Originally posted by: Ben87

Hi,

i have the following problem: I use the OplScript (Main) to solve two different models after each other. In the first model i have a boolean decision variable, which contains a two dimensional Array. This Array i want to use in my second model as a data-input with the data type "int". To pass the variables I use the "IloDataElements" function.

While doing that, i get the following error "Internal error: dvar unexpected at ..\..\..\src\script\proxies.cpp:999".

So my question is, if it is possible in general to use a boolean decision variable as an data-input of type int?

here is some code to clarify the problem:

First, this is my decision variable of the first model: "dvar boolean AllokationErsteWerkeProdukte;"

this is the declaration of the Array in the second model, which I want to use as a data-input: "int AllokationWerkeProdukte=...;"

the second model in the main script is introduced in following way:
var ZweiteStufeSource = new IloOplModelSource("ZweiteStufe.mod");
var ZweiteStufeDef = new IloOplModelDefinition(ZweiteStufeSource);
var ZweiteStufeCplex = new IloCplex();
var ZweiteStufeData = new IloOplDataElements();

ZweiteStufeData.Allokation = ErsteStufeOpl.AllokationErste; //getting the boolean Array from the first model
.......

var ZweiteStufeOpl = new IloOplModel(ZweiteStufeDef, ZweiteStufeCplex);
ZweiteStufeOpl.addDataSource(ZweiteStufeData);
ZweiteStufeOpl.generate();
ZweiteStufeCplex.solve();

the above-described error occurs when i want to read the Array, for example in the following way:
var test = ZweiteStufeData.Allokation;
writeln("Array = ", test);

Also my second model yields an incorrect objective value.

Regards,
Ben
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

> ZweiteStufeData.Allokation = ErsteStufeOpl.AllokationErste; //getting
> the boolean Array from the first model

You should not directly assign the AllokationErste variable (which is a dvar) to the
non-dvar variable Allokation.

Instead, you should declare a variable to store the optimal values of the AllokationErste
variable from the 1st model run, update it in a post-processing execute block and then
use that to initialize the Allokation variable in the second model. This scheme could
look like the following:

<1stMODEL.mod> 

int AllokationErste_Optimal; ..... [subject to>] ..... execute post_process
{ 

for(var iter in WerkeProdukte) AllocationErste_Optimal[iter]=AllocationErste[iter]; 
}   <MainScript> .... MasterOpl.postProcess(); 
//this will make sure that the AllocationErste_Optimal gets initialized properly with the optimal values from the 1st run .... var ZweiteStufeSource = 

new IloOplModelSource(
"ZweiteStufe.mod"); var ZweiteStufeDef = 

new IloOplModelDefinition(ZweiteStufeSource); var ZweiteStufeCplex = 

new IloCplex(); var ZweiteStufeData = 

new IloOplDataElements();   ZweiteStufeData.Allokation = ErsteStufeOpl.AllokationErste_Optimal; .... var ZweiteStufeOpl = 

new IloOplModel(ZweiteStufeDef, ZweiteStufeCplex); ZweiteStufeOpl.addDataSource(ZweiteStufeData); ZweiteStufeOpl.generate(); ZweiteStufeCplex.solve();

This approach seems to work for a sample on my side. If you are still having problems, please attach a sample which can reproduce the error.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Mathsg

Hello, 

I am very sorry to disturbed you in your busy time but i really need few seconds of your valueble time. i am facing the same error, i.e., passing variable elements from one model to another through scripting part of OPL Cplex. I have divided my problem into two problem i.e., Master and sub problem. In the Master problem i have dvar boolean variable which is multi dimentional array and i am interested to use it in the sub model as a input parameter for solving the sub problem and then use the solution of sub problem to update this array for next iteration. I am getting an error scripting runtime error: Data element z does not exist while z is the new array created in script to store the value of decsion variable which is used in master problem. Here is my main script part of my code.

