These sources are unregistered candidates. They are not part of the accepted root imports and do not establish a completed classification or the complete hyperplane theorem. The first hosted checks found ordinary elaboration errors; the reached errors are being repaired and checked again. New adapters remain explicit candidates until their full gates pass.
The separate Lean loneliness geometry candidates workflow checks independent scopes. It rebuilds every project dependency, checks each required endpoint, audits all owned theorem declarations including private declarations, and independently replays the exact dependency closure with stock leanchecker. Only propext, Classical.choice, and Quot.sound are permitted. Receipts record source, configuration, verification-code, and compiled-artifact hashes. Failed receipts are retained with their error and completed checks.
| Scope | Candidate statement | Remaining mathematical inputs or work |
|---|---|---|
prime-bounds |
Small denominators force a divisor of the speed product. Exact integer certificates select an avoiding prime for six through thirteen speeds. | The natural squared-norm bound is explicit. The six-speed list uses the first eight original primes; the other lists are unchanged. The combined arithmetic source exceeded its memory limit; the same facts are now split into rational-product and per-count numerical modules. |
basis |
A primitive integer vector extends to an integer basis. Minimizing a natural prefix-Gram potential supplies a basis fixing that vector and satisfying the adjacent squared-length ratio. The Gram product is the literal determinant. | The first-vector-preserving ratio basis is not asserted to satisfy the full Korkine–Zolotarev conditions. The complete basis component has passed source, axiom, and kernel checks; it remains unregistered. |
nearest-plane-value |
Rounding modulo a prefix gives a bound on the value of the containing rational subspace. Coordinate coalescence gives the squared prefix inequalities. | Lower-dimensional runner bounds remain explicit. Spectrum arguments need the additional plane-gap hypotheses specified in the written proofs. |
norm-geometry |
Construct the actual tail Gram squares from a primitive speed vector and derive a norm cutoff from a checked product bound. | Only strictly lower-dimensional runner theorems are used. Their proofs and the numerical product certificate remain explicit inputs. The complete norm-geometry closure has passed its source, axiom, and kernel checks. |
small-norm |
The actual prefix geometry and finite product tables imply natural squared-norm ceilings and avoiding-prime selection for six through eight speeds. The six-speed ceiling is 34,333,151. | Lower-dimensional runner theorems remain induction inputs. The six- through eight-speed adapters have passed their full checks. The nine- through thirteen-speed adapter is now checked by the separate large-norm extension task. |
cube |
A partial-bit decision tree certifies a full binary branch of the existing closed lifting checker. | The complete initial residue enumeration and generated ten-speed trees remain separate obligations. Only a four-assignment worked certificate is included in this candidate. |
cube-probe |
Three explicit modulo-127 roots have compact binary lifting certificates with 11, 39, and 20 leaves. | This is a numerical probe, not the full ten-speed enumeration. Its real-value exclusion retains the nine-speed runner bound. |
initial-cover |
A finite support search, state-reference composition, and multiplicity-aware coordinate permutation provide exhaustive modular normalization. | The generic checker and ten-speed wrappers passed their full checks at a4d614f. Full numerical support-coverage tables are separate work. |
six-support |
The generic normalized support cover has exact bit-mask and checked memo-table interfaces. A six-speed wrapper turns complete support certificates into the consecutive-direction classification. | The finite coverage certificates remain explicit inputs. The real-value conclusion retains the five-speed runner theorem. These sources await their first gate. |
six-support-probe |
A complete modulo-257 minimum-pair case with second coordinate 20 has 171 stored search nodes, 129 coverage masks, and 61 compatibility masks. | This is one complete minimum-pair case, not the complete prime or six-speed theorem. Its arithmetic and generic soundness are checked in a separate closure. |
small-sharp |
The exact six-spectrum arithmetic table implies the stronger conditional global norm cutoff 21,829, which also suffices for the displayed four- and five-parameter searches. | Basis, prefix, and determinant premises remain explicit. The generic rational-subspace covolume statement is separate. |
hyperplane |
Fourier character orthogonality, finite-frequency density control, an exceptional numerator count, an explicit denominator count from real admissibility, and the uniform probability rate for the literal ordered positive distinct box. Real and torus shift events are identified. | The complete uniform probability component has passed all three gates. The recursive subspace theorem, exact quasipolynomials, endpoint density, and independent-shift limiting-law assembly remain open. |
The deterministic hyperplane maximum and the existing finite arithmetic tables have separate earlier receipts. This document does not extend those receipts to the new candidates. The Haar-distribution candidate remains in its existing hyperplane-haar lane. Initial residue enumeration and ten-speed lifting trees remain under the separate finite-certificate work.
Completed component checks
The following unregistered closures have passed all three gates. Their scope does not include the other candidates in the table.
In run 44455928210, the nearest-plane value closure passed all 15 modules and its 234-theorem ownership audit. The small-sharp closure passed all eight modules and its 264-theorem audit. Both runs used source commit
488301d.In run 44457211754, the cube checker passed all 25 modules and its 576-theorem audit at source commit
779f3c5. The new numerical probe was not part of that run.The separate Haar lane passed all 11 modules and its 200-theorem audit in run 44457211742. This proves the Haar law of the predicted hyperplane maximum; the sampled velocity limiting-law assembly is separate.
In run 44459759714, the three numerical cube probes passed their full 26-module closure and 582-theorem audit at
3349777. The cube, nearest-plane, and small-sharp closures also passed again. The basis, prime-arithmetic, support-cover, and probability jobs failed at concrete source errors or the recorded arithmetic memory limit; they have no completed receipts from this run.In run 44459759796, the closed modular lifting engine passed all 30 project modules and its 603-theorem audit at
3349777. This includes the direction-leaf interface; the generated six- and higher-speed coverage tables remain separate.In run 44461710960, the basis component passed 16 modules and 285 audited theorems, prime arithmetic passed 21 modules and 463 theorems, and the uniform hyperplane probability component passed 21 modules and 383 theorems at
9a7cb8a.In run 44462446786, norm geometry passed 31 modules and 525 audited theorems, and the complete six- through eight-speed norm adapter passed 49 modules and 903 theorems at
de486df. Lower-dimensional runner bounds remain explicit inputs.In run 44464817263, initial coverage and the ten-speed orbit interpretation passed their 23-module closure and 603-theorem audit at
a4d614f. The memo-based six-speed probe failed at a separate source lemma and has no numerical verification receipt from this run.
To inspect a closure without running Lean:
python3 scripts/verify_loneliness_geometry.py basis --planTo run its source, axiom, and kernel gates:
python3 scripts/verify_loneliness_geometry.py basis