On Sections of Convex Bodies in John's Position and of Generalised $B_p^n$ Balls

Basic Information

• Paper ID: 2510.14047
• Title: On Sections of Convex Bodies in John's Position and of Generalised $B_p^n$ Balls
• Authors: David Alonso-Gutiérrez, Silouanos Brazitikos, Giorgos Chasapis
• Classification: math.MG (Metric Geometry), math.FA (Functional Analysis)
• Publication Date: October 17, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.14047

Abstract

This paper revisits a clever argument of K. Ball to provide sharp estimates for the volume of sections of convex bodies in John's position. The authors' technique combines the geometric Brascamp-Lieb inequality with generalized Parseval-type identities. This enables them to supplement some earlier results of the first two authors and extend the classical estimates of Meyer-Pajor and Koldobsky concerning extremal sections of $B_p^n$ balls to a broader family of norms induced by John identity decompositions in $\mathbb{R}^n$.

Research Background and Motivation

1. Core Problem: The study of volume estimates for hyperplane sections and projections of convex bodies is a classical and actively developing field in modern convex geometry, with deep connections to functional analysis, geometric tomography, and high-dimensional probability.
2. Problem Significance:
   • Extremal problems for k-dimensional sections (such as determining maximum or minimum volumes of such sections) encode subtle quantitative information about the geometry of geometric bodies
   • Related to central problems in the field, such as the Busemann-Petty problem and the slicing problem (recently resolved in KL)
3. Limitations of Existing Methods:
   • Ball's method heavily relies on the product structure of the cube: the Fourier transform of the indicator function of a product body decomposes as a product of one-dimensional Fourier transforms
   • For general geometric bodies in John's position, this coordinate independence is lacking, and direct product decomposition is no longer available; the Fourier method appears a priori inapplicable
4. Research Motivation:
   • To supplement and extend results in AB concerning sections of convex bodies in John's position
   • To overcome the requirement of product structure in Ball's method
   • To generalize classical estimates for sections of $\ell_p^n$ balls to a broader family of norms

Core Contributions

1. Methodological Innovation: Demonstrates how to circumvent the requirement of cubic product structure in Ball's argument by combining Parseval-type identities with the Brascamp-Lieb inequality
2. Sharp Estimates: Provides sharp volume estimates for k-dimensional sections of centrally symmetric convex bodies in John's position, particularly in the case $k \geq n/2$
3. Counterexample Construction: Constructs a convex polytope $L$ in John's position, proving that Ball's bound of $2^{\frac{n+k}{2}}$ no longer holds when $k \geq n/2$ and all projections $P_H(\sqrt{c_j}v_j)$ are large
4. Generalization of Classical Results: Extends known bounds of Meyer-Pajor and Koldobsky concerning sections of $\ell_p^n$ balls to the family of generalized $\ell_p$ norms induced by John decompositions
5. Asymmetric Case: Establishes improved results for hyperplane sections of convex bodies in John's position without imposing symmetry assumptions

Detailed Methodology

Problem Formulation

Study volume estimates for sections $K \cap H$ of convex bodies K in John's position by k-dimensional linear subspaces H. Given a John decomposition $\text{Id}_n = \sum_{j=1}^m c_j v_j \otimes v_j$, the goal is to establish upper bounds for $\text{vol}_k(K \cap H)$.

Core Technical Framework

1. Parseval-Type Identity (Proposition 1)

For a family of functions $(f_j)_{j=1}^m$ where $f_j \in S(\mathbb{R}^{n_j})$:

$\int_H \prod_{j=1}^m f_j(P_{\mathbb{R}^{n_j}} y) dy = \frac{1}{(2\pi)^{N-k}} \int_{H^⊥} \prod_{j=1}^m \hat{f}_j(P_{\mathbb{R}^{n_j}} z) dz$

2. Application of Geometric Brascamp-Lieb Inequality (Corollary 3)

For a John decomposition $(c_j, v_j)$:

