Demo available at https://lorenzogatto.github.io/GAPsolvers/index.html
Implemented metaheuristics are:
- Simulated annealing
- Tabu search
- Iterated local search
- Variable neighborhood search
The latter two algorithms are the ones giving the best performance, about 3% above the optimum on large instances.
Costs of optimal solutions can be found at http://astarte.csr.unibo.it/gapdata/gapinstances.html for most instances.
The original instace files can be found at http://people.brunel.ac.uk/~mastjjb/jeb/orlib/gapinfo.html and http://www.al.cm.is.nagoya-u.ac.jp/~yagiura/gap/.