TrafficAssignment.jl

Julia package for finding traffic user equilibrium flow
Popularity
35 Stars
Updated Last
1 Year Ago
Started In
February 2015

TrafficAssignment.jl

Build Status codecov

This package for the Julia Language does basically two tasks: (1) loading a network data and (2) finding a user equilibrium traffic pattern. See Traffic Assignment.

Install

julia> Pkg.add("TrafficAssignment")

This will install LightGraphs.jl and Optim.jl, if you don't have them already.

To check if works

julia> Pkg.test("TrafficAssignment")

load_ta_network

This function loads a network data available in this TNTP github repository. The network name must match with the directory name in the TNTP repository.

Example:

using TrafficAssignment
ta_data = load_ta_network("SiouxFalls")
# ta_data = load_ta_network("Anaheim")
# ta_data = load_ta_network("Barcelona")
# ta_data = load_ta_network("Winnipeg")

The return value is of the TA_Data type, which is defined as

mutable struct TA_Data
    network_name::String

    number_of_zones::Int
    number_of_nodes::Int
    first_thru_node::Int
    number_of_links::Int

    init_node::Array{Int,1}
    term_node::Array{Int,1}
    capacity::Array{Float64,1}
    link_length::Array{Float64,1}
    free_flow_time::Array{Float64,1}
    b::Array{Float64,1}
    power::Array{Float64,1}
    speed_limit::Array{Float64,1}
    toll::Array{Float64,1}
    link_type::Array{Int64,1}

    total_od_flow::Float64

    travel_demand::Array{Float64,2}
    od_pairs::Array{Tuple{Int64,Int64},1}

    toll_factor::Float64
    distance_factor::Float64

    best_objective::Float64
end

ta_frank_wolfe

This function implements methods to find traffic equilibrium flows: currently, Frank-Wolfe (FW) method, Conjugate FW (CFW) method, and Bi-conjugate FW (BFW) method.

References:

Example:

link_flow, link_travel_time, objective = ta_frank_wolfe(ta_data, log="off", tol=1e-2)

Available optional arguments:

  • method=:fw / :cfw / :bfw (default=:bfw)
  • step="exact" / "newton" : exact line search using golden section / a simple Newton-type step (default=:exact)
  • log=:on / :off : displays information from each iteration or not (default=:off)
  • max_iter_no=integer value : maximum number of iterations (default=2000)
  • tol=numeric value : tolerance for the Average Excess Cost (AEC) (default=1e-3)

For example, one may do:

ta_data = load_ta_network("SiouxFalls")
link_flow, link_travel_time, objective =
ta_frank_wolfe(ta_data, method=:cfw, max_iter_no=50000, step=:newton, log=:on, tol=1e-5)

The total system travel time can be simply computed as

using LinearAlgebra
system_travel_time = dot(link_travel_time, link_flow)

Contributor

This package is written and maintained by Changhyun Kwon.