4.2.1 How does the algorithm work

1. During initialization, the agents are distributed, every town gets one agent. The start city is entered into the respective History list and the Iterator initialized.
   

2. The agents now move from city to city, where the next city is chosen with a specific probability, depending on the length and marker thickness. For every decision, the attended city is entered into the history list. This is repeated until all agents have ended their round trips.
   

3. Before the evaluation, the agents are reset onto their start city. The overall length of every agent tour based on the history lists (the lenght of the round trip) will then be computed. If this is shorter than the up to now shortest one, this one will be observed. For every edge, the length dependent evaluations of all agents are summed up.
   

4. The old markers evaporate (normally through a factor less than 1), then the new evaluations will be added. The iteration counter will be incremented.
   

5. If the iteration counter has the maximum iteration number the Algorithm stops and the up to now shortest tour will be the result. Otherwise all history lists are cleared and the algorithm is executed once more from the 2nd point.