I'm having a small issue during solving: I want to set up a search strategy which will prioritize decisions on a selected group of variables (ie. setting this group at the top of the decision tree to make the search faster) but I can't seem to find a pre-built way to.
Does it exist?
I've tried using "
inputOrderLBSearch" but I get some strange behavior (it seems some constraints are not taken into account properly).
If the above search strategy is the correct one to use, do I need to specify a default strategy for all the other variables?
Example of pseudo-code if the groups were reduced to a single variable :
IntVar A; //primary variable
IntVar Ax;
model.element(A, list, Ax);
IntVar B; //less important variable
IntVar Bx;
model.element(B, list, Bx);
... Various constraints between Ax and Bx ...
Solver solver = model.getSolver();
solver.setSearch(inputOrderLBSearch(A));
Thanks for your help! Regards,
Pierre-Louis Sauvage