51
82

Testing kk-Modal Distributions: Optimal Algorithms via Reductions

Abstract

We give highly efficient algorithms, and almost matching lower bounds, for a range of basic statistical problems that involve testing and estimating the L_1 distance between two k-modal distributions pp and qq over the discrete domain {1,,n}\{1,\dots,n\}. More precisely, we consider the following four problems: given sample access to an unknown k-modal distribution pp, Testing identity to a known or unknown distribution: 1. Determine whether p=qp = q (for an explicitly given k-modal distribution qq) versus pp is \eps\eps-far from qq; 2. Determine whether p=qp=q (where qq is available via sample access) versus pp is \eps\eps-far from qq; Estimating L1L_1 distance ("tolerant testing'') against a known or unknown distribution: 3. Approximate dTV(p,q)d_{TV}(p,q) to within additive \eps\eps where qq is an explicitly given k-modal distribution qq; 4. Approximate dTV(p,q)d_{TV}(p,q) to within additive \eps\eps where qq is available via sample access. For each of these four problems we give sub-logarithmic sample algorithms, that we show are tight up to additive \poly(k)\poly(k) and multiplicative \polyloglogn+\polylogk\polylog\log n+\polylog k factors. Thus our bounds significantly improve the previous results of \cite{BKR:04}, which were for testing identity of distributions (items (1) and (2) above) in the special cases k=0 (monotone distributions) and k=1 (unimodal distributions) and required O((logn)3)O((\log n)^3) samples. As our main conceptual contribution, we introduce a new reduction-based approach for distribution-testing problems that lets us obtain all the above results in a unified way. Roughly speaking, this approach enables us to transform various distribution testing problems for k-modal distributions over {1,,n}\{1,\dots,n\} to the corresponding distribution testing problems for unrestricted distributions over a much smaller domain {1,,}\{1,\dots,\ell\} where =O(klogn).\ell = O(k \log n).

View on arXiv
Comments on this paper