Wick theorem for analytic functions of Gaussian fields

✨

This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to understand a massive, chaotic crowd of people. In the world of physics and math, this crowd is a Gaussian Field. Think of it like a giant, invisible trampoline that is constantly rippling up and down due to random wind gusts. Every point on the trampoline has a height (a value), and these heights are related to each other in a specific, predictable way (the "covariance").

Usually, scientists only care about the simple ripples (the raw heights). But in the real world, things are rarely that simple. We often care about complex things built from these ripples: the square of the height (energy), the exponential of the height (probability of extreme events), or even the gradient (how steep the slope is).

This paper is a new instruction manual for calculating the "average behavior" of these complex things. Here is the breakdown using simple analogies:

1. The Problem: Too Many Variables

If you want to know the average of a simple ripple, it's easy. But if you want the average of a complicated formula (like $e^{\text{height}}$ ), you have to multiply thousands of random numbers together. Doing this by hand is impossible.

For simple ripples, there's an old rule called Wick's Theorem (or Isserlis' Theorem). It says: "To find the average of a bunch of ripples, just pair them up."

Analogy: Imagine you have a room full of people holding hands. To count the total number of handshakes, you don't need to ask everyone individually. You just look at the pairs. If everyone is holding hands in pairs, the math is simple.

2. The Innovation: Pairing "Complex" People

The authors of this paper asked: "What if the people in the room aren't just holding hands, but are wearing complex costumes (analytic functions)?"

They developed a new, super-charged version of Wick's Theorem. Instead of just pairing simple ripples, they showed how to pair up complex costumes.

The Tool: They used Multigraphs (a type of drawing).
The Metaphor: Imagine each point on the trampoline is a node in a drawing. When you calculate the average of a complex function, you draw lines (edges) between these nodes.
- If you have a function like $x^4$ , it's like a node with 4 "arms" reaching out to grab other nodes.
- The paper provides a recipe to draw all possible ways these arms can grab each other without getting tangled in themselves.
- Every unique drawing (multigraph) corresponds to a specific term in the final math equation.

3. The "Zoom Out" Effect (Scaling Limits)

The paper starts with a discrete world (a grid, like pixels on a screen) and zooms out to the continuous world (a smooth, real-life surface).

Analogy: Imagine looking at a digital photo. Up close, you see individual square pixels (the lattice). As you zoom out, the pixels blur together, and you see a smooth, continuous image (the continuum).
The authors proved that their "pairing rules" for the pixelated world work perfectly when you zoom out. The messy grid of connections transforms into a smooth, elegant formula used in advanced physics (Fock space fields). This means their math works for both computer simulations and real-world physics.

4. The Twist: Bosons vs. Fermions (The "Good Cop, Bad Cop" Duo)

In physics, there are two types of particles:

Bosons: The "social" particles. They like to be in the same place and behave nicely (positive correlations).
Fermions: The "antisocial" particles. They hate being in the same place and follow strict "no-touch" rules (negative correlations/anti-commuting).

Usually, calculating things for Fermions is a nightmare because of their "antisocial" nature. However, the authors discovered a surprising duality (a mirror image relationship).

The Discovery: They showed that for certain even-powered functions (like squaring the field), the messy math of the "social" Bosons is mathematically identical to the "antisocial" Fermions, just with a sign flip (positive becomes negative).
The Analogy: It's like realizing that a chaotic dance party (Bosons) and a strict, silent library (Fermions) actually produce the exact same pattern of movement if you look at them from a specific angle, provided you flip the lights.
The Catch: This only works if a specific algebraic puzzle (finding the "principal minors" of a matrix) can be solved. The paper highlights this as a deep, unsolved mystery in algebra that connects these two very different worlds.

Summary

In short, this paper is a universal translator for random fields.

It takes complex, non-linear functions (like exponentials or squares) of random noise.
It translates them into a language of drawings (graphs) where you just count connections.
It proves these rules work whether you are looking at a pixelated grid or a smooth surface.
It reveals a hidden secret: that the chaotic "social" particles and the strict "antisocial" particles are actually two sides of the same coin, linked by a deep algebraic structure.

This allows physicists and mathematicians to solve problems that were previously too messy to calculate, bridging the gap between abstract algebra, probability theory, and quantum physics.

1. Problem Statement

The paper addresses the challenge of computing correlations and cumulants for analytic functions of general Gaussian fields (both discrete on lattices and continuous in the scaling limit). While Wick's theorem and Isserlis' theorem provide systematic methods for computing expectations of products of linear Gaussian variables (or polynomials), extending these to general analytic functions (e.g., exponentials $e^{\alpha \phi}$ , trigonometric functions, or general power series) in the context of interacting quantum field theories and statistical mechanics is non-trivial.

Specifically, the authors aim to:

Derive a combinatorial framework for the correlation functions of $:f(\phi(x)):,$ where $f$ is an analytic function and $\phi$ is a Gaussian field.
Establish the scaling limits of these correlations as discrete lattice fields converge to continuous Gaussian Free Fields (GFF) or fractional GFFs.
Connect these bosonic (real/complex) field correlations to fermionic Gaussian states, investigating whether a duality exists between the cumulants of even powers of bosonic fields and specific fermionic observables.

2. Methodology

The authors employ a combination of probabilistic combinatorics, Fock space theory, and Grassmann-Berezin calculus.

