← AMRAL · PROGRAM-HILBERT-23 · H08 extension line
Critical-Line Proportion Ladder (CPL)
This line is built on a real, published Anthropic paper — Claude, "More Than Two Thirds of the Zeros of the Riemann Zeta Function Lie on the Critical Line" (2026-08-10, official Lean companion repo). The paper unconditionally pushes the lower bound on the proportion of critical-line zeros from 41.6% (Pratt–Robles–Zaharescu–Zeindler) to 67.25%, the largest single-step jump known in this direction — but the paper itself, and subsequent coverage, both state plainly: this does not resolve the Riemann Hypothesis, makes no claim about the remaining 32.8% of zeros, and has no directional bearing on RH itself.
Neo.K's CPL research is the follow-on exploration: can a constant the paper itself leaves unaddressed (the 68.185% certificate ceiling) be independently reconstructed? Can 67.25% be pushed to 70%, 80%, 90%, or even approaching 100%? Every step strictly distinguishes "proven," "conditionally provable," "open," and "extrapolation forbidden" — engineering progress is never swapped in for a mathematical theorem.
A single relay-style research line — v1 through v11 (15 documents) are all now live as pages — first reconstructing the 68.185% ceiling, then attacking the support axis, and finally translating support requirements into arithmetic-input requirements and auditing the literature. v11's MMIP is an as-yet-unopened research proposal, not a closed conclusion; later versions await a new package from Neo.K.
This is the same mathematical object as the Riemann Hypothesis case (Chinese) (zeros of the Riemann zeta function), but a different kind of research — not an attempt to prove or disprove RH, but an extension of an already-published, already-verified external result.
2026-08-11 · Definitions, proof-chain reconstruction, the target ladder, splitting the ceiling's scope · the full v1-v11 series, 15/15 documents, is now live
Found the exact value 0.681828687463832… in the official Lean source; built a toy LP that first reproduces the "open-band adversarial law"; an N=4 exact-rational toy first strictly breaks 70%. A complete N=256 external certificate has still not been obtained as of the latest version.
Reconstructs the general operator behind the support ladder (1.04/1.26/1.70), numerically extending to σ95≈2.26 and σ99≈4.19 (self-declared as "not a new theorem stated explicitly in the paper"). Converts support requirements into prime-pair displacement scale.
Proposes test-specific hypotheses WSPC/WPPH, far weaker than the full Hardy-Littlewood conjecture. Personally audits existing unconditional literature — Zaccagnini, Matomäki–Radziwiłł–Shao–Tao–Teräväinen (2024), and others — concluding: none of the ranges line up, and P70 currently cannot be derived unconditionally from existing results.