Bounds on the Number of Longest Common Subsequences

Abstract

This paper performs the analysis necessary to bound the running time of known, efficient algorithms for generating all longest common subsequences. That is, we bound the running time as a function of input size for algorithms with time essentially proportional to the output size. This paper considers both the case of computing all distinct LCSs and the case of computing all LCS embeddings. Also included is an analysis of how much better the efficient algorithms are than the standard method of generating LCS embeddings. A full analysis is carried out with running times measured as a function of the total number of input characters, and much of the analysis is also provided for cases in which the two input sequences are of the same specified length or of two independently specified lengths.