Exponential decay of correlations at high temperature in $H^{2|2n}$ nonlinear sigma models

This paper establishes the exponential decay of two-point correlations for nonlinear sigma models with $H^{2|2n}$ target spaces in the high-temperature regime by combining cluster expansion, exact combinatorics, and Grassmann norm bounds to reduce the problem to a marginal fermionic theory.

The Gist

Imagine you are trying to understand how a crowd of people moves through a giant, chaotic city. In physics, this "crowd" is often a collection of particles, and the "city" is a grid (like a lattice). Sometimes, the city is full of random obstacles (disorder), making it hard to predict how the particles will behave.

This paper is about a specific, very complex mathematical model used to study these chaotic systems. The authors, Margherita Disertori, Javier Durán Fernández, and Luca Fresta, have proven a very important rule about how these particles interact when the "temperature" is high.

Here is the breakdown in simple terms, using some creative analogies.

1. The Setting: A City with Ghosts and Superpowers

The model they are studying is called the $H^{2|2n}$ nonlinear sigma model. That sounds scary, but let's break it down:

• The Grid ($Z^d$): Think of the city as a giant grid of street corners.
• The Particles: At every corner, there is a "super-particle."
• The Twist (Supersymmetry): These aren't normal particles. They have two types of "legs":
  • Real legs (Bosons): These are like normal people walking around.
  • Ghost legs (Fermions/Grassmann variables): These are like invisible, anti-social ghosts. They follow weird rules: if two ghosts try to occupy the same spot, they cancel each other out (this is called the Pauli exclusion principle, but in a math sense).
• The Target Space ($H^{2|2n}$): This is the "shape" the particles can take. It's a weird, hyperbolic shape (like a saddle or a Pringles chip) that exists in a world where ghosts and people mix. The number $n$ tells us how many pairs of "ghost legs" each particle has.

2. The Problem: How Far Do They Talk?

In physics, we often ask: "If I wiggle a particle at point A, how much does a particle at point B feel it?"

• Low Temperature (Cold): The particles are sluggish and stuck. They might get "stuck" in one spot (localization).
• High Temperature (Hot): The particles are energetic and moving fast. Usually, in a hot system, the influence of one particle on another dies out very quickly as you get further away. This is called exponential decay of correlations.

The big question for this specific model (with the ghost legs) was: Does this "fast forgetting" happen even when the system is full of ghosts and disorder?

3. The Discovery: The "Ghost" Advantage

The authors proved that YES, it does happen.

They showed that if the temperature is high enough (specifically, if the inverse temperature $\beta$ is small enough relative to the number of ghosts $n$), the connection between two points drops off incredibly fast.

The Analogy of the "Ghost Tax":
Imagine the particles are trying to send a message to their neighbors.

• In a normal system, the message gets weaker as it travels.
• In this model, the "ghost legs" act like a tax on the message. Every time the message tries to hop from one corner to another, it has to pay a "ghost tax."
• The more ghosts ($n$) you have, the higher the tax.
• The authors proved that if the "heat" (energy) isn't strong enough to pay this tax, the message dies out almost instantly. The influence decays exponentially, meaning if you double the distance, the connection doesn't just get half as strong; it gets squared (or cubed, etc.) weaker.

4. The Method: How They Solved It

Proving this wasn't easy because the math involves "Grassmann variables" (the ghosts), which are notoriously difficult to handle. You can't just use standard calculus; you have to use a special kind of algebra.

The authors used a clever trick called a High-Temperature Expansion.

• The Metaphor: Imagine trying to count all the ways a crowd can move. Instead of looking at the whole crowd at once, they broke the problem down into tiny "clusters" or "polymer groups."
• They treated the interactions between particles like a game of connecting dots. They proved that if the "cost" of connecting dots (the interaction strength) is low enough, the only way to connect two far-away points is to build a massive, expensive chain of connections.
• Because the "ghost tax" makes these chains incredibly expensive, the probability of a long chain existing becomes tiny.

