对加速自适应Perry-共轭梯度法全局收敛性的进一步研究

Further Discussion on the Global Convergence of Accelerated Adaptive Perry Conjugate Gradient Algorithms

提出了一类有效的求解大规模优化问题的共轭梯度法(AGGSSV),但其全局收敛性是在目标函数为一致凸的条件下成立,研究了目标函数不是凸函数的条件下,共轭梯度法(AGGSSV)的全局收敛性.

In the report,a class of efficient conjugate gradient algorithms ( ACGSSV) was proposed to solve large-scale unconstrained optimization problems,however,its global convergence property was established under the condition that the objective function is uniformly convex. The global convergence of ACGSSV without convexity assumption on the objective function was also discussed.