 

main {
     

  var status = 0;
  thisOplModel.generate();
  
  var RC_EPS = 1.0e-6;

  var masterDef = thisOplModel.modelDefinition;
  var masterCplex = cplex;
  var masterData = thisOplModel.dataElements;
   
  var subSource = new IloOplModelSource("PP_Jan.mod");
  var subDef = new IloOplModelDefinition(subSource);
  var subData = new IloOplDataElements();
  var subCplex = new IloCplex();
   
   
  var best;
  var curr = Infinity;

  while ( best != curr ) {
    best = curr;

    var masterOpl = new IloOplModel(masterDef, masterCplex);
    masterOpl.addDataSource(masterData);
    masterOpl.generate();
    masterOpl.convertAllIntVars();
        
    writeln("Solve master.");
    if ( masterCplex.solve() ) {
      curr = masterCplex.getObjValue();
      writeln();
      writeln("OBJECTIVE: ",curr);
    } 
    else {
      writeln("No solution!");
      var z =  new Array();
  for (var i in thisOplModel.flows){
    z[i] = new Array();
    for ( var j in thisOplModel.routes){
       z[i][j] = new Array();
       for (var u in thisOplModel.nodes){
    z[i][j][u] = new Array();
    for (var v in thisOplModel.nodes){
       z[i][j][u][v] = new Array();
       z[i][j][u][v] = 0;
     }}}}       
      for (j in thisOplModel.routes){
      for (u in thisOplModel.nodes){
      for (v in thisOplModel.nodes){
      for (v in thisOplModel.nodes){
      z[i][j][u][v] = masterOpl.lemda[i][j][u][v];
      masterData.z[i][j][u][v] = z[i][j][u][v];
       masterOpl.addDataSource(z[i][j][u][v]);
    }}}}    
      masterOpl.end();
     
      break;
    }
//subData.lemda = masterOpl.lemda;
  subData.nmbr_routes = masterOpl.nmbr_routes;
  subData.nbnodes = masterOpl.nbnodes;
  subData.flows = masterOpl.flows;
  subData.Lincards = masterOpl.Lincards;
  subData.hope_count = masterOpl.hope_count;
  subData.link = masterOpl.link;
  subData.q = masterOpl.q;
  subData.ECcards = masterOpl.ECcards;
  subData.Duals = masterOpl.Duals;
        
 for (var i in masterOpl.routes) {
 subData.Duals[i] = masterOpl.Const10[i].dual;
 }
    var subOpl = new IloOplModel(subDef, subCplex);
    subOpl.addDataSource(subData);
    subOpl.generate();

    writeln("Solve sub.");
    if ( subCplex.solve() ) {
      writeln();
      writeln("OBJECTIVE: ",subCplex.getObjValue());
    }
    else {
      writeln("No solution!");
      subOpl.end();
      masterOpl.end();
      break;
    }

    if (subCplex.getObjValue() >-RC_EPS) { 
      subOpl.end();
      masterOpl.end();
      break;
    }

    
    // Prepare the next iteration:
   //masterData.routes.add(masterData.routes.size,1,subOpl.getObjValue());
  
     for ( var j in subOpl.routes){
        masterData.z[1][j][1][1].add(masterData.z[1][j][1][1],size+1,subOpl.lemda[Opl.first(subOpl.flows)][j][1][1].solutionValue); //'Here i am getting an error
     //(subOpl.f[Opl.first(subOpl.Lincards)][j][1][1].solutionValue);
     }
    //masterData.routes.add(subOpl.getObjValue());
    
  subOpl.end();
  masterOpl.end();
  }
  writeln("Relaxed model search end.");

  masterOpl = new IloOplModel(masterDef,masterCplex);
  masterOpl.addDataSource(masterData);
  masterOpl.generate();   

  writeln("Solve integer master.");  
  if ( masterCplex.solve() ) {
    writeln();
    writeln("OBJECTIVE: ",masterCplex.getObjValue());
    if (Math.abs(masterCplex.getObjValue() - 122)>=0.0001) {
      writeln("Unexpected objective value");
      status = -1;
      
    }
    /*for(i in  masterData.Patterns) {
      if (masterOpl.Cut[i].solutionValue > 0) {
        writeln("Pattern : ", i, " used ", masterOpl.Cut[i].solutionValue << " times");
      }
    }*/
  }
  masterOpl.end();
  status;
}
  Kindly help and it would be highly appricated. I am looking forward to hearing from you. thank you

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Ben87

Thank you for your help. This works fine.

Regards,
Benjamin
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

> Thank you for your help. This works fine.

Am glad to note that. Thank you for letting us know.
#DecisionOptimization
#OPLusingCPLEXOptimizer