[gecode-users] Skipping branching alternatives?
Luis Quesada
luque at info.ucl.ac.be
Mon Jul 10 15:37:59 CEST 2006
Guido Tack wrote:
> [...]
>
>This is slightly against the general architecture of Gecode. One of the main
>achievements in CP research was the clear separation of concerns between
>propagation, branching, and search. Propagation does the inferences within
>one space, branching determines the shape of the search tree, and search
>explores the tree.
>Cutting off branches is normally considered part of exploration. [...]
>
I don't agree with this bit Guido. Cutting off branching (i.e. pruning)
is the very purpose of propagation. When you remove 5 from X's domain
you are cutting off the branch associated with X=5.
Luis
More information about the gecode-users
mailing list