=========

Master thesis implementation - Traffic Assignment Problem : Shortest Path Optimization in Directed Graph with Multi-agent Population and Central (partial) Authority

##CONTENTS

Split into three parts:

• Sim - The actual C++ simulation (requires jansson library) ; also contains the JSON file used as input in the result generation/analysis part of the thesis.
• Graph Creation Tool - The processing.js applet to create/edit/visualize graphs used as input for the simulation (requires processing.js). Also contains the python(3.3) script used to "assemble" the separate .pde files composing the applet
• OSMImport - The python (3.3) script used to import, filter and transform a .osm file (exported from OpenStreetMap) to a JSON file compatible with the simulation and graph creation tool (requires OSM filter)

The master thesis can also be found here in pdf format.

##INSTALLATION:

A (rudimentary) makefile is present and should be used to create the binary called TAPSim It requires the installation of the JSON IO handling for C/C++ called jansson

##UNINSTALLATION:

Use make clean and delete TAPSim binary file

##USE:

path_to_binary/TAPSim <network_file> <other_option>

Where: <network_file> : the file containing information about the traffic network to simulate

<other_option> : list of other options ; can contain the following:

• --preset <preset_id> : load a preset (with all necessary options covered) ;
• any options set this way will be overridden by any other options used
• -simplified : Use simplified learner ;only keeping an average travel time for each road
• --medium <nb_intervals> : Use more advanced learner ; keeping average travel time over multiple time interval
• --mediumWs <weights> : Use alternative weights vector in medium learner ; specified by a list of floats whose sum should be 1
• --lastV <nb_entries> <weights> : Use last visited learner ; keeping only the nb_entries last passage through each road as data - needs to specify the weights ;
• --eBackprop <learning_factor> : Use error back-propagation learner with specified learning factor ;
• -eBackprop : Use error back-propagation learner with default learning factor ;
• --learnerWs <weights> : Set weights of learners - only if multiple learners , weights should be given in same order as learners were set up ;
• -simpleR : Use simplest route generation mode ; each agent has a single OD pair randomly generated and a departure time within the day
• -smarteR : Use smart(er) route generation mode ;
• -realR : Use realistic(ish) route generation mode ;
• -A* : Use a request handler using basic A* behavior to find shortest route.
• --RedA* <depth> : Use a request handler using reduced A* behavior to find shortest route with given depth.
• -RedA* : Use a request handler using reduced A* behavior to find shortest route with default depth.
• --Expl <expl_prob> : Adds exploration behavior to request handler with given probability
• -Expl : Adds exploration behavior to request handler with default probability
• --depth <pos_int> : Set depth to be used by request handler of type A* (without this set the default value is 1)
• -marginal : Set request handler to use marginal travel time cost ;
• --agents <nbFreeAgents> <nbInformedAgents> <nbDirectedAgents> : Set number of (each type of) agents to be used
• --seed <pos_int> : Set seed of the random number generator
• --days <pos_int> : Set number of (virtual) days the simulation will run for -v : Set verbosity to true ; more information will be displayed throughout execution
• --o"" <output_file_prefix> : Set prefix (including path) for output file (if not set, none is produced)