Correlation function of the cubic–quintic scheme
DefinitionGrothendieckLB_Supplement_cqCorrgaussian-analysisgrothendieck-constantkrivine-roundingp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let and be standard Gaussian vectors in such that the two coordinate pairs are independent, and . The correlation function of the cubic–quintic limiting scheme , , , is
Formalization Note Defined for all real parameters ; the paper's scheme takes the values (10).
Definition code
import Mathlib
import Definitions.Def_GrothendieckLB_Supplement_multiCorr
import Definitions.Def_GrothendieckLB_Supplement_cqRhoPoly
import Definitions.Def_GrothendieckLB_Supplement_cqF
import Definitions.Def_GrothendieckLB_Supplement_cqG
namespace GrothendieckLB
namespace Supplement
/-- The correlation function `H` of the cubic–quintic limiting scheme
`S = (f, g, ρ₁, ρ₂)`, `ρ₁ = ρ`, `ρ₂(t) = t` (Saha et al., *New Lower and Upper Bounds for the Grothendieck Constant*, arXiv:2608.11158 (held version, sha256 e7eb5d65…), Definition 4.1 and (13), p.24):
`H(t) = (π/2) E[f(W, X) g(W′, X′)]`, where `(W, X)` and `(W′, X′)` are standard Gaussian vectors,
the two coordinate pairs are independent, `corr(W, W′) = ρ(t)` and `corr(X, X′) = t`.
**Formalization Note** The two coordinates of `ℝ²` are `z 0 = w` and `z 1 = x`; the
correlation vector is `(ρ(t), t)`. This is `multiCorr` of the pair `(f, g)` of (12), i.e. the
formula of `LimitingScheme.corr` for this scheme. The parameters are free reals; the paper's
are (10). -/
noncomputable def cqCorr (η s₃ s₅ : ℝ) (t : ℝ) : ℝ :=
multiCorr (fun z : Fin 2 → ℝ => cqF η s₃ s₅ (z 0) (z 1))
(fun z : Fin 2 → ℝ => cqG η s₃ s₅ (z 0) (z 1)) ![(cqRhoPoly s₃ s₅).eval t, t]
end Supplement
end GrothendieckLB
Source
Saha et al., New Lower and Upper Bounds for the Grothendieck Constant, arXiv:2608.11158 (held version, sha256 e7eb5d65…), Definition 4.1, (13), p.24