🔢 mathematics

Generic regularity of equilibrium measures for the logarithmic potential with external fields

By leveraging the connection between minimizing measures and thin obstacle problems, this paper proves the conjecture that external potentials are generically "off-critical" (or "regular") for equilibrium measures in logarithmic potential theory and $\beta$ -models.

Original authors: Giacomo Colombo, Alessio Figalli

Published 2026-08-11

📖 8 min read🧠 Deep dive

CC BY 4.0

Original authors: Giacomo Colombo, Alessio Figalli

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ 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 a crowded dance floor where everyone wants to find the perfect spot to stand. If the dancers are repelled by each other—like magnets with the same pole—they will naturally spread out to avoid bumping into one another. But they also have to deal with the shape of the room itself; maybe there are walls pushing them in, or a spotlight that makes them want to huddle in the center. In the world of physics and mathematics, this scenario is modeled by something called "Coulomb gases" or "beta-ensembles." Scientists study how these particles arrange themselves to reach a state of perfect balance, or "equilibrium." The big question is: what does that final arrangement look like? Does it form a smooth, predictable pattern, or does it get messy and chaotic?

For a long time, mathematicians have suspected that if you pick a "typical" or "generic" set of rules for the room (the external potential), the dancers will arrange themselves in a very specific, orderly way. They thought that, usually, the crowd would settle into a few distinct, separate groups, and the density of people at the edges of these groups would fade away smoothly, like a gentle slope. However, they also knew that if you picked a very strange or "weird" set of rules, this smooth pattern could break down. The big mystery was whether these smooth, orderly patterns were just a lucky fluke for special cases, or if they were actually the rule for almost every possible scenario. This paper steps in to settle that debate, proving that yes, for a vast majority of situations, the dancers do indeed form that beautiful, predictable pattern.

The Dance of the Equilibrium Measures

In this paper, authors Giacomo Colombo and Alessio Figalli tackle a famous conjecture in the theory of how particles interact. They are looking at a specific type of energy equation that describes how a cloud of particles (like electrons or charged balls) settles down when they repel each other but are also pushed around by an external force, like a magnetic field or a container wall.

The central idea is the "equilibrium measure." Think of this as the final snapshot of the dance floor once everyone has stopped moving and found their perfect spot. For a long time, mathematicians have assumed that for most "generic" external forces, this final snapshot looks very nice and regular. Specifically, they assumed the particles would cluster into a few separate islands, and the density of particles at the edge of these islands would drop off like a square root (a smooth curve that gets steeper as it hits the edge). This is called the "regularity" or "off-criticality" assumption.

The problem is that while this looks true for simple, perfectly smooth forces (like a perfect parabola), it was known to fail for some very weird, jagged forces. The big question was: Is the "nice" behavior just a rare exception, or is it actually the norm? The authors prove that it is the norm. They show that if you take a wide class of forces (specifically, those that are twice differentiable with a bit of extra smoothness, known as $C^{2,\alpha}$ ), and you slightly tweak them, you will almost always get that nice, regular pattern. In fact, they prove that the set of forces that produce this regular pattern is "open and dense," meaning if you pick a random force from this class, it is almost guaranteed to be regular, and if you have a weird one, you can nudge it just a tiny bit to make it regular.

How They Solved the Puzzle

To prove this, the authors used a clever trick involving a different branch of mathematics called "obstacle problems." Imagine you have a rubber sheet stretched over a frame, and you push a bumpy object (the "obstacle") up from underneath. The sheet will drape over the object, touching it in some places and floating above it in others. The line where the sheet touches the object is called the "free boundary."

The authors realized that the problem of finding the equilibrium position of the particles is mathematically identical to finding the shape of this rubber sheet. The particles' density is related to how much the sheet bends. By translating the particle problem into this "thin obstacle problem," they could use powerful tools from the study of partial differential equations (PDEs).

However, there was a catch. The known mathematical tools for proving that the "free boundary" is smooth only work if you have a family of solutions that change in a very specific, monotone way (like a sheet that gets pushed up higher and higher without ever dipping down). The authors had to find a way to create such a family. They did this by varying the "mass constraint"—essentially, changing how many particles are on the dance floor. They showed that as they increased the number of particles, the solution changed in a strictly monotone way. This allowed them to apply the existing theory of "generic regularity" for obstacle problems to their specific case.

The Discrete Case and the "Two-Phase" Twist

The paper also tackles a "discrete" version of the problem, which is relevant for things like orthogonal polynomials and discrete particle systems. In this version, there is an extra rule: the density of particles cannot exceed a certain limit (like a maximum crowd density). This turns the problem into a "two-phase thin obstacle problem." Imagine the rubber sheet now has two different obstacles: one pushing up from below (the floor) and one pushing down from above (a ceiling). The sheet has to navigate between them.

