SPS Past Prize Winners

Celebrating excellence in Stochastic Programming—discover the past prize events that recognized outstanding contributions and achievements in the field.

2019: Dupačová-Prékopa Student Paper Prize Awards

First Prize: Philip Thompson (Instituto de Matemática Pura e Aplicada (IMPA), Rio de Janeiro)
Variance-Based Extragradient Methods with Line Search for Stochastic Variational Inequalities, SIAM Journal on Optimization, 29(1):175-206, 2019.

Advisors: Alfredo N. Iusem and Alejandro Jofre

Citation: The paper proposes the first provably convergent stochastic approximation method with variance reduction, either for stochastic variational inequalities or stochastic optimization, which is robust with respect to an unknown Lipschitz constant. This widens the applicability and improves the desired efficient acceleration of the existing variance reduction methods, all of which still assume knowledge of a Lipschitz constant. This paper breaks with the prevailing approach in stochastic optimization and obtains an optimal stochastic gradient descent algorithm with line search. This allows the authors to make only local assumptions. The paper is technically deep and includes advances in concentration inequalities as well as optimization theory. The paper has exceptional reach and is expected to significantly influence a large number of communities. Philip was the leading developer and author as confirmed with his co-authors.

Runner-up:

Xie Weijun (School of Industrial & Systems Engineering, Georgia Institute of Technology)
On Deterministic Reformulations of Distributionally Robust Joint Chance Constrained Optimization Problems, SIAM Journal on Optimization, 28(2):1151-1182, 2018.

Advisor: Shabbir Ahmed

Citation: he paper studies tractable reformulations of a very general class of distributionally robust joint chance constrained optimization problems (DRCCP) where the joint chance constraint is required to hold for all probability distributions of the stochastic parameters from a convex moment ambiguity set. In recent years there has been a flurry of activity in deriving convex reformulations of distributionally robust chance constraints in a variety of different settings. This paper unifies several existing proof techniques for reformulating DRCCP, and provides substantial extensions by considering joint chance constraints. Xie was the leading developer and author as confirmed with his author.

Prize Committee: Nilay Noyan (chair), Alejandro Jofre, and Johannes Royset