2015 INFORMS Simulation Society Research Workshop

Simulation: At the Interface of Simulation and Optimization

July 25-27, 2015 at Purdue University, West Lafayette, IN, U.S.A.

The 2015 Simulation Workshop was co-hosted by the Department of Statistics and the Department of Industrial Engineering at Purdue University, West Lafayette, IN. The workshop, with the theme “At the Interface of Simulation and Optimization,” brought together a small group of carefully selected researchers from different sub-cultures (e.g., stochastic programming, approximate dynamic programming, simulation optimization, machine learning) in an earnest attempt at cross-fertilizing and reconciling what appear to be contemporaneous advances on several important questions.

There were a diverse set of talks on topics such as function structure detection and exploitation in high-dimensional optimization, optimal stochastic sampling within iterative schemes, fundamental relationships and performance limits for algorithm and problem classes, alternative tractable stochastic optimization formulations, high-dimensional data driven models and optimization, learning in optimization, iterative and computing adaptations for high-dimensional optimization, and the treatment of multiple stochastic performance measures in optimization. Over the three-day period July 25 — July 27, 2015 the workshop featured three plenary talks, and a single track of 30-minute invited talks with ample time for discussion.

Large Deviations and Selection of the Best System
Peter Glynn, Stanford University

Markov Chain Monte Carlo for Continuous-Time Discrete-State Systems
Vinayak Rao, Purdue University

A Multi-Resolution Gaussian Markov Random Field Model for Discrete Optimization via Simulation
Barry Nelson, Northwestern University 

Particle Filtering Approaches for Dynamic Stochastic Optimization
John Birge, University of Chicago

Bayesian Optimization
Peter Frazier, Cornell University

Efficient Ranking and Selection in Parallel Computing Environments
Shane Henderson, Cornell University

Multi-objective Simulation Optimization on Finite Sets
Susan Hunter, Purdue University

On Multiple Roles of Regularization in Stochastic Programming
 Suvrajeet Sen, University of Southern California

Data and Decision: A Risk-Averse Perspective
 Johannes Royset, Naval Postgraduate School

Efficient and Optimal Mode Estimation using kNN graphs
Samory Kpotufe, Princeton University

Clearing the Jungle of Stochastic Optimization
Warren Powell, Princeton University

Tractable Simulation-Optimization: A Robust Optimization Approach
Chaithanya Bandi, Northwestern University 

Budget-Constrained Stochastic Approximation
Uday Shanbhag, Penn. State Univ

Adaptive Sampling for Large-Scale Sim. Opt. & An Inexact Line-Search Algorithm
 Raghu Pasupathy, Purdue University

QNSTOP: A Computational Laboratory for Stochastic Optimization
Michael Trosset, Indiana University

Unconstrained trust region based stochastic optimization with biased and unbiased noise
 Katya Scheinberg, Lehigh University

A Model-based Trust Region Method for Stochastic Derivative-free Optimization
Jeff Larson, Argonne National Labs 

In-network Nonconvex Big Data Optimization
Gesualdo Scutari, Purdue University

Vanishing Duality Gap in Large Collaborative Optimization
Mengdi Wang, Princeton University

Convex hull of permutation invariant sets and applications
Mohit Tawarmalani, Purdue University

Risk neutral, risk averse and distributional robust approaches to multistage stochastic programming
Alexander Shapiro, Georgia Tech

On the Use of Phi-Divergences for Data-Driven Stochastic Optimization
Güzin Bayraksan, Ohio State University

Simulation optimization under input uncertainty
Enlu Zhou, Georgia Tech

Exploiting Structure in Approximate Dynamic Programming
Warren Powell, Princeton University

New Approaches to Information Collection
 Roberto Szechtman, NPS

Online A-B Testing
Vivek Farias, MIT

A New Framework for Multi-fidelity Simulation Optimization
Jie Xu, George Mason University

Optimal Resource Capacity Management in Stochastic Networks
Soumyadip Ghosh, IBM TJ Watson Research

Tail analysis without tail information: A worst-case perspective
 Henry Lam, University of Michigan

Lasso-Optimal Supersaturated Screening Designs and Analysis
Hong Wan, Purdue University