This is much harder because, in this two-phase scenario, it wasn't known if the "phases" (the regions where the sheet touches the floor vs. the ceiling) could touch each other in a messy way. The authors proved a crucial new result: in this specific setup, the phases with opposite signs cannot touch. This means the "messy" middle ground where things get weird simply doesn't exist. This discovery not only solved their main problem but also fills a gap in the literature regarding fracture mechanics models.

What They Found

The main finding is a resounding confirmation of the "generic regularity" conjecture. For the continuous case (standard particle clouds), they proved that for almost every scaling of a potential function, the equilibrium measure is regular. This means the particles form a finite number of disjoint intervals, and the density of particles at the edges of these intervals behaves like a square root, just as the "off-critical" assumption predicted.

For the discrete case (where particle density is capped), they proved the same thing: for almost every scaling of the potential and almost every choice of mass constraints, the minimizing measure is regular. The density will be smooth inside the allowed regions and will vanish or hit the cap in a predictable, square-root fashion at the boundaries.

They also showed that if the external force is very smooth (specifically, if it belongs to a certain class of functions denoted as $C^{k+1/2+\beta}$ ), then the function describing the particle density is also very smooth (specifically, $C^{k-1,\beta}$ ). This establishes a direct link between the smoothness of the environment and the smoothness of the particle arrangement.

What They Didn't Find (and What They Ruled Out)

It is important to note what this paper does not claim. They do not say that every possible potential leads to a regular pattern. They explicitly acknowledge that there are "bad" potentials where the pattern breaks down (for example, where the support of the measure is not a finite union of intervals). Their result is that these "bad" potentials are rare; they are not the norm. The set of "bad" potentials is so small that if you were to pick a potential at random from the class they studied, the chance of picking a "bad" one is zero.

Furthermore, they do not claim to have solved the problem for all types of potentials. Their proof relies on the potential being at least twice differentiable with some Hölder continuity ( $C^{2,\alpha}$ ). They do not address potentials that are less smooth than this. They also do not claim to have solved the problem for all dimensions in the context of Riesz potentials (a more general type of interaction), though they mention their strategy likely works for low dimensions.

The Bottom Line

Colombo and Figalli have successfully bridged the gap between the theory of Coulomb gases and the theory of obstacle problems. By showing that the "regular" behavior of equilibrium measures is not just a lucky accident for special cases but is actually the generic, expected outcome for a wide class of smooth potentials, they have solidified a foundational assumption in the field. Their work confirms that nature, when left to its own devices with smooth rules, tends to arrange itself in beautifully predictable, orderly patterns. The "messy" cases are the outliers, not the rule.

Technical Summary: Generic Regularity of Equilibrium Measures for the Logarithmic Potential with External Fields

1. Problem Statement

The paper addresses a central question in the theory of Coulomb gases and the mean-field limit of $\beta$ -ensembles: the behavior of probability measures $\mu_V$ that minimize the energy functional
$E_V(\mu) := 2 \int_{\mathbb{R}} V(x) \, d\mu(x) - \int_{\mathbb{R}} \int_{\mathbb{R}} \log |x - y| \, d\mu(x) \, d\mu(y).$
A standard assumption in the field, often termed "off-criticality" or "regularity," posits that for a generic potential $V$ , the minimizing measure $\mu_V$ is supported on a finite union of disjoint compact intervals. Furthermore, within each interval $[a, b]$ , the density of the measure is expected to take the form:
$\frac{d\mu_V}{dx} = \frac{Q_V(x)}{\sqrt{(x-a)(b-x)}},$
where $Q_V$ is a strictly positive and bounded function.

While this property is known to hold for analytic potentials (established in [KM00]) and is conjectured to be true generically for broader classes, it is known to fail for arbitrary potentials. Prior to this work, the generic validity of this regularity assumption was unknown for non-analytic potentials, specifically within the $C^{2,\alpha}$ class. The paper also extends this inquiry to discrete $\beta$ -models, where the measure's density is constrained by an upper bound $\theta$ , leading to a "two-phase" scenario where the density vanishes at the lower bound and saturates at the upper bound.

2. Methodology

The authors employ a strategy that bridges the theory of minimizing measures with the regularity theory of free boundary problems, specifically the thin obstacle problem.

Connection to Obstacle Problems

The core methodological step involves associating the minimizing measure $\mu_V$ with a solution $u_V$ to a thin obstacle problem in $\mathbb{R}^2$ .

