arXiv++ Combinatorics

For a fixed poset $\mathcal P$ we say that a family $\mathcal F\subseteq\mathcal P([n])$ is $\mathcal P$-saturated if it does not contain an induced copy of $\mathcal P$, but whenever we add a new set to $\mathcal F$, we form an induced copy of $\mathcal P$. The size of the smallest such family is denoted by $\text{sat}^*(n, \mathcal P)$.\par For the diamond poset $\mathcal D_2$ (the two-dimensional Boolean lattice), while it is easy to see that the saturation number is at most $n+1$, the best known lower bound has stayed at $O(\sqrt n)$ since the introduction of the area of poset saturation. In this paper we prove that $\text{sat}^*(n, \mathcal D_2)\geq \frac{n+1}{5}$, establishing that the saturation number for the diamond is linear. The proof uses a result about certain pairs of set systems which may be of independent interest.

The dyadic dual VC-dimension of a set system \( \mathcal{F} \) is the largest integer \( \ell \) such that there exist \( \ell \) sets \( F_1, F_{2}, \dots, F_\ell \in \mathcal{F} \), where every pair \( \{i, j\} \in \binom{[\ell]}{2} \) is witnessed by an element \( a_{i,j} \in F_i \cap F_j \) that does not belong to any other set \( F_k \) with \( k \in [\ell] \setminus \{i, j\} \). In this paper, we determine the largest dyadic dual VC-dimension of a non-piercing family is exactly $4$, providing a rare example where the maximum of this parameter can be determined for a natural family arising from geometry. As an application, we give a short and direct proof that the transversal number \( τ(\mathcal{F}) \) of any non-piercing family is at most \(Cν(\mathcal{F})^9 \), where \( ν(\mathcal{F}) \) is the matching number and $C$ is a constant. This improves a recent result of Pálvölgyi and Zólomy.

We analyze generalized progressions in some nonabelian groups using a measure of complexity called VC-dimension, which was originally introduced in statistical learning theory by Vapnik and Chervonenkis. Here by a "generalized progression" in a group $G$, we mean a finite subset of $G$ built from a fixed set of generators in analogy to a (multidimensional) arithmetic progression of integers. These sets play an important role in additive combinatorics and, in particular, the study of approximate groups. Our two main results establish finite upper bounds on the VC-dimension of certain set systems of generalized progressions in finitely generated free groups and also the Heisenberg group over $\mathbb{Z}$.

We prove that for every $d\in \mathbb{N}$ and a graph class of bounded expansion $\mathscr{C}$, there exists some $c\in \mathbb{N}$ so that every graph from $\mathscr{C}$ admits a proper coloring with at most $c$ colors satisfying the following condition: in every ball of radius $d$, every color appears either zero times or an odd number of times. For $d=1$, this provides a positive answer to a question raised by Goetze, Klute, Knauer, Parada, Peña, and Ueckerdt [ArXiv 2505.02736] about the boundedness of the strong odd chromatic number in graph classes of bounded expansion. The key technical ingredient towards the result is a proof that the strong odd coloring number of a sets system can be bounded in terms of its semi-ladder index, 2VC dimension, and the maximum subchromatic number among induced subsystems.

We determine the largest size of $3$-uniform set systems on $[n]$ with VC-dimension $2$ for all $n$.

Extremal problems on set systems with restricted intersections have been an important part of combinatorics in the last 70 year. In this paper, we study the following Ramsey version of these problems. Given a set $L\subseteq \{0,\dots,k-1\}$ and a family $\mathcal{F}$ of $k$-element sets which does not contain a sunflower with $m$ petals whose kernel size is in $L$, how large a subfamily of $\mathcal{F}$ can we find in which no pair has intersection size in $L$? We give matching upper and lower bounds, determining the dependence on $m$ for all $k$ and $L$. This problem also finds applications in quantum computing. As an application of our techniques, we also obtain a variant of Füredi's celebrated semilattice lemma, which is a key tool in the powerful delta-system method. We prove that one cannot remove the double-exponential dependency on the uniformity in Füredi's result, however, we provide an alternative with significantly better, single-exponential dependency on the parameters, which is still strong enough for most applications of the delta-system method.

