Renormalised two-point functions of CLE$_4$ gaskets

This paper provides a purely probabilistic calculation, using Brownian loop soups and the Gaussian free field, of renormalised two-point functions for nested CLE$_4$ gaskets, thereby establishing a connection to the scaling limit of the Ashkin-Teller model and recovering Ising correlations at the decoupling point.

The Gist

Imagine you are standing in a vast, empty room (a mathematical "domain") and you start dropping tiny, invisible ink blots onto the floor. These blots aren't random; they follow very specific, magical rules. Sometimes they form single, isolated rings. Other times, they form a Russian nesting doll structure, where a ring contains another ring, which contains another, and so on, forever.

This paper is about understanding the geometry of these invisible rings and how likely it is for two specific points in the room to end up inside the same ring (or the same nested set of rings).

Here is a breakdown of the paper's big ideas using simple analogies:

1. The Main Characters: The "CLE4" Rings

The authors are studying a specific type of random loop pattern called CLE4 (Conformal Loop Ensemble with parameter 4).

• The Analogy: Think of CLE4 as a magical, self-replicating soap bubble foam.
  • Simple CLE4: Imagine a single layer of bubbles floating on a pond. They don't overlap, and they don't touch the shore.
  • Nested CLE4: Now, imagine that inside every bubble, a new, smaller layer of bubbles instantly appears. And inside those, even smaller ones. It's an infinite fractal of bubbles within bubbles.
• The "Gasket": The authors call the solid "skin" of these bubbles a "gasket." If you were to trace the outline of a bubble and all the bubbles inside it, that outline is the gasket.

2. The Big Question: "Are we neighbors?"

The paper asks a very specific question: If I pick two points, $z_1$ and $z_2$, in this room, what are the odds that they end up inside the same bubble (or the same nested set of bubbles)?

In the real world, this is like asking: "If I drop two grains of sand on a beach, what are the chances they are trapped inside the same wave?"

The authors found a precise mathematical formula for this probability. It's not just a simple number; it depends on:

• How close the points are to each other.
• How close they are to the walls of the room.
• A special mathematical "shape factor" (involving things called Theta functions, which are like complex, wavy patterns).

3. The Secret Weapon: The "Brownian Loop Soup"

How did they solve this? They didn't just stare at the bubbles. They used a tool called a Brownian Loop Soup.

• The Analogy: Imagine a pot of soup where the ingredients are tiny, jittery, invisible paths (like drunk ants walking randomly). These paths loop around and cross each other.
• The Magic: The authors discovered that if you let these "drunk ant" paths interact in a specific way, they naturally form the exact same bubble patterns (CLE4) as the ones in the physics models they are trying to understand.
• By studying the "soup," they could calculate the probability of the bubbles without having to simulate the bubbles directly. It's like figuring out the shape of a cloud by studying the wind patterns that create it.

4. The Connection to Physics: The "Ashkin-Teller" Model

Why do we care about these bubbles? Because they represent the behavior of materials at the "critical point"—the exact moment a material changes state (like ice melting into water, or a magnet losing its magnetism).

• The Ising Model (The Classic): This is the simplest model of a magnet. The authors show that their bubble math perfectly recovers the known results for this simple magnet.
• The Ashkin-Teller Model (The Complex Cousin): This is a more complex model where you have two interacting magnets. The authors found that their bubble math describes this complex system too!
  • The Twist: They found that the "odd" numbered bubbles correspond to one type of magnetic behavior, and the "even" numbered bubbles correspond to another. By mixing them, they can describe the complex Ashkin-Teller model.

5. The "Renormalized" Probability

You might notice the math involves terms like $\epsilon^{-1/8}$. This is a bit technical, but here's the simple version:

• The Problem: If you try to calculate the chance of two points being in the same bubble, and you make the points infinitely small, the probability usually drops to zero. It's like asking, "What are the odds two specific atoms are in the same house?" The odds are zero because atoms are tiny.
• The Fix: The authors "renormalize" the answer. They multiply the tiny probability by a huge number (based on how small the points are) to get a meaningful, finite answer.
• The Result: They found that even though the points are tiny, the relative chance of them being neighbors follows a beautiful, predictable pattern dictated by the geometry of the room.

6. The "GFF" (Gaussian Free Field)

The paper also connects these bubbles to something called the Gaussian Free Field (GFF).

