Topics in Gaussian Wiener chaos expansion

This document is a set of lecture notes titled "Topics in Gaussian Wiener Chaos Expansion" by Nils Berglund. It is designed for a summer school of mathematicians and physicists.

To explain this to a general audience, imagine you are trying to understand a very complex, noisy, and chaotic system—like the weather, a stock market, or a quantum field. The paper provides a mathematical "toolkit" for taking that chaos, breaking it down into simple, understandable pieces, and then rebuilding it to make predictions.

Here is the breakdown of the paper's journey, using everyday analogies:

1. The Foundation: The "Gaussian" and the "Dice"

The paper starts with the basics: Gaussian random variables.

The Analogy: Imagine rolling a single die. The result is random. Now imagine rolling millions of dice and adding them up. The result will almost always form a perfect "bell curve" (the Gaussian distribution).
The Problem: In physics, we often deal with functions of these random variables (like the energy of a system). Calculating the average outcome of these functions is hard because the "dice" are interacting in complex ways.
The Solution (Hermite Polynomials): The author introduces Hermite polynomials. Think of these as a special set of "Lego bricks." Just as you can build any complex shape out of Lego bricks, you can build any random function out of these specific polynomials. The paper shows how to create these bricks and how they fit together perfectly without overlapping (orthogonality).

2. The Big Idea: "Wiener Chaos Expansion"

This is the core concept of the paper.

The Analogy: Imagine a piece of music. It sounds complex, but it is actually just a sum of simple notes (frequencies).
The Concept: The Wiener Chaos Expansion says that any random variable (any "song" in the universe of probability) can be broken down into a sum of these Hermite polynomial "notes."
- The first note is the average (the silence).
- The second note is the first layer of noise.
- The third note is a more complex layer of noise, and so on.
Why it matters: Instead of trying to solve the whole messy equation at once, you can solve it note by note. This turns a terrifyingly hard problem into a series of manageable steps.

3. Moving to Many Dimensions: The "Fock Space"

The paper then moves from one variable to many (multivariate).

The Analogy: Imagine a choir. One singer is easy to analyze. But a choir of 100 singers? That's chaotic.
The Concept: The author uses a concept called Fock space (borrowed from quantum physics). Think of this as a "library of states."
- Level 0: No singers (silence).
- Level 1: One singer.
- Level 2: Two singers interacting.
- Level $n$ : $n$ singers interacting.
The Magic: The paper shows that you can treat the interactions between these "singers" (random variables) using a special math trick called the Wick product. This is like a rulebook that tells you how to multiply two complex songs together without creating a mess. It separates the "pure" interaction from the "noise" that just cancels itself out.

4. The Infinite Case: White Noise and Fields

The paper then scales this up to infinite dimensions, dealing with Gaussian Fields (like a field of grass where every blade is moving randomly).

The Analogy: Imagine White Noise. It's like static on a radio. It's so chaotic that at any single point, the value is infinite and undefined. It's "rougher" than a function; it's more like a "distribution" (a mathematical ghost).
The Gaussian Free Field (GFF): This is a slightly smoother version of white noise. Imagine a rubber sheet being shaken randomly. The sheet has a shape, but it's very bumpy.
The Challenge: In 1 dimension (a line), this rubber sheet is smooth enough to touch. In 2 or 3 dimensions (a surface or a volume), it becomes so bumpy that you can't even define its height at a specific point. It's "too rough."

5. The Climax: The $\Phi^4$ Model and "Renormalization"

The final and most complex part of the paper deals with the $\Phi^4$ model. This is a famous toy model in physics used to describe how particles interact.

