New nonlinear conjugate gradient formulas for large-scale unconstrained optimization problems (Q849738)

Some new nonlinear conjugate gradient (CG) formulas for computing the search directions for unconstrained optimization problems are proposed. The significant differences of this approach from the classical studies on CG methods are that: 1) some proposed formulas possess the sufficient descent property without any line searches, and 2) the global convergence results for some of these given formulas with the standard Armijo line search are also given. The new formulas turn out to be the conjugate descent formula if exact line searches are made. General convergence results for the proposed formulas with the weak Wolfe-Powell conditions are studied. It is proved that some of the formulas with the steplength technique which ensures the Zoutendijk condition to be held are globally convergent. Preliminary numerical results show that the proposed methods are very promising.