Cyrus Derman

Brief Biography

Cyrus Derman was a John von Neumann Theory Prize recipient who made fundamental contributions to performance analysis and the optimization of stochastic systems. Growing up in Collingdale, Pennsylvania, Derman’s dream was to become a concert violinist. He attended the University of Pennsylvania where he studied music and mathematics. Upon finishing his bachelors degree, Derman decided to pursue graduate work in the latter and went on Columbia University. There, he had the opportunity to work with some of the leading statisticians and probability theorists of the day. He wrote his dissertation on contributions to Markov chain theory and received his PhD in 1954.

Derman joined Columbia’s Department of Industrial Engineering as an Instructor in Operations Research after spending a year at Syracuse University. In 1965, he rose to rank of Professor and remained at the university until his retirement in 1992. Derman was a prominent figure in Columbia’s operations research community as many of his PhD students, including Sheldon M. Ross, Peter J. Kolesar, and John von Neumann prize-winner Arthur Veinott Jr., went on to become OR leaders in their own right. With Ross, he wrote Statistical Aspects of Quality Control (1997), a compact text that stresses the key ideas and concepts in the area rather than trying to cover each method in complete depth. In addition to his positon at Columbia, Derman spent most of the summers from 1971 to 1989 at Stanford University and held visiting appointments at a number of institutions around the globe.

In 1982, Derman co-authored an article on the consecutive-k-of-n system with Ross and Gerald Lieberman. In it, they introduced a beautiful solution of a stochastic assignment problem that can be applied to optimally assigning workers with different skills to randomly arriving jobs of varying types.

Derman and Donald L. Iglehart were named co-recipients of the John von Neumann Theory Prize in 2002. Derman was lauded for his book, Finite State Markovian Decision Processes (1970), a culmination of a series of papers that fundamentally advanced finite-state-and-action Markov decision processes. In it, he showed that starting from a state, the set of state-action frequencies over all policies is the convex hull of the finite set of state-action frequencies over all stationary determinist Markov policies. These contributions played a fundamental role in solving problem in the presence of linear constraints on the state-action frequencies, such as the frequency of unfavorable events like shortages and accidents. Derman was also recognized for his work with Liberman and Ross as well as with Morton Klein, another former student and long-time colleague. He and Klein originated and did pioneering work characterizing FIFO and LIFO when issuing policies are optimal for items whose values or field life depends on their age. Derman did this by finding a minimax inspection for randomly failing equipment.

Other Biographies

Wikipedia Entry for Cyrus Derman