Loading [MathJax]/jax/element/mml/optable/BasicLatin.js

重要定理及其证明

重要定理及其证明

一、数列收敛的Cauchy收敛准则

数列{an}的充要条件是:对任意的ε>0,存在NN+,当m,n>N时,有

|aman|<ε

证明:

必要性:设lim

于是任意的\varepsilon >0,存在N\in {{N}^{+}},当m,n>N时,有\left| {{a}_{m}}-A \right|<\frac{\varepsilon }{2},\left| {{a}_{n}}-A \right|<\frac{\varepsilon }{2}

于是\left| {{a}_{m}}-{{a}_{n}} \right|<\left| {{a}_{m}}-A \right|+\left| {{a}_{n}}-A \right|<\varepsilon

充分性:先证明\{{{a}_{n}}\}有界

由题可知:存在{{\varepsilon }_{0}}=1,存在{{N}_{0}}\in {{N}^{+}},当n>{{N}_{0}}时,有\left| {{x}_{n}}-{{x}_{{{N}_{0}}+1}} \right|<1

M=\max \{\left| {{x}_{1}} \right|,\left| {{x}_{2}} \right|,\cdots ,\left| {{x}_{{{N}_{0}}}} \right|,\left| {{x}_{{{N}_{0}}+1}} \right|+1

于是对一切n\in N,有\left| {{a}_{n}} \right|\le M

由致密性定理,在\{{{x}_{n}}\}中必有收敛子列:\underset{k\to +\infty }{\mathop{\lim }}\,{{a}_{{{n}_{k}}}}=\xi

有条件,对任意的\varepsilon >0,存在N\in {{N}^{+}},当m,n>N时,有\left| {{a}_{m}}-{{a}_{n}} \right|<\varepsilon

{{a}_{m}}={{a}_{{{n}_{k}}}},其中k充分大,满足{{n}_{k}}>N,并且令k\to +\infty ,于是得到\left| {{a}_{n}}-\xi  \right|\le \frac{\varepsilon }{2}<\varepsilon

即数列\{{{a}_{n}}\}收敛

二、归结原则

f{{U}^{0}}({{x}_{0}};\delta ')上有定义,\underset{x\to {{x}_{0}}}{\mathop{\lim }}\,f(x)存在的充要条件是:对任何含于{{U}^{0}}({{x}_{0}};\delta ')且以{{x}_{0}}为极限的数列\{{{x}_{n}}\},极限\underset{n\to +\infty }{\mathop{\lim }}\,f({{x}_{n}})都存在且相等

证明:

必要性:设\underset{x\to {{x}_{0}}}{\mathop{\lim }}\,f(x)=A

则则任给\varepsilon >0,存在\delta >0,其中\delta <\delta ',当0<\left| x-{{x}_{0}} \right|<\delta 时,\left| f(x)-A \right|<\varepsilon

另一方面,设数列\{{{x}_{n}}\}\subset {{U}^{0}}({{x}_{0}};\delta ')\underset{n\to +\infty }{\mathop{\lim }}\,{{x}_{n}}={{x}_{0}}

则对上述\delta >0,存在N>0,使得当n>N时,有0<\left| {{x}_{n}}-{{x}_{0}} \right|<\delta

从而有\left| f({{x}_{n}})-A \right|<\varepsilon

\underset{n\to +\infty }{\mathop{\lim }}\,f({{x}_{n}})=A

充分性:设对任何数列\{{{x}_{n}}\}\subset {{U}^{0}}({{x}_{0}};\delta ')\underset{n\to +\infty }{\mathop{\lim }}\,{{x}_{n}}={{x}_{0}},有\underset{n\to +\infty }{\mathop{\lim }}\,f({{x}_{n}})=A

反证法:若x\to {{x}_{0}}f(x)不以A为极限

则存在{{\varepsilon }_{0}}>0,对任何\delta >0,存在x,当\left| x-{{x}_{0}} \right|<\delta 时,\left| f(x)-A \right|\ge {{\varepsilon }_{0}}

现依次取\delta =\delta ',\frac{\delta '}{2},\cdots ,\frac{\delta '}{n},\cdots ,则存在相应的点{{x}_{1}},{{x}_{2}},\cdots ,{{x}_{n}},\cdots ,使得

0<\left| {{x}_{n}}-{{x}_{0}} \right|<\frac{\delta '}{n}\left| f({{x}_{n}})-A \right|\ge {{\varepsilon }_{0}},n=1,2,\cdots

显然数列\{{{x}_{n}}\}\subset {{U}^{0}}({{x}_{0}};\delta ')\underset{n\to +\infty }{\mathop{\lim }}\,{{x}_{n}}={{x}_{0}},但\underset{n\to +\infty }{\mathop{\lim }}\,f({{x}_{n}})\ne A矛盾

