Calculus-A2-homework all
本文最后更新于:1 年前
\(\mathbf{PDF}\) 文件下载请点击这里 微积分A(2)作业
\(\large\textcolor{blue}{微积分第一次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.2.25}\)
1
设 \(\|\cdot\|\) 是范数,证明 \(\mathbb{C}=\{\vec{x}\in\mathbb{R}^m\mid\|\vec{x}\|\leq 1\}\) 是闭凸集。
( \(\mathbb{C}\in\mathbb{R}^{m}\) 是凸集\(\ \Longleftrightarrow \ \forall\ \vec{x},\vec{y}\in\mathbb{C},\ \forall \ t\in(0,1),\ (1-t)\vec{x}+t\vec{y}\in\mathbb{C}\))
证明:先证明其为凸集,由范数的性质,满足三角不等式和正齐次性有 \[ \|\vec{x}\|,\|\vec{y}\|\leq 1\in\mathbb{C},\ \forall \ t\in(0,1),\ \|(1-t)\vec{x}+t\vec{y}\|\leq(1-t)\|\vec{x}\|+t\|\vec{y}\|\leq 1 \] 考虑 \(\mathbb{C}\) 的补集 \(\mathbb{A}=\{\vec{x}\in\mathbb{R}^m\mid\|\vec{x}\|> 1\}\),对于该集合中任意一点 \(\vec{a}\in\mathbb{A}\),令 \(\|\vec{a}\|=\lambda>1\) \[ \forall \ \vec{a}\in\mathbb{A},\ \exists \ \delta=\dfrac{\lambda-1}{2},\ s.t.\forall \ \|\vec{x}-\vec{a}\|<\delta,\|\vec{x}\|\geq\|\vec{a}\|-\|\vec{a}-\vec{x}\|=\|\vec{a}\|-\|\vec{x}-\vec{a}\|>1 \] 从而 \(\mathbb{A}\) 满足任意一点总存在该点领域全在 \(\mathbb{A}\) 中,\(\mathbb{A}\) 为开集,\(\mathbb{C}\) 为 \(\mathbb{A}\) 的补集,则 \(\mathbb{C}\) 为闭集
再结合其为凸集,得 \(\mathbb{C}\) 为闭凸集 \(□\)
2
设 \(\|\cdot\|\) 满足正定、正齐次性,证明:\(\|\cdot\|\) 是范数 \(\Longleftrightarrow\) \(\mathbb{C}=\{\vec{x}\in\mathbb{R}^m\mid\|\vec{x}\|\leq 1\}\) 是凸集。
证明:由第一题结论可得,\(\|\cdot\|\) 是范数 \(\Longrightarrow\) \(\mathbb{C}=\{\vec{x}\in\mathbb{R}^m\mid\|\vec{x}\|\leq 1\}\) 是凸集。
又若 \(\mathbb{C}=\{\vec{x}\in\mathbb{R}^m\mid\|\vec{x}\|\leq 1\}\) 是凸集,即 \(\forall\ \vec{x},\vec{y}\in\mathbb{C},\ \forall \ t\in(0,1),\ (1-t)\vec{x}+t\vec{y}\in\mathbb{C}\)。若
\(\|\vec{x}\|\) 或 \(\|\vec{y}\|\) 中有一者为 \(0\),由 \(\|\cdot\|\) 的正定性,可得 \(\vec{x}\) 或 \(\vec{y}\) 中一者为 \(\boldsymbol 0\in \mathbb{R}^m\),此时三角不等式
\(\|\vec{x}\|+\|\vec{y}\|\geq\|\vec{x}+\vec{y}\|\) 成立。则对两者范数均大于 \(0\) 的情况, \(\forall\ \vec{x},\vec{y}\),令\(\|\vec{x}\|=l_1>0\)
\(\|\vec{y}\|=l_2>0\),单位化有 \(\vec{n}_1=\dfrac{\vec{x}}{l_1},\vec{n}_2=\dfrac{\vec{y}}{l_2}\),满足 \(\vec{n}_1,\vec{n}_2\in\mathbb{C}\), 由凸集性质 \[ \begin{gathered} \because \forall \ t\in(0,1),\|(1-t)\vec{n}_1+t\vec{n}_2\|\in\mathbb{C},\vec{n}_1=\dfrac{\vec{x}}{l_1},\vec{n}_2=\dfrac{\vec{y}}{l_2}\\ \therefore \forall\ t\in (0,1),\|\dfrac{1-t}{l_1}\cdot \vec{x}+\dfrac{t}{l_2}\cdot \vec{y}\|\leq 1\ \therefore \mbox{set}\ \ \ t=\dfrac{l_2}{l_1+l_2},\|\dfrac{\vec{x}+\vec{y}}{l_1+l_2}\|\leq1 \end{gathered} \] 由正齐次性,\(\|\dfrac{\vec{x}+\vec{y}}{l_1+l_2}\|=\dfrac{1}{l_1+l_2}\|\vec{x}+\vec{y}\|\leq 1\),则有 \(\|\vec{x}+\vec{y}\|\leq l_1+l_2=\|\vec{x}\|+\|\vec{y}\|\)
则 \(\mathbb{C}=\{\vec{x}\in\mathbb{R}^m\mid\|\vec{x}\|\leq 1\}\) 是凸集 \(\Longrightarrow \|\cdot\|\) 为范数,两者等价 \(□\)
3
证明 \(\|\vec{x}\|_{\infty}=\underset{1\leq i\leq m}{\max}|x^{i}|\) 是范数。
证明:显然正定性满足;且 \(\|\lambda\vec{x}\|_{\infty}=\underset{1\leq i\leq m}{\max}|\lambda x^{i}|=|\lambda|\underset{1\leq i\leq m}{\max}|x^{i}|=|\lambda|\|\vec{x}\|_{\infty}\) 满足正齐次性,
有 \(\vec{x}=(x^1,\cdots,x^m)^{T},\vec{y}=(y^1,\cdots,y^m)^{T}\),设两者最大值为 \(|x^{max}|,|y^{max}|\) \[ \begin{gathered} \|\vec{x}+\vec{y}\|_{\infty}=(x^1+y^1,x^2+y^2,\cdots,x^m+y^m)\leq(|x^{1}|+|y^1|,\cdots,|x^{m}|+|y^m|)\\ \leq(|x^{max}|+|y_1|,\cdots,|x^{max}|+|y^m|)=|x^{max}|+(|y^1|,\cdots,|y^m|)\leq|x^{max}|+|y^{max}|\\ \end{gathered} \] 即有 \(\forall \ \vec{x},\vec{y},\ \|\vec{x}+\vec{y}\|_{\infty}\leq\|\vec{x}\|_{\infty}+\|\vec{y}\|_{\infty}\),满足三角不等式,从而 \(\|\vec{x}\|_{\infty}\) 是(无穷)范数 \(□\)
4
设 \(p>1\),证明 \(\|\vec{x}\|_{p}=(\displaystyle \sum_{i=1}^m|x^i|^p)^{\frac{1}{p}}\) 是范数。
证明:显然 \((\displaystyle \sum_{i=1}^m|x^i|^p)^{\frac{1}{p}}\geq 0\) 满足正定性,若乘以 \(\lambda\) 倍,有 \[ \|\lambda\vec{x}\|_{p}=\|\vec{x}\|_{\infty}=(\displaystyle \sum_{i=1}^m|\lambda x^i|^p)^{\frac{1}{p}}=\|\vec{x}\|_{\infty}=(\lambda^p \displaystyle \sum_{i=1}^m|x^i|^p)^{\frac{1}{p}}=\lambda \cdot(\displaystyle \sum_{i=1}^m|x^i|^p)^{\frac{1}{p}}=\lambda\|\vec{x}\|_{\infty} \] 从而 \(\|\vec{x}\|_p\) 满足正齐次性,对三角不等式,先证明两个不等式
引理一:\(\mbox{Young}\) 不等式:设 \(a,b>0,q,p>1,\dfrac{1}{p}+\dfrac{1}{q}=1\),有以下不等式成立 \[ ab\leq \dfrac{a^p}{p}+\dfrac{b^q}{q} \] 构造函数 \(f(x)=\dfrac{1}{p}x^{p}+\dfrac{1}{q}-x\ (x>0)\),求导有 \(f'(x)=x^{p-1}-1\),当 \(x\in(0,1)\) 时,
\(f'(x)<0\),\(f(x)\) 单调递减;当 \(x\in (1,+\infty)\) 时,\(f'(x)>0\),\(f(x)\) 单调递增。从而
\(f(x)\geq f(1)=\dfrac{1}{p}+\dfrac{1}{q}-1=0\)。令 \(x=\dfrac{a}{b^{\frac{q}{p}}}\) 有 \(\dfrac{1}{p}\dfrac{a^p}{b^{q}}+\dfrac{1}{q}\geq\dfrac{a}{b^{\frac{q}{p}}}\),两边乘以 \(b^q\),代入
\(\dfrac{1}{p}=1-\dfrac{1}{q}\),则有 \(\dfrac{a^p}{p}+\dfrac{b^q}{q}\geq a\cdot b^{q-\frac{q}{p}}=ab\)
引理二:\(\mbox{Hölder}\) 不等式:设 \(a_k,b_k>0,p,q\geq1,k=1,2,\cdots,n,\dfrac{1}{p}+\dfrac{1}{q}=1\),有以下不等式成立 \[ \displaystyle \sum_{k=1}^na_kb_k\leq(\sum_{k=1}^na_k^p)^{\frac{1}{p}}\cdot(\sum_{k=1}^nb_k^q)^{\frac{1}{q}} \] 考虑两者相除运用 \(\mbox{Young}\) 不等式即可得证 \[ \begin{gathered} \dfrac{\displaystyle \sum_{i=1}^na_kb_k}{\displaystyle (\sum_{k=1}^na_k^p)^{\frac{1}{p}}\cdot(\sum_{k=1}^nb_k^q)^{\frac{1}{q}}}=\displaystyle\large \sum_{k=1}^n((\normalsize\dfrac{a_k^p}{\normalsize \displaystyle \sum_{k=1}^na_k^p})^{\frac{1}{p}}\cdot(\normalsize \dfrac{b_k^q}{\normalsize \displaystyle \sum_{k=1}^nb_k^q})^{\frac{1}{q}})\normalsize \\ \leq\dfrac{1}{p}\sum_{i=1}^n\dfrac{a_k^p}{\normalsize \displaystyle \sum_{k=1}^na_k^p}+\dfrac{1}{q}\sum_{i=1}^n\dfrac{b_k^q}{\normalsize \displaystyle \sum_{k=1}^nb_k^q}=\dfrac{1}{p}+\dfrac{1}{q}=1 \end{gathered} \]
拆开和的 \(p-\)范数有 \(\displaystyle \sum_{i=1}^{n}\left(x^{i}+y^{i}\right)^{p}=\sum_{i=1}^{n} x^{i}\left(x^{i}+y^{i}\right)^{p-1}+\sum_{i=1}^{n} y^{i}\left(x^{i}+y^{i}\right)^{p-1}\)
令 \(\dfrac{1}{p}+\dfrac{1}{q}=1\),利用 \(\mbox{Hölder}\) 不等式放缩两者有 \[ \begin{gathered} \sum_{i=1}^{n} x^{i}(x^{i}+y^{i})^{p-1} \leq(\sum_{i=1}^{n} (x^{i})^{p})^{\frac{1}{p}}(\sum_{i=1}^{n}(x^{i}+y^{i})^{(p-1) q})^{\frac{1}{q}}=(\sum_{i=1}^{n} (x^{i})^{p})^{\frac{1}{p}}\cdot (\sum_{i=1}^{n}(x^{i}+y^{i})^{p})^{\frac{1}{q}}\\ \sum_{i=1}^{n} y^{i}(x^{i}+y^{i})^{p-1} \leq(\sum_{i=1}^{n} (y^{i})^{p})^{\frac{1}{p}}(\sum_{i=1}^{n}(x^{i}+y^{i})^{(p-1) q})^{\frac{1}{q}}=(\sum_{i=1}^{n} (y^{i})^{p})^{\frac{1}{p}}\cdot (\sum_{i=1}^{n}(x^{i}+y^{i})^{p})^{\frac{1}{q}} \end{gathered} \] 从而求和得到 \(\displaystyle \sum_{i=1}^n(x^i+y^i)^p\leq \Large(\normalsize (\displaystyle \sum_{i=1}^n(x^i)^p)^{\frac{1}{p}}+(\sum_{i=1}^n(y^i)^p)^{\frac{1}{p}}\Large)\normalsize \cdot (\sum_{i=1}^{n}(x^{i}+y^{i})^{p})^{\frac{1}{q}}\),移项即得 \[ (\displaystyle \sum_{i=1}^n(x^i)^p)^{\frac{1}{p}}+(\sum_{i=1}^n(y^i)^p)^{\frac{1}{p}}\geq (\sum_{i=1}^n(x^i+y^i)^p)^{1-\frac{1}{q}}=(\sum_{i=1}^n(x^i+y^i)^p)^{\frac{1}{p}} \] \(\|\vec{x}\|_{p}=(\displaystyle \sum_{i=1}^m|x^i|^p)^{\frac{1}{p}}\) 满足正定、正齐次、三角不等式,从而为范数 \(□\)
5
设 \(m>1\),证明 \(\mathbb{A}=\{\vec{x}\in\mathbb{R}^m\mid \|x\|=1\}\) 是道路连通集。
证明:\(\forall \ \vec{x},\vec{y}\in E\),令 \(f(t)=\dfrac{(1-t)\vec{x}+t\vec{y}}{\|(1-t)\vec{x}+t\vec{y}\|}\),对 \(\vec{x}=\vec{y}\) 时,显然 若 \(。\vec{x}\neq\vec{y}\),\(f(0)=\vec{x}\),
\(f(1)=\vec{y}\),先证明该函数映射连续性。有线性函数 \(g_1:[0,1]\longrightarrow \mathbb{R}^m,t \longmapsto (1-t)\vec{x}+t\vec{y}\) 连
续,考虑函数 \(g_2:[0,1]\longrightarrow [0,1]t\longmapsto\|(1-t)\vec{x}+t\vec{y}\|\),由范数的正齐次性和三角不等式得 \[ \forall \ t_1,t_2\in[0,1],|g_2(t_1)-g_2(t_2)|=|\|\vec{x}+t_1(\vec{y}-\vec{x})\|-\|\vec{x}+t_2(\vec{y}-\vec{x})\||\leq|t_1-t_2|\|\vec{y}-\vec{x}\| \] 而 \(\|\vec{y}-\vec{x}\|\leq \|\vec{y}\|+\|-\vec{x}\|=2\) 为有限量,则有 \[ \forall \ \epsilon>0,\exists\ \delta<\dfrac{\epsilon}{2},s.t.\forall \ t_1,t_2\in[0,1],|t_1-t_2|<\delta,|g(t_1)-g(t_2)|\leq|t_1-t_2|\|\vec{y}-\vec{x}\|<\epsilon \] 从而 \(g_2\) 为连续函数,同时由范数的正定性,\(\|(1-t)\vec{x}+t\vec{y}\|\geq0\),若 \(\|(1-t)\vec{x}+t\vec{y}\|=0\)
