伪Newton-B族的导出及其性质

Derivation and Property of Pesudo-Newton-N Class

本文对无约束优化问题提出了一类新的近似牛顿法（伪牛顿-B族）,此方法同样具有二次终止性,产生的矩阵序列保持正定对称传递性。并证明了算法的全局收敛性和超级性收敛性。

In this paper, a new Newton-like method (pesudo-Newton-B class) is presented for unconstrained optimization, this method possesses the property of finite convergence for quadratic function and the sequence of matrix updated by the method maitain sysmmetric and positive definite. The global convergence and superlinear convergence are proved.