Introduction to a Newton-type method for solving nonlinear equations (Q2470174)

A new Newton-type method called rational Newton method to find a simple root of a nonlinear equation \(f(x)= 0\) is introduced. In the classical Newton method the iteration formula is \(x_n= x_{n-1} x-{f(x_{n-1})\over f'(x_{n-1})}\), where \(x_0\) is an initial approximation of the root. In the proposed method we compute first \(y_{n-1}= x_{n-1}- {f(x_{n-1})\over f'(x_{n-1})}\) and next \[ x_n= x_{n-1}- {f(x_{n-1}\cdot f'(y_{n-1})^2+ f(x_{n-1})\cdot f'(x_{n-1})^2\over f'(x_{n-1})\cdot f'(y_{n-1})^2+ f'(x_{n-1})\cdot f'(y_{n-1})^2}. \] In contrast to the quadratic convergence rate of the classical Newton method, it is proved that the order of convergence of the proposed method is cubical. The effectiveness of the rational Newton method is compared with known methods as Halley's rational method, \textit{J. Kou, Y. Li} and \textit{X. Wang}'s method [Appl. Math. Comput. 181, No. 2, 1106--1111 (2006; Zbl 1172.65021)] and \textit{S. Weerakoon} and \textit{T. G. I. Fernando}'s method [Appl. Math. Lett. 13, No. 8, 87--93 (2000; Zbl 0973.65037)] by several numerical examples. The proposed method is found to be substantially more stable and accurate.