An asymptotic shape optimization problem for Riesz means of Laplacian eigenvalues

This paper reviews recent results and presents new findings on the asymptotic optimization of Riesz means of Laplacian eigenvalues over convex sets and their disjoint unions, demonstrating that for certain exponents, the optimal shapes converge to a ball as the cut-off parameter tends to infinity.

The Gist

Imagine you are an architect tasked with designing a room (a shape) that has a fixed amount of floor space (volume). Your goal is to arrange the walls of this room in a specific way to either maximize or minimize a very specific "energy score" associated with the sound waves that can bounce around inside it.

This paper, written by mathematicians Rupert Frank and Simon Larson, is about finding the perfect shape for this room when the "sound" we are looking at becomes incredibly high-pitched (a mathematical concept called the limit where a parameter $\lambda$ goes to infinity).

Here is the breakdown of their discovery using simple analogies:

1. The Game: Tuning the Room

Think of a room as a musical instrument. When you clap your hands, sound waves bounce off the walls. Every room has a set of natural "notes" (eigenvalues) it can play.

• The Dirichlet Case (The "Silent" Room): Imagine the walls are soundproof and absorb all noise. The sound waves must be zero at the walls. The goal here is to find the shape that maximizes the total energy of the low-frequency notes.
• The Neumann Case (The "Reflective" Room): Imagine the walls are perfectly hard mirrors. The sound waves bounce off without losing energy. The goal here is to find the shape that minimizes the total energy.

The mathematicians are asking: "As we look at higher and higher frequencies (like a scream getting higher and higher), what shape of room gives us the best score?"

2. The Intuition: The Ball is King

In the world of geometry, there is a famous rule called the Isoperimetric Inequality. It basically says: "For a fixed amount of area, the circle (or sphere in 3D) has the shortest perimeter."

The authors suspected that as the sound gets higher and higher, the "perfect" room shape would eventually look more and more like a perfect ball. Why?

• In the "Silent Room" scenario, you want to minimize the "friction" the sound feels against the walls. A ball has the smallest wall surface for its size.
• In the "Reflective Room" scenario, the math works out similarly: the ball is the most efficient shape.

3. The Twist: It's Not Always a Ball

The paper confirms that yes, the ball is the winner, but only under certain conditions.

The authors discovered a "tipping point" (called a critical exponent).

• If the sound is "loud" enough (high exponent): The optimal shape is definitely a ball. If you try to build a long, skinny room or a square room, you will lose the competition. As the sound gets higher, the optimal shape morphs into a perfect sphere.
• If the sound is "quiet" (low exponent): The ball might not be the winner. In fact, the optimal shape might break apart into many tiny, disconnected bubbles.

4. The "Swiss Cheese" Surprise

One of the most interesting findings in the paper concerns what happens if you allow the room to be made of multiple disconnected pieces (like a block of Swiss cheese rather than a single solid room).

• The Ball Rule: If the sound frequency is high enough, the "Swiss cheese" collapses. All the tiny holes merge, and the optimal shape becomes one single, solid ball.
• The Bubble Rule: If the frequency is too low, the optimal shape refuses to be a single ball. Instead, it splits into thousands of tiny, separate bubbles. It's as if the system decides, "I can't win as one big room, so I'll win by being a million tiny rooms."

5. The Connection to a Famous Mystery

The paper also touches on a famous unsolved problem in math called Pólya's Conjecture.

• Think of this as a "Grand Challenge" in math: "Is the ball always the best shape, no matter what?"
• The authors show that proving their result about the ball winning in the high-frequency limit is just as hard as solving Pólya's Conjecture. If you can prove the ball always wins, you solve the big mystery. If you can't, you can't prove the ball wins in this specific limit either.

Summary

In plain English, this paper says:

"We studied the best shapes for rooms to hold sound energy. We found that if the sound is high-pitched enough, the perfect shape is always a ball. However, if the sound is too low-pitched, the best shape might be a swarm of tiny bubbles. Proving that the ball is always the winner is one of the hardest puzzles in modern mathematics."

