Past Awards
The 2024 INFORMS John von Neumann Theory Prize is awarded to Jim Dai for his fundamental and sustained contributions to stochastic systems theory, most prominently for his seminal work on stochastic network stability and heavy traffic diffusion approximations.
Dai’s most important contribution is his 1995 paper that relates the stability of fluid models to the positive recurrence of Markov processes. Although heuristic fluid models, in conjunction with Foster’s criterion, had been used before to establish stability for some models involving countable Markov chains, Dai’s seminal contribution was to develop a general, direct, and rigorous connection between the stability of fluid models and the positive recurrence of general state-space Markov processes, such as those that describe queueing networks. More concretely, he showed that stability of a deterministic fluid model implies stability of the stochastic process model, in a very general setting. Since then, Dai’s approach has become a centerpiece in the field of stochastic networks and the foundation for a wide array of subsequent results by Dai and others.
Beyond fluid analysis, Dai has made major contributions to the broader field of stochastic networks, such as developing heavy traffic diffusion approximations for certain models (e.g., multiclass service stations with Markovian feedback), producing a famous counterexample (in collaboration with Wang) about the (non)existence of diffusion approximations, developing new results and insights on reflected Brownian motion models, introducing a new approach for obtaining quantitative bounds for diffusion approximations based on Stein’s method, generalizing the scope and asymptotic optimality of max-weight scheduling, and many others.
Dai’s work is motivated by models that arise in manufacturing and networking, which he connects to paradigmatic mathematical problems. His work is distinguished by intellectual taste, mathematical depth, and profound originality. In addition, he has been a leading educator and mentor with a very strong record of service to the profession.
"On Positive Harris Recurrence of Multiclass Queueing Networks: A Unified Approach Via Fluid Limit Models," Annals of Applied Probability, 5:49-77, 1995.