Variants of Newton's method for functions of several variables

From MaRDI portal





Newton-like methods are discussed for finding a real solution of a system of nonlinear equations: \(F(x)=0\) in \(\mathbb R^n\). The authors propose a midpoint Newton method: \[ x^{(k+1)} = x^{(k)} - J_F((x^{(k)} + z^{(k)})/2)^{-1}F(x^{(k)}, \quad k=0, 1, \dots. \] Here \(J_F(x)\) is the Jacobian matrix of the function \(F\). \(z^{(k)}\) is defined via a Newton step as \[ z^{(k)} = x^{(k)}-J_F(x^{(k)})^{-1}F(x^{(k)}). \] The midpoint Newton method is proven to be of quadratic covergence and illustrated to be better than the Newton method itself with numerical examples. But it is not compared with the two step Newton method and the cost of evaluation of the function \(F\) and its Jacobian at each iteration step is not considered.




Cited in
(68)








This page was built for publication: Variants of Newton's method for functions of several variables

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q864757)