It's still a non-linear growth, I believe. It's less likely the ram,
then the simple cost of running through all of the combination, even
with pruning.
I ported this same algorithm to Python (no, it's too ugly to share),
and I think I hit around the same limits.
You might have to do something similar and do offline processing,
where maybe you use your own matrix of data and give the obvious
distances a really large number so they get ignored sooner. (say,
opposite ends of the state or something). Then also consider tweaking
the variables in the code.
mrc
> --
> You received this message because you are subscribed to the Google Groups
> "Google Maps TSP Solver" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to
google-maps-tsp-s...@googlegroups.com.
> For more options, visit
https://groups.google.com/d/optout.