• The Analogy: Imagine a trampoline. If you bounce on it, it creates a bumpy surface. The GFF is a random, bumpy surface that exists everywhere in the universe.
• The Link: The authors proved that the "bubbles" (CLE4) are actually the contour lines (like the lines on a topographic map) of this random bumpy surface. If you draw a line where the trampoline is exactly 5 units high, you get a bubble. If you draw the line for 10 units, you get a bubble inside the first one.
• This connection allows them to use the physics of the "bumpy trampoline" to solve the geometry of the "bubbles."

Summary: What did they actually achieve?

1. They mapped the bubbles: They wrote down the exact formula for the probability that two points share a bubble in this infinite nesting doll structure.
2. They bridged two worlds: They showed that the geometry of these random loops (CLE4) is the same as the physics of complex magnets (Ashkin-Teller model).
3. They provided a new lens: They proved that you can understand these complex physical systems by looking at "Brownian loop soups" and "bumpy trampolines" (GFF), offering a purely probabilistic way to solve problems that were previously thought to require heavy physics theories.

In a nutshell: The authors took a complex, abstract problem about random loops and magnets, translated it into the language of "drunk ants" and "bumpy trampolines," and used that to write a new, precise rulebook for how these systems behave.

Technical Summary

1. Problem Statement and Context

The paper addresses the computation of renormalised two-point correlation functions for the Conformal Loop Ensemble with parameter $\kappa=4$ (CLE$_4$). Specifically, the authors aim to calculate the probabilities that two distinct points, $z_1$ and $z_2$, in a simply-connected domain $D$ belong to the same CLE$_4$ "gasket" (the union of loops forming a cluster).

The study focuses on two specific configurations:

1. Nested CLE$_4$: The full collection of loops obtained by iteratively sampling independent CLE$_4$ inside the loops of the previous generation.
2. Outermost (Simple) CLE$_4$: The collection of only the outermost loops (the boundary of the union of all loops).

These calculations are motivated by the conjecture that CLE$_4$ describes the scaling limit of interfaces in the Ashkin-Teller (AT) model on its critical line, as well as the 4-Potts model and the Ising model (at specific points). The authors seek to establish a rigorous geometric representation (an FK-type representation) for the spin-spin correlations of these models using CLE$_4$ and the Gaussian Free Field (GFF).

2. Methodology

The authors employ a purely probabilistic approach, avoiding direct reliance on Conformal Field Theory (CFT) integrability, though they acknowledge the connection. Their methodology relies on three main pillars:

A. Brownian Loop Soups (BLS) and Twist Fields

The core of the calculation involves Brownian loop soups at critical intensity ($\alpha = 1/2$, corresponding to central charge $c=1$).

• The authors define "twist field correlations" via the probability that no cluster of the loop soup (plus boundary excursions) disconnects $z_1$ from $z_2$.
• They introduce two regularizations (cut-offs) for these probabilities:
  1. Time cut-off: Restricting loops to have duration $t > \delta$.
  2. Spatial cut-off: Restricting loops to avoid small disks $D(z_i, \varepsilon_i)$ around the points.
• They prove the equivalence of these definitions and derive explicit asymptotic expansions for the loop soup measures as the cut-offs vanish. This involves mapping the domain to an annulus or a strip and using heat kernel estimates and Jacobi theta functions.

B. Coupling with the Gaussian Free Field (GFF)

The paper utilizes the deep connection between BLS and the GFF (Isomorphism Theorems):

• Sign Excursion Sets: The clusters of a BLS with boundary excursions correspond to the sign excursion sets of a GFF with constant boundary conditions.
• Two-Valued Sets (TVS): The authors use the geometry of Two-Valued Sets $A_{-a, b}$ of the GFF. Specifically, the CLE$_4$ gasket corresponds to the TVS $A_{-2\lambda, 2\lambda}$ where $2\lambda = \sqrt{\pi/2}$ is the height gap.
• Iterated TVS: To model the nested structure and the AT model, they construct an iterative process: sampling a CLE$_4$ gasket, then sampling specific TVS inside the resulting components, and repeating. This iteration mimics the level lines of a compactified GFF.

C. Renormalisation and Conformal Geometry

The probabilities of points belonging to a gasket vanish as the points approach the gasket. The authors define renormalised probabilities by multiplying the raw probability by $\varepsilon_1^{-1/8}\varepsilon_2^{-1/8}$.

