Exécution de notre outil de synthèse sur un nouvel article
Vérifiez l’article ci-dessous, puis lancez sa synthèse.
Partial Results on Four Open Problems in Discrete Geometry, Learning Theory, and Graph Theory
Zheng Yang (Chinese University of Hong Kong)
Résumé : This deposit contains partial research results on four open problems selected from recent workshop problem lists and conference open-problem sessions (AimPL, Oberwolfach Reports, COLT): 1. Great-circle cell expectations (OWR 2024 Problem 8, Xavier Goaoc). We prove the exact value M(8) = 113/29 by exhaustive enumeration over all 3,315 realizable 8-point order types, and establish the lower bound M(n) ≥ 8 − O(1/n) from regular near-pencil arrangements. The asymptotic upper bound for general arrangements remains open. 2. Crossing-number edge-addition constant (AimPL Albertson Problem 1.1, Jacob Fox). We prove c_opt ≥ 2/9, improving the classical 1/8 lower bound, via path-stacked triangulations with a matching upper bound for that family. Whether any universal c < 1 exists remains open. 3. Closed-simplex weighted data selection for d=2, n=3 (COLT 2025 Open Problem, Hanneke/Moran/Shlimovich/Yehudayoff). We give a continuous one-parameter family of planar datasets whose closed-simplex weighted-selection ratio is exactly 3/2. The universal worst-case value for d=2, n=3 and the general d < n < 2d regime remain open. 4. Optimal exponent σ(B_2) for the bipartite B_2 incidence-graph Turán problem (OWR 2024 Problem 9, Ji Zeng). We prove σ(B_2) ≥ 18/7 ≈ 2.571, the first explicit lower bound above the trivial 2. The exact value of σ(B_2) and the matching upper bound remain open. Each direction includes theorem statements and proof notes, reproduction scripts, data certificates, adversarial review reports, novelty audits, and a public note. Some oversized raw enumeration outputs and heuristic search probes were omitted because they are not required for the stated theorems; see the README and directions/OP-2026-0037/data/README.md for details. Original authors have not yet been contacted. These are preliminary research records, not published papers.