A reverse isoperimetric inequality in three-dimensional space forms

This paper proves a sharp reverse isoperimetric inequality in three-dimensional space forms of constant curvature, demonstrating that among all $\lambda$-convex bodies with a fixed surface area, the $\lambda$-convex lens uniquely minimizes volume, thereby confirming Borisenko's Conjecture for non-zero curvature cases.

The Gist

Here is an explanation of the paper, translated from mathematical jargon into everyday language using analogies.

The Big Picture: The "Reverse" Balloon Problem

Imagine you have a piece of elastic fabric (like a balloon skin). In the world of geometry, there is a famous rule called the Isoperimetric Inequality. It basically says: "If you have a fixed amount of fabric (surface area), the shape that holds the most air (volume) is a perfect sphere."

This paper is about the Reverse Isoperimetric Inequality. Instead of asking, "What shape holds the most?" it asks: "What shape holds the least?"

But there's a catch. You can't just make a shape that holds almost nothing (like a flat pancake) because the paper adds a rule: The shape must be "stiff."

The Rules of the Game

1. The "Stiffness" Rule ($\lambda$-convexity):
   Imagine you are trying to roll a ball under the surface of your shape. The rule says: "You must be able to roll a specific-sized ball (let's call it a Stiffness Ball) under the surface everywhere, without the surface bending inward too sharply."

   • If the surface bends too sharply (like a sharp corner or a deep dent), the Stiffness Ball won't fit.
   • So, a "$\lambda$-convex body" is a shape that is "round enough" everywhere. It can't have sharp spikes or deep valleys.

2. The Goal:
   You have two shapes, Shape A and Shape B.

   • They both have the exact same amount of skin (Surface Area).
   • They both obey the Stiffness Rule.
   • Question: Which one holds less air (Volume)?

The Winner: The "Lens"

The authors prove that the shape that holds the least amount of air is a $\lambda$-convex Lens.

• What is a Lens? Imagine two round, convex bubbles (like the two halves of a contact lens) pressed together. The shape formed where they overlap is the "Lens."
• The Result: If you take any other stiff shape with the same amount of skin, it will always hold more air than this Lens. The Lens is the "most efficient" way to waste space while keeping the skin tight and stiff.

The Setting: Where does this happen?

The paper solves this problem in three different "universes" (Space Forms):

1. Flat Space (Euclidean): Like a normal sheet of paper extended infinitely. (This was already solved for 3D recently).
2. Spherical Space: Like the surface of a giant ball (positive curvature).
3. Hyperbolic Space: Like a Pringles chip or a saddle shape that curves away in all directions (negative curvature).

The authors specifically solved the mystery for Spherical and Hyperbolic spaces in 3D. They confirmed a guess (conjecture) made by a mathematician named Borisenko: In these curved universes, the Lens is still the champion of minimal volume.

How Did They Prove It? (The "Peeling" Analogy)

To prove this, the authors used a clever method involving "peeling" the shapes layer by layer.

1. The Onion Method: Imagine taking your shape and shaving off a tiny layer of skin from the outside, moving inward. You keep doing this until the shape disappears.
2. Tracking the Skin: As you shave off layers, the surface area changes.
   • For a perfect sphere, the skin shrinks at a steady, predictable rate.
   • For a Lens, the skin shrinks at a specific rate.
   • For any other stiff shape, the authors proved that the skin shrinks slower than the Lens does.

The Logic:

• If Shape A and the Lens start with the same amount of skin.
• And as you peel them, the Lens loses skin faster than Shape A.
• Then, by the time you reach the center, the Lens must have had less total volume to begin with.

If Shape A had less volume, it would have run out of skin before the Lens did, which contradicts the fact that they started with the same amount of skin. Therefore, Shape A must have had more volume.

Why Does This Matter?

• Solving a Mystery: It confirms a long-standing guess about how geometry works in curved spaces.
• New Tools: The math they developed to prove this (involving "Gauss-Bonnet" theorems and "rolling balls") is a new toolkit that can help solve other problems in geometry.
• Real World: While we don't live in a hyperbolic universe, understanding how shapes behave in curved spaces helps in fields like general relativity (how gravity bends space) and material science (how stiff materials deform).

