2013 COSP Student Paper Prize Awards
In 2013, the prize committee decided to award two first prizes, one for theory and the other for modeling
First Prize - Theory
Ruiwei Jiang (University of Florida)
Data Driven Chance-Constrained Stochastic Program
Advisor: Yongpei Guan
Prize Committee: Georg Pflug (chair), Suvrajeet Sen, and Güzin Bayraksan
2010 COSP Student Paper Prize Awards
Uma Ravat (University of Illinois at Urbana-Champaign)
On the characterization of solution sets of smooth and nonsmooth stochastic Nash games
Advisor: Uday V. Shanbhag
Citation: The inaugural COSP Student Paper Prize has been awarded to Uma Ravat for the paper entitled 'On the Characterization of Solution Sets of Smooth and Nonsmooth Stochastic Nash Games'. Uma Ravat is currently a Ph.D. candidate in the Mathematics Department at the University of Illinois at Urbana-Champaign, working in the research group of Uday Shanbhag.
The paper studies game theoretic stochastic programs, in particular, a class of continuous strategy Nash games with uncertain data where the players solve expected value problems. A comprehensive analysis of the structural properties of these games is presented by providing conditions under which the corresponding sample point problems exhibit similar structural properties in an almost sure sense or with some positive measure. Several cases such as smooth and nonsmooth Nash games and games with coupled stochastic constraints are examined. The concepts are illustrated through important examples such as risk-neutral and risk-averse Nash-Cournot games and with expected value constraints.
The concept of Nash equilibrium was introduced by Nash in early 1950s and the stochastic programming models were introduces in mid-to-late fifties by Dantzig, 1955, Beale 1955, and Charnes and Cooper, 1959. While the two areas have had a steady stream of contributions, the avenues at the intersection of the two have been less explored. The paper makes a fundamental contribution in this avenue. Mathematically the paper is very mature, elegantly combining techniques from different areas. The exposition is clear and the theory is substantiated with important special cases. The paper opens up a new realm of applications in computational game theory.
Christian Kuechler (Humboldt University of Berlin)
On stability of multistage stochastic programs
Advisor: Werner Römisch
Citation: This citation goes to Christian Kuchler for his paper "On stability of multistage stochastic programs". The paper was written while the author was a doctoral student at Humboldt-University Berlin in the research group of Werner R\"omisch. It has been published in Volume 19, 2008, of SIAM Journal on Optimization.
In research on multistage linear stochastic programs a lasting discrepancy can be observed. On one hand, there is an abundance of algorithmic work, mainly for models with discrete probability distributions. On the other, fairly little is known on structure, and, more so, on stability of models in general settings. This is somewhat unsatisfactory since, quite often, models with discrete distributions arise from approximations of general models. Then it would be good to have some safeguards that solutions to approximate problems stay close to the solutions of the original one.
Kuchler's paper contributes to milden this discrepancy. The main result identifies sufficient conditions for the Lipschitz continuity of optimal values if the underlying stochastic data processes are subjected to perturbations. Some of the preparatory results, on Lipschitz continuity of recourse functions and calmness of optimal solutions, are substantial on their own.
The paper shows a remarkable analytical mastership. Profound knowledge of optimization and probability theory meets technical maturity in handling non-trivial estimates and recursions. Finding the "right" concepts is another strength of the paper. This concerns assumptions, which are strong enough for the desired conclusions but stay sufficiently general to include relevant problem classes.The paper is very well written. Despite the inevitable technical complexity of the topic the reader is never left alone. Coming investigations are skillfully motivated. Indispensability of assumptions and their role in related research are carefully discussed. With competence, concepts and results of the paper are put into perspective with the existing literature.
Prize Committee: Alan King (chair), Ruediger Schultz, and Güzin Bayraksan