The combinatorial discrepancy of arithmetic progressions inside $[N] := \{1, \ldots, N\}$ is the smallest integer $D$ for which $[N]$ can be colored with two colors so that any arithmetic progression in $[N]$ contains at most $D$ more elements from one color class than the other. Bounding the discrepancy of such set systems is a classical problem in discrepancy theory. More recently, this problem was generalized to arithmetic progressions in grids like $[N]^d$ (Valk{ó}) and $[N_1]\times \ldots \times [N_d]$ (Fox, Xu, and Zhou). In the latter setting, Fox, Xu, and Zhou gave upper and lower bounds on the discrepancy that match within a $\frac{\log |Ω|}{\log \log |Ω|}$ factor, where $Ω:= [N_1]\times \ldots \times [N_d]$ is the ground set. In this work, we use the connection between factorization norms and discrepancy to improve their upper bound to be within a $\sqrt{\log|Ω|}$ factor from the lower bound. We also generalize Fox, Xu, and Zhou's lower bound, and our upper bounds to arithmetic progressions in arbitrary convex bodies.

In its Euclidean form, the Dense Neighborhood Lemma (DNL) asserts that if $V$ is a finite set of points of $\mathbb{R}^N$ such that for each $v \in V$ the ball $B(v,1)$ intersects $V$ on at least $δ|V|$ points, then for every $\varepsilon >0$, the points of $V$ can be covered with $f(δ,\varepsilon)$ balls $B(v,1+\varepsilon)$ with $v \in V$. DNL also applies to other metric spaces and to abstract set systems, where elements are compared pairwise with respect to (near) disjointness. In its strongest form, DNL provides an $\varepsilon$-clustering with size exponential in $\varepsilon^{-1}$, which amounts to a Regularity Lemma with 0/1 densities of some trigraph. Trigraphs are graphs with additional red edges. They are natural instances of partial concept classes, introduced by Alon, Hanneke, Holzman and Moran [FOCS 2021]. This paper is mainly a combinatorial study of the generalization of Vapnik-Cervonenkis dimension to partial concept classes. The main point is to show how trigraphs can sometimes explain the success of random sampling even though the VC-dimension of the underlying graph is unbounded. All the results presented here are effective in the sense of computation: they primarily rely on uniform sampling with the same success rate as in classical VC-dimension theory. Among some applications of DNL, we show that $\left(\frac{3t-8}{3t-5}+\varepsilon\right)\cdot n$-regular $K_t$-free graphs have bounded chromatic number. Similarly, triangle-free graphs with minimum degree $n/3-n^{1-\varepsilon}$ have bounded chromatic number (this does not hold with $n/3-n^{1-o(1)}$). For tournaments, DNL implies that the domination number is bounded in terms of the fractional chromatic number. Also, $(1/2-\varepsilon)$-majority digraphs have bounded domination, independently of the number of voters.

Directed acyclic graphs (DAGs) are fundamental structures used across many scientific fields. A key concept in DAGs is the least common ancestor (LCA), which plays a crucial role in understanding hierarchical relationships. Surprisingly little attention has been given to DAGs that admit a unique LCA for every subset of their vertices. Here, we characterize such global lca-DAGs and provide multiple structural and combinatorial characterizations. We show that global lca-DAGs have a close connection to join semi-lattices and establish a connection to forbidden topological minors. In addition, we introduce a constructive approach to generating global lca-DAGs and demonstrate that they can be recognized in polynomial time. We investigate their relationship to clustering systems and other set systems derived from the underlying DAGs.

Let $X$ and $Y$ be upward closed set systems in the lattice of $\{0,1\}^n$. The celebrated Harris-Kleitman inequality implies that if $|X|=α2^n$, $|Y|=β2^n$, the density of the set of points in exactly one of $X$ and $Y$ is maximal when $X$ and $Y$ are independent, meaning $|X\cap Y|=αβ2^n$. Is the same true of three upward closed systems, $X$, $Y$, and $Z$? Suppose $|X|=|Y|=|Z|$. Kahn asked whether the set of points in exactly one of $X$, $Y$, $Z$ has density at most $\frac49$. We answer this question in the negative.

We establish a theorem regarding the maximum size of an {\it{induced}} matching in the bipartite complement of the incidence graph of a set system $(X,\mathcal{F})$. We show that this quantity plus one provides an upper bound on the colorful Helly number of this set system, i.e. the minimum positive integer $N$ for which the following statement holds: if finite subfamilies $\mathcal{F}_1,\ldots, \mathcal{F}_{N} \subset \mathcal{F}$ are such that $\cap_{F \in \mathcal{F}_{i}} F = 0$ for every $i=1,\ldots,N$, then there exists $F_i \in \mathcal{F}_i$ such that $F_1 \cap \ldots \cap F_{N} = \emptyset$. We will also discuss some natural refinements of this result and applications.

