Skip to content

Counterfeit ladder: kernel-checked ζ_K (ℚ(√−5)) vs Epstein x²+5y² window separation + orbit-6 witnesses - #631

Draft
DrMurphyIsIn wants to merge 4 commits into
cl/crux3bfrom
cl/counterfeit-ladder
Draft

DrMurphyIsIn wants to merge 4 commits into
cl/crux3bfrom
cl/counterfeit-ladder

Conversation

@DrMurphyIsIn

Copy link
Copy Markdown
Owner

Stacked on #622 (Crux3), which is stacked on #615.

This PR adds the first kernel-checked window separation between an L-function that has an Euler product and a same-functional-equation counterfeit that doesn't.

theorem CF.window_separation_cE :
    0 < ∫ x, ‖v x‖^2 ∧ 3 * ∫‖v‖^2 ≤ Re W_{ζ_K}(v⋆ṽ) ∧ Re W_E(v⋆ṽ) ≤ -(3/20) * ∫‖v‖^2

The objects

  • Test function: one nonzero 9-mode test at window x = e^{18π/17} ≈ 27.8.
  • ζ_K: the Dedekind zeta of ℚ(√−5), with weights Λ(n)(1+χ₋₂₀(n)).
  • E: the Epstein zeta ½Σ′(x²+5y²)^{−s}, normalized so a(1) = 1. Its lattice counts are computed in the kernel.
    • E's −E′/E coefficient recursion is a hypothesis of the theorem. It is discharged by cE_conv, and eq_cE_of_conv proves the coefficients are unique.
  • Archimedean side: Γ_ℂ with conductor 20, with both pole terms kept.

Other results

  • Lemma O witnesses: c_D(6) = (1+κ²) log 6 ≠ 0, c_E(6) = 2 log 6 ≠ 0, and 6 is not a prime power. Each counterfeit has a "prime orbit" at a composite length.
  • Tamper controls: the kernel rejects all three.

Verification

  • Guard: CF_AxiomGuard has 54 lines, all standard. My rebuild ran 8788 jobs.
  • CI: the anchors are also appended to the required AxiomGuardRvMBridge.
  • Own-session skeptic: not refuted. It re-derived Γ_ℂ with conductor 20 and reproduced the −E′/E recursion, the zero-side cross-checks and E's off-line quadruple (−1.825).

Not formalized

  • The explicit formula linking these functionals to their zero sums. It's checked numerically, to about 4e-5.
  • ζ_K positivity on the whole 9-mode family. Only at-the-test positivity is proved; the span's minimum eigenvalue is about 1e-4.

Lane B numerics

These are in telperion/research/counterfeit_ladder_CF/laneB/.

  • Certified gap: full-class λ_min(ζ_K) ≥ 3.44e-3 at x = 20, where E is certified negative (x_E ≈ 19.82).
  • Independent confirmation: a peer-session skeptic confirmed this and proved the Loewner off-diagonal identity (PEER_SKEPTIC_VERDICT.md).
  • Lean cost: a full-class Lean theorem is about 10³–10⁴× current budgets unless restructured. That restructuring is the Step 2 lane.

conjecture1_proved = False.

🤖 Generated with Claude Code

https://claude.ai/code/session_01LMeoWeTz2Q3iSeLfqxfYo6

Dr. Murphy and others added 4 commits September 24, 2026 23:59
…paration at x = e^{18pi/17} (~27.8) + orbit-6 witnesses; lane B gate numerics

CF.window_separation_cE: on one nonzero 9-mode test, Re W_{zeta_K}(v*v~) >= 3||v||^2 and
Re W_E(v*v~) <= -(3/20)||v||^2, where E = (1/2) sum' (x^2+5y^2)^{-s} (a(1) = 1, lattice counts in the
kernel) with E's -E'/E recursion discharged by cE_conv (uniqueness via eq_cE_of_conv), zeta_K
weights Lambda(n)(1+chi_-20(n)). Gamma_C archimedean side with conductor 20, pole terms kept.
Lemma O witnesses: c_D(6) = (1+kappa^2) log 6 != 0, c_E(6) = 2 log 6 != 0, 6 not a prime power.
Tamper controls rejected. 54 guard lines, all standard; my rebuild 8788 jobs. Skeptic: not refuted.
Not formalized: the explicit formula linking these functionals to zero sums (numerically ~4e-5).
Lane B (numerics, telperion/research/counterfeit_ladder_CF/laneB): certified full-class
lambda_min(zeta_K) >= 3.44e-3 at x = 20 where E is certified negative (x_E ~ 19.82); independently
confirmed by a peer-session skeptic (Loewner identity proved there). conjecture1_proved = False.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LMeoWeTz2Q3iSeLfqxfYo6
…ge (append-only: 2 imports, 10 #print axioms)

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LMeoWeTz2Q3iSeLfqxfYo6
…Lemma O to #630 (peer skeptic nits, comment-only)

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LMeoWeTz2Q3iSeLfqxfYo6
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant