Informs Annual Meeting 2017

MC73

INFORMS Houston – 2017

MC71

for logistics during extreme events, as it can parallelize the last mile of delivery. It also enables city logistics, where smaller, more flexible, and maneuverable vehicles are in charge of final delivery. Last, with the increasing availability of unmanned vehicles, it helps automate the supply process, while also serving areas that are difficult to reach or are rendered inaccessible. 2 - Toward the Development of Resilience Indices for Multi-echelon Assembly Supply Chain Networks Huy Q. Nguyen, Rensselaer Polytechnic Institute, 110 8th Street, 5119 Center for Industrial Innovation, Troy, NY, 12180, United States, nguyeh7@rpi.edu, Thomas Sharkey, William A. Wallace, John E.Mitchell We develop scheduling rules to optimize the recovery performance of multi- echelon assembly supply chain networks (MEASC) from large-scale disruptive events. Each supplier within the MEASC network assembles a component from a series of sub-components received from other suppliers. Utilizing these decision rules, we develop resilience indices to help supply chain managers evaluate the vulnerability of their networks to different types of disruptive events. Our decision rules and indices are applied on a data set modeling an industry partner’s supply chain. 3 - An Optimization Model for Multiagent Path Selection and Scheduling with Geographic Conflict Navid Matin Moghaddam, nmatinmo@asu.edu, Jorge A. Sefair We study the problem of finding the minimum arrival time of multiple agents traveling between known origins and destinations in a network. Paths must not be conflicting, meaning that there must be a minimum distance between agents at any time. We discuss the problem complexity and present an optimization model to find the path and schedule for each agent to traverse the network. Typical examples of this problem include the transportation of hazardous materials and transportation operations under geographic failures. We present a cutting-plane exact solution approach to solve this problem. 4 - Restoration of Interdependent Infrastructures using Integrated Network Design and Scheduling Problems with Machine Movement Interdependence Aniela Garay Sianca, University of Arkansas, Fayetteville, AR, 72701, United States, agaraysi@email.uark.edu, Sarah G. Nurre Researchers have examined how to restore interdependent infrastructure systems by deciding what components are repaired, when the repairs are scheduled, and who performs each repair using integrated network design and scheduling problems. A main limitation of this work is the assumption that a machine can feasibly move using the transportation network when assigned two consecutive tasks. Flooding and other obstructions on the transportation network often prohibit such assignments. We present a new optimization model by explicitly including the machine movement interdependence. We present the results of computational experiments for the interdependent infrastructure networks of Panama. 372B Statistical Learning and Optimization Invited: Operations Research and the Future of Computing Invited Session Chair: Emma Frejinger, Universite de Montreal, Universite de Montreal, Montreal, QC, H3C 3J7, Canada, emma.frejinger@cirrelt.ca 1 - On the Role of (Machine) Learning in (Mathematical) Optimization Andrea Lodi, École Polytechnique Montréal, Montréal, QC, Canada, andrea.lodi@polymtl.ca In this talk, I try to explain my point of view as a Mathematical Optimizer — especially concerned with discrete (integer) decisions — on Big Data. I advocate a tight integration of Machine Learning and Mathematical Optimization (among others) to deal with the challenges of decision-making in Data Science and I discuss in details two areas in which machine learning techniques have been (successfully) applied to mixed-integer programming. 2 - Discrete-continuous Maximum Likelihood Estimation Anna Fernandez-Antolin, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland, anna.fernandezantolin@epfl.ch, Virginie Lurkin, Matthieu de Lapparent, Michael Bierlaire In statistics, maximum likelihood estimation involves solving a nonlinear optimization problem, that is often non-convex. It generates statistical estimates of the unknown parameters of a model. In the context of discrete choice modeling, the structure of the model is also unknown. For instance, a nested logit model assumes a partition of the choice set. Obtaining the partition from the data assumes involving discrete variables into the maximum likelihood estimation problem. In this paper, we aim at formulating the problem as a mixed integer linear problem (MILP). MC73

