Unitary ensembles with a critical edge point, their multiplicative statistics and the Korteweg-de-Vries hierarchy

This paper demonstrates that the multiplicative statistics of unitary random matrices with a critical edge point, where the limiting density vanishes as a power of 5/2, are governed by the first three equations of the Korteweg-de Vries hierarchy, and it analyzes the asymptotic behavior of the corresponding solutions.

The Gist

Imagine you are looking at a crowd of people, but instead of people, they are invisible particles called "eigenvalues" that belong to a special type of random matrix. In the world of mathematics and physics, these particles don't just sit randomly; they have a specific way of arranging themselves, especially near the very edge of the crowd.

This paper is about what happens at a very specific, "critical" edge of this crowd. Usually, the density of these particles fades away gently, like a hill sloping down. But in this specific scenario, the crowd thins out much more dramatically—like a cliff that drops off steeply. The authors are studying the "multiplicative statistics" of this crowd. In plain English, this means they are asking: "If we randomly decide to keep or remove each particle based on a specific rule, what are the odds that the entire crowd disappears?"

Here is a breakdown of their journey and discoveries using everyday analogies:

1. The Setup: A Special Crowd and a Rule

Think of the particles as guests at a party. The "edge" of the party is where the music stops and the guests thin out.

• The Critical Edge: In most parties, the crowd fades out slowly. Here, the authors are looking at a "super-critical" edge where the crowd vanishes incredibly fast (mathematically, like a power of 5/2).
• The Rule (Thinning): They introduce a rule, represented by a function called $\sigma$. Imagine a bouncer who lets each guest stay with a certain probability and sends them home with the rest. The paper calculates the probability that no one is left at the party after this bouncer does their job.

2. The Discovery: The Crowd Follows a "Wave"

The most surprising finding is that the probability of the party emptying out isn't just a random number. It is governed by a famous set of mathematical rules known as the Korteweg-de Vries (KdV) hierarchy.

• The Analogy: Think of the KdV equations as the "laws of physics" for water waves. They describe how a wave moves, changes shape, and interacts with itself.
• The Connection: The authors proved that the probability of the party emptying out behaves exactly like a complex water wave. Specifically, the "shape" of this probability wave is dictated by the first three equations of the KdV hierarchy. It's as if the random arrangement of these invisible particles is secretly dancing to the same rhythm as ocean waves.

3. The Three Different "Weather Patterns"

The paper doesn't just stop at finding the wave; it studies how this wave behaves under three different "weather conditions" (mathematical regimes). They use a technique called the Riemann-Hilbert problem, which is like a sophisticated map-making tool that helps them navigate the complex landscape of these probabilities.

• Regime 1 (The Calm Morning): When the parameters are set one way, the probability wave looks very much like a specific, well-known solution to the wave equations. It's stable and predictable.
• Regime 2 (The Stormy Middle): When the parameters shift, the wave changes shape. It starts to look like a different, more complex type of wave (related to the "Painlevé II" hierarchy). This is like the water turning turbulent and forming a new kind of structure.
• Regime 3 (The Edge of the Cliff): When the parameters get very close to a critical limit, the wave behaves like a "Bessel function" (a type of wave often seen in circular ripples). Here, the probability of the party emptying out is determined by a specific, unique solution to a mathematical puzzle.

4. The "Magic" of the Math

The authors use a powerful tool called Riemann-Hilbert problems. You can think of this as a way of solving a jigsaw puzzle where the pieces are defined by how they "jump" or change when you cross a line. By solving this puzzle, they can translate the messy, random behavior of the particles into the clean, structured language of the KdV wave equations.

Summary

In simple terms, this paper shows that even in a system that looks completely random and chaotic (a crowd of random matrix particles at a critical edge), there is a hidden, beautiful order. The probability of this system disappearing follows the exact same mathematical laws that govern water waves. The authors mapped out exactly how this "probability wave" behaves in three different scenarios, proving that the universe of random matrices and the universe of water waves are speaking the same secret language.

Technical Summary

Technical Summary: Unitary ensembles with a critical edge point, their multiplicative statistics and the Korteweg-de-Vries hierarchy

Problem Statement
This paper investigates the multiplicative statistics associated with a specific limiting determinantal point process that describes the eigenvalues of unitary random matrix ensembles near a critical edge point. Unlike standard edge points where the eigenvalue density vanishes as a power of $1/2$ (governed by the Airy kernel), this critical edge is characterized by a density vanishing as a power of $5/2$. The relevant kernel for this regime, introduced by Claeys and Vanlessen, is of integrable type and is linked to the second member of the Painlevé I hierarchy, denoted $PI(2)$. The authors aim to characterize the multiplicative statistics of this process—defined as the probability of a thinned point process being empty—and determine their relationship to integrable partial differential equations, specifically the Korteweg-de-Vries (KdV) hierarchy.

