arXiv subjects

Julian Sahasrabudhe

Publications and source records attributed to Julian Sahasrabudhe.

36 records · Page 2Linked to original sources

arXiv2020

Random polynomials: the closest roots to the unit circle

Let $f = \sum_{k=0}^n \varepsilon_k z^k$ be a random polynomial, where $\varepsilon_0,\ldots ,\varepsilon_n$ are iid standard Gaussian random variables, and let $ζ_1,\ldots,ζ_n$ denote the roots of $f$. We show that the point process determined by the magnitude of the roots $\{ 1-|ζ_1|,\ldots, 1-|ζ_n| \}$ tends to a Poisson point process at the scale $n^{-2}$ as $n\rightarrow \infty$. One consequence of this result is that it determines the magnitude of the closest root to the unit circle. In particular, we show that \[ \min_{k} ||ζ_k| - 1|n^2 \rightarrow \mathrm{Exp}(1/6),\] in distribution, where $\mathrm{Exp}(λ)$ denotes an exponential random variable of mean $λ^{-1}$. This resolves a conjecture of Shepp and Vanderbei from 1995 that was later studied by Konyagin and Schlag.

math.PR↗

arXiv2020

Cosine polynomials with few zeros

In a celebrated paper, Borwein, Erdélyi, Ferguson and Lockhart constructed cosine polynomials of the form \[ f_A(x) = \sum_{a \in A} \cos(ax), \] with $A\subseteq \mathbb{N}$, $|A|= n$ and as few as $n^{5/6+o(1)}$ zeros in $[0,2π]$, thereby disproving an old conjecture of J.E. Littlewood. Here we give a sharp analysis of their constructions and, as a result, prove that there exist examples with as few as $C(n\log n)^{2/3}$ roots.

math.CA↗

arXiv2019

Central limit theorems and the geometry of polynomials

Let $X \in \{0,\ldots,n \}$ be a random variable, with mean $μ$ and standard deviation $σ$ and let \[f_X(z) = \sum_{k} \mathbb{P}(X = k) z^k, \] be its probability generating function. Pemantle conjectured that if $σ$ is large and $f_X$ has no roots close to $1\in \mathbb{C}$ then $X$ must be approximately normal. We completely resolve this conjecture in the following strong quantitative form, obtaining sharp bounds. If $δ= \min_ζ|ζ-1|$ over the complex roots $ζ$ of $f_X$, and $X^{\ast} := (X-μ)/σ$, then \[ \sup_{t \in \mathbb{R}} \left|\mathbb{P}(X^{\ast} \leq t) - \mathbb{P}( Z \leq t) \, \right| = O\left(\frac{\log n}{δσ} \right) \] where $Z \sim \mathcal{N}(0,1)$ is a standard normal. This gives the best possible version of a result of Lebowitz, Pittel, Ruelle and Speer. We also show that if $f_X$ has no roots with small argument, then $X$ must be approximately normal, again in a sharp quantitative form: if we set $δ= \min_ζ|\arg(ζ)|$ then \[ \sup_{t \in \mathbb{R}} \left|\mathbb{P}(X^{\ast} \leq t) - \mathbb{P}( Z \leq t) \, \right| = O\left(\frac{1}{δσ} \right). \] Using this result, we answer a question of Ghosh, Liggett and Pemantle by proving a sharp multivariate central limit theorem for random variables with real-stable probability generating functions.

math.PR↗

arXiv2019

Flat Littlewood Polynomials Exist

We show that there exist absolute constants $Δ> δ> 0$ such that, for all $n \geqslant 2$, there exists a polynomial $P$ of degree $n$, with $\pm 1$ coefficients, such that $$δ\sqrt{n} \leqslant |P(z)| \leqslant Δ\sqrt{n}$$ for all $z\in\mathbb{C}$ with $|z|=1$. This confirms a conjecture of Littlewood from 1966.

math.CA↗

arXiv2019

Counting Zeros of Cosine Polynomials: On a Problem of Littlewood

We show that if $A$ is a finite set of non-negative integers then the number of zeros of the function \[ f_A(θ) = \sum_{a \in A} \cos(aθ), \] in $[0,2π]$, is at least $(\log \log \log |A|)^{1/2-\varepsilon}$. This gives the first unconditional lower bound on a problem of Littlewood, solves a conjecture of Borwein, Erdélyi, Ferguson and Lockhart and improves upon work of Borwein and Erdélyi. We also prove a result that applies to more general cosine polynomials with "few" distinct rational coefficients. One of the main ingredients in the proof is perhaps of independent interest: we show that if $f$ is an exponential polynomial with "few" distinct integer coefficients and $f$ "correlates" with a low-degree exponential polynomial $P$, then $f$ has a very particular structure.

math.CA↗

arXiv2019

The Erdős-Selfridge problem with square-free moduli

A covering system is a finite collection of arithmetic progressions whose union is the set of integers. The study of covering systems with distinct moduli was initiated by Erdős in 1950, and over the following decades numerous problems were posed regarding their properties. One particularly notorious question, due to Erdős, asks whether there exist covering systems whose moduli are distinct and all odd. We show that if in addition one assumes the moduli are square-free, then there must be an even modulus.

