Asynchronous Approximation of a Single Component of the Solution to a Linear System

C.E. Lee, A. Ozdaglar, D. Shah
arXiv preprint arXiv:1411.2647

We present a distributed asynchronous algorithm for approximating a single component of the solution to a system of linear equations Ax=b, where A is a positive definite real matrix, and bn. This is equivalent to solving for xi in x=Gx+z for some G and z such that the spectral radius of G is less than 1. Our algorithm relies on the Neumann series characterization of the component xi, and is based on residual updates. We analyze our algorithm within the context of a cloud computation model, in which the computation is split into small update tasks performed by small processors with shared access to a distributed file system. We prove a robust asymptotic convergence result when ρ(G̃ )<1, regardless of the precise order and frequency in which the update tasks are performed, where G̃ ij=|Gij|. We provide convergence rate bounds which depend on the order of update tasks performed, analyzing both deterministic update rules via counting weighted random walks, as well as probabilistic update rules via concentration bounds. The probabilistic analysis requires analyzing the product of random matrices which are drawn from distributions that are time and path dependent. We specifically consider the setting where n is large, yet G is sparse, e.g., each row has at most d nonzero entries. This is motivated by applications in which G is derived from the edge structure of an underlying graph. Our results prove that if the local neighborhood of the graph does not grow too quickly as a function of n, our algorithm can provide significant reduction in computation cost as opposed to any algorithm which computes the global solution vector x. Our algorithm obtains an ϵx2 additive approximation for xi in constant time with respect to the size of the matrix when d=O(1) and 1/(1G2)=O(1).