由正定性 \((1-t)\vec{x}+t\vec{y}=0\) 得 \(\vec{x}=-\vec{y}\) ,只要 \(\vec{x}\neq -\vec{y}\Longrightarrow \|(1-t)\vec{x}+t\vec{y}\|\neq 0\)
令 \(g_3:x\in(0,1]\longmapsto \dfrac{1}{x}\in[1,+\infty)\) 为连续函数,复合函数 \(f(t)=(g_3\circ g_2)\cdot g_1\) 也连续
而 \(\|f(t)\|=\dfrac{\|(1-t)\vec{x}+t\vec{y}\|}{\|(1-t)\vec{x}+t\vec{y}\|}=1\),则对 \(t\in[0,1]\),都有 \(f(t)\in \mathbb{A}\),满足道路连通集条件
而针对特殊情况(两向量恰好反向), \(\vec{x}=-\vec{y}\),可选取 \(\vec{z}\neq \vec{x},\vec{z}\neq\vec{y}\),令 \(\vec{x}\) 到 \(\vec{z}\) 的连续映射为
\(f_1(t)\),\(\vec{z}\) 到 \(\vec{y}\) 的连续映射为 \(f_2(t)\),可以构造连续映射 \[ f(t)=\begin{cases}f_1(2t),t\in[0,\dfrac{1}{2})\\f_2(2t-1),t\in[\dfrac{1}{2},1]\end{cases} \] 也满足道路连续性的条件,从而 \[ \forall \ \vec{x},\vec{y}\in \mathbb{A},\ \exists\ 连续\ f(t),f(0)=\vec{x},f(1)=\vec{y},s.t.\forall\ t\in[0,1],f(t)\in\mathbb{A} \] 即 \(\mathbb{A}\) 为道路连通集 \(□\)
6
设 \(m>1\),证明 \(\mathbb{D}=\{A\mid A\ 是\ m\ 阶方阵,\det A>0\}\) 是道路连通集。
证明:首先证明单位阵 \(I\) 到 \(\mathbb{D}\) 中任意元素 \(A\in\mathbb{D},\det A>0\) 是连通的,对单位阵 \(I\),总存在一系列初等变换有 \[ E_1E_2\cdots E_nIE_1'E_2'\cdots E_n'=A \] 若 \(E_i\) 是倍乘变换,倍乘系数为 \(k\neq0\),则对 \(\forall\ B\in \mathbb{D}\),作连续映射 \[ f_i(t):\mathcal{GL}^{+}(m) \longrightarrow \mathcal{GL}^{+}(m)\quad B\longmapsto(1+(k-1)t)E_iB \] 若 \(E_i\) 是倍加变换,倍加系数为 \(m>0\),则对 \(\forall \ C\in\mathbb{D}\),作连续映射 \[ g_i(t):\mathcal{GL}^{+}(m) \longrightarrow \mathcal{GL}^{+}(m)\quad C\longmapsto(1+(k-1)t)E_iC \] 若 \(E_i\) 是换行 \(/\) 换列变换并对其中一行 $/ $ 列乘以 \(-1\) (保证行列式 \(>0\)),对改变的两行 \(/\) 两列有连续复合映射 \[ (A,B)\stackrel{E_1}{\longrightarrow} (A+B,B)\stackrel{E_2}{\longrightarrow}(A+B,-A)\stackrel{E_3}{\longrightarrow}(B,-A)\stackrel{E_4(-1)}{\longrightarrow}(B,A) \] 综上,对所有初等矩阵 \(E_i\),均存在连续映射 \(f_{E_{i}}(t)\) 从 \(I\) 到 \(E_i\),使得 \(I\) 到 \(A\in\mathbb{D}\) 是连通的
对 \(\forall \ A_1,A_2\in \mathbb{D}\),对 \(A_2\) 实行上述步骤得到连续映射 \(I\longmapsto A_2,F(t):\displaystyle \prod_{i=1}^n f_{E_i}\),则 \(A_1^{-1}\circ F\) 满足\(A_1\longmapsto A_2\) 连续映射,且由于 \(f_{i}(t)\) 和 \(g_{i}(t)\) 都映射到 \(\mathbb{D}\) 中,\(\forall \ t\in[0,1],A_1^{-1}\circ F(t)\in\mathbb{D}\)
,即 \(\mathbb{D}\) 道路连通 \(□\)
7
设 \(f:\mathbb{R}^m\diagdown\{0\}\longmapsto \mathbb{R}\),且 \(f\) 连续,满足 \(\forall \ t>0,\vec{x}\neq0 ,f(t\vec{x})=f(\vec{x})\),证明 \(f\) 有最大值和最小值。
证明:令单位“实心球”集合 \(\mathbb{S}=\{\vec{y}\mid \|\vec{y}\|\leq 1\}\) 为有界闭集,对 \(\forall\ \vec{y}\in\mathbb{R}^m\diagdown\{0\}\),若 \(\|\vec{y}\|\leq1\),
则 \(\vec{y}\in\mathbb{S}\),否则 \(\|\vec{y}\|>1\),令 \(\vec{z}=\dfrac{\vec{y}}{\|\vec{y}\|}\),其范数 \(\|\vec{z}\|=1 ,\vec{z}\in\mathbb{S}\),\(f(\vec{y})=f(\|\vec{y}\|\cdot\vec{z})=f(\vec{z})\)
定义映射 \(g:\mathbb{S}\longmapsto\mathbb{R},g(\vec{x})=f(\vec{x})\),由知 \(f\) 和 \(g\) 的值域相同,且 \(g\) 为有界闭集上的连续映射
从而 \(f\) 和 \(g\) 都有最大值和最小值 \(□\)
8
设 \(f(x,y,z)=\dfrac{x+2y+3z}{x^2+y^2+z^2+1}\),证明 \(f\) 有正的最大值和负的最小值。
证明:令三维欧式空间向量 \(\vec{r}=(x,y,z)\),有 \[ f(x,y,z)=f(\vec{r})=\dfrac{\vec{r}\cdot(1,2,3)}{\|\vec{r}\|^2+1}=\dfrac{\|\vec{r}\|\cdot\sqrt{14}\cos \theta}{\|\vec{r}\|^2+1} \] 由均值不等式 \(\dfrac{\|\vec{r}\|}{\|\vec{r}\|^2+1}\leq\dfrac{1}{2}\),以及 \(\cos \theta\in[-1,1]\),得 \(f(\vec{r})\in[-\dfrac{\sqrt{14}}{2},\dfrac{\sqrt{14}}{2}]\) 有界
而 \(f(x,y,z)\) 为连续函数,从而有最值,且当 \(x,y,z>0\) 时 \(f(x,y,z)>0\);当 \(x,y,z<0\) 时
\(f(x,y,z)<0\),则 \(f\) 有正的最大值和负的最小值 \(□\)
\(\large\textcolor{blue}{微积分第二次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.3.4}\)
1
1、(连续性与偏连续性)设 \(f:[a,b]\times [c,d]\rightarrow \mathbb{R}\)。
- 如果 \(f\) 是连续函数, 那么是否成立: 对任意 \(y \in[c, d], f(\cdot, y)\) : \([a, b] \rightarrow \mathbb{R}\) 连续; 对任意 \(x \in[a, b]\)
\(f(x, \cdot):[c, d] \rightarrow \mathbb{R}\) 连续?
- 如果对任意 \(y \in[c, d], \quad f(\cdot, y):[a, b] \rightarrow \mathbb{R}\) 连续; 对任意 \(x \in[a, b]\), \(f(x, \cdot):[c, d] \rightarrow \mathbb{R}\) 连续, 那么 \(f\) 是否为连续函数? 考虑 \[ f(x, y)= \begin{cases}\dfrac{x y}{x^{2}+y^{2}}, & x y \neq 0 \\ 0, & x y=0\end{cases} \]
- 设 \(x_{0} \in[a, b], y_{0} \in[c, d], f\) 满足: \(f\left(\cdot, y_{0}\right):[a, b] \rightarrow \mathbb{R}\) 连续; 并且对 任意 \(\varepsilon>0\), 存在
\(\delta(\varepsilon)>0\) 使得对任意 \(x \in[a, b]\) 以及任意 \(y \in[c, d]\), 只要 \(\left|y-y_{0}\right|<\delta(\varepsilon)\), 就有
\(\left|f(x, y)-f\left(x, y_{0}\right)\right|<\varepsilon\) 。证明 \(f\) 在 \(\left(x_{0}, y_{0}\right)\) 连续。
- 设 \(f(x, y)\) 分别对每个变量 \(x\) 和 \(y\) 是连续的,且对 \(y\) 是单调不减的, 证明 \(f\) 关于 \((x, y)\) 是连续
的。
解:\((a)\) 是。由连续函数定义,对 \(\mathbf{x}^*=(x^1,x^2)\) \[ \forall \ \mathbf{x}^*\in [a,b]\times [c,d],\forall \ \epsilon>0,\exists\ \delta>0,s.t.\forall\ \|x-x^*\|\leq\delta,\|f(\mathbf{x})-f(\mathbf{x^*})\|<\epsilon \] 则固定 \(x^1\) 时,仅取 \(\|x-x^*\|\leq \delta\) 中 \(x\) 第一分量为 \(x^1\) 的点集,由上述定义知 \[ \begin{gathered} \forall \ 固定\ x^1\in [a,b],\forall \ x^2\in [c,d],\mathbf{x}^*=(x^1,x^2),s.t.\forall \ \epsilon>0,\exists\ \delta>0,\\ \forall\ \|x-x^*\|\leq\delta,\|f(\mathbf{x})-f(\mathbf{x^*})\|<\epsilon \end{gathered} \] 从而固定 \(x^1\) 时,\(f(x^1,x^2)\) 随 \(x^2\) 连续;由对称性可得,固定 \(x^2\) 时,\(f(x^1,x^2)\) 随 \(x^1\) 连续
\((b)\) 否。考虑函数 \(f(x,y)=\begin{cases}\dfrac{xy}{x^2+y^2},& xy\neq 0\\0& xy=0\end{cases}\quad x,y\in[-1,1]\),由对称性,固定 \(x\) 有
若 \(x=0\),则 \(f(x,y)=0\) 在 \(y\in[-1,1]\) 区间上均满足 \(f(0,y)\) 随 \(y\) 连续;
若 \(x\neq 0\),则函数 \(g_1(y)=x^2+y^2\neq 0\),\(g_2(z)=\dfrac{1}{z}(z\neq 0)\), \(g_3(y)=xy\) 三者均连续
则 \(f(x,y)=(g_2\circ g_1)\cdot g_3\) 复合函数连续,即 \(\forall\ x\in [-1,1],f(x,\cdot):[c,d]\rightarrow \mathbb{R}\) 连续
同理 \(\forall\ y\in [-1,1],f(\cdot,y):[c,d]\rightarrow \mathbb{R}\) 连续,而考虑 \(f(x,y)\) 在原点处的连续性 \[ \lim\limits_{(x,y)\to (0,0)}\dfrac{xy}{x^2+y^2}\Large |\normalsize _{(x,y)\to (x,kx)}=\dfrac{k}{1+k^2}\in[-\dfrac{1}{2},\dfrac{1}{2}] \] \(k\) 值变化时,\(f(x,y)=\dfrac{xy}{x^2+y^2}\) 在 \((0,0)\) 处的极限不存在,从而 \(f\) 不一定为连续函数
\((c)\) 证明:由 \(f\) 满足 \(f(\cdot ,y_0):[a,b]\to \mathbb{R}\) 连续,翻译成 $-$ 语言有 \[ \forall \ y_0\in[c,d],\forall \ \epsilon >0 ,\exists\ \delta_1 >0,|x-x_0|<\delta_1(\epsilon ),|f(x_0,y_0)-f(x,y_0)|<\epsilon \] 由 $$ 的任意性,则 \(\forall \ \epsilon '>0\),上述取 \(\epsilon=\dfrac{1}{2}\epsilon'\),另一条件为 \[ \forall\ \epsilon >0,\exists \ \delta_2>0,\forall \ x\in[a,b],y\in[c,d],|y-y_0|<\delta_2(\epsilon),|f(x,y)-f(x,y_0)|<\epsilon \] 由后者条件中 \(x\) 的任意性,取 $=' $,则考虑欧几里得空间中的 \(2-\)范数有 \[ \forall \ \epsilon' >0,\exists \ \delta =\min(\delta_1,\delta_2)>0,\forall \ x\in[a,b],y\in[c,d]\\ \|x-x^*\|=\sqrt{(x-x_0)^2+(y-y_0)^2}<\delta\Longrightarrow |x-x_0|<\delta_1 ,|y-y_0|<\delta_2 \\|f(x,y)-f(x_0,y_0)|\leq |f(x,y)-f(x,y_0)|+|f(x_0,y_0)-f(x,y_0)|<\dfrac{1}{2}\epsilon'+\dfrac{1}{2}\epsilon '=\epsilon' \] 则 \(f(x,y)\) 在 \((x_0,y_0)\) 处连续
\((d)\) 证明:由于 \(f(x,y)\) 对 \(y\) 单调不减,则有对 \(\forall \ x\in[a,b]\), 不妨只考虑 \(y>y_0\) 的部分
\(\delta(y_0)>0,0<f(x,y_0+\delta(y_0))-f(x,y_0)<\epsilon\),取 \(y=y_0+\delta(y_0)\),考虑 \[ |f(x,y)-f(x_0,y_0)|=|f(x,y_0+\delta(y_0))-f(x_0,y_0)|\\\leq |f(x,y_0+\delta(y_0))-f(x,y_0)|+|f(x,y_0)-f(x_0,y_0)|<\epsilon+\epsilon=2\epsilon \] 其中后者由于 \(f(x,y)\) 对 \(x\) 连续可以保证,从而 \(f(x,y)\) 在 \((x_0,y_0)\) 处连续
2
设 \(U \subseteq \mathbb{R}^{2}\) 是开集, \((a, b) \in U, f\) 在 \(U \backslash\{(a, b)\}\) 上有定义, 且满足极限 \[ A=\lim _{(x, y) \rightarrow(a, b)} f(x, y), \quad B=\lim _{y \rightarrow b} \lim _{x \rightarrow a} f(x, y) \] 都存在。证明 \(A=B\) 。由此知道, 如果 \[ \lim _{y \rightarrow b} \lim _{x \rightarrow a} f(x, y) \neq \lim _{x \rightarrow a} \lim _{y \rightarrow b} f(x, y) \] 则极限 \(\lim\limits _{(x, y) \rightarrow(a, b)} f(x, y)\) 不存在。
