Sharp dependence inequalities have two complementary aspects: the largest possible factorization defect and the cost of imposing a mixing or a multilinear structure. Starting from the universal characteristic-function bound \(C_d\), we identify sharp covariance constants \(4K_G(k)\) for real Hilbert-space-valued observables, including \(4√2\) for complex scalars; we establish the small-mixing slope for several factors and quantify restricted-phase and centred-interaction variants. For \(n\) \(σ\)-fields, the sharp real and complex multilinear factors are \(2^n\) and \(2^nκ_n\), respectively, where \(κ_n\) is the inverse planar GHZ local-model visibility. A square-wave/Möbius construction gives the matching formula \(κ_n=(π/2)^nζ(n)/(2(1+2⁻ⁿ)ζ(2n))\) for every \(n≥4\); in particular \(κ_4=105/34\). For three parties a finite-harmonic Bell functional built from the divisors of \(105\) and the primes \(11,13,17,19,23,29,31\) yields the strengthened bound \(κ_3≥2.0349681535…\), whereas \(κ_3≤105ζ(3)/(2π^3)\) and exact equality remains open. The three-party lower bound is computer-assisted: the analytic reduction is completed by 272 explicitly specified inequalities verified using exact rational arithmetic and directed bounds, documented in a reproducible proof supplement. We state precisely where this exceptional finite verification enters the argument and separate it from the independent results on mixing and interaction measures.
No takes yet. Share an insight, caveat, or question.
Alexandr Martinevski (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: