Infeasible Solution Handling with Python cvxpy and Python_MIP() (CBC solver)

Question

In an MIP tool which utilizes cvxpy with Python MIP (CBC) as the solver, I am looking for a method to produce specifics of an infeasible solution for logging purposes. Does anyone know if this exists? I was reading that CPLEX has this capability (function) to point out specific constraints violations.

Answer 1

in python docplex you can get relaxation and conflict.

See example https://github.com/AlexFleischerParis/zoodocplex/blob/master/zoorelaxationandconflict.py

So in your case you could either:

• Turn your model in python docplex
• Keep cvxpy, export mps file and then import mps file (with model.import_model) in docplex and then run relaxation in docplex

Here the model:

from docplex.mp.model import Model
from docplex.mp.relaxer import Relaxer

from docplex.mp.conflict_refiner import ConflictRefiner

mdl = Model(name='buses')
nbbus40 = mdl.integer_var(name='nbBus40')
nbbus30 = mdl.integer_var(name='nbBus30')

mdl.add_constraint(nbbus40*40 + nbbus30*30 >= 300, 'kids')
mdl.add_constraint(nbbus40 + nbbus30 <= 7, 'maxTotalBuses')

mdl.minimize(nbbus40*500 + nbbus30*400)

mdl.solve()

mdl.report()
print(f"* solve status is: '{mdl.solve_details.status}'") #infeasible model

print()
print("------- starting relaxation")
print()

rx = Relaxer()
rx.relax(mdl)

print ("number_of_relaxations= " + str(rx.number_of_relaxations))
rx.print_information()

mdl.report()
print(f"* status after relaxation is: {mdl.solve_details.status}")
#print(mdl.solution)

print()
print("------ starting conflict refiner")
print()

cr=ConflictRefiner()
conflicts=cr.refine_conflict(mdl)
conflicts.display()