证明:由于对 \(\forall \ 0<|y-b|<\delta_0\),\(\varphi(y)=\lim\limits_{x\to a}f(x,y)\) 存在,即 \[ \forall \ \epsilon>0,y\in U,\exists\ \delta_{1}(y),s.t. |x-a|<\delta_{1}(y),|f(x,y)-\varphi(y)|<\epsilon \] 而又有 \(\lim\limits_{y\to b}\varphi(y)=B\) 存在,即 \(\forall \ \epsilon>0,\exists\ \delta_2 >0,|y-b|<\delta_2,|\varphi(y)-B|<\epsilon\)
而 \(\lim\limits_{(x,y)\to (a,b)}f(x,y)=A\),即有 \[ \forall\ \epsilon >0,\exists \ \delta_0>0,\forall \ x,y\in U,\|\boldsymbol v^*-\boldsymbol v\|<\delta_0(\epsilon),|f(x,y)-A|<\epsilon \] 则取 \(|y-b|<\min(\delta_0,\delta_2),|x-b|<\min(\delta_1(y),\delta_0)\),总有 \(|f(x,y)-B|<\epsilon+\epsilon=2\epsilon\)
从而 \(|A-B|<\epsilon+2\epsilon=3\epsilon\),则 \(A=B\),且若 \(\displaystyle \lim _{y \rightarrow b} \lim _{x \rightarrow a} f(x, y) \neq \lim _{x \rightarrow a} \lim _{y \rightarrow b} f(x, y)\) 两者极限存在
反证法,若极限 \(\lim\limits _{(x, y) \rightarrow(a, b)} f(x, y)\) 存在,则由上述证明以及对称性知,三者相等,与假设矛盾,从而
极限 \(\lim\limits _{(x, y) \rightarrow(a, b)} f(x, y)\) 不存在
3
设
极限 \(\displaystyle H(y)=\lim _{x \rightarrow a} f(x, y)\) 对任意 \(y \neq b\) 存在;
极限 \(\displaystyle G(x)=\lim _{y \rightarrow b} f(x, y)\) 对任意 \(x\) 存在,且存在 \(\delta_{0}>0\) 使得:对任意 \(\varepsilon>0\),存在
\(\delta_{2}(\varepsilon)>0\) 对任意 \(0<|y-b|<\delta_{2}(\varepsilon)\) 任意 \(0<|x-a|<\delta_{0}\) 有 \(|f(x, y)-G(x)|<\varepsilon\),则
极限 \(\displaystyle \lim _{x \rightarrow a} \lim _{y \rightarrow b} f(x, y), \lim _{y \rightarrow b} \lim _{x \rightarrow a} f(x, y)\) 和 \(\displaystyle \lim _{x \rightarrow a, y \rightarrow b} f(x, y)\) 都存在, 且
\[ \lim _{x \rightarrow a, y \rightarrow b} f(x, y)=\lim _{y \rightarrow b} \lim _{x \rightarrow a} f(x, y)=\lim _{x \rightarrow a} \lim _{y \rightarrow b} f(x, y) \]
证明:(Cauchy准则 )\(\forall \ 0<|y_1-b|<\delta_2(\epsilon),0<|y_2-b|<\delta_2(\epsilon)\),对 \(|x-a|<\delta_{0}\),都有 \[ |f(x,y_1)-G(x)|<\dfrac{\epsilon}{2},|f(x,y_2)-G(x)|<\dfrac{\epsilon}{2}\ \therefore |f(x,y_1)-f(x,y_2)|<\dfrac{\epsilon}{2}\cdot 2=\epsilon \] 则令 \(x\to a\),由于 \(H(y)=\lim\limits_{x\to a}f(x,y)\) 存在,则有 \(|H(y_1)-H(y_2)|<\epsilon\),即 \[ \forall \ \epsilon >0,\exists \ \delta>0,\forall \ y_i(i=1,2),0<|y_i-b|<\delta ,|H(y_1)-H(y_2)|<\epsilon \] 由柯西收敛准则得到 \(\lim\limits_{y\to b}\lim\limits_{x\to a}f(x,y)\) 极限存在,由 \(0<|y_0-b|<\delta_2(\epsilon)\) \[ |G(x_1)-G(x_2)|\leq|G(x_1)-f(x_1,y_0)|+|G(x_2)-f(x_2,y_0)|+|f(x_1,y_0)-f(x_2,y_0)| \] 只需要 \(0<|x-a|<\delta_0(\dfrac{1}{3}\epsilon)\),就有 \(|G(x_1)-f(x_1,y_0)|<\dfrac{1}{3}\epsilon,|G(x_2)-f(x_2,y_0)|<\dfrac{1}{3}\epsilon\)
而由于 \(\displaystyle H(y)=\lim _{x \rightarrow a} f(x, y)\) 对 \(y\neq b\) 时均存在,则添加条件 \(0<|x-a|<\delta_1(\dfrac{1}{3}\epsilon)\)
就有 \(|f(x_1,y_0)-f(x_2,y_0)|<\dfrac{1}{3}\epsilon\),则满足下列条件 \[ 0<|x-a|<\min(\delta_{0}(\dfrac{1}{3}\epsilon),\delta_{1}(\dfrac{1}{3}\epsilon)),0<|y-b|<\delta_{1}(\epsilon),|G(x_1)-G(x_2)|<\dfrac{1}{3}\epsilon\cdot 3=\epsilon \] 由柯西收敛准则得 \(\lim\limits_{x\to a}\lim\limits_{y\to b}f(x,y)\) 存在,由作业第2题知,三种极限存在且相等
4
设
\[ f(x,y)=\begin{cases}x\sin (\dfrac{1}{y}),& y\neq 0\\ 0,&y=0 \end{cases} \]
讨论极限 \(\lim\limits_{(x,y)\to (0,0)}f(x,y),\lim\limits_{y\to 0}\lim\limits_{x\to 0}f(x,y),\lim\limits_{x\to 0}\lim\limits_{y\to 0}f(x,y)\)
解:显然 \(\lim\limits_{x\to 0}f(x,y)=0\),则有 \(\lim\limits_{y\to 0}\lim\limits_{x\to 0}f(x,y)=0\),而对 \(y\to 0\) 的性质有 \[ \forall \ |x|>0,f(x,y)=\begin{cases}x\sin(\dfrac{1}{y}),&y\neq 0\\0,&y=0\end{cases} \] 对 \(y_{n}=\dfrac{1}{2n\pi +\dfrac{\pi}{2}},y_{m}=\dfrac{1}{2m\pi -\dfrac{\pi}{2}}\),总有 \(f(x,y_n)-f(x,y_m)=x-(-x)=2x\)
由柯西收敛准则,\(\forall \ x\neq 0,\lim\limits_{y\to 0}x\sin(\dfrac{1}{y})\) 不存在,从而 \(\lim\limits_{x\to 0}\lim\limits_{y\to 0}f(x,y)\) 极限不存在
而对 \(\forall \ \epsilon >0,\exists\ \delta=\epsilon,\sqrt{x^2+y^2}<\delta,|f(x,y)-0|\leq |x||\sin(\dfrac{1}{y})|\leq |x|<\delta=\epsilon\)
从而 \(\lim\limits_{(x,y)\to (0,0)}f(x,y)\) 极限存在,且极限为 \(0\)
5
判断以下极限的存在性; 若存在, 求出极限值。
\[ \begin{aligned} &(1) \lim _{(x, y) \rightarrow(0,0)} \frac{\arcsin \left(x^2+y^2\right)}{x^2+y^2}\\ &(3) \lim _{(x, y) \rightarrow(0,0)}\left(x^2+y^2\right) \mathrm{e}^{x-y}\\ &(5) \lim _{(x, y) \rightarrow(0,0)} \frac{x+y}{|x|+|y|}\\ &(7) \lim _{(x, y) \rightarrow \infty} \frac{x+y}{x^2+x y+y^2}\\ &(9) \lim _{(x, y) \rightarrow \infty}\left(\frac{|x y|}{x^2+x y+y^2}\right)^{x^2} \end{aligned} \]
解:\((1)\) 令 \(z=x^2+y^2\) , \(\lim\limits_{z\to 0}\dfrac{\arcsin(z)}{z}=1\),而当 \(x,y\to 0\) 时, \(z=x^2+y^2\neq 0\) 则 \[ \displaystyle \lim _{(x, y) \rightarrow(0,0)} \frac{\arcsin \left(x^{2}+y^{2}\right)}{x^{2}+y^{2}}=0 \] \((3)\) 由极限的连续性,且 \(\lim\limits_{(x,y)\to (0,0)}x^2+y^2=0,\lim\limits_{(x,y)\to (0,0)}e^{x-y}=0\) 则极限存在,且极限为 \(0\)
\((5)\) 只需取 \(k>0\),\(\lim\limits_{(x,kx)\to (0,0)}\dfrac{x+y}{|x|+|y|}=\dfrac{(1+k)x}{(1+k)|x|}\) 从而极限不存在
\((7)\) 放缩有 \(|\dfrac{x+y}{x^2+xy+y^2}|\leq \dfrac{|x+y|}{|x^{2}+y^2|-|xy|}\leq \dfrac{|x+y|}{|2xy|-|xy|}\leq\dfrac{|x|+|y|}{|xy|}\leq \dfrac{1}{|x|}+\dfrac{1}{|y|}\)
从而 \(\displaystyle \lim _{(x, y) \rightarrow \infty} \frac{x+y}{x^{2}+x y+y^{2}}=0\)
\((9)\) 若取 \(y=x\),有 \(\lim\limits_{(x,x)\to \infty}\dfrac{|xy|}{x^2+xy+y^2}=\pm\dfrac{1}{3}\),此时 \(\lim\limits_{x\to \infty}(\pm\dfrac{1}{3})^{x^2}=0\)
又若取 \(y=-x\),有 \(\lim\limits_{(x,x)\to \infty}\dfrac{|xy|}{x^2+xy+y^2}=\pm1\),此时 \(\lim\limits_{x\to \infty}(\pm\dfrac{1}{3})^{x^2}\neq 0\)
从而 \(\displaystyle \lim _{(x, y) \rightarrow \infty}\left(\frac{|x y|}{x^{2}+x y+y^{2}}\right)^{x^{2}}\) 极限不存在
6
设 \(\Omega \subseteq \mathbb{R}^{m}\) 是一个非空集合。
证明函数 \(\displaystyle f: \mathbb{R}^{m} \rightarrow \mathbb{R}, f(\mathbf{x})=\inf _{\mathbf{y} \in \Omega}\|\mathbf{y}-\mathbf{x}\|\), 是一个连续函数。
若 \(\Omega\) 是闭集, 证明对任意 \(\mathbf{x}\) 不在 \(\Omega\) 中, 存在 \(\mathbf{y}^{*} \in \Omega\) 使得 \(f(\mathbf{x})=\left\|\mathbf{y}^{*}-\mathbf{x}\right\|\), 并 且 \(\mathbf{y}^{*} \in \partial \Omega\) (即 \(\mathbf{y}^{*}\) 的任意邻域中都有不属于 \(\Omega\) 的点)。
若 \(\Omega_{1}, \Omega_{2}\) 是闭集且其中至少一个有界, 证明存在 \(\mathbf{x}_{k}^{*} \in \Omega_{k}\) 使得
\[ \left\|\mathbf{x}_{1}^{*}-\mathbf{x}_{2}^{*}\right\|=\inf _{\mathbf{x}_{1} \in \Omega_{1}, \mathbf{x}_{2} \in \Omega_{2}}\left\|\mathbf{x}_{1}-\mathbf{x}_{2}\right\| . \] (d) 举例说明上述 “闭”、“有界”的条件是不能或缺的。
证明:\((a)\) \(\forall \ \epsilon>0,\exists\ \delta>0,||\mathbf{x}-\mathbf{x}_0||<\delta\),范数三角不等式 \(\|\mathbf{y}-\mathbf{x}\|\leq \|\mathbf{y}-\mathbf{x_0}\|+\|\mathbf{x}_0-\mathbf{x}\|\)
左侧放缩 \(\displaystyle \inf _{\mathbf{y} \in \Omega}\|\mathbf{y}-\mathbf{x}\|\leq\|\mathbf{y}-\mathbf{x}\|\leq \|\mathbf{y}-\mathbf{x}_0\|+\|\mathbf{x}_0-\mathbf{x}\|\) 对所有的 \(\mathbf{y}\) 成立,则取特定 \(\mathbf{y}\) 得 \[ \displaystyle \inf _{\mathbf{y} \in \Omega}\|\mathbf{y}-\mathbf{x}\|\leq \inf _{\mathbf{y} \in \Omega}\|\mathbf{y}-\mathbf{x}_0\|+\|\mathbf{x}_0-\mathbf{x}\| \] 同理交换 \(\mathbf{x}\) 和 \(\mathbf{x}_0\) 得到 \(|\displaystyle \inf _{\mathbf{y} \in \Omega}\|\mathbf{y}-\mathbf{x}\|-\displaystyle \inf _{\mathbf{y} \in \Omega}\|\mathbf{y}-\mathbf{x}_0\||\leq \|\mathbf{x}-\mathbf{x_0}\|\),则取 $=\(,就有\)$ >0, =,||-_0||<,|f()-f(_0)|= $$ 则 \(\displaystyle f(\mathbf{x})=\inf _{\mathbf{y} \in \Omega}\|\mathbf{y}-\mathbf{x}\|\) 连续
\((b)\) 由三角不等式 \(\|\mathbf{y^*}-\mathbf{y_n}\|+\|\mathbf{y_n}-\mathbf{x}\|\geq \|\mathbf{y^*-\mathbf{x}}\|\) 若 \(f(\mathbf{x})\) 中 \(\mathbf{y}\) 的取值为 \(\Omega\) 的内点,则
\(\forall \ \epsilon_0>\epsilon>0,\exists \ \|\mathbf{y^*}-\mathbf{y}_n\|<\epsilon\),从而可以找到其领域中的 \(\mathbf{y}_n\) 并尽量使三者共线,从而矛盾
从而 \(\mathbf{y}^*\in \partial \Omega\)
