Originally posted by: BerkUstun

I'm currently coding on C++ along with the CPLEX API and have recently discovered that there are matrix arrays in CPLEX (i.e. IloNumArray2) but there is no documentation on them.

I am wondering how I can use matrices within CPLEX.

1. Can I get away with using matrices to set up constraints without a for loop (i.e. .add(Ax = 0) where c is an A is a IloNumArray2 and x is a IloNumVarArray?

2. Can I add two matrices together (i.e. C = A + B) or multiply them by scalars (i.e. C = 2*A). Here, A, B and C are IloNumArrays2 (actually, it would also help to know whether I can do this for vectors)

3. Can I append a matrix? If so, how?

Anything helps!
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Answer to question 1: I don't think so. However, if you have to do that frequently it should be easy to write a function that does that.

Answer to question 2: I don't think so either. But it is again easy to define you own operator+ for that purpose if you want to that a lot (untested):

struct SizeMismatchException {
};
 
IloNumArray2 operator+(IloNumArray2 const& a, IloNumArray2 const& b)
{
   if (a.getSize() != b.getSize())
      throw SizeMismatchException();
 
   IloNumArray2 c(a.getEnv());
 
   for (IloInt i = 0; i < a.getSize(); ++i) {
      IloNumArray const& arow = a[i];
      IloNumArray const& brow = b[i];
      if (arow.getSize() != brow.getSize()) {
         for (IloInt j = 0; j < c.getSize(); ++j)
            c[i].end();
         c.end();
         throw SizeMismatchException();
      }
      IloNumArray row(c.getEnv());
      for (IloInt j = 0; j < arow.getSize(); ++j) {
         row.add(arow[j] + brow[j]);
      }
      c.add(row);
   }
 
   return c;
}

A similar operator can be defined for two-dimensional arrays of variables or for vectors.

Answer to question 3: IloNumArray2 is just an IloArray of IloNumArray instances. You can easily append new rows or append elements to rows:

IloNumArray2 A(env);
 
// Append a row:
// 1. create the new row.
IloNumArray row(env);
...
// 2. append the new row.
A.add(row);
 
// Append an element to row i:
A[i].add(3.58);

Be careful, IloNumArray2 implements a dense matrix. If you get large dimensions you may run out of memory. You may want to consider using sparse matrix implementations.
#CPLEXOptimizers
#DecisionOptimization