An SDP-Based Algorithm for Linear-Sized Spectral Sparsification
This solves a fundamental problem in graph theory and algorithms for researchers in spectral graph theory and machine learning, providing an efficient method for sparsification with optimal edge count and near-linear time.
The paper tackles the problem of constructing a linear-sized spectral sparsifier for undirected weighted graphs, achieving an algorithm that outputs a (1+ε)-spectral sparsifier with O(n/ε²) edges in Õ(m/ε^O(1)) time, answering an open question affirmatively.
For any undirected and weighted graph $G=(V,E,w)$ with $n$ vertices and $m$ edges, we call a sparse subgraph $H$ of $G$, with proper reweighting of the edges, a $(1+\varepsilon)$-spectral sparsifier if \[ (1-\varepsilon)x^{\intercal}L_Gx\leq x^{\intercal} L_{H} x\leq (1+\varepsilon) x^{\intercal} L_Gx \] holds for any $x\in\mathbb{R}^n$, where $L_G$ and $L_{H}$ are the respective Laplacian matrices of $G$ and $H$. Noticing that $Ω(m)$ time is needed for any algorithm to construct a spectral sparsifier and a spectral sparsifier of $G$ requires $Ω(n)$ edges, a natural question is to investigate, for any constant $\varepsilon$, if a $(1+\varepsilon)$-spectral sparsifier of $G$ with $O(n)$ edges can be constructed in $\tilde{O}(m)$ time, where the $\tilde{O}$ notation suppresses polylogarithmic factors. All previous constructions on spectral sparsification require either super-linear number of edges or $m^{1+Ω(1)}$ time. In this work we answer this question affirmatively by presenting an algorithm that, for any undirected graph $G$ and $\varepsilon>0$, outputs a $(1+\varepsilon)$-spectral sparsifier of $G$ with $O(n/\varepsilon^2)$ edges in $\tilde{O}(m/\varepsilon^{O(1)})$ time. Our algorithm is based on three novel techniques: (1) a new potential function which is much easier to compute yet has similar guarantees as the potential functions used in previous references; (2) an efficient reduction from a two-sided spectral sparsifier to a one-sided spectral sparsifier; (3) constructing a one-sided spectral sparsifier by a semi-definite program.