Highly efficient family of iterative methods for solving nonlinear models (Q1624624): Difference between revisions

This manuscript deals with a family of iterative methods for solving nonlinear scalar equations \(f(x)=0\), where \(f\) is a real valued function. The proposed class has the following iterative expression: \begin{align*} &y_n = x_n-\dfrac{2}{3}\dfrac{f(x_n)}{f'(x_n)},\\ &z_n = x_n - \left[a_1 + a_2\left( \dfrac{f'(x_n)}{f'(y_n)} \right)^2 \right] \dfrac{f(x_n)}{f'(x_n)},\\ &x_{n+1} = z_n- \left[\dfrac{f'(x_n)+b_1f'(y_n)}{b_2f'(x_n)+b_3f'(y_n)} \right] \dfrac{f(z_n)}{f'(x_n)}, \ \ n=0,1,\ldots \end{align*} where \(a_1\), \(a_2\), \(b_1\), \(b_2\) and \(b_3\) are real parameters. Under standard conditions on the function \(f\), on the initial estimation \(x_0\) and on the parameters, the authors prove the sixth-order convergence of the proposed family. Since all the denominators of the above iterative expressions are first derivatives of the function \(f\), it is easy to extend this class of methods to the context of systems of nonlinear equations, \(F(x)=0\), replacing the derivative by the Jacobian matrix of \(F\). For this type of models, the authors prove the sixth-order convergence under the same conditions. A dynamical study of some elements of the above family is presented, determining the fixed and critical points of the rational operator obtained when the method is applied on a generic quadratic polynomial. In order to show the applicability of the proposed methods, the authors choose some real world problems, e.g., kinematic syntheses, boundary value, Bratu's 2D problem and Fisher's and Hammerstein integral problems in the case of nonlinear systems. In addition, numerical computations on simple scalar examples are carried out to show the performance of the proposed techniques.