SOME THREE-TERM CONJUGATE GRADIENT METHODS FOR SOLVING UNCONSTRAINED OPTIMIZATION PROBLEMS

摘要

In this paper, we investigate some three-term conjugate gradient methods, since these methods are an efficient class for unconstrained optimization problems. Three-term conjugate gradient methods are obtained to improve traditional conjugate gradient methods. Conjugate gradient methods have low memory requirements and strong local and global convergence properties. Here, we focus on the three-term conjugate gradient methods which generated directions satisfy the descent condition, the sufficient descent condition and the Dai-Liao conjugacy condition. We compare some three-term conjugate gradient methods on the unconstrained CUTEst test problems.