Is there a docplex function for getting solution of more than one variables for using in lazy cut?

Question

I code an optimization problem in python (docplex) base on here but my cut is something like that sum(X) <= N + M*sum(Y, Z), X, Y and Z are decision vars. I do not know how I can pass the solution of X, Y, Z to self.get_cpx_unsatisfied_cts([ct], solution, tolerance=1e-6). when I use self.get_values(), I get this Error: Error: Internal error in CPLEX solve: AttributeError: 'list' object has no attribute '_get_var_value' Is there anyone can help ?

Answer 1

In the example [here] (Implementing TSP with Lazy constraint callback), callbacks are aconnected to DOcplex throught the mixin class MosdelCallbackMixin. This class has a make_complete_solution method which returns a solution object (see documentation for docplex.mp.SolveSolution) from the callback internals.

In a second step get_cpx_unsatisfied_cts returns a list of unsatisfied constraints from a given list of constraints. In your case, if you need to check only one constraint, it is simpler to use the is_satisfied methods for constraints, which takes a solution and a tolerance (default is 1e-6, that is:

   if not myct.is_satisfied(sol, tolerance):
       print(f" constraint {myct} is violated")

To summarize:

• use make_complete_solution to generate a solution object when callback is called
• use ct.is_satisfied to check whether is is satisfied or not.