于是必有\underset{x\to {{x}_{0}}}{\mathop{\lim }}\,f(x)=A

三、函数收敛的Cauchy准则

设函数f{{U}^{0}}({{x}_{0}};\delta ')上有定义,\underset{x\to {{x}_{0}}}{\mathop{\lim }}\,f(x)存在的充要条件是:任给\varepsilon >0,存在\delta >0,其中\delta <\delta ',使得对任何{{x}_{1}},{{x}_{2}}\in {{U}^{0}}({{x}_{0}};\delta ),有\left| f({{x}_{1}})-f({{x}_{2}}) \right|<\varepsilon

证明:

必要性:设\underset{x\to {{x}_{0}}}{\mathop{\lim }}\,f(x)=A

则任给\varepsilon >0,存在\delta >0,其中\delta <\delta ',使得对任何x',x''\in {{U}^{0}}({{x}_{0}};\delta ),有

\left| f({{x}_{1}})-A \right|<\frac{\varepsilon }{2},\left| f({{x}_{2}})-A \right|<\frac{\varepsilon }{2}

于是\left| f({{x}_{1}})-f({{x}_{2}}) \right|\le \left| f({{x}_{1}})-A \right|+\left| f({{x}_{2}})-A \right|<\varepsilon

充分性:设数列\{{{x}_{n}}\}\subset {{U}^{0}}({{x}_{0}};\delta )\underset{n\to +\infty }{\mathop{\lim }}\,{{x}_{n}}={{x}_{0}}

按假设,任给\varepsilon >0,存在\delta >0,其中\delta <\delta ',使得对任何{{x}_{1}},{{x}_{2}}\in {{U}^{0}}({{x}_{0}};\delta ),有

\left| f({{x}_{1}})-f({{x}_{2}}) \right|<\varepsilon

由于\underset{n\to +\infty }{\mathop{\lim }}\,{{x}_{n}}={{x}_{0}},则对上述\delta >0,存在N>0,使得当n>N时,有{{x}_{n}},{{x}_{m}}\in {{U}^{0}}({{x}_{0}};\delta )

从而\left| f({{x}_{m}})-f({{x}_{n}}) \right|<\varepsilon

于是按照数列的Cauchy收敛准则,\{f({{x}_{n}})\}的极限存在,记为A,即\underset{n\to +\infty }{\mathop{\lim }}\,f({{x}_{n}})=A

对任意的x\in {{U}^{0}}({{x}_{0}};\delta ),当n>N时,有\left| f(x)-f({{x}_{n}}) \right|<\varepsilon

n\to +\infty ,则\left| f(x)-A \right|\le \varepsilon

于是\underset{n\to +\infty }{\mathop{\lim }}\,f(x)=A

四、一致收敛

函数f(x)定义在区间I上,试证f(x)I上一致收敛的充要条件为:对任何数列

\{x_{n}^{'}\},\{x_{n}^{''}\}\subset I,若\underset{n\to +\infty }{\mathop{\lim }}\,(x_{n}^{'}-x_{n}^{''})=0,则\underset{n\to +\infty }{\mathop{\lim }}\,[f(x_{n}^{'})-f(x_{n}^{''})]=0

证明:

必要性:若函数在I上一致收敛

则对任意的\varepsilon >0,存在\delta >0,对任意的x',x''\in I,\left| x'-x'' \right|<\delta ,有\left| f(x')-f(x'') \right|<\varepsilon

I上两个数列\{x_{n}^{'}\},\{x_{n}^{''}\},满足\underset{n\to +\infty }{\mathop{\lim }}\,(x_{n}^{'}-x_{n}^{''})=0

则对上述\delta >0,存在N>0,当n>N时,\left| x_{n}^{'}-x_{n}^{''} \right|<\delta

于是\left| f(x_{n}^{'})-f(x_{n}^{''}) \right|<\varepsilon

\underset{n\to +\infty }{\mathop{\lim }}\,[f(x_{n}^{'})-f(x_{n}^{''})]=0

充分性:对任何数列\{x_{n}^{'}\},\{x_{n}^{''}\}\subset I,若\underset{n\to +\infty }{\mathop{\lim }}\,(x_{n}^{'}-x_{n}^{''})=0,则\underset{n\to +\infty }{\mathop{\lim }}\,[f(x_{n}^{'})-f(x_{n}^{''})]=0

下证f(x)I上一致收敛

采用反证法:设f(x)I上不一致收敛,则

存在{{\varepsilon }_{0}}>0,对任意的\delta >0,满足\left| x'-x'' \right|<\delta ,但是\left| f(x')-f(x'') \right|\ge {{\varepsilon }_{0}}

