random
6952 papers tagged with this keyword
Exponential Lower Bounds for Many Pivot Rules for the Simplex Method
The existence of a pivot rule for the simplex method that guarantees a strongly polynomial run-time is a longstanding, fundamental open problem in the theory of linear programming. The leading pivot rule in theory is the shadow pivot rule, which solves a linear program by projecting the feasible region onto a polygon. It has been shown to perform in expected strongly polynomial time on uniformly random instances and in smoothed analysis. In practice, the pivot rule of choice is the steepest edge rule, which normalizes the set of improving neighbors and then chooses a maximally improving normalized neighbor. Exponential lower bounds are known for both rules in worst-case analysis. However, for the shadow simplex method, all exponential examples were only proven for one choice of projection, and for the steepest edge rule, the lower bounds were only proven for the Euclidean norm. In this work, we construct linear programs for which any choice of projection for shadow rule variants will lead to an exponential run-time and exponential examples for any choice of norm for a steepest edge variant.
A class of polynomial recurrences resulting in $(n/\log n, n/\log^2n)$-asymptotic normality
Published
• View Publication
• BIB
We consider sequences of polynomials that satisfy differential-difference recurrences. Polynomials satisfying such recurrences frequently appear as generating polynomials of integer valued random variables that are of interest in discrete mathematics. It is, therefore, of interest to understand the properties of such polynomials and their probabilistic consequences. We identify a class of polynomial recurrences that lead to a normal law with the expected value and the variance proportional to $n/\log n$ and $n/\log^2n$, respectively. Examples include Stirling number of the second kind and other polynomials concerning set partitions as well as polynomials related to Whitney numbers of Dowling lattices.
Bounds for Rainbow-uncommon Graphs
We say a graph $H$ is $r$-rainbow-uncommon if the maximum number of rainbow copies of $H$ under an $r$-coloring of $E(K_n)$ is asymptotically (as $n \to \infty$) greater than what is expected from uniformly random $r$-colorings. Via explicit constructions, we show that for $H\in\{K_3,K_4, K_5\}$, $H$ is $r$-rainbow-uncommon for all $r\geq {|V(H)|\choose 2}$. We also construct colorings to show that for $t \geq 6$, $K_t$ is $r$-rainbow-uncommon for sufficiently large $r$.
Monotonicity, Topology, and Convexity of Recurrence in Random Walks
We consider non-homogeneous random walks on the two-dimensional positive quadrant $\mathbb{N}^2$ and the one-dimensional slab $\{0,1,\dots,k\}\times\mathbb{N}$. In the 1960's the following question was asked for $\mathbb{N}^2$: is it true if such a random walk $X$ is recurrent and $Y$ is another random walk that at every point is more likely to go down and more likely to go left than $Y$, then $Y$ is also recurrent?
We provide an example showing that the answer is negative. We also show, via a coupling argument, that if either the random walk $X$ or $Y$ is sufficiently homogeneous then the answer is in fact positive. In addition, we show using the Rayleigh monotonicity principle that the analogous question for random walks on trees is positive.
These results show that the subset of parameter space that yields recurrent random walks possesses some geometric properties, in this case the structure of an order ideal. Motivated by this perspective, we consider the more symmetric setting of homogeneous random walks on finitely generated abelian groups, and ask when this subset possesses other geometric properties, namely various topological properties and convexity. We answer some of these questions: in particular, we show that this subset is closed, and under a symmetric support condition, show it is path-connected and additionally show it is convex if and only if its effective dimension is at most 2. We also show its complement is in some sense typically path-connected but not convex. We finally propose some related open problems.
Better-than-average uniform random variables and Eulerian numbers, or: How many candidates should a voter approve?
Consider $n$ independent random numbers with a uniform distribution on $[0,1]$. The number of them that exceed their mean is shown to have an Eulerian distribution, i.e., it is described by the Eulerian numbers. This is related to, but distinct from, the well known fact that the integer part of the sum of independent random numbers uniform on $[0,1]$ has an Eulerian distribution. One motivation for this problem comes from voting theory.
