Multiple depot vehicle routing problem

This source code represents solution for multiple depot vehicle routing problem. Ant Colony Optimization algorithm is used. This solution was developed as an assignment for Heuristic Optimization Methods class at Faculty of Electrical Engineering and Computing, University of Zagreb.

This implementation uses Repa library that makes the code automatically concurrent. This opens up the possibility od utilizing as many cores of the processor as one likes, provided one supplies +RTS -N on the command line when running the program.

Authors: Teon Banek, Vlatko Klabučar, Damir Tomić

Ant Colony Optimization pseudocode for solving multiple depot vehicle routing problem.

Adjustable parameters are in stars, e.g. *alpha*.

repeat max-iterations:
    # Generate *m* feasible solutions.
    repeat *m*:
        solutions += generate-solution(*alpha*, *beta*, pheromones)
    # Generate new solutions from existing by mutating each with *mut-prob*.
    mutations = mutate-solutions(*mut-prob*, solutions)
    # Find best among all current solutions.
    current-best = find-best(mutations, solutions)
    # Update pheromones according to current best solution.
    update-pheromones(*decay-rate*, *deposit-rate*, pheromones, current-best)
    # Set all time best
    if current-best < best:
        best = current-best

fn generate-solution(*alpha*, *beta*, pheromones):
    until visited-all(visited)
      # Start new route if current truck has unsufficient cargo.
      next-city = probability-select(*alpha*, *beta*, pheromones)
      visited += next-city
    return best-2-opt-exchange(visited)