Searcharxiv⌕ Search

arXiv subjects

Maksim Zhukovskii

Publications and source records attributed to Maksim Zhukovskii.

At least 19 recordsLinked to original sources

Canonization of a random graph by two matrix-vector multiplications

We show that a canonical labeling of a random $n$-vertex graph can be obtained by assigning to each vertex $x$ the triple $(w_1(x),w_2(x),w_3(x))$, where $w_k(x)$ is the number of walks of length $k$ starting from $x$. This takes time $O(n^2)$, where $n^2$ is the input size, by using just two matrix-vector multiplications. The linear-time canonization of a random graph is the classical result of Babai, Erdős, and Selkow. For this purpose they use the well-known combinatorial color refinement procedure, and we make a comparative analysis of the two algorithmic approaches.

cs.CC↗

A Local Central Limit Theorem for Clique Counts in Sparse Random Graphs

Let $X_H$ denote the number of copies of a fixed graph $H$ in $G_{n, p}$. Gilmer and Kopparty conjectured that $X_H$ satisfies a local central limit theorem (LCLT) provided that $H$ is connected, $p \gg n^{-1/m(H)}$, and $n^2 (1-p) \gg 1$, where $m(H)$ is the maximum density. Following the work of Berkowitz, Sah and Sawhney confirmed this conjecture for every constant $p$, leaving the regime where $p=o(1)$ open. In this regime, the only case addressed in the literature is when $H=K_3$, where, in a recent paper, Araújo and Mattos confirmed the conjecture for $p \in (4n^{-1/2}, 1/2)$. This, together with a general result of Röllin and Ross, essentially settles the conjecture for the triangle. We generalise these results by showing that an LCLT holds for $H = K_r$ (for any fixed $r \ge 3$) in the regime $n^{-1/m(H)}\ll p\leq 1/2$, essentially settling the conjecture for cliques.

math.CO↗

The critical activation density in graph bootstrap percolation

In graph bootstrap percolation, edges of an Erdős-Rényi random graph ${\mathcal G}_{n,p}$ are initially active, and activation spreads to other edges of $K_n$ via the combinatorics of a fixed graph $H$: an edge becomes active whenever it is the unique inactive edge in a copy of $H$. The process $H$-percolates if all edges of $K_n$ are eventually activated. While classical cases such as $H=K_3$ (connectivity) and $H=K_4$ (related to $2$-neighbor bootstrap percolation) have been studied extensively, general graphs $H$ can exhibit wildly different behaviors. In this work, we determine the critical $H$-percolation threshold $p_c(n,H)$ for every graph $H$, fully resolving a longstanding open question of Balogh, Bollobás, and Morris. The location of $p_c(n,H)$ is governed by a new, universal parameter $ρ(H)$, which measures the maximal efficiency of witness graphs that activate an edge. To achieve this, we introduce a novel framework based on the unfolding and refolding of witness graphs. While previous works were restricted to specific families of $H$, our approach provides a unified strategy for all $H$. Inspired by algebraic topology, we lift witness graphs to covering graphs and algorithmically embed folded versions into ${\mathcal G}_{n,p}$ via a sequence of extensions. Crucially, this allows us to incorporate highly efficient witness graphs of unbounded size, which are potentially far larger than ${\mathcal G}_{n,p}$ itself. Beyond resolving $p_c(n,H)$, our framework recovers and strengthens several existing bounds in the literature. Finally, we initiate the study of the universal density parameter $ρ(H)$ and pose central open questions regarding its computability and its exact correspondence with the sharpness of the $H$-percolation threshold.

math.PR↗

Universality in random graphs via optimal linking systems: trees and beyond

We develop a framework for proving universality results in sparse random graphs. As a first application, we show that there exists an absolute constant $C>1$ such that, with high probability, for every fixed constant $Δ$, the binomial random graph $G(n,C\ln n/n)$ contains every $n$-vertex tree with maximum degree at most $Δ$. This answers a question of Montgomery (Advances in Mathematics, 2019). We also determine, for every $p$ satisfying $C\ln n/n\leq p=n^{-1+o(1)}$, the minimum girth $\ell$ (up to an absolute multiplicative constant) for which with high probability $G(n,p)$ contains all cycle factors of girth at least $Ω(\ell)$. In particular, with high probability $G(n,C\ln n/n)$ contains all cycle factors of girth at least $100\ln n/\ln\ln n$, which is optimal up to a constant factor. This extends an earlier result of Ferber, Kronenberg, and Luh (Transaction of the American Mathematical Society, 2019) and significantly improves a corollary of a deep result of Kahn, Lubetzky, and Wormald (Communications on Pure and Applied Mathematics, 2017). One of the key ingredients in the proofs is establishing the optimal depth of linking systems in sparse random graphs.