\((c)\) 假设 \(\Omega_2\) 为有界集合,固定在 \(\Omega_1\) 中一点,由在有界闭集上的连续函数有界有 \[ \forall \ \mathbf{x}_1\in\Omega_{1},\exists \ \mathbf{x}_2^*\in \Omega_{2},\|\mathbf{x}_1-\mathbf{x}_2^*\|=\inf_{\mathbf{x_2}\in \Omega_2}\|\mathbf{x_2}-\mathbf{x_1}\| \] 则取所有 \(\mathbf{x}_1\) 中范数最小的,则存在 \(\displaystyle \left\|\mathbf{x}_{1}^{*}-\mathbf{x}_{2}^{*}\right\|=\inf _{\mathbf{x}_{1} \in \Omega_{1}, \mathbf{x}_{2} \in \Omega_{2}}\left\|\mathbf{x}_{1}-\mathbf{x}_{2}\right\| .\)
\((d)\) 取 \(\Omega:x^2+y^2<1\),范数定义为二维平面中的欧式距离,则 \(\Omega\) 与 \((2,0)\) 就没有最小值
而两个无界区域之间有可能没有范数最小值,如 \(\{(x,y)\big |y=\tan (x),0\leq x<\dfrac{\pi}{2}\}\)
和 \(\{(x,y)\big |x=\dfrac{\pi}{2}\}\),两者的欧式距离没有最小值(无限趋于 \(0\) )
\(\large\textcolor{blue}{微积分第三次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.\Large\pi}\)
教材习题1.4.4
求下列函数的全微分 \[ (1)u=\sin \dfrac{1}{\sqrt{x^2+y^2+z^2}}\ \ \mbox{at}\ \ (\dfrac{\sqrt{2}}{2}.\dfrac{1}{2},-\dfrac{1}{2}) \quad (2)\arccos e^{xy} \]
解:\((1)\) \(du=\dfrac{\partial u}{\partial x}dx+\dfrac{\partial u}{\partial y}dy+\dfrac{\partial u}{\partial z}dz\),代入有 \[ \begin{gathered} du=\cos\dfrac{1}{\sqrt{x^2+y^2+z^2}}\cdot (-\dfrac{1}{2}(x^2+y^2+z^2)^{-\frac{3}{2}})(2xdx+2ydy+2zdz)\\ =-\cos 1\cdot (xdx+ydy+zdz)=-\cos 1\cdot (\dfrac{\sqrt{2}}{2}dx+\dfrac{1}{2}dy-\dfrac{1}{2}dz) \end{gathered} \] \((2)\) 由 \((\arccos x)'=-\dfrac{1}{\sqrt{1-x^2}}\) \[ du=\dfrac{\partial u}{\partial x}dx+\dfrac{\partial u}{\partial y}dy=-\dfrac{1}{\sqrt{1-e^{2xy}}}(ye^{xy}dx+xe^{xy}dy)=-\dfrac{1}{\sqrt{e^{-2xy}-1}}(ydx+xdy) \]
教材习题1.4.6
已知扇形中心角 \(\alpha=60^\circ\),半径 \(R=20\ \mbox{cm}\),当 \(\alpha\) 增加 \(1^\circ\) 时,为使扇形面积保持不变,其半径的增加量 \(\Delta R\) 近似等于多少?
解:由面积公式 \(S=\dfrac{1}{2}R^{2}\alpha\),要求 \(dS=\dfrac{1}{2}R^{2}d\alpha+R\alpha dR=0\) 取 \(dR=\Delta R,d\alpha =\Delta \alpha\)
解得 \(\Delta R=-\dfrac{R}{2}\dfrac{\Delta \alpha}{\alpha}=-\dfrac{1}{6}\ \mbox{cm}\)
\(\dfrac{1}{6}\approx0.16666667\),与直接计算 \(|\Delta R|=0.16461185\ \mbox{cm}\) 相对误差为 \(1.2\%\)
教材习题1.4.11(4)
求下列函数在点 \(P_{0}\) 处沿方向 \(l\) 的方向导数。
\((4)\) \(z=\ln(x_1+x_2+\cdots+x_n),P_n=(1,0,\cdots,0),l=(1,1,\cdots,1)\)
解:由于方向导数针对单位向量,先单位化 \(l_{\small \mbox{unit}}=\dfrac{1}{\sqrt{n}}(1,1,\cdots,1)\) 则由方向导数定义 \[ \partial _{\vec{l}}\large z\normalsize=\dfrac{d}{dt}\ln(1+n\cdot \dfrac{t}{\sqrt{n}})\big |_{t=0}=\sqrt{n} \]
教材习题1.4.12(2)
求下列数量场的梯度。\((2)\) \(u(x,y,z)=\dfrac{xyz}{x+y+z}\)
解:代入标准正交基 \(x,y,z\) 下的梯度 \[ \nabla u=\dfrac{\partial u}{\partial x}\hat{x}+\dfrac{\partial u}{\partial y}\hat{y}+\dfrac{\partial u}{\partial z}\hat{z}=\dfrac{yz(y+z)}{(x+y+z)^2}\hat{x}+\dfrac{xz(x+z)}{(x+y+z)^2}\hat{y}+\dfrac{xy(x+y)}{(x+y+z)^2}\hat{z} \]
教材习题1.4.15
证明下列函数满足相应的等式。
\((3)\) \(\begin{cases}u=e^{x}\cos y\\v=e^{x}\sin y\end{cases}\) 满足 \(\mbox{Cauchy-Riemann}\) 条件 \(\begin{cases}\dfrac{\partial u}{\partial x}=\dfrac{\partial v}{\partial y}\\\dfrac{\partial u}{\partial y}=-\dfrac{\partial v}{\partial x}\end{cases}\),且分别满足 \(\Delta u=0\)
\((4)\) \(n>2\),\(u=(\sqrt{x_1^{2}+x_2^2+\cdots+x_n^2})^{2-n}\) 满足 \(\dfrac{\partial ^2u}{\partial x_{1}^{2}}+\dfrac{\partial ^2u}{\partial x_{2}^{2}}+\cdots+\dfrac{\partial ^2u}{\partial x_{n}^{2}}=0\)
证明:\((3)\) 代入有 \(\dfrac{\partial u}{\partial x}=e^{x}\cos y=\dfrac{\partial v}{\partial y}=e^{x}cos y,\dfrac{\partial u}{\partial y}=-e^{x}\sin y=-\dfrac{\partial v}{\partial x}\)
则若 \(\dfrac{\partial^2 f}{\partial x^{2}}+\dfrac{\partial^2 f}{\partial y^{2}}=0\),\(f(u,v)=f(e^{x}\cos y,e^{x}\sin y)\)代入计算两次偏导数有 \[ \begin{gathered} \dfrac{\partial f}{\partial x}=\dfrac{\partial f}{\partial u}e^{x}\cos y+\dfrac{\partial f}{\partial v}e^{x}\sin y=u\dfrac{\partial f}{\partial u}+v\dfrac{\partial f}{\partial v}\\ \dfrac{\partial f}{\partial y}=-\dfrac{\partial f}{\partial u}e^{x}\sin y+\dfrac{\partial f}{\partial v}e^{x}\cos y=-v\dfrac{\partial f}{\partial u}+u\dfrac{\partial f}{\partial v} \end{gathered} \]
\[ \begin{gathered} \dfrac{\partial ^2f}{\partial x^2}=u\dfrac{\partial f}{\partial u}+u^2\dfrac{\partial^2 f}{\partial u^2}+v\dfrac{\partial f}{\partial v}+v^2\dfrac{\partial^2 f}{\partial v^2}\\ \dfrac{\partial ^2f}{\partial y^2}=-u\dfrac{\partial f}{\partial u}+v^2\dfrac{\partial^2 f}{\partial u^2}-v\dfrac{\partial f}{\partial v}+u^2\dfrac{\partial^2 f}{\partial v^2} \end{gathered} \]
从而 \(\dfrac{\partial^2 f}{\partial x^{2}}+\dfrac{\partial^2 f}{\partial y^{2}}=0\Longrightarrow (u^2+v^2)(\dfrac{\partial^2 f}{\partial u^{2}}+\dfrac{\partial^2 f}{\partial v^{2}})=0\Longrightarrow \dfrac{\partial^2 f}{\partial u^{2}}+\dfrac{\partial^2 f}{\partial v^{2}}=0\)
\((4)\) 对 \(\forall \ i\in[1,n]\),有 \(\dfrac{\partial u}{\partial x_i}=\dfrac{(1-\dfrac{n}{2})\cdot 2x_{i}}{(x_1^2+\cdots+x_n^2)^{\frac{n}{2}}}=\dfrac{(2-n)\cdot x_{i}}{(x_1^2+\cdots+x_n^2)^{\frac{n}{2}}}\)
则二阶导 \(\dfrac{\partial^2 u}{\partial x_i^2}=\dfrac{2-n}{(x_1^2+\cdots+x_n^2)^{\frac{n}{2}}}+\dfrac{(n-2)\cdot \dfrac{n}{2}x_i\cdot 2x_i}{(x_1^2+\cdots+x_n^2)^{\frac{n}{2}+1}}\),则对所有二阶导求和 \[ \sum_{i=1}^{n}\dfrac{\partial^2 u}{\partial x_i^2}=(x_1^2+\cdots+x_n^2)^{-\frac{n}{2}}((2-n)\cdot n+(n-2)\cdot n)=0 \]
教材习题1.5.3(3)
求下列复合函数的偏导数 \(\dfrac{\partial z}{\partial x},\dfrac{\partial z}{\partial y}\)(已知 \(f\) 为可微函数)
\((3)\) \(z=f(x^2-y^2,e^{xy})\)
解:令 \(u=x^2-y^2,v=e^{xy}\) 则代入计算偏导数 \[ \dfrac{\partial z}{\partial x}=\dfrac{\partial z}{\partial u}\cdot 2x+\dfrac{\partial z}{\partial v}\cdot ye^{xy},\dfrac{\partial z}{\partial y}=\dfrac{\partial z}{\partial u}\cdot (-2y)+\dfrac{\partial z}{\partial v}\cdot xe^{xy} \]
教材习题1.5.5
已知函数 \(u=f(x,y)\),其中 \(x=r\cos \theta,y=r\sin \theta\), \(f\) 可微,证明 \[ (\dfrac{\partial u}{\partial r})^2+(\dfrac{1}{r}\dfrac{\partial u}{\partial \theta})^2=(\dfrac{\partial u}{\partial x})^2+(\dfrac{\partial u}{\partial y})^2 \]
证明:对 \(u=f(x,y)=f(r\cos \theta,r\sin \theta)\) 对 \(r,\theta\) 分别求偏导有 \[ \dfrac{\partial u}{\partial r}=\dfrac{\partial u}{\partial x}\cdot \cos \theta+\dfrac{\partial u}{\partial y}\cdot \sin \theta,\dfrac{\partial u}{\partial \theta}=\dfrac{\partial u}{\partial x}\cdot (-r\sin \theta)+\dfrac{\partial u}{\partial y}\cdot(r\cos \theta) \] 则计算 \((\dfrac{\partial u}{\partial r})^2+(\dfrac{1}{r}\dfrac{\partial u}{\partial \theta})^2=(\dfrac{\partial u}{\partial x}\cdot \cos \theta+\dfrac{\partial u}{\partial y}\cdot \sin \theta)^{2}+(\dfrac{\partial u}{\partial x}\cdot (-\sin \theta)+\dfrac{\partial u}{\partial y}\cdot(\cos \theta))^2\)
\(=(\dfrac{\partial u}{\partial x})^2+(\dfrac{\partial u}{\partial y})^2\) 从而 \((\dfrac{\partial u}{\partial r})^2+(\dfrac{1}{r}\dfrac{\partial u}{\partial \theta})^2=(\dfrac{\partial u}{\partial x})^2+(\dfrac{\partial u}{\partial y})^2\)
教材习题1.5.7
设 \(f\in C^{2}(\mathbb{R}^2)\) 满足 \(\mbox{Laplace}\) 方程 \((\dfrac{\partial f}{\partial x})^2+(\dfrac{\partial f}{\partial y})^2=0\),证明: \[ u(x,y)=f(\dfrac{x}{x^2+y^2},\dfrac{y}{x^2+y^2}) \] 也满足 \(\mbox{Laplace}\) 方程。
证明:(使用平面极坐标下拉普拉斯算子简化计算)\(\Delta f_{\mbox{polar}}=\dfrac{\partial ^2f}{\partial r^2}+\dfrac{1}{r}\dfrac{\partial f}{\partial r}+\dfrac{1}{r^2}\dfrac{\partial ^2f}{\partial \theta^2}\)
则对于满足直角坐标系下 \(\mbox{Laplace}\) 方程 \(\Delta f_{\mbox{rectan}}=0=\Delta f_{\mbox{polar}}\),转换 \(u(x,y)=f(\dfrac{1}{r},\theta)\)
而 \(\dfrac{\partial u}{\partial (\frac{1}{r})}=-r^2\dfrac{\partial u}{\partial r}\),\(\dfrac{\partial^2 u}{\partial (\frac{1}{r})^2}=(-r^2)(-2r\dfrac{\partial u}{\partial r}-r^2\dfrac{\partial^2 u}{\partial r^2})=r^{3}(2\dfrac{\partial u}{\partial r}+r\dfrac{\partial^2 u}{\partial r^2})\)