Continuous Case: The function $u_V(x) = -(\log|\cdot| * \mu_V)(x) - C_V$ solves a variational inequality where the obstacle is $-V$ . The support of the measure corresponds to the contact set $\{u_V = -V\}$ .
Discrete Case: When an upper bound $\theta$ is imposed on the density, the problem transforms into a two-phase thin obstacle problem. The solution $u$ satisfies $\Delta u = 0$ outside the support, while inside the support, the normal derivative $\partial_2 u$ is constrained between $0$ and $-\pi\theta$ , switching values depending on whether the density is below $\theta$ or saturated.

Generic Regularity via Monotone Families

To prove that regular potentials are dense (and thus generic), the authors utilize a perturbation argument inspired by [KM00] but adapted to non-analytic settings:

Scaling and Mass Rescaling: Instead of directly perturbing the potential $V$ , the authors consider a family of scaled potentials $V_{s,\gamma}(x) = s^{-1}V(s^\gamma x)$ .
Monotonicity: They establish that the associated solutions $u_s$ to the obstacle problems form a strictly monotone family with respect to the scaling parameter $s$ . Specifically, $u_s$ is strictly decreasing in $s$ .
Application of Known Theory: This monotonicity allows the application of existing generic regularity results for the thin obstacle problem (specifically from [FRTL23]), which assert that for a strictly monotone family of solutions, the free boundary is regular for almost every parameter value.
Discrete Case Novelty: For the discrete case with density constraints, the authors face the lack of generic regularity theory for the two-phase thin obstacle problem. They overcome this by proving a key auxiliary result (Appendix B): in the two-phase problem, phases with opposite signs cannot touch. This result, which fills a gap in [CCF20], ensures that the free boundary structure remains manageable enough to apply the generic regularity arguments.

Stability and Higher Regularity

The authors also prove the stability of regular potentials under $C^{2,\alpha}$ perturbations and establish that higher regularity of the potential $V$ (specifically $C^{k+1/2+\beta}$ ) yields higher regularity of the density function $Q_V$ (specifically $C^{k-1,\beta}$ ). This is achieved through a detailed analysis of the Taylor expansion of the solution near the free boundary.

3. Key Contributions and Results

Main Theorems

Theorem 1.1 / 2.3 (Continuous Case): For potentials $V \in C^{2,\alpha}_{\text{loc}}(\mathbb{R})$ that grow faster than $\log|x|$ at infinity, the set of potentials satisfying the regularity assumption (A) is open and dense. Specifically, for the family $V_{s,\gamma}(x) = s^{-1}V(s^\gamma x)$ , the potential is regular for almost every $s > 0$ .
Theorem 1.3 / 2.6 (Discrete Case): For discrete $\beta$ -ensembles with density constraints, the set of potentials satisfying the two-phase regularity assumption (B) is open and dense in the appropriate topology. This holds both without and with additional mass constraints on specific intervals.
Regularity of the Density: If the potential $V$ is of class $C^{k+1/2+\beta}$ , the function $Q_V$ describing the density is of class $C^{k-1,\beta}$ .

Technical Advances

Two-Phase Obstacle Problem: The paper provides a rigorous proof that in the two-phase thin obstacle problem, the contact sets for the two phases (where the density is 0 and where it is $\theta$ ) cannot touch. This is a fundamental tool for studying free boundaries in cohesive zone models for fracture mechanics.
Extension to Non-Analytic Potentials: The results generalize the known generic regularity from analytic potentials to $C^{2,\alpha}$ potentials, a significant step as many physical models involve non-analytic external fields.
Riesz Potentials: The authors note in Appendix D that their strategy extends to Riesz potentials (energies with kernel $|x-y|^{-\sigma}$ ) in low dimensions, proving the generic validity of the corresponding regularity assumption (A $\sigma$ ).

4. Significance and Claims

The paper claims to resolve a long-standing conjecture in the theory of Coulomb gases and $\beta$ -ensembles regarding the generic nature of equilibrium measures. By establishing that "off-critical" or "regular" potentials are dense in the space of $C^{2,\alpha}$ potentials, the authors justify the widespread use of the semicircle law and its generalizations (form (1.2)) in the study of these systems, provided the external potential is not specifically tuned to be singular.

The work is significant for two main communities:

Probability and Statistical Physics: It provides a rigorous foundation for the "generic" behavior of minimizing measures in mean-field limits, validating assumptions previously taken for granted or only proven for analytic cases.
Partial Differential Equations (Free Boundaries): It introduces new tools for the two-phase thin obstacle problem, specifically the non-touching result for opposite phases, which has independent interest in fracture mechanics and cohesive zone models.

The authors emphasize that their results are structural and rely on the connection between energy minimization and obstacle problems, rather than specific properties of the logarithmic kernel, suggesting the robustness of the approach for related energy functionals like Riesz potentials.