The height gap of planar Brownian motion is $\frac{5}{\pi}$

This paper demonstrates that the occupation measure of planar Brownian motion exhibits a constant height gap of $5/\pi$ across its outer boundary, a result derived by leveraging the expected area of the domain delimited by a Brownian bridge and drawing parallels to height gap properties in the Gaussian free field and Brownian loop soup.

The Gist

Imagine you are watching a drunk person wandering aimlessly through a city park. This person is our Planar Brownian Motion. They start at a fountain, wander around randomly, bumping into trees, benches, and other people, and eventually stop after a certain amount of time.

Now, imagine you draw a line around the entire area this person has visited. This line is the Outer Boundary. It's a very messy, jagged, fractal line (like a coastline seen from space) that encloses everything the walker touched.

This paper is about a very specific, surprising property of the "time" the walker spent in different parts of the park, right next to that messy boundary line.

The Core Discovery: The "Height Gap"

Think of the time the walker spent in a specific spot as the height of a pile of sand at that location.

• If the walker spent a lot of time there, the sand pile is high.
• If they barely stopped, the sand pile is low.

The authors discovered a magical rule about the edge of the park (the outer boundary):

1. Outside the boundary: If you stand just outside the fence and look at the sand, the height is zero. The walker never went there, so there is no sand.
2. Inside the boundary: If you stand just inside the fence, right next to it, the sand pile suddenly jumps up to a specific, constant height.

The paper calculates exactly what that height is. It turns out to be 5 divided by Pi ($5/\pi$).

The Analogy: Imagine the boundary is a cliff. On the outside, it's a flat, empty plain (height 0). On the inside, right at the edge of the cliff, there is a sudden, perfectly flat plateau that is exactly $5/\pi$ meters high. No matter where you stand along the jagged, winding edge of the park, if you look just inside, the "time sand" is always at that same height.

Why is this surprising?

Usually, when you have a random, messy path like this, you expect the "time spent" to vary wildly. Some parts of the boundary might be touched often, others rarely. You would expect the sand height to be bumpy and irregular.

But this paper proves that nature smooths it out. Even though the path is chaotic, the average amount of time spent right next to the edge is perfectly constant. It's like a law of physics for random walks.

How did they figure this out?

The authors didn't just guess; they used a clever trick involving mirrors and maps.

1. The Magic Map: They imagined taking the messy, jagged shape of the park and stretching it out until it became a perfect, smooth circle (like a pizza). In math, this is called a "conformal map."
2. The Symmetry: Once the shape was a perfect circle, the problem became much easier because a circle is perfectly symmetrical. The authors realized that if the shape is a circle, the "time sand" must be distributed evenly around the edge.
3. The Calculation: They used a known result from a different field of math (about the average area of a shape made by a random bridge) to calculate the exact amount of "sand" needed to fill that circle.
4. The Result: When they did the math, the numbers canceled out beautifully to leave exactly $5/\pi$.

The Big Picture Connection

This result isn't just about a drunk walker in a park. It connects to some of the most advanced theories in modern physics and mathematics:

• The Gaussian Free Field: This is a mathematical model used to describe things like magnetic fields or the surface of a liquid. It also has "height gaps" (sudden jumps in value) across certain curves.
• The Limit: The authors show that their result is the "zero intensity" limit of a more complex system called a "Brownian loop soup" (imagine a pot full of many random loops). As you reduce the number of loops to just one, the complex physics simplifies down to this clean, constant gap of $5/\pi$.

Summary in One Sentence

This paper proves that if you look at the very edge of a random path in a 2D plane, the amount of time the path spends right next to that edge is always a constant, predictable value of $5/\pi$, creating a perfect "step" in height between the inside and the outside of the path.

It's a beautiful example of how, even in total chaos and randomness, there are hidden, perfect constants waiting to be discovered.

Technical Summary

Here is a detailed technical summary of the paper "The height gap of planar Brownian motion is $5/\pi$" by Antoine Jego, Titus Lupu, and Wei Qian.

1. Problem Statement and Context

The paper addresses the geometric and probabilistic properties of the occupation measure of planar Brownian motion, specifically focusing on its behavior near the outer boundary of the trajectory.

• Background: The outer boundary of a planar Brownian motion is known to be a fractal curve with Hausdorff dimension $4/3$, distributed as an$\text{SLE}_{8/3}$ (Schramm-Loewner Evolution) curve.
• The Phenomenon: The authors investigate the "height gap" of the occupation measure across this boundary. The occupation measure, denoted $L_x(P)dx$, represents the time the Brownian path $P$ spends in a region.
• Analogy: The problem is motivated by celebrated results concerning the Gaussian Free Field (GFF) and Schramm-Loewner Evolution (SLE). In the GFF context (Miller-Sheffield, Schramm-Sheffield), the field exhibits a constant height jump across $\text{SLE}_4$ curves. Similarly, for a Brownian loop soup with intensity $\theta \in (0, 1/2]$, a conformally invariant field $h_\theta$ exhibits a height gap across $\text{CLE}_\kappa$ curves.
• The Specific Question: What is the value of the height gap for the occupation measure of a single planar Brownian motion (or loop) across its outer boundary? This corresponds to the limit $\theta \to 0^+$ of the loop soup results, where the field becomes the local time of a single loop.

