Analysis of sparse quasi-Newton updates with positive definite matrix completion (Q489095)

Yamashita proposed a quasi-Newton update, called MCQN, for solving the unconstrained optimization problem. Under suitable assumptions, the local and superlinear convergence of the MCQN update with Broyden's convex family is established. A global analysis is provided for the MCQN update with Broyden's convex family assuming that the objective function is uniformly convex and its dimension is two. Some numerical study on five test problems of the MCQN update with Broyden's positive family is presented.