对Choleski-PCG Newton算法的一些改进

Some Improvement on Choleski-PCG Newton Method

对无约束最优化问题提出了一种不精确牛顿算法模型ACPN（α），是对DengN．Y．和WangZ．Z文（CanNewtonmethodbesurpassed．见ChineseScienceBulletin，1998，Vol．43，No．20，p．132～134）中Choleski-PCGNewton算法的改进。新算法对于变量个数在35～186范围内的无约束问题更有效，并打破了所构造的点列必须恰Q－2阶收敛的局限，对进一步改进算法有提示作用。

An inexact Newton method to the unconstrained optimization problem, algorithmACPN (α), is proposed as some improvement on Choleski-PCG Newton method given inreference Deng N, Y. and Wang Z. Z. (Can Newton method be surpassed. In: Chinese ScienceBulletion, 1998,Vol. 43, No. 20,p. 132 - 134). It's more efficient when variable quantities arebetween 35 and 186 without the constraints of precisely quadratic convergence.