توسیعی از یک روش گرادیان مزدوج سه بخشی مبتنی بر مقادیر تابع هدف با تضمین همگرایی بدون فرض تحدب

With respect to importance of the conjugate gradient methods for large-scale optimization, in this study a descent three-term conjugate gradient method is proposed based on an extended modified secant condition. In the proposed method, objective function values are used in addition to the gradient information. Also, it is established that the method is globally convergent without convexity assumption on the objective function. Numerical experiments show reasonable computational efficiency of the suggested method.