Therefore, x3 will be considerably smaller in magnitude still.
Thus, for small x, f(x) ≈ 3kx2 − 3k2x + k3 − a, and our problem
can be reduced to that of finding the roots of the corresponding
quadratic function g (x) = 3kx2 − 3k2x + k3 − a.
This presents a small problem — while f(x) has only one root
(of multiplicity 3), g(x) has two roots. Which of these roots is the
approximation of b? Since x (the root of f (x)) is assumed to be so
close to zero, the root of g(x) that we seek is logically the smaller
of the two roots. So if x0 is root of g (x) of lesser magnitude, and
x∗ is the only root of f (x) , then we will have