[gecode-users] Breaking symmetries with LDSB in Matrix model.

[Chris Mears]

On Thu, 2 Oct 2014 14:54:19 +0100
Bouchene Sabrina <genius_linda1987 at yahoo.fr> wrote:

> Thank you Mr Mears, 
> If I understand correctly, to execute the combination of symmetries I
> must to add all the relevant declarations of symmetries to the same
> object Symmetries s. and the execution of the combination is done
> sequentially, So, we have the execution of the first decalaration
> symmetry is done and next the second definition of symmetry is
> executed ...etc, but they do not run in parrallel. 

Each symmetry is processed separately, but they are re-run repeatedly
until there is no more work to be done.  In this sense they operate
similarly to constraint propagators.

For your example, I cannot see how posting multiple branchers over the
same variables will do what you want.  I suggest that you put all the
symmetry definitions into one Symmetries object, pass that to a single
call to "branch", and see if that gives you the right behaviour.