Summary in One Sentence

If you have a fixed amount of "stiff" skin in a curved universe, the shape that traps the least amount of space inside is a Lens (two bubbles stuck together), and no other shape can beat it.

Technical Summary

Here is a detailed technical summary of the paper "A Reverse Isoperimetric Inequality in Three-Dimensional Space Forms" by Drach, Solanes, and Tatarko.

1. Problem Statement

The paper addresses the Reverse Isoperimetric Inequality within the class of $\lambda$-convex bodies in three-dimensional space forms $M^3(c)$ (manifolds of constant sectional curvature $c$).

• Context: In standard isoperimetric problems, one seeks to maximize volume for a fixed surface area. The reverse problem seeks to minimize volume for a fixed surface area within a specific class of convex sets.
• $\lambda$-Convexity: A convex body $K$ is $\lambda$-convex if its boundary $\partial K$ has principal curvatures bounded below by a constant $\lambda > 0$ (in a weak sense). This generalizes standard convexity (where $\lambda=0$) and implies the body is "curved outward" at least as much as a sphere of curvature $\lambda$.
• The Extremal Candidate: The conjectured minimizer is the $\lambda$-convex lens ($L$), defined as the domain bounded by two totally umbilical hypersurfaces (caps) of normal curvature $\lambda$.
• Borisenko's Conjecture: The central open problem addressed is Borisenko's Conjecture, which posits that for any $\lambda$-convex body $K$ and a $\lambda$-convex lens $L$ with equal surface areas ($|\partial K| = |\partial L|$), the volume satisfies $|K| \ge |L|$, with equality if and only if $K$ is a lens.
• Gap in Literature: While the conjecture was known for Euclidean space ($c=0$) in dimensions $n=2$ and recently $n=3$, and for all 2D space forms, it remained open for 3D space forms with non-zero curvature ($c \neq 0$, i.e., spherical and hyperbolic spaces).

2. Methodology

The authors employ a combination of differential geometry, convex geometry, and calculus of variations using the method of inner parallel bodies.

A. Definitions and Preliminaries

• $\lambda$-Spheres and $\lambda$-Balls: Defined as totally umbilical hypersurfaces of constant normal curvature $\lambda$. In hyperbolic space, these include geodesic spheres ($\lambda > 1$), horospheres ($\lambda = 1$), and equidistant hypersurfaces ($0 < \lambda < 1$).
• $\lambda$-Convex Polyhedra: Defined as intersections of finitely many $\lambda$-balls. The authors first prove the result for polyhedra and then extend it to general bodies via approximation.
• Inner Parallel Bodies: For a body $K$, the inner parallel body $K_t$ at distance $t$ is the set of points at distance $\ge t$ from the boundary. As $t$ increases, $K_t$ shrinks until it vanishes at the inradius $r(K)$.

B. Key Technical Tools

1. Gauss–Bonnet Theorem for $\lambda$-Convex Polyhedra (Theorem 3.1):
   The authors derive a specific version of the Gauss–Bonnet theorem for $\lambda$-convex polyhedra in $M^3(c)$. For a polyhedron with edges $E$ and vertices $V$:
   $$(\lambda^2 + c)|\partial K| + 2\lambda \sum_{E} \ell_E \tan\left(\frac{\beta_E}{2}\right) + \sum_{v} |u(K, v)| = 4\pi$$
   where $\ell_E$ is edge length, $\beta_E$ is the exterior dihedral angle, and $|u(K, v)|$ is the spherical area of the normal cone at vertex $v$. This formula relates surface area, curvature, and combinatorial geometry.

2. Variation of Surface Area (Theorem 4.2):
   They establish the derivative of the surface area of the inner parallel body $K_t$ with respect to $t$:
   $$\frac{d}{dt}|\partial K_t| = -(n-1)\lambda(t)|\partial K_t| - 2 \sum_{E \in E(t)} \ell_E \tan\left(\frac{\beta_E}{2}\right)$$
   Here, $\lambda(t)$ evolves according to $\lambda'(t) = \lambda(t)^2 + c$. This formula allows the comparison of how surface area shrinks for different shapes.

