无约束多变量问题最优化方法──最佳方向法

The Optimization Method for Solving Problems of Unconstrained Multi-variable──Direction Method of Optimality

根据最速下降法，提出了无约束多变量问题新的最优化方法──最佳方向法．所谓最佳方向法指的是从初始点出发，确定某一搜索方向，沿着这个方向直接找到极值点．本文阐述了此方法的基本原理、计算过程和计算框图．此方法既保持了最速下降法的优点，又消除了其锯齿现象，有效提高了计算速度，与其它有关的这一类算法相比，最佳方向法也显示出计算简单、占用计算机内存空间少和收敛速度快的优点。

A new optimization method is pot forward for solving problems of unconstrained multi-variable - direction method of optimality on the basis of steepest descent method. The direction method of optimality means to determine a direction of search, along which the extreme point can be obtained, from a initial point. This paper expounds the method's principle, computational program and flow diagram. This method keeps the advantages of steepest descent method and overcomes the phenomenon of zigzag, and thus increases convergence rate.