Tales of Huffman

Abstract: We study the new problem of Huffman-like codes subject to individual restrictions on the code-word lengths of a subset of the source words. These are prefix codes with minimal expected code-word length for a random source where additionally the code-word lengths of a subset of the source words is prescribed, possibly differently for every such source word. Based on a structural analysis of properties of optimal solutions, we construct an efficient dynamic programming algorithm for this problem, and for an integer programming problem that may be of independent interest.