Multigraph Representation: Instead of standard Feynman diagrams (which pair individual field insertions), the authors generalize the contraction patterns to multigraphs. For a product of Wick-ordered powers $: \phi(x_i)^{n_i} :$ , the correlations are expressed as a sum over undirected multigraphs where vertices represent the field insertion points and edges represent covariance contractions.
Wick Products and Orthogonal Projections: The analysis relies on the definition of Wick products via orthogonal projections in Gaussian Hilbert spaces ( $L^2$ ). This allows the decomposition of analytic functions into orthogonal chaos components.
Scaling Limits: The paper utilizes the convergence of discrete covariance matrices $G_\epsilon$ to continuous kernels $g_U$ (e.g., Green's functions of Laplacians) to derive the continuum limits of the discrete correlation formulas.
Fock Space Isomorphism: The authors map the discrete correlations to the tensor product structure of Bosonic Fock spaces. They define "Fock space fields" as formal Wick products and show that the scaling limit of discrete correlations corresponds to a specific tensor product of correlation functionals in the continuum.
Fermionic Duality: To investigate the boson-fermion correspondence, the authors use Grassmann variables and the Berezin integral. They define fermionic Gaussian states and compare their cumulants with those of bosonic fields using determinant and permanent identities.

3. Key Contributions and Results

A. Generalized Wick Theorem for Analytic Functions

The central result is Theorem 2.4, which provides a closed-form expression for the expectation of a product of Wick-ordered analytic functions of a Gaussian field.

Formula: For an analytic function $f(x) = \sum a_n x^n$ and points $x_1, \dots, x_N$ , the expectation is:
$E\left[ \prod_{i=1}^N :f(\phi(x_i)): \right] = \sum_{n_1, \dots, n_N} \left( \prod a_{n_i} \right) \sum_{q \in \text{MG}_0} \delta(q) \prod_{i,j} G(i,j)^{q_{ij}}$
where the inner sum runs over multigraphs $q$ with no self-loops, $G$ is the covariance matrix, and $\delta(q)$ is a combinatorial constant counting the number of Feynman diagrams that map to the multigraph $q$ .
Cumulants: The paper further derives that the joint cumulants correspond to the sum over connected multigraphs only (Equation 2.17).

B. Scaling Limits and Fock Space Connection

Proposition 3.1 establishes the link between discrete lattice models and continuous field theory.

It proves that the scaling limit of the discrete correlations of analytic functions converges to the tensor product of correlation functionals in the Bosonic Fock space.
Specifically, $\lim_{\epsilon \to 0} \eta(\epsilon)^N E[\dots] = Y_1 \bullet \dots \bullet Y_N$ , where $Y_j$ are Fock space fields and $\bullet$ denotes a specific correlation functional tensor product.
This applies to various models, including the Gaussian Free Field (GFF), its gradient, the membrane model (bi-Laplacian), and Fractional Gaussian Fields.

C. Boson-Fermion Duality and Principal Minors

The paper investigates the relationship between the cumulants of even powers of bosonic fields and fermionic fields.

Theorem 4.3: For the case of squares ( $r=1$ ), the authors prove a duality: the cumulants of the squared complex bosonic field $|Z_i|^2$ are equal (up to a sign $(-1)^{|A|-1}$ ) to the cumulants of fermionic bilinears $\psi_i \bar{\psi}_i$ , provided the covariance matrices match ( $C=G$ ).
Generalization to $2r$ -powers (Lemma 4.6): The authors generalize this to higher even powers ( $r > 1$ $r > 1$ ). They show that a similar duality holds if and only if a specific principal minor assignment problem for the inverse of the fermionic covariance matrix is solvable.
- The condition requires that the inverse of the restricted covariance matrix $(C^{-1})_{A_r, A_r}$ can be expressed as a sum over multigraphs weighted by signs and combinatorial factors.
Open Problem: The authors note that solving this principal minor assignment for general $r$ and arbitrary covariance matrices is a long-standing open problem in algebra, implying that the boson-fermion duality for higher-order powers is not guaranteed for all Gaussian states.

4. Significance

Unification of Diagrammatics: The paper unifies the treatment of nonlinear observables in Gaussian fields by replacing standard Feynman diagrams with multigraphs, providing a rigorous combinatorial tool for non-perturbative calculations in statistical mechanics and quantum field theory.
Rigorous Scaling Limits: It provides a mathematically rigorous bridge between discrete lattice models (like the DGFF) and their continuous counterparts (Fock space fields), validating the use of continuum Fock space techniques for discrete observables.
Supersymmetric Insights: By characterizing the conditions under which bosonic and fermionic cumulants coincide, the paper sheds light on the underlying supersymmetric structures in Gaussian systems. It clarifies that while such duality is exact for squares ( $r=1$ ), it imposes strict algebraic constraints (principal minor identities) for higher powers.
Applications: The results are directly applicable to models involving Liouville quantum gravity, $\phi^4$ theory, sine-Gordon models, and Gaussian multiplicative chaos, where analytic functions of fields are fundamental.

In summary, this work extends the classical Wick theorem to a broad class of analytic functionals, formalizes their continuum limits via Fock space tensor products, and deepens the understanding of the algebraic relationship between bosonic and fermionic Gaussian states.