Originally posted by: hllsen

To use in a MIP model, I created 3D variable X_{k,j,t} with sizes n, m and H, respectively. Type definition and initialization is like this:

typedef IloArray<IloArray<IloNumVarArray > >     IloNumVarArray3; IloNumVarArray3 X( env );

Then array is filled like this:

X.add( m, IloArray<IloNumVarArray >( env ) ); 

for (k = 0; k < m; ++k ) X[k].add( n, IloNumVarArray( env, H, 0, IloInfinity, IloNumVar::Float ));

After considerable amount of debugging time, I noticed that this method does not create n*m*H number of variables. That is, "id" of different k or j index variables has the same "id". Therefore, I ended up with only H number of variables.

Variable id
X[0][0][0] 1
X[0][1][0] 1
X[1][2][0] 1
X[0][0][1] 2
X[0][1][1] 2
X[1][2][1] 2

But, the code below just works fine and creates n*m*H number of variables.

for (k = 0; k < m; ++k ) 
{ X.add( IloArray<IloNumVarArray >( env ) ); 

for (j = 0; j < n; ++j) 
{ X[k].add( IloNumVarArray( env, H, 0, IloInfinity, IloNumVar::Float ) ); 
} 
}

AFAIK, my problem is solved but I still couldn't quite understand why first method does not work. An explanation would be very much appreciated. Thank you.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

> hllsen wrote:

>

> X.add( m, IloArray<IloNumVarArray >( env ) );
>

I think what this is doing is creating one 2D array and adding that one array m times -- meaning all m entries are pointers to the same array. The C++ manual says that the overload of add() you are using adds the same second argument m times. The constructor in the second argument is only called once.

Paul

Mathematicians are like Frenchmen: whenever you say something to them, they translate it into their own language, and at once it is something entirely different. (Goethe)
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: hllsen

Thanks for the clarification Paul...
#CPLEXOptimizers
#DecisionOptimization