• They relate these renormalised probabilities to the conformal radius $CR(z, D)$ and the Dirichlet Green's function $G_D(z_1, z_2)$.
• The final formulas are expressed in terms of Jacobi Theta functions ($\theta_2, \theta_3$) and the nome $q$, which is uniquely determined by the conformal invariant $G_D(z_1, z_2)$.

3. Key Contributions and Results

A. Explicit Formulas for Correlation Functions

The paper provides the first rigorous computation of two-point functions for CLE$_4$ in domains with non-trivial modular parameters (i.e., not just the whole plane or infinite strip).

Theorem 1.1 (Nested CLE$_4$):
The renormalised probability that $z_1$ and $z_2$ belong to the same nested CLE$_4$ gasket is:
$$\lim_{\varepsilon \to 0} \varepsilon_1^{-1/8}\varepsilon_2^{-1/8} P(\text{same gasket}) = C_0 CR(z_1, D)^{-1/8} CR(z_2, D)^{-1/8} \exp\left(\frac{\pi}{2} G_D(z_1, z_2)\right)$$
This is identified with the two-point function of the imaginary multiplicative chaos $:e^{i\sqrt{\pi/2}\Phi_D}:$.

Theorem 1.2 (Outermost CLE$_4$):
The renormalised probability for the outermost gasket involves a ratio of Theta functions:
$$\lim_{\varepsilon \to 0} \varepsilon_1^{-1/8}\varepsilon_2^{-1/8} P(\text{outermost}) = C_1 CR(z_1, D)^{-1/8} CR(z_2, D)^{-1/8} \sqrt{\frac{\theta_3(q^{1/4})}{\theta_2(q^{1/4})}}$$
where $q$ is determined by $\exp(\pi G_D(z_1, z_2)) = \theta_3(q^{1/2})/\theta_2(q^{1/2})$.

B. Connection to Statistical Physics Models

The authors demonstrate that their probabilistic constructions recover known results for specific models:

• Ising Model ($g=2$): By restricting the iteration to specific boundary conditions (or alternating CLE$_4$ with specific TVS), they recover the Ising spin-spin correlation function, providing a CLE$_4$-based FK representation for the Ising model.
• 4-Potts Model ($g=4$): Restricting to odd or even gaskets corresponds to the FK representation of the 4-Potts model with wired or free boundary conditions.
• Ashkin-Teller Model: They generalize the construction to a parameter $g > 1$, proposing a continuum FK representation for the entire critical line of the AT model.

C. Geometric Representation of Spin Fields

The paper conjectures (and proves for specific cases) that the scaling limit of single spins in the AT model can be represented as a sum over CLE$_4$ gaskets weighted by their Minkowski content and a sign determined by the "height" of the gasket in the coupled GFF.
$$\sigma(z) \sim \sum_{C \in \text{CLE}_4} \nu(C) s_C$$
where $s_C$ is a sign determined by the random walk of heights associated with the nested gaskets.

4. Significance and Impact

1. Rigorous CFT-Probability Bridge: The work provides a rigorous probabilistic derivation of correlation functions that were previously only accessible via non-rigorous CFT arguments or specific lattice integrability. It explicitly links the geometry of CLE$_4$ and GFF level lines to the $c=1$ CFT of the Ashkin-Teller model.
2. New FK Representations: It proposes and partially proves a new geometric (FK-type) representation for the continuum limit of the Ising and Ashkin-Teller spin fields using CLE$_4$. This is significant because the standard FK representation for the Ising model usually involves CLE$_3$, not CLE$_4$. The paper suggests CLE$_4$ plays a role via the "dual" of sign excursions.
3. Technical Innovation: The paper develops sophisticated tools for handling iterated Two-Valued Sets and renormalised probabilities in the continuum. The use of Brownian loop soups to compute "twist field" correlations serves as a powerful new technique for studying critical phenomena.
4. Universality: The results reinforce the universality of CLE$_4$ as the scaling limit for a wide class of models (Ising, 4-Potts, AT, Double-Dimer) and clarify the role of the GFF in organizing these limits.

In summary, Aru and Lupu successfully translate the complex algebraic structures of the Ashkin-Teller model into the geometric language of random loops and free fields, offering a unified probabilistic framework for understanding critical spin correlations in 2D statistical mechanics.