Hello,
I used Simulated Annealing (SA), I know TSP, Python is one of my favourite language.
But I never apply SA to TSP. I guess the algorithm with following:
- At high temperature: probability to find complete new path (the set of indices is chosen by random) is high, if not complete new path: permute the current set (always storage the best set)
- Temperature is always reduced.
- At low temperature: probability to find complete new path is low, if not complete new path: permute the current set (always storage the best set)
TSP is NP-hard problem, no chance to find the global optimal solution. With SA algorithms, it searchs the solution in global area first and later it searchs the solution in local area.
If my understanding is correct or close with your algorithm, I could help you in this project because I have practice with Python and algorithms a lot. (I am PhD student in Signal Processing)