An accelerated subspace minimization three-term conjugate gradient algorithm for unconstrained optimization (Q457047)

An algorithm based on minimizations of subspaces is presented to solve large-scale unconstrained optimization problems. The algorithm is a three-term conjugate gradient method satisfying both the descent and the conjugacy conditions. The global convergence of the algorithm is established for uniformly convex functions. Some experimental results on 750 unconstrained optimization test problems are presented to compare the proposed method with four existing conjugate gradient algorithms and two other large-scale unconstrained optimization methods.