The authors used advanced tools (like "semiclassical analysis," which is like looking at sound waves through a microscope that blurs the line between waves and particles) to prove that, eventually, nature prefers the smooth, round ball over any jagged or complex shape.

Technical Summary

1. Problem Statement

The paper investigates shape optimization problems concerning the Riesz means of Laplace eigenvalues for open sets $\Omega \subset \mathbb{R}^d$ ($d \geq 2$) with a fixed volume $|\Omega|=1$.

Let $-\Delta^\sharp_\Omega$ denote the Laplacian with either Dirichlet ($\sharp=D$) or Neumann ($\sharp=N$) boundary conditions. For a parameter $\lambda \geq 0$ and an exponent $\gamma > 0$, the Riesz mean is defined as:
$$\text{Tr}(-\Delta^\sharp_\Omega - \lambda)^\gamma_- = \sum_{j} (\lambda_j - \lambda)_-^\gamma$$
where $\lambda_j$ are the eigenvalues and $(x)_- = \max(0, -x)$.

The authors study the behavior of the optimal shapes as the cut-off parameter $\lambda \to \infty$. Specifically, they address two classes of optimization problems:

1. Convex Sets ($C_d$): Optimizing over the class of open, bounded, convex sets.
   $$M^\sharp_\gamma(\lambda) := \sup/\inf \{ \text{Tr}(-\Delta^\sharp_\Omega - \lambda)^\gamma_- : \Omega \in C_d, |\Omega|=1 \}$$
2. Disjoint Unions of Convex Sets ($\tilde{C}_d$): Optimizing over sets that are disjoint unions of convex components.
   $$\tilde{M}^\sharp_\gamma(\lambda) := \sup/\inf \{ \text{Tr}(-\Delta^\sharp_\Omega - \lambda)^\gamma_- : \Omega \in \tilde{C}_d, |\Omega|=1 \}$$

The Central Question: Do the optimizers (or near-optimizers) for these problems converge, up to translation, to a ball as $\lambda \to \infty$?

2. Methodology and Theoretical Framework

The authors employ a combination of semiclassical analysis, spectral inequalities, and geometric measure theory.

A. Weyl Asymptotics and Subleading Terms

The analysis relies on the two-term Weyl asymptotic expansion for the trace:
$$\text{Tr}(-\Delta^\sharp_\Omega - \lambda)^\gamma_- \sim L_{\gamma,d}^{sc} |\Omega| \lambda^{\gamma + d/2} \mp \frac{1}{4} L_{\gamma-1,d}^{sc} H_{d-1}(\partial \Omega) \lambda^{\gamma + (d-1)/2}$$
where $L_{\gamma,d}^{sc}$ is the semiclassical constant and $H_{d-1}(\partial \Omega)$ is the surface area.

• Dirichlet ($D$): The subleading term is negative; maximizing the trace requires minimizing surface area.
• Neumann ($N$): The subleading term is positive; minimizing the trace requires minimizing surface area.
  Since the volume is fixed, minimizing surface area implies the shape should approach a ball (by the isoperimetric inequality). However, this argument is only valid if the optimizers behave "regularly" enough for the asymptotics to hold.

B. Critical Riesz Exponents

A central concept introduced is the critical exponent $\gamma^\sharp_d$, defined as the smallest $\gamma$ such that the Riesz mean is bounded by the leading Weyl term (i.e., the semiclassical constant is sharp for convex sets):
$$\gamma^\sharp_d := \inf \left\{ \gamma \geq 0 : \text{Tr}(-\Delta^\sharp_\Omega - \lambda)^\gamma_- \leq (\geq) L_{\gamma,d}^{sc} |\Omega| \lambda^{\gamma + d/2} \quad \forall \Omega \in C_d \right\}$$

• The validity of Pólya's conjecture for convex sets is equivalent to $\gamma^\sharp_d = 0$.
• The authors prove (Theorem 2.1) that $\gamma^\sharp_d < 1$ for all dimensions $d$.
• The behavior of the optimization problem changes drastically depending on whether $\gamma$ is above or below these critical thresholds.

