Next:
Introduction
Up:
Multi-agent transportation simulation
Previous:
Multi-agent transportation simulation
Contents
Introduction
Introduction
A quick tour
Introduction
Demand generation
Trip generation
Route generation
Traffic simulation
Feedback
Analysis
A do-it-yourself simulation package
Motivational start: Roundabout
Some basics of object-oriented programming
Introduction
Compilation of programs under Unix
Pointers
Structs
Classes and minimal memory management
Encapsulation
Constructors
Arrays of classes
The Standard Template Library (STL)
Associative arrays/maps
Methods; Inlining
References (``&'') in subroutine calls
``.'' vs. ``->''
General code structure
Review
Some programming recommendations
General
Programming language
Compiler error messages for STL code
Iterators
Tokenizer
Street network data and data structures
Introduction
Network file formats
Node class
SimWorld class
Nodes input
Link class
Links input
Incoming/outgoing links
Cellular automata micro-simulation
Introduction
Vehicles
Vehicles on links
Random moves through intersections
Fairer intersections
Initializing vehicles for testing purposes
Main program
Visualizer
Introduction
Vehicle output
Visualization via gnuplot
Testing the current status of the simulation
Plans following in the micro-simulation
Plans
Vehicle class
Plans format
ReadPlans
Class Plan
Park queue
Wait queue
Vehicle insertion
Plans following and vehicle arrival
Computational Speed
Events output
Modularization, inheritance, templates, and code re-use
Introduction
Links, Simlinks, and Inheritance
Templates
What belongs into the base class?
Route planner
Introduction
Fastest Path
Link travel times
Library support for graph algorithms
General structure
Input file: Trips
FindPath and Dijkstra
Plans output
Congestion-dependent router
Link travel times and congestion
Congestion dependency: Link travel times
Feedback/System integration
Introduction
Subset of trips file
Calling the router
Merging of the routes
Traffic simulation
Iterations
Activities planner: Adjust trip starting times
Introduction
Utilities
Basic idea
Dependence on departure time
Departure time selection
Operationalization
Input data: Activities file
Origin-destination travel times
Departure time choice
Feedback
Do-it-yourself transportation planning simulation: Summary
File formats summary
Nodes file
Links file
Snapshot file (visualizer output)
Plans file
Events file
Trips file
Activities file
Improvements
More realistic CA traffic simulation logic
Introduction
The stochastic traffic cellular automaton (STCA)
Some validation of the STCA
Lane changing
Validation of lane changing rules
Traffic signals
Validation of traffic signal rules
Unprotected turns
Validation of rules for unprotected turns
Discussion
The queue model for traffic dynamics
Introduction
General
Fair intersections
Limitations of the queue model
Routing
Time aggregation
Generalized cost functions
Alternative routes
Logit for routes
Planning for given arrival time
Mental maps
Non-car modes of transportation
Routing
Simulation
Demand
Origin-destination matrices
Activities-based demand modeling
Feedback
Introduction
Global trip times table
Agent data base
Day-to-day vs. within-day re-planning
Other Modules
Better file formats
Introduction
Use header line
XML
Some discussion
Parallel computing
Introduction
Micro-simulation parallelization: Domain decomposition
Graph partitioning
Adaptive Load Balancing
Performance prediction for the Transims micro-simulation
Speed-up and efficiency
Other modules
Summary
Distributed computing and truly distributed intelligence
Some background
Traffic flow theory
Introduction
Traffic flow measurements
Speed
Flow
Density
Fundamental diagrams
Car following
Reaction time argument for car following
Discrete space and discrete time: Cellular automata rules
Continuous space and continuous time
Discrete time and continuous space car following
Kinematic waves and fluid-dynamics
The Lighthill-Whitham-Richards equation
Linearization
Macroscopic shocks
The deterministic CA in terms of kinematic waves
More advanced fluid-dynamical models
Capacities, especially at bottlenecks
Cost-flow curves for static assignment
Static assignment
Introduction
Equilibrium principle
Beckmann's mathematical programming formulation
Constrained optimization
Uniqueness
Convexity of
Convexity of the feasible region
A solution method
Summary
Discrete choice theory
Introduction
Binary choice
Systematic vs random component of utility
Choice based on random utilities
Linear decomposition of systematic part of utility
Simple example
2nd example
Probability distributions, generating functions, etc.
Binary Probit (Randomness is Gaussian)
Gumbel distribution
Combination of Gumbel-distributed variables
Logistic distribution
Binary logit (randomness is Gumbel distributed)
Multinomial choice
Multinomial logit (MNL)
Discussion of modeling assumptions
Independence from irrelevant alternatives (IID)
Maximum likelihood estimation
... for binary choice in general
... for binary logit model
Discussion
The beta parameter from earlier
Summary
Axhausen lecture
Learning and feedback
Introduction
Additional aspects of day-to-day learning
Individualization of knowledge
Classifier System and Agent Database
Individual plans storage
Interpretation as dynamical system
Deterministic systems
Stochastic systems
Transients
Relation to game theory
Relation to machine learning
Smart agents and non-predictability
Conclusion
Calibration and validation
Traffic flow characteristics
Introduction
Validation, Calibration, etc.
The Transims microsimulation approach
Rules of the model
Single lane uni-directional traffic
Lane changing for passing
Lane changing for plan following
Unprotected turning movements
Signalized intersections
Unsignalized intersections
Parking locations
Parallel logic
Complete scheduling
Towards a standardized flow test suite for simulation models
Measured quantities
Test networks
Results
Yield sign behavior
Comparison to Case Study Logic
Short discussion
Summary and conclusion
Intersection test suite
Routing
A Dallas case - do I want this??
A Portland/Oregon case
Introduction
Problem statement
Our approach
Related work
Experimental setup and simulation results
Comparison to field data and to emme/2 study results
Discusssion
Summary
Acknowledgments
A Switzerland case
Acknowledgments
Bibliography
About this document ...
2004-02-02