We investigate the longstanding problem of determining the maximum size of a $(d+1)$-uniform set system with VC-dimension at most $d$. Since the seminal 1984 work of Frankl and Pach, which established the elegant upper bound $\binom{n}{d}$, this question has resisted significant progress. The best-known lower bound is $\binom{n-1}{d} + \binom{n-4}{d-2}$, obtained by Ahlswede and Khachatrian, leaving a substantial gap of $\binom{n-1}{d-1}-\binom{n-4}{d-2}$. Despite decades of effort, improvements to the Frankl--Pach bound have been incremental at best: Mubayi and Zhao introduced an $Ω_d(\log{n})$ improvement for prime powers $d$, while Ge, Xu, Yip, Zhang, and Zhao achieved a gain of 1 for general $d$. In this work, we provide a purely combinatorial approach that significantly sharpens the Frankl--Pach upper bound. Specifically, for large $n$, we demonstrate that the Frankl--Pach bound can be improved to $\binom{n}{d} - \binom{n-1}{d-1} + O_d(n^{d-1 - \frac{1}{4d-2}})=\binom{n-1}{d}+O_d(n^{d-1 - \frac{1}{4d-2}})$. This result completely removes the main term $\binom{n-1}{d-1}$ from the previous gap between the known lower and upper bounds. It also offers fresh insights into the combinatorial structure of uniform set systems with small VC-dimension. In addition, the original Erdős--Frankl--Pach conjecture, which sought to generalize the EKR theorem in the 1980s, has been disproven. We propose a new refined conjecture that might establish a sturdier bridge between VC-dimension and the EKR theorem, and we verify several specific cases of this conjecture, which is of independent interest.

We show that the shatter function of a semilinear set system on $\mathbb{R}^m$ is asymptotic to a polynomial. This confirms, for the structure $(\mathbb{R}; +, <)$, a conjecture of Chernikov and is a step towards characterizing model-theoretic linearity via shatter functions.

The VC-dimension of a family of sets is a measure of its combinatorial complexity used in machine learning theory, computational geometry, and even model theory. Computing the VC-dimension of the $k$-fold union of geometric set systems has been an open and difficult combinatorial problem, dating back to Blumer, Ehrenfeucht, Haussler, and Warmuth in 1989, who ask about the VC-dimension of $k$-fold unions of half-spaces in $\mathbb{R}^d$. Let $\mathcal{F}_1$ denote the family of all lines in $\mathbb{R}^2$. It is well-known that $\mathsf{VC}\text{-}\mathsf{dim}(\mathcal{F}_1) = 2$. In this paper, we study the $2$-fold and $3$-fold unions of $\mathcal{F}_1$, denoted $\mathcal{F}_2$ and $\mathcal{F}_3$, respectively. We show that $\mathsf{VC}\text{-}\mathsf{dim}(\mathcal{F}_2) = 5$ and $\mathsf{VC}\text{-}\mathsf{dim}(\mathcal{F}_3) = 9$. Moreover, we give complete characterisations of the subsets of $\mathbb{R}^2$ of maximal size that can be shattered by $\mathcal{F}_2$ and $\mathcal{F}_3$, showing they are exactly two and five, respectively, up to isomorphism in the language of the point-line incidence relation.

Let $q=p^α$ be a fixed prime power, $k\geq 2$ be an integer. We give a new upper bound for the size of $k$-wise $q$-modular $L$-avoiding $L$-intersecting set systems, where $L$ is any proper subset of $\{0, \ldots , q-1\}$. Our proof is based on the linear algebra bound method and basic number theory.

Let $N$ be a finite set and $\mathcal{F}$, an intersection-closed family of subsets. Frankl conjectured that there always exists an element in $N$ which is contained in at most half the number of sets in $\mathcal{F}$ unless $\mathcal{F} =\{E\}$. We prove the validity of a class of inequalities which imply Frankl's conjecture.

For any positive integers $n\ge d+1\ge 3$, what is the maximum size of a $(d+1)$-uniform set system in $[n]$ with VC-dimension at most $d$? In 1984, Frankl and Pach initiated the study of this fundamental problem and provided an upper bound $\binom{n}{d}$ via an elegant algebraic proof. Surprisingly, in 2007, Mubayi and Zhao showed that when $n$ is sufficiently large and $d$ is a prime power, the Frankl-Pach upper bound is not tight. They also remarked that their method requires $d$ to be a prime power, and asked for new ideas to improve the Frankl-Pach upper bound without extra assumptions on $n$ and $d$. In this paper, we provide an improvement for any $d\ge 2$ and $n\ge 2d+2$, which demonstrates that the long-standing Frankl-Pach upper bound $\binom{n}{d}$ is not tight for any uniformity. Our proof combines a simple yet powerful polynomial method and structural analysis.

