Originally posted by: beratpostalci

Hi all,

 

I'm learning Cplex Java API to solve TSP. I am able to generate obj (paths between cities with random weights) like below:

 

randValue x12 + randValue x13 + randValue x14 + ...

 

with this code:

for(int i = 0; i < n; i++) {
        
        x[i] = cplex.numVarArray(n, 0.0, Double.MAX_VALUE);
        for(int j = 0; j < n; j++) {
                if(i != j) {
                        if (i + 1 > 10 && j + 1 < 10) {
                                x[i][j].setName("x" + (i + 1) + "0" + (j + 1));
                        } else {
                                x[i][j].setName("x" + (i + 1) + (j + 1));
                        }                                     
                        expr.addTerm(50 + r.nextInt(951), x[i][j]);
                } 
        }

    
}

Then, I want to add a constraint for 5 cities like below:

 

x12+x13+x14+x15 = 1

 

with this code:

 

for(int i = 0; i < n; i++) {
        for(int j = 0; j < n; j++) {
                if(i != j) {
                        if(j >= n - 1) {
                                cplex.addEq(cplex.sum(x[i][j],0), 1);
                        } else {
                                cplex.addEq(cplex.sum(x[i][j],x[i][j + 1]), 1);
                        }
                        j++;
                }
        }
}

 

However when I write System.out.println( cplex.toString() ); It gives output as below:

 

IloModel  {
IloMinimize  : (738.0*x12 + 968.0*x13 + 268.0*x14 + 732.0*x15 + 431.0*x21 + 173.0*x23 + 675.0*x24 + 174.0*x25 + 371.0*x31 + 831.0*x32 + 236.0*x34 + 279.0*x35 + 293.0*x41 + 119.0*x42 + 673.0*x43 + 465.0*x45 + 125.0*x51 + 433.0*x52 + 875.0*x53 + 561.0*x54)
IloRange  : 1.0 <= (1.0*x12 + 1.0*x13) <= 1.0
IloRange  : 1.0 <= (1.0*x14 + 1.0*x15) <= 1.0
IloRange  : 1.0 <= (1.0*x21 + 1.0*[0.0..infinity]) <= 1.0
IloRange  : 1.0 <= (1.0*x23 + 1.0*x24) <= 1.0
IloRange  : 1.0 <= (1.0*x25) <= 1.0
IloRange  : 1.0 <= (1.0*x31 + 1.0*x32) <= 1.0
IloRange  : 1.0 <= (1.0*x34 + 1.0*x35) <= 1.0
IloRange  : 1.0 <= (1.0*x41 + 1.0*x42) <= 1.0
IloRange  : 1.0 <= (1.0*x43 + 1.0*[0.0..infinity]) <= 1.0
IloRange  : 1.0 <= (1.0*x45) <= 1.0
IloRange  : 1.0 <= (1.0*x51 + 1.0*x52) <= 1.0
IloRange  : 1.0 <= (1.0*x53 + 1.0*x54) <= 1.0

}

 

How can I generate constraint  x12 + x13 + x14 + x15 = 1   ?

 

 

 

 

 

 

 

#CPLEXOptimizers
#DecisionOptimization

It's a bit hard to see what is going on here, particularly since your code for assigning names looks buggy to me. (One symptom: The "1.0*[0.0..infinity])" suggests that at least one variable did not get a name.) If you're looking for suggestions on how to add up a bunch of variables in a constraint, there are multiple ways to do it.

One approach is as follows.

• Create a vector to hold the variables: IloNumVar[] temp = new IloNumVar[<number of terms>];
• Plug each variable you want to add into a slot in temp (using a loop, or nested loops).
• Use cplex.addEq(cplex.sum(temp), ...) to create the constraint.

A variation on that approach is to make temp an instance of ArrayList<IloNumVar>. That saves you having to count the number of terms up front and keep track of an index; you just use the ArrayList.add() method to populate temp. You then need to use ArrayList.toArray() to export it to IloNumVar[] and feed that to cplex.sum().

A different approach is as follows.

• Create an empty expression: IloLinearNumExpr temp = cplex.linearNumExpr();
• Inside a loop, or nested loop, add your variables to the expression: temp.addTerm(1.0, <next variable>);
• Use cplex.addEq(temp, ...) to create the constraint.

There are other variations, but that should be enough to get you rolling.

 

 

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: beratpostalci

Thanks for your answer. Since I eliminate main diagonal in matrix with i != j statement their names are not assigned. I'll try out your suggestions.

#CPLEXOptimizers
#DecisionOptimization