An Efficient Parallel Algorithm for Spectral Sparsification of Laplacian and SDDM Matrix Polynomials
Abstract
For "large" class of continuous probability density functions (p.d.f.), we demonstrate that for every there is mixture of discrete Binomial distributions (MDBD) with distinct Binomial distributions that -approximates a discretized p.d.f. for all , where . Also, we give two efficient parallel algorithms to find such MDBD. Moreover, we propose a sequential algorithm that on input MDBD with for that induces a discretized p.d.f. , that is either Laplacian or SDDM matrix and parameter , outputs in time a spectral sparsifier of a matrix-polynomial, where notation hides factors. This improves the Cheng et al.'s [CCLPT15] algorithm whose run time is . Furthermore, our algorithm is parallelizable and runs in work and depth . Our main algorithmic contribution is to propose the first efficient parallel algorithm that on input continuous p.d.f. , matrix as above, outputs a spectral sparsifier of matrix-polynomial whose coefficients approximate component-wise the discretized p.d.f. . Our results yield the first efficient and parallel algorithm that runs in nearly linear work and poly-logarithmic depth and analyzes the long term behaviour of Markov chains in non-trivial settings. In addition, we strengthen the Spielman and Peng's [PS14] parallel SDD solver.
Keywords
Cite
@article{arxiv.1507.07497,
title = {An Efficient Parallel Algorithm for Spectral Sparsification of Laplacian and SDDM Matrix Polynomials},
author = {Gorav Jindal and Pavel Kolev},
journal= {arXiv preprint arXiv:1507.07497},
year = {2016}
}