An heuristic & genetic algorithm to approach the TSA problem with a nice result in a convenient time.
Basically, one gene is at the beginning a random set of (x,y) points peeked from the initial city points.
Each generation is thus a number of genes which will learn by themselves to fit the lowest total length condition.
The crossover step consists in choosing the best genes of the precedent generation and to cross them to get a new gene with
similar aspects, and so on. The mutation rate allows the algorithm to avoid the stagnation on a bad solution by adding random modifications on each gene.
The four main methods of the algorithms are:
population.naturalSelection();
population.newGeneration();
population.fitness();
population.evaluate();- the natural selection part is where the algorithm selects the genes with the best results considering the last generation.
- then, a new generation is built from these champion genes.
- fitness is just here to normalize the results the gene values.
- the evaluation method checks wether the final score is enough good to stop the algorithm. Otherwise, it selects the best genes for the next generation.