RailNL

Group: Team Gekke Hackers Members: Sietze Berends, Floris Holstege & Daan Uittenhout

Case: RailNL - finding an optimal Lijnvoering for a fictional railway company. For more information: http://heuristieken.nl/wiki/index.php?title=RailNL

Getting started

Prerequisites

1. Powershell/Cmd
2. Python 3.6.3
3. run pip install -r requirements.txt for external packages

Running the program

The application is run by executing application.py with command line arguments:

  application.py arg1 arg2 arg3 arg4 arg5

The following arguments should be given:

arg1: Choose which map to use:

  National
  Holland

arg2: Amount of Hillclimbers that are created:

  Positive integer 

arg3: Amount of iterations that every Hillclimber does -> change the value of iterationsInHillclimber:

  Positive integer

arg4: Turn simulated annealing on or off for every Hillclimber and choose a cooling strategy:

  a: off
  b: linear
  c: exponential
  d: Geman & Geman
  e: hardcoded on a very low acceptation chance

arg5: print additional details e.g. new highscores that are reached while the algorithm is running:

  True
  False

Application.py will run the hillclimber first with the configuration chosen by the user. After that, a Greedy for 4 trajectories will be runned.

Structure

0. The application is run by executing application.py from the main folder
1. All classes are kept in a seperate 'Classes' folder
2. Algorithms are kept in an 'Algorithm' folder
3. Files necessary for experiments are kept in an 'Experiments'

Experiments:

The following experiments have been done:

0 - Manipulating the score function 1 - Different simulated annealing cooling strategies and temperatures 2 - Hillclimber with or without simulated annealing 3 - Disable Utrecht 4 - All stations critical

All experiments are placed in a folder called Experiments and have their own subfolder with a Readme. Please refer to these Experiments to get the description, results and plots of each experiments.

Visualisation:

Our visualisation depicts the final solutions that we have found. The following link will bring you to our index page, where you then will be taken to our visualisation.

https://sietzeberends.github.io/Gekke-Hackers-RailNL/Visualisation/mapSolution.html