Originally posted by: lriWANGCHEN

Hello,everyone,

I want to define a matrix like this Y[t][i][s][j], and the element in the matrix is float, I write the C++ codes with cplex like this:

NumVarMatrix Y(env,n);
for(int i=0;i<n;i++)
{
for(int j=0;j<n;j++)
{
Y[i][j]=IloNumVarArray(env,n,0.0,1.0,ILOFLOAT);
}
}

But it is not right, so could someone can help me? Thank you so much.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

I see three different ways to implement that:
1. Probably the easiest but not nicest: You the IloArray template:

typedef IloArray<IloNumVarArray> NumVarArray2; 
// 2D array typedef IloArray<NumVarArray2>   NumVarArray3; 
// 3D array typedef IloArray<NumVarArray3>   NumVarArray4; 
// 4D array   NumVarArray4 Y(env); 

for (

int t = 0; t < ...; ++t) 
{ NumVarArray a3(env); 

for (

int i = 0; i < ...; ++i) 
{ NumVarArray a2(env); 

for (

int s = 0; s < ...; ++s) 
{ a2.add(IloNumVarArray(env, n, 0.0, 1.0, ILOFLOAT)); 
} a3.add(a2); 
} Y.add(a3); 
}

2. Use a flat IloNumVarArray and a macro:

#define Y(t,i,s,j) varY[(t) * maxI * maxS * maxJ + (i) * maxS * maxJ + (s) * maxJ + (j)] 

int maxT = ..., maxI = ..., maxS = ..., maxJ = ...; IloNumVarArray varY(env); 

for (

int t = 0; t < maxT; ++t) 
{ NumVarArray a3(env); 

for (

int i = 0; i < maxI; ++i) 
{ NumVarArray a2(env); 

for (

int s = 0; s < maxS; ++s) 
{ 

for (

int j = 0; j < maxJ; ++j) 
{ varY.add(IloNumVar(env, 0.0, 1.0, ILOFLOAT)); 
} 
} a3.add(a2); 
} 
} 
// Now access Y[0][1][2][3] like as Y(0, 1, 2, 3)

3. Use recursive templates. This is IMO the most elegant approach but it requires some understanding about templates. I have posted example code for this on the Forum before.
#CPLEXOptimizers
#DecisionOptimization