KPZ universality through ASEP

KPZ universality is the discovery that a vast collection of seemingly unrelated random growth and transport models — growing interfaces, the asymmetric exclusion process, last-passage percolation, polymers in random media, even certain traffic and bacterial-colony models — share the same large-scale statistics, governed not by the Gaussian distribution but by the Tracy-Widom laws of random matrix theory. The asymmetric simple exclusion process (ASEP and its totally asymmetric case TASEP) is the cleanest exactly-analysable window onto this class: it is the interacting particle system through which the KPZ scaling exponents and limiting distributions were first computed rigorously. The name is for Kardar, Parisi and Zhang, who in 1986 wrote down the conjectural continuum equation for such growth.

The KPZ universality class is characterised by its scaling exponents. For a one-dimensional growing interface of height h(t, x), fluctuations grow as t^(1/3) (not the t^(1/2) of a Gaussian / Edwards-Wilkinson interface), spatial correlations extend over a length t^(2/3), and the rescaled height fluctuation ( h(t, x) - v t ) / (c t^(1/3)) converges to a Tracy-Widom distribution — specifically GUE Tracy-Widom for curved (droplet) initial data and GOE Tracy-Widom for flat initial data, with the geometry of the initial condition selecting the limit law. Mapped to ASEP, the interface height is the integrated current of particles: the number that have crossed the origin by time t fluctuates on scale t^(1/3) with a Tracy-Widom law, and the spatial structure is the Airy process. The conjectural continuum limit is the KPZ equation partial_t h = (1/2) partial_x^2 h + (1/2)(partial_x h)^2 + white noise, whose solution theory required the breakthroughs of regularity structures (Hairer) and paracontrolled calculus to make rigorous, the nonlinear (partial_x h)^2 term being what drives the system out of the Gaussian Edwards-Wilkinson class. For ASEP/TASEP the exact t^(1/3) and Tracy-Widom results were obtained through exactly-solvable structure — the Bethe ansatz, determinantal / Pfaffian formulas, the RSK correspondence linking TASEP to random matrices and the Tracy-Widom distribution — and are theorems, not just conjectures.

This matters because it is one of the great unifying discoveries of modern probability: a single universality class, with Tracy-Widom (not Gaussian) statistics, governing one-dimensional growth and driven transport across physics, and ASEP is its mathematical anchor. Several honesty points. The full universality — that all these models share the limit — is rigorously established only for special, exactly-solvable representatives (TASEP, certain last-passage models, the KPZ equation itself, the KPZ fixed point as the universal scaling limit); for generic models in the class it remains conjectural, the central open problem of the field. The dimension matters: the clean t^(1/3) / Tracy-Widom theory is a one-dimensional (1+1) phenomenon; higher dimensions are far less understood and may not even share a single exponent. And the appearance of Tracy-Widom — born in random matrix theory as the law of the largest eigenvalue of a GUE matrix — in a particle-hopping model is a genuinely deep coincidence, mediated by the combinatorics of the RSK correspondence, not a superficial analogy.