Nonlinear successive over-relaxation (Q1059370): Difference between revisions

Let \(\Phi\) be a real strictly convex functional defined and twice continuously differentiable on a convex domain in \({\mathbb{R}}^ n\). To find its minimum the authors solve a system of nonlinear equations \(F(x)=0\) in \({\mathbb{R}}^ n\), where F denotes grad \(\Phi\). They present two theorems giving sufficient conditions of convergence for: (i) a nonlinear analogue of the Gauss-Seidel method for positive-definite matrices, (ii) a nonlinear successive overrelaxation method, used for such a system of equations. The theorems are parallel to results of Schechter (1962, 1968) but give more general sufficient conditions and may be applied for a more general class of functionals whose Hessian matrix may be singular.