OOPSLA 2002


Tracks
Technical Program
Tutorials
Workshops
DesignFest
Educators' Symposium
Doctoral Symposium
Demonstrations
Posters
Student Research
Competition

Student Volunteers
Special Events
Exhibits
Housing Information
Registration Information
Transportation

Thursday 13:00-13:45 – Convention Ctr - Exhibit Hall 4A
Friday 11:00-11:45 – Convention Ctr - Exhibit Hall 4A

20 UrbanSim

Bjorn Freeman-Benson
Department of Computer Science & Engineering, University of Washington
info@urbansim.org
David Socha
Department of Computer Science & Engineering, University of Washington
Julian Walker
Department of Computer Science & Engineering, University of Washington
Zach Musgrave
Department of Computer Science & Engineering, University of Washington
Derek Mehlhorn
Department of Computer Science & Engineering, University of Washington
Belen Roca
Department of Computer Science & Engineering, University of Washington

UrbanSim (http://www.urbansim.org /) is a software-based simulation model designed to be used for integrated planning and analysis of urban development, incorporating the interactions between land use, transportation, and public policy. It is intended for use by Metropolitan Planning Organizations and others needing to interface existing travel models with new land use forecasting and analysis capabilities.

UrbanSim is an open source (GPL) project written entirely in Java. The system has obvious and natural objects: households, jobs, land parcels, zoning rules, and so on. However, obvious and natural does not necessarily equate to small and efficient. Thus the interesting part of UrbanSim from the OOPSLA perspective is how we deal with the computational and storage problems that our obvious and natural objects give us.

The biggest requirement UrbanSim has is that its users (the modelers and planners) understand the system and trust the results. Thus our solution to the inefficient “obvious and natural” objects cannot be to make the objects efficient but convoluted and complex to the point where only an OOPSLA attendee would enjoy reading the code.

The system we have come up with is both easy for the non-programmer modelers to understand, and efficient enough for the MPOs to use. We have applied the system to Eugene, Honolulu, and Salt Lake City. We are currently applying it to Houston, Denver, Salt Lake City (again), and the Seattle/Puget Sound area.