{{\delta }_{1}}=1,存在x_{1}^{'},x_{1}^{''}\in I,\left| x_{1}^{'}-x_{1}^{''} \right|<1,有\left| f(x_{1}^{'})-f(x_{1}^{''}) \right|\ge {{\varepsilon }_{0}}

{{\delta }_{2}}=\frac{1}{2},存在x_{2}^{'},x_{2}^{''}\in I,\left| x_{2}^{'}-x_{2}^{''} \right|<\frac{1}{2},有\left| f(x_{2}^{'})-f(x_{2}^{''}) \right|\ge {{\varepsilon }_{0}}

\vdots

{{\delta }_{n}}=\frac{1}{n},存在x_{n}^{'},x_{n}^{''}\in I,\left| x_{n}^{'}-x_{n}^{''} \right|<\frac{1}{n},有\left| f(x_{n}^{'})-f(x_{n}^{''}) \right|\ge {{\varepsilon }_{0}}

\vdots

于是\underset{n\to +\infty }{\mathop{\lim }}\,(x_{n}^{'}-x_{n}^{''})=0,但\underset{n\to +\infty }{\mathop{\lim }}\,[f(x_{n}^{'})-f(x_{n}^{''})]\ne 0矛盾

于是f(x)I上一致收敛

 

五、函数列一致收敛的Cauchy准则

函数列\{{{f}_{n}}\}在数集D上一致收敛的充要条件是:对任意的\varepsilon >0,总存在N\in {{N}^{+}},使得当

m,n>N时,都有\left| {{f}_{m}}(x)-{{f}_{n}}(x) \right|<\varepsilon

证明:

必要性:设{{f}_{n}}(x)\overrightarrow{\to }f(x)(n\to +\infty )x\in D

即对任意的\varepsilon >0,总存在N\in {{N}^{+}},使得当n>N时,都有\left| {{f}_{n}}(x)-f(x) \right|<\varepsilon

于是当m,n>N时,都有\left| {{f}_{m}}(x)-{{f}_{n}}(x) \right|\le \left| {{f}_{m}}(x)-f(x) \right|+\left| {{f}_{n}}(x)-f(x) \right|<\varepsilon

充分性:若对任意的\varepsilon >0,总存在N\in {{N}^{+}},使得当m,n>N时,都有\left| {{f}_{m}}(x)-{{f}_{n}}(x) \right|<\varepsilon

由数列的Cauchy准则,\{{{f}_{n}}(x)\}D上任意一点都收敛,设\underset{n\to +\infty }{\mathop{\lim }}\,{{f}_{n}}(x)=f(x),x\in D

现固定已知中的n,令m\to +\infty

于是当n>N时,对一切x\in D,都有\left| {{f}_{n}}(x)-f(x) \right|\le \varepsilon

于是{{f}_{n}}(x)\overrightarrow{\to }f(x)(n\to +\infty )

六、可微的充分条件

若函数z=f(x,y)的偏导数在({{x}_{0}},{{y}_{0}})的某邻域上存在,且{{f}_{x}}{{f}_{y}}在点({{x}_{0}},{{y}_{0}})连续,则函数f在点({{x}_{0}},{{y}_{0}})可微。

证明:由于

\Delta z=f({{x}_{0}}+\Delta x,{{y}_{0}}+\Delta y)-f({{x}_{0}},{{y}_{0}})=[f({{x}_{0}}+\Delta x,{{y}_{0}}+\Delta y)-f({{x}_{0}},{{y}_{0}}+\Delta y)]+[f({{x}_{0}},{{y}_{0}}+\Delta y)-f({{x}_{0}},{{y}_{0}})]

Lagrange中值定理知:

\Delta z={{f}_{x}}({{x}_{0}}+{{\theta }_{1}}\Delta x,{{y}_{0}}+\Delta y)\Delta x+{{f}_{y}}({{x}_{0}},{{y}_{0}}+{{\theta }_{2}}\Delta x)\Delta y,0<{{\theta }_{1}},{{\theta }_{2}}<1

由于{{f}_{x}}{{f}_{y}}在点({{x}_{0}},{{y}_{0}})连续,因此有

{{f}_{x}}({{x}_{0}}+{{\theta }_{1}}\Delta x,{{y}_{0}}+\Delta y)={{f}_{x}}({{x}_{0}},{{y}_{0}})+\alpha

{{f}_{y}}({{x}_{0}},{{y}_{0}}+{{\theta }_{2}}\Delta x)={{f}_{y}}({{x}_{0}},{{y}_{0}})+\beta

