Tetris–Hashing or optimal table compression Original Research Article

In this paper, we present a new method for mapping a static set of n keys, each an integer between 0 and N−1, into a hash table of size n without any collision. Our data structure requires only an additional array of n integers, each less than n, and achieves a worst-case lookup time of O(1). This method is based on a randomized compression scheme, and it finds a minimal perfect hash function in average time O(n). Our concept can be easily adapted for dynamic key sets. Then, the hash table has no longer minimal size but the storage location remains very small. Because of its simplicity our approach is particularly interesting for practical purposes.