Originally posted by: KarolisK.

Hello all,

I am using column generation to solve my mathematical problem.

The problem that I encounter is that the decision variable in my submodel is three dimensional z[j][k][c] which is used later to add new patterns to a tuple by using following line of script:

masterData.LocPattern.add(masterData.LocPattern.size, 1, subOpl.z.solutionValue);

As it is known a tuple does not support multidimensional arrays. So my question is how can a three dimensional dvar be converted into a flat one dimensional array which could be used in a tuple afterwards?

 

Originally posted by: KarolisK.

I have found that 3-dimensional array can be transformed into 1-dimensional array like this:

z[j][k][c]  -->> z[j*K*C+k*C+c]   where capital K and C are sizes of arrays.

But my question still remains the same: how and where this should be implemented in OPL?..

Hi

could

int C=4;
  int J=5;
  int K=2;
 
 
  int z[c in 1..C][j in 1..J][k in 1..K]=c+j+k;
 
  int Z[1..C*J*K]=[ 1+(c-1)*J*K+(j-1)*K+k-1: z[c][j][k] | c in 1..C,j in 1..J,k in 1..K];

help ?