OctoInformBeta

Conformal universality for weakly interacting and random-bond Ising models

The Lean library studies how prescribing spins on one part of a finite ferromagnetic Ising graph changes the full conditional spin law on another part in the presence of common pinned spins. Let G=(V,E)G=(V,E) be a finite graph with nonnegative couplings KeK_e and finite real fields hvh_v, and let μ\mu be its Ising Gibbs measure. Let I,J,F⊆VI,J,F\subseteq V be pairwise disjoint sets, with I,JI,J nonempty, and let τ∈{−1,+1}F\tau\in\{-1,+1\}^F prescribe the common spins on FF. Let σ\sigma denote the spin configuration on VV. For a probability measure λ\lambda on {−1,+1}J\{-1,+1\}^J, define μλ(a)=∑bλ(b)μ(σI=a∣σJ=b,σF=τ)\mu_\lambda(a)=\sum_b\lambda(b)\mu(\sigma_I=a\mid\sigma_J=b,\sigma_F=\tau) for a∈{−1,+1}Ia\in\{-1,+1\}^I. Let G0G_0 be the graph obtained by deleting FF and its incident edges and separately contracting II and JJ, adding parallel couplings. Let q0q_0 be the probability that the two terminals are connected in the zero-field FK model on G0G_0 with cluster weight 22 and edge parameters pe=1−e−2Kep_e=1-e^{-2K_e}.

Duminil-Copin, Hongler, and Nolin proved polynomial relative decorrelation for separated FK events. Ding, Song, and Sun proved that spin covariance in arbitrary fields is bounded above by the corresponding zero-field two-point function.

The preprint proves e−4arctanh⁡q0≤μλ(a)/μλ′(a)≤e4arctanh⁡q0e^{-4\operatorname{arctanh}q_0}\leq\mu_\lambda(a)/\mu_{\lambda'}(a)\leq e^{4\operatorname{arctanh}q_0} for every configuration aa on II and every two probability measures λ,λ′\lambda,\lambda' on the source prescriptions. At the critical nearest-neighbor coupling on the square lattice, the preprint derives uniform pointwise comparison across a separating band of fixed positive relative width; this geometric consequence is not contained in the Lean library.

Lean has checked every result of the Lean library, so the Lean library is the reference for the statements on these pages. Each result is stated as in the Lean library and linked to its declaration, with any differing statement of the preprint given beside it. The Lean library proves the arbitrary-field covariance inequality used in the comparison. The preprint is a guide only to the order and names of the results.