Recursive bounds for perfect hashing

Abstract Let k ⩽ b be positive integers. A family C of sequences of length t over an alphabet of size b is called k -separated if for any k distinct members of C , there is a coordinate in which they mutually differ. Let N ( t , b , k ) denote the maximum size of such a family. This function has been studied extensively, mainly in the context of perfect hashing. Here we slightly improve a recent bound of Dyachkov, showing that for all t k ⩽ b , N ( t , b , k )⩽ tb −( k −1)( t −1). This implies that if k ⩽ b and t is divisible by k −1, then N ( t , b , k )⩽( k −1) b t /( k −1) −( k −1) 2 .

[1]  Andrew Chi-Chih Yao,et al.  Should Tables Be Sorted? , 1981, JACM.

[2]  L. A. Bassalygo,et al.  Hash codes , 1997, Proceedings of IEEE International Symposium on Information Theory.

[3]  A. Nilli Perfect Hashing and Probability , 1994, Combinatorics, Probability and Computing.

[4]  János Körner,et al.  New Bounds for Perfect Hashing via Information Theory , 1988, Eur. J. Comb..

[5]  János Körner,et al.  Compressing inconsistent data , 1994, IEEE Trans. Inf. Theory.

[6]  J. Körner Fredman-Kolmo´s bounds and information theory , 1986 .

[7]  Zoltán Füredi,et al.  Families of Finite Sets in Which No Set Is Covered by the Union of Two Others , 1982, J. Comb. Theory, Ser. A.

[8]  Erdal Arikan,et al.  An upper bound on the zero-error list-coding capacity , 1994, IEEE Trans. Inf. Theory.