其中当(\Delta x,\Delta y)\to (0,0)时,\alpha \to 0,\beta \to 0

于是\Delta z={{f}_{x}}({{x}_{0}},{{y}_{0}})\Delta x+{{f}_{y}}({{x}_{0}},{{y}_{0}})\Delta y+\alpha \Delta x+\beta \Delta y

有定义知f(x,y)在点({{x}_{0}},{{y}_{0}})可微

七、偏导数交换顺序

{{f}_{xy}}(x,y){{f}_{yx}}(x,y)都在点({{x}_{0}},{{y}_{0}})连续,则{{f}_{xy}}(x,y)={{f}_{yx}}(x,y)

证明:令

F(\Delta x,\Delta y)=f({{x}_{0}}+\Delta x,{{y}_{0}}+\Delta y)-f({{x}_{0}}+\Delta x,{{y}_{0}})-f({{x}_{0}},{{y}_{0}}+\Delta y)+f({{x}_{0}},{{y}_{0}})

\varphi (x)=f(x,{{y}_{0}}+\Delta y)-f(x,{{y}_{0}})

于是

F(\Delta x,\Delta y)=\varphi ({{x}_{0}}+\Delta x)-\varphi ({{x}_{0}})

由于函数f(x,y)存在关于x的偏导数,所以\varphi (x)可导

由一元函数中值定理知:

\varphi ({{x}_{0}}+\Delta x)-\varphi ({{x}_{0}})=\varphi '({{x}_{0}}+{{\theta }_{1}}\Delta x)\Delta x=\left[ {{f}_{x}}({{x}_{0}}+{{\theta }_{1}}\Delta x,{{y}_{0}}+\Delta y)-{{f}_{x}}({{x}_{0}}+{{\theta }_{1}}\Delta x,{{y}_{0}}) \right]\Delta x(0<{{\theta }_{1}}<1)

又由于{{f}_{x}}(x,y)存在关于y的偏导数,故对以y为自变量的函数{{f}_{x}}({{x}_{0}}+{{\theta }_{1}}\Delta x,y)应用一元微分中值定理,得

\varphi ({{x}_{0}}+\Delta x)-\varphi ({{x}_{0}})={{f}_{xy}}({{x}_{0}}+{{\theta }_{1}}\Delta x,{{y}_{0}}+{{\theta }_{2}}\Delta y)\Delta x\Delta y(0<{{\theta }_{1}},{{\theta }_{2}}<1)

于是

F(\Delta x,\Delta y)={{f}_{xy}}({{x}_{0}}+{{\theta }_{1}}\Delta x,{{y}_{0}}+{{\theta }_{2}}\Delta y)\Delta x\Delta y(0<{{\theta }_{1}},{{\theta }_{2}}<1)

如果令\psi (y)=f(x+\Delta x,y)-f({{x}_{0}},y)

F(\Delta x,\Delta y)=\psi ({{y}_{0}}+\Delta y)-\psi ({{y}_{0}})

同上可证F(\Delta x,\Delta y)={{f}_{yx}}({{x}_{0}}+{{\theta }_{3}}\Delta x,{{y}_{0}}+{{\theta }_{4}}\Delta y)\Delta x\Delta y(0<{{\theta }_{3}},{{\theta }_{4}}<1)

\Delta x,\Delta y不为0时,于是

{{f}_{xy}}({{x}_{0}}+{{\theta }_{1}}\Delta x,{{y}_{0}}+{{\theta }_{2}}\Delta y)={{f}_{xy}}({{x}_{0}}+{{\theta }_{3}}\Delta x,{{y}_{0}}+{{\theta }_{4}}\Delta y)

由于{{f}_{xy}}(x,y){{f}_{yx}}(x,y)都在点({{x}_{0}},{{y}_{0}})连续

故当\Delta x\to 0,\Delta y\to 0时,有上式两边极限存在且

{{f}_{xy}}({{x}_{0}},{{y}_{0}})={{f}_{xy}}({{x}_{0}},{{y}_{0}})

posted @   牙膏高露洁  阅读(542)  评论(0编辑  收藏  举报
编辑推荐:
· .NET 原生驾驭 AI 新基建实战系列:向量数据库的应用与畅想
· 从问题排查到源码分析:ActiveMQ消费端频繁日志刷屏的秘密
· 一次Java后端服务间歇性响应慢的问题排查记录
· dotnet 源代码生成器分析器入门
· ASP.NET Core 模型验证消息的本地化新姿势
阅读排行:
· 从零开始开发一个 MCP Server!
· ThreeJs-16智慧城市项目(重磅以及未来发展ai)
· .NET 原生驾驭 AI 新基建实战系列(一):向量数据库的应用与畅想
· Ai满嘴顺口溜,想考研?浪费我几个小时
· Browser-use 详细介绍&使用文档
点击右上角即可分享
微信分享提示