English

Optimal Gossip Algorithms for Exact and Approximate Quantile Computations

Data Structures and Algorithms 2017-11-28 v1 Distributed, Parallel, and Cluster Computing

Abstract

This paper gives drastically faster gossip algorithms to compute exact and approximate quantiles. Gossip algorithms, which allow each node to contact a uniformly random other node in each round, have been intensely studied and been adopted in many applications due to their fast convergence and their robustness to failures. Kempe et al. [FOCS'03] gave gossip algorithms to compute important aggregate statistics if every node is given a value. In particular, they gave a beautiful O(logn+log1ϵ)O(\log n + \log \frac{1}{\epsilon}) round algorithm to ϵ\epsilon-approximate the sum of all values and an O(log2n)O(\log^2 n) round algorithm to compute the exact ϕ\phi-quantile, i.e., the the ϕn\lceil \phi n \rceil smallest value. We give an quadratically faster and in fact optimal gossip algorithm for the exact ϕ\phi-quantile problem which runs in O(logn)O(\log n) rounds. We furthermore show that one can achieve an exponential speedup if one allows for an ϵ\epsilon-approximation. We give an O(loglogn+log1ϵ)O(\log \log n + \log \frac{1}{\epsilon}) round gossip algorithm which computes a value of rank between ϕn\phi n and (ϕ+ϵ)n(\phi+\epsilon)n at every node.% for any 0ϕ10 \leq \phi \leq 1 and 0<ϵ<10 < \epsilon < 1. Our algorithms are extremely simple and very robust - they can be operated with the same running times even if every transmission fails with a, potentially different, constant probability. We also give a matching Ω(loglogn+log1ϵ)\Omega(\log \log n + \log \frac{1}{\epsilon}) lower bound which shows that our algorithm is optimal for all values of ϵ\epsilon.

Keywords

Cite

@article{arxiv.1711.09258,
  title  = {Optimal Gossip Algorithms for Exact and Approximate Quantile Computations},
  author = {Bernhard Haeupler and Jeet Mohapatra and Hsin-Hao Su},
  journal= {arXiv preprint arXiv:1711.09258},
  year   = {2017}
}
R2 v1 2026-06-22T22:56:48.923Z