math.CO↗

New bounds for the optimal density of covering single-insertion codes via the Turán density

We prove that the density of any covering single-insertion code $C\subseteq X^r$ over the $n$-symbol alphabet $X$ cannot be smaller than $1/r+δ_r$ for some positive real $δ_r$ not depending on $n$. This improves the volume lower bound of $1/(r+1)$. On the other hand, we observe that, for all sufficiently large $r$, if $n$ tends to infinity then the asymptotic upper bound of $7/(r+1)$ due to Lenz et al (2021) can be improved to $4.911/(r+1)$. Both the lower and the upper bounds are achieved by relating the code density to the Turán density from extremal combinatorics. For the last task, we use the analytic framework of measurable subsets of the real cube $[0,1]^r$.

math.CO↗

Diameter and mixing time of the giant component in the percolated hypercube

We consider bond percolation on the $d$-dimensional binary hypercube with $p=c/d$ for fixed $c>1$. We prove that the typical diameter of the giant component $L_1$ is of order $Θ(d)$, and the typical mixing time of the lazy random walk on $L_1$ is of order $Θ(d^2)$. This resolves long-standing open problems of Bollobás, Kohayakawa and Łuczak from 1994, and of Benjamini and Mossel from 2003. A key component in our approach is a new tight large deviation estimate on the number of vertices in $L_1$ whose proof includes several novel ingredients: a structural description of the residue outside the giant component after sprinkling, a tight quantitative estimate on the spread of the giant in the hypercube, and a stability principle which rules out the disintegration of large connected sets under thinning. This toolkit further allows us to obtain optimal bounds on the expansion in $L_1$.

math.PR↗

What can be computed in average anonymous networks?

We study what deterministic distributed algorithms can compute on random input graphs in extremely weak models of distributed computing: all nodes are anonymous, and in each communication round, nodes broadcast a message to all their neighbors, receive a (multi)set of messages from their neighbors, and update their local state. These correspond to the SB and MB models introduced by Hella et al. [PODC 2012] and are strictly weaker than the standard port-numbering PN and LOCAL models. We investigate what can be computed almost surely on random input graphs. We give a one-round deterministic SB-algorithm using $O(\log n)$-bit messages that computes unique identifiers with high probability on anonymous networks sampled from $G(n,p)$, where $n^{\varepsilon-1} \le p \le 1/2$ and $\varepsilon>0$ is an arbitrarily small constant. This algorithm is inspired by canonical labeling techniques in graph isomorphism testing and can be used to "anonymize" existing distributed graph algorithms designed for the broadcast CONGEST and LOCAL models. In particular, we give a new anonymous algorithm that finds a triangle in $O(1/\varepsilon)$ rounds on the above input distribution. We also investigate computational power of natural analogs of "Monte Carlo" and "Las Vegas" distributed graph algorithms in the random graph setting, and establish some new collapse and hierarchy results. For example, our work shows the collapse of the weak model hierarchy of Hella et al. on $G(n,p)$, as apart from a vanishingly small fraction of input graphs, the SB model is as powerful as LOCAL.

cs.DC↗

Canonical labelling of random regular graphs

We prove that whenever $d=d(n)\to\infty$ and $n-d\to\infty$ as $n\to\infty$, then with high probability for any non-trivial initial colouring, the colour refinement algorithm distinguishes all vertices of the random regular graph $\mathcal{G}_{n,d}$. This, in particular, implies that with high probability $\mathcal{G}_{n,d}$ admits a canonical labelling computable in time $O(\min\{n^ω,nd^2+nd\log n\})$, where $ω<2.372$ is the matrix multiplication exponent.

math.CO↗

A logical approach to concentration

Concentration results say that a sequence of random variables becomes progressively concentrated around the mean. Such results are common in the study of functions of random graphs. We introduce a real-valued logic with various aggregate operators on graphs, including summation, and prove that every term in the language, seen as a random variable on random graphs within the classical Erdős-Rényi random graph model, is concentrated. We prove this for dense and sparse variants of Erdős-Rényi graphs. On the one hand, our results extend the line of work originating with Fagin and Glebskii et al. on zero-one laws for dense random graphs, as well as the zero-one law of Shelah and Spencer for sparse random graphs. On the other hand, they can be seen as a meta-theorem for inferring concentration results on random graphs, and we give examples of such applications.

