9:10am-10:00am
Urban Network Design
Hani S. Mahmassani, PhD, William A. Patterson Distinguished Chair in Transportation, Northwestern University
In this talk, Hani Mahmassani will highlight developments in technology, methodology and operational concepts for the design of integrated management strategies for urban transportation systems—to reduce congestion and improve environmental quality. The main premise is that these developments enable existing infrastructure to deliver significantly higher throughput than under business-as-usual operation. However, existing deployments have fallen short of lofty expectations put forward by advocates. The intelligence for effective application of these technologies has been inadequate, and the deployment of physical infotech infrastructure has lagged behind. Mahmassani will propose effective algorithms based on prediction-based look-ahead, given sensor and probe data on current conditions, and make the case for the role of pricing and real-time information, in conjunction with advanced traffic control measures, in the management of transportation networks and delivery of urban transportation services. He will discuss methodological implications and approaches for both off-line evaluation of these strategies, as well as real-time operational decision-making in this context. Developments in dynamic network modeling tools will also be discussed, with particular focus on new methods to evaluate the impact of congestion pricing in networks with heterogeneous users. Models of user route choice and trip timing decisions in response to pricing, information and travel time reliability will be incorporated in simulation-based dynamic assignment algorithms. The role of pricing as a real-time management tool, in conjunction with information supply is discussed, along with strategies for anticipatory pricing in conjunction with online network state prediction tools.
10:30am-11:20pm
Open-Source Software and Operations Research
Brady Hunsaker, PhD, Software Engineer, Google Pittsburgh
While the term "open-source software" has been around for 10 years, the software itself has been in existence much longer. It appears regularly in the news as more and more companies make use of it to varying degrees. Nevertheless, there remains considerable confusion about what exactly open source is and isn't. This tutorial will introduce open-source software and help put it in context, while also providing an overview of software for operations research. Topics will include:
• What is and is not open-source software;
• Similar and related concepts, such as "free software" and shareware;
• Common myths about open-source software;
• Differentiating software licenses from business and development models;
• Advantages and disadvantages of open-source software;
• Relationship to other "open" collaboration models, such as for text or art;
• Open-source software projects relating to operations research.
11:30am-12:20pm
Nonlinear Complementarity Problems and Extensions
Michael C. Ferris, PhD, Professor of Computer Sciences and Industrial & Systems Engineering, University of Wisconsin
While optimizers are familiar with complementary slackness within the optimality conditions of linear and nonlinear programming, there are numerous complementarity problems arising naturally in many practical applications from engineering and economics. These include applied general equilibrium modeling, traffic network design, structural engineering and finance. In this session, Ferris will outline several examples, together with an overview of modern, practical modeling and solution techniques within this field. Since complementarity allows for competition among players, optimization problems that involve complementarity constraints, and models with embedded complementarities are becoming increasingly important within applications. Ferris will introduce these new ideas, highlight several computational schemes and explain their utility by application.
2:00pm-2:50pm
Probability Management: A Cure for the Flaw of Averages
Sam L. Savage, PhD, Consulting Professor of Management Science and Engineering, Stanford University, and Senior Associate, Cambridge University Judge School of Business
The flaw of averages occurs when uncertainties are replaced by single average numbers in business or military planning. Consider, for example, the statistician who drowns fording a river with an average depth of three feet. Probability management addresses this problem through a combination of technologies and management practices. In this tutorial, Savage will explore:
• Interactive visual simulation, doing for distributions what the spreadsheet did for numbers;
• 2nd generation stochastic information systems for storing distributions instead of numbers;
• The role of the chief probability officer, CPO;
• Examples applications:
1. Managing the exploration portfolio of a major oil firm,
2. Interactive simulation of retirement plans at a trust company,
3. Incentive planning at a large bank.
For more information, see www.ProbabilityManagement.org
3:30pm-4:20pm
What's the Use of System Dynamics?
James Hines, Ventana Systems
?System dynamics is an approach to understanding the system structure that causes patterns of events. The system dynamics "trick" is to first see one's problem as a pattern of behavior and to then think of which feedback loops (i.e. structure) can generate it. Computer simulation brings big benefits, helping us grasp the impact of important feedback loops which then lead us to better, sometimes surprising, solutions. This session illuminates system dynamics in the context of a story: "There once was a bank threatened by losses on mortgage loans…"
4:30pm-5:20pm
Achieving Optimal Solution Performance for Your Optimization Modeling Project
Bjarni Kristjansson, BS, President, Maximal Software, Inc.
When working with real-world, large-scale optimization models, there are many factors that come into play on how to gain the optimum solution speed. Most of the attention has typically been on choosing the fastest solver algorithms, but there are other factors that can be just as important.
