Further data for the paper

Krzysztof Socha, Joshua Knowles, Michael Sampels:
A MAX-MIN Ant System For the University Timetabling Problem
3rd International Workshop on Ant Algorithms (ANTS 2002)

Problem instances
Algorithms
Diagrams

Problem instances

We generated a set of instances for testing both algorithms. We used Ben Paechter's problem instance generator, which is briefly described in the paper.

Small Medium Large
small1.tim medium1.tim large.tim
small2.tim medium2.tim
small3.tim medium3.tim
small4.tim medium4.tim
small5.tim medium5.tim


Algorithms

The algorithms were implemented in C++. We used the GNU compiler gcc version 2.95.3. The following files contain the source code of the implemented algorithms:

MAX-MIN Ant System (MMAS)
Random Restart Local Search (RRLS)

They contain a makefile. The executable can be generated by simply typing make. Both algorithms use the same following command line arguments:
<algorithm> -i <instance> -t <time limit in seconds per try> -n <number of trials> -s <seed for random number generator> [-o <output file>]

Additionally to the parameters specified above, the mmas uses the following command line parameters:

-ant <number of ants>
-evp <evaporation coefficiant (rho)>
-min <minimal pheromone level (tau_min)>


Diagrams of the comparisons

Boxplot comparing MMAS and RRLS for the small instances (eps-format)
Boxplot comparing MMAS and RRLS for the medium instances (eps-format)
Boxplot comparing MMAS and RRLS for the large instances (eps-format)