[pgrouting-dev] Do we need a new and better TSP?

[Mario Basa]

Daniel,

I am trying to understand this: the mutation computation will be done
in SQL? Wouldn't that be expensive?

For me, implementing the GA specifically for pgRouting will be the
best way to remove GAUL dependency. No idea how complex that will be
though.

Mario.

On Sun, Sep 18, 2011 at 2:50 AM, Daniel Kastl <daniel at georepublic.de> wrote:
> Hi list,
> On Friday at FOSS4G in Denver I discussed with Steve about TSP, and I want
> to share a bit of what we came up with:
>
> According to Steve there is a way to solve TSP without GA (Genetic
> Algorithm), and that way we could get rid off GAUL dependency and make it a
> "core" function.
> Instead of calculation of a distance matrix within the TSP function, the
> distance matrix could be passed as a "SELECT". Then the user could decide
> the way distances should be calculated. And APSP would be a nice way to
> calculate a distance matrix for example.
>
> Any other ideas how to improve the current implementation of TSP?
> Do you have some other use cases with specific requirements?
> Daniel
>
>
> --
> Georepublic UG & Georepublic Japan
> eMail: daniel.kastl at georepublic.de
> Web: http://georepublic.de
>
> _______________________________________________
> pgrouting-dev mailing list
> pgrouting-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/pgrouting-dev
>
>