Showing 20 papers (Total: 81623 papers, 47312 authors in database)
Decomposition Polyhedra of Piecewise Linear Functions
Published
• View Publication
• BIB
In this paper we contribute to the frequently studied question of how to decompose a continuous piecewise linear (CPWL) function into a difference of two convex CPWL functions. Every CPWL function has infinitely many such decompositions, but for applications in optimization and neural network theory, it is crucial to find decompositions with as few linear pieces as possible. This is a highly challenging problem, as we further demonstrate by disproving a recently proposed approach by Tran and Wang [Minimal representations of tropical rational functions. Algebraic Statistics, 15(1):27-59, 2024]. To make the problem more tractable, we propose to fix an underlying polyhedral complex determining the possible locus of nonlinearity. Under this assumption, we prove that the set of decompositions forms a polyhedron that arises as intersection of two translated cones. We prove that irreducible decompositions correspond to the bounded faces of this polyhedron and minimal solutions must be vertices. We then identify cases with a unique minimal decomposition, and illustrate how our insights have consequences in the theory of submodular functions. Finally, we improve upon previous constructions of neural networks for a given convex CPWL function and apply our framework to obtain results in the nonconvex case.
A lonely weak tile
Published
• View Publication
• BIB
The notion of weak tiling was a key ingredient in the proof of Fuglede's spectral set conjecture for convex bodies \cite{conv}, due to the fact that every spectral set tiles its complement weakly with a suitable Borel measure. In this paper we review the concept of weak tiling, and answer a question raised in \cite{weak} by giving an example of a set $T$ which tiles its complement weakly, but $T$ is neither spectral, nor a proper tile.
A tower lower bound for the degree relaxation of the Regularity Lemma
Published
• View Publication
• BIB
It is well-known that if $(A,B)$ is an $\tfrac{\varepsilon}{2}$-regular pair (in the sense of Szemerédi) then there exist sets $A'\subset A$ and $B'\subset B'$ with $|A'|\leq \varepsilon|A|$ and $|B'|\leq \varepsilon|B|$ so that the degrees of all vertices in $A\setminus A'$ differ by at most $\varepsilon|B|$ and the degrees of all vertices in $B\setminus B'$ differ by at most $\varepsilon|A|$. We call such a property "$\varepsilon$-degularity". This leads to the notion of an "$\varepsilon$-degular" partition of a graph in the same way as the definition of $\varepsilon$-regular pairs leads to the notion of $\varepsilon$-regular partitions. We show that there exist graphs in which any $\varepsilon$-degular partition requires the number of clusters to be $\mathrm{tower}(Θ(\varepsilon^{-1/3}))$. That is, even though degularity is a substantial relaxation of regularity, in general one cannot improve much on the bounds that come with Szemerédi's regularity lemma.
The chain algebra of a pure poset
Published
• View Publication
• BIB
We extend the notion of chain algebra, originally defined in \cite{GN} for finite distributive lattices, to that of finite pure posets. We show this algebra corresponds to the Ehrhart ring of a (0,1)-polytope, termed the chain polytope, and characterize the indecomposability of this polytope. Furthermore, we prove the normality of the chain algebra, describe its canonical module, and extend one of main results from \cite{GN} by computing its Krull dimension. For width-2 pure posets, we determine the algebra's regularity and conditions for it to be Gorenstein or nearly Gorenstein.
Some line and conic arrangements and their Waldschmidt constants
We study the Waldschmidt constant of some configurations in the projective plane. In the first part, we show that the Waldschmidt constant of a set $\mathbb{X}$ of $n$ points where at least $n-3$ points among them lie on a line is either equal to $1, \frac{2n-3}{n-1}, 2, \frac{16}{7}, \frac{7}{3}, \frac{17}{7},$ or $\frac{5}{2}$. Together with the Hilbert polynomials, this gives a complete geometric characterization for $\mathbb{X}$. Next, we study some specific configurations whose Waldschmidt constants are bounded from above by $\frac{5}{2}$. Under this condition, we describe all configurations of $n$ points with $n-1$ points among them lying on an irreducible conic, and we also study some specific configurations of $9$ points.
