Originally posted by: AndyHam

(I don't think this forum is not for this topic, but I could not find the right forum. Any help will be appreciated. Thanks, Andy)

Dear IBM,
I have a flow control model for MIP-CP linking. The flowcontrol model works well within OPL environment. However, when I called the flowcontrol model via doOPL, it does not function. doOPL works well with stand-alone OPL models. The following is what I got when I call "main.py". Basically, it does not do anything. There are codes for main.py and flowcontrol.mod in this post. Would you please take a look at the codes and let me know what I have done wrong?

C:\Users\Google Drive\Codes\OPL>main.py

Tried aggregator 1 time.
No LP presolve or aggregator reductions.
Presolve time = 0.02 sec. (0.00 ticks)
oplmodel:
     mod file: C:\Users\AndyHam\Google Drive\Codes\OPL\FlowControl.mod
     0 .dat files
     0 python structures

 

//main.py

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: VincentBeraudier

This is a bug in doopl when using a .mod with a main.

We will issue a fix very soon.

I don't see any workaround.

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: AndyHam

Thanks for checking out this issue. If you can provide me a rough timeline of the fix, it will be very appreciated.

#DecisionOptimization
#OPLusingCPOptimizer

Hi,

let me offer you a workaround in the meantime.

From python you may call a .mod file without any main block that will call a .mod file with a main block through IloOplExec:

Let me give you some details out of the example from https://www.linkedin.com/pulse/making-decision-optimization-simple-alex-fleischer/

 

zoo.mod

tuple bus
    {
    key int nbSeats;
    float cost;
    }

    tuple kid
    {
        string name;
        int nbKids;
    }

    
    // This is a tuple set
    {bus} buses=...;
    {kid} kids=...;

    // asserts help make sure data is fine
    assert forall(b in buses) b.nbSeats>0;
    assert forall(b in buses) b.cost>0;

    // decision variable array
    dvar int+ nbBus[buses];
    dvar int nbKids;

    // objective
    minimize
     sum(b in buses) b.cost*nbBus[b];
     
    // constraints
    subject to
    {
     sum(b in buses) b.nbSeats*nbBus[b]>=nbKids;
    }

    tuple solution
    {
      int nbBus;
      int sizeBus;
    }

    {solution} solutions={<nbBus[b],b.nbSeats> | b in buses};

    main
    {
        var f=new IloOplOutputFile("solution.txt");
        thisOplModel.generate();
        for(var scenari in thisOplModel.kids)
        {
            thisOplModel.nbKids.LB=scenari.nbKids;
            thisOplModel.nbKids.UB=scenari.nbKids;
            cplex.solve();
            f.writeln("scenarii = ",scenari.name);
            thisOplModel.postProcess();
            f.writeln(thisOplModel.solutions);
            
        }
        f.close();

 

    }    

 

zoomain.mod

int nbKids=300;

    // a tuple is like a struct in C, a class in C++ or a record in Pascal
    tuple bus
    {
    key int nbSeats;
    float cost;
    }

    tuple kid
    {
        string name;
        int nbKids;
    }

    
    // This is a tuple set
    {bus} buses=...;
    {kid} kids=...;

    // asserts help make sure data is fine
    assert forall(b in buses) b.nbSeats>0;
    assert forall(b in buses) b.cost>0;

    execute
    {
        var f=new IloOplOutputFile("zoo.dat");
        f.write("buses=");
        f.write(buses);
        f.writeln(";");
        f.write("kids=");
        f.write(kids);
        f.writeln(";");
        f.close();
        
        IloOplExec("C:\\ILOG\\CPLEX_Studio128\\opl\\bin\\x64_win64\\oplrun.exe c:/zoo.mod zoo.dat",true);
    }
    

    

 

and then callmainbus.py

from doopl.factory import *

# Data

Buses=[
    (40,500),
    (30,400)
    ]

Kids=[("scenario1",300),("scenario2",310)]

# Create an OPL model from a .mod file
with create_opl_model(model="zoomain.mod") as opl:
    # tuple can be a list of tuples, a pandas dataframe...
    opl.set_input("buses", Buses)
    opl.set_input("kids",Kids)

    # Generate the problem and solve it.
    opl.run()

when you run that Python program you get solution.txt

scenarii = scenario1
 {<6 40> <2 30>}
scenarii = scenario2
 {<7 40> <1 30>}

regards

Alex Fleischer

PS:

Many how to with OPL at https://www.linkedin.com/pulse/how-opl-alex-fleischer/

Many examples from a very good book : https://www.linkedin.com/pulse/model-building-oplcplex-alex-fleischer/

Making optimization simple : https://www.linkedin.com/pulse/making-decision-optimization-simple-alex-fleischer/

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: AndyHam

Thanks for the feedback.
However, my client does not like the idea of using "IloOplExec".  They are waiting for doOPL fix. In fact, they plan to submit a formal request.

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: AndyHam

My client is waiting for the doOPL fix. If you can provide me a rough timeline of the fix, it will be very appreciated. Thanks!

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: VincentBeraudier

Can you try the new version that was uploaded few days ago on pypi and conda?

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: AndyHam

I just tried doOPL version 18.1 after upgrading, but it is still not working. When I run FlowControl.mod on OPL environment, it works great. But, Py does not.

***********************************************

/pip-18.1-py2.py3-none-any.whl (1.3MB)
    100% 1.3MB 5.1MB/s
Installing collected packages: pip
  Found existing installation: pip 18.0
    Uninstalling pip-18.0:
      Successfully uninstalled pip-18.0
Successfully installed pip-18.1

C:\Users\Google Drive\Codes\OPL\main.py
Tried aggregator 1 time.
No LP presolve or aggregator reductions.
Presolve time = 0.00 sec. (0.00 ticks)
oplmodel:
     mod file: C:\Users\Google Drive\Codes\OPL\FlowControl.mod
     0 .dat files
     0 python structure
***********************************************

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: Viu-Long Kong

Hi,

I think you confused pip and doopl. The log fragment you pasted shows that you upgraded pip (the python package manager), but not doopl.

Can you please try again after you update doopl with:

 

pip install --upgrade doopl

 

You can check the installed version with:

 

pip feeze

(then looking for the line with doopl== )

 

Thanks

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: AndyHam

It was my bad.  I did not upgrade doOPL as you found ^^.
After upgrading doOPL, The flowControl for linking MIP and CP works great!
Thanks a lot!

 

Successfully installed doopl-12.8.0.8

C:\Users\AndyHam\flowcontrol.py
//MIP
Total (root+branch&cut) =    0.16 sec. (0.40 ticks)
maxjobCountOn  =      38
minjobCountOn  =      12
travelDistance  =      679
//CP
 ! Search terminated by limit, 20 solutions found.
 ! Best objective         : 898 (gap is 7.35%)
 ! Total memory usage     : 9.2 MB (9.0 MB CP Optimizer + 0.3 MB Concert)
 ! Time spent in solve    : 5.02s (5.00s engine + 0.02s extraction)
 

#DecisionOptimization
#OPLusingCPOptimizer