最优化理论与方法lec6unconstrained课件_第1页
最优化理论与方法lec6unconstrained课件_第2页
最优化理论与方法lec6unconstrained课件_第3页
最优化理论与方法lec6unconstrained课件_第4页
最优化理论与方法lec6unconstrained课件_第5页
已阅读5页,还剩38页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

1、Models for trust-region methodsModels for trust-region methodsThe trust-region Newton method has proved to be highly effective in practice Line Search MethodsIntroductionDescent directionThe search direction often has the form:steepest descent method Newtons method quasi-Newton method an approximati

2、on to the Hessian Step length The process of determining the minimum point on a given line is called line search, which are really procedures for solving one-dimensional minimization problems The popular methods for resolving the line search problem include the Fibonacci and Golden section search, w

3、hich have a certain degree of theoretical elegance. The method determines the minimum value of a function f over a closed interval c1, c2. f is assumed unimodal, that is, it has a single relative minimum.单峰函数Unimodal function7 The minimum point of f is to be determined, at least approximately, by me

4、asuring the value of f at a certain number of points. How to successively select N measurement points so that we can determine the smallest possible region of uncertainty in which the minimum must lie.Fibonacci search8Fibonacci search9Golden section search10 If the number N of allowed measurement po

5、ints in a Fibonacci search is made to approach infinity, we obtain the golden section method.The interval of uncertainty at any point in the process has width:Step lengthInexact line search Two Stages : A bracketing phase finds an interval A bisection or interpolation phase computes a good step leng

6、th Termination conditionThe Wolfe conditionsArmijo condition Sufficient decrease condition The Wolfe conditionsThe Wolfe conditionsCurvature condition The Wolfe conditions The Wolfe conditionsThe strong Wolfe conditionsThe existence of the step lengthThe existence of the step length(1)(2)Combining (

7、1) and (2) givesThe strong Wolfe conditions hold in the same interval. The Goldstein conditionsSufficient decrease condition Introduced to control the step length from below The Goldstein conditionsConvergence of line search methodsConvergence of line search methodsZoutendijk Condition Convergence o

8、f line search methodswhile the Lipschitz condition implies thatSubstituting this inequality into the first Wolfe condition, we obtainConvergence of line search methodssumming allConvergence of line search methodsThis limit can be used in turn to derive global convergence results for line search algo

9、rithms. globally convergent Convergence of line search methodsSteepest descent methods Now we consider the ideal case, in which the objective function is quadratic and the line searches are exact.symmetric and positive definite Steepest descent methodsSteepest descent methods To quantify the rate of

10、 convergence, we introducethe weighted normThe bound of decreaseThe bound of decreaseThe rate-of-convergence behavior of the steepest descent method is essentially the same on general nonlinear objective functions. Quasi-Newton methodsQuasi-Newton methodsConvergence TheoremConvergence TheoremProof. We first show that Convergence TheoremFor the remainder of the

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论