Spring 2009 UMASS Amherst
Operations Research / Management Science Seminar Series

Date: Friday, April 24, 2009

Time: 11:00 AM
Location: Isenberg School of Management, Room 112

Speaker: Professor Hillel Bar-Gera

Dept of Industrial Engineering and Management
Ben Gurion University of the Negev
visiting Department of Civil Engineering, Purdue University

Biography: Dr. Hillel Bar-Gera is a tenured Senior Lecturer of Industrial Engineering and Management at Ben-Gurion University of the Negev, Beer-Sheva, Israel. His Ph.D. thesis on the origin-based assignment algorithm won the 2000 dissertation award for best Ph.D. thesis in transportation from the Transportation Science Section of INFORMS. He has been a faculty member at Ben-Gurion University since 2001. He received his B.Sc. in Mathematics, Physics and Computer Science as well as his M.Sc. in Mathematics from the Hebrew University, Jerusalem, Israel. He received his Ph.D. in Civil and Materials Engineering from the University of Illinois at Chicago. He serves as a member of the Intelligent Transportation Systems Committee of the Transportation Research Board, and as an Associate Editor of the journal Networks and Spatial Economics.

TITLE: Traffic Assignment by Paired Alternative Segments (TAPAS)
Abstract: The user equilibrium (UE) traffic assignment is a cornerstone in travel forecasting and traffic impact analysis. Many algorithms have been proposed over the years for solving the UE model. Recently, attention is given not only to the total link flow results of the UE model, but also to the route flow results of the model, and particularly to the Maximum Entropy User Equilibrium (MEUE) route flow solution that maintains desirable consistency properties.

In this research we experiment with a new algorithmic approach that focuses on local paired alternative route segments. The approach relies on a fundamental understanding that paired alternative segments "capture" the essence of route choices in the UE model in an extremely compact manner.

Traffic Assignment by Paired Alternative Segments (TAPAS) is superior to previous algorithms in several respects. First, it converges to equilibrium very quickly, as we demonstrate with several large-scale networks. In that respect it is a competitive candidate relative to other recent alternatives. In addition, the set of routes used by the method is relatively consistent throughout the iterative process, thus it allows achieving reasonable consistency even at moderate levels of convergence. Last, but not least, while certain aspects of the algorithm rely on non-trivial theoretical insights, the key principle is rather simple and likely to be widely understood by the general professional transportation community.

This series is organized by the UMASS Amherst INFORMS Student Chapter. Support for this series is provided by the Isenberg School of Management, the Department of Finance and Operations Management, INFORMS, and the John F. Smith Memorial Fund.

Dr. Anna Nagurney, the John F. Smith Memorial Professor of Operations Management in the Isenberg School of Management,  is the Faculty Advisor of the Speaker Series.