Wednesday, August 19

7:30-8:50 Registration and breakfast

8:50-9:00 Opening remarks (Tamás Terlaky)

9:00-10:00 Natalia Alexandrov, NASA Langley Research Center

Toward Optimal Transport Systems Chair: Larry Snyder Strictly reactive, evolutionary approaches to improving the air transport system no longer suffice in the face of the predicted growth in demand and must be supplemented with active design methods. The ability to actively design, optimize and control a system presupposes the existence of predictive modeling and reasonably well-defined functional dependencies among the controllable variables of the system and objective and constraint functions for optimization. We investigate functional relationships that govern the performance of transport systems with the aim of arriving at substantiated modeling, design optimization, and control methods.

10:00-10:30 Coffee break

10:30-12:00 Parallel technical sessions

Applications Chair: Suriya Ruangpattana	Geometric aspects of optimization Chair: Serge Kruk
Strategic Planning: OR to the Rescue Francis J. Vasko, Kutztown University	Face Lattice Computation Under Symmetry Adrian Burlacu, McMaster University
Application of Optimization to the Design of a Predictive Controller with Variable Time Switches Kaska Kowalska, McMaster University	Recent progress on the polytope diameter bounds David Bremner, University of New Brunswick
Motion Analysis and Modeling in Image-Guided Radiation Therapy Olesya Peshko, McMaster University	Self-correcting geometry in model based derivative-free optimization Katya Scheinberg, Columbia University

12:00-13:00 Lunch
13:00-14:30 Parallel technical sessions

Distribution networks Chair: Larry Snyder	Semidefinite programming and its applications Chair: Hayato Waki
Warehouse-Retailer Network Design: New Formulation and Approximation Algorithm Xi Wang and Jiawei Zhang, New York University	Shape-constrained spline estimation of multivariate functions using semidefinite programming David Papp, Rutgers Center for Operations Research
A new efficient heursitic for the design and analysis of distribution systems Zumbul Bulut, Lehigh University	New Relaxation Schemes for Polynomial Programing Juan Vera, University of Waterloo
A tree-based model for redundant multicast routing problem with shared risk link group diverse constraints Wanpracha Art Chaovalitwongse, Rutgers University	Identification and Elimination of Interior Points for the Minimum Enclosing Ball Problem Selin D. Ahipasaoglu, Cornell University

14:30-15:00 Coffee break
15:00-16:30 Parallel technical sessions

Financial optimization Chair: Aurélie Thiele	Algorithms Chair: Frank Curtis
Constant Rebalanced Portfolio Optimization under Nonlinear Transaction Costs Yuichi Takano, University of Tsukuba	The harmony search algorithm Zong Woo Geem, Johns Hopkins University
Log-robust portfolio management Ban Kawas, Lehigh University	On-line bin covering algorithms Pall Jensson, University of Iceland
Fractional Brownian motion and optimization by ellipsoids Vladimir Dobric, Lehigh University	Level set methods for finding critical points of mountain pass type C.H. Jeffrey, Pang, Fields Institute

16:30-16:45 Break
16:45-17:45 John M. Mulvey, Princeton University

Financial Risk Management and Portfolio Optimization: Lessons from the 2008 Economic Crash Chair: Aurélie Thiele Traditional (static) portfolio optimization models employed by many institutional and individual investors failed to protect the investor’s capital during the 2008 economic crash. As an alternative, we propose a short-horizon, multi-period portfolio model. The framework takes into account changing market conditions and possible contagion during turbulent periods. The system responses quickly to minimize draw down and related risks. Empirical results over past periods show the benefits of a dynamic, anticipatory approach.

18:30-21:30 Student social, Graduate Student Center

Thursday, August 20

8:00-9:00 Ravindra K. Ahuja, University of Florida, Gainesville

A Combinatorial Approach for Railroad Operating Plan Design Chair: Ted Ralphs A freight railroad’s Operating Plan dictates the movement of railcars, blocks, trains, crew and locomotives in a railroad’s network. It consists of consolidating railcars into blocks of cars, determining train schedules (how many trains to run; the origin, destination, and route of each train and the days of operation; the train arrival and departure times for each station at which it stops), and the assignment of blocks of cars to trains. Developing the railroad operating plan consists of solving two decision problems: the railroad blocking problem, and the train schedule design problem, each of which is a very large-scale decision problem containing billions of decision variables. We describe combinatorial algorithms to solve these two problems and provide computational results of these algorithms on the data provided by US freight railroads. In this talk, we will also share our experiences of working with industries and what it takes to build successful applications.

