Moments at the hard edge and Rayleigh functions

This paper investigates inverse power trace moments of the Laguerre random matrix ensemble in the hard edge regime, deriving explicit results for classical cases, partition-based formulas for general $\beta$, and demonstrating that the low-temperature limit yields expressions involving the Bessel zeta function.

The Gist

Imagine you have a giant, chaotic crowd of numbers (eigenvalues) dancing inside a box. In the world of Random Matrix Theory, these numbers aren't just random; they follow specific rules of attraction and repulsion, much like charged particles in a gas. This paper is about studying what happens when we look at the "hardest" part of this crowd: the very edge of the box where the numbers are squeezed tightest against a wall.

Here is a simple breakdown of what the authors, Anna Maltsev and Nick Simm, discovered, using some everyday analogies.

1. The Setup: The "Hard Edge" Party

Imagine a party where guests (the numbers) are dancing on a floor that stretches from zero to infinity.

• The Wall: There is a solid wall at zero. Guests can't go past it.
• The Repulsion: The guests don't like to stand too close to each other (they repel), but they are also pushed by the wall.
• The "Hard Edge": This is the area right next to the wall where the guests are packed most tightly.

Usually, if you look at the whole party, you see a smooth, predictable shape (like a hill). But if you zoom in on the wall, the shape changes completely. The authors wanted to know: If we look at the "inverse" of these numbers (1/number), what happens when the party gets infinitely huge?

2. The Mystery of the "Inverse"

Why look at $1/\text{number}$?
Think of it like this: If a guest is very close to the wall (a tiny number), their "inverse" (1/tiny) becomes a giant number.

• The Problem: In the middle of the room, the average size of these "inverse" guests is easy to calculate. But right at the wall, the tiny numbers make the calculation blow up (diverge). It's like trying to calculate the average height of a crowd if one person is infinitely short; the math breaks.
• The Goal: The authors wanted to fix this math so they could understand the behavior of these tiny numbers at the edge.

3. The Three Special Cases (The "Classic" Rules)

First, they looked at three specific types of parties where the rules are well-known (called $\beta = 1, 2, 4$). These correspond to real, complex, and quaternion numbers.

• The Discovery: They found that for these three cases, the "inverse moments" (the average of the inverses) follow a very specific, elegant pattern.
• The Magic Mirror: For the complex case ($\beta=2$), they found a "mirror symmetry." If you flip the calculation around a certain point, the answer stays the same. It's like looking in a mirror and seeing your reflection is identical to the original, just flipped.
• The Result: They wrote down exact formulas for these three cases using special mathematical functions (like Bessel functions, which describe ripples in a pond).

4. The General Case (The "Wild" Party)

What about parties with any set of rules ($\beta > 0$)? This is much harder because the "mirror" trick doesn't work.

• The Puzzle Piece Approach: Instead of a single formula, they used a method involving "partitions." Imagine breaking a number (like 5) into smaller chunks (5, or 4+1, or 3+2, etc.).
• The Formula: They showed that the answer is a sum of many different "partition" pieces. It's like calculating the total weight of a pile of bricks by adding up the weight of every possible way you could stack them.
• The Duality: They found a cool trick: If you swap the rules of the party in a specific way (changing $\beta$ to $4/\beta$), the answer transforms in a predictable way. It's like realizing that a party with 2 dancers per table behaves mathematically like a party with 0.5 dancers per table, just scaled differently.

5. The Grand Finale: The "Bessel Zeta" Connection

This is the most exciting part. The authors looked at what happens when the "temperature" of the party gets extremely low (mathematically, $\beta \to \infty$).

• Freezing the Crowd: As the temperature drops, the chaotic dancing stops. The guests freeze into a perfect, rigid crystal formation.
• The Bessel Connection: In this frozen state, the positions of the guests perfectly match the "zeros" of a specific mathematical wave called the Bessel function.
• The Zeta Function: The authors proved that the "inverse moments" of this frozen crowd are exactly the same as a famous mathematical object called the Bessel Zeta function.
  • Analogy: It's like discovering that the pattern of footprints left by a frozen crowd is identical to the pattern of notes in a specific musical scale (the Riemann Zeta function is the "standard" scale; the Bessel Zeta is the "cylindrical" scale).

Why Does This Matter?