则代入以 \((\dfrac{1}{r},\theta)\) 为度量的极坐标拉普拉斯算子判定有 \[ \begin{gathered} \Delta f_{\mbox{polar'}}=\dfrac{\partial ^2f}{\partial (\frac{1}{r})^2}+\dfrac{1}{\frac{1}{r}}\dfrac{\partial f}{\partial (\frac{1}{r})}+\dfrac{1}{(\frac{1}{r})^2}\dfrac{\partial ^2f}{\partial \theta^2}\\=r^{3}(2\dfrac{\partial f}{\partial r}+r\dfrac{\partial^2 f}{\partial r^2})+r(-r^2\dfrac{\partial f}{\partial r})+r^{2}\dfrac{\partial ^2f}{\partial \theta^2}=r^{3}(\dfrac{\partial f}{\partial r}+r\dfrac{\partial^2 f}{\partial r^2})+r^2\dfrac{\partial ^2f}{\partial \theta^2}=r^{4}\Delta f_{\mbox{polar}} \end{gathered} \] 从而 \(\Delta f_{\mbox{polar'}}=0\) 即 \(f(\dfrac{1}{r},\theta)=f(\dfrac{x}{x^2+y^2},\dfrac{y}{x^2+y^2})\) 满足 \(\mbox{Laplace}\) 方程
讲义习题2.2.8
记 \(\mathbb{R}^{m} \times \mathbb{R}=\left\{\left(X^{1}, \ldots, X^{m}, X^{m+1}\right) \mid X^{k} \in \mathbb{R}\right\}, \mathbb{R}^{m}=\left\{\left(x^{1}, \ldots, x^{m}\right) \mid x^{k} \in \mathbb{R}\right\}\)
\(N=(0, \ldots, 0,2) \in \mathbb{R}^{m} \times \mathbb{R}, S^{m}=\left\{\left(X^{1}, \ldots, X^{m}, X^{m+1}\right) \mid\left(X^{1}\right)^{2}+\cdots+\right.\) \(\left.\left(X^{m}\right)^{2}+\left(X^{m+1}-1\right)^{2}=1\right\}\) 。定义球极投影,对 \(\left(x^{1}, \ldots, x^{m}\right) \in \mathbb{R}^{m}\), 连接 \(\left(x^{1}, \ldots, x^{m}, 0\right)\) 与 \(N\) 的直线交 \(S^{m}\) 于 \(\left(X^{1}, \ldots, X^{m}, X^{m+1}\right)\) 。证明映射 \[ \left(X^{1}\left(x^{1}, \ldots, x^{m}\right), \ldots, X^{m}\left(x^{1}, \ldots, x^{m}\right), X^{m+1}\left(x^{1}, \ldots, x^{m}\right)\right) \] 是可微映射, 并求它的 \(\mbox{Jacobi}\) 矩阵。
解:连接 \(N=(0,\cdots,0,2)\) 与 \(\mathbf{x}=(x^1,\cdots,x^m,0)\) 的直线参数方程为 \[ \vec{l}=(0,\cdots ,0,2)+t(x^1,\cdots,x^m,-2)=(tx^1,tx^2,\cdots,tx^m,2-2t) \] 代入“高维球”方程得 \(t^{2} \displaystyle\sum_{i=1}^{n}\left(x^{i}\right)^{2}+(1-2 t)^{2}=1 \Longrightarrow t=\dfrac{4}{4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}}\)
则球极投影 \(f(\mathbf{x})=(\dfrac{4 x^{1}}{4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}}, \dfrac{4 x^{2}}{4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}}, \cdots, \dfrac{4 x^{m}}{4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}}, \dfrac{2 \displaystyle \sum_{4}\left(x^{i}\right)^{2}}{4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}})\)
这是由于每个分量对 \(x^i\) 均是初等函数,从而该球极投影映射为可微映射
雅可比矩阵定义 \(\displaystyle \left[\begin{array}{ccc} \dfrac{\partial X_{1}}{\partial x_{1}} & \cdots & \dfrac{\partial X_{1}}{\partial x_{m}} \\ \vdots & \ddots & \vdots \\ \dfrac{\partial X_{m+1}}{\partial x_{1}} & \cdots & \dfrac{\partial X_{m+1}}{\partial x_{m}} \end{array}\right]\),记该矩阵第 \(k\) 行第 \(l\) 列元素为 \(A_{k,l}\)
计算有 \(A_{k, l}= \begin{cases}4 \cdot \dfrac{4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}-2\left(x^{k}\right)^{2}}{\left(4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}\right)^{2}} & l=k \leq m \\ \dfrac{-8 x^{k} x^{j}}{\left(4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}\right)^{2}} & l \neq k \leq m \\ \dfrac{16 x^{l}}{\left(4+\displaystyle \sum_{i=0}^{n}\left(x^{i}\right)^{2}\right)^{2}} & k=m+1\end{cases}\) 为雅可比矩阵的每一项
讲义习题2.3.10
\(\mathbb{R}^{3}\) 中的球坐标系。其中极径 \(r\), 纬度 \(\theta\) 和经度 \(\varphi\) 如图所示。
\((a)\) 试写出用球坐标 \((r, \theta, \varphi)\) 表达的直角坐标 \((x(r, \theta, \varphi), y(r, \theta, \varphi), z(r, \theta, \varphi))\) 的表达式, 并证明它是可微映射, 求出该映射的 \(\mbox{Jacobi}\) 矩阵。该矩阵是否可逆? 为什么?
\((b)\) 记 \[ \mathbf{e}_{r}=\left(\begin{array}{l} \dfrac{\partial x}{\partial r} \\ \dfrac{\partial y}{\partial r} \\ \dfrac{\partial z}{\partial r} \end{array}\right), \quad \mathbf{e}_{\theta}=\left(\begin{array}{l} \dfrac{\partial x}{\partial \theta} \\ \dfrac{\partial y}{\partial \theta} \\ \dfrac{\partial z}{\partial \theta} \end{array}\right), \quad \mathbf{e}_{\varphi}=\left(\begin{array}{l} \dfrac{\partial x}{\partial \varphi} \\ \dfrac{\partial y}{\partial \varphi} \\ \dfrac{\partial z}{\partial \varphi} \end{array}\right) \] 求它们给出的度量矩阵。
\((c)\) 计算 \((x(r, \theta, \varphi), y(r, \theta, \varphi), z(r, \theta, \varphi))\) 的 \(\mbox{Jacobi}\) 矩阵的行列式, 并对结果给出几何解释。
\((d)\) 对可微函数 \(f: \mathbb{R}^{3} \rightarrow \mathbb{R}\), 求在球坐标系下 \(f\) 的梯度 \(\nabla f\) (用偏导数 \(\dfrac{\partial f}{\partial r}, \dfrac{\partial f}{\partial \theta}, \dfrac{\partial f}{\partial \varphi}\) 表示)。
\((e)\) 记 \(S^{2}=\left\{(x, y, z) \in \mathbb{R}^{3} \mid x^{2}+y^{2}+z^{2}=1\right\}\) 。如果 \(f(\theta, \varphi)\) 是 \(S^{2}\) 上的可微函数, 求 \(\nabla f\) (用偏导数 \(\dfrac{\partial f}{\partial \theta}, \dfrac{\partial f}{\partial \varphi}\) 表示)。
解:\((a)\) 由几何关系得 \(\left(\begin{array}{l} x \\ y\\z \end{array}\right)=\left(\begin{gathered} r \sin\theta\cos\varphi \\ r \sin \theta\sin\varphi\\ r\cos\theta \end{gathered}\right)\),显然该映射可微,对其求导得到雅克比矩阵 \[ \left(\begin{gathered}\mathrm{d}x\\\mathrm{d}y\\\mathrm{d}z\end{gathered}\right)= \begin{pmatrix} \sin\theta\cos\varphi&r\cos\theta\cos\varphi&-r\sin\theta\sin\varphi\\ \sin\theta\sin\varphi&r\cos\theta\sin\varphi&r\sin\theta\cos\varphi\\ \cos\theta&-r\sin\theta&0 \end{pmatrix}\left(\begin{gathered}\mathrm{d}r\\\mathrm{d}\theta\\\mathrm{d}\varphi\end{gathered}\right)=J\left(\begin{gathered}\mathrm{d}r\\\mathrm{d}\theta\\\mathrm{d}\varphi\end{gathered}\right) \] \((b)\) 代入计算有 \(\mathbf{e}_r=\left(\begin{gathered}\sin\theta\cos\varphi\\ \sin \theta\sin\varphi\\ \cos\theta\end{gathered}\right), \mathbf{e}_\theta=\left(\begin{gathered}r\cos\theta\cos\varphi\\ r\cos \theta\sin\varphi\\ -r\sin\theta\end{gathered}\right), \mathbf{e}_\varphi=\left(\begin{gathered}-r\sin\theta\sin\varphi\\ r\sin \theta\cos\varphi\\ 0\end{gathered}\right)\) \[ \begin{gathered}\langle\mathbf{e}_r,\mathbf{e}_r\rangle=1,\langle\mathbf{e}_r,\mathbf{e}_\theta\rangle=0,\langle\mathbf{e}_r,\mathbf{e}_\varphi\rangle=0,\langle\mathbf{e}_\theta,\mathbf{e}_\theta\rangle=r^2,\langle\mathbf{e}_\theta,\mathbf{e}_\varphi\rangle=0,\langle\mathbf{e}_\varphi,\mathbf{e}_\varphi\rangle=r^2\sin^2\theta\end{gathered} \] 则度量矩阵为 \(\left(\begin{array}{} 1 & 0 &0\\ 0 & r^{2}&0\\ 0&0&r^2\sin^2\theta \end{array}\right)\)
\((c)\) 由 \(\small(a)\) 中雅可比矩阵计算行列式 \[ \begin{gathered} \det J=r^2\sin ^3\theta\sin ^2\varphi+r^2\sin \theta\cos^2 \theta\sin ^2\varphi+r^2\sin \theta\cos^2 \theta\cos ^2\varphi+r^2\sin ^3\theta\cos ^2\varphi\\ =r^2\sin \theta\sin ^2\varphi+r^2\sin \theta\cos ^2\varphi=r^2\sin \theta \end{gathered} \] 该行列式代表三维坐标下微体积元与三维直角坐标系中微体积元之比
\((d)\) 由梯度表达式计算得 \[ \begin{gathered}\nabla f(r, \theta,\varphi)=\left(\mathbf{e}_{r}, \mathbf{e}_{\theta},\mathbf{e}_\varphi\right)\left(\begin{array}{} 1 & 0 &0\\ 0 & r^{2}&0\\ 0&0&r^2\sin^2\theta \end{array}\right)^{-1}\left(\begin{array}{c} \partial_{r} f \\ \partial_{\theta} f\\ \partial_{\varphi} f \end{array}\right)\\=\partial_{r} f \mathbf{e}_{r}+\frac{1}{r^{2}} \partial_{\theta} f \mathbf{e}_{\theta}+\frac{1}{r^2\sin^2\theta}\partial_\varphi f\mathbf{e}_\varphi=\dfrac{\partial f}{\partial r}\mathbf{e}_{r}+\dfrac{1}{r^2}\dfrac{\partial f}{\partial \theta}\mathbf{e}_{\theta}+\dfrac{1}{r^2\sin^2 \theta}\dfrac{\partial f}{\partial \varphi}\mathbf{e}_{\varphi}\end{gathered} \] \((e)\) 相当于是固定 \(r\equiv1\),则 \(\nabla f=\dfrac{\partial f}{\partial \theta}\mathbf{e}_{\theta}+\dfrac{1}{\sin^2 \theta}\dfrac{\partial f}{\partial \varphi}\mathbf{e}_{\varphi}\)
讲义习题2.3.11
环面上, 纬度 \(\theta\) 和经度 \(\varphi\) 如图所示。设环面轴径为 \(a\), 经线半径为 \(b\), \(0<b<a\) ,如图所示。
\((a)\) 试写出用环面坐标 \((\theta, \varphi)\) 表达的直角坐标 \((x(\theta, \varphi), y(\theta, \varphi), z(\theta, \varphi))\) 的表达式, 并证明它是可微映射, 求出该映射的 \(\mbox{Jacobi}\) 矩阵。该矩阵是否可逆? 为什么?
