Método do ponto proximal inexato e uma técnica de busca linear não monótona para otimização irrestrita

Abstract

This paper shows algorithms for problem solving Unrestricted. Initially it will address the Proximal Point Algorithm Inaccurate using classical algorithms for solving the problem of regularization of convex function, continuously differentiable and decisive the Hessian near zero. Then the Search Algorithm Nonlinear monotone that aims to improve the likelihood of find a global optimum, using traditional methods to decrease obtaining the step size, moreover, they can improve the speed of convergence in specific cases of monotonous scheme. At the end we will implementation of quadratic functions and analysis of results