point |
SHA256SUMS.txt |
Integrity hashes |
release/Hsin_Heart_FINAL_Breakthrough_Exact84_Rigidity_Stability.zip |
Frozen all-in-one archive |
Reproduce
Requires Python 3 only; the verifier uses the standard library.
python verification/hsin84_final_verify.py \
verification/core38_final_data.json \
verification/hsin84_final_certificate.json
Expected headline output:
PASS
38 direct maximal doubly-even [21,9] core representatives verified
Non-Golay sharp upper: 81
Golay/PG(2,4) exact upper: 84
Extremizers per fixed PG(2,4) core: 476837158203125
84-witness size: 84
81 non-Golay witness size: 81
Main proof architecture
- Complement the 13-sets to weight-8 binary incidence vectors.
- Pairwise intersections (2,4) imply pairwise binary orthogonality.
- Their span is therefore a doubly-even binary code.
- The even-weight quadratic space in length 21 has Witt index 9, forcing dimension at most 9.
- Every candidate family lies in a maximal doubly-even ([21,9]) core.
- A published classification gives exactly 38 permutation classes of such cores.
- The release verifies upper certificates over one representative of every class.
- Every non-(PG(2,4)) core has maximum at most 81.
- The (PG(2,4)) core decomposes its conflict graph into 21 disjoint Petersen graphs, giving (21\cdot4=84).
- Equality in every Petersen component forces a star, producing the complete extremizer classification.
New-claim boundary
The release does not claim discovery of the classical ingredients. In particular, the following are prior art and are cited in the manuscript/audit:
- the historical 105 Delsarte bound;
- Gosselin's earlier 105-block incidence formulation;
- Calderbank's quasi-symmetric-design obstruction;
- doubly-even / self-orthogonal binary-code theory;
- classification of doubly-even ([21,9]) codes;
- the unique self-orthogonal ([21,9,8]) code;
- its (PG(2,4)) realization and Golay-related geometry.
The candidate new conclusions are the exact value 84, the sharp 81/82 rigidity threshold, the problem-specific 21-Petersen decomposition as used here, and the complete selector classification of extremizers.
Citation
A machine-readable CITATION.cff is included. Until external review establishes final bibliographic status, cite this repository by its Hugging Face URL and release commit/revision.
Licensing
This repository uses a mixed license:
- research prose and tables: CC BY 4.0;
- verification/source code: MIT;
- third-party mathematical results remain subject to their original publications and are cited rather than relicensed.
See LICENSE_RESEARCH.txt and LICENSE_CODE.txt.
- Downloads last month
- 9