Arrow Research search
Back to STOC

STOC 2025

Linear Hashing Is Optimal

Conference Paper Session 2D Algorithms and Complexity · Theoretical Computer Science

Abstract

We prove that hashing n balls into n bins via random 2 -linear maps yields expected maximum load O (log n / loglog n ), resolving an open question of Alon, Dietzfelbinger, Miltersen, Petrank, and Tardos (STOC ’97, JACM ’99). More generally, we show that the maximum load exceeds r · log n /loglog n with probability at most O (1/ r 2 ). Our proof uses potential functions to detect heavy bins.

Authors

Keywords

  • balls and bins
  • pseudorandomness
  • universal hash functions

Context

Venue
ACM Symposium on Theory of Computing
Archive span
1969-2025
Indexed papers
4364
Paper id
25852890554860416
v2026.09.13