Seymour's second neighbourhood conjecture: random graphs and reductions
A longstanding conjecture of Seymour states that in every oriented graph there is a vertex whose second outneighbourhood is at least as large as its outneighbourhood. In this short note we show that, for any fixed $p\in[0,1/2)$, a.a.s. every orientation of $G(n,p)$ satisfies Seymour's conjecture (as well as a related conjecture of Sullivan). This improves on a recent result of Botler, Moura and Naia. Moreover, we show that $p=1/2$ is a natural barrier for this problem, in the following sense: for any fixed $p\in(1/2,1)$, Seymour's conjecture is actually equivalent to saying that, with probability bounded away from $0$, every orientation of $G(n,p)$ satisfies Seymour's conjecture. This provides a first reduction of the problem.
For a second reduction, we consider minimum degrees and show that, if Seymour's conjecture is false, then there must exist arbitrarily large strongly-connected counterexamples with bounded minimum outdegree. Contrasting this, we show that vertex-minimal counterexamples must have large minimum outdegree.
Galled Tree-Child Networks
We propose the class of galled tree-child networks which is obtained as intersection of the classes of galled networks and tree-child networks. For the latter two classes, (asymptotic) counting results and stochastic results have been proved with very different methods. We show that a counting result for the class of galled tree-child networks follows with similar tools as used for galled networks, however, the result has a similar pattern as the one for tree-child networks. In addition, we also consider the (suitably scaled) numbers of reticulation nodes of random galled tree-child networks and show that they are asymptotically normal distributed. This is in contrast to the limit laws of the corresponding quantities for galled networks and tree-child networks which have been both shown to be discrete.
The clique chromatic number of sparse random graphs
The clique chromatic number of a graph is the smallest number of colors in a vertex coloring so that no maximal clique is monochromatic. In this paper, we determine the order of magnitude of the clique chromatic number of the random graph G_{n,p} for most edge-probabilities p in the range n^{-2/5} \ll p \ll 1. This resolves open problems and questions of Lichev, Mitsche and Warnke as well as Alon and Krievelevich.
One major proof difficulty stems from high-degree vertices, which prevent maximal cliques in their neighborhoods: we deal with these vertices by an intricate union bound argument, that combines the probabilistic method with new degree counting arguments in order to enable Janson's inequality. This way we determine the asymptotics of the clique chromatic number of G_{n,p} in some ranges, and discover a surprising new phenomenon that contradicts earlier predictions for edge-probabilities p close to n^{-2/5}.
Binary search trees of permuton samples
Published in Advances in Applied Mathematics, Volume 162, January 2025, 102774
• View Publication
• BIB
Binary search trees (BST) are a popular type of data structure when dealing with ordered data. Indeed, they enable one to access and modify data efficiently, with their height corresponding to the worst retrieval time. From a probabilistic point of view, binary search trees associated with data arriving in a uniform random order are well understood, but less is known when the input is a non-uniform random permutation. We consider here the case where the input comes from i.i.d. random points in the plane with law $μ$, a model which we refer to as a permuton sample. Our results show that the asymptotic proportion of nodes in each subtree depends on the behavior of the measure $μ$ at its left boundary, while the height of the BST has a universal asymptotic behavior for a large family of measures $μ$. Our approach involves a mix of combinatorial and probabilistic tools, namely combinatorial properties of binary search trees, coupling arguments, and deviation estimates.
Random Generation of Git Graphs
Published in GASCom 2024, Jun 2024, Talence, France, France
• Search Publication
Version Control Systems, such as Git and Mercurial, manage the history of a project as a Directed Acyclic Graph encoding the various divergences and synchronizations happening in its life cycle. A popular workflow in the industry, called the feature branch workflow, constrains these graphs to be of a particular shape: a unique main branch, and non-interfering feature branches. Here we focus on the uniform random generation of those graphs with n vertices, including k on the main branch, for which we provide three algorithms, for three different use-cases. The first, based on rejection, is efficient when aiming for small values of k (more precisely whenever k = O($\sqrt$ n)). The second takes as input any number k of commits in the main branch, but requires costly precalculation. The last one is a Boltzmann generator and enables us to generate very large graphs while targeting a constant k/n ratio. All these algorithms are linear in the size of their outputs.