\((b)\) 记 \[ \mathbf{e}_{\theta}=\left(\begin{array}{l} \dfrac{\partial x}{\partial \theta} \\ \dfrac{\partial y}{\partial \theta} \\ \dfrac{\partial z}{\partial \theta} \end{array}\right), \quad \mathbf{e}_{\varphi}=\left(\begin{array}{l} \dfrac{\partial x}{\partial \varphi} \\ \dfrac{\partial y}{\partial \varphi} \\ \dfrac{\partial z}{\partial \varphi} \end{array}\right) \] 求它们给出的度量矩阵。
\((c)\) 如果 \(f(\theta, \varphi)\) 是环面上的可微函数, 求 \(\nabla f\) (用偏导数 \(\dfrac{\partial f}{\partial \theta}, \dfrac{\partial f}{\partial \varphi}\) 表示)。
解:\((a)\) 由几何关系得 \(\left(\begin{array}{l} x \\ y\\z \end{array}\right)=\left(\begin{gathered} (a+b\cos \theta)\cos \varphi \\ (a+b\cos \theta)\sin \varphi\\ b\sin\theta \end{gathered}\right)\),显然其可微,对其求导有 \[ \left(\begin{array}{l} \mathrm{d} x \\ \mathrm{~d} y \\ \mathrm{~d} z \end{array}\right)=\left(\begin{array}{cc} -b \sin \theta \cos \varphi & -(a+b \cos \theta) \sin \varphi \\ -b \sin \theta \sin \varphi & (a+b \cos \theta) \cos \varphi \\ b \cos \theta & 0 \end{array}\right)\left(\begin{array}{c} \mathrm{d} \theta \\ \mathrm{d} \varphi \end{array}\right)=J\left(\begin{array}{c} \mathrm{d} r \\ \mathrm{~d} \theta \\ \mathrm{d} \varphi \end{array}\right) \] \((b)\) 有 \(\mathbf{e}_\theta=\left(\begin{gathered}-b\sin \theta\cos \varphi\\ -b\sin \theta\sin \varphi\\ -b\cos\theta\end{gathered}\right), \mathbf{e}_\varphi=\left(\begin{gathered}-(a+b\cos \theta)\sin \varphi\\(a+b\cos \theta)\cos \varphi\\ 0\end{gathered}\right)\),计算度量矩阵 \[ \begin{gathered}\langle\mathbf{e}_{\theta},\mathbf{e}_{\theta}\rangle=b^2,\langle\mathbf{e}_{\varphi},\mathbf{e}_\theta\rangle=0,\langle\mathbf{e}_{\varphi},\mathbf{e}_\varphi\rangle=(a+b\cos \theta)^{2}\end{gathered} \] \((c)\) 由梯度表达式 \(\nabla f(\theta,\varphi)=\left( \mathbf{e}_{\theta},\mathbf{e}_\varphi\right)\left(\begin{array}{} b^2 & 0 \\ 0 & (a+b\cos \theta)^2 \end{array}\right)^{-1}\left(\begin{array}{c} \partial_{\theta} f\\ \partial_{\varphi} f \end{array}\right)\\\)
得到 \(\nabla f(\theta,\varphi)=\dfrac{1}{b^2}\dfrac{\partial f}{\partial \theta}\mathbf{e}_{\theta}+\dfrac{1}{(a+b\cos \theta)^{2}}\dfrac{\partial f}{\partial \varphi}\mathbf{e}_{\varphi}\)
讲义习题2.4.5
平面极坐标系下 \(\mbox{Laplace}\) 方程为 \(\dfrac{\partial^{2} u}{\partial r^{2}}+\dfrac{1}{r} \dfrac{\partial u}{\partial r}+\dfrac{1}{r^{2}} \dfrac{\partial^{2} u}{\partial \theta^{2}}=0\) 。
求分离变量形式的解 \(u(r, \theta)=X(r) Y(\theta)\) 。
解:代入分离变量形式有 \(\Delta u=Y(\theta)\left(\partial_{r, r}^{2} X(r)+\dfrac{1}{r} \partial_{r} X(r)\right)+\dfrac{X(r)}{r^{2}} \partial_{\theta, \theta} Y(\theta)=0\) 变形为 \[ \frac{1}{X(r)}\left(r^{2} \partial_{r, r}^{2} X(r)+r \partial_{r} X(r)\right)+\frac{1}{Y(\theta)} \partial_{\theta, \theta} Y(\theta)=0 \] 由于前两项与后一项无关联,令 \(r^{2} \partial_{r, r}^{2} X(r)+r \partial_{r} X(r)=C X(r),\partial_{\theta, \theta} Y(\theta)=-CY(\theta)\)
前者为欧拉方程,取 \(t=\ln r\),化简得 \(\dfrac{\partial^2 X(t)}{\partial t^2}=CX(t)\),则分类讨论有 \[ \begin{aligned} &X(r)=k_{1} \cosh (\sqrt{C} \ln r)+k_{2} \sinh (\sqrt{C} \ln r), & C>0 \\ &Y(\theta)=k_{3} \cos (\sqrt{C} \theta)+k_{4} \sin (\sqrt{C} \theta) ; \\ &X(r)=k_{1} \cos (\sqrt{-C} \ln r)+k_{2} \sin (\sqrt{-C} \ln r), &C<0 \\ &Y(\theta)=k_{3} \cosh (\sqrt{-C} \theta)+k_{4} \sinh (\sqrt{-C} \theta) ; \\ &X(r)=k_{1} \ln r+k_{2}, Y(\theta)=k_{3} \theta ; &C=0 \end{aligned} \] 从而合并所有解令 \(C>0\) 有 \[ u(r, \theta)=\left\{\begin{array}{l} \left(C_{1} r^{C}+C_{2} r^{-C}\right) \cos \left(C \theta+C_{3}\right) \\ \cos \left(C \ln r+C_{1}\right)\left(C_{2} \mathrm{e}^{C \theta}+C_{3} \mathrm{e}^{-C \theta}\right) \\ C_{1}\left(\ln r+C_{2}\right) \theta \end{array}\right. \]
讲义习题2.4.6
求直角坐标系下的 \(\mbox{Laplace}\) 方程 \(\Delta f=\dfrac{\partial^{2} f}{\partial x^{2}}+\dfrac{\partial^{2} f}{\partial y^{2}}+\dfrac{\partial^{2} f}{\partial z^{2}}=0\) 在空间球坐标系下表达式。
解:利用 \(\Delta=\nabla\cdot \nabla\),且由讲义 \(2.3.10\) 习题知 \[ \begin{gathered} \nabla f(r, \theta,\varphi)=\dfrac{\partial f}{\partial r}\mathbf{e}_{r}+\dfrac{1}{r^2}\dfrac{\partial f}{\partial \theta}\mathbf{e}_{\theta}+\dfrac{1}{r^2\sin^2 \theta}\dfrac{\partial f}{\partial \varphi}\mathbf{e}_{\varphi}\\ \mathbf{e}_r=\left(\begin{gathered}\sin\theta\cos\varphi\\ \sin \theta\sin\varphi\\ \cos\theta\end{gathered}\right), \mathbf{e}_\theta=\left(\begin{gathered}r\cos\theta\cos\varphi\\ r\cos \theta\sin\varphi\\ -r\sin\theta\end{gathered}\right), \mathbf{e}_\varphi=\left(\begin{gathered}-r\sin\theta\sin\varphi\\ r\sin \theta\cos\varphi\\ 0\end{gathered}\right) \end{gathered} \] 且不同坐标的偏导有,计算拉普拉斯算子得到 \[ \begin{gathered} \partial_{r} \mathbf{e}_{r}=0, \partial_{r} \mathbf{e}_{\theta}=\frac{\mathbf{e}_{\theta}}{r}, \partial_{r} \mathbf{e}_{\varphi}=\frac{\mathbf{e}_{\varphi}}{r} \\ \partial_{\theta} \mathbf{e}_{r}=\frac{\mathbf{e}_{\theta}}{r}, \partial_{\theta} \mathbf{e}_{\theta}=-r \mathbf{e}_{r}, \partial_{\theta} \mathbf{e}_{\varphi}=\cot \theta \mathbf{e}_{\varphi} \\ \partial_{\varphi} \mathbf{e}_{r}=\frac{\mathbf{e}_{\varphi}}{r}, \partial_{\varphi} \mathbf{e}_{\theta}=\cot \theta \mathbf{e}_{\varphi}, \partial_{\varphi} \mathbf{e}_{\varphi}=\left[\begin{array}{c} -r \sin \theta \cos \varphi \\ -r \sin \theta \sin \varphi \\ 0 \end{array}\right]\\ \Delta=\left(\mathbf{e}_{r} \partial_{r}+\frac{\mathbf{e}_{\theta}}{r^{2}} \partial_{\theta}+\frac{\mathbf{e}_{\varphi}}{r^{2} \sin ^{2} \theta} \partial_{\varphi}\right) \cdot\left(\mathbf{e}_{r} \partial_{r}+\frac{\mathbf{e}_{\theta}}{r^{2}} \partial_{\theta}+\frac{\mathbf{e}_{\varphi}}{r^{2} \sin ^{2} \theta} \partial_{\varphi}\right) \\ =\partial_{r, r}+0+0+\frac{1}{r} \partial_{r}+\frac{1}{r^{2}} \partial_{\theta, \theta}+0+\frac{1}{r} \partial_{r}+\frac{\cot \theta}{r^{2}} \partial_{\theta}+\frac{1}{r^{2} \sin ^{2} \theta} \partial_{\varphi, \varphi} \\ =\partial_{r, r}+\frac{2}{r} \partial_{r}+\frac{1}{r^{2}} \partial_{\theta, \theta}+\frac{\cot \theta}{r^{2}} \partial_{\theta}+\frac{1}{r^{2} \sin ^{2} \theta} \partial_{\varphi, \varphi} \end{gathered} \]
这与球坐标系下拉普拉斯算子的一般表达式(来源于网络)等价 \[ \Delta=\frac{1}{r^{2}} \frac{\partial}{\partial r}\left(r^{2} \frac{\partial u}{\partial r}\right)+\frac{1}{r^{2} \sin \theta} \frac{\partial}{\partial \theta}\left(\sin \theta \frac{\partial u}{\partial \theta}\right)+\frac{1}{r^{2} \sin ^{2} \theta} \frac{\partial^{2} u}{\partial \varphi^{2}} \]
\(\large\textcolor{blue}{微积分第四次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.3.22}\)
教材第一章总复习11
已知偏微分方程 (输运方程) \(\left\{\begin{array}{l}\dfrac{\partial z}{\partial t}=a \dfrac{\partial z}{\partial x}+b \dfrac{\partial z}{\partial y}, \\ z(x, y, 0)=z_{0}(x, y),\end{array}\right.\) 证明它的解为 \(z=\) \(z_{0}(x+a t, y+b t) .\)
解:显然当 \(z=z_{0}(x+at,y+bt)\) 时,\(\dfrac{\partial z}{\partial t}=\dfrac{\partial z}{\partial x}\cdot a+\dfrac{\partial z}{\partial y}\cdot b\) 满足条件
令 \(z(s)=z(x-as,y-bs,t+s)\),满足 \(\dfrac{\partial z}{\partial s}=-a\dfrac{\partial z}{\partial x}-b\dfrac{\partial z}{\partial y}+\dfrac{\partial z}{\partial t}=0\),则对每个 \(x,y,t\)
都有 \(z(x-as,y-bs,t+s)=z(0)=z(x,y,t)\),令 \(t=-s\) 代入得 \[ z(x+at,y+bt,0)=z_0(x+at,y+bt)=z(x,y,t) \] 则 \(z(x,y,t)=z_0(x+at,y+bt)\) 为满足条件的唯一解
教材第一章总复习12(1)
(1)设函数 \(f, g \in C^{2}\), 证明: \(u(x, y, z, t)=\dfrac{1}{r}(f(t+r)-g(t-r))\) 满足弦振动方程
\(\dfrac{\partial^{2} u}{\partial t^{2}}=\dfrac{\partial^{2} u}{\partial x^{2}}+\dfrac{\partial^{2} u}{\partial y^{2}}+\dfrac{\partial^{2} u}{\partial z^{2}}\), 其中 \(r=\sqrt{x^{2}+y^{2}+z^{2}}\);
证明:对 \(u(x, y, z, t)=\dfrac{1}{r}(f(t+r)-g(t-r))\) 求偏导有 \[ \begin{gathered} \dfrac{\partial u}{\partial t}=\dfrac{f'(t+r)-g'(t-r)}{r},\dfrac{\partial^2 u}{\partial t^2}=\dfrac{f''(t+r)-g''(t-r)}{r}\\ \dfrac{\partial u}{\partial x}=\dfrac{f'(t+r)\cdot x+g'(t-r)\cdot x-(f(t+r)-g(t-r))\cdot \dfrac{x}{r}}{x^2+y^2+z^2}\\ \dfrac{\partial^2 u}{\partial x^2}=\dfrac{(f'(t+r)+g'(t-r)+f''(t+r)\cdot \dfrac{x^2}{r}-g''(t-r)\cdot \dfrac{x^2}{r})\cdot r^2}{(x^2+y^2+z^2)^2}\\ +\dfrac{-(f'(t+r)+g'_r(t-r))\cdot \dfrac{x^2}{r^2}\cdot r^2-(f(t+r)-g(t-r))\cdot \dfrac{y^2+z^2}{r^3}\cdot r^2}{(x^2+y^2+z^2)^2}\\ -\dfrac{(f'(t+r)\cdot x+g'(t-r)\cdot x-(f(t+r)-g(t-r))\cdot \dfrac{x}{r})\cdot 2x}{(x^2+y^2+z^2)^2}\\ \end{gathered} \] \[ \begin{gathered} =\dfrac{-(f(t+r)-g(t-r))\cdot (\dfrac{-2x^2+y^2+z^2}{r})+(-3x^2+r^2)\cdot (f'(t+r)+g'(t-r))}{r^4}\\ +\dfrac{(f''(t+r)-g''(t-r))\cdot x^2r}{r^4} \end{gathered} \] 则对 \(x,y,z\) 求和有 \[ \begin{gathered} \dfrac{\partial^{2} u}{\partial x^{2}}+\dfrac{\partial^{2} u}{\partial y^{2}}+\dfrac{\partial^{2} u}{\partial z^{2}}=\dfrac{-(f(t+r)-g(t-r))\cdot \displaystyle \sum\dfrac{-2x^2+y^2+z^2}{r}}{r^4}\\ +\dfrac{(f'(t+r)+g'(t-r))\cdot \displaystyle \sum (-3x^2+r^2)}{r^4}+\dfrac{(f''(t+r)-g''(t-r))\cdot \displaystyle \sum x^2r}{r^4}\\ =0+0+\dfrac{f''(t+r)-g''(t-r)}{r}=\dfrac{\partial ^2u}{\partial t^2} \end{gathered} \]
教材第一章总复习15
设 \(f(x, y)\) 是可微函数, 且满足以下条件 \(\lim\limits _{x^{2}+y^{2} \rightarrow+\infty} \dfrac{f(x, y)}{\sqrt{x^{2}+y^{2}}}=+\infty\), 试证明: 对于任意的 \(\boldsymbol{v}=\left\{v_{1}, v_{2}\right\}\), 都存在点 \(\left(x_{0}, y_{0}\right)\), 使得 \(\operatorname{grad} f\left(x_{0}, y_{0}\right)=v\).