371F New Results on Integer Programming Sponsored: Optimization, Global Optimization Sponsored Session Chair: Daniel Bienstock, Columbia University, New York, NY, 10027, United States, dano@columbia.edu 1 - Extension Complexity Lower Bounds for Mixed-integer Extended Formulations Robert Hildebrand, IBM T.J. Watson Research Center, Yorktown Heights, NY, United States, robDhildebrand@gmail.com We prove that any mixed-integer linear extended formulation for the matching polytope of the complete graph on n vertices, with a polynomial number of constraints, requires $\Omega(\sqrt{\sfrac{n}{\log n}})$ many integer variables. By known reductions, this result extends to the traveling salesman polytope. This lower bound has various implications regarding the existence of small mixed- integer mathematical formulations of common problems in operations research. This provides a first non-trivial lower bound on the number of integer variables needed in such settings.This is joint work with Robert Weismantel and Rico Zenklusen 2 - Cutting Planes from Reverse Polars of Disjunctive Sets Egon Balas, Professor, Carnegie Mellon University, Posner Hall Rm 232B, 5000 Forbes Avenue, Pittsburgh, PA, 15213-3890, United States, eb17@andrew.cmu.edu, Aleksandr Kazachkov Lift-and-project cuts from a union of polyhedra can also be obtained without lifting, from the intersection of the reverse polars of those polyhedra, involving their V-polyhedral representation.. In particular, strong cuts can be obtained from the intersection of the reverse polars of polyhedra representing the active leaves of a branch-and-bound tree. We explore and test this approach. 3 - Aggregation-based Cutting-planes Santanu Subhas Dey, Georgia Institute of Technology, School of Industrial and Systems Engineering, 765 Ferst Drive Nw,, Atlanta, GA, 30332, United States, santanu.dey@isye.gatech.edu, Merve Bodur, Alberto Del Pia, Marco Molinaro, Sebastian Pokutta We study the strength of aggregation cuts for packing and covering integer programs (IPs). Aggregation cuts are obtained as follows: Given an IP formulation, first generate a single implied inequality using aggregation of the original constraints, then obtain the integer hull of the set defined by this single inequality with variable bounds, and finally use the inequalities describing the integer hull as cutting-planes. We show that for packing and covering IPs, the aggregation closures can be 2-approximated by simply generating cuts from original formulation constraints. On the other hand, aggregation cuts can be arbitrarily Andres Gomez, University of California, 2020 Delaware Street, Apt 3, Berkeley, CA, 94709, United States, a.gomez@berkeley.edu, Alper Atamturk We describe convex valid inequalities for conic quadratic mixed-0-1 optimization. The inequalities are based on decomposing the quadratic term into two components, one of which is submodular. We prove that the proposed inequalities completely describe the convex hull of a single conic constraint over binary variables and unbounded continuous variables, and are also sufficient to describe the convex hull of rotated cone constraints with a diagonal quadratic term. Computational experiments with value-at-risk minimization and assortment optimization problems indicate that the inequalities strengthen the convex relaxations substantially and lead to significant performance improvements. 372A Network Optimization Models and Applications Sponsored: Optimization, Network Optimization Sponsored Session Chair: Jorge A. Sefair, Arizona State Univerity, Tempe, AZ, 85287-8809, United States, jorge.sefair@asu.edu 1 - A Two Echelon Humanitarian Supply Chain Model Chrysafis Vogiatzis, North Dakota State University, 1057 35th Street North, Apt 105, Fargo, ND, 58102, United States, chrysafis.vogiatzis@ndsu.edu In this talk, we discuss two-echelon VRPs for logistics operations during a humanitarian crisis. It differs from typical multi-echelon supply chain management systems as there exist no intermediate facilities to organize goods for delivery; instead all decisions are made en-route. This paradigm offers advantages stronger than cuts from individual constraints for general IPs. 4 - On Exploiting Submodularity in Conic Quadratic Mixed-0-1 Optimization MC72

229