Bounds for the periodized cutoff covariance #
The torus covariance is written in DetParametrix.Kernels as a lattice
periodization of the compactly supported Euclidean covariance. This file
proves that, for 0 < ε ≤ 1, only a fixed finite box of period vectors can
contribute. It then turns the Euclidean uniform bound into the paper-scale
O(ε⁻⁴) torus bound without any summability hypothesis.
A fixed integer half-width containing every period vector which can
meet the support of η_ε, uniformly for 0 < ε ≤ 1.
Instances For
The corresponding finite box in ℤ⁴.
Equations
- ρ.covariancePeriodBox = Fintype.piFinset fun (x : Fin Anderson4D.dim) => Finset.Icc (-ρ.covariancePeriodRadius) ρ.covariancePeriodRadius
Instances For
One summand in the lattice periodization defining etaEpsT4.
Equations
- ρ.etaPeriodTerm ε z k = ε⁻¹ ^ Anderson4D.dim * ρ.eta fun (i : Fin Anderson4D.dim) => ε⁻¹ * (Anderson4D.torusLift z i + 2 * Real.pi * ↑(k i))
Instances For
Passing from a Euclidean representative to the product quotient does not increase the fixed sup norm.
Every period representative of a torus point dominates its quotient norm.
A nonzero periodization term belongs to the fixed period box.
The lattice periodization is an actual finite sum on the fixed box.
Every periodization term is nonnegative at positive scale.
Positivity of the torus covariance. The fourth rescaling power is nonnegative even outside the theorem's positive-scale regime.
Squared support indicator in exactly the form used by Proposition 4.1.
The factor 4 combines the product quotient norm and
torusDistSq ≤ 4‖z‖².
A Euclidean bound η ≤ C yields the paper-scale torus bound, with
the finite period-box cardinality kept explicit.