Asymptotic normality arising in Baxter permutations
Baxter permutations arose in the study of fixed points of the composite of commuting functions by Glen Baxter in 1964. This type of permutations are counted by Baxter numbers $B_n$. It turns out that $B_n$ enumerate a lot of discrete objects such as the bases for subalgebras of the Malvenuto-Reutenauer Hopf algebra, the pairs of twin binary trees on $n$ nodes, or the diagonal rectangulations of an $n\times n$ grid. The refined Baxter number $D_{n,k}$ also count many interesting objects including the Baxter permutations of $n$ with $k-1$ descents and $n-k$ rises, twin pairs of binary trees with $k$ left leaves and $n-k+1$ right leaves, or plane bipolar orientations with $k+1$ faces and $n-k+2$ vertices. In this paper, we obtain the asymptotic normality of the refined Baxter number $D_{n,k}$ by using a sufficient condition due to Bender. In the course of our proof, the computation involving $B_n$ and some related numbers is crucial, while $B_n$ has no closed form which make the computation untractable. To address this problem, we employ the method of asymptotics of the solutions of linear recurrence equations. Our proof is semi-automatic. All the asymptotic expansions and recurrence relations are proved by utilizing symbolic computation packages.
Canonical Landau-Ginzburg models for cominuscule homogeneous spaces
Published
• View Publication
• BIB
We present a type-independent Landau-Ginzburg (LG) model $(X_\mathrm{can}, \mathcal{W}_\mathrm{can})$ for any cominuscule homogeneous space $X=G/P$. We give a fully combinatorial construction for our superpotential $\mathcal{W}_\mathrm{can}$ as a sum of $n+1$ rational functions in the (generalized) Plücker coordinates on the "Langlands dual" minuscule homogeneous space $\mathbb{X}=P^\vee\backslash G^\vee$. Explicitly, we define the denominators $\mathcal{D}_{i_*}$ of these rational functions using the combinatorics of order ideals of the corresponding minuscule poset, which can be interpreted as (generalized) Young diagrams, by a process that can be described by "moving boxes" and hence is easily implemented. To construct the corresponding numerators, we define derivations $δ_{i_1}$ on $\mathbb{C}[\mathbb{X}]$ that act by "adding an appropriate box if possible" and then we apply each $δ_{i_1}$ to the corresponding $\mathcal{D}_{i_*}$. By studying certain Weyl orbits in the fundamental representations of $\widetilde{G}^\vee$ and exploiting the existence of a certain dense algebraic torus in $\mathbb{X}$, we show that the polynomials $\mathcal{D}_{i_*}$ coincide with the generalized minors $φ_{i_*}$ appearing in the cluster structures for homogeneous spaces studied by Geiß-Leclerc-Schröer in arXiv:math/0609138. We then define the mirror variety $X_\mathrm{can}=\mathbb{X}\setminus D_\mathrm{ac}$ to be the complement of the anticanonical divisor $D_\mathrm{ac} = \sum_{i_*}\{\mathcal{D}_{i_*}=0\}$ formed by the $\mathcal{D}_{i_*}$. Moreover, we show that the LG models $(X_\mathrm{can},\mathcal{W}_\mathrm{can})$ are isomorphic to the Lie-theoretic LG-models $(X_\mathrm{Lie},\mathcal{W}_\mathrm{Lie})$ constructed by Rietsch in arXiv:math/0511124 and our models naturally generalize the type-dependent Plücker coordinate LG-models previously studied by various authors.
Counterexample to Babai's lonely colour conjecture
Motivated by colouring minimal Cayley graphs, in 1978, Babai conjectured that no-lonely-colour graphs have bounded chromatic number. We disprove this in a strong sense by constructing graphs of arbitrarily large girth and chromatic number that have a proper edge-colouring in which each cycle contains no colour exactly once.
K-Independent Boolean Networks
Published
• View Publication
• BIB
This paper proposes a new parameter for studying Boolean networks: the independence number. We establish that a Boolean network is $k$-independent if, for any set of $k$ variables and any combination of binary values assigned to them, there exists at least one fixed point in the network that takes those values at the given set of $k$ indices. In this context, we define the independence number of a network as the maximum value of $k$ such that the network is $k$-independent. This definition is closely related to widely studied combinatorial designs, such as "$k$-strength covering arrays", also known as Boolean sets with all $k$-projections surjective. Our motivation arises from understanding the relationship between a network's interaction graph and its fixed points, which deepens the classical paradigm of research in this direction by incorporating a particular structure on the set of fixed points, beyond merely observing their quantity. Specifically, among the results of this paper, we highlight a condition on the in-degree of the interaction graph for a network to be $k$-independent, we show that all regulatory networks are at most $n/2$-independent, and we construct $k$-independent networks for all possible $k$ in the case of monotone networks with a complete interaction graph.
New formulas involving Bernoulli and Stirling numbers of both kinds
This paper is devoted to establishing several new formulas relating Bernoulli and Stirling numbers of both kinds.
