Previously, I was using OPL Script, but I decided to switch to Python in order to incorporate machine learning. However, since I still want to use the existing OPL model, I am calling the oplrun command from Python using the subprocessmodule. The OPL model is an MIQP.

In the previous OPL Script, there was a part where the convertAllIntVars method was used to solve a continuous relaxation of the problem. I am having trouble figuring out how to solve the continuous relaxation using the oplruncommand. Do I have no choice but to manually convert the model into its relaxed form?

------------------------------
Toshiyuki Miyamoto
------------------------------

Hi

in docplex you can relax integrity constraints as can be seen at https://github.com/AlexFleischerParis/zoodocplex/blob/master/zoorelaxintegrity.py

from docplex.mp.model import Model
from docplex.mp.relax_linear import LinearRelaxer

def make_bus_model():
    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)
    return mdl

if __name__ == '__main__':
    bm1 = make_bus_model()
    bm1.print_information()
    s1 = bm1.solve(log_output=True)
    s1.display()

    bmr = LinearRelaxer.make_relaxed_model(bm1)
    bmr.print_information()
    rs = bmr.solve(log_output=True)
    rs.display()

    duals = bmr.get_constraint_by_name("kids").dual_value

    print("dual of the 300 kids constraint = ",duals)

    for v in bmr.iter_continuous_vars():
       print(v," = ",v.solution_value)
       rc=v.reduced_cost
       print("reduced cost =",rc)

You can also do that directly in OPL as can be seen at https://github.com/AlexFleischerParis/zooopl/blob/master/zoorelaxinteger.mod

    int nbKids=300;
    float costBus40=500;
    float costBus30=400;
     
    dvar int+ nbBus40;
    dvar int+ nbBus30;
     
    minimize
     costBus40*nbBus40  +nbBus30*costBus30;
     
    subject to
    {
     ctKids:40*nbBus40+nbBus30*30>=nbKids;

    }

    main {
      var status = 0;
      thisOplModel.generate();
      if (cplex.solve()) {
        writeln("Integer Model");   
        writeln("OBJECTIVE: ",cplex.getObjValue());   
        
      }

      // relax integrity constraint
      thisOplModel.convertAllIntVars();
     
      if (cplex.solve()) {
        writeln("Relaxed Model");   
        writeln("OBJECTIVE: ",cplex.getObjValue());  
        
        writeln("dual of the kids constraint = ",thisOplModel.ctKids.dual);
        writeln("reduced costs for nbbus40 : ",thisOplModel.nbBus40.reducedCost);
        writeln("reduced costs for nbbus30 : ",thisOplModel.nbBus30.reducedCost);
      }
       
     
    }

regards

------------------------------
[Alex] [Fleischer]
[Data and AI Technical Sales]
[IBM]
------------------------------

Original Message:
Sent: Thu July 24, 2025 03:09 AM
From: Toshiyuki Miyamoto
Subject: continuous relaxation on oplrun

Previously, I was using OPL Script, but I decided to switch to Python in order to incorporate machine learning. However, since I still want to use the existing OPL model, I am calling the oplrun command from Python using the subprocessmodule. The OPL model is an MIQP.

In the previous OPL Script, there was a part where the convertAllIntVars method was used to solve a continuous relaxation of the problem. I am having trouble figuring out how to solve the continuous relaxation using the oplruncommand. Do I have no choice but to manually convert the model into its relaxed form?

------------------------------
Toshiyuki Miyamoto
------------------------------