3. Comparison Principle:
   The proof strategy relies on comparing the surface area functions $f_K(t) = |\partial K_t|$ and $f_L(t) = |\partial L_t|$.

   • They utilize a known result (Theorem 4.1) stating that if $|\partial K| = |\partial L|$, then the inradius $r(K) \ge r(L)$, with equality iff $K$ is a lens.
   • They prove that if $K$ is not a lens, the rate of decrease of surface area for $K$ is strictly slower (or the area remains larger) than that of the lens $L$ at corresponding stages of shrinking.

3. Key Contributions and Results

Main Theorem (Reverse Isoperimetric Inequality in Nonflat Space Forms)

Statement: Let $K \subset M^3(c)$ with $c \neq 0$ be a $\lambda$-convex body, and let $L$ be a $\lambda$-convex lens. If $|\partial K| = |\partial L|$, then:
$$|K| \ge |L|$$
Equality holds if and only if $K$ is a $\lambda$-convex lens.

Proof Logic:

1. Polyhedral Case: Assume $K$ is a $\lambda$-convex polyhedron.
2. Initial Condition: At $t=0$, $|\partial K| = |\partial L|$.
3. Gauss-Bonnet Comparison: Using the derived Gauss-Bonnet formula, the authors show that for a fixed surface area, the sum of edge terms $\sum \ell_E \tan(\beta_E/2)$ is minimized by the lens.
4. Differential Inequality: This implies that the derivative of the surface area of the lens, $f_L'(t)$, is more negative (decreases faster) than that of $K$, $f_K'(t)$, whenever their areas are equal.
5. Contradiction Argument: If $K$ were not a lens, $f_K(t)$ would stay strictly above $f_L(t)$ for small $t$. If they were to cross at some $t_0$, the derivative condition would prevent the lens from "overtaking" the body, leading to a contradiction unless $K$ is already a lens.
6. Volume Integration: Since $|K| = \int_0^{r(K)} |\partial K_t| dt$ and $r(K) \ge r(L)$, and $|\partial K_t| \ge |\partial L_t|$ for all $t$, it follows that $|K| \ge |L|$.
7. Generalization: The result is extended from polyhedra to arbitrary $\lambda$-convex bodies via metric approximation, relying on the rigidity of the inradius inequality.

Secondary Result: 2D Hyperbolic Space

The authors provide an alternative proof for the reverse isoperimetric inequality in the 2-dimensional hyperbolic space $H^2(-1)$.

• They adapt the variation method to 2D polygons.
• They use the 2D Gauss-Bonnet theorem: $2\pi = -|K| + \lambda|\partial K| + \sum \beta_i$.
• By analyzing the monotonicity of the function $x \mapsto \tan(x/2)$, they show that the lens minimizes volume, offering a new perspective distinct from previous proofs (e.g., [Dr1]).
• Note: The authors explicitly state this method fails for the 2-sphere ($S^2$) because the sign of the area term in the Gauss-Bonnet formula changes.

4. Significance

• Resolution of a Major Conjecture: The paper completes the verification of Borisenko's Conjecture in all three-dimensional space forms. Combined with the recent Euclidean result [DT], the conjecture is now fully resolved for $n=3$ across all constant curvature models.
• Geometric Rigidity: It establishes the $\lambda$-convex lens as the unique minimizer of volume under fixed surface area constraints in curved spaces, highlighting the role of curvature $c$ in geometric inequalities.
• Methodological Advancement: The derivation of the Gauss-Bonnet theorem for $\lambda$-convex polyhedra in space forms (Theorem 3.1) is a novel contribution. It bridges discrete geometry (polyhedra) with continuous curvature constraints in non-Euclidean settings.
• Unified Approach: The use of inner parallel bodies and variation formulas provides a robust framework that works uniformly for spherical ($c>0$) and hyperbolic ($c<0$) geometries, differing only in the evolution of the curvature parameter $\lambda(t)$.

In summary, this work settles a long-standing problem in convex geometry by proving that among all $\lambda$-convex bodies with a given surface area in 3D curved spaces, the "lens" shape encloses the least volume, and it does so by developing powerful new tools for analyzing the geometry of $\lambda$-convex sets.