math.NT↗

arXiv2018

On the Erdős Covering Problem: the density of the uncovered set

Since their introduction by Erdős in 1950, covering systems (that is, finite collections of arithmetic progressions that cover the integers) have been extensively studied, and numerous questions and conjectures have been posed regarding the existence of covering systems with various properties. In particular, Erdős asked if the moduli can be distinct and all arbitrarily large, Erdős and Selfridge asked if the moduli can be distinct and all odd, and Schinzel conjectured that in any covering system there exists a pair of moduli, one of which divides the other. Another beautiful conjecture, proposed by Erdős and Graham in 1980, states that if the moduli are distinct elements of the interval $[n,Cn]$, and $n$ is sufficiently large, then the density of integers uncovered by the union is bounded below by a constant (depending only on $C$). This conjecture was confirmed (in a strong form) by Filaseta, Ford, Konyagin, Pomerance and Yu in 2007, who moreover asked whether the same conclusion holds if the moduli are distinct and sufficiently large, and $\sum_{i=1}^k \frac{1}{d_i} < C$. Although this condition turns out not to be sufficiently strong to imply the desired conclusion, as the main result of this paper we will give an essentially best possible condition which is sufficient. Our method has a number of further applications. Most importantly, we prove the conjecture of Schinzel stated above, which was made in 1967. We moreover give an alternative (somewhat simpler) proof of a breakthrough result of Hough, who resolved Erdős' minimum modulus problem, with an improved bound on the smallest difference. Finally, we make further progress on the problem of Erdős and Selfridge.

math.NT↗

arXiv2018

A Stability Theorem for Maximal $K_{r+1}$-free Graphs

For $r \geq 2$, we show that every maximal $K_{r+1}$-free graph $G$ on $n$ vertices with $(1-\frac{1}{r})\frac{n^2}{2}-o(n^{\frac{r+1}{r}})$ edges contains a complete $r$-partite subgraph on $(1 - o(1))n$ vertices. We also show that this is best possible. This result answers a question of Tyomkyn and Uzzell.

math.CO↗

arXiv2018

Central limit theorems from the roots of probability generating functions

For each $n$, let $X_n \in \{0,\ldots,n\}$ be a random variable with mean $μ_n$, standard deviation $σ_n$, and let \[ P_n(z) = \sum_{k=0}^n \mathbb{P}( X_n = k) z^k ,\] be its probability generating function. We show that if none of the complex zeros of the polynomials $\{ P_n(z)\}$ are contained in a neighbourhood of $1 \in \mathbb{C}$ and $σ_n > n^{\varepsilon}$ for some $\varepsilon >0$, then $ X_n^* =(X_n - μ_n)σ^{-1}_n$ tends to a normal random variable $Z \sim \mathcal{N}(0,1)$ in distribution as $n \rightarrow \infty$. Moreover, we show this result is sharp in the sense that there exist sequences of random variables $\{X_n\}$ with $σ_n > C\log n $ for which $P_n(z)$ has no roots near $1$ and $X_n^*$ is not asymptotically normal. These results disprove a conjecture of Pemantle and improve upon various results in the literature. We go on to prove several other results connecting the location of the zeros of $P_n(z)$ and the distribution of the random variables $X_n$.

math.PR↗

arXiv2018

Dense Subgraphs in Random Graphs

For a constant $γ\in[0,1]$ and a graph $G$, let $ω_γ(G)$ be the largest integer $k$ for which there exists a $k$-vertex subgraph of $G$ with at least $γ\binom{k}{2}$ edges. We show that if $0<p<γ<1$ then $ω_γ(G_{n,p})$ is concentrated on a set of two integers. More precisely, with $α(γ,p)=γ\log\fracγ{p}+(1-γ)\log\frac{1-γ}{1-p}$, we show that $ω_γ(G_{n,p})$ is one of the two integers closest to $\frac{2}{α(γ,p)}\big(\log n-\log\log n+\log\frac{eα(γ,p)}{2}\big)+\frac{1}{2}$, with high probability. While this situation parallels that of cliques in random graphs, a new technique is required to handle the more complicated ways in which these "quasi-cliques" may overlap.

math.CO↗

arXiv2017

Ramsey graphs induce subgraphs of many different sizes

A graph on $n$ vertices is said to be \emph{$C$-Ramsey} if every clique or independent set of the graph has size at most $C \log n$. The only known constructions of Ramsey graphs are probabilistic in nature, and it is generally believed that such graphs possess many of the same properties as dense random graphs. Here, we demonstrate one such property: for any fixed $C>0$, every $C$-Ramsey graph on $n$ vertices induces subgraphs of at least $n^{2-o(1)}$ distinct sizes. This near-optimal result is closely related to two unresolved conjectures, the first due to Erdős and McKay and the second due to Erdős, Faudree and Sós, both from 1992.