Balls in groups: volume, structure and growth
We give sharp bounds in Breuillard, Green and Tao's finitary version of Gromov's theorem on groups with polynomial growth. Precisely, we show that for every non-negative integer d there exists $c=c(d)>0$ such that if $G$ is a group with finite symmetric generating set $S$ containing the identity and $|S^n|\le cn^{d+1}|S|$ for some positive integer $n$ then there exist normal subgroups $H\leΓ\le G$ such that $H\subseteq S^n$, such that $Γ/H$ is $d$-nilpotent (i.e. has a central series of length $d$ with cyclic factors), and such that $[G:Γ]\le g(d)$, where $g(d)$ denotes the maximum order of a finite subgroup of $GL_d(\mathbb{Z})$. The bounds on both the nilpotence and index are sharp; the previous best bounds were $O(d)$ on the nilpotence, and an ineffective function of $d$ on the index. In fact, we obtain this as a small part of a much more detailed fine-scale description of the structure of $G$. These results have a wide range of applications in various aspects of the theory of vertex-transitive graphs: percolation theory, random walks, structure of finite groups, scaling limits of finite vertex-transitive graphs.... We obtain some of these applications in the present paper, and treat others in companion papers. Some are due to or joint with other authors.
The $k$-representation number of the random graph
The $k$-representation number of a graph $G$ is the minimum cardinality of the system of vertex subsets with the property that every edge of $G$ is covered at least $k$ times while every non-edge is covered at most $(k-1)$ times. In particular, for $k=1$ this notion is equivalent to the clique number of a graph $G$. Extending results of Frieze and Reed, and Eaton and Grable, we study the $k$-representation number of $G(n,1/2)$. As a tool, we will prove a sharp concentration result counting the number of induced subgraphs of $G(n,1/2)$ with density $(\frac{1}{2}+α)$. In Lemma 3.7, we will show that the number of such subgraphs is close to its expected value with probability $1-\exp(-n^C)$.
Circulant graphs with valency up to 4 that admit perfect state transfer in Grover walks
We completely characterize circulant graphs with valency up to $4$ that admit perfect state transfer. Those of valency $3$ do not admit it. On the other hand, circulant graphs with valency $4$ admit perfect state transfer only in two infinite families: one discovered by Zhan and another new family, while no others do. The main tools for deriving these results are symmetry of graphs and eigenvalues. We describe necessary conditions for perfect state transfer to occur based on symmetry of graphs, which mathematically refers to automorphisms of graphs. As for eigenvalues, if perfect state transfer occurs, then certain eigenvalues of the corresponding isotropic random walks must be the halves of algebraic integers. Taking this into account, we utilize known results on the rings of integers of cyclotomic fields.
The Maker-Breaker percolation game on a random board
The $(m,b)$ Maker-Breaker percolation game on $(\mathbb{Z}^2)_p$, introduced by Day and Falgas-Ravry, is played in the following way. Before the game starts, each edge of $\mathbb{Z}^2$ is removed independently with probability $1-p$. After that, Maker chooses a vertex $v_0$ to protect. Then, in each round Maker and Breaker claim respectively $m$ and $b$ unclaimed edges of $G$. Breaker wins if after the removal of the edges claimed by him the component of $v_0$ becomes finite, and Maker wins if she can indefinitely prevent Breaker from winning.
We show that for any $p < 1$, Breaker almost surely has a wining strategy for the $(1,1)$ game on $(\mathbb{Z}^2)_p$. This fully answers a question of Day and Falgas-Ravry, who showed that for $p = 1$ Maker has a winning strategy for the $(1,1)$ game. Further, we show that in the $(2,1)$ game on $(\mathbb{Z}^2)_p$ Maker almost surely has a winning strategy whenever $p > 0.9402$, while Breaker almost surely has a winning strategy whenever $p < 0.5278$. This shows that the threshold value of $p$ above which Maker has a winning strategy for the $(2,1)$ game on $\mathbb{Z}^2$ is non-trivial. In fact, we prove similar results in various settings, including other lattices and biases $(m,b)$.
These results extend also to the most general case, which we introduce, where each edge is given to Maker with probability $α$ and to Breaker with probability $β$ before the game starts.