9:00-9:15 Break
9:15-10:45 Parallel technical sessions

Mixed-integer NLP Chair: Pietro Belotti	Energy Chair: Suriya Ruangpattana
Extending SCIP for MINLP Stefan Vigerske, Humboldt University Berlin	Formulations for the Unit Commitment Problem Jim Ostrowski, University of Waterloo
Mixed Integer Bilevel Programming Ted Ralphs, Lehigh University	Cancelled
Parametric nonlinear discrete optimization Jon Lee, IBM Research	Optimal Division of the Load Duration Curve for Managing Fuel Diversification Suriya Ruangpattana, Purdue University

10:45-11:15 Coffee break
11:15-12:45 AIMMS-CPLEX/MOPTA Optimization Modeling Competition Final

The three finalists are

AIMMS-CPLEX/MOPTA Optimization Modeling Competition Final

Team Konrad’s Truckers, Zuse Institute, Berlin, Germany Stefan Heinz, Jonas Schweiger, Thomas Schlechte, Advisor: Rüdiger Stephan

Team Twente, University of Twente, The Netherlands Ruben Hoeksma, Arjan Thomas, Erik-Jan Krijgsman, Advisor: Jacob Jan Paulus

Team SMU, Southern Methodist University, Dallas, TX, USA T. Jason Kratz, Angelika Leskovskaya, Ron Dearing, Advisor: Dick Barr

See the competition page for technical details. The winner will be announced at the conference banquet. Also, see the AIMMS session on Friday for an official solution of the problem.

12:45-13:45 Lunch
13:45-15:15 Parallel technical sessions

Nonlinear programming Chair: C. H. Jeffrey Pang	Discrete optimization Chair: Ted Ralphs
Mathematical programming is Turing-complete Leo Liberti, Ecole Polytechnique, LIX	On mixing inequalities: rank, closure and cutting plane proofs Oktay Gunluk, IBM Research
Network Equilibrium Problems and Hybrid Dynamical Systems Scott Greenhalgh, University of Guelph	The master equality polyhedron Sanjeeb Dash, IBM Research
Entropic optimization: a novel framework for optimization/decision making under uncertainty Eugene Perevalov, Lehigh University	Facets of multiple all-different constraints Serge Kruk, Oakland University

15:15-15:45 Coffee break
15:45-16:45 Panel discussion

Trends: the future of computational optimization and software

Ravindra Ahuja, University of Florida, Gainesville

Natalia Alexandrov, NASA Langley Research Center

Gertjan de Lange, Paragon Decision Technology

Jon Lee, IBM Research

Irv Lustig, ILOG

Tamás Terlaky, Lehigh University (moderator)

16:45-17:45 Pablo A. Parrilo, Massachusetts Institute of Technology

Bandgap Optimization of Photonic Crystals using Semidefinite Programming Chair: Imre Pólik Photonic crystals are structures with very interesting optical properties, that allow for the control and manipulation of light propagation. In this talk, we consider the optimal design of photonic crystal band structures on two-dimensional lattices. The mathematical formulation of the band gap optimization problem leads to an infinite dimensional eigenvalue optimization problem, parametrized by the dielectric material and the wave vector. Our computational approach to this challenging design problem is based on finite element techniques in combination with subspace methods and semidefinite programming. The solutions obtained by our numerical techniques exhibit patterns which go far beyond typical physical intuition on periodic media design (joint work with H. Men, C. Nguyen, R. Freund, J. Peraire).

18:30-21:30 Conference banquet, Banana Factory

Friday, August 21

8:00-9:00 Robert Weismantel, Otto-von-Guericke Universität Magdeburg

Mathematical proofs for questions arising in the synthesis and reconstruction of complex processes Chair: Pietro Belotti The first part of the talk is dedicated to the synthesis of chemical and, more specifically, chromatographic processes. In this context we study the problem of designing an ideal separation of a binary mixture. Interesting questions in this context are to detect the minimal number of stages per zone for given purity requirements etc. In order to answer questions of this kind, we develop proofs for the infeasibility of nonlinear mixed integer optimization models. Recent developments for the latter topic are discussed. The second part of the talk deals with the question of reconstructing the set of all possible explanations for a given series of observed experimental molecular data. We present an exact algorithm for solving this problem that is based on combinatorial arguments.

