Coordinate Methods for Accelerating ℓ_∞ Regression and Faster Approximate Maximum Flow
share
In this paper we provide faster algorithms for approximately solving ℓ_∞ regression, a fundamental problem prevalent in both combinatorial and continuous optimization. In particular we provide an accelerated coordinate descent method which converges in k iterations at a O(1/k) rate independent of the dimension of the problem, and whose iterations can be implemented cheaply for many structured matrices. Our algorithm can be viewed as an alternative approach to the recent breakthrough result of Sherman which achieves a similar running time improvement over classic algorithmic approaches, i.e. smoothing and gradient descent, which either converge at a O(1/√(k)) rate or have running times with a worse dependence on problem parameters. We show how to use our algorithm to achieve a runtime of Õ(m + √(ns)/ϵ) to compute an ϵ-approximate maximum flow, for an undirected graph with m edges, n vertices, and where s is the squared ℓ_2 norm of the congestion of any optimal flow. As s = O(m) this yields a running time of Õ(m + √(nm)/ϵ), generically improving upon the previous best known runtime of Õ(m/ϵ) by Sherman whenever the graph is slightly dense. Moreover, we show how to achieve improved exact algorithms for maximum flow on a variety of unit capacity graphs. We achieve these results by providing an accelerated coordinate descent method exploiting dynamic measures of coordinate smoothness for smoothed ℓ_∞ regression. Our analysis leverages the structure of the Hessian of the smoothed problem via a bound on its trace, as well as techniques for exploiting sparsity of the constraint matrix for faster sampling and smoothness estimates.
READ FULL TEXT
