Originally posted by: Dmartinc

Hi,

I am trying to use a variable as an index in a coeffiecient matrix,
What should I do in order to make it work?

/* //Objective function
IloObjective Obj_Func_1(env);
for (j = 0; j < NTasks; j++)
{
Obj_Func_1 += c[Tasksj][j];
}
model.add(IloMinimize(env, Obj_Func_1));

//Tasks all different
model.add(IloAllDiff(env, Tasks)); */

David
#CPOptimizer
#DecisionOptimization

Originally posted by: mslusky

If c and Tasks are one-dimensional arrays, you could use IloElement:

Obj_Func_1 += IloElement(c, Tasks[j]);

Marla
#CPOptimizer
#DecisionOptimization

Originally posted by: Dmartinc

Thank you Marla,

but in the case that I have a 2 dimension array, does anyone know how to make it work?

David
#CPOptimizer
#DecisionOptimization

Originally posted by: ChrisBr

Hello David,

I think some character have been eaten by the writer, so I guess you wanted to write

c[ Tasks[j][j] ] 
//case 1

or

c[ Tasks[j] ][j] 
//case 2

In case 1
I suppose that Task is an IloArray<IloIntVarArray> (so 2 dimensions) and c is either an IloIntArray or a IloNumArray (so 1 dimension)

Note that in this case you cannot write IloAllDiff(env, Tasks), you have to flatten the array Tasks before.

Here is an sample of code which looks like what you want:

//model.add(IloAllDiff(env, Tasks)); // cannot compile! IloIntVarArray Tasks2(env, NTasks*NTasks); 

for (i=0; i<NTasks; ++i) 

for (j=0; j<NTasks; ++j) Tasks2[i*NTasks+j] = Tasks[i][j]; model.add(IloAllDiff(env, Tasks2));   IloExpr Obj_Func_1(env, 0); 

for (j = 0; j < NTasks; j++) 
{ Obj_Func_1 += c[Tasks[j][j]]; 
} IloObjective objective = IloMinimize(env, Obj_Func_1); model.add(objective);

In case 2
I suppose that Task is an IloIntVarArray (so 1 dimension) and c is either an IloArray<IloIntArray> or a IloArray<IloNumArray> (so 2 dimensions)
You have to flatten the array c

Here is an sample of code which looks like what you want:

model.add(IloAllDiff(env, Tasks)); IloIntArray c2(env, NTasks*NTasks); 

for (i=0; i<NTasks; ++i) 

for (j=0; j<NTasks; ++j) c2[i*NTasks+j] = c[i][j];   IloExpr Obj_Func_1(env, 0); 

for (j = 0; j < NTasks; j++) 
{ 
//Obj_Func_1 += c[Tasks[j]][j]; // cannot compile! Obj_Func_1 += c2[Tasks[j]*NTasks +j]; 
} IloObjective objective = IloMinimize(env, Obj_Func_1); model.add(objective);

I hope this helps,

Chris
#CPOptimizer
#DecisionOptimization

Originally posted by: Dmartinc

Thanks Chris,

it was the second case:

c[Tasks[j]][j]

Now, following the same approach, I have M Employees who can have or not the enough skills, in order to carry out a tasks. Thus,
I have an elegibility matrix, with domain http://0..1, being 1 if the employee i is elgibiled to carry out task j, otherwise cannot.

Would be the following contraint properly defined according to the previous statement?

//Elegibility constraint IloIntArray e2(env, MEmployees*NTasks); 

for (i=0; i<MEmployees; ++i) 

for (j=0; j<NTasks; ++j) e2[i*NTasks+j] = e[i][j];   

for(i=0; i< MEmployees; ++i) 
{ 

for(j=0; j< NTasks; ++j) 
{ model.add(e2[Tasks[j]*NTasks+j]==1); 
} 
}

David
#CPOptimizer
#DecisionOptimization

Originally posted by: Dmartinc

Sorry, it should be like this:

for(j=0; j< NTasks; ++j) 
{ model.add(e2[Tasks[j]*NTasks+j]==1); 
}

I am not quite sure if the approach it´s the correct one.

David
#CPOptimizer
#DecisionOptimization

Originally posted by: ChrisBr

Hello David,

Please, note that if the variable is used as index in last dimension, you don't need to flatten the array, so you can write:

for(i=0; i< MEmployees; ++i) 
{ 

for(j=0; j< NTasks; ++j) 
{ model.add(e[i][Tasks[j]]==1); 
} 
}

Regards,

Chris
#CPOptimizer
#DecisionOptimization

Originally posted by: Dmartinc

Thanks Chris,

That it´s good to know, because my current approach is to assign Employees to Tasks,
but it could be the other way around.

Thanks again.

David
#CPOptimizer
#DecisionOptimization