Wolfe线搜索充分下降的修正DY共轭梯度法

摘要

基于DY和DL共轭梯度法,给出一个新的βk公式,在精确线搜索下该公式等价于βDYk.基于新参数公式建立了采用Wolfe线搜索的共轭梯度算法,证明了算法满足充分下降性和全局收敛性,初步的数值试验结果表明该方法是有效的,适合于求解非线性无约束优化问题.%Based on the DY and DL conjugate gradient methods,a new parameter formulaβk was presented,which is the same as,βDYk when the line search is exact.Based on the new formula,a new conjugate gradient algorithm with the Wolfe line search was proposed,and the sufficient descent property and the global convergence of the algorithm were proved.Preliminary numerical results show that the new method was effective and it suitable for solving nonlinear unconstrained optimization problems.