An inequality for the compositions of convex functions with convolutions and an alternative proof of the Brunn-Minkowski-Kemperman inequality

Takashi Satomi

Home/Archive/arXiv

arXiv · 2111.15349

An inequality for the compositions of convex functions with convolutions and an alternative proof of the Brunn-Minkowski-Kemperman inequality

Takashi Satomi

View original source ↗Read source document ↗Export metadata ↓

Abstract

Let $m(G)$ be the infimum of the volumes of all open subgroups of a unimodular locally compact group $G$. Suppose integrable functions $ϕ_1 , ϕ_2 \colon G \to [0,1]$ satisfy $\| ϕ_1 \| \leq \| ϕ_2 \|$ and $\| ϕ_1 \| + \| ϕ_2 \| \leq m (G)$, where $\| \cdot \|$ denotes the $L^1$-norm with respect to a Haar measure $dg$ on $G$. We have the following inequality for any convex function $f \colon [0, \| ϕ_1 \| ] \to \mathbb{R}$ with $f(0) = 0$: \begin{align*} \int_{G}^{} f \circ ( ϕ_1 * ϕ_2 ) (g) dg \leq 2 \int_{0}^{\| ϕ_1 \|} f(y) dy + ( \| ϕ_2 \| - \| ϕ_1 \| ) f( \| ϕ_1 \| ). \end{align*} As a corollary, we have a slightly stronger version of Brunn-Minkowski-Kemperman inequality. That is, we have \begin{align*} \mathrm{vol}_* ( B_1 B_2 ) \geq \mathrm{vol} ( \{ g \in G \mid 1_{B_1} * 1_{B_2} (g) > 0 \} ) \geq \mathrm{vol} (B_1) + \mathrm{vol} (B_2) \end{align*} for any non-null measurable sets $B_1 , B_2 \subset G$ with $\mathrm{vol} (B_1) + \mathrm{vol} (B_2) \leq m(G)$, where $\mathrm{vol}_*$ denotes the inner measure and $1_B$ the characteristic function of $B$.

Explore related subjects

math.MG math.CO math.FA math.GR

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Takashi Satomi. 2023-02-18. An inequality for the compositions of convex functions with convolutions and an alternative proof of the Brunn-Minkowski-Kemperman inequality. https://arxiv.org/abs/2111.15349

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

An approximate counterexample to the Barker--Larman problem in dimension $4$

The Barker--Larman problem asks if a convex body $K \subseteq \mathbb R^n$ containing the Euclidean ball $\mathbb B_n$, such that all the sections of $K$ by hyperplanes tangent to $\mathbb B_n$ have constant $(n-1)$-dimensional volume, must necessarily be a Euclidean ball. In this paper we show a result pointing to a negative answer in dimension $4$. Taking $λ_0 = 4 \sqrt{3}π$ and any $N \in \mathbb N$, we obtain the existence of a family of convex bodies $K_{λ,N}$ with $λ\in (λ_0-r_N, λ_0 + r_N)$, such that the sections of $K_{λ,N}$ by hyperplanes tangent to the Euclidean ball, have area within $c |λ- λ_0|^{N+1}$ of $λ$, while the difference between outradius and inradius of $K_{λ,N}$ is larger than $C |λ- λ_0|$. The bodies $K_{λ,N}$ are constructed via radial functions as \[ρ_{K_{λ,N}}(t) = \cos\left( \sum_{n=0}^N \frac{(λ-λ_0)^n}{n!} φ_n(t) \right)^{-1},\] where $t \in [0,2π), λ\in \mathbb R$ and $φ_n$ are trigonometric polynomials that can be computed explicitly. The convergence of the inner power series when $N \to \infty$ (which is left open) would imply a negative answer to the Barker--Larman problem in dimension $4$. As an example we obtain a convex body whose outradius and inradius differ by more than $0.176$, and the area of the sections oscillate by less than $3 \times 10^{-7}$.

math.MG↗

arXiv2026

On convex bodies with rotationally symmetric planar projections

Let $n\ge 3$ and let $K\subset\mathbb R^n$ be a convex body. For a two-dimensional linear subspace $P\subset\mathbb R^n$, let $K|P$ be the orthogonal projection of $K$ onto $P$. We prove that if for every two-dimensional subspace $P$, the planar convex body $K|P$ has $q$-fold rotational symmetry up to translation, then for $q\ge4$, this forces $K$ to be an Euclidean ball. The case $q=3$ is exceptional: non-spherical examples exist.

math.MG↗

arXiv2026

A Spectral Proof of Khachiyan's Ellipsoid Conjecture

For a convex body $K\subset\mathbb{R}^n$, let $w(K)$ denote the volume of its maximum-volume inscribed ellipsoid. We prove that every closed halfspace $H$ whose boundary passes through the center of the maximizing ellipsoid satisfies \[ w(K\cap H)\le\frac{\sqrt e}{2}\,w(K). \] The constant is optimal uniformly over all dimensions, as witnessed by a family of circular cones, thereby establishing Khachiyan's conjecture. The proof converts containment, maximality, and the central-cut condition into algebraic constraints on positive definite matrices. Two complementary spectral bounds from a diagonal model extend to arbitrary center displacements through a directional rank-one estimate for fractional trace powers. Concavity determines their joint optimum. We also derive finite-dimensional bounds and a necessary condition for near equality, with self-contained supporting proofs and an alternative resolvent argument. An AI language model discovered the proof in a human-directed research process. Lean 4 with mathlib verifies the main theorem, sharpness, and supporting results.

math.MG↗