Spectral rigidity among ellipses, Bialy's conjecture and local extrema of Mather's beta function

This paper proves Bialy's conjecture by demonstrating that two ellipses are identical if their Mather beta functions coincide at one or two nonzero rotation numbers (with the latter condition requiring equal perimeters for a single rotation number), and discusses the implications of this result for local extremizers of the beta function.

The Gist

Imagine you have a magical drum. If you hit it, it sings a specific song made of pure tones (frequencies). In the 1960s, a famous mathematician named Mark Kac asked a mind-bending question: "Can you hear the shape of a drum?" In other words, if you only listen to the song the drum makes, can you figure out exactly what shape the drum is? Is it a circle? A square? A weird blob?

For a long time, mathematicians thought the answer was "maybe, but it's really hard." This paper by Corentin Fierobe tackles a specific, slightly different version of this puzzle, but instead of sound waves, we are looking at billiard balls.

The Billiard Table Analogy

Imagine a billiard table with a perfectly smooth, curved edge. You shoot a ball, and it bounces around forever.

• If the table is a perfect circle, the ball bounces in a very predictable, symmetrical way.
• If the table is an ellipse (a stretched circle, like a rugby ball), the ball still bounces in a very special, orderly way.
• If the table is a weird, lumpy shape, the ball's path becomes chaotic and unpredictable.

Mathematicians have a special tool called Mather's Beta Function (let's call it the "Bouncing Score"). This score tells us the maximum distance a ball can travel in one loop if it bounces in a specific pattern. The "pattern" is defined by a rotation number, which is just a fancy way of saying "how many times the ball goes around the table before it repeats its path."

The Big Question: The "Two-Point" Test

The paper addresses a conjecture (a guess) made by a mathematician named Bialy. Here is the scenario:

Imagine you have two mystery billiard tables. You don't know if they are circles, ellipses, or something else. You are told they are both ellipses.

1. You shoot a ball on Table A and Table B with a specific pattern (Rotation Number 1). You measure the "Bouncing Score" for both. They are identical.
2. You shoot a ball with a different pattern (Rotation Number 2). You measure the score again. They are identical again.

The Conjecture: If the scores match for two different patterns, are the tables actually the same shape (just maybe rotated or moved)?

The Answer: YES.
Fierobe proves that if two ellipses match on two different "bouncing patterns," they are mathematically identical. You cannot trick the system. If you try to stretch one ellipse to match the score of the first pattern, it will automatically fail to match the second pattern unless it is exactly the same shape as the first one.

The "Same Size" Twist

The paper goes even further. What if you only have one pattern to test? Can you tell the shapes apart then?

• Scenario: You have two ellipses with the exact same perimeter (the total length of the edge is identical).
• Test: You check the "Bouncing Score" for just one pattern.
• Result: If the scores match, the shapes must be identical.

Think of it like this: If you have two rubber bands of the same total length, and you stretch them into oval shapes, the "tightness" of the bounce for a specific angle will be unique to that specific oval. You can't stretch one into a different oval without changing the bounce score.

The "Perfect Circle" is the King

The paper also discusses who is the "champion" of bouncing.

• Among all shapes with the same perimeter, the perfect circle (disk) always gives the highest possible "Bouncing Score" for almost every pattern.
• The paper proves that if you have a shape that is almost a circle (but slightly squished), and you try to wiggle it to see if you can get a better score, you can't. The circle is the local champion.
• Furthermore, no ellipse (that isn't a perfect circle) can ever be a local champion. If you have a rugby-ball-shaped table, you can always wiggle it slightly to get a better score. It is never the "best" it can be.

Why This Matters

This is like solving a puzzle where you only have a few clues.

• Old thinking: "We need to know the score for every single possible pattern to know the shape."
• New thinking (This paper): "No! We only need two clues (or even just one if we know the size) to know the shape, provided we know the shape is an ellipse."

Summary in a Nutshell

1. The Drum: We are trying to identify shapes by how balls bounce inside them.
2. The Ellipse Rule: If two ellipses bounce the same way on two different tracks, they are the same shape.
3. The Size Rule: If two ellipses are the same size and bounce the same way on one track, they are the same shape.
4. The Circle Crown: The perfect circle is the only shape that is unbeatable. Any ellipse that isn't a perfect circle can be tweaked to bounce better, meaning it's not the "best" shape.

This paper closes the door on a long-standing guess, proving that ellipses are uniquely identifiable by very little data, and confirming that the perfect circle holds a special, unbeatable status in the world of bouncing balls.

Technical Summary

Here is a detailed technical summary of the paper "Spectral rigidity among ellipses, Bialy's conjecture and local extrema of Mather's beta function" by Corentin Fierobe.

1. Problem Statement and Context

The paper addresses inverse spectral problems in the context of planar billiard dynamics. Specifically, it investigates whether the shape of a strictly convex domain $\Omega$ is uniquely determined by the Mather beta function $\beta_\Omega(\rho)$.

• Mather Beta Function: Defined as $\beta_\Omega(p/q) = -\frac{1}{q} L_{p/q}$, where $L_{p/q}$ is the maximal perimeter of a periodic orbit with rotation number $\rho = p/q$ ($q$ bounces, $p$ winding number). This function extends to a convex map on $[0, 1/2]$.
• The Core Question: If two domains (specifically ellipses) share the same values of $\beta_\Omega(\rho)$ at specific rotation numbers, are they isometric (identical up to translation, rotation, and reflection)?
• Bialy's Conjecture: Proposed by M. Bialy, this conjecture posits that if two ellipses $E$ and $E'$ satisfy $\beta_E(\rho_0) = \beta_{E'}(\rho_0)$ and $\beta_E(\rho_1) = \beta_{E'}(\rho_1)$ for two distinct rotation numbers $\rho_0, \rho_1 \in (0, 1/2]$, then $E$ and $E'$ must be isometric.
• Secondary Question: Can a single value $\beta_\Omega(\rho)$ determine an ellipse if the perimeter is fixed? Furthermore, what are the local extrema of $\beta(\rho)$ within the space of convex domains?

