Here are some TSP references: articles and videos.

This video illustrates (without explanation) four approaches on 200 cities.

Approach                                         Final path length

Random network                               327,452.4

Greedy search                                    36,226   (We saw this in class.)

2-Opt                                                   31,887   (We discussed this in class. 2-opt is the "untwist" heuristic. But done here with nothing to guide it.)

Simulating Annealing with 2-Opt.       30,944.3 (Simulating Annealing has been described as a version of genetic algorithms with a population size of 1. I.e., it has no crossover. The only evolution operation is mutation. One is willing to do more extreme mutations at the start ("higher temperature"). As the process continues one does mutations that are less and less extreme. )