The Problem: When you try to calculate the energy of this system in 2 or 3 dimensions, you get infinity. The math breaks down because the "bumps" in the rubber sheet are too wild.
The Solution (Renormalization): This is the paper's most dramatic moment. To fix the infinity, the author uses a technique called Renormalization.
- The Analogy: Imagine you are trying to measure the weight of a feather, but your scale is broken and adds 1,000 lbs to every reading. You can't measure the feather directly. Instead, you measure the feather plus the broken scale, and then you mathematically subtract the 1,000 lbs (the "counterterm") to get the true weight.
- In the Paper: The author shows that by adding specific "counterterms" (mathematical adjustments) to the energy equation, you can cancel out the infinities.
- The "Wick Map": The paper introduces a clever tool called the Wick Map (using Bell polynomials in higher dimensions). Think of this as a "translator" that automatically knows which parts of the equation are the "broken scale" (the infinities) and removes them, leaving you with a finite, meaningful answer.

Summary of the Journey

Start: We have random noise (Gaussian variables).
Tool: We break it down into simple building blocks (Hermite polynomials).
Expansion: We build a library of all possible interactions (Wiener Chaos).
Scaling: We apply this to infinite, rough systems (Fields).
Crisis: The math explodes into infinity when we try to calculate energy in 3D.
Resolution: We use a sophisticated "subtraction" technique (Renormalization via Wick maps) to cancel the infinity and get a real, finite result.

What the paper claims (and what it doesn't):
The paper claims to provide a rigorous mathematical framework for these steps. It proves that these "renormalized" calculations work and stay finite under certain conditions. It does not claim to solve real-world engineering problems, predict stock markets, or cure diseases. It is purely a theoretical guide for mathematicians and physicists on how to handle the "infinite" nature of quantum fields and random systems using the language of probability and chaos.

Technical Summary: Topics in Gaussian Wiener Chaos Expansion

Problem Statement
The lecture notes address the mathematical foundations of Gaussian Wiener chaos expansions, extending from finite-dimensional settings to infinite-dimensional Gaussian fields, and applying these tools to the rigorous analysis of the $\Phi^4_d$ model in Euclidean Quantum Field Theory. The central problem is the construction and analysis of non-linear functionals of Gaussian fields, particularly when the underlying field possesses low regularity (e.g., white noise or the Gaussian free field in dimensions $d \geq 2$ ). In such regimes, standard pointwise operations (like taking powers of the field) are ill-defined due to divergences. The notes aim to provide a systematic framework for defining these non-linearities via Wick ordering and renormalization, and to analyze the convergence of perturbative expansions for partition functions and correlation functions.

Methodology
The exposition proceeds through a structured progression of mathematical tools:

One-Dimensional and Finite-Dimensional Foundations:
- The notes begin with the properties of Gaussian random variables and the construction of Hermite polynomials via Gram–Schmidt orthogonalization of monomials.
- Key algebraic structures are introduced, including the generating functions of Hermite polynomials, their relationship to cumulants, and the combinatorial interpretation via pairings (Isserlis' theorem/Wick's theorem).
- The Wick calculus is formalized using convolution algebras and Hopf algebra structures (specifically the antipode and coproduct), linking probabilistic operations to algebraic manipulations of power series.
- The Wiener chaos decomposition is established as an orthogonal decomposition of the $L^2$ space of functionals of Gaussian variables into homogeneous chaos subspaces. The Wiener isometry maps symmetric tensor products of the underlying Hilbert space to these chaos subspaces.
- Hypercontractivity of the Ornstein–Uhlenbeck semigroup is utilized to prove the equivalence of moments (Nelson's estimate), establishing that $L^p$ norms of chaos variables are controlled by their $L^2$ norm.
Infinite-Dimensional Gaussian Fields:
- The framework is extended to isonormal Gaussian processes on separable Hilbert spaces, specifically $L^2(\mathbb{T}^d)$ .
- Two primary fields are constructed: Gaussian white noise (a distribution of regularity $C^\alpha$ for $\alpha < -d/2$ ) and the Gaussian free field (GFF) (regularity $C^\alpha$ for $\alpha < 1 - d/2$ ).
- The notes demonstrate how to define Wick powers ( $:\phi^n:$ ) for these fields by subtracting divergent constants (variances) using scaled Hermite polynomials. This is shown to yield well-defined random variables in the $n$ -th Wiener chaos with uniform bounds on their moments.
Application to the $\Phi^4_d$ Model:
- The $\Phi^4_d$ model is defined via a Gibbs measure with energy functional containing a quartic interaction term. Since the Lebesgue measure on the space of fields is non-existent, expectations are defined relative to the Gaussian free field measure.
- Dimension $d=1$ : The model is well-defined without renormalization beyond Wick ordering. The partition function ratio is expanded into a series of Feynman diagrams (vacuum diagrams). The linked-cluster theorem is invoked to show that the cumulant expansion involves only connected diagrams. The series is shown to be an asymptotic (Gevrey-1) expansion rather than a convergent one.
- Dimension $d=2$ : The variance of the GFF diverges logarithmically. The model requires a mass renormalization counterterm ( $\beta_N$ ) dependent on the cutoff $N$ . Nelson's estimate is used to prove the uniform boundedness of the partition function ratio despite the diverging counterterm.
- Dimension $d=3$ : The divergence is stronger (linear in $N$ ). The notes introduce BPHZ renormalization (Bogoliubov–Parasiuk–Hepp–Zimmermann) to handle subdivergences. This involves the Connes–Kreimer Hopf algebra of Feynman diagrams, where the antipode map systematically extracts and subtracts divergent subgraphs. A Wick map is constructed to translate the complex diagrammatic renormalization into a polynomial transformation involving Bell polynomials (generalizing Hermite polynomials).
- Dimension $d \in (3,4)$ : The notes briefly discuss the emergence of new subdivergences as $d$ approaches 4, requiring the use of Bell polynomials in the Wick map to handle the increasing complexity of divergent substructures.