1. Mathematical Unity: It connects three different worlds: Random Matrices (chaos), Bessel Functions (waves/ripples), and Zeta Functions (number theory). It shows that deep down, these different areas of math are speaking the same language.
2. Physics Applications: These "hard edge" calculations help physicists understand things like:
   • Quantum Chaos: How energy levels behave in complex atoms.
   • Entanglement: How particles are linked in quantum computers.
   • Geometry: The shape of drums, cones, and spheres (since Bessel functions describe how sound vibrates on a drum).

In a nutshell: The authors took a messy, chaotic problem at the edge of a mathematical box, found that for specific rules it has a beautiful symmetry, and discovered that when the system gets "cold" enough, it reveals a hidden, perfect order that matches the fundamental vibrations of the universe (Bessel functions).

Technical Summary

1. Problem Statement and Motivation

The paper investigates the inverse power trace moments of the Laguerre ensemble (also known as the Wishart-Laguerre ensemble) in the hard edge scaling regime.

• The Ensemble: The Laguerre ensemble consists of $N \times N$ matrices (or eigenvalues $\lambda_1, \dots, \lambda_N$) with a joint probability density function proportional to:
  $$\prod_{j=1}^N \lambda_j^\alpha e^{-\lambda_j} \prod_{1 \le i < j \le N} |\lambda_j - \lambda_i|^\beta$$
  where $\beta > 0$ is the inverse temperature parameter and $\alpha$ is a shape parameter.
• The Quantity of Interest: The authors study the expectation of the sum of inverse powers of eigenvalues:
  $$M_N^{(\beta)}(-k, \alpha) = \mathbb{E}\left[ \sum_{j=1}^N \frac{1}{\lambda_j^k} \right], \quad k > 0$$
• The Regime: The study focuses on the hard edge limit, where $N \to \infty$ while $\alpha$ is kept fixed. In this regime, the eigenvalues accumulate near zero (the "hard edge" of the spectrum), making the calculation of inverse moments non-trivial because the standard Marchenko-Pastur law (which describes the bulk spectrum) diverges for inverse powers near zero.
• Motivation: The work is driven by the analogy between spectral moments of random matrices and spectral zeta functions. Specifically, the authors aim to connect these moments to the Bessel zeta function and Rayleigh functions (values of the Bessel zeta function at even integers).

2. Methodology

The authors employ distinct mathematical techniques depending on the value of $\beta$:

A. Classical Cases: $\beta \in \{1, 2, 4\}$

For these values, the ensemble corresponds to real, complex, and quaternion matrices, respectively. The methodology relies on determinantal and Pfaffian point processes:

1. Eigenvalue Density: The authors utilize explicit formulas for the one-point eigenvalue density $\rho_N^{(\beta)}(x)$ expressed in terms of Laguerre polynomials (for $\beta=2$) or skew-orthogonal polynomials (for $\beta=1, 4$).
2. Hard Edge Scaling: They apply the hard edge scaling limit ($x \to u/4N$) to these densities. Using asymptotic expansions of Laguerre polynomials, the densities converge to limiting profiles involving Bessel functions ($J_\nu$).
3. Mellin Transforms: The limiting moments are computed as Mellin transforms of these Bessel-based limiting densities.
4. Alternative Proofs: For $\beta=2$, they provide a second proof utilizing the functional equation of the spectral zeta function established in prior work (Theorem 1.1) combined with the bulk Marchenko-Pastur limit.

B. General Case: $\beta > 0$

For arbitrary $\beta$, orthogonal polynomial techniques fail. The authors utilize a combinatorial approach:

1. Partition Sums: They rely on a result by Fyodorov and Le Doussal which expresses the moments as a sum over integer partitions $\eta$.
2. Asymptotic Analysis: They analyze the large $N$ and large $\beta$ limits of the coefficients in these partition sums.
3. Low Temperature Limit: They consider a specific scaling where $\beta \to \infty$ as $N \to \infty$ with $\alpha = \frac{\beta}{2}(\nu + 1)$. This corresponds to a "low temperature" limit where the system becomes deterministic.
4. Matrix Model Connection: They utilize the tridiagonal matrix model representation of the Laguerre ensemble (where eigenvalues are squares of singular values of a bidiagonal matrix with $\chi$-distributed entries). In the $\beta \to \infty$ limit, these random variables converge to their means, reducing the problem to the eigenvalues of a deterministic matrix related to Laguerre polynomials, which in turn converge to Bessel zeros.