We say that two partial orders on $[n]$ are compatible if there exists a partial order that refines both of them. This compatibility relation induces a natural set system structure between the collection $\mathcal{F}$ of all partial orders and the collection $\mathcal{G}$ of all total orders on $[n]$, where each order is associated with the set of orders compatible with it. In this note, we determine the VC-dimension of $\mathcal{F}$ with respect to $\mathcal{G}$, proving that $\operatorname{VC}_{\mathcal{G}}(\mathcal{F}) = \lfloor\frac{n^2}{4}\rfloor$ for $n \ge 4$. We also establish bounds on the dual VC-dimension, showing that $2(n-3) \le \operatorname{VC}_{\mathcal{F}}(\mathcal{G}) \le n \log_2 n$ for all $n \ge 1$.

We explore the connections between clusters and least common ancestors (LCAs) in directed acyclic graphs (DAGs), focusing on the interplay between so-called $I$-lca-relevant DAGs and DAGs with the $I$-lca-property. Here, $I$ denotes a set of integers. In $I$-lca-relevant DAGs, each vertex is the unique LCA for some subset $A$ of leaves of size $|A|\in I$, whereas in a DAG with the $I$-lca-property there exists a unique LCA for every subset $A$ of leaves satisfying $|A|\in I$. We elaborate on the difference between these two properties and establish their close relationship to pre-$I$-ary and $I$-ary set systems. This, in turn, generalizes results established for (pre-)binary and $k$-ary set systems. Moreover, we build upon recently established results that use a simple operator $\ominus$, enabling the transformation of arbitrary DAGs into $I$-lca-relevant DAGs. This process reduces unnecessary complexity while preserving key structural properties of the original DAG. The set $C_G$ consists of all clusters in a DAG $G$, where clusters correspond to the descendant leaves of vertices. While in some cases $C_H = C_G$ when transforming $G$ into an $I$-lca-relevant DAG $H$, it often happens that certain clusters in $C_G$ do not appear as clusters in $H$. To understand this phenomenon in detail, we characterize the subset of clusters in $C_G$ that remain in $H$ for DAGs $G$ with the $I$-lca-property. Furthermore, we show that the set $W$ of vertices required to transform $G$ into $H = G \ominus W$ is uniquely determined for such DAGs. This, in turn, allows us to show that the ``shortcut-free'' version of the transformed DAG $H$ is always a tree or a galled-tree whenever $C_G$ represents the clustering system of a tree or galled-tree and $G$ has the $I$-lca-property. In the latter case $C_H = C_G$ always holds.

We study and solve several problems in two closely related settings: set families in $2^{[n]}$ with many disjoint pairs of sets and low rank matrices with many zero entries. - More than 40 years ago, Daykin and Erdős asked for the maximum number of disjoint pairs of sets in a family $F\subseteq 2^{[n]}$ of size $2^{(1/2+δ)n}$ and conjectured it contains at most $o(|F|^2)$ such pairs. This was proven by Alon and Frankl in 1985. In this paper we completely resolve this problem, proving an optimal dependence of the number of disjoint pairs on the size of family $F$. We also prove the natural variant of the Daykin-Erdős conjecture in which disjoint pairs are replaced by pairs with intersection $λ\neq 0$. - Motivated by a conjecture of Lovett related to the famous log-rank conjecture, Singer and Sudan asked to show that for two families $A, B \subseteq 2^{[n]}$ with a positive constant fraction of set pairs $(a,b)\in A\times B$ being disjoint, there are $R\subset A$ and $S\subset B$ such that all set pairs $(r, s)\in R\times S$ are disjoint, and $|R|\geq 2^{-O(\sqrt{n})}|A|$ and $|S|\geq 2^{-O(\sqrt{n})}|B|$. We prove this conjecture in a strong quantitative form. - We prove the following generalizations of the best known bounds for the log-rank conjecture. If $M$ is an $n\times n$ non-negative integer matrix of rank $r$ in which the average of the entries is $\varepsilon\leq 1/2$, then $M$ contains an all-zero submatrix of size at least $2^{-O(\sqrt{\varepsilon r})}n$. Unlike the known bounds for the log-rank conjecture, this result is optimal. Moreover, using similar methods, we also prove that any $n\times n$ matrix of rank $r$ with entries from $\{0,\dots,t\}$ contains a constant submatrix of size at least $2^{-O(t\sqrt{r})}n$. Our proofs use probabilistic, entropy and discrepancy methods and explore connections to additive combinatorics and coding theory.