Let's start by looking at how we'll initialize this objective function. The complete source is available, but since it's so big, let's look at it piece by piece again:

import org.coinor.opents.*;

public class MyObjectiveFunction implements ObjectiveFunction
{
    public double[][] matrix;
    
    public TSPObjectiveFunction( double[][] customers ) 
    {   matrix = createMatrix( customers );
    }   // end constructor

    public double[] evaluate( Solution solution, Move move)
    {
        ...
    }   // end evaluate

    ...
    
}   // end class MyObjectiveFunction

The constructor expects to be passed that nx2 matrix of customer locations that we generated in the main program. We then create a matrix of distances among these customers in the createMatrix(...) method. Let's look at that:

import org.coinor.opents.*;

public class MyObjectiveFunction implements ObjectiveFunction
{
    ...

    /** Create symmetric matrix. */
    private double[][] createMatrix( double[][] customers )
    {
        int len = customers.length;
        double[][] matrix = new double[len][len];
        
        for( int i = 0; i < len; i++ )
            for( int j = i+1; j < len; j++ )
                matrix[i][j] = matrix[j][i] = norm(
                    customers[i][0], customers[i][1],
                    customers[j][0], customers[j][1] );
        return matrix;
    }   // end createMatrix
    
    /** Calculate distance between two points. */
    private double norm( double x1, double y1, double x2, double y2 )
    {   
        double xDiff = x2 - x1;
        double yDiff = y2 - y1;
        return Math.sqrt( xDiff*xDiff + yDiff*yDiff );
    }   // end norm
    
}   // end class MyObjectiveFunction

In fact we used two methods: createMatrix(...) and norm(...). The second method simply applies the Pythagorean theorem to the two sets of (x,y) coordinates.

The matrix is symmetric, so we could get away with creating a double matrix that was not square, but then we'd have to worry about making sure we called for [i][j] instead of [j][i]: too confusing. The little extra memory needed for a full, square matrix is worth the fewer headaches.

We still haven't seen how we're going to evaluate a solution, so we'll look at that next.