English

Beating Fredman-Koml\'{o}s for perfect $k$-hashing

Information Theory 2018-05-14 v1 Discrete Mathematics Data Structures and Algorithms Combinatorics math.IT

Abstract

We say a subset C{1,2,,k}nC \subseteq \{1,2,\dots,k\}^n is a kk-hash code (also called kk-separated) if for every subset of kk codewords from CC, there exists a coordinate where all these codewords have distinct values. Understanding the largest possible rate (in bits), defined as (log2C)/n(\log_2 |C|)/n, of a kk-hash code is a classical problem. It arises in two equivalent contexts: (i) the smallest size possible for a perfect hash family that maps a universe of NN elements into {1,2,,k}\{1,2,\dots,k\}, and (ii) the zero-error capacity for decoding with lists of size less than kk for a certain combinatorial channel. A general upper bound of k!/kk1k!/k^{k-1} on the rate of a kk-hash code (in the limit of large nn) was obtained by Fredman and Koml\'{o}s in 1984 for any k4k \geq 4. While better bounds have been obtained for k=4k=4, their original bound has remained the best known for each k5k \ge 5. In this work, we obtain the first improvement to the Fredman-Koml\'{o}s bound for every k5k \ge 5. While we get explicit (numerical) bounds for k=5,6k=5,6, for larger kk we only show that the FK bound can be improved by a positive, but unspecified, amount. Under a conjecture on the optimum value of a certain polynomial optimization problem over the simplex, our methods allow an effective bound to be computed for every kk.

Keywords

Cite

@article{arxiv.1805.04151,
  title  = {Beating Fredman-Koml\'{o}s for perfect $k$-hashing},
  author = {Venkatesan Guruswami and Andrii Riazanov},
  journal= {arXiv preprint arXiv:1805.04151},
  year   = {2018}
}

Comments

19 pages