Originally posted by: SystemAdmin

[ncasti said:]

Hello everybody,
With the next problem...

0 <= x <= 5<br />0 <= y <= 5<br />x >= 3
x > y

[b]is possible to access and display the simplified constraints of the IloSolver?[/b]
I want to display something like this:
[color=brown]x in 3..5
y in 0..4
x < y[/color]<br />
I have seen that by default solver tries to propagate all the constraints and later do a labelling for all the variables.
Can I approach a solver problem from the "interval solution" point of view?

I know how to extract a model and propagate all the constraints.

I know hoy to "simulate" the "interval solution" behavior for the variables, using IlcIntVar ilc_x = IloSolver::getIntVar(x);
Now I can ask ilc_x.getMin(), getMax(), getSize() and even iterate by its values (as you show me how to do it a few weeks ago). 

This helped me in the past to "avoid" my real problem for the variables. But, now I have to deal with my real problem.
Many times I don't want to find a labelling solution for my problem (e.g. x = 3, y = 0 in this problem). What I really want is to show the internal state of the solver. The state in which the solver has simplified the constraints of the model.
In this example to say:
x in 3..5
y in 0..4
is not just a solution for the model.

Of course this is the result of propagate:
0 <= x <= 5<br />0 <= y <= 5<br />x >= 3
x > y

but the solution would be something like:
x in 3..5
y in 0..4
x < y<br />
which says all the possible values you can choose for your variables, but imposes a few constraints on your choice.

[b]Is it possible to do that?[/b]

Thank you very much in advance,
Nacho Castiñeiras

#CPOptimizer
#DecisionOptimization

Originally posted by: SystemAdmin

[shaw said:]

Have you looked at IloCP::propagate()?  I think it may be what you need.


Paul
#CPOptimizer
#DecisionOptimization