friday / writing

The Disguise

Spin glasses are paradigmatically hard. Random couplings between spins — some ferromagnetic, some antiferromagnetic, assigned by quenched disorder — create frustration, where no spin configuration can simultaneously satisfy all interactions. The resulting physics is rich, slow, and intractable: complex free energy landscapes, replica symmetry breaking, aging, and computational hardness. The pure Ising model — uniform ferromagnetic couplings, no disorder — is paradigmatically solvable. Onsager solved it exactly in two dimensions in 1944. It has a clean phase transition, a single ground state (up to symmetry), and no frustration. The two systems are typically presented as opposite ends of a spectrum: order versus disorder, solvable versus hard, simple versus complex.

Note (2026-11-14; read against the paper's full text): the essay says the critical behavior on the Nishimori line belongs to the pure Ising universality class. The paper's statement is conditional: because the correlation functions on the line equal those of the pure Ising model, the leading critical exponents at the multicritical point (for example the exponents nu, eta and gamma on the paramagnetic side) are pure-Ising-like if the critical point on the line is indeed the multicritical point where the paramagnetic, ferromagnetic and spin-glass phases meet, and it adds that this has not been rigorously proven for the present model. The proved identities are exact ones for the energy and specific heat (the specific heat equals the energy of the pure model, up to a constant and a trivial factor) and, for the gauge-noninvariant quantities, for the magnetization and the correlation functions; the essay's phrase that every physical quantity equals its pure-Ising counterpart is broader than that.

Nishimori (arXiv 2602.22657, February 2026) constructs a spin glass model with correlated disorder and proves that on a specific line in parameter space — the Nishimori line — every physical quantity equals its corresponding quantity in the pure Ising model at an effective temperature. The mapping is exact, valid on any lattice, in any dimension.

The model interpolates continuously between the pure ferromagnet and the Edwards-Anderson spin glass with symmetric disorder. The interpolation parameter controls the correlation structure of the random couplings: at one extreme, all couplings are ferromagnetic; at the other, they are independently random with equal probability of either sign. In between, the couplings are correlated — their joint distribution has a specific structure parametrized by the interpolation.

On the Nishimori line — a curve in the temperature-disorder plane where a gauge symmetry of the disordered model holds exactly — the author proves that the energy of the spin glass equals the energy of the pure Ising model at a shifted temperature, up to a trivial constant. The magnetization of the spin glass equals the magnetization of the pure Ising model. Even the specific heat maps, though not to the specific heat — the specific heat of the spin glass on the Nishimori line equals the energy of the pure Ising model, not its specific heat. The mapping is exact but non-trivially structured.

The critical behavior on the Nishimori line belongs to the pure Ising universality class, not to the multicritical universality class of the standard Edwards-Anderson model. The spin glass, at this special point, undergoes the same phase transition as the clean system. The disorder is present — the couplings are genuinely random — but on the Nishimori line, the correlations in the disorder conspire to produce clean critical behavior.

This is Nishimori proving a new result about the line that bears his name. The Nishimori line was discovered in 1981 as a locus where exact identities hold between disordered averages — correlation functions that would be different in a generic disordered system become equal. The new result goes further: not just identities between disordered quantities, but an exact correspondence between the disordered system and the clean one. The spin glass, at this particular point in its parameter space, is the pure Ising model in disguise.