RE: [boost] FW: Dijkstra with several starting point

From: Reid Sweatman (drunkardswalk_at_[hidden])
Date: 2003-07-28 16:33:06

> -----Original Message-----
> From: boost-bounces_at_[hidden]
> [mailto:boost-bounces_at_[hidden]]On Behalf Of JOUBERT Antoine
> Sent: Monday, July 28, 2003 4:14 AM
> To: 'boost_at_[hidden]'
> Subject: [boost] FW: Dijkstra with several starting point
>
> -----Original Message-----
> From: JOUBERT Antoine
> Sent: 28 July 2003 12:09
> To: 'boost_at_[hidden]'
> Subject: Dijkstra with several starting point
>
>
> I want a shortest path calculation from several starting point .
> I could of course reapply the dijkstra for every starting point,
> but i would rather initialize Disjkstra with a container of vertex .
>
> So instead of initialize d[firstPoint]=0
> I would for(all point in starting list ) d[point]=0
> I need some advice , I have tried adding in dijkstra and in bfs
> some thing like ( vector<vertex> ) but it is not working

You've piqued my curiosity now.  If you can do so, could you explain the
problem this is meant to solve?  I've implemented Dijkstra (or one the
variants, including the obvious one) a number of times, but have never come
across a need for what (I think) you're describing.  Ah, for the record, my
use was for path-finding in games, although I'm certainly aware that the
algorithm is a much more generic graph algorithm.  That's probably coloring
my understanding of what you're asking, and I'd like to understand the
request better.

Is this an attempt to re-use search data for multiple paths that may share a
portion of the search space, or is it something more esoteric, sort of an
algorithmic "variational" method, where you're looking for the best path,
but can also choose your start point?

Reid Sweatman