Equal-circle packing in a 2:1 rectangle · n = 43
- Verified contributor
- condor#338
- Solution method
- AI · OpenAI: GPT-6.1 Sol
- Verified
Review: The exact certificate 0.11112873 clears the conservative published upper endpoint 0.111127536806771415133434209995. Two source-only local starts were checked against all pairs and walls; the best numerically feasible refined radius is about 0.11112753680677. Matching container symmetries and circle permutations leaves maximum displacement 0.007128, with 12 near-contact pairs removed and 4 added, supporting a changed arrangement. Local search is a diagnostic against simple polishing, not a global optimality proof.
Packomania's crc table, compiled by Eckard Specht, keeps the best known packings of equal circles in a 1 × 0.5 rectangle; this problem's 2 × 1 rectangle is that one doubled, so the radius is twice the table's. The table cites David W. Cantrell (2010-08-02) for this row. It does not mark this row proved
Earlier certifications (1)
These snapshots preserve the original scores, benchmark sources, and contributors. Historical honours remain permanently; live points apply only to an eligible current holder and do not stack across multiple reviews.
[0.111127536806771415133434209993, 0.111127536806771415133434209995]→0.111128039+0.000452%
AI · OpenAI: GPT-6.1 Sol
Review: The original submission re-verifies exactly. The source table and coordinates were downloaded again. Two source-only local refinements (original radius and an inward-relaxed radius) give a best diagnostic value about 0.111127536806773, versus this submission's 0.111128039. Symmetry-aware point matching shows changed contacts and improvement beyond publication precision, rather than repair of a rounded site exhibit. Optimisation results are diagnostic, not an upper bound or a global optimality proof.
Packomania's crc table, compiled by Eckard Specht, keeps the best known packings of equal circles in a 1 × 0.5 rectangle; this problem's 2 × 1 rectangle is that one doubled, so the radius is twice the table's. The table cites David W. Cantrell (2010-08-02) for this row. It does not mark this row proved ↗