Key Contributions and Results

Unified Algebraic Framework: The notes provide a cohesive derivation of Hermite polynomials, Wick products, and chaos expansions using the language of convolution algebras and Hopf algebras, clarifying the combinatorial structure of Gaussian moments.
Rigorous Construction of Wick Powers: A detailed proof is provided for the existence and uniform moment bounds of Wick powers of the Gaussian free field in dimensions $d=1, 2, 3$ , establishing the necessary regularity for defining non-linear interactions.
Renormalization of $\Phi^4_3$ : The text outlines a proof of the existence of the $\Phi^4_3$ model (in the sense of bounded perturbative expansions) by combining the linked-cluster theorem with BPHZ renormalization. It explicitly constructs the necessary mass and energy counterterms ( $\beta_N, \gamma_N$ ) to cancel divergences.
Diagrammatic Analysis: The notes detail the degree of divergence for Feynman diagrams using Hepp sectors and provide a criterion for non-divergence based on subgraph degrees.
Commutative Diagrams: A key result is the demonstration of a commutative diagram linking the algebraic renormalization of Feynman diagrams (via the Connes–Kreimer antipode) to the probabilistic renormalization of polynomial functionals (via the Wick map and Bell polynomials).

Significance
The paper serves as a pedagogical and technical bridge between classical probability theory (Gaussian variables, Hermite polynomials) and modern constructive quantum field theory. It claims significance by:

Offering a self-contained introduction to the Wiener chaos expansion and its isometric properties, which are fundamental to Malliavin calculus and stochastic analysis.
Providing a clear, step-by-step derivation of the renormalization procedure for the $\Phi^4$ model across dimensions, highlighting the transition from trivial cases ( $d=1$ ) to cases requiring sophisticated algebraic renormalization ( $d=3$ ).
Explicitly connecting the linked-cluster theorem and BPHZ renormalization, showing how the combinatorial structure of connected diagrams ensures the cancellation of divergences in the partition function.
Presenting the Wick map as a unifying tool that simplifies the handling of subdivergences, replacing complex diagrammatic subtraction with algebraic operations on polynomials.

The notes emphasize that while the perturbative expansions for the partition functions are generally asymptotic (non-convergent), they provide a rigorous framework for defining the model and computing physical quantities order-by-order. The work relies on and synthesizes results from standard references in the field (e.g., Nualart, Hairer) to present a coherent narrative on the construction of Gaussian fields and their non-linear interactions.