Target cell: weight-8 × unrestricted. The board's weight-8 cell at n = 144 was thin: the incumbent at this (n, k) was [[144,16,8]], so a same-size code one point higher in distance was a concrete frontier opening. The source is the published census of Lu, Yang and Guo (arXiv:2609.06572), which develops the structure theory of BB-type codes with weight-4 generator polynomials (weight-8 checks) and reports this code with an exactly verified distance of 10. This entry is a faithful reproduction of their Table 1, entry 1, not a new search: the parameters exist in the literature, and the contribution here is an independently rebuilt, witness-backed board entry.
No search. The generator polynomials were transcribed from the paper's Table 1 and rebuilt with the repo kit: QC(A, B) on Z_12 x Z_6 with
(exponents written x^a y^b). The rebuild reproduced the paper's parameters exactly before packaging: n = 144, k = 16 (recomputed by the verifier's own rank arithmetic, not trusted from the paper), max check weight 8.
verifier, cross-validated against Magma's Words enumeration (their Remark 4.4), and their census reproduces the BB benchmark d = 12 as a calibration check.
sides, so the staged claim is a witness-backed upper bound d <= 10. The validation gate (verify/validate_candidate.py) ran an 8000-trial fresh-seed RIS refutation and found no lighter logical; the gate reports the code as board-advancing in the weight-8 x unrestricted cell.
upper_bound, d <= 10. The paper's exhaustive result saysthe true distance is exactly 10; per board policy that d= tier belongs to the server certifier, not to this submission.
Not applicable to a reproduction. One adjacent finding from evaluating the paper's full Table 1 against the board: most of its weight-8 members ([[72,14,8]], [[144,10,12]], [[144,14,10]], [[144,20,8]], [[144,18,8]], [[144,8,12]]) are dominated by existing weight-8-cell entries here — e.g. [[64,18,8]] and the weight-6 BB benchmark [[144,12,12]] — so only this code and the paper's [[144,6,d>=15]] advance this board.
Human-authored source paper (Lu, Yang, Guo, arXiv:2609.06572). Reproduction harness: this repository's kit — research/kit/bb.py (build_bb), research/kit/css.py (k recomputation), research/kit/submit.py and the CLI ./qldpc submit (witness search, verification), verify/validate_candidate.py (gate). No AI model produced the code; provenance.model is human.
Rebuild (H_X, H_Z) with the kit on the torus Z_12 x Z_6 (n = 2*12*6 = 144):
from bb import build_bb
HX, HZ = build_bb(12, 6,
A_terms=[(0,0),(11,0),(0,5),(9,5)],
B_terms=[(3,2),(8,2),(4,3),(9,3)])
Then ./qldpc submit <npz> --authors @MathysRennela --family bivariate-bicycle. The construction string in the JSON carries the same polynomials. Source: arXiv:2609.06572 (https://arxiv.org/abs/2609.06572), Table 1, entry 1.