[gecode-users] staged search

Denys Duchier denys.duchier at univ-orleans.fr
Wed Oct 8 23:03:38 CEST 2008


you may still be wondering why I would want to do something like this.
I am really, really, really embarassed to admit this, but what prompted
this line of thought was the "historical" example about how constraint
and search evolved:

	- from generate and test
        - to test and generate
        - then on to propagate and distribute

see e.g. the introduction to that rather old lecture series I gave at
ESSLLI in 2000:

http://www.univ-orleans.fr/lifo/Members/duchier/teaching/ESSLLI-2000/node1.html#chapter.introduction

which shows the search trees corresponding to each of the techniques.

Especially, I was wondering how to demonstrate in Gist the "generate and
test" case.  This was very easy to model in Mozart/Oz: first distribute,
then post the constraints.

This is what started me thinking about "staged" search (which is a
generalization of what is required for that example).

So, it's a bit embarassing because my original motivation was so
ridiculously trivial, but I think that the generalization raises
interesting questions... you may of course disagree, and I wouldn't
blame you :-)

Cheers,

--Denys




More information about the gecode-users mailing list