The latest release is multi threaded to solve combinatoric optimization problems (like the Traveling Salesman Problem, or the Knapsack Problem). Simply feed the constructor a dict mapping your node names to coordinates of those nodes and give it a distance function call back that can take the coordinates and it will solve it using the ACO algorithm as described.
An effort has been made to initialize the algorithm's variables (alpha, beta, Q, etc) to sensible defaults. But your TSP problem may differ significantly from the testing set (see test directory) and may perform more poorly than optimal.
The latest release is multithreaded.
A full description of the behavior of the algorithm can be found here: https://en.wikipedia.org/wiki/Ant_colony_optimization_algorithms#Example_pseudo-code_and_formula
Note: Tests that implement finding actual optimal paths may fail occasionally, as this algorithm is semi-random. The default configuration should make this a rare occurrence.