Universal Perfect Samplers for Incremental Streams
Abstract
If , the -moment of a vector is and the -sampling problem is to select an index according to its contribution to the -moment, i.e., such that . Approximate -samplers may introduce multiplicative and/or additive errors to this probability, and some have a non-trivial probability of failure. In this paper we focus on the exact -sampling problem, where is selected from the class of Laplace exponents of non-negative, one-dimensional L\'evy processes, which includes several well studied classes such as th moments , , logarithms , Cohen and Geri's soft concave sublinear functions, which are used to approximate concave sublinear functions, including cap statistics. We develop -samplers for a vector that is presented as an incremental stream of positive updates. In particular: * For any , we give a very simple -sampler that uses 2 words of memory and stores at all times a , such that is exactly . * We give a ``universal'' -sampler that uses words of memory w.h.p., and given any at query time, produces an exact -sample. With an overhead of a factor of , both samplers can be used to -sample a sequence of indices with or without replacement. Our sampling framework is simple and versatile, and can easily be generalized to sampling from more complex objects like graphs and hypergraphs.
Cite
@article{arxiv.2407.04931,
title = {Universal Perfect Samplers for Incremental Streams},
author = {Seth Pettie and Dingyu Wang},
journal= {arXiv preprint arXiv:2407.04931},
year = {2024}
}