Genetic Algorithm Solving TSP (100 cities)

Genetic Algorithm developed by myself. Solving the Traveling Salesman Problem with 100 cities (plotted on a circle).

Parameters:

Order Crossover: 60 %
Mutation: 10 %
Selection: Elitism
Population: 100
Generations: 10000

~ Sirar Salih

Category:

Science & Technology

Tags:

• Genetic
• Algorithm
• Traveling
• Salesman
• Problem
• circle
• cities
• algorithms
• computer

[YCDPKLD]

omg randomizing it would've been faster.

[Sirar]

@YCDPKLD Faster? Of course. More precise? Of course not. Randomizing something leaves much to coincidence.

[yonnelol]

nice presentation! your mutation rate seems to be rather high though. its not high enough to protect the most fit ones. as you can see there the fitness value detoriorates every now and then.

[Sirar]

@yonnelol Thanks, I agree! :)

[txikipantxo]

genetic-cumagic in youtube.

[Sirar]

@TwoKingdoms That is correct. Instead of restricting the visual representation to showing shorter distances to complete the circle, my algorithm is instead giving you a visual representation of continuous calculations to reach the best solution (a complete circle). Basically, I thought it looked more nifty to show you how the algorithm actually worked behind the scene. Later, I also added text on the screen showing the number of generations and the best so far solution.

[TwoKingdoms]

@taostoner1 Only if the virtual positioning of the cities represents their physical positioning and the distances between them, which isn't the case in this video (given the algorithm is correct). The algorithm uses the edge weights (like distances) to solve this graph problem which aren't shown in this visual representation.

@artefom Useful comment...

[artefom]

@taostoner1 HAHAHAHHAH

[taostoner1]

Shouldn't the solution just be to follow along the edge of the circle?