SURFACE / ORDERREFINED ENVELOPE

SIXTH-POWER-FREE RADICANDS · A PINNED PREPRINT INPUT

The physical family matters.
The signed target remains.

Apply de Faveri’s fixed-order large-sieve theorem to the actual sixth-power-free radicands, with every original zero retained. At H = D3/4, the global unsigned power drops from 7/6 to 9/8. The A₆ = 1 signed slice reaches 13/12. The fixed-profile target 3/4 remains open.

These are analytic upper-bound powers, with constants and (HD)ε suppressed. They are not simulated row energies or a proof of GRH.

H = D3/4 · 0 ≤ a ≤ 1/4

Keep the whole face in view.

A₁ = A₂ = Da
A₃ = D1/4 − a
A₄ = A₅ = A₆ = 1

Historical and current powers across the entire former saturation faceOld power 1.25. Historical cube-only refinement min(1.25,1+2a). Current ordinary-conductor refinement1 throughout. Target0.75.
The entire former face is retained. The new ordinary term D + R² has leading power 1 everywhere on this slice. The earlier cube-only refinement remains valid as a historical comparison; it no longer describes the best available bound.
  • Old BGL · 5/4
  • Earlier cube-only minimum
  • Current · ordinary conductor
  • Sixth-free comparison · 13/12
  • Target H · 3/4

The monomials, evaluated

Approximate algebraic values at the selected D. These sums omit unknown theorem constants and (HD)ε; they are not numerical bounds for Λ.

TermPower of DValue ≈
Old monomial sum
Quadratic monomial sum
Earlier cube/BGL minimum
Ordinary D + R²
Sixth-free F₆(H,D)
Current minimum of all four

At finite D, sums include lower-order terms. Their numerical crossover need not coincide with the asymptotic power crossover.

Geometry and capacity

Aⱼ are dyadic norm scales, not six sampled ideals. Capacity is checked for these scales; fixed annular constants are suppressed. Globally r₆ is arbitrary and may overlap other factors. A₆ = 1 is a choice of this slice, not a global squarefree or coprimality restriction.

A₆ = 1 · CURRENT SLICE POWER 13/12

The remaining two-pivot slice.

Set A₁ = Dx₁, A₂ = Dx₂, and A₃ = ⋯ = A₆ = 1. The capacity triangle is x₁ + 2x₂ ≤ 3/4. Color shows the minimum of the old BGL, ordinary and physical sixth-free upper-bound powers.

Click inside the triangle, or focus the plot and use arrow keys. Buttons select exact witnesses. Teal: power 1; amber: power 13/12. These are upper-envelope witnesses, not arithmetic energy lower bounds.
1 · stronger bound13/12 · remaining slice

Here r = x₁+x₂ and m = max(x₁,x₂). Old power = max(r,1+r−m,2/3+r−m/3); ordinary power = max(1,2r). The sixth-free comparison uses h = x₁ + 2x₂ and F₆ powers h, 1, 5h/6 + 1/3, h/3 + 5/6. At each selected capacity-boundary witness it gives 13/12, improving the previous 7/6. Its gap above target is 1/3.

1 ≤ H ≤ D · H = Dθ

Global factor counting and the sixth-free slice.

Freezing the arbitrary, possibly overlapping r₆ preserves its actual collision mask. Summing its sixth-power capacity cost produces the global DH1/6 term. A₆ = 1 avoids that factor count.

Upper-bound powers, with constants and arbitrary ε losses suppressed. These curves do not measure an operator norm or a row energy.
  • Previous global envelope
  • Sixth-free A₆ = 1
  • Current global envelope
  • Target H

For 1 ≤ H ≤ D the global expression is ≪ DH1/6. For fixed 0 < θ < 1, the power is sharp in the previously constructed class of profiles that vary with D. That lower result does not refute a bound for one fixed W or an appropriate bounded-seminorm class.

CURRENT EXTERNAL INPUT AND ACTUAL FAMILY

Retain the physical sixth-free norm.

De Faveri’s Theorem 1.1, arXiv:2610.04045v1, is a pinned submitted-preprint input. For sixth-power-free radicands and columns, its comparison is

F₆(H,D) = H + D + H5/6D1/3 + H1/3D5/6.

The book proves the application to literal primary generators in Q(ω), including lower character orders, common-prime zeros, ray factors and capacities. The row size is the norm of the physical radicand; replacing it by its radical would change the theorem’s input. Freezing r₆ gives the global expression

Φ(H,D) = H + DH1/6 + H5/6D1/3 + H1/3D5/6.

At H = D3/4, Φ has power 9/8, improving the previous 7/6 by 1/24. The A₆ = 1 mean-retaining signed common-divisor bound has power 13/12. These are applications of the external theorem, with independent reviews of the family match and assembly; they are not a new proof of that theorem or a new large sieve.

The full former cube face remains controlled by the older ordinary comparison at power 1. The historical curves are preserved above. The fixed-common-profile signed target, principal/short assembly and independent arithmetic determinant/zero-location bridge remain open. There is no new Lean verification or GRH conclusion.

Classical row norms apply to the original unweighted family or its stated bounded natural multiplicities. Arbitrary positive row weights require their actual weighted operator norm.