Tight Bounds for Lp Samplers, Finding Duplicates in Streams, and Related Problems

TL;DR

This paper introduces near-optimal space algorithms for Lp-sampling and duplicate detection in data streams, improving bounds for various p values and establishing tight lower bounds, thus advancing streaming algorithm efficiency.

Contribution

It provides improved space bounds for Lp-samplers across different p ranges, introduces a zero relative error L0-sampler, and establishes tight lower bounds for related problems.

Findings

O(psilon^{-p} log^2 n) space for p  (1,2) Lp-samplers

O(log^2 n) bits zero relative error L0-sampler

Matching  log^2 n lower bound for sampling from 0, b1 1 vectors

Abstract

In this paper, we present near-optimal space bounds for Lp-samplers. Given a stream of updates (additions and subtraction) to the coordinates of an underlying vector x \in R^n, a perfect Lp sampler outputs the i-th coordinate with probability |x_i|^p/||x||_p^p. In SODA 2010, Monemizadeh and Woodruff showed polylog space upper bounds for approximate Lp-samplers and demonstrated various applications of them. Very recently, Andoni, Krauthgamer and Onak improved the upper bounds and gave a O(\epsilon^{-p} log^3 n) space \epsilon relative error and constant failure rate Lp-sampler for p \in [1,2]. In this work, we give another such algorithm requiring only O(\epsilon^{-p} log^2 n) space for p \in (1,2). For p \in (0,1), our space bound is O(\epsilon^{-1} log^2 n), while for the $p=1$ case we have an O(log(1/\epsilon)\epsilon^{-1} log^2 n) space algorithm. We also give a O(log^2 n) bits zero relative error L0-sampler, improving the O(log^3 n) bits algorithm due to Frahling, Indyk and Sohler. As an application of our samplers, we give better upper bounds for the problem of finding duplicates in data streams. In case the length of the stream is longer than the alphabet size, L1 sampling gives us an O(log^2 n) space algorithm, thus improving the previous O(log^3 n) bound due to Gopalan and Radhakrishnan. In the second part of our work, we prove an Omega(log^2 n) lower bound for sampling from 0, \pm 1 vectors (in this special case, the parameter p is not relevant for Lp sampling). This matches the space of our sampling algorithms for constant \epsilon > 0. We also prove tight space lower bounds for the finding duplicates and heavy hitters problems. We obtain these lower bounds using reductions from the communication complexity problem augmented indexing.