They combined this with Grassmann Norms (a way of measuring the "size" of these ghostly math objects) to show that the math works out perfectly, even as the number of ghosts ($n$) gets huge.

5. Why Does This Matter?

This isn't just abstract math. These models are used to understand disordered systems, like:

• Random Materials: Metals that aren't perfect crystals.
• Quantum Chaos: How electrons move through messy environments.
• The Anderson Transition: A famous phenomenon where a material switches from being an electrical conductor to an insulator.

By proving that correlations decay exponentially in this high-temperature regime, the authors have provided a rigorous mathematical foundation for understanding how disorder and "ghostly" quantum effects interact. It confirms that in these chaotic, hot environments, local disturbances don't spread forever; they die out quickly.

Summary

Think of the paper as a proof that in a chaotic, super-hot city filled with invisible ghosts, if you shout from one corner, the people at the other end of the city won't hear you. The "ghosts" make the noise cancel itself out so effectively that the signal vanishes before it can travel far. The authors figured out exactly how hot the city needs to be for this to happen, no matter how many ghosts are in the crowd.

Technical Summary

1. Problem Statement and Context

The paper investigates the $H^{2|2n}$ nonlinear sigma model, a supersymmetric statistical mechanical model defined on a lattice $\mathbb{Z}^d$.

• Target Space: The model's target space is the hyperbolic supermanifold $H^{2|2n}$ ($n > 1$), which generalizes Zirnbauer's well-studied $H^{2|2}$ model. It consists of two real variables and $2n$ Grassmann (anticommuting) variables.
• Physical Motivation: These models are crucial for understanding disordered systems, specifically the Anderson metal-insulator transition and random band matrices. The $H^{2|2}$ case ($n=1$) is known to exhibit an Anderson transition (localization at low temperature, quasi-diffusion at high temperature).
• The Gap: While the $n=1$ and $n=2$ cases have been analyzed using probabilistic representations (Vertex-Reinforced Jump Process and Arboreal Gas, respectively), the behavior of the general $H^{2|2n}$ model for $n > 1$ remained less understood, particularly regarding rigorous bounds on correlation decay in the high-temperature regime.
• Objective: The authors aim to prove exponential decay of the two-point correlation function in the high-temperature regime (small inverse temperature $\beta$) for any dimension $d \geq 1$ and any $n > 1$.

2. Methodology

The proof relies on a combination of supersymmetric localization, high-temperature cluster expansions, and Grassmann algebraic techniques.

A. Reduction to a Marginal Fermionic Model

Using the supersymmetric localization theorem, the authors reduce the original $H^{2|2n}$ model (involving both commuting and anticommuting variables) to a purely fermionic $H^{0|2m}$ model (where $m = n-1$).

• The commuting variables ($x, y, \xi, \eta$) are integrated out exactly.
• The resulting model involves only Grassmann variables $\psi$ and $\bar{\psi}$ with a specific interaction structure derived from the hyperbolic constraint.
• This reduction transforms the problem into analyzing a purely fermionic system with a Hamiltonian that is a polynomial perturbation of a Gaussian.

B. High-Temperature Cluster Expansion

The authors perform a high-temperature expansion of the generating function for the $H^{0|2m}$ model.

• Polymer Representation: The Gibbs weight is expanded into connected components, representing the system as a hard-core polymer gas.
• Activity Definition: The "activity" of a polymer (a subset of lattice sites $Y$) is defined via the connected part of the interaction weight.
• Correlation Bounds: The two-point correlation function is expressed as a sum over polymers connecting two points. The decay is controlled by bounding the magnitude of these activities.

C. Grassmann Norms and Combinatorial Analysis

A central technical challenge is obtaining bounds that are optimal in the parameter $n$ (or $m$).

