English

Unbiased Insights: Optimal Streaming Algorithms for $\ell_p$ Sampling, the Forget Model, and Beyond

Data Structures and Algorithms 2026-04-07 v3

Abstract

We study p\ell_p sampling and frequency moment estimation in a single-pass insertion-only data stream. For p(0,2)p \in (0,2), we present a nearly space-optimal approximate p\ell_p sampler that uses O~(lognlog(1/δ))\widetilde{O}(\log n \log(1/\delta)) bits of space and for p=2p = 2, we present a sampler with space complexity O~(log2nlog(1/δ))\widetilde{O}(\log^2 n \log(1/\delta)). This space complexity is optimal for p(0,2)p \in (0, 2) and improves upon prior work by a logn\log n factor. We further extend our construction to a continuous p\ell_p sampler, which outputs a valid sample index at every point during the stream. Leveraging these samplers, we design nearly unbiased estimators for FpF_p in data streams that include forget operations, which reset individual element frequencies and introduce significant non-linear challenges. As a result, we obtain near-optimal algorithms for estimating FpF_p for all pp in this model, originally proposed by Pavan, Chakraborty, Vinodchandran, and Meel [PODS'24], resolving all three open problems they posed. Furthermore, we generalize this model to what we call the suffix-prefix deletion model, and extend our techniques to estimate entropy as a corollary of our moment estimation algorithms. Finally, we show how to handle arbitrary coordinate-wise functions during the stream, for any gGg \in \mathbb{G}, where G\mathbb{G} includes all (linear or non-linear) contraction functions.

Keywords

Cite

@article{arxiv.2508.07067,
  title  = {Unbiased Insights: Optimal Streaming Algorithms for $\ell_p$ Sampling, the Forget Model, and Beyond},
  author = {Honghao Lin and Hoai-An Nguyen and William Swartworth and David P. Woodruff},
  journal= {arXiv preprint arXiv:2508.07067},
  year   = {2026}
}

Comments

To appear in PODS 2026

R2 v1 2026-07-01T04:42:38.152Z