When you add a segment E you have ONE of the routes that lead to that segment. But if c(E) fluctuates wildly according to which route you used to get there, then it may be worthwhile to take a longer route to get to the start of E, just so that you get a lower c(E).
On Sat, Sep 20, 2008 at 10:57 AM, OJ W <[EMAIL PROTECTED]> wrote: > I must be implementing A* wrong then, because every time I add a new > segment to a potential route in pyroute, I have the history showing > how it got to that segment. > > > > On Sat, Sep 20, 2008 at 4:34 AM, Stefan Pflumm <[EMAIL PROTECTED]> > wrote: > > Hello, > > > > I'm searching for an routing algorithm who can handle this two > properties: > > > > 1. the edge weight function c(E) of the edge E depends on the costs of > > the predecessor edges. A* for example, can't handle this, because it > > connects existings paths without updating the successors. If the path > > previous to an edge changes and the edge is already expanded the costs > > of the edge and all successors will also change. > > > > 2. the edge weight function c(E) of the node N depends on the successor > > edge. For example: E_1 and E_2 are neighbour edges of E. c(E) to E_1 > > must be not equal to c(E) to E_2 > > > > I hope somebody can give me an advice, thanks > > > > Stefan > > > > > > > > > > > > _______________________________________________ > > Routing mailing list > > [email protected] > > http://lists.openstreetmap.org/listinfo/routing > > > > _______________________________________________ > Routing mailing list > [email protected] > http://lists.openstreetmap.org/listinfo/routing >
_______________________________________________ Routing mailing list [email protected] http://lists.openstreetmap.org/listinfo/routing
