Abstract
For a convex body $K$ in $\mathbb{R}^n$, the classical inequalities of Rogers-Shephard and Zhang provide fundamental bounds on the volume of derived bodies, such as the difference body $DK$ and the polar projection body $\Pi^\circ K$. Specifically, $\text{vol}_n(DK) \leq \binom{2n}{n} \text{vol}_n(K) \leq \text{vol}_n(n\text{vol}_n(K)\Pi^\circ K)$, with equality if and only if $K$ is an $n$-dimensional simplex. This work introduces a unified framework that generalizes these operators to an $m$th-order setting using weighted measures with non-uniform densities. We establish a nested chain of set inclusions for radial mean bodies that interpolates between the $m$th-order difference body and the $m$th-order polar projection body, proving that the simplex remains the unique extremal case under these generalized conditions.
Introduction
In the study of convex geometry, researchers often analyze "derived bodies"—new geometric objects constructed from an initial convex set $K$. Two of the most significant derived bodies are the difference body ($DK$), representing the set of all differences between points in $K$, and the projection body ($\Pi K$), which is intrinsically linked to the surface area of $K$’s orthogonal projections.
Historically, the relationships between these bodies were governed by the Rogers-Shephard and Zhang’s projection inequalities, typically formulated using the standard Lebesgue measure (uniform volume). However, recent advancements have pushed these boundaries in two directions: higher-order intersections ($m$th-order) and weighted measures (measures with non-uniform density functions). This paper synthesizes these two tracks, providing a comprehensive theory for $m$th-order weighted operators.
The $m$th-Order Weighted Framework
The core of this research involves extending the concept of the difference body to $m$ translates. For $m \in \mathbb{N}$, the $m$th-order difference body is defined as: $$D^m(K) = \{(x_1, \dots, x_m) \in \mathbb{R}^{nm} : K \cap \bigcap_{i=1}^m (K + x_i) \neq \emptyset\} \subset \mathbb{R}^{nm}$$
To move beyond standard volume, the authors replace the Lebesgue measure with a measure $\mu$ possessing a density function $\phi(x)$. This allows for the analysis of "mass" distribution within the geometry. The bridge between these concepts is the weighted $m$th-order covariogram, $g_{\mu,m}$, which measures the $\mu$-measure of the intersection of $K$ and its $m$ translates.
Variational Foundations
A primary contribution of this work is the rigorous geometric link established via variational derivatives. In Theorem 1.3, the authors demonstrate that the infinitesimal change in the $m$th-order covariogram as one moves radially from the origin is precisely the support function of the weighted projection body.
To ensure these functions remain well-behaved, the authors utilize the theory of $s$-concave measures. These measures satisfy a generalized Brunn-Minkowski inequality, providing the necessary structural "curvature" to prove set inclusions. Furthermore, the authors extend Aleksandrov’s Variational Formula to arbitrary measures, linking changes in a body's measure to its weighted surface area.
The Interpolation Chain
The paper highlights the role of radial mean bodies ($R_p K$) as a continuous bridge between the difference body and the projection body. As the parameter $p$ varies, these bodies "morph" from the $m$th-order difference body toward the $m$th-order projection body.
Theorem 4.5 establishes a nested relationship for $s$-concave measures: $$D^m(K) \subseteq \text{scaled } R^m_q K \subseteq \text{scaled } R^m_p K \subseteq \text{scaled } \Pi^{\circ,m}_\mu K$$
This inclusion chain demonstrates that the geometry of the difference body is always contained within the geometry of the projection body, with radial mean bodies filling the intermediate space.
Key Results and Extremality
- Weighted Zhang’s Inequality: The authors establish a lower bound for the measure of the polar weighted $m$th-order projection body (Theorem 3.5).
- Simplex Extremality: Despite the complexity introduced by weights and higher orders, the $n$-dimensional simplex remains the unique extremal case. Equality in the inclusion chains holds if and only if $K$ is a simplex, the measure is essentially Lebesgue, and the concavity parameter $s$ is exactly $1/n$.
- Generalized Volume: The final section extends these results to "generalized volumes," suggesting that the underlying geometric principles are robust even beyond measures with densities.
Discussion and Future Work
While this framework provides a powerful unification, several open questions remain. The results largely depend on the measure being $s$-concave; whether these inequalities hold for measures without such concavity is currently unknown. Additionally, the convexity of radial mean bodies $R^m_p K$ for the parameter range $p \in (-1, 0)$ remains unproven, even in the unweighted case. Finally, while the paper addresses non-symmetric bodies, further refinements may be possible for origin-symmetric sets, potentially touching upon the Gardner-Zvavitch conjecture.
Summary updated