Even the most basic matrix operations of addition, subtraction, and, especially, multiplication can put a strain on a computer's computational capabilities. The matrix operands do not have to be very big before they generate large numbers of individual floating-point operations. This is prime breeding ground for accumulated roundoff errors.
In this chapter, we'll begin to develop the
numbercruncher.matrix package. Its classes include
ColumnVector. We'll add class
LinearSystem for solving systems of simultaneous linear equations in Chapter 10 and class
InvertibleMatrix in Chapter 11. Figure 9-1 shows the inheritance tree for these classes.
Figure 9-1. The inheritance ...