C. Uniform Semiclassics and Geometric Inequalities

The proofs utilize:

• Uniform semiclassical estimates for convex sets (Theorem 3.3), which provide error bounds depending on the inradius $r_{in}(\Omega)$.
• Improved Berezin-Li-Yau and Kröger inequalities to handle cases where $\lambda$ is small relative to the geometry.
• Blaschke's Selection Theorem to ensure the existence of convergent subsequences of convex sets in the Hausdorff metric.
• Anisotropic scaling arguments to analyze degenerate limits where the set becomes "flat" in certain directions.

3. Key Contributions and Results

A. Convergence to a Ball for Convex Sets (Section 3)

Theorem 1.1 & 3.2:
For the class of convex sets ($C_d$), if $\gamma$ is sufficiently large, any sequence of almost-optimizers converges to a ball.

• Condition: $\gamma > (\gamma^\sharp_{d-1} - 1/2)_+$.
• Result: If $\{ \Omega_j \}$ are near-optimizers for $M^\sharp_\gamma(\lambda_j)$ with $\lambda_j \to \infty$, then $\Omega_j$ converges (up to translation) in the Hausdorff sense to a unit ball.
• Significance: This confirms the intuition that the ball is the optimizer in the high-energy limit, provided the exponent $\gamma$ is large enough to suppress "flat" or degenerate shapes. The threshold depends on the critical exponent in dimension $d-1$.

B. Multi-Component Optimization (Section 4)

Theorem 1.2 & 4.2:
For the class of disjoint unions of convex sets ($\tilde{C}_d$), the behavior is more complex and depends on the critical exponent in dimension $d$ ($\gamma^\sharp_d$).

• Case 1: $\gamma > \gamma^\sharp_d$
  The optimizers converge to a single ball. The system does not benefit from splitting into multiple components.
• Case 2: $\gamma < \gamma^\sharp_d$
  The optimizers do not converge to a single ball. Instead, the mass splits into many small components.
  • The number of connected components grows as $\sim \lambda^{d/2}$.
  • The individual components become vanishingly small (measure $\to 0$) or have vanishing inradius.
• Case 3: $\gamma = \gamma^\sharp_d$
  The behavior is critical and depends on the specific constants.

C. Connection to Pólya's Conjecture

The paper establishes a profound link between the asymptotic shape optimization and Pólya's conjecture.

• Theorem 1.3 & 1.4: The convergence of optimizers to a ball for all $\gamma > 0$ is equivalent to the validity of Pólya's conjecture (i.e., $\gamma^\sharp_d = 0$) for convex sets.
• Since Pólya's conjecture is open for $d \geq 2$, the unconditional convergence of optimizers to a ball for all $\gamma$ remains an open problem. The authors show that proving convergence for small $\gamma$ is as hard as proving Pólya's conjecture.

4. Significance and Implications

1. Resolution of the "Ball" Conjecture in Specific Regimes: The paper rigorously proves that for sufficiently large Riesz exponents, the ball is indeed the asymptotic optimizer, even when allowing for disjoint unions of convex sets.
2. Phase Transition in Shape: The work reveals a phase transition in the optimal shape based on the exponent $\gamma$.
   • High $\gamma$: The system prefers a single, compact, spherical shape (minimizing boundary effects).
   • Low $\gamma$: The system prefers fragmentation into many small components to exploit the spectral properties of the boundary terms more effectively.
3. Methodological Advance: The authors develop a robust framework for analyzing shape optimization in the semiclassical limit, specifically handling the interplay between the geometry of the domain (convexity, inradius) and the spectral parameter $\lambda$.
4. Bibliographic Context: This work extends previous studies on individual eigenvalue optimization to the broader context of Riesz means, which are more stable and physically relevant in many quantum mechanical applications (e.g., density of states).

Summary Conclusion

Frank and Larson demonstrate that the asymptotic shape of domains optimizing Riesz means of Laplace eigenvalues is governed by the critical Riesz exponent $\gamma^\sharp_d$. For large $\gamma$, the ball is the unique asymptotic optimizer. For small $\gamma$, the optimal shape fragments into a cloud of small convex sets. The paper effectively reduces the problem of unconditional convergence to a ball to the famous open problem of Pólya's conjecture for convex sets.