[Paper Review] Integrable fluctuations in the KPZ universality class

Daniel Remenik|arXiv (Cornell University)|May 3, 2022

Complex Systems and Time Series Analysis4 citations

TL;DR

This paper establishes that the KPZ fixed point, the universal scaling limit of one-dimensional growth models in the KPZ universality class, is a stochastic integrable system whose finite-dimensional distributions satisfy the Kadomtsev-Petviashvili (KP) equation. The key result is that the generating function for the narrow wedge initial condition solution of the KPZ equation satisfies the KP equation, linking universal fluctuations to integrable PDEs via exact solvability and Fredholm determinants.

ABSTRACT

The KPZ fixed point is a scaling invariant Markov process which arises as the universal scaling limit of a broad class of models of random interface growth in one dimension, the one-dimensional KPZ universality class. In this survey we review the construction of the KPZ fixed point and some of the history that led to it, in particular through the exact solution of the totally asymmetric simple exclusion process, a special solvable model in the class. We also explain how the construction reveals the KPZ fixed point as a stochastic integrable system, and how from this it follows that its finite dimensional distributions satisfy a classical integrable dispersive PDE, the Kadomtsev-Petviashvili (KP) equation.

Motivation & Objective

• To establish the KPZ fixed point as a stochastic integrable system with exact solvability.
• To demonstrate that finite-dimensional distributions of the KPZ fixed point satisfy the Kadomtsev-Petviashvili (KP) equation.
• To connect the universal scaling limits of KPZ models to integrable PDEs via exact formulas derived from solvable models like TASEP and the stochastic heat equation.
• To show that the generating function for the narrow wedge initial condition of the KPZ equation solves the KP equation, revealing deep integrability.
• To extend this integrability to other initial data types, including flat, Brownian, and stationary initial conditions.

Proposed method

• The construction of the KPZ fixed point via the exact solution of the totally asymmetric simple exclusion process (TASEP), a solvable model in the KPZ class.
• Use of Fredholm determinant formulas for the KPZ generating function, derived from the stochastic heat equation with multiplicative noise.
• Application of the Brownian scattering operator and its spectral properties to derive differential equations for the generating function.
• Transformation of the resulting equations into the KP equation through self-similar solutions and Miura-type transformations.
• Use of Painlevé II transcendents and Tracy–Widom distributions to analyze tail behavior and connect to random matrix theory.
• Derivation of the KP equation for special solutions of the KPZ equation by verifying that the second logarithmic derivative of the generating function satisfies the KP equation.

Experimental results

Research questions

• RQ1 Does the finite-dimensional distribution of the KPZ fixed point satisfy a classical integrable PDE?
• RQ2 Can the generating function for the narrow wedge initial condition of the KPZ equation be shown to solve the KP equation?
• RQ3 What is the role of exact solvability in linking universal fluctuations to integrable systems?
• RQ4 How do different initial conditions (narrow wedge, flat, Brownian) affect the integrability structure of the KPZ fixed point?
• RQ5 Is the KP equation structure preserved beyond the narrow wedge case, and for which other initial data does it hold?

Key findings

• The finite-dimensional distributions of the KPZ fixed point satisfy the Kadomtsev-Petviashvili (KP) equation, establishing a direct link between universal fluctuations and integrable PDEs.
• The generating function for the narrow wedge initial condition of the KPZ equation solves the KP equation, as shown through the second logarithmic derivative of the Fredholm determinant.
• The solution for the narrow wedge case is connected to the Tracy–Widom GUE distribution via the Painlevé II equation and the Hastings-McLeod solution.
• For the flat initial condition, the generating function satisfies the KP equation and is related to the Tracy–Widom GOE distribution through a different Painlevé II solution.
• The integrability structure extends to other initial data, including half-Brownian, two-sided Brownian, and stationary initial conditions, via similar Fredholm determinant and differential equation techniques.
• The KPZ fixed point is identified as a stochastic integrable system, with its dynamics governed by a hierarchy of integrable dispersive PDEs, including the KP equation.