NEW ALGORITHMS FOR AN ANCIENT SCHEDULING PROBLEM

We consider the on-line version of the original m-machine scheduling p roblem: given m machines and n positive real jobs, schedule the n jobs on the m machines so as to minimize the makespan, the completion time of the last job. In the on-line version, as soon as job j arrives, it must be assigned immediately to one of the m machines. We present two main results. The first is a (2 - epsilon)-competitive deterministic algorithm for all m. The competitive ratio of all previous algorithms approaches 2 as m --> infinity. Indeed, the problem of improving the c ompetitive ratio for large m had been open since 1966, when the first algorithm for this problem appeared. The second result is an optimal r andomized algorithm for the case m = 2. To the best of our knowledge, our 4/3-competitive algorithm is the first specifically randomized alg orithm for the original, m-machine, on-line scheduling problem. (C) 19 95 Academic Press, Inc.