For example:
• How to formulate the model to best utilize the sparsity of the data;
• Type of data sources (text, binary, XML, spreadsheet, databases);
• The solver technology that best fits the overall project;
• Tuning solver parameters to achieve optimal performance;
• Type of hardware (PC vs. Unix, 32-bit vs. 64-bit, Intel vs. AMD);
• Advanced technologies that have recently become popular, such as multi-core CPU's and virtualization.
In this presentation, Kristjansson will discuss his group’s experiences with the above and report benchmark testing they have conducted.
Tuesday, April 15
9:10am-10:00am
The Art of Modeling
Jeffrey D. Camm, PhD, Professor and Department Head, Quantitative Analysis and Operations Management, University of Cincinnati College of Business
This tutorial will be a hands-on session on modeling. Camm will begin with a brief discussion of how to bring structure to the process of going from “mess to model.” The focus will then narrow to some useful techniques for optimization models including some discussion of:
• Modeling for insight – model so you can answer every question the client might ask;
• Modeling for solvability - using sets to avoid more binary variables and other tricks of the trade;
• The practice of the alternative - - giving your client some alternatives.
Only a basic understanding of linear and binary optimization models is necessary for this session to be valuable.
10:30am-11:20am
Multi-Echelon Inventory Optimization
Sean Willems, Boston University
?From a planning perspective, two significant events have occurred in the past five years. First, leading-edge companies completed their implementation of advanced planning and scheduling (APS) systems. Second, companies formalized sales and operations planning (S&OP) processes to better match supply and demand. This establishes a foundation where once a plan is agreed to, it can be executed successfully. However, companies are not realizing all the benefits they expected from these initiatives. Existing APS systems and S&OP processes do a fine job of deterministic optimization but a poor job of variability optimization. The next step in continuing to improve corporate performance involves establishing a platform for variability optimization within the company. The easiest first area to optimize relates to inventory.
This talk will cover:
• An overview of the different technology choices available to optimize inventory levels, spanning single-stage calculations in Excel to multi-echelon tools;
• Best practices on integrating inventory planning into the S&OP process;
• A road-map for a supply chain professional (as opposed to an operations researcher) looking to take their employer's Inventory practices to the next level.
This talk will not delve into the mathematical underpinnings of multi-echelon inventory optimization. Instead, it will serve as a practical tutorial on how best to implement this kind of technology.
11:30am-12:20pm
Probability Management: A Cure for the Flaw of Averages
Sam L. Savage, PhD, Consulting Professor of Management Science and Engineering, Stanford University, and Senior Associate, Cambridge University Judge School of Business
The flaw of averages occurs when uncertainties are replaced by single average numbers in business or military planning. Consider, for example, the statistician who drowns fording a river with an average depth of three feet. Probability management addresses this problem through a combination of technologies and management practices. In this tutorial, Savage will explore:
• Interactive visual simulation, doing for distributions what the spreadsheet did for numbers;
• 2nd generation stochastic information systems for storing distributions instead of numbers;
• The role of the chief probability officer, CPO;
• Examples applications:
4. Managing the exploration portfolio of a major oil firm,
5. Interactive simulation of retirement plans at a trust company,
6. Incentive planning at a large bank.
For more information, see www.ProbabilityManagement.org
3:10pm-4:00pm
Designing Reverse Supply Chains to Maximize Value Recovery for Product Returns
V. Daniel R. Guide, Jr., Associate Professor of Operations & Supply Chain Management, Smeal College of Business, The Pennsylvania State University
The increasing flow of product returns has become a major concern for many manufacturers. Annually, the total value of product returned by consumers in the U.S. alone exceeds $100 billion. Conventional wisdom dictates that manufacturers minimize the costs of these returns, but this implies there is a single design for reverse supply chains. Using simple analytic models that take into account the marginal value of time, Guide will explain how to develop and implement a decision-making tool that shows the right reverse supply chains design may be responsive or efficient. He will explore the drivers that influence the design to maximize value-recovery based on projects with multinational corporations, and discuss a variety of related issues, including:
• How product recovery forms the foundation for sustainability,
• The five basic processes required for reverse supply chains,
• Different kinds of product returns and their design implications.
4:10pm -5:00pm
Prediction and Decision Markets
Robin Hanson, PhD, Research Associate, Future of Humanity Institute at Oxford University, and Associate Professor of Economics, George Mason University
Prediction markets have been in the news a lot lately, forecasting the twists and turns of the presidential election. But the real action has been corporations pioneering the use of such markets to forecast sales, schedule dates, and much more. This tutorial will review:
• Track records and experiences of firms so far;
• Various trading and survey mechanisms;
• Incentives, such as cash, bragging rights, and info accounting;
• Ways to deter foul play, such as sabotage, manipulation, and retribution;
• Legal issues, such as gambling law, insider-trading law, and anti-trust law;
• Ways for a few participants to produce combinatorial spaces of estimates. |
