• 5959595


    证明:必要性.若存在$X$上的连续线性泛函$f$满足条件,则egin{align*}left| {sumlimits_{upsilon = 1}^k {{t_upsilon }{alpha _upsilon }} } ight| &= left| {sumlimits_{upsilon = 1}^k {{t_upsilon }fleft( {{x_upsilon }} ight)} } ight| = left| {fleft( {sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} } ight)} ight|\
    &le left| f ight| cdot left| {sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} } ight| le Mleft| {sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} } ight|
    end{align*}

    充分性.若对任意数,有$left| {sumlimits_{upsilon = 1}^k {{t_upsilon }{alpha _upsilon }} } ight| le Mleft| {sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} } ight|$

    令$M = spanleft{ {{x_1},{x_2}, cdots ,{x_k}} ight}$,则$M$为赋范线性空间$X$的子空间.对任意$sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} in M$,定义$M$上的线性泛函$f_0$:[{f_0}left( {sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} } ight) = sumlimits_{upsilon = 1}^k {{t_upsilon }{alpha _upsilon }} ]由于[left| {{f_0}left( {sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} } ight)} ight| = left| {sumlimits_{upsilon = 1}^k {{t_upsilon }{alpha _upsilon }} } ight| le Mleft| {sumlimits_{upsilon = 1}^k {{t_upsilon }{x_upsilon }} } ight|]则$left| {{f_0}} ight| le M$,即$f_0$为$M$上的连续线性泛函,于是由$f{Hahn-Banach}$泛函延拓定理知,存在$X$上的连续线性泛函$f$,使得当$x in M$时,有$fleft( x ight) = {f_0}left( x ight)$,且${left| f ight|_X} = {left| {{f_0}} ight|_M}$,所以充分性得证

  • 相关阅读:
    CNN结构:SPP-Net为CNNs添加空间尺度卷积-神经元层
    VR: AR和VR演进哲学
    OpenCV:使用 随机森林与GBDT
    VTK初始化New返回Null问题
    VTK嵌入MFC同步显示
    VTK:VTK嵌入MFC成功
    深度学习的技术困难
    编译Caffe-Win错误集锦
    Reducing the Dimensionality of Data with Neural Networks:神经网络用于降维
    VC维与DNN的Boundary
  • 原文地址:https://www.cnblogs.com/ly758241/p/3800786.html
Copyright © 2020-2023  润新知