Semilocal convergence of a sixth-order method in Banach spaces (Q1762497): Difference between revisions

The paper deals with the approximate solution of a nonlinear equation \(F(x)= 0\), where \(F\) is a mapping of a convex set \(\Omega\) of a Banach space \(X\) in a Banach space \(Y\). It is assumed that \(F\) is Fréchet-differentiable of order 3. To solve the equation numerically we know Newton's method, Chebyshev's method, Halley's method, Newton-likes methods, the super-Halley's method. For the solution of nonlinear Hammerstein integral equations some fourth-order methods are used. In this paper, a new iterative method of order six is proposed: Suppose \(x_n\) is given. Then denote \[ \begin{aligned}\Gamma_n &= [F'(x_n)]^{-1},\\ u_n &= x_n-\textstyle{{1\over 3}}\Gamma_n F(x_n),\\ K(x_n) &= \Gamma F''(u_n)\Gamma_n F(x_n),\\ z_n &= x_n -\{I+\textstyle{{1\over 2}} K(x_n)[I- K(x_n)]^{01}\} \Gamma_n F(x_n),\\ x_{n+1} &= z_n- [I+ \Gamma_n F''(u_n)(z)_n- x_n)]^{-1} \Gamma_n F(x_n).\end{aligned} \] The semilocal convergence of the iterations is proved. The existence and the uniqueness of the solution of the nonlinear equation can be proved under certain assumptions. An a priori error estimate is given. Two numerical applications (the integral equation of Chandrasekhar and a nonlinear Hammerstein integral equation of second kind) demonstrate the proposed method. In these examples the numerical results are better than those of other known methods.