Methodology
The analysis relies on the Riemann-Hilbert (RH) problem formulation, a standard technique in the theory of integrable systems and random matrices. The methodology proceeds through the following steps:

1. RH Characterization: The authors reformulate the multiplicative statistics $Q_\sigma$ as a Fredholm determinant of an integrable kernel. This determinant is linked to a specific $2 \times 2$ matrix-valued Riemann-Hilbert problem, denoted as Problem 2.9, involving a jump matrix dependent on a function $\sigma$ and the solution $\Phi$ of a base RH problem (Problem 2.1) associated with the Claeys-Vanlessen kernel.
2. Lax Pair and Integrable Equations: By combining the base RH problem with the one characterizing the statistics, the authors construct a new matrix function $X(\zeta)$. They demonstrate that $X(\zeta)$ satisfies a set of Lax equations. By analyzing the asymptotic expansion of $X(\zeta)$ at infinity, they derive differential equations for the coefficients of this expansion. These coefficients are shown to satisfy the first three equations of the KdV hierarchy and the corresponding potential KdV equations.
3. Asymptotic Analysis: To understand the behavior of the solutions in different regimes, the authors employ the Deift-Zhou nonlinear steepest descent method. They analyze three distinct asymptotic regimes as the parameter $\tau_7$ (related to the scaling of the point process) approaches zero:
   • Regime 1: Large $\tau_5$ relative to $\tau_7$. Here, the jump matrix of the RH problem is exponentially close to the identity, allowing for a direct approximation using the solution of the base RH problem.
   • Regime 2: All scaled variables $\tau_j$ are bounded. The authors construct a global parametrix based on the solution to a RH problem associated with the third member of the Painlevé II hierarchy ($P_{II}^{(3)}$) and a local parametrix near the origin.
   • Regime 3: A specific scaling where $\tau_5$ is negative and large in magnitude. This regime involves a local parametrix constructed from a RH problem related to the Bessel kernel (specifically, the solution to a problem studied in the context of the Airy point process), adapted to the specific scaling of the critical edge.

Key Contributions and Results

• Connection to KdV Hierarchy: The primary result (Theorem 1.5) establishes that the multiplicative statistics $Q_\sigma$, when expressed in terms of specific variables $\tau_1, \tau_3, \tau_5, \tau_7$, generate solutions to the first three equations of the KdV hierarchy. Specifically, the function $v = \partial^2_{\tau_1} \log Q_\sigma + \dots$ satisfies the KdV equations, while $u = \partial_{\tau_1} \log Q_\sigma + \dots$ satisfies the potential KdV equations. This generalizes previous results for the Airy kernel to the $5/2$-power vanishing density case.
• Asymptotic Behavior: The paper provides detailed asymptotic formulas for the functions $u$ and $v$ in the three regimes mentioned above (Theorem 1.10):
  • In the first regime, $u$ and $v$ are asymptotically close to solutions of the potential KdV and $PI(2)$ equations, respectively, with exponentially small error terms.
  • In the second regime, the solutions are expressed in terms of the solution $q$ to the third member of the Painlevé II hierarchy and an auxiliary function $p$, linked via a Miura transformation.
  • In the third regime (where $\iota=1$), the solutions are characterized by functions $p_\sigma$ and $q_\sigma$ arising from a RH problem related to the Bessel kernel, recovering behavior similar to the Airy point process but with specific modifications due to the critical edge nature.
• Explicit Computations: The authors provide explicit derivations of the Lax matrices and the recursive relations defining the KdV hierarchy polynomials, confirming the integrable structure of the problem.

Significance and Claims
The paper claims to establish a rigorous link between the multiplicative statistics of unitary random matrices at a specific multicritical edge ($5/2$ vanishing density) and the KdV hierarchy. This extends the known universality of integrable PDEs in random matrix theory beyond the standard Airy and Bessel kernels.

The authors note that their results serve as the analog for the Claeys-Vanlessen kernel of theorems previously established for the Airy kernel (specifically referencing work by Bothner, Buckingham, Deift, and Kriecherbauer). They highlight that while the stationary reductions of these equations relate to known finite-gap and multi-soliton solutions, the specific connection to the multiplicative statistics of this critical edge process is a novel contribution. Furthermore, the paper distinguishes its findings from previous studies on the Painlevé II hierarchy (e.g., Claeys and Vanlessen [12]) by focusing on the double log-derivative with respect to $\tau_1$ rather than derivatives with respect to the variables appearing in the cumulative distribution function of the largest particle. The work provides a comprehensive framework for analyzing these statistics through the lens of Riemann-Hilbert problems and integrable hierarchies, offering asymptotic expansions that are consistent with known results in limiting cases.