Past Awards
The John von Neumann Theory Prize is awarded annually by INFORMS to a scholar who has made outstanding contributions to the theory of operations research and management science. The emphasis is on extraordinary contributions that have stood the test of time.
This year the prize was awarded to Peter Whittle for the remarkable variety and high caliber of his lifetime's work and its profound impact on operations research, particularly his contributions to time series analysis, stochastic optimization, stochastic network processes and self-organizing systems.
Whittle began his research career in the 1950's with a development of the asymptotic inference theory for Gaussian processes and related spatial processes. This work, which he presented in several papers and a book, was ahead of its time in its consideration of power law covariance functions, now central in image analysis.
His contributions to stochastic optimization have been broad and deep. Much of this is contained in "Risk Sensitive Optimal Control" and the two-volume "Optimization Over Time." In his seminal work on risk-sensitive linear/quadratic/Gaussian treatments, he discovered a generalization of the certainty equivalence principle for solving this and related problems. In his work on stochastic dynamic programming, he helped lay the foundation of the multi-armed bandit model (or which project to select) by providing a new proof of the Gittens index. This cleared up key loose ends and opened the way to many applications. He also expanded the theory to relaxed bandits (where projects change state even when they are not active)
.Whittle played a prominent role in developing the theory of stochastic network processes. This began with his studies of spatial processes, reversibility of birth and death processes and reversible network migration processes. His articles in the 1960s paralleled those of Jackson and others that have led to a theory of queueing networks. He pioneered the use of partial balance in Markov processes and used it to prove insensitivity properties for queueing networks (e.g., the equilibrium behavior of a network with exponential service times is the same for service times with general distributions). His book "Systems in Stochastic Equilibrium," for which he was awarded the Lanchester Prize in 1987, contains a wealth of information on stochastic networks and queueing.
He made significant contributions to self-organizing systems, including critical effects in polymerization and fundamentals of neural nets. In addition, he wrote key papers on sequential analysis and design, threshold models for epidemics and wrote a book on the foundations of probability.
In a biographical essay he wrote in 1994 titled "Almost Home," he said the following: "I would name two personal qualities, which I recognise in myself and which I think have been useful, although some will rightly say that they have their negative obverses. One is laziness. If a piece of work is heavy and complicated then it is wrong. In calculations one will of course persist for quite a way with mounting complication and obscurity, but there has to be a resolution, a simplification at some point, if one is on the right path. Of course, the simplicity may be 'simplicity' in a subtler sense than one would before have understood. This raises the second quality, which is almost stochastic in nature. One has to keep a particular openness of mind. Solving a problem is like going to a strange place, not to subdue it, but simply to spend time there, to preserve one's openness, to wait for the signals, to wait for the strangeness to dissolve into sense".
Whittle was born and raised in Wellington, New Zealand. He received a PhD in mathematical statistics in 1951 at Uppsala University, Sweden.
In 1957 he was appointed Lecturer in Mathematics at the University of Cambridge. He left in 1961 to become Professor of Mathematical Statistics at the University of Manchester. In 1967 he returned to Cambridge to take up the Churchill Chair of Mathematics for Operational Research. From this position, which he held to his retirement in 1994, he was a leading force in making operations research a recognized discipline in England.
Two people were awarded the 1986 Lanchester Prize: Alexander Schrijver for his research monograph titled Theory of Linear and Integer Programming, and Peter Whittle for his research monograph Systems in Stochastic Equilibrium. Both works were published by John Wiley in 1986.
The citation for Peter Whittle's monograph:
- "One of the most striking achievements in Operations Research has been the development of the theory of queueing networks. This began with the independent works of J.R. Jackson in 1963, W.J. Gordon and G.F. Newell in 1967, and P. Whittle in 1967, 1968. Since then, Peter Whittle has made several significant contributions to this subject culminating in this research monograph titled Systems in Stochastic Equilibrium.
-
"This is a study of a system of interacting components (e.g., queues, computers, work stations, molecules, organisms, subsystems) between which move units (e.g., customers, data packets, material, quanta of energy or resources). Whittle focuses on what he calls systems of weakly coupled components. Building on his earlier contributions to reversible Markov processes and partial balance equations, he establishes the equilibrium distribution for a variety of systems of weakly coupled components. In doing so, he shows that weak coupling underlies statistical mechanics models with a Gibbs distribution, kinetics of chemical reactions, resource exchange in ecological and economic systems, polymerization, randomly changing graphs, and queueing networks. The major topic of queueing networks includes novel treatments of functionally dependent queues, optimization of customer routing to avoid congestion, and insensitivity of equilibrium distributions to general service times. An additional feature is a lucid introduction to random fields (viz., Markovian and Gaussian fields), whose spatial and time evolution properties resemble those of systems of interacting components.
- "Whittle's book is an eloquent penetration into the underlying structure of a wide class of stochastic systems. It is remarkable in its breadth of applications as well its depth of ideas. Its refreshing style of writing leaves the reader unencumbered by heavy technicalities. Its new perspective on the modeling of stochastic systems go beyond the ordinary. We therefore award it the Lanchester Prize for 1986."