Beyond Cheeger's constant

Lorenzo Brasco

2024-04-05

Abstract:The Cheeger constant of an open set of the Euclidean space is defined by minimizing the ratio "perimeter over volume", among all its smooth compactly contained subsets. We consider a natural variant of this problem, where the volume of admissible sets is raised to any positive power. We show that for {\it sublinear} powers, all these generalized Cheeger constants are equivalent to the standard one, by means of a universal two-sided estimate. We also show that this equivalence breaks down for {\it superlinear} powers. In this case, some weird phenomena appear. We finally consider the case of convex planar sets and prove an existence result, under optimal assumptions.

Analysis of PDEs,Functional Analysis

What problem does this paper attempt to address?

The problems that this paper attempts to solve mainly revolve around the generalization of the Cheeger constant and its properties. Specifically, the author explores variants of the Cheeger constant defined in Euclidean space, where the volume term is raised to an arbitrary positive power. The main problems of the paper can be summarized as follows:

1. **Equivalence of the generalized Cheeger constant**:
- For sub - linear powers (i.e., \(0 < q < 1\)), are all these generalized Cheeger constants equivalent to the standard Cheeger constant?
- For super - linear powers (i.e., \(q > 1\)), does this equivalence still hold?

2. **New phenomena and characteristics**:
- When the power exceeds a certain critical value, will some strange phenomena emerge? For example, the behavior of minimizing sequences, the existence and regularity of optimal solutions, etc.

3. **Existence results for convex planar sets**:
- In the case of convex planar sets, does an optimal solution exist? If so, what are the conditions?

4. **Generalization of the Poincaré inequality**:
- Is the generalized Cheeger constant associated with some form of the Poincaré - Sobolev inequality?

### Specific problems

- **Problem (Q1)**: Can a "universal" lower bound similar to the Cheeger inequality be proved, but with \(h_q\) instead of \(h_1\)?

- **Problem (Q2)**: Can \(h_1(\Omega)\) and \(h_q(\Omega)\) be compared (when \(q\neq1\))?

- **Problem (Q3)**: For each \(0 < q <\frac{N}{N - 1}\), is it always the case that \(\lambda_{1,q}(\Omega)=h_q(\Omega)\)?

### Main conclusions

The main conclusions of the paper include:

1. **Comparison of Cheeger constants**:
- For each \(1 < q <\frac{N}{N - 1}\) and each open set \(\Omega\subseteq\mathbb{R}^N\), we have
\[
C_1\left(h_1(\Omega)\right)^{\frac{N}{q-(N - 1)}}\leq h_q(\Omega)\leq C_2\left(h_1(\Omega)\right)^{\frac{N}{q-(N - 1)}}
\]
where \(C_1,C_2 > 0\) depend only on \(N\) and \(q\).
- For each \(0 < q < 1\), we have
\[
h_q(\Omega)\leq C_3\left(h_1(\Omega)\right)^{\frac{N}{q-(N - 1)}}
\]
and this estimate cannot be reversed.

2. **Poincaré constants**:
- For each \(1 < q <\frac{N}{N - 1}\), the constant \(h_q(\Omega)\) is always equal to a sharp Poincaré - Sobolev constant.
- For each \(0 < q < 1\), generally only \(h_q(\Omega)\geq\lambda_{1,q}(\Omega)\), and the equality does not hold for some specific shapes.

3. **Existence of convex set extrema**:
- In the two - dimensional case, for each \(1 < q < 2\), if \(\Omega\subseteq\mathbb{R}^2\) is an open convex set (not necessarily bounded), then
\[
h_q(\Omega)=\inf\left\{\frac{H^1(\partial E)}{|E|^{\frac{1}{q}}}:E\subseteq\Omega\text{ is an open bounded convex set}\right\}
\]
and if and only if the high ridge of \(\Omega\)