math.CO↗

arXiv2017

On Existentially Complete Triangle-free Graphs

For a positive integer $k$, we say that a graph is $k$-existentially complete if for every $0 \leq a \leq k$, and every tuple of distinct vertices $x_1,\ldots,x_a$, $y_1,\ldots,y_{k-a}$, there exists a vertex $z$ that is joined to all of the vertices $x_1,\ldots,x_a$ and none of the vertices $y_1,\ldots,y_{k-a}$. While it is easy to show that the binomial random graph $G_{n,1/2}$ satisfies this property with high probability for $k \sim c\log n$, little is known about the "triangle-free" version of this problem; does there exist a finite triangle-free graph $G$ with a similar "extension property". This question was first raised by Cherlin in 1993 and remains open even in the case $k=4$. We show that there are no $k$-existentially complete triangle-free graphs with $k >\frac{8\log n}{\log\log n}$, thus giving the first non-trivial, non-existence result on this "old chestnut" of Cherlin. We believe that this result breaks through a natural barrier in our understanding of the problem.

math.CO↗

arXiv2017

Partitioning a graph into monochromatic connected subgraphs

A well-known result by Haxell and Kohayakawa states that the vertices of an $r$-coloured complete graph can be partitioned into $r$ monochromatic connected subgraphs of distinct colours; this is a slightly weaker variant of a conjecture by Erdős, Pyber and Gyárfás that states that there exists a partition into $r-1$ monochromatic connected subgraphs. We consider a variant of this problem, where the complete graph is replaced by a graph with large minimum degree, and prove two conjectures of Bal and DeBiasio, for two and three colours.

math.CO↗

arXiv2016

Exponential Patterns in Arithmetic Ramsey Theory

We show that for every finite colouring of the natural numbers there exists $a,b >1$ such that the triple $\{a,b,a^b\}$ is monochromatic. We go on to show the partition regularity of a much richer class of patterns involving exponentiation. For example, as a corollary to our main theorem, we show that for every $n \in \mathbb{N}$ and for every finite colouring of the natural numbers, we may find a monochromatic set including the integers $x_1,\ldots,x_n >1$; all products of distinct $x_i$; and all "exponential compositions" of distinct $x_i$ which respect the order $x_1,\ldots,x_n$. In particular, for every finite colouring of the natural numbers one can find a monochromatic quadruple of the form $\{ a,b,ab,a^b \}$, where $a,b>1$.

math.CO↗

arXiv2016

The multiplication table problem for bipartite graphs

We investigate the following generalisation of the 'multiplication table problem' of Erdős: given a bipartite graph with $m$ edges, how large is the set of sizes of its induced subgraphs? Erdős's problem of estimating the number of distinct products $ab$ with $a,b \le n$ is precisely the problem under consideration when the graph in question is the complete bipartite graph $K_{n,n}$. In this note, we prove that the set of sizes of the induced subgraphs of any bipartite graph with $m$ edges contains $Ω(m/(\log m)^{12})$ distinct elements.

math.CO↗

arXiv2016

Monochromatic Solutions to Systems of Exponential Equations

Let $n\in \mathbb{N}$, $R$ be a binary relation on $[n]$, and $C_1(i,j),\ldots,C_n(i,j) \in \mathbb{Z}$, for $i,j \in [n]$. We define the exponential system of equations $\mathcal{E}(R,(C_k(i,j)_{i,j,k})$ to be the system \[ X_i^{Y_1^{C_1(i,j)} \cdots Y_n^{C_n(i,j)} } = X_j , \text{ for } (i,j) \in R ,\] in variables $X_1,\ldots,X_n,Y_1,\ldots,Y_n$. The aim of this paper is to classify precisely which of these systems admit a monochromatic solution ($X_i,Y_i \not=1)$ in an arbitrary finite colouring of the natural numbers. This result could be viewed as an analogue of Rado's theorem for exponential patterns.

math.CO↗

arXiv2015

On the maximum running time in graph bootstrap percolation

Graph bootstrap percolation is a simple cellular automaton introduced by Bollobás in 1968. Given a graph $H$ and a set $G \subseteq E(K_n)$ we initially "infect" all edges in $G$ and then, in consecutive steps, we infect every $e \in K_n$ that completes a new infected copy of $H$ in $K_n$. We say that $G$ percolates if eventually every edge in $K_n$ is infected. The extremal question about the size of the smallest percolating sets when $H = K_r$ was answered independently by Alon, Kalai and Frankl. Here we consider a different question raised more recently by Bollobás: what is the maximum time the process can run before it stabilizes? It is an easy observation that for $r=3$ this maximum is $\lceil \log_2 (n-1) \rceil $. However, a new phenomenon occurs for $r=4$ when, as we show, the maximum time of the process is $n-3$. For $r \geq 5$ the behaviour of the dynamics is even more complex, which we demonstrate by showing that the $K_r$-bootstrap process can run for at least $n^{2-\varepsilon_r}$ time steps for some $\varepsilon_r$ that tends to $0$ as $r \to \infty$.

math.CO↗

arXiv2011

On abelian and additive complexity in infinite words

The study of the structure of infinite words having bounded abelian complexity was initiated by G. Richomme, K. Saari, and L. Q. Zamboni. In this note we define bounded additive complexity for infinite words over a finite subset of Z^m. We provide an alternative proof of one of the results of Richomme, Saari, and Zamboni.

math.CO↗