证明:对 \(\forall \ v_1,v_2\in R,\boldsymbol v=\{v_1,v_2\}\),定义 \(g(x,y)=f(x,y)-(v_1x+v_2y)\)
则 \(\mbox{grad}\ f(x_0,y_0)=\boldsymbol v\Longleftrightarrow \mbox{grad}\ g(x_0,y_0)=0\),而 \(\dfrac{|x_0x+y_0y|}{\sqrt{x^2+y^2}}\leq \dfrac{\sqrt{x_0^2+y_0^2}\cdot \sqrt{x^2+y^2}}{\sqrt{x^2+y^2}}\)
后者为一常量 \(<+\infty\),则 \(\lim\limits_{x^2+y^2\to +\infty}\dfrac{g(x,y)}{\sqrt{x^2+y^2}}=+\infty\),由极限定义知 \[ \forall \ M>0,\exists\ N>0,\forall \ x^2+y^2>N,\dfrac{g(x,y)}{\sqrt{x^2+y^2}}>M \] 则任取 \(M_0\),令 \(\mathbb{S}=\{(x,y)\mid x^2+y^2\leq N\}\),该集合为定义在 \(\mathbb{R}^2\to \mathbb{R}\) 上的有界闭集
对应值域有最小值 \(f(x_1,y_1)=m_{min}\),结合上述可知,该坐标为整个定义域上的最小值
则一定 \(\exists \ x_1,y_1,s.t.\mbox{grad}\ g(x_1,y_1)=0\),从而原命题成立
讲义2.4.8
求 \(\partial^{2} \operatorname{det}(A)\) 以及 \(\operatorname{det}\) 在 \(I\) 处的 \(2\) 阶 \(Taylor\) 展开式。(提示: \(A A^{* T}=\) \(\left.\operatorname{det}(A) I_{\circ}\right)\)
解:由 \(D\det A(B_1)=\text{tr}(A^{*T}B_1)=\det A\cdot \text{tr}(A^{-1}B_1)\),继续变化 \(A\) 为 \(B_2\) 求导有
\(D(\text{tr}(A^{-1}B_1))(B_2)=\text{tr}((A+B_2)^{-1}B_1)-\text{tr}(A^{-1}B_1)\) 由可逆运算的一阶“泰勒”展开
\((A+B_2)^{-1}=(A(I+A^{-1}B_2))^{-1}=(I+A^{-1}B_2)^{-1}A^{-1}=(I-A^{-1}B_2+o(B_2))A^{-1}\)
则有 \(D(\text{tr}(A^{-1}B_1))(B_2)=-\text{tr}(A^{-1}B_2A^{-1}B_1)\),则由链式法则对 \(\det A\) 的二阶展开式 \[ \partial ^2\det A=\det A\cdot (\tr(A^{-1}B_2)\cdot \tr(A^{-1}B_1)-\tr (A^{-1}B_2A^{-1}B_1)) \] 则 $$ 在 \(I\) 处的 \(Taylor\) 展开式有 \(1+\text{tr}(B)+\dfrac{1}{2}(\text{tr}^2(B)-\text{tr}(B^2))+o(\text{tr}(B^2))\)
讲义2.5.3
求以下函数的极值和最值
\(u=\sin x+\sin y+\sin z-\sin (x+y+z)(0 \leq x, y, z \leq \pi)\)
\(f(\mathbf{x})=\dfrac{\langle A \mathbf{x}, \mathbf{x}\rangle}{\langle\mathbf{x}, \mathbf{x}\rangle}(\mathbf{x} \neq \mathbf{0})\), 其中 \(A\) 是一个对称矩阵。
\(x^{y}-(x-1) y+(x-1)^{2}(x>0, y \in \mathbb{R})\) 。
解:\((1)\) 由对该三元函数求一阶导得
\[ \dfrac{\partial u}{\partial x}=\cos x-\cos (x+y+z), \dfrac{\partial u}{\partial y}=\cos y-\cos (x+y+z), \dfrac{\partial u}{\partial z}=\cos z-\cos (x+y+z) \] 当三者均为 \(0\) 时有 \[ \cos x=\cos y=\cos z=\cos (x+y+z),x,y,z\in[0,{\pi}] \] 由于 \(\cos x\) 函数在 \([0,\pi]\) 区间上是单调的,从而 \(x=y=z\),代入三倍角公式有 \[ \cos x=\cos (3x)=-3\cos x+4\cos ^3x \] 求二阶导 \[ \dfrac{\partial ^2 u}{\partial x^2}=-\sin x+\sin (x+y+z), \dfrac{\partial ^2 u}{\partial x \partial y}=\sin (x+y+z), \dfrac{\partial ^2 u}{\partial x\partial z}=\sin (x+y+z)\\ \dfrac{\partial ^2 u}{\partial ^2y}=-\sin y+\sin (x+y+z), \dfrac{\partial ^2 u}{\partial y\partial z}=\sin (x+y+z), \dfrac{\partial ^2 u}{\partial z^2}=-\sin z+\sin (x+y+z) \\ \] 解得 \(x=0\ 或\ \dfrac{\pi}{2}\),而当 \(x=0\) 为边界点,不可能为极值点,代入 \(x=\dfrac{\pi}{2}\) 计算 \(Hesse\) 矩阵有 \[ x=\dfrac{\pi}{2}, \begin{bmatrix} \dfrac{\partial ^2 u}{\partial x^2} &\dfrac{\partial ^2 u}{\partial x\partial y}&\dfrac{\partial ^2 u}{\partial x\partial z}\\ \dfrac{\partial ^2 u}{\partial y\partial x} &\dfrac{\partial ^2 u}{\partial y^2}&\dfrac{\partial ^2 u}{\partial y\partial z}\\ \dfrac{\partial ^2 u}{\partial z\partial x}&\dfrac{\partial ^2 u}{\partial z\partial y}&\dfrac{\partial ^2 u}{\partial z^2} \end{bmatrix}= \begin{bmatrix} -2&-1&-1\\-1&-2&-1\\-1&-1&-2 \end{bmatrix} \] 计算该矩阵的特征值 \[ \det(\boldsymbol H-\lambda \boldsymbol I)= \begin{vmatrix} -2-\lambda&-1&-1\\ -1&-2-\lambda&-1\\ -1&-1&-2-\lambda \end{vmatrix}=-(\lambda+1)^2(\lambda+4),\lambda=-1,-1,-4 \] 从而 \(\boldsymbol H\) 矩阵负定,从而 \(u(x,y,z)=\sin x+\sin y+\sin z-\sin (x+y+z)\) 在 \((\dfrac{\pi}{2},\dfrac{\pi}{2},\dfrac{\pi}{2})\) 有极大值 \(4\),同时也是最大值,对最小值有三角恒等式 \[ \sin (x)+\sin (y)+\sin (z)-\sin (x+y+z)=4 \sin \left(\frac{x+y}{2}\right) \sin \left(\frac{x+z}{2}\right) \sin \left(\frac{y+z}{2}\right) \] 代入取值范围有 \(\sin (\dfrac{x+y}{2}),\sin (\dfrac{x+z}{2}),\sin (\dfrac{y+z}{2})\geq 0\),则最小值为 \(0\)
综上所述原函数有极大值 \(4\),无极小值,有最大值 \(4\),最小值 \(0\)
\((2)\) 对该函数求导有 \[ \mathbf{d} f(\mathbf{x}) \mathbf{v}=\frac{2 \mathbf{v}^{T} A \mathbf{x}}{\mathbf{x}^{T} \mathbf{x}}-\frac{\mathbf{x}^{T} A \mathbf{x}}{\left(\mathbf{x}^{T} \mathbf{x}\right)^{2}} 2 \mathbf{v}^{T} \mathbf{x}=\frac{2 \mathbf{v}^{T}}{\mathbf{x}^{T} \mathbf{x}}[A \mathbf{x}-f(\mathbf{x}) \mathbf{x}] \] \(\mathrm{d} f(\mathbf{x}) \mathbf{v}=0(\forall \mathbf{v})\) 当且仅当 \(A \mathbf{x}=f(\mathbf{x}) \mathbf{x}\), 即 \(\mathbf{x}\) 是 \(A\) 的特征向量。 此时, \(f(\mathbf{x})\) 是相应的特征值。 在 \(A\) 的单位特征向量 \(\mathrm{x}_{0}\) 处, \(\mathbf{v}=\mu \mathbf{x}_{0}+(1+\mu) \mathbf{v}^{\perp}\) \[ f\left(\mathbf{x}_{0}+\mathbf{v}\right)=\lambda+\left(\mathbf{v}^{\perp}\right)^{T}(A-\lambda I) \mathbf{v}^{\perp}+o\left(\left\|\mathbf{v}^{\perp}\right\|^{2}\right) \] 如果 \(\lambda\) 是 \(A\) 的最大特征值(或最小特征值), 则 \(\mathrm{x}_{0}\) 是 \(f\) 的最大值点 (相应地,最小值点)。如果特征值 \(\lambda\) 既非 \(A\) 的最大特征 值也非最小特征值, 则 \(\mathbf{x}_{0}\) 不是 \(f\) 的极值点。
\((3)\) 对二元函数 \(z(x,y)=x^y-(x-1)y+(x-1)^2\) 求偏导等于 \(0\) 有 \[ \begin{gathered} \dfrac{\partial z}{\partial x}=yx^{y-1}-y+2(x-1)=0,\dfrac{\partial z}{\partial y}=x^{y}\ln x-(x-1)=0\\ \dfrac{\partial ^2z}{\partial x^2}=y(y-1)x^{y-2}+2,\dfrac{\partial ^2z}{\partial y^2}=x^{y}\ln ^2x,\dfrac{\partial ^2z}{\partial x\partial y}=x^{y-1}+yx^{y-1}\ln x-1 \end{gathered} \] 固定 \(y_0\) 有 \(y_0\leq0\) 或 \(y_0\leq 1\) 时 \(\dfrac{\partial ^2z}{\partial x^2}\geq2>0\),则 \(\dfrac{\partial z}{\partial x}\) 随 \(x\) 单调递增,有且仅有 \(x=1\) 该根
若 \(0<y_0<1\) ,有 \(y-2<0\),\(\dfrac{\partial ^3z}{\partial x^3}=y(y-1)(y-2)x^{y-3}>0\),\(\dfrac{\partial ^2z}{\partial x^2}\) 随 \(x\) 单调递增
取点 \(\dfrac{\partial ^2z}{\partial x^2}\big{|}_{x\to 0^{+}}\to -\infty\),\(\dfrac{\partial ^2z}{\partial x^2}\big{|}_{x=1}=2>0\),由零点存在定理, \[ \exists\ !\ x_{0}\in (0,1),s.t.\dfrac{\partial ^2z}{\partial x^2}\big{|}_{x=x_0}=0 \] 则 \(\dfrac{\partial z}{\partial x}\) 当 \(x\in (0,x_0)\) 单调递减,当 \(x\in (x_0,+\infty)\) 单调递增,又 \(\dfrac{\partial z}{\partial x}\big{|}_{x=1}=0\) 则有
\(\dfrac{\partial z}{\partial x}\big{|}_{x=x_0}<0\) 结合单调性可知,\(\dfrac{\partial z}{\partial x}\) 在 \((0,+\infty)\) 上有且仅有两个零点,一者 \(x_1\in (0,x_0)\) ,一者
为 \(0\) ,则 \(z\) 关于 \(x\) 在区间 \((0,x_1),(x_1,1),(1+\infty)\) 上分别单调递增、单调递减、单调递增
但对于 \(x_1\neq 1\) 其在固定 \(y\) 的截面上为极大值,而在固定 \(x\) 的截面时 \(\dfrac{\partial ^2z}{\partial y^2}=x^{y}\ln^2 x>0\) 为极小值
从而对所有的 \(x_1\) 均为函数的鞍点,非极值点。而只要极值点坐标 \(x=1\),代入 \(\mbox{Hesse}\) 矩阵有 \[ \mathbf{H}=\left[\begin{array}{cc}\dfrac{\partial^{2} z}{\partial x^{2}} & \dfrac{\partial^{2} z}{\partial x \partial y} \\ \dfrac{\partial^{2} z}{\partial x \partial y} & \dfrac{\partial^{2} z}{\partial y^{2}}\end{array}\right]=\left[\begin{array}{cc}y^{2}-y+2 & 0 \\ 0 & 0\end{array}\right] \] 为正定矩阵,从而所有 \(x=1\) 的点(一根直线)都是极小值点,极小值为 \(1\),无极大值
\(y\leq 0\) 或 \(y\geq 1\),\(x=1\) 为最小值,\(0<y<1\) 时,由单调性 \(\min z(x,y)=\min(1,z(x\to 0,y))\)
而 \(z(0,y)=1-y>0\),从而 \(z(x,y)\) 有最小值 \(1\),无最大值,其图像如下
讲义2.5.4
设 \(\gamma\) 是由抛物线 \(y=c^{2}-x^{2}(-c \leq x \leq c)\) 和 \(x\) 轴上的线段组成的简单封闭曲线。求 \(\gamma\) 的内接四边形的最大面积。
解:由于抛物线为凸函数,若四个点均在对称轴一侧,则固定其中三个点,将第四点移动到另一侧显然面积会增加;又若依据对称轴恰好一侧为三个点,一侧为一个点,则按照 \(y\) 轴坐标大小排序中间两个点相连总有斜率 \(k\),若 \(k=0\),则可以调整上下任何一个点到以抛物线对称轴为对称中心的对称点使得面积一样;若 \(k\neq 0\),则总有一侧的点转移至相应对称点后总面积增加。综上,内接四边形面积最大时对称轴分开一定每侧两个点。(假设对称轴上的点划分到右侧)
