The distribution of the length of the longest run has wide applications in regard to reliability and DNA sequencing. Statistical tests based on the longest well-matching run are usually considered to be more reasonable than tests based on the perfect-matching run. In this paper, a method adopted from Fu and Koutras (J. Amer. Statist. Assoc. 89 (1994) 1050) is proposed to improve the efficiency of computing the exact distribution of length. We used the result to investigate the accuracy of some approximations of the distribution. (C) 2003 Elsevier Science B.V. All rights reserved.
關聯:
STATISTICS & PROBABILITY LETTERS v.63 n.3 pp.267-274