$\int_{\mathbb{R}^n} \prod_{j=1}^m f_j(\sqrt{c_j}\langle x, v_j \rangle) dx \leq \frac{1}{(2\pi)^{m-n}} \prod_{j=1}^m \left(\int_\mathbb{R} |\hat{f}_j^{\frac{1}{1-c_j}}(\sqrt{1-c_j}t)| dt\right)^{1-c_j}$

3. Key Construction (Proposition 2)

Given $H \in G_{m,k}$ and $(u_j)_{j=1}^m \subseteq S^{m-1} \cap H$, $(c_j)_{j=1}^m \subseteq (0,\infty)$ satisfying $\text{Id}_H = \sum_{j=1}^m c_j u_j \otimes u_j$, there exist orthonormal bases $(x_j)_{j=1}^m$ and $(w_j)_{j=1}^m \subseteq S^{m-1} \cap H^⊥$ such that:

• $P_H x_j = \sqrt{c_j} u_j$
• $P_{H^⊥} x_j = \sqrt{1-c_j} w_j$

Technical Innovation Points

1. Role of Parseval Representation: In the setting of this paper, the Parseval representation plays the role of product decomposition in the cubic case: the volume is represented as an integral of products of univariate factors, but the exponents and prefactors reflect the geometric data of the John decomposition rather than coordinate independence
2. Brascamp-Lieb Reduction: Once the Parseval representation is in place, the Brascamp-Lieb reduction converts multidimensional integrals into products of one-dimensional integrals, with exponents determined by the Euclidean norms of vector projections
3. Dual Representation: The volume of a symmetric polytope admits two equivalent representations: the integral of products of indicator functions and the integral of Fourier transforms (on complementary subspaces)

Main Results

Centrally Symmetric Case (Theorem 6)

For $k \geq n/2$:

• If $\tilde{c}_j \geq \frac{1}{2}$ for all $j$, then: $\text{vol}_k(L \cap H) \leq 2^{\frac{m_0+k}{2}} \prod_{j=1}^{m_0} c_j^{\frac{\tilde{c}_j}{2t_j^2}}$
• Otherwise: $\text{vol}_k(L \cap H) \leq 2^{\frac{n-k}{2}} \text{vol}_k(B_\infty^k)$

Counterexample Construction (Theorem 5)

There exist arbitrarily large $k$ such that for each $n$ satisfying $n/2 \leq k \leq n$, there exists a convex body $L$ in John's position with: $\text{vol}_k(L \cap \mathbb{R}^k) = \binom{n}{k}^{\frac{k}{2}} \text{vol}_k(B_\infty^k)$

Generalized $\ell_p$ Balls (Theorems 16 and 23)

For generalized $\ell_p$ norms induced by John decompositions $\|x\|_{K_p} = \left(\sum_{j=1}^m \alpha_j |\langle x, v_j \rangle|^p\right)^{\frac{1}{p}}$:

Upper bound for $p=1$ case: $\text{vol}_k(K_1 \cap H) \leq \text{vol}_k(B_1^k) \prod_{j=1}^m \left(\frac{\sqrt{c_j}}{\alpha_j}\right)^{c_j\|P_H v_j\|_2^2}$

General case for $p \in [1,2]$: $\text{vol}_k(K_p \cap H) \leq \prod_{j \in J} \left(\frac{\sqrt{c_j}}{\alpha_j^{1/p}}\right)^{\tilde{c}_j} \text{vol}_k(B_p^k)$

Asymmetric Case (Theorem 26)

For a convex body $K$ in John's position and $a \in S^{n-1}$, if $\kappa_j \geq 1/2$ for all $j \in J$: $\text{vol}_{n-1}(K \cap a^⊥) \leq \frac{1}{\sqrt{2}} \sqrt{\frac{n+1}{n}} \left(\frac{n+1}{n-1}\right)^{\frac{n-1}{2}} \text{vol}_{n-1}(S^{n-1})$