9:00-9:15 Break
9:15-10:45 Parallel technical sessions

Software Chair: Stefan Vigerske	Supply chains and pricing Chair: Larry Snyder
Choosing your own adventure: Automatic taxonomy expansion to permit many paths Xiaoguang Qi, Dawei Yin, Zhenzhen Xue, Brian D. Davison, Computer Science & Engineering, Lehigh University	Pricing with Markups under Horizontal and Vertical Competition Roger Lederman, Columbia Business School
The flexibility of the constraint integer programming solver SCIP Stefan Heinz, Zuse Institute Berlin	Distributing the Inter-Pool Moves and Asset Replacement in the Car Rental Business Gen-Han Wu, The Pennsylvania State University
Bigger, Better, Faster: update on Cbc, the world’s fastest open-source MIP solver Laszlo Ladanyi, IBM Research	A Network Optimization Model for a Candy Supply Chain Lawrence V. Snyder and Zeliha Akca, Lehigh University

10:45-11:15 Coffe break
11:15-12:45 Parallel technical sessions

AIMMS tutorial and demonstration Chair: Imre Pólik	Interior-point methods Chair: Tamás Terlaky
Basics Gertjan de Lange, Peter Nieuwesteeg, Paragon Decision Technology	Full-Newton step polynomial-time methods for LO based on locally self-concordant barrier functions Kees Roos, TU Delft
AIMMS-CPLEX/MOPTA Optimization Modeling competition solution Gertjan de Lange, Peter Nieuwesteeg, Paragon Decision Technology	Primal-dual Interior Point Algorithms for LCPs with Arbitrary Matrices Marianna Nagy, Eötvös Loránd University of Sciences, Budapest
Case studies Gertjan de Lange, Peter Nieuwesteeg, Paragon Decision Technology	Strange Behaviors of Interior-point Methods for Solving Semidefinite Programming Problems in Polynomial Optimization Hayato Waki, The University of Electro-Communications

12:45-13:45 Lunch
13:45-15:15 Parallel technical sessions

Multiobjective and nonsmooth optimization Chair: Robert Weismantel	Simulation and modeling Chair: Ben Felzer
A Sequential Quadratic Programming Method for Nonsmooth Optimization Frank Curtis, Lehigh University	Modeling of Birth Seasonality in Canada and the Provinces Arzu Sardarli, First Nations University of Canada
Multiobjective Optimization via Parametric Programming: Algorithms and Applications Oleksandr Romanko, McMaster University	Stochastic Sequencing and Scheduling of an Operating Room Camilo Mancilla, Lehigh University
Allocation of Overlapping Inventory in Online Advertising via MultiObjective Optimization John A. Tomlin, Yahoo! Labs	Using NCAR CCSM3 to Explore how Future Environmental Stresses to Plant Physiology Affect the Ocean’s Global ‘Conveyor Belt’ Benjamin S. Felzer and Kristian M. Douma, Lehigh University

15:15-15:30 Break
15:30-16:30 Paul I. Barton, Massachusetts Institute of Technology

Global Optimization of Algorithms Chair: Robert Storer The ability to evaluate convex and concave relaxations of nonconvex functions plays a crucial role in modern deterministic global optimization algorithms. One of the best methods for this is the recursive relaxation of factorable functions proposed by McCormick. We show how the basic ideas of McCormick can be extended to evaluate convex and concave relaxations of algorithms, in a similar fashion to the differentiation of algorithms by automatic differentiation. Moreover, since the resulting relaxations are usually nonsmooth functions, it is also possible to evaluate subgradients that are accumulated in either forward or backward modes. The notion of evaluating relaxations and subgradients of algorithms is then extended to the more general setting of fixed-point iterations for solving systems of nonlinear equations. In this, and other cases, the necessary propagation of the ranges of functions using interval arithmetic becomes very weak, and more tailored approaches have to be developed. We present our development of parametric interval Newton methods for this purpose. The ability to evaluate relaxations of algorithms motivates the development of a reduced-space approach to global optimization. Suppose that a model involves a few inputs and outputs, but a large number of internal state variables and equations that describe the mapping of the inputs to the outputs. Examples include chemical processes and unit operations, nonlinear networks, biological systems, etc. If the input-output mapping can be formulated as an algorithm, this can be relaxed directly, obviating the need to include the large number of internal state variables explicitly in the optimization problem. Since deterministic global optimization algorithms have worst-case exponential complexity, this reduced-space approach may make otherwise intractable problems solvable in reasonable times. We present our progress in developing reduced-space branch-and-bound, outer approximation and primal-relaxed dual algorithms that combine relaxations of algorithms, their subgradients and nonsmooth constrained convex solvers.