Table of Contents
Fetching ...

Near-Optimal Four-Cycle Counting in Graph Streams

Sebastian Lüderssen, Stefan Neumann, Pan Peng

Abstract

We study four-cycle counting in arbitrary order graph streams. We present a 3-pass algorithm for $(1+\varepsilon)$-approximating the number of four-cycles using $\widetilde{O}(m/\sqrt{T})$ space, where $m$ is the number of edges and $T$ the number of four-cycles in the graph. This improves upon a 3-pass algorithm by Vorotnikova using space $\widetilde{O}(m/T^{1/3})$ and matches a multi-pass lower bound of $Ω(m/\sqrt{T})$ by McGregor and Vorotnikova.

Near-Optimal Four-Cycle Counting in Graph Streams

Abstract

We study four-cycle counting in arbitrary order graph streams. We present a 3-pass algorithm for -approximating the number of four-cycles using space, where is the number of edges and the number of four-cycles in the graph. This improves upon a 3-pass algorithm by Vorotnikova using space and matches a multi-pass lower bound of by McGregor and Vorotnikova.

Paper Structure

This paper contains 53 sections, 25 theorems, 73 equations, 4 figures, 3 algorithms.

Key Result

Theorem 1.1

There exists a $3$-pass algorithm that obtains a graph $G$ as an arbitrary-order edge stream and returns a $(1\pm\varepsilon)$-approximation of the number of four-cycles in $G$ using space $\widetilde{O}(m/\sqrt{T})$ with high probability.

Figures (4)

  • Figure 1: Illustrations of the three hard instances we consider in \ref{['sec:challenges-node-sampling']}.
  • Figure 2: Illustration of our assigned configurations $(A,\kappa,x,y)$. If $x$ and $y$ are opposite nodes in a four-cycle $A$, we will only consider $A$ if $x,y\in S_{1,\kappa}$ and its other two nodes are in $S_{2,\kappa}$and it does not contain an edge $e$ of heaviness $t(e)>\sqrt{T}$. Otherwise, if $x$ and $y$ are adjacent, we only consider $A$ if $x\in R_{1a,\kappa}$ and $y\in R_{1b,\kappa}$and it has exactly one edge $e$ of heaviness $t(e)\geq \sqrt{T}$. The dashed line represents the heavy edge $e$ and we note that our algorithm collects it in the second pass.
  • Figure 3: Visualization of the probabilities $p_{(z,\kappa,\ell)}$ and the heaviness thresholds $\theta_{(z,\kappa,\ell)}$ for given labeled substructures $(z,\kappa,\ell)$. In the first two figures, we annotated each vertex $v$ with its label $\ell(v)$; for the probabilities, we omitted the dependency on $\kappa$ for better readability. The third and fourth figure include the dependency on $\kappa$ and list the heaviness thresholds for each node, each edge, and two of the four wedges.
  • Figure 4: Example of a substructure with non-monotone heaviness. Here, we consider a node $v$ with label $\ell(v)=S_1$ which is contained in $2T^{3/4}/\delta^{1.5}$ edge-disjoint four-cycles. This makes $v$ heavy for $\kappa=T^{1/4}$ as $t(v,T^{1/4},\ell)>\theta_{(v,T^{1/4},\ell)}=T^{3/4}/\delta^{1.5}$. Additionally, $v$ is endpoint of $T^{1/8}/\delta^{1.5}$ onions of width $\kappa'=T^{3/8}$. Then all four-cycles in the onions do not count towards $t(v,\kappa'/2,\ell)$ but do count towards $t(v,\kappa',\ell)$. As $t(v,T^{3/8},\ell)>\theta_{(v,T^{3/8},\ell)}=T^{7/8}/\delta^{1.5}$, we observe that $v$ is heavy for $\kappa=T^{1/4}$ and $\kappa=\kappa'$ but light for $\kappa=\kappa'/2$.

Theorems & Definitions (48)

  • Theorem 1.1
  • Theorem 5.1
  • Definition 5.2
  • Definition 5.3
  • Proposition 5.6
  • Lemma 5.7: mcgregor2020triangle
  • Lemma 5.8
  • proof
  • Lemma 5.9: Combination lemma
  • proof
  • ...and 38 more