2015 Informs Annual Meeting

MA02

INFORMS Philadelphia – 2015

Monday, 8:00am - 9:30am

5 - Design of a Responsive Vaccine Supply Chain under Supply and Demand Uncertainty Stef Lemmens, KU Leuven, Naamsestraat 69 Box 3555, Leuven, 3000, Belgium, stef.lemmens@kuleuven.be, Nico Vandaele, Catherine Decouttere Both literature and industrial evidence emphasize the challenge and the importance of the design of a responsive vaccine supply chain. We model the interrelationships between multi-echelon inventory, production capacity and lead time and take supply and demand uncertainty into account by the use of a methodology which combines the guaranteed service approach and queueing theory. Furthermore, we show the results of applying our methodology to a real- life industrial rotavirus vaccine supply chain. Software Demonstration Cluster: Software Demonstrations Invited Session 1 - Artelys - See the Artelys KNITRO 10.0 Optimization Solver in Action Richard Waltz, Senior Scientist, Artelys Corp KNITRO is the premier solver for nonlinear optimization and recent winner of the GECCO 2015 black-box optimization competition, finishing first among 28 solvers. This software demonstration will highlight the latest KNITRO features, including a new object-oriented interface and new SQP algorithm for derivative- free optimization (DFO). The demo will also provide an overview of how to effectively use KNITRO in a variety of environments and applications, and present Steve Dirkse, Director of Optimization, GAMS Development Corp This workshop will show you how to use the General Algebraic Modeling System (GAMS) in an efficient and productive way. There will be an introduction to the system and a presentation of the key concepts in GAMS. The largest part of the workshop consists of hands-on exercises. Amongst others, it will be demonstrated how GAMS interacts with other applications and you will see how to analyze and debug problems using the tools available within GAMS. recent benchmarking results for DFO and nonlinear optimization. 2 - GAMS Development Corp - GAMS – An Introduction SD79 79-Room 302, CC

MA01 01-Room 301, Marriott Military O.R. and Applications III Sponsor: Military Applications Sponsored Session

Chair: Michael Hirsch, ISEA TEK, 620 N. Wymore Rd., Ste. 260, Maitland, FL, 32751, United States of America, mhirsch@iseatek.com 1 - Electronic Attack Decision Framework using Pomdp Brandon Ha, Sr. Systems Engr Ii, Raytheon Company, 2000 E. El Segundo Blvd, E1/B2208D, El Segundo, CA, 90245, United States of America, Brandon.C.Ha@raytheon.com The objective of this research is to develop a suite of machine learning algorithms to address the need for engaging with future advanced unknown agile RF threats and co-evolve with the adversary’s response. Our approach uses POMDP to represent the unobservable elements in the environment and actions that can provide partial information about these elements – to learn (characterize) the unknown emitters and then to predict the intent and deploy optimal EA technique(s). 2 - Pursuit on a Graph using Partial Information: Max-delay David Casbeer, Dr., Air Force Research Laboratory, 2210 8th Street, B20146 R300, Wright Patterson AFB, OH, 45433, United States of America, david.casbeer@us.af.mil, Krishna Kalyanam, Meir Pachter The optimal control of a “blind” pursuer searching for an evader on a graph is presented. At specific locations on the graph (road network), unattended ground sensors (UGS) have been placed which detect the intruder. The pursuer (UAV) visits the sensors and decides where to travel in order to capture the evader. An algorithm is presented to compute the maximum initial delay for which capture is guaranteed. The algorithm also returns the corresponding optimal pursuit policy. 3 - Improved Sensor Placement in Multistatic Sonar Networks Multistatic sonar networks containing non-collocated sources and receivers represent an important generalization of traditional sonar systems. Although they convey many tactical and operational advantages, multistatic sensor networks are difficult to model and to employ optimally. We discuss the multistatic sensing problem and describe algorithms for placing sources and receivers. 4 - Optimal Deployment of Network Defenses David Myers, Research Engineer, United States Air Force, 26 Bedford Drive, Whitesboro, NY, 13492, United States of America, david.djm.myers@gmail.com Optimally deploying an ever-growing slate of network defense capabilities, while maintaining the ability to perform the mission, is a critical component of future USAF operations. Utilizing a system’s attack graph, this defense configuration problem (DCP) is a network interdiction problem where the network defender is the interdictor and the attacker is the evader. The purpose of this presentation is to formulate and describe the DCP and discuss extensions into a dynamic posture problem. MA02 02-Room 302, Marriott Game Theory in Practice for Homeland Security Cluster: Homeland Security Invited Session Chair: Milind Tambe, USC, 941 Bloom Walk, Los Angeles, CA, United States of America, tambe@usc.edu 1 - Game Theoretic Applications in Coast Guard Operations Erich Stein, USCG, 1 Chelsea St., New London, CT, 06339, United States of America, Erich.V.Stein@uscg.mil, Craig Baldwin, Sam Cheung The Coast Guard has tested and operationalized game theory applications in several mission areas including port security and fisheries. A model was created to mitigate effects of illegal fishing and generate schedules for USCG assets. The Port Resilience Operational Tactical Enforcement to Counter Terrorism (PROTECT) game model optimizes limited security resource allocations. Finally, development of innovative patrol strategies for drug and migrant interdiction efforts is ongoing. W. Matthew Carlyle, Naval Postgraduate School, mcarlyle@nps.edu, Emily Craparo, Mumtaz Karatas, Christoph Hof

145