Proximal-Point Algorithm for Minimizing Quadratic Functions
Abstract
The proximal-point algorithm is studied in the framework of the minimization of a quadratic function. In this context it can also be used to compute the least eigenvalue of the hessian and an associated eigenvector. The speed of convergence is analysed and a connection with the power method is pointed out. A modified algorithm with quadratic convergen- ce is studied. Numerical tests assess the theoretical results