A note on Two-Point Concentration of the Independence Number of $G_{n,m}$
We show that the independence number of $ G_{n,m}$ is concentrated on two values for $ n^{5/4+ ε} < m \le \binom{n}{2}$. This result establishes a distinction between $G_{n,m}$ and $G_{n,p}$ with $p = m/ \binom{n}{2}$ in the regime $ n^{5/4 + ε} < m< n^{4/3}$. In this regime the independence number of $ G_{n,m}$ is concentrated on two values while the independence number of $ G_{n,p}$ is not; indeed, for $p$ in this regime variations in $ α( G_{n,p})$ are determined by variations in the number of edges in $ G_{n,p}$.
Neighborhood Balanced 3-Coloring
A graph is said to be neighborhood 3-balanced if there exists a vertex labeling with three colors so that each vertex has an equal number of neighbors of each color. We give order constraints on 3-balanced graphs, determine which generalized Petersen and Pappus graphs are 3-balanced, discuss when being 3-balanced is preserved under various graph constructions, give two general characterizations of cubic 3-balanced graphs, and classify cubic 3-balanced graphs of small order.
Characterizing real-representable matroids with large average hyperplane-size
Generalizing a theorem of the first two authors and Geelen for planes, we show that, for a real-representable matroid $M$, either the average hyperplane-size in $M$ is at most a constant depending only on its rank, or each hyperplane of $M$ contains one of a set of at most $r(M)-2$ lines. Additionally, in the latter case, the ground set of $M$ has a partition $(E_{1}, E_{2})$, where $E_{1}$ can be covered by few flats of relatively low rank and $|E_{2}|$ is bounded. These results extend to complex-representable and orientable matroids. Finally, we formulate a high-dimensional generalization of a classic problem of Motzkin, Grünbaum, Erdős and Purdy on sets of red and blue points in the plane with no monochromatic blue line. We show that the solution to this problem gives a tight upper bound on $|E_{2}|$. We also discuss this high-dimensional problem in its own right, and prove some initial results.
On the local convergence of integer-valued Lipschitz functions on regular trees
Published
• View Publication
• BIB
We study random integer-valued Lipschitz functions on regular trees. It was shown by Peled, Samotij and Yehudayoff that such functions are localized, however, finer questions about the structure of Gibbs measures remain unanswered. Our main result is that the weak limit of a uniformly chosen 1-Lipschitz function with 0 boundary condition on a $d$-ary tree of height $n$ exists as $n \to \infty$ if $2 \le d \le 7$, but not if $d \ge 8$, thereby partially answering a question posed by Peled, Samotij and Yehudayoff. For large $d$, the value at the root alternates between being almost entirely concentrated on 0 for even $n$ and being roughly uniform on $\{-1,0,1\}$ for odd $n$, leading to different limits as $n$ approaches infinity along evens or odds. For $d \ge 8$, the essence of this phenomenon is preserved, which obstructs the convergence. For $d \le 7$, this phenomenon ceases to exist, and the law of the value at the root loses its connection with the parity of $n$. Along the way, we also obtain an alternative proof of localization. The key idea is a fixed point convergence result for a related operator on $\ell^\infty$, and a procedure to show that the iterations get into a `basin of attraction' of the fixed point. We also prove some accompanying analogous `even-odd phenomenon' type results about $M$-lipschitz functions on general non-amenable graphs with high enough expansion (this includes for example the large $d$ case for regular trees). We also prove a convergence result for 1-Lipschitz functions with $\{0,1\}$ boundary condition. This last result relies on an absolute value FKG for uniform 1-Lipschitz functions when shifted by $1/2$.
Misere Connect Four is Solved
Published
• View Publication
• BIB
Connect Four is a two-player game where each player attempts to be the first to create a sequence of four of their pieces, arranged horizontally, vertically, or diagonally, by dropping pieces into the columns of a grid of width seven and height six, in alternating turns. Misere Connect Four is played by the same rules, but with the opposite objective: do not connect four. This paper announces that Misere Connect Four is solved: perfect play by both sides leads to a second-player win. More generally, this paper also announces that Misere Connect $k$ played on a $w \times h$ board is also solved, but the outcome depends on the game's parameters $k$, $w$, and $h$, and may be a first-player win, a second-player win, or a draw. These results are constructive, meaning that we provide explicit strategies, thus enabling readers to impress their friends and foes alike with provably optimal play in the misere form of a table-top game for children.
On multivalued groups of order 3
Published
• View Publication
• BIB
A complete classification of the multivalued coset groups of order $3$ is given. The proof is based on the classification of rank $3$ groups having regular normal subgroups.
Multipartite tournaments in which any two vertices have an $(i,j)$-step common out-neighbor
We say that a digraph $D$ is $(i,j)$-step competitive if any two vertices have an $(i,j)$-step common out-neighbor in $D$ and that a graph $G$ is $(i,j)$-step competitively orientable if there exists an $(i,j)$-step competitive orientation of $G$.
