🔢 mathematics

One-phase Free Boundary Problems on RCD Metric Measure Spaces

This paper establishes the existence and local Lipschitz regularity of solutions to a vector-valued one-phase Bernoulli-type free boundary problem on non-collapsed $RCD(K,N)$ metric measure spaces, and proves that the resulting free boundary is an $(N-1)$ -dimensional topological manifold outside a singular set of Hausdorff dimension at most $N-3$ .

Original authors: Chung-Kwong Chan, Hui-Chun Zhang, Xi-Ping Zhu

Published 2026-04-22

📖 5 min read🧠 Deep dive

CC0 1.0

Original authors: Chung-Kwong Chan, Hui-Chun Zhang, Xi-Ping Zhu

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ 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 an architect trying to design the most efficient shape for a new city. You have a piece of land, and you need to decide exactly where to build houses (the "active" zone) and where to leave it as empty parkland (the "inactive" zone).

Your goal is to minimize a specific "cost." The cost has two parts:

Construction Cost: The effort to build the walls of the houses (represented by the gradient of your building plan).
Land Cost: The price of the land you actually occupy (represented by the area where houses exist).

The tricky part is that you don't know the boundary between the houses and the park beforehand. This boundary is the Free Boundary. It's "free" because it's not drawn on a map yet; it emerges naturally from your attempt to save money.

This paper is about solving this architectural puzzle, but with a massive twist: The ground itself isn't flat like a normal city.

The Setting: A "Weird" World

Usually, mathematicians solve this problem on flat, Euclidean space (like a standard sheet of graph paper). But in this paper, the authors (Chan, Zhang, and Zhu) are working in a world called RCD Spaces.

Think of an RCD space as a terrain that might be:

Crumpled: Like a crumpled piece of paper.
Singular: Like a pyramid tip or a cone where the geometry gets weird.
Curved: Like the surface of a sphere or a saddle.

These spaces satisfy a condition called RCD(K, N), which is a fancy way of saying, "This world has a specific type of curvature (like gravity) and a specific dimension, even if it looks broken or crumpled in places."

The Big Questions

The authors wanted to know:

Does a solution exist? Can we always find a "best" way to build our city on this weird terrain?
Is the plan smooth? If we zoom in on the boundary between the houses and the park, is it a nice, straight line (or smooth curve), or is it jagged and messy?
How messy can it get? If the boundary is messy, how big is the mess?

The Journey of Discovery

1. Finding the Blueprint (Existence)

First, they proved that no matter how weird the terrain is (as long as it's "non-collapsed," meaning it doesn't flatten out into nothingness), there is always a perfect blueprint. You can always find the optimal way to split the land between "built" and "empty."

2. The Smoothness Guarantee (Lipschitz Continuity)

In the flat world, we know the boundary is usually smooth. But on a crumpled, singular terrain, things get tricky.

The Metaphor: Imagine trying to draw a straight line on a crumpled piece of paper. It's hard.
The Result: The authors proved that even on this crumpled paper, the boundary between the houses and the park is locally Lipschitz.
Translation: This means the boundary is "well-behaved." It doesn't have infinite spikes or wild oscillations. If you zoom in close enough, it looks like a straight line, even if the ground underneath is bumpy. They achieved this by using a special "gradient estimate" (a tool to measure how steep the building plan is) adapted for these weird spaces.

3. The Shape of the Boundary (Regularity)

This is the most exciting part. They asked: How smooth is the boundary, and where does it break?

The Regular Part: Most of the boundary is a beautiful, smooth $(N-1)$ -dimensional surface. If you are in a 3D world, the boundary is a 2D sheet (like a wall). If you are in a 4D world, it's a 3D volume.
The Singular Part (The "Glitches"): Sometimes, the boundary hits a "singularity" in the ground (like the tip of a cone). At these points, the boundary might look weird or jagged.
The Dimension Estimate: The authors proved that these "glitchy" points are very rare.
- In a 3D world, the bad points are just isolated dots (0-dimensional).
- In a 4D world, the bad points form a line (1-dimensional).
- The Rule: The "bad" set is always at least 3 dimensions smaller than the space itself.
- Analogy: If you are building a city in a 100-dimensional universe, the places where your city limits get messy are so small they are practically invisible (only 97 dimensions).

Why This Matters

This isn't just about abstract math.

Real-World Geometry: Many real-world phenomena (like the shape of the universe, or the structure of complex networks) can be modeled as these "crumpled" spaces.
Robustness: The authors showed that the rules of optimization (finding the best shape) are robust. They hold true even when the underlying geometry is broken or singular.
Sharpness: They even gave an example (a "double triangle" glued together) to show that their estimate is the best possible. You can't say the bad points are even smaller than they are; nature forces them to be at least that big.

Summary in a Nutshell

The authors took a classic problem about finding the best shape for a boundary and solved it on a terrain that can be crumpled, curved, or broken. They proved:

A solution always exists.
The boundary is mostly smooth.
The "messy" parts are incredibly small (at least 3 dimensions smaller than the space).

It's like proving that even if you try to build a perfect city on a pile of rubble, the city limits will still be mostly straight lines, and the jagged bits will be so tiny they barely matter.

1. Problem Statement

The paper investigates the vector-valued one-phase Bernoulli-type free boundary problem within the context of RCD(K, N) metric measure spaces. These spaces generalize Riemannian manifolds with Ricci curvature bounded below by $K$ and dimension bounded above by $N$ , encompassing non-smooth spaces such as Ricci limit spaces and Alexandrov spaces.

