Originally posted by: naveendivakaran

Hi,

I am trying to run a 2nd solve on a model that is basically a modified version of the original model, in which I have to modify the coefficients of an array of variables in a constraint. I am trying to do that without having to run the generate function. The way I tried to do this is:

 

for(var i in opl.setWW)

        {

            opl.ctBacklog[i].UB = 1000;

            iconst = Math.ceil((opl.ObjWeeklyBacklog[i])*1/scale_factor);

            opl.ctScaled_Backlog[i].setCoef(opl.Scaled_ObjBacklog[i],-iconst);

            

        }

where ctBacklog and ctScaled_Backlog are constraints on array of variables. I get a "No Solutions Error" when it is trying to run this for loop. What am I doing wrong? The approach that I am taking to meet my goal for now is to create a fresh opl object after the first solve, and modify coefficients there which works fine. But I have to generate the model which consumes 3 minutes, that I would like to avoid. Please let me know if there is a way without having to run the generate method.

 

Method that works: (which requires regeneration 3 extra minutes)

 

             vardef2 = opl.modelDefinition;

             var data2 = opl.dataElements;

             var cplex2 = new IloCplex();

             var opl2 = new IloOplModel(def2,cplex2);        

             opl2.addDataSource(data2); 

                

        for(var i6 in opl.setWW)

        {

            opl2.ctBacklog[i6].UB = scale_factor+ub_tolerance;

            iconst = Math.ceil((opl.ObjWeeklyBacklog[i6])*1/scale_factor);

            opl2.ctScaled_Backlog[i6].setCoef(opl2.Scaled_ObjBacklog[i6],-iconst);

            

        }

 

Please help me with this task.

 

-Naveen

 

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

some examples at

https://www.ibm.com/developerworks/community/forums/html/topic?id=9ad7f60c-e801-410f-8faf-0a6dcacd00de&ps=25

regards

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: naveendivakaran

Hi,
I read the examples. I dont think it addresses what I am  looking for. Maybe a simple mod and dat file can explain what I am looking for. Let me know if it makes sense. It is just a dumb problem to explain what I am trying, as I cannot share my entire problem which is pretty huge.
I have attached a Trial.mod, and Trial.dat file here. So here I am trying to get the results of the first solve and update the coefficients of a constraint in the original model with the first solve results, after updating the Upper Bound on another constraint. It gives the error: 

Exception from IBM ILOG CPLEX: CPLEX Error  1217: No solution exists.    

I hope you can understand what I am trying to do in the main script. What is the best way to achieve it without having to create a fresh opl model and running the generate function?
Your help on this is greatly appreciated. Happy New Year!!
-Naveen

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

let me apply what I meant in your example. (See attached trial.mod)

What I did is store in temp the values before changing the coeffiients:

for(var i7 in thisOplModel.setWW)
  thisOplModel.temp[i7]=Math.ceil((thisOplModel.a[i7]));
 
 
  for(var i7 in thisOplModel.setWW)
    {
        thisOplModel.ctPain[i7].UB = 100;
        writeln("attempt: ",i7)
        iconst = thisOplModel.temp[i7]; //Math.ceil((thisOplModel.a[i7]));
        thisOplModel.ctScaledPain[i7].setCoef(thisOplModel.a[i7],-iconst);
        
    }

regards

 

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: naveendivakaran

Awesome!!

Thank you very much. Can you also shed some light to why I need a temp variable? Just so that I understand the need for this solution?

-Naveen

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

modifying a model will invalidate all previous model solution structures: including thedecision variable data, and any post-processing data. 

So as soon as you do

thisOplModel.ctPain[i7].UB

you cannot read

thisOplModel.a[i7]

any longer.

That is why you need a temporary structure like temp

regards

 

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: naveendivakaran

Fantabulous!! Thank you very much!!

#DecisionOptimization
#OPLusingCPLEXOptimizer