I would like to obtain more than one solution with SCIP. Using SAT (cp_model.CpModel()) I can collect all the possible solutions, I would like to be able get the top N best solutions with SCIP, is it possible?

(I say SCIP because I’ve read that this can be done only using Gurobi or SCIP through OR-Tools).

Instead of cleaning out my actual problem, I’ll use this OR-Tools MIP example, because it has a nice graphic that shows the possible solutions; for the sake of this example, I would like to get them all:

from ortools.linear_solver import pywraplp

solver = pywraplp.Solver.CreateSolver("SCIP")

infinity = solver.infinity()
x = solver.IntVar(0.0, infinity, "x")
y = solver.IntVar(0.0, infinity, "y")

solver.Add(x + 7 * y <= 17.5)
solver.Add(x <= 3.5)

solver.Maximize(x + 10 * y)

print(f"Solving with {solver.SolverVersion()}")
status = solver.Solve()

if status == pywraplp.Solver.OPTIMAL:

    for _ in range(5):

        print(f"Objective value = {solver.Objective().Value():0.3f}")
        print("x =", x.solution_value())
        print("y =", y.solution_value())

        if solver.NextSolution():
            print(f"There is another solution: ")
        else:
            print("No more solutions")
            break

else:
    print("The problem does not have an optimal solution.")

I get this:

Solving with SCIP 9.0.0 [LP solver: Glop 9.10]
Objective value = 23.000
x = 3.0
y = 2.0
There is another solution:
Objective value = 0.000
x = 0.0
y = 0.0
No more solutions

I tried configuring SCIP like this, but that didn’t help:

solver = pywraplp.Solver.CreateSolver("SCIP")

solver.SetSolverSpecificParametersAsString("limits/solutions=-1")
solver.SetSolverSpecificParametersAsString("limits/bestsol=-1")
solver.SetSolverSpecificParametersAsString("constraints/countsols/sollimit=-1")
solver.SetSolverSpecificParametersAsString("constraints/countsols/collect=TRUE")
solver.SetSolverSpecificParametersAsString("constraints/countsols/discardsols=FALSE")
...

Is there something I’m doing wrong or misunderstanding? Is this possible?

Thanks