Originally posted by: hugotorres

I need to create a variable which has 3 indexes, let's say they're i for resource, j for zone and k for subzone. I have the resources organized in 11x11 matrixes, related to bases vs subzones. Each of this matrixes has the name Zone_X in excel and I need to import each one of them to the variable, to have it in a format described above. Can this be done by defining the range of the zones, and then dynamically allocating the matrixes to the vector? something like:

z = 2000

range zones = 1..z.;

//create a cycle here

var[zones] = zone_(zones)

Would something like this work? I'm asking to get ahead on my work, since I don't have the excel file ready yet, just want to discuss the best way to do this

Hi,

could

https://www.ibm.com/developerworks/community/forums/html/topic?id=bcaea400-c839-4f12-85bc-4750c48d68f3&ps=25

help ?

regards

Originally posted by: hugotorres

Now that I see it, I might have phrased it wrongly. 

I have a ~8400*11 matrix, and this is what I want to import to CPLEX. This matrix aggregates all data from all zones I have to consider, each zone in a 11x11 matrix.

What I want to do is to create a cycle where zones[i] corresponds to the matrix starting in the i*11 row and ends on the (i+1)*11 row. Each 11x11 matrix has bases as rows and subzones as columns, so in the end, the matrix zones will be something like zones[i][base][subzone].

Hi,

do you mean your 2D and 3D array should have the same total number of cells ?

regards

Originally posted by: hugotorres

Yes, sort of. I have an example I can use to illustrate this:

The 3D array array[x][y][z] should have x as 1,2, etc, the ones on the far left side, and inside each one, the're supposed to have the matrix in front of it, which are [y]*[z]

Hi,

what about something like

int N=11;
int M=10;

int A[ij in 1..M*N][k in 1..N]=rand(10);

int B[i in 1..M][j in 1..N][k in 1..N]=
A[1+(i-1)*N+(j-1)][k];

assert sum(ij in 1..M*N,k in 1..N) A[ij][k]==
sum(i in 1..M,j in 1..N,k in 1..N) B[i][j][k];

?

regards