The problem involves minimizing the energy functional:
$J_Q(u) = \int_{\Omega} \left( |\nabla u|^2 + Q \chi_{\{|u|>0\}} \right) d\mu$
where:

$(X, d, \mu)$ is an RCD(K, N) space.
$\Omega \subset X$ is a bounded domain.
$u: \Omega \to [0, \infty)^m$ is a vector-valued map ( $m \geq 1$ ).
$Q \in L^\infty(\Omega)$ is a positive weight function satisfying $0 < Q_{\min} \leq Q(x) \leq Q_{\max} < \infty$ .
The set $\Omega_u = \{x \in \Omega : |u(x)| > 0\}$ is the free boundary region, and its boundary $\partial \Omega_u \cap \Omega$ is the free boundary.

The goal is to establish the existence, regularity, and geometric structure of the minimizers and their free boundaries in this non-smooth setting.

2. Methodology

The authors adapt classical techniques from Euclidean free boundary theory (specifically the work of Alt-Caffarelli, Caffarelli-Shahgholian-Yeressian, and Mazzoleni-Terracini-Velichkov) to the synthetic curvature setting. Key methodological pillars include:

Non-Collapsed Assumption: The analysis primarily assumes the space is non-collapsed (ncRCD), meaning the measure $\mu$ is the $N$ -dimensional Hausdorff measure $\mathcal{H}^N$ . This is crucial for obtaining sharp dimension estimates and Lipschitz regularity.
Gradient Estimates and Mean Value Inequalities: Instead of relying on the Poisson formula (which is unavailable in general RCD spaces), the authors utilize Cheng-Yau gradient estimates for harmonic functions and derive a new mean value inequality (Lemma 5.3) tailored to the RCD setting.
Blow-up Analysis: The authors employ a compactness argument using pointed measured Gromov-Hausdorff (pmGH) convergence. By rescaling the space and the minimizer, they analyze the behavior of the free boundary at infinitesimal scales, leading to tangent cones and homogeneous global minimizers.
Dimension Reduction: A variant of the classical dimension reduction argument is used to estimate the size of the singular set. This involves iterating blow-ups to show that singularities cannot persist beyond a certain codimension.
Weiss-Type Monotonicity: The authors establish a Weiss-type monotonicity formula for the energy density on metric cones, which is essential for characterizing regular points and proving the homogeneity of blow-up limits.

3. Key Contributions and Results

A. Existence and Regularity of Minimizers

Existence: The paper proves the existence of a minimizer for the functional $J_Q$ under Dirichlet boundary conditions (Proposition 1.3).
Lipschitz Continuity: A major contribution is proving that local minimizers are locally Lipschitz continuous (Theorem 1.4). This is non-trivial in RCD spaces because standard Euclidean proofs rely on tools (like the Poisson formula or harmonicity of gradients) that fail in singular spaces. The proof relies on the non-collapsed condition and Cheng-Yau estimates.
Subharmonicity: The components of the minimizer are shown to be subharmonic in the distributional sense, and harmonic in the positivity set $\{|u|>0\}$ .

B. Structure of the Free Boundary

Local Finiteness of Perimeter: The free boundary $\partial \{|u|>0\} \cap \Omega$ is shown to be a set of locally finite perimeter (Theorem 1.6).
Euler-Lagrange Equation: The authors derive the distributional Euler-Lagrange equation:
$\Delta u_i = q_i \cdot \mathcal{H}^{N-1} \llcorner (\partial \{|u|>0\} \cap \Omega)$
where $q_i$ are non-negative Borel functions satisfying the algebraic constraint $\sum_{i=1}^m q_i^2(x) = Q(x)$ for $\mathcal{H}^{N-1}$ -almost every point on the free boundary.
Regularity and Singular Set Dimension:
- The free boundary is shown to be an $(N-1)$ -dimensional topological manifold away from a singular set.
- Main Theorem (Theorem 1.9): The singular set $S_{\Omega_u}$ of the free boundary has Hausdorff dimension $\dim_H(S_{\Omega_u}) \leq N - 3$ .
- This bound is proven to be sharp via a counter-example involving the product of a line and a doubled equilateral triangle (Example 1.8), where the singular set has dimension $N-3$ .
- If $N=3$ , the singular set consists of at most isolated points.

C. Compactness and Blow-up Limits

The paper establishes a compactness theorem (Theorem 7.1) for sequences of minimizers on pmGH-converging ncRCD spaces. This ensures that limits of minimizers are themselves minimizers on the limit space.
Classification of Blow-ups: It is shown that blow-up limits at regular points are 1-homogeneous global minimizers on Euclidean space $\mathbb{R}^N$ , while blow-ups at singular points lead to lower-dimensional structures.

4. Significance

Extension to Non-Smooth Spaces: This work significantly extends the classical theory of Bernoulli-type free boundary problems from smooth Riemannian manifolds and Euclidean spaces to the broad class of RCD spaces, which include limits of manifolds with lower Ricci curvature bounds and singular spaces like Alexandrov spaces.
Overcoming Analytic Obstacles: The paper successfully navigates the lack of smooth structure by developing new analytic tools (mean value inequalities, gradient estimates) that do not depend on the existence of a smooth Riemannian metric.
Sharp Dimension Estimates: The result $\dim_H(S) \leq N-3$ is the optimal generalization of the Euclidean result (where the singular set has codimension at least 3). It highlights that the singularities of the free boundary are strictly controlled by the dimension of the ambient space, even in the presence of space singularities, provided the space is non-collapsed.
Vector-Valued Case: The results apply to the vector-valued case ( $m \geq 2$ ), which is more complex than the scalar case due to the interaction between components, generalizing recent Euclidean results by Caffarelli-Shahgholian-Yeressian and others.

In summary, this paper provides a comprehensive regularity theory for one-phase free boundary problems in a synthetic geometric setting, bridging the gap between geometric analysis on singular spaces and free boundary problems.