Originally posted by: Ben87

Hallo,
I want to add iterative a two-dimensional matrix to a tuple in the Main-Script, so I can use it in a submodel (which I also solve iterative) as a constraint. I do not know the number of elements or the values of the elements in advance. For example I want to add in every iteration a matrix like that:

[[0 -1350 0 0] [0 0 0 -500] [-600 -225 -250 -1680]]

While using just single values it works fine, when doing it in the following way in Opl Script:

SomeOpl.BetaTuple.add(0); SomeOpl.BetaTuple.add(1.1);

or:

SomeOpl.BetaTuple.add(someVariable);

and in my submodel I use it in the following way:

forall ( f in BetaTuple) ctKoeffizienten: f.b <= Theta;

My overall main goal with these matrices and the single values is to do something like that in the sub-model:

tuple BCuts
{ 

float b; 
} 
{BCuts
} BetaTuple=
{
};   tuple ACuts
{ 

float a[Werke][Produkte]; 
}
// I know that this is not possible, but some like that i want to do.   
{ACuts
} AlphaTuple=
{
}; … Minimize… Subject to
{ … forall (e in AlphaTuple, f in BetaTuple, i in facility, j in products) ctCoefficient: (sum(i in facility, j in products)e.a[i][j] * Allocation[i][j]) + f.b <= Theta; …
}

Maybe the mathematical formulation helps to understand the problem a bit better:

Min ( (sum(i in facility, j in products) (CostAllocation[i][j] * Allocation[i][j])) + Theta)   s.t.  Theta >= ( sum(i in facility, j in products)a_k[i][j]*y[i][j]) + b_k ; forall(k)

k = Iteration;
Allocation[i][j] is a boolean;

I know that multidimensional Arrays are not allowed in Tuples. So my question is, if there is any other method or work-around for doing something I have described above?

Best Regards,
Benjamin
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

> I know that multidimensional Arrays are not allowed in Tuples. So my
> question is, if there is any other method or work-around for doing
> something I have described above?

How about changing the ACuts tuple structure to something like:

tuple ACuts{
        //float a[Werke][Produkte];
        int facility;
        int products;
        float a;
}

So, you will have one tuple entry for every element in your 2d array, consisting of the 1st index, 2nd index and the actual value. Thus if your variables look like:

BetaTuple = {<1>,<1.1>};
AlphaTuple = {<1,3,7>,<1,4,8>,<2,3,9>,<2,4,10>};
facility={1,2};
products={3,4};
a[facility][products]=[[7,8],[9,10]];

you should be able to add them like such:

subOpl.AlphaTuple.add(1,3,7);
   subOpl.AlphaTuple.add(1,4,8);
   subOpl.AlphaTuple.add(2,3,9);
   subOpl.AlphaTuple.add(2,4,10);

Hope this helps.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Ben87

Thanks for the idea. I should really work more often with tuples.

Best regards,
Benjamin
#DecisionOptimization
#OPLusingCPLEXOptimizer