100

Moments, Random Walks, and Limits for Spectrum Approximation

Annual Conference Computational Learning Theory (COLT), 2023
Abstract

We study lower bounds for the problem of approximating a one dimensional distribution given (noisy) measurements of its moments. We show that there are distributions on [1,1][-1,1] that cannot be approximated to accuracy ϵ\epsilon in Wasserstein-1 distance even if we know \emph{all} of their moments to multiplicative accuracy (1±2Ω(1/ϵ))(1\pm2^{-\Omega(1/\epsilon)}); this result matches an upper bound of Kong and Valiant [Annals of Statistics, 2017]. To obtain our result, we provide a hard instance involving distributions induced by the eigenvalue spectra of carefully constructed graph adjacency matrices. Efficiently approximating such spectra in Wasserstein-1 distance is a well-studied algorithmic problem, and a recent result of Cohen-Steiner et al. [KDD 2018] gives a method based on accurately approximating spectral moments using 2O(1/ϵ)2^{O(1/\epsilon)} random walks initiated at uniformly random nodes in the graph. As a strengthening of our main result, we show that improving the dependence on 1/ϵ1/\epsilon in this result would require a new algorithmic approach. Specifically, no algorithm can compute an ϵ\epsilon-accurate approximation to the spectrum of a normalized graph adjacency matrix with constant probability, even when given the transcript of 2Ω(1/ϵ)2^{\Omega(1/\epsilon)} random walks of length 2Ω(1/ϵ)2^{\Omega(1/\epsilon)} started at random nodes.

View on arXiv
Comments on this paper