I'm looking for an instance of Euclidean TSP problem (shortest path among a number of points) on a complete graph with a known perfect solution. Has anybody encountered such examples? Or is there a simple algorithm to generate such instance that there will certainly be no shorter route than generated?
1 Answers
I'm pretty sure there are libraries of problems for this. Looking at http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/TSPFAQ.html I see
Q: Are the given solution values only the best ones known?.
A: No, for every problem either the value of a provably optimal solution or an interval given by the best known lower and upper bound is listed. The optimality of solutions has been proven by branch-and-cut or branch-and-bound algorithms.
Also see at http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/STSP.html
When I published TSPLIB more than 10 years ago, I expected that at least solving the large problem instances to proven optimality would pose a challange for the years to come.
However, due to enormous algorithmic progress all problems are now solved to optimality!!

- 19,301
- 2
- 19
- 25