2. Methodology

The authors employ a combination of conformal invariance, loop measure theory, and precise asymptotic analysis.

A. Conformal Mapping and Conditioning

• Loop Measure: The authors work with the Brownian loop measure $\mu_{\text{loop}}$, which is conformally invariant. They consider loops conditioned on their outer boundary being a specific curve $\gamma$.
• Wired Measure: By conformally mapping the interior of the outer boundary $\gamma$ to the unit disk $\mathbb{D}$, they define a "wired" probability measure $\mathbb{P}^{\text{wired}}_{\mathbb{D}, \partial\mathbb{D}}$. This measure represents a Brownian loop conditioned to have its outer boundary coincide with the unit circle $\partial\mathbb{D}$.
• Symmetry: Due to the conformal invariance of the loop measure, the expectation of the occupation measure under this wired law is invariant under conformal automorphisms of the disk.

B. Decomposition into Excursions

• Point Process of Excursions: A key technical tool is the decomposition of the wired Brownian loop into a locally finite point process of Brownian excursions attached to the boundary. This relies on the Brownian bubble measure and results from previous work (Jego, Lupu, Qian [6]) regarding conformal restriction.
• Correlation Functions: The authors define and compute "correlation functions" (densities of the occupation measure).
  • First Moment: The density of the expected occupation measure $\langle L_x(P) \rangle$.
  • Second Moment: The density of the second moment $\langle L_x(P)L_y(P) \rangle$, which involves Green's functions and Poisson kernels.

C. Asymptotic Analysis and Decorrelation

• Concentration: To prove the height gap is constant, they must show that the occupation measure concentrates around its expectation when integrated against test functions supported near the boundary.
• Decorrelation: A major technical hurdle is proving that occupation measures of two small sets near the boundary, but at a macroscopic distance from each other, become asymptotically uncorrelated as the sets shrink. This involves a partial exploration of the outer boundary and analyzing the independence of excursions.

3. Key Contributions and Results

Main Theorem (Theorem 1.1 & 2.3)

The central result establishes that for a planar Brownian motion (or loop) $P$, the occupation measure exhibits a constant height gap of $5/\pi$ across its outer boundary.

• Formal Statement: Let $\gamma$ be the outer boundary. For a sequence of test functions $f_\varepsilon$ concentrating on the boundary from the inside (with $\int f_\varepsilon = 1$), the integral of the occupation measure converges in $L^1$:
  $$\int f_\varepsilon(x) L_x(P) dx \xrightarrow{\varepsilon \to 0} \frac{5}{\pi}$$
• Interpretation: Approaching the boundary from the inside (the bounded component), the local time density is $5/\pi$. Approaching from the *outside*, it is$0$.

Determination of the Constant

The value $5/\pi$ is derived through a specific calculation involving the expected area of the domain delimited by the outer boundary:

1. Symmetry Argument: The authors prove that under the wired measure, the expected occupation density is constant, say $\lambda_0$.
2. Area Relation: They relate $\lambda_0$ to the expected area of the interior of the loop. Specifically, they use the identity:
   $$\lambda_0 \times \mathbb{E}[\text{Area}(\text{int}(\gamma))] = \mathbb{E}[\text{Duration}(P)]$$
   (in a renormalized sense).
3. Garban-Trujillo Ferreras Result: They rely on a result by Garban and Trujillo Ferreras [3], which states that the expected area of the domain delimited by the outer boundary of a Brownian bridge of duration 1 is $\pi/5$.
4. Calculation: Combining the duration expectation (normalized to 1) and the area expectation ($\pi/5$), they solve for $\lambda_0$:
   $$\lambda_0 \cdot \frac{\pi}{5} = 1 \implies \lambda_0 = \frac{5}{\pi}$$

Technical Lemmas

• Lemma 4.9: Proves the constancy of the first moment density $\langle L_x(P) \rangle$ in the unit disk.
• Lemma 4.14: Proves the decorrelation of the second moment for points separated by a macroscopic distance, ensuring the variance vanishes in the limit.

4. Significance and Implications

• Resolution of the $\theta \to 0$ Limit: The paper provides the explicit value of the height gap for the limiting case of a Brownian loop soup with intensity $\theta \to 0$. While the general case for $\theta \in (0, 1/2]$ was studied in [6], the exact value was only known for $\theta=1/2$ (GFF, gap $2\lambda$). This paper fills the gap for$\theta \to 0$.
• Universality of Height Gaps: It reinforces the deep connection between the geometry of random curves (SLE) and the behavior of associated random fields (occupation measures, GFF). The constant $5/\pi$ is a universal constant for planar Brownian motion.
• Non-Poissonian Nature: The authors note that unlike the critical loop soup ($\theta=1/2$) where boundary excursions form a Poisson point process, the excursions induced by a single Brownian loop do not form a Poisson process. Despite this intricate dependence structure, the boundary occupation time remains constant, which is a non-trivial result.
• Methodological Advance: The techniques developed for decomposing the wired loop into excursions and proving decorrelation estimates for non-Poissonian point processes of Brownian excursions are significant contributions to the field of stochastic analysis and SLE theory.

In summary, the paper rigorously proves that the local time of a planar Brownian motion accumulates at a rate of $5/\pi$ along its outer boundary, bridging the gap between loop soup theories and the classical properties of Brownian motion.