Clique Decompositions in Random Graphs via Refined Absorption
We prove that if $p\ge n^{-\frac{1}{3}+β}$ for some $β> 0$, then asymptotically almost surely the binomial random graph $G(n,p)$ has a $K_3$-packing containing all but at most $n + O(1)$ edges. Similarly, we prove that if $d \ge n^{\frac{2}{3}+β}$ for some $β> 0$ and $d$ is even, then asymptotically almost surely the random $d$-regular graph $G_{n,d}$ has a triangle decomposition provided $3 \mid d \cdot n$. We also show that $G(n,p)$ admits a fractional $K_3$-decomposition for such a value of $p$. We prove analogous versions for a $K_q$-packing of $G(n,p)$ with $p\ge n^{-\frac{1}{q+0.5}+β}$ and leave of $(q-2)n+O(1)$ edges, for $K_q$-decompositions of $G_{n,d}$ with $(q-1)~|~d$ and $d\ge n^{1-\frac{1}{q+0.5}+β}$ provided $q\mid d\cdot n$, and for fractional $K_q$-decompositions.
Thresholds for $(n,q,2)$-Steiner Systems via Refined Absorption
We prove that if $p \geq n^{-(q-6)/2}$, then asymptotically almost surely the binomial random $q$-uniform hypergraph $G^{(q)}(n,p)$ contains an $(n,q,2)$-Steiner system, provided $n$ satisfies the necessary divisibility conditions.
A Conditional Extension of the Park-Pham Theorem
Neglecting many motivating details for the Park-Pham theorem (previously known as the Kahn-Kalai conjecture), the result starts with a finite set $X$, a non-trivial upper set $\mathcal{F} \subseteq 2^X$, and a particular parameterized family of random variables $X_p$, then proceeds to provide an upper bound on the value $p_c(\mathcal{F})$ such that $\mathbb{P}(X_{p_c(\mathcal{F})} \in \mathcal{F}) = 1/2$. A natural question to ask is if there is an analog to the Park-Pham theorem for upper sets in finite posets other than $2^X$ and other parameterized families of random variables taking values in them. In this short note, we show that there is, with minor adjustments, in at least one circumstance. This is done by producing a conditional version of the $ε$-dependent form of the Park-Pham theorem, which has practical implications for the study of large neural networks and may also be interesting in its own right.
Marginal Independence and Partial Set Partitions
We establish a bijection between marginal independence models on $n$ random variables and split closed order ideals in the poset of partial set partitions. We also establish that every discrete marginal independence model is toric in cdf coordinates. This generalizes results of Boege, Petrovic, and Sturmfels and Drton and Richardson, and provides a unified framework for discussing marginal independence models. Additionally, we provide an axiomatic characterization of marginal independence and we show that our set of axioms are sound and complete in the set of probability distributions. This follows the work of Geiger, Paz and Pearl who provided an analogous characterization of independence for statements involving 2 sets of random variables.
Sparse Autoregressive Neural Networks for Classical Spin Systems
Published in Mach. Learn.: Sci. Technol. 5 025074 (2024)
• View Publication
• BIB
Efficient sampling and approximation of Boltzmann distributions involving large sets of binary variables, or spins, are pivotal in diverse scientific fields even beyond physics. Recent advances in generative neural networks have significantly impacted this domain. However, these neural networks are often treated as black boxes, with architectures primarily influenced by data-driven problems in computational science. Addressing this gap, we introduce a novel autoregressive neural network architecture named TwoBo, specifically designed for sparse two-body interacting spin systems. We directly incorporate the Boltzmann distribution into its architecture and parameters, resulting in enhanced convergence speed, superior free energy accuracy, and reduced trainable parameters. We perform numerical experiments on disordered, frustrated systems with more than 1000 spins on grids and random graphs, and demonstrate its advantages compared to previous autoregressive and recurrent architectures. Our findings validate a physically informed approach and suggest potential extensions to multivalued variables and many-body interaction systems, paving the way for broader applications in scientific research.
Symmetric functions and the explicit moment problem for abelian groups
Recently, Sawin and Wood (arXiv:math/2210.06279) proved a formula for the distribution of a random abelian group $G$ in terms of its $H$-moments $\mathbb{E} [\#\operatorname{Sur}(G,H)]$. We show that properties of Macdonald polynomials yield an alternate proof.