math.PR↗

Saturation in Random Hypergraphs

Let $K^r_n$ be the complete $r$-uniform hypergraph on $n$ vertices, that is, the hypergraph whose vertex set is $[n]:=\{1,2,...,n\}$ and whose edge set is $\binom{[n]}{r}$. We form $G^r(n,p)$ by retaining each edge of $K^r_n$ independently with probability $p$. An $r$-uniform hypergraph $H\subseteq G$ is $F$-saturated if $H$ does not contain any copy of $F$, but any missing edge of $H$ in $G$ creates a copy of $F$. Furthermore, we say that $H$ is weakly $F$-saturated in $G$ if $H$ does not contain any copy of $F$, but the missing edges of $H$ in $G$ can be added back one-by-one, in some order, such that every edge creates a new copy of $F$. The smallest number of edges in an $F$-saturated hypergraph in $G$ is denoted by $sat(G,F)$, and in a weakly $F$-saturated hypergraph in $G$ by $wsat(G,F)$. In 2017, Korándi and Sudakov initiated the study of saturation in random graphs, showing that for constant $p$, with high probability $sat(G(n,p),K_s)=(1+o(1))n\log_{\frac{1}{1-p}}n$, and $wsat(G(n,p),K_s)=wsat(K_n,K_s)$. Generalising their results, in this paper, we solve the suturation problem for random hypergraphs for every $2\le r < s$ and constant $p$.

math.CO↗

On the threshold for triangulations inside convex polygons

Start with a large convex polygon and add all other edges inside independently with probability $p$. At what critical threshold $p_c$ do triangulations of the polygon begin to appear? The first author and Gravner asked this question, and observed that $p_c=Θ(1)$, using the relationship with the Catalan numbers and a coupling with oriented site percolation on ${\mathbb Z}^2$. More recently, Archer, Hartarsky, the first author, Olesker-Taylor, Schapira and Valesin proved that $1/4<p_c<p_c^o$, where $1/4$ is the Catalan exponential growth rate and $p_c^o$ is the critical threshold for oriented percolation. The upper bound is strict, but non-quantitative, and follows by a renormalization argument. We show that $p_c<1/2$ using a simple ear clipping algorithm, which can be analyzed using the gambler's ruin problem. This bound is closer to the truth (perhaps near $0.4$) and shows that most configurations of edges inside large convex polygons contain triangulations.

math.PR↗

Weak saturation numbers of large complete bipartite graphs

An $n$-vertex graph $G$ is weakly $F$-saturated if $G$ contains no copy of $F$ and there exists an ordering of all edges in $E(K_n) \setminus E(G)$ such that, when added one at a time, each edge creates a new copy of $F$. The minimum size of a weakly $F$-saturated graph $G$ is called the weak saturation number $\mathrm{wsat}(n, F)$. We obtain exact values and new bounds for $\mathrm{wsat}(n, K_{s,t})$ in the previously unaddressed range $s+t < n < 3t-3$, where $3\leq s\leq t$. To prove lower bounds, we introduce a new method that takes into account connectivity properties of subgraphs of a complement $G'$ to a weakly saturated graph $G$. We construct an auxiliary hypergraph and show that a linear combination of its parameters always increases in the process of the deletion of edges of $G'$. This gives a lower bound which is tight, up to an additive constant.

math.CO↗

Canonization of a random circulant graph by counting walks

It is well known that almost all graphs are canonizable by a simple combinatorial routine known as color refinement, also referred to as the 1-dimensional Weisfeiler-Leman algorithm. With high probability, this method assigns a unique label to each vertex of a random input graph and, hence, it is applicable only to asymmetric graphs. The strength of combinatorial refinement techniques becomes a subtle issue if the input graphs are highly symmetric. We prove that the combination of color refinement and vertex individualization yields a canonical labeling for almost all circulant digraphs (i.e., Cayley digraphs of a cyclic group). This result provides first evidence of good average-case performance of combinatorial refinement within the class of vertex-transitive graphs. Remarkably, we do not even need the full power of the color refinement algorithm. We show that the canonical label of a vertex $v$ can be obtained just by counting walks of each length from $v$ to an individualized vertex. Our analysis also implies that almost all circulant graphs are compact in the sense of Tinhofer, that is, their polytops of fractional automorphisms are integral. Finally, we show that a canonical Cayley representation can be constructed for almost all circulant graphs by the more powerful 2-dimensional Weisfeiler-Leman algorithm.

