Larry Stockmeyer: 1948-2004

Larry was born in Evansville, Indiana, in 1948. He attended college at MIT, receiving a B.S. in mathematics and an M.S. in electrical engineering in 1972, and a Ph.D. in computer science in 1974. He spent all of his professional career at IBM. He started in 1974 at the Thomas J. Watson Research Center, and in 1982 moved to what is now the Almaden Research Center. He retired from IBM in November 2003. In 1996 he was named an ACM Fellow. Larry passed away on July 31, 2004, at age 55. Larry made many important contributions to the theory of computation. One – extremely important– example is his discovery (with Albert Meyer in FOCS 1972) of the polynomial time hierarchy. He also did important work in parallel and distributed computing. He had three influential papers about consensus in partial synchrony models. After the FLP result that proved consensus to be impossible in an asynchronous system where one process can crash, finding realistic systems where consensus could be solved became an active research area. Larry’s paper “On the Minimal Synchronism Needed for Distributed Consensus” (with Danny Dolev and Cynthia Dwork in FOCS 1983 and JACM 1987) helped explaining why consensus is possible in certain models but not possible in others. It considered various synchrony conditions, examined how varying these affects the number of faults that can be tolerated by a consensus algorithm. In “Consensus in the presence of partial synchrony” (with Dwork and Nancy Lynch, JACM 1988), the concept of partial synchrony in a distributed system is studied, and consensus protocols are given for various cases of partial synchrony and various fault models. Lower bounds that show in most cases that the protocols are optimal with respect to the number of faults tolerated are also given. Also, they introduced fault-tolerant clock synchronization protocols, and used them in their protocols. In “Bounds on the time to reach agreement in the presence of timing uncertainty” (with Hagit Attiya, Dwork and Lynch in JACM 1994) upper and lower bounds are proved for the time complexity of solving consensus in a crash failure model, where time for message delivery is at most d, and the amount of time between any two consecutive steps of any nonfaulty process is at least c1 and at most c2; thus, C = c2/c1 is a measure of the timing uncertainty. On a different subject, “Lower bounds on the competitive ratio for mobile user tracking and distributed job scheduling” (with Noga Alon, Gil Kalai and Moty Ricklin, TCS 1994) proves a lower bound of Ω(log n/ log log n) on the competitive ratio for solving a mobile user problem and a Distributed Job Scheduling problem. His paper “What can be computed locally?” (with Moni Naor, SICOMP 1995) studies computation that can be done within time independent of the size of the network. These distributed computing papers are consistent with others by Larry – serious, deep work. They further illustrate how broad his research was. He will be missed by many people.