Constructing Huffman Trees in Parallel

We present a parallel algorithm for the Huuman Coding problem. We reduce the Huuman Coding problem to the Concave Least Weight Subsequence problem and give a parallel algorithm that solves the latter problem in O(p n log n) time with n processors on a CREW PRAM. This leads to the rst sublinear time o(n 2)-total work parallel algorithm for Huuman Coding. This reduction of the Huuman Coding problem to the Concave Least Weight Subsequence problem also yields an alternative O(n log n)-time (or linear time { for a sorted input sequence) algorithm for Huuman Coding.