• Grassmann Norms: The authors utilize $\ell_1$-type norms on the Grassmann algebra to bound the Berezin integrals.
• The "Norm Problem": Standard norm estimates for the variable $z = \sqrt{1 + \psi \cdot \psi}$ grow as $\|z\| \sim m^m$, which would lead to suboptimal bounds.
• Refined Estimation: The authors develop a refined combinatorial analysis. They carefully track the powers of $z$ in the expansion and utilize the specific structure of the single-site partition function $\tilde{Z}_{\varepsilon, m}$.
• Key Insight: They prove that the ratio of the norm of the interaction terms to the partition function remains bounded uniformly in $m$ (specifically, the effective coupling scales as $\beta m$). This is achieved by exploiting the exact cancellation properties within the single-site model (Section 3.1) and the Battle-Brydges-Federbusch formula for connected components.

3. Key Contributions and Results

Main Theorem (Theorem 1.1)

The authors prove that for any dimension $d \geq 1$ and $n > 1$, there exists a constant $C_0$ such that if the high-temperature condition $\beta n C_0 < 1$ holds, the two-point correlation function decays exponentially (or according to the interaction decay) uniformly in the volume and the external field $\varepsilon$.

Specifically, for nearest-neighbor interactions:
$$|\langle \xi_{i_0} \eta_{j_0} \rangle| \lesssim (C_0 \beta n)^{|i_0 - j_0|}$$
Similar bounds are established for:

1. Exponentially decaying interactions: Decay rate matches the interaction range.
2. Polynomially decaying interactions: Decay rate matches the interaction exponent $a$.

Technical Innovations

1. Optimal $n$-dependence: The proof establishes that the critical threshold for the high-temperature phase scales as $\beta \sim 1/n$. This is shown to be optimal because the $n$ fermionic degrees of freedom effectively sum up the interaction strength, analogous to the $1/N$ scaling in classical $O(N)$ models.
2. Uniformity in $\varepsilon$: Unlike the $n=1$ case where correlations can diverge as the pinning field $\varepsilon \to 0$, the $n > 1$ case exhibits uniform bounds even as $\varepsilon \to 0$. This is attributed to the "surplus" of Grassmann variables preventing the divergence.
3. Generalized Decay: The results extend to truncated correlation functions of arbitrary order, showing decay based on the minimal tree connecting the observation domains.

Connection to Probabilistic Models

The paper discusses the relationship between $H^{2|2n}$ and probabilistic models:

• $n=1$: Linked to the Vertex-Reinforced Jump Process (VRJP).
• $n=2$: Linked to the Arboreal Gas (acyclic percolation).
• $n > 2$: While no direct probabilistic representation (like the arboreal gas) is known for general $n$, the authors' rigorous bounds provide a mathematical foundation for the expected behavior of these systems, suggesting a universal mechanism for localization/delocalization transitions.

4. Significance

• Rigorous Control of Supersymmetry: The paper provides one of the first rigorous proofs of exponential decay for a broad class of supersymmetric sigma models with target spaces beyond $H^{2|2}$.
• Resolution of Scaling: It clarifies the role of the number of Grassmann components ($n$) in the high-temperature expansion, resolving the scaling of the critical temperature.
• Methodological Advance: The combination of Grassmann norm estimates with precise combinatorial analysis of the single-site partition function offers a new toolkit for handling non-Gaussian fermionic systems where standard cluster expansion techniques might fail due to large norms.
• Implications for Disordered Systems: By establishing the high-temperature (localized) phase rigorously for $n > 1$, the work supports the physical intuition that these models exhibit Anderson-type transitions, providing a stepping stone for future analysis of the low-temperature (delocalized) phase and the critical point.

In summary, the paper successfully extends the rigorous understanding of hyperbolic supersymmetric sigma models from the specific cases of $n=1, 2$ to the general $n > 1$ regime, proving that high-temperature correlations decay exponentially with a rate determined by the product of the temperature and the number of fermionic components.