A functional Loomis-Whitney type inequality in the Heisenberg group and projection theorems over finite fields

Technical Summary: Functional Loomis–Whitney Inequalities in the Finite Heisenberg Group

Problem Statement
This paper investigates functional Loomis–Whitney type inequalities within the noncommutative setting of the Heisenberg group over a finite field, denoted as $H_n(\mathbb{F}_q)$, where $q$ is an odd prime power. The classical Loomis–Whitney inequality bounds the volume of a $d$-dimensional body by the volumes of its $(d-1)$-dimensional coordinate projections. In the discrete setting, this relates the size of a set $A$ to the sizes of its projections $\pi_i(A)$.

The authors aim to establish analogous functional inequalities for the Heisenberg group. Specifically, they seek to determine the sharp region of exponents $(u_1, \dots, u_{2n})$ such that the inequality
$$\frac{1}{q^{2n+1}} \sum_{(x,t) \in H_n(\mathbb{F}_q)} \prod_{j=1}^{2n} f_j(\pi_j(x, t)) \lesssim \prod_{j=1}^{2n} \|f_j\|_{L^{u_j}(\mathbb{F}_q^{2n}, dx)}$$
holds uniformly in $q$. Here, $\pi_j$ are specific vertical coordinate projections mapping $H_n(\mathbb{F}_q)$ to subgroups $W_j \cong \mathbb{F}_q^{2n}$. The study also addresses the corresponding set-theoretic inequalities, bounding the cardinality $|K|$ of a subset $K \subset H_n(\mathbb{F}_q)$ in terms of the sizes of its projections $\{|\pi_j(K)|\}_{j=1}^{2n}$.

Methodology
The paper employs a combination of harmonic analysis, combinatorial geometry, and induction on the dimension $n$.

1. Structural Analysis of Subgroups: The authors first analyze the subgroup structure of $H_n(\mathbb{F}_q)$. They distinguish between general subgroups (when $q$ is prime) and homogeneous subgroups (invariant under the dilation action). They characterize these subgroups using isotropic subspaces of the underlying symplectic space $\mathbb{F}_q^{2n}$ and establish counting formulas for these subgroups.
2. Case $n=1$ (Planar Heisenberg Group): For $n=1$, the problem is reduced to analyzing a bilinear operator $L(f_1, f_2)$. The authors utilize Hölder's inequality and the nesting property of norms. They determine the sharp region of exponents by testing specific functions and applying interpolation theorems. The proof relies on establishing operator bounds $A(s \to r)$ for a specific linear operator $A$ derived from the group multiplication structure.
3. General Case $n > 1$: For higher dimensions, the authors do not attempt a full characterization of the admissible exponent region, which is significantly more complex. Instead, they focus on the symmetric critical exponent $u = \frac{n(2n+1)}{n+1}$.
   • Induction on Dimension: The proof of the symmetric estimate relies on an induction scheme inspired by continuous settings but adapted for finite fields. The core strategy involves "freezing" one symplectic coordinate pair to rewrite the $n$-dimensional form as an averaged $(n-1)$-dimensional Heisenberg Loomis–Whitney form.
   • Asymmetric Estimates: The authors first establish a family of asymmetric estimates (Theorem 4.1) where one symplectic pair of inputs is measured in $L^{\frac{2n+1}{2}}$ and the remaining inputs in $L^{2n+1}$.
   • Multilinear Interpolation: These asymmetric bounds are combined using a multilinear interpolation scheme to reach the symmetric estimate at the critical exponent.
4. Incidence Geometry (Special Case $n=1$): When $n=1$ and $|K| > q$, the authors utilize Vinh's point-line incidence theorem to derive a stronger lower bound for the maximum projection size, improving upon the general functional inequality in this specific regime.

Key Contributions and Results

• Sharp Exponents for $n=1$: The paper determines the exact region of exponents $(u_1, u_2)$ for which the inequality holds uniformly in $q$. The condition is given by the system:
  $$\frac{1}{u_1} + \frac{2}{u_2} \leq 2 \quad \text{and} \quad \frac{2}{u_1} + \frac{1}{u_2} \leq 2.$$
  This includes the endpoint estimate $L^{3/2} \times L^{3/2} \to L^1$, which is shown to be optimal.
• Symmetric Multilinear Estimate for General $n$: The authors prove the symmetric multilinear estimate at the endpoint exponent $u = \frac{n(2n+1)}{n+1}$. This is established via the induction and interpolation strategy described above.
• Sharp Set Inequalities: Specializing the functional results to indicator functions yields a sharp Loomis–Whitney type set inequality. For any finite set $K \subset H_n(\mathbb{F}_q)$, the size is bounded by:
  $$|K| \lesssim q^{\frac{1}{2n+1}} \prod_{j=1}^{2n} |\pi_j(K)|^{\frac{n+1}{n(2n+1)}}.$$
  Consequently, the paper derives a lower bound for the maximum projection size:
  $$\max_{1 \leq j \leq 2n} |\pi_j(K)| \gtrsim_n |K|^{\frac{2n+1}{2(n+1)}} q^{-\frac{1}{2(n+1)}}.$$
  The authors demonstrate that this result is optimal up to absolute constants by constructing specific examples.
• Improved Bound for $n=1$: For $n=1$ and large sets ($|K| > q$), the paper provides a stronger bound using incidence geometry:
  $$\max\{|\pi_1(K)|, |\pi_2(K)|\} \gtrsim \min\left\{ |K|^{1/2}q^{1/2}, |K|q^{-1/2} \right\}.$$

Significance and Connections
The paper positions its results as a bridge between discrete harmonic analysis and combinatorial geometry in non-abelian groups.

• Structural Insight: The inequalities shed light on the structural properties of the Heisenberg group over finite fields, particularly regarding how the group's non-commutative structure (via the symplectic form) influences projection behavior.
• Connection to Boundedness Problems: The authors highlight a direct connection to the boundedness problem for multilinear forms/operators over finite fields studied by Bhowmik, Iosevich, Koh, and Pham (2025). Specifically, the $n=1$ case of their inequality is shown to coincide with the boundedness of a specific multilinear operator associated with a complete graph on two vertices and a kernel defined by a quadratic relation.
• Orthogonal Projections: The results are compared to orthogonal projection and covering questions in $\mathbb{F}_q^{2n+1}$ studied by Chen (2018). The authors show that the family of Heisenberg projections $\{\pi_j\}$ provides a strictly stronger lower bound for the maximal projection size compared to the bounds derived from Chen's results on orthogonal projections in the vector space setting.
• Future Directions: The paper notes that the classification of all subgroups of $H_n(\mathbb{F}_q)$ (specifically non-product type subgroups) remains an open problem. Furthermore, the authors mention plans to investigate discrete analogues of covering problems for the real Heisenberg group (related to work by Balogh, Fässler, Mattila, and Tyson) in a sequel paper.

The work is presented as a foundational step in extending Loomis–Whitney theory to non-commutative finite groups, offering sharp constants and identifying critical exponents that were previously unknown in this setting.