2. Methodology

The author employs a combination of variational calculus, symplectic geometry, and analytic perturbation theory.

• Support Functions and Generating Functions: The billiard map is treated as an exact symplectic twist map generated by a function $S(\phi_1, \phi_2) = -2h(\psi)\sin\delta$, where $h$ is the support function of the domain.
• Action-Parametrization: The proof relies on the existence of invariant curves (caustics) for the billiard map, parametrized by an action variable $\Theta$. This allows $\beta_\Omega(\rho)$ to be expressed as an integral over the invariant curve:
  $$\beta_\Omega(\rho) = -2 \int_0^1 h(\psi(\Theta)) \sin \delta(\Theta) \, d\Theta$$
• Variational Analysis (First Variations): The core technique involves computing the derivative of $\beta_\Omega(\rho)$ with respect to a deformation parameter $\tau$ of the domain family. Using Corollary 3 (derived from [7]), the derivative is:
  $$\frac{d}{d\tau} \beta_{\Omega_\tau}(\rho) = \int_0^1 \partial_\tau h_\tau(\psi) \sin \delta_\tau(\Theta) \, d\Theta$$
• Elliptic Integrals and Monotonicity: For the specific case of ellipses, the author derives explicit integral formulas involving the eccentricity $e$. The proof of monotonicity relies on analyzing the sign of the derivative by comparing integrals of the form $\int \frac{\sin^2 \psi}{\sqrt{(1-e^2\sin^2\psi)(1+k^2\sin^2\psi)}} d\psi$. The author uses a double integral argument to show that the derivative vanishes only if the rotation numbers are identical.
• KAM Theory: For the local extrema results, the paper utilizes Lazutkin's theorem on the existence of invariant curves (KAM curves) for Diophantine rotation numbers in $C^r$-smooth domains.

3. Key Contributions and Results

A. Proof of Bialy's Conjecture (Theorem 1)

The paper proves that two ellipses are isometric if they share $\beta$-values at two distinct rotation numbers.

• Mechanism: The author constructs an analytic one-parameter family of ellipses $(E_e)_{e \in [0,1)}$ where the eccentricity $e$ varies, but $\beta_{E_e}(\rho_0)$ is held constant.
• Monotonicity: It is proven that for any $\rho_1 \neq \rho_0$, the map $e \mapsto \beta_{E_e}(\rho_1)$ is strictly monotone.
• Conclusion: If $\beta_E(\rho_0) = \beta_{E'}(\rho_0)$ and $\beta_E(\rho_1) = \beta_{E'}(\rho_1)$, the eccentricities must be identical, implying the ellipses are isometric.

B. Rigidity with Fixed Perimeter (Theorem 2 & Corollary 1)

The paper establishes that an ellipse is uniquely determined by its perimeter and a single rotation number value.

• Theorem 2: For a fixed perimeter $p$, the map $e \mapsto \beta_{E_e}(\rho)$ (where $E_e$ is the unique ellipse of perimeter $p$ and eccentricity $e$) is strictly decreasing for $\rho \in (0, 1/2]$.
• Corollary 1: If two ellipses have the same perimeter and $\beta_E(\rho) = \beta_{E'}(\rho)$ for a single $\rho$, they are isometric.

C. Local Extrema of Mather's Beta Function (Theorem 4)

The paper characterizes local maximizers and minimizers of $\beta(\rho)$ within a bounded set of $C^r$-smooth domains.

• Result: There exists a set of rotation numbers $R'$ (containing Diophantine numbers) of positive measure such that:
  1. There are no local $C^r$-minimizers of $\beta(\rho)$.
  2. The only local $C^r$-maximizers are disks.
• Implication: Non-circular ellipses are not local maximizers of $\beta(\rho)$ for any $\rho \in (0, 1/2]$. This is a stronger statement than global maximality; it rules out ellipses as even local optima in the space of convex shapes.

D. Connection to Invariant Curves

The proof of the local extremum result relies on the fact that if a domain is a local extremizer, it must possess an invariant curve of constant angle. By Gutkin's results, only disks possess such curves for Diophantine rotation numbers, forcing the domain to be a disk.

4. Significance

• Resolution of a Conjecture: The paper provides a complete proof of Bialy's conjecture, settling a specific case of the inverse spectral problem for ellipses.
• Spectral Rigidity: It demonstrates that the "spectrum" (represented by $\beta$ at finite points) is highly rigid for ellipses. Unlike general domains where the length spectrum might not determine the shape, ellipses are uniquely determined by very sparse data (two points or one point + perimeter).
• Optimization Landscape: The results clarify the landscape of Mather's beta function. It confirms that disks are the unique global and local maximizers, while ellipses (despite being natural generalizations of disks) do not occupy local extrema positions.
• Methodological Advancement: The use of analytic families and the strict monotonicity of the beta function with respect to eccentricity provides a robust framework for future studies on spectral rigidity in other convex domains.

In summary, Fierobe's work rigorously establishes that the Mather beta function is a powerful invariant for ellipses, capable of distinguishing them from all other ellipses with minimal data, and proves that among all convex shapes, only the disk can locally maximize this function.