Technical Details

Application of Wills Functional

The authors also investigate the Wills functional for sections of convex bodies in John's position: $W(\lambda(K \cap H)) = \sum_{i=0}^k V_i(\lambda(K \cap H)) = \sum_{i=0}^k \lambda^i V_i(K \cap H)$

Through the relationship between the Wills functional and volume and mean width:

• $\text{vol}_k(K \cap H) = \lim_{\lambda \to \infty} \frac{W(\lambda(K \cap H))}{\lambda^k}$
• $V_1(K \cap H) = \lim_{\lambda \to 0^+} \frac{W(\lambda(K \cap H)) - 1}{\lambda}$

Fourier Analysis Techniques

For the function $f(x) = e^{-|x|^p}$, its Fourier transform $\gamma_p(y) = \int_\mathbb{R} e^{ixy} e^{-|x|^p} dx$ satisfies the key estimate (Lemma 24): $\int_\mathbb{R} \gamma_p\left(\frac{t}{\sqrt{s}}\right)^s dt \leq 2\pi \left(2\Gamma\left(1 + \frac{1}{p}\right)\right)^{s-1}$

Related Work

1. Ball's Pioneering Work: Utilizes the geometric Brascamp-Lieb inequality and Fourier analysis techniques to study cube sections
2. Meyer-Pajor and Koldobsky: Establish classical estimates for sections of $\ell_p^n$ balls
3. Prior Work in AB: Systematically studies sections of convex bodies in John's position; this paper supplements and extends those results
4. John's Theorem: Provides canonical standardization of convex bodies, forming the foundation of this paper's technical framework

Conclusions and Discussion

Main Conclusions

1. Successfully overcomes Ball's method's dependence on product structure, providing a systematic approach to volume estimates for sections of general convex bodies in John's position
2. Establishes sharp bounds in both centrally symmetric and asymmetric cases, identifying new threshold phenomena
3. Extends classical $\ell_p$ ball results to broader families of norms, providing estimates more sensitive to geometric structure

Limitations

1. Some estimates require specific conditions on projection lengths (e.g., $\tilde{c}_j \geq 1/2$)
2. For the general case $p > 1$, intermediate bounds similar to the $p=1$ case cannot be provided due to inability to explicitly compute $\gamma_p$
3. The technical approach is primarily applicable to finite-dimensional Euclidean spaces

Future Directions

1. Explore applications to more general geometric bodies
2. Investigate high-dimensional asymptotic behavior
3. Develop more refined geometric inequality techniques

In-Depth Evaluation

Strengths

1. Strong Methodological Innovation: Cleverly combines Parseval identities and Brascamp-Lieb inequalities, breaking through limitations of traditional methods
2. Completeness of Results: Covers both symmetric and asymmetric cases, providing a systematic theoretical framework
3. Technical Depth: Involves deep techniques from Fourier analysis, convex geometry, and functional analysis
4. Practical Value: Extends important classical results and provides new tools for related research

Weaknesses

1. Technical Complexity: The method involves multiple sophisticated techniques, potentially limiting its scope of application
2. Condition Restrictions: Some results require specific geometric conditions, limiting applicability
3. Computational Complexity: Practical application may involve complex computations

Impact

1. Theoretical Contribution: Provides important new techniques and profound results for convex geometry
2. Methodological Influence: The Parseval-Brascamp-Lieb combination method may inspire research on related problems
3. Application Prospects: Has potential applications in high-dimensional data analysis, optimization theory, and related fields

Applicable Scenarios

1. Theoretical research in convex and integral geometry
2. High-dimensional probability and random geometry
3. Inequality theory in functional analysis
4. Geometric tomography and inverse problems

References

The paper cites important literature in the field, including:

• Ball's classical work on cube slicing
• Geometric forms of the Brascamp-Lieb inequality
• John's theorem and foundations of convex body theory
• Classical results of Meyer-Pajor, Koldobsky, and others on $\ell_p$ balls