In [Choi et al. Competitively orientable complete multipartite graphs. Discrete Mathematics, 345(9):112950, 2022], Choi et al. introduce the notion of competitive digraph and completely characterize competitively orientable complete multipartite graphs in terms of the sizes of its partite sets. Here, a competitive digraph means a $(1,1)$-step competitive digraph. In this paper, the result of Choi et al. has been extended to a general characterization of $(i,j)$-step competitively orientable complete multipartite graphs.
Non-conflicting no-where zero $Z_2\times Z_2$ flows in cubic graphs
Published in AUSTRALASIAN JOURNAL OF COMBINATORICS 91(3), 2025, 392--413
• Search Publication
Let $Z_2\times Z_2=\{0, α, β, α+β\}$. If $G$ is a bridgeless cubic graph, $F$ is a perfect matching of $G$ and $\overline{F}$ is the complementary 2-factor of $F$, then a no-where zero $Z_2\times Z_2$-flow $θ$ of $G/\overline{F}$ is called non-conflicting with respect to $\overline{F}$, if $\overline{F}$ contains no edge $e=uv$, such that $u$ is incident to an edge with $θ$-value $α$ and $v$ is incident to an edge with $θ$-value $β$. In this paper, we demonstrate the usefulness of non-conflicting flows by showing that if a cubic graph $G$ admits such a flow with respect to some perfect matching $F$, then $G$ admits a normal 6-edge-coloring. We use this observation in order to show that claw-free bridgeless cubic graphs, bridgeless cubic graphs possessing a 2-factor having at most two cycles admit a normal 6-edge-coloring. We demonstrate the usefulness of non-conflicting flows further by relating them to a recent conjecture of Thomassen about edge-disjoint perfect matchings in highly connected regular graphs. In the end of the paper, we construct infinitely many 2-edge-connected cubic graphs such that $G/\overline{F}$ does not admit a non-conflicting no-where zero $Z_2\times Z_2$-flow with respect to any perfect matching $F$.
A combinatorial approach to nonlinear spectral gaps
A seminal open question of Pisier and Mendel--Naor asks whether every degree-regular graph which satisfies the classical discrete Poincaré inequality for scalar functions, also satisfies an analogous inequality for functions taking values in \textit{any} normed space with non-trivial cotype. Motivated by applications, it is also greatly important to quantify the dependence of the corresponding optimal Poincaré constant on the cotype $q$. Works of Odell--Schlumprecht (1994), Ozawa (2004), and Naor (2014) make substantial progress on the former question by providing a positive answer for normed spaces which also have an unconditional basis, in addition to finite cotype. However, little is known in the way of quantitative estimates: the mentioned results imply a bound on the Poincaré constant depending super-exponentially on $q$.
We introduce a novel combinatorial framework for proving quantitative nonlinear spectral gap estimates. The centerpiece is a property of regular graphs that we call \emph{long range expansion}, which holds with high probability for random regular graphs. Our main result is that any regular graph with the long-range expansion property satisfies a discrete Poincaré inequality for any normed space with an unconditional basis and cotype $q$, with a Poincaré constant that depends \emph{polynomially} on $q$, which is optimal. As an application, any normed space with an unconditional basis which admits a low distortion embedding of an $n$-vertex random regular graph, must have cotype at least polylogarithmic in $n$. This extends a celebrated lower-bound of Matoušek for low distortion embeddings of random graphs into $\ell_q$ spaces.
Log-Concave Sequences in Coding Theory
Published
• View Publication
• BIB
We introduce the notion of logarithmically concave (or log-concave) sequences in Coding Theory. A sequence $a_0, a_1, \dots, a_n$ of real numbers is called log-concave if $a_i^2 \ge a_{i-1}a_{i+1}$ for all $1 \le i \le n-1$. A natural sequence of positive numbers in coding theory is the weight distribution of a linear code consisting of the nonzero values among $A_i$'s where $A_i$ denotes the number of codewords of weight $i$. We call a linear code log-concave if its nonzero weight distribution is log-concave. Our main contribution is to show that all binary general Hamming codes of length $2^r -1$ ($r=3$ or $r \ge 5$), the binary extended Hamming codes of length $2^r ~(r \ge 3)$, and the second order Reed-Muller codes $R(2, m)~ (m \ge 2)$ are all log-concave while the homogeneous and projective second order Reed-Muller codes are either log-concave, or 1-gap log-concave. Furthermore, we show that any MDS $[n, k]$ code over $\mathbb F_q$ satisfying $3 \leqslant k \leqslant n/2 +3$ is log-concave if $q \geqslant q_0(n, k)$ which is the larger root of a quadratic polynomial. Hence, we expect that the concept of log-concavity in coding theory will stimulate many interesting problems.