Originally posted by: C.Magno

Hi everyone!

I'm coding a heuristic with exact local search.

In my main block script code I'm trying to access the value of a decision variable.

My model is in a separated file "LP-CA.mod".

This is the important segments of the code in this file:

...

dvar boolean ypc[P][C]; //decision variable that i'm trying access

...

minimize sum (d in Demands) xd[d]*d.bandwidth;
 
 subject to{
  
  //ypc is used in the constraints
  
  forall(d in Demands)
    sum(p in pat[d])
         sum (c in channels[d])
            ypc[p][c] + xd[d] == 1;     
                  
  forall(e in Links)
   forall(s in Slots)
    sum (d in Demands)
      sum(p in pat[d]: deltape[p][e] == 1 ) 
        sum (c in channels[d]: gamacs[c][s] == 1) 
           deltape[p][e]*ypc[p][c]*gamacs[c][s] <= 1;
        
  forall(d in Demands: selected[d] == 1)
        xd[d] == 1;  
 }

As we can see. The ypc[ ][ ] is used in the constraints. In my main script block I'm checking the cplex status before access the ypc[ ][ ] variable. Like this:

main{
    
    ... //load model and data files...
    
    function runHeuristic(opl){ //opl is an instance of IloOplModel
        
        ...
        
        if(cplex.solve()){
                        
                        if(cplex.status == 1) //Optimal solution is available.
                            writeln(opl.ypc); //Not returning a string like [[0,0,0,0,][0,0,1,0][1,0,0,0]]
                        
                        objAux = cplex.getObjValue();
                        writeln("Obj. Value: " + objAux);
                        if(pertubationIdx >= 0){
                        localSearchCounter++;
                }                               
        }
        
        ...
    }
    
    ...
}

But i'm getting as result just a string whit the name of the variable: "ypc".

Someone can help me with this issue?

Thanks in advanced.

Update 1:

When I try access element by element like this:

for(var p in opl2.P)
   for(var c in opl2.C)
      writeln(opl2.ypc[p][c]);

I get this string "[a IloNumVar]".

Update 2:

Since I'm using the ILOG IDE version 12.5, I must access the decision variable value like this:

opl2.ypc[p][c].solutionValue

But now, I'm getting a new problem.
I'm getting a error message saying that "has no available solution" and it's "impossible to process".

But how is it possible if the cplex status says "1" that means "Optimal solution is available"?

Hi,

you may notice in the example at

https://www.ibm.com/developerworks/community/forums/html/topic?id=a63c5e3e-b205-4707-b63f-86ee3b78a544&ps=25

that I wrote

sol[w]=thisOplModel.Open[w].solutionValue

in order to get all the values and then later on I use those values to update the model lower and upper bounds.

You should do the same since as soon as you change a bound the previous solution gets invalid and that is why you get an error

regards

Originally posted by: C.Magno

Hi Alex, thanks for the support.

I've noticed that the error is caused because I make changes in the external data and generate again the opl model before solve.

So, this is a more detailed code:

main{
    
    var cplex = new IloCplex();
    var modSource =  new IloOplModelSource(".../file.mod");
    var def = new IloOplModelDefinition(modSource);
    var data = new IloOplDataSource(".../file.dat");
    var opl = new   IloOplModel(def, cplex);

    opl.addDataSource(data);
        
    runHeuristic(opl, cplex);
    
    function runHeuristic(opl, cplex){
          
       opl.generate();
                
       cplex.tilim = 1; //I'm limiting the time because the heuristic
                
       var obj = Number.MAX_VALUE;
       var objAux = Number.MAX_VALUE;
                
       /*In this part I put this code as a test
       **and it worked well. No problems.*/
                
       write("Solving... ");
       if(cplex.solve()){
          objAux = cplex.getObjValue();
          writeln("Obj. Value: " + objAux);
                  
          if(cplex.status == 1)
             for(var p in  opl.P)
                for(var c in opl.C)
                   writeln(opl.ypc[p][c].solutionValue);                               
          }
        
        /*****************************************/
                
        while(Some Condition){
                    
           /*This function changes external data elements of the model
           **and generate it again.*/
           selectPercentOfDemandsRandomly(cplex, opl);
                        
           write("Solving... ");
           if(cplex.solve()){
               objAux = cplex.getObjValue();
               writeln("Obj. Value: " + objAux);
               if(cplex.status == 1)
                  for(var p in  opl.P)
                      for(var c in opl.C)
                          writeln(opl.ypc[p][c].solutionValue); //This line is getting the error.
           }
           ...
     }
}

The external data I change  is a vector called selected[ ] :

 minimize sum (d in Demands) xd[d]*d.bandwidth;
 
 subject to{
 
  Only_one_path_and_channel:
  forall(d in Demands)
    sum(p in pat[d])
         sum (c in channels[d])
             ypc[p][c] + xd[d] == 1;  
  
  Only_one_demand_per_channel:    
  forall(e in Links)
   forall(s in Slots)
    sum (d in Demands)
      sum(p in pat[d]: deltape[p][e] == 1 ) 
        sum (c in channels[d]: gamacs[c][s] == 1) 
          deltape[p][e] * ypc[p][c] * gamacs[c][s] <= 1;
        
  forall(d in Demands: selected[d]==1) //I change this vector before solve.
        xd[d] == 1; //that changes this dvar boolean array too, indirectly.
 }

This is the function where i make the changes and generate the model:

function selectPercentOfDemandsRandomly(cplex, opl){
    var dat = opl.dataElements;
    
    ...
    /*So, some times, for a given [d] I do:*/
    dat.selected[d] = 1;
    /*or*/
    dat.selected[d] = 0;
    /*And finally*/
    
    cplex.clearModel();
    opl = new IloOplModel(mod, cplex);
    opl.addDataSource(dat);
    opl.generate();
}

Is there something that i can do to be able to access the decision variable value without error after make the changes on the data?

Regards

Hi,

could you have a look at the example I provided at https://www.ibm.com/developerworks/community/forums/html/topic?id=71ac9d4e-fcc6-4ef9-a7c5-8f2a257e5cdd&ps=25 ?

regards

Originally posted by: C.Magno

Hi Alex,

Thank you very much!

I studied the example code you mentioned.

I made changes in my code and it worked for me!

Regards.