摘要
对无约束最优化问题提出了一类非拟Ｎｅｗｔｏｎ族算法，它不再是Ｈｕａｎｇ族中的成员．与拟Ｎｅｗｔｏｎ法相比，新给出的校正公式，在不增加计算量的前提下，能吸纳更多的信息，且仍保持正定对称传递性．对一致凸的目标函数，证明了算法的整体收敛性．且结论对众多类型的精确与非精确线搜索均能成立，而这些线搜索在最优化算法中是比较有效且常用的．
A class of Non Quasi Newton method is presented,it does not belong to the Huang class.Compared the new method with Quasi Newton method, it does not add capcity of computing and may draw more information. The sequence of matrices updated by the new algorithm remain symmetric and positive definite. The global convergence of such a class of methods with many kinds of exact/inexact line search is proved when applied to the uniform convex objective function.