3. Key Contributions and Results

A. Explicit Formulas for $\beta \in \{1, 2, 4\}$

The authors derive explicit limiting formulas for the scaled moments $M^{(\beta)}(-s, \alpha) = \lim_{N \to \infty} N^{-s} M_N^{(\beta)}(-s, \alpha)$:

• $\beta = 2$: The limit is expressed via Gamma functions:
  $$M^{(2)}(-s, \alpha) = \frac{4^{s-1}\Gamma(\alpha + 1 - s)\Gamma(s - 1/2)}{\sqrt{\pi}\Gamma(s+1)\Gamma(s+\alpha)}$$
• $\beta = 1, 4$: The results are expressed in terms of hypergeometric functions (${}_3F_2$ and ${}_4F_3$). For example, for $\beta=4$:
  $$M^{(4)}(-s, \alpha) = 2^{s-1}M^{(2)}(-s, \alpha) - \text{correction terms involving } {}_3F_2$$
• Integer Moments: For integer $k$, these hypergeometric series terminate, yielding finite sums involving Gamma functions. The authors also establish linear recurrence relations for these integer moments.

B. General $\beta$ and Partition Sums

• Corollary 1.6: Provides a general formula for the limiting moments for any $\beta > 0$ as a sum over partitions $\eta$ of $k$:
  $$M^{(\beta)}(-k, \alpha) = \sum_{|\eta|=k} A_\eta^{(\beta)} \alpha^{-\eta}$$
  where $A_\eta^{(\beta)}$ are explicit coefficients depending on $\beta$ and the partition structure.
• Duality: The moments satisfy a duality relation under the mapping $\beta \to 4/\beta$.

C. Connection to Bessel Zeta Functions (The Main Theorem)

The most significant result (Theorem 1.8) establishes the link to the Bessel zeta function $\zeta_\nu(s) = \sum_{n=1}^\infty j_{\nu,n}^{-s}$, where $j_{\nu,n}$ are the zeros of the Bessel function $J_\nu$.

• Scaling: Under the scaling $\alpha = \frac{\beta}{2}(\nu + 1)$ and taking the limit $\beta \to \infty$ (with $N \to \infty$):
  $$\lim_{N \to \infty} \left( \frac{\beta_N}{8N} \right)^k M_N^{(\beta_N)}\left(-k, \frac{\beta_N}{2}(\nu+1)\right) = \zeta_\nu(2k)$$
• Rayleigh Functions: This implies that the inverse moments in the low-temperature limit are exactly the Rayleigh functions (values of the Bessel zeta function at even integers). The authors verify that their derived formulas for low-order moments ($k=1, 2, 3, 4$) match the known explicit values of $\zeta_\nu(2k)$ (e.g., $\zeta_\nu(2) = \frac{1}{4(\nu+1)}$).

4. Significance

1. Bridging Random Matrix Theory and Spectral Geometry: The paper rigorously connects the statistical mechanics of random matrices (specifically the hard edge of the Laguerre ensemble) to spectral geometry. It shows that the eigenvalues of the Laplacian on domains with cylindrical symmetry (governed by Bessel zeros) emerge naturally as the low-temperature limit of random matrix ensembles.
2. Extension of Zeta Function Theory: It extends the understanding of spectral zeta functions beyond the classical Riemann zeta function (which corresponds to $\nu=1/2$) to general Bessel zeta functions, providing a probabilistic interpretation for their values at even integers.
3. Analytical Tools: The paper provides a systematic toolkit for calculating hard edge moments, including new recurrence relations and explicit hypergeometric representations for $\beta=1, 4$, which were previously less understood than the $\beta=2$ case.
4. Universality: The results demonstrate the universality of the Bessel zeta function as the limiting object for inverse moments in the hard edge regime under high inverse temperature scaling.

In summary, Maltsev and Simm successfully characterize the inverse moments of the Laguerre ensemble in the hard edge regime, deriving exact formulas for classical symmetry classes and proving that in the general low-temperature limit, these moments converge to the Rayleigh functions associated with Bessel zeros.