cs.CC↗

Sums along the edges of bounded degree graphs

Let $G$ be a graph on $n$ vertices and $(H,+)$ be an abelian group. What is the minimum size ${\sf S}_H(G)$ of the set of all sums $A(u)+A(v)$ over all injections $A:V(G)\to H$? In 2012, the first author, Angel, the second author, and Lubetzky proved that, for expander graphs and $H=\mathbb{Z}$, this minimum is at least $Ω(\log n)$, and this bound is tight -- there exists a regular expander $G$ with ${\sf S}_{\mathbb{Z}}(G)=O(\log n)$. We prove that, for every constant $d\geq 3$, the random $d$-regular graph $\mathcal{G}_{n,d}$ has significantly larger sum-sets: with high probability, for every abelian group $H$, ${\sf S}_H(\mathcal{G}_{n,d})=Ω(n^{1-2/d})$. In particular, this proves that, for every $\varepsilon>0$, there exists a regular graph with $O(n)$ edges and with sum-sets of size at least $n^{1-\varepsilon}$, for all abelian groups. The bound ${\sf S}_H(\mathcal{G}_{n,d})=Ω(n^{1-2/d})$ is tight up to a polylogarithmic factor: We show that, for every $3\leq d\leq \ln n/ \ln \ln n$, there exists an abelian group $H$ such that, for every graph $G$ on $n$ vertices with maximum degree at most $d$, ${\sf S}_H(G) \leq n^{1-2/d}(\log n)^{O(1)}$. We also prove that, for $d\gg\ln^2 n$, with high probability, for every abelian group $H$, ${\sf S}_H(\mathcal{G}_{n,d})=n(1-o(1))$ and determine the second-order term, up to a polylogarithmic factor.

math.CO↗

A very sharp threshold for first order logic distinguishability of random graphs

In this paper we find an integer $h=h(n)$ such that the minimum number of variables of a first order sentence that distinguishes between two independent uniformly distributed random graphs of size $n$ with the asymptotically largest possible probability $\frac{1}{4}-o(1)$ belongs to $\{h,h+1,h+2,h+3\}$. We also prove that the minimum (random) $k$ such that two independent random graphs are distinguishable by a first order sentence with $k$ variables belongs to $\{h,h+1,h+2\}$ with probability $1-o(1)$.

math.CO↗

When does a tree activate the random graph?

Let $F$ and $G$ be two graphs. A spanning subgraph $H$ of $G$ is called weakly $F$-saturated if one can add to $H$ the edges of $G \setminus H$ in some order, so that whenever a new edge is added, a new copy of $F$ is formed. Obtaining lower bounds for the minimum size $\mathrm{wsat}(G,F)$ of such an $H$ is a classical problem in extremal combinatorics. In particular, in the past 40 years, various algebraic tools have been developed to prove lower bounds on the weak saturation number $\mathrm{wsat}(G,F)$. Our paper uncovers a new connection of weak saturation to topology of clique complexes, that allows to prove tight lower bounds in some cases when the algebraic tools are not efficient. It is easy to see that the smallest $K_3$-saturating graphs in $K_n$ are trees, thus $\mathrm{wsat}(K_n,K_3)=n-1$. In 2017, Korándi and Sudakov proved that this is also the case in dense random graphs $G\sim G_{n,p}$, $p=\mathrm{const}\in(0,1)$, and posed the question of determining the smallest $p$ for which $G_{n,p}$ contains a $K_3$-saturating tree with high probability. Using the new topological connection, we show that this critical $p$ is of order $n^{-1/3-o(1)}$. Inspired by Gromov's local-to-global principle for hyperbolic groups, we further develop our topological approach and determine the critical probability up to a constant factor, for trees with diameter at most $n^{c}$, for some $c>0$. The new connection also enables us to improve the best known upper bound on the threshold probability for simple connectivity of the 2-dimensional clique complex of $G_{n,p}$, due to Kahle.

math.CO↗

Majority dynamics on finite trees

For an arbitrary finite tree $T$, we find the exact value of the wort-case stabilisation time of majority dynamics on $T$. We also prove that for a perfect rooted cubic tree $T$ with diameter $D$ and uniformly random initial opinions, the dynamics stabilises in time $τ\in(D/4,D/3)$ with high probability.

math.CO↗

Sharp thresholds for spanning regular subgraphs

We prove that $(1+o(1))\sqrt{e/n}$ is the sharp threshold for the appearance of the square of a Hamilton cycle in $G(n,p)$, confirming the conjecture of Kahn, Narayanan, and Park. We also find the exact asymptotics of the threshold for the emergence of a spanning subgraph isomorphic to a fixed graph $F$ for a wide family of $d$-regular graphs $F$. This family includes almost all $d$-regular graphs.

math.CO↗