An improved data stream algorithm for frequency moments

  • Authors:
  • Don Coppersmith;Ravi Kumar

  • Affiliations:
  • IBM T. J. Watson Research Center, Yorktown Heights, NY;IBM Almaden Research Center, San Jose, CA

  • Venue:
  • SODA '04 Proceedings of the fifteenth annual ACM-SIAM symposium on Discrete algorithms
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present a simple, one-pass, Õ(√n)-space data stream algorithm for approximating the third frequency moment. This is the first improvement to the Õ(n2/3)-space data stream algorithm of Alon, Matias, and Szegedy [AMS99]. the current known lower bound for this problem is Ω(n1/3) [BJKS02a].Our algorithm can also be generalized to an Õ(n1-1/(k-1))-space data stream algorithm for approximating the k-th frequency moment. Besides improving the Õ(n1--1/k)-space upper bound [AMS99], our algorithm beats the Ω(n1--1/k)-sampling lower bound [BKS01] for this problem.Our method suggests a unified perspective of space-efficient data stream algorithms for all frequency moments.