James D. McCaffrey published a fascinating article on "Natural Algorithms"
Use Bee Colony Algorithms to Solve Impossible Problems in the April 2011 edition of MSDN Magazine.
I've ported his VB.Net code to Jabaco:
SimulatedBeeColony.zip
During my experiments, I've made the following observations:
- The Jabaco application finds the optimal solution most of the time while the published code doesn't
(I have no clue, why this happens. This is left as an exercise to the reader ...)
- The Jabaco app takes some 30s to execute from within the IDE but only 2s being executed as Exe file.
- The VB.Net/C# code takes 30ms
- Char arrays in Jabaco don't work. I replaced them by Integer arrays
(this seems to be a bug or an omission in the Framework)
It would be nice to integrate this code into my
Traveling Salesman Sample.
Any volunteers?
Greetings