Skip to content

Hilbert 8 / Goldbach — Chebyshev ψ Spectral Cascade (Redirect 2 of 3)

Parent: hilbert_08_goldbach_conjecture/REPORT.md Status: cascade dispatched 2026-05-23 Class cascade: A ∘ J ∘ L ∘ K Source: srmech catalog hilbert_08_goldbach_chebyshev_psi (100 rows)

Cascade design

Compute ψ(N) = Σ_{p^k ≤ N} log(p) for N ∈ {20, 40, …, 2000}. Residual = ψ(N) − N (the prime-number-theorem error term, RH-bounded by O(√N · log² N)). Class L applied to residual time-series path graph weighted by |Δresidual|.

Findings (2026-05-23)

Stat Value
Sample range N ∈ [20, 2000]
Primes ≤ 2000 303
ψ(2000) 1994.45
Residual at N_max −5.55
rel_residual at N_max (in units of √N) −0.1241
Mean |rel_residual| 0.1853
RH bound O(log² N) at N=2000 ~57.8
Margin to RH bound ~466× (rel_residual ≪ bound)
Class L Fiedler 0.0034
Class L spectral radius 67.68
Spectral gap 5e-5

Structural observation: ψ(N) − N is well within the RH-predicted O(√N · log² N) bound at N=2000 — the cascade confirms the residual is bounded in the predicted way. Per Class K asymptotic-DoF reading: the residual is the line-projection of the prime-distribution loop; its bounded oscillation is exactly what the framework predicts for a substrate-asymptotic-wave whose loop is asymptotically projected to a line.

The high spectral-gap-ratio (max/fiedler ≈ 20000) of the residual time-series path Laplacian indicates the residual's variation is concentrated at specific scales (high-frequency oscillation around the asymptotic), consistent with the explicit-formula picture where ψ(N) − N is a sum over zeta zeros.

Run

python docs/unsolved-maths/hilbert/hilbert_08_goldbach_chebyshev_psi/generate_catalog.py