Decision Optimization

Decision Optimization

Delivers prescriptive analytics capabilities and decision intelligence to improve decision-making.

 View Only
Back to discussions
Expand all | Collapse all

How find bounds on objective after early termination

  • 1.  How find bounds on objective after early termination

    Posted Mon June 28, 2021 11:43 AM
    I have a cplex MIP model which I'd like to run with a node limit.  After it terminates, I'd like to find the best lower bound (if I'm minimizing, or upper bound if I'm maximizing) for the objective.  I'm using the Python interface.   Is there a method that I can call after termination to retrieve this?  Right now I'm using a MIPInfoCallback to record the best_bound on the unevaluated nodes, and take the maximum over all of these.  Does that sound correct?  Is it guaranteed that the last such bound that I find before termination will be the best?  Is that the only way?

    Victor

    ------------------------------
    Victor Miller
    ------------------------------

    #DecisionOptimization


  • 2.  RE: How find bounds on objective after early termination

    Posted Mon June 28, 2021 01:59 PM
    Hi

    let me share a tiny example:

    from docplex.mp.model import Model
from docplex.mp.progress import *

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.minimize(nbbus40*500 + nbbus30*400)

mdl.parameters.mip.limits.solutions=1

while (1==1):
    sol=mdl.solve(log_output=False)
    for v in mdl.iter_integer_vars():
       print(v," = ",v.solution_value)

    print("objective = ",sol.get_objective_value())
    print("best bound = ",mdl.solve_details.best_bound)
    print("mip gap = ",mdl.solve_details.mip_relative_gap)  
       
    print("status : ",mdl.solve_details.status)   
    if ("optimal solution" in str(mdl.solve_details.status)):
        break​


    which gives

    nbBus40  =  8.0
nbBus30  =  0
objective =  4000.0
best bound =  0.0
mip gap =  0.999999999999975
status :  solution limit exceeded
nbBus40  =  7.0
nbBus30  =  1.0
objective =  3900.0
best bound =  3750.0
mip gap =  0.03846153846153748
status :  solution limit exceeded
nbBus40  =  6.0
nbBus30  =  2.0
objective =  3800.0
best bound =  3800.0
mip gap =  0.0
status :  integer optimal solution


    ------------------------------
    [Alex] [Fleischer]
    [EMEA CPLEX Optimization Technical Sales]
    [IBM]
    ------------------------------

    Original Message


Related Content