设横坐标从左往右依次为 \(-c\leq x_1<x_2<0\leq x_3<x_4\leq c\),固定 \(x_x,x_3,x_4\) 有内接四边形面积仅与横坐标为 \(x_1\) 的点到经过 \(x_3,x_4\) 直线的距离有关,令直线方程 \(y=kx+b,k<0,b>0\)
点到直线距离公式 \(d=\dfrac{|kx_1+x_1^2-c^2+b|}{\sqrt{k^2+1}}\),其中 \(k=-(x_3+x_4)\in(-2c,0)\),\(b=x_3x_4+c^2\)
代入 \(d=\dfrac{|kx_1+x_1^2+x_3x_4|}{\sqrt{k^2+1}}\),有 \(|kx_1+x_1^2+x_3x_4|\leq |k||c|+c^2+x_3x_4\),条件 \(x_1=-c\)
由对称性可知 \(x_1=-c,x_4=c\)。划分为两个三角形相加得到 \[ S\left(x_{2}, x_{3}\right)=\frac{1}{2} \cdot 2 c \cdot\left(c^{2}-x_{2}^{2}\right)+\frac{1}{2} \operatorname{det}\left|\begin{array}{cc}c-x_{2} & x_{2}^{2}-c^{2} \\ x_{3}-x_{2} & x_{2}^{2}-x_{3}^{2}\end{array}\right|=c^{3}+\frac{1}{2}\left(\left(x_{3}-x_{2}\right) c^{2}-\left(x_{2}^{2}+x_{3}^{2}\right) c+\left(x_{3}-x_{2}\right) x_{2} x_{3}\right) \] 有 \(\dfrac{\partial S}{\partial x_2}=-\dfrac{1}{2}c^2-cx_2+\dfrac{1}{2}x_3^2-x_2x_3,\dfrac{\partial S}{\partial x_3}=\dfrac{1}{2}c^2-cx_3-\dfrac{1}{2}x_2^2+x_2x_3\)
当两者等于 \(0\) 时,\(-2x_2+x_3=2x_3-x_2=c\),则 \(x_2=-\dfrac{1}{3}c\),\(x_3=\dfrac{1}{3}c\)
对应 \(\mbox{Hesse}\) 矩阵为 \(\mathbf{H}=\left[\begin{array}{cc}\dfrac{\partial^{2} S}{\partial x^{2}} & \dfrac{\partial^{2} S}{\partial x \partial y} \\ \dfrac{\partial^{2} S}{\partial x \partial y} & \dfrac{\partial^{2} S}{\partial y^{2}}\end{array}\right]=\left[\begin{array}{cc}-c-x_{3} & x_{3}-x_{2} \\ x_{3}-x_{2} & -c+x_{2}\end{array}\right]=\frac{c}{3}\left[\begin{array}{cc}-2 & 2 \\ 2 & -4\end{array}\right]\)
为极大值,且最大值点在极大值处取到,\(S_{\max}=\dfrac{32}{27}c^3\)
讲义2.5.5
证明: \(f(x, y)=-y+x^{2}+x y+x^{4}+y^{4}\) 有唯一极值点, 并判断该极值点的类型。
证明:对该二元函数求偏导得 \[ \begin{gathered} \dfrac{\partial f}{\partial x}=2x+y+4x^{3},\dfrac{\partial f}{\partial y}=-1+x+4y^3 \end{gathered} \] 则极值点对应 \(y=-2x(1+2x^2),4y^3=1-x\),对应 \(\mbox{Hesse}\) 矩阵有
\[ \mathbf{H}=\left[\begin{array}{cc}\dfrac{\partial^{2} z}{\partial x^{2}} & \dfrac{\partial^{2} z}{\partial x \partial y} \\ \dfrac{\partial^{2} z}{\partial x \partial y} & \dfrac{\partial^{2} z}{\partial y^{2}}\end{array}\right]=\left[\begin{array}{cc}2+8 x^{2} & 1 \\ 1 & 12 y^{2}\end{array}\right] \] 则由于 \(2+8x^2>2>0\),则该对称矩阵 \(\mathbf{H}\) 只能为正定矩阵,其行列式大于等于 \(0\) \[ 24y^2(1+4x^2)\geq1 \] 联立对应方程为 \(-32x^3(1+2x^2)^3=1-x\),则解不为 \(0\), \(f(x)=32x^3(1+2x^2)^3-x+1\)
\(f'(x)=32(3x^2(1+2x^2)^3)+x^3\cdot 12x(1+2x^2)^2-1=96x^2(1+2x^2)^2(1+6x^2)-1\)
考虑 \(x>0\) 的情况,\(f'(x)\) 在 \(x>0\) 时单调递增,\(f(0)=-1\),且有\(f(\dfrac{1}{4})>6-1>0\),由零点存在定理 \(\exists \ x_0\in (0,\dfrac{1}{4}),f'(x_0)=0\),且有对 \(x>0\),\(f(x)\) 在\((0,x_0)\) 单调递减,在 \((x_0,+\infty)\)
单调递增。而 \(f(\dfrac{1}{4})=\dfrac{1}{2}\cdot \dfrac{9}{8}^3-\dfrac{3}{4}=-\dfrac{39}{1024}<0\),有 \(f(x)\) 在 \((\dfrac{1}{4},+\infty)\) 单调递增
且 \(f(1)>1\),\(f(x_0)<f(\dfrac{1}{4})<0\),则在 \(x>0\) 处存在两个零点 \(x_1,x_2\),且有 \(x_1,x_2<1\)
放缩不等式有 \(48y^2\cdot \dfrac{y}{-2x}=24y^2(2+4x^2)\geq24y^2(1+4x^2)\geq 1\),代入 \(4y^3=1-x\)
得到 \(\dfrac{6(1-x)}{-x}\geq 1\Longrightarrow \dfrac{1}{x}\leq \dfrac{5}{6}\),若 \(x>0\),则至少需要 \(x\geq 1.2\) 才能满足正定矩阵条件,矛盾
从而 \(x<0\),由导函数知 \(f(x)\) 在 \((-x_0,0)\) 单调递减,\(f(x_0)>f(0)>1\),而在 \((-\infty,x_0)\)
单调递增,从而由零点唯一性定理,存在唯一负数 \(x_3<0\),使得 \(f(x_3)=0\)
此时有放缩 \(1-\dfrac{1}{x}\geq \dfrac{1}{3}\Longrightarrow \dfrac{12y^3}{-x}=24y^2\dfrac{y}{-2x}\geq 1\Longrightarrow 24y^2(1+2x^2)\geq 1\)
则满足条件 \(24y^2(1+4y^2)\geq 24y^2(1+2y^2)\geq 1\),则只有一个极值点,为极小值点,图像如下
讲义2.8 习题讨论课3.3
求以下函数在指定点的 \(Peano\) 余项 \(Taylor\) 展开:
\((a)\) \(x^{y}\) 在 \((x, y)=(1,0)\) 处展开到二阶;
\((b)\) \(\lambda x+\mathrm{e}^{-\left(x^{2}+y^{2}\right)}\) 在它的临界点处展开到二阶。
解:\((a)\) 对多元函数 \(z(x,y)=x^y\),有一阶导: \[ \dfrac{\partial z}{\partial x}=y\cdot x^{y-1}=0,\dfrac{\partial z}{\partial y}=\ln (x)\cdot x^y=0 \] 二阶导: \[ \dfrac{\partial^2 z}{\partial x^2}=y(y-1)\cdot x^{y-2}=0,\dfrac{\partial ^2 z}{\partial y^2}=\ln ^2(x)\cdot x^y=0,\dfrac{\partial ^2 z}{\partial x\partial y}=x^{y-1}+y\ln x\cdot x^{y-1}=1 \] 从而展开有 \[ z=1+(\dfrac{\partial }{\partial x}+\dfrac{\partial }{\partial y})z+\dfrac{1}{2!}(\dfrac{\partial }{\partial x}+\dfrac{\partial }{\partial y})^2z=1+(x-1)y \] 即有带 \(\mbox{peano}\) 余项的 \(\mbox{Taylor}\) 展开式为 \(z=1+y(x-1)+o(x^2+y^2)\)
\((b)\) 对多元函数 \(z(x,y)=\lambda x+e^{-(x^2+y^2)}\) 有一阶导 \[ \dfrac{\partial z}{\partial x}=\lambda-2xe^{-(x^2+y^2)},\dfrac{\partial z}{\partial y}=-2ye^{-(x^2+y^2)} \] 对于临界点两者取零 \(y=0,\lambda =2xe^{-x^2}=f(x)\) 有对 \(f(x)\) 求导 \(f'(x)=(2-4x^2)e^{-x^2}\)
\(x\in (-\infty,-\dfrac{\sqrt{2}}{2})\ \cap\ (\dfrac{\sqrt{2}}{2},+\infty),f'(x)<0\),\(f(x)\) 单调递减;\(x\in (-\dfrac{\sqrt{2}}{2},\dfrac{\sqrt{2}}{2})\)
\(f'(x)>0\),\(f(x)\) 单调递增。当 \(x\to -\infty\) 时,\(2xe^{-x^2}\to 0^{-}\),\(f(-\dfrac{\sqrt{2}}{2})=-\sqrt{\dfrac{2}{e}}\)
\(f(\dfrac{\sqrt{2}}{2})=\sqrt{\dfrac{2}{e}}\),结合单调性有 \(f(x)\in (-\sqrt{\dfrac{2}{e}},\sqrt{\dfrac{2}{e}})\),当且仅当 \(|\lambda|\leq \sqrt{\dfrac{2}{e}}\) 时有临界点
而该多元函数的二阶导有 \[ \dfrac{\partial^2 z}{\partial x^2}=(4x^2-2)e^{-(x^2+y^2)},\dfrac{\partial^2 z}{\partial y^2}=(4y^2-2)e^{-(x^2+y^2)},\dfrac{\partial^2 z}{\partial x \partial y}=4xye^{-(x^2+y^2)} \] 记满足 \(2xe^{-x^2}=\lambda\) 方程的解为 \(x_0\)(若有多个解,考虑其中一个解)对原函数展开有 \[ \begin{gathered} \lambda x+e^{-(x^2+y^2)}=\lambda x_0+e^{-x_0^2}+\dfrac{1}{2}(\dfrac{\partial }{\partial x}+\dfrac{\partial }{\partial y})^2z=\lambda x_0+e^{-x_0^2}+(2x_0^2-1)e^{-x_0^2}(x-x_0)\\ -e^{-x_0^2}y+0+o(x^2+y^2)=\lambda(x_0+\dfrac{1-y}{2x_0}+(2x_0-\dfrac{1}{x_0})(x-x_0))+o(x^2+y^2) \end{gathered} \]
讲义2.8 习题讨论课3.4
对参数 \(\lambda\), 讨论函数 \(\lambda x+\mathrm{e}^{-\left(x^{2}+y^{2}\right)}\) 的极值情况。
解:由上一题可知,当且仅当 \(-\sqrt{\dfrac{2}{e}}\leq \lambda \leq \sqrt{\dfrac{2}{e}}\) 时有临界点,而对应 \(\mbox{Hesse}\) 矩阵有 \[ \mathbf{H}=\left[\begin{array}{cc}\dfrac{\partial^{2} z}{\partial x^{2}} & \dfrac{\partial^{2} z}{\partial x \partial y} \\ \dfrac{\partial^{2} z}{\partial x \partial y} & \dfrac{\partial^{2} z}{\partial y^{2}}\end{array}\right]=2 e^{-\left(x^{2}+y^{2}\right)}\left[\begin{array}{cc}2 x^{2}-1 & 2 x y \\ 2 x y & 2 y^{2}-1\end{array}\right] \] 求解对应特征值 \[ \det(\mathbf{H-\lambda \mathbf{I}})=(2x^2-1-\lambda)\cdot (2y^2-1-\lambda)-4x^2y^2=(\lambda+1)^2-2(\lambda+1)(x^2+y^2) \] 则特征值为 \(\lambda_1=-1,\lambda_2=2x^2+2y^2-1\),则若函数存在极值点,只能 \(\lambda_1,\lambda_2\) 均小于零
为极大值点,此时 \(2x^2+2y^2\leq1\Longrightarrow x^2\leq\dfrac{1}{2}\),由上一题的单调性可知 \(-\sqrt{\dfrac{2}{e}}\leq \lambda \leq \sqrt{\dfrac{2}{e}}\)
综上所述,\(-\sqrt{\dfrac{2}{e}}\leq \lambda \leq \sqrt{\dfrac{2}{e}}\) 时有极大值点,令 \(2xe^{-x^2}=\lambda\) 绝对值不超过 \(\dfrac{\sqrt{2}}{2}\) 的解为 \(x_0\),该极大值点位于 \((x_0,0)\),极大值为 \(e^{-x_0^2}+\lambda x_0=\lambda(x_0+\dfrac{1}{2x_0})\),否则无极值点
\(\large\textcolor{blue}{微积分第五次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.3.29}\)
讲义3.1.4
证明 \(-y+x^{2}+x y+x^{4}+y^{4}=0\) 在 \((0,0)\) 的一个邻域内定义了一个 \(C^{\infty}\) 隐函数 \(y=h(x)\), 并求 \(h\) 在 \(x=0\) 处的四阶泰勒展开式。
解:对 \(-y+x^2+xy+x^4+y^4=0\) 令 \(x,y\to 0\) 使用 \(o\) 语言有 \[ -y+x^2+o(y)+o(x^2)+o(y)=0\Longrightarrow y=x^2+o(x^2) \] 令 \(y=x^2+u\) 代入有 \[ -u+x^2(x+u)+x^4+(x^2+u)^4=0=-u+x^3+o(u)+o(x^3)+o(x^3)+o(u) \] 则 \(u=x^3+o(x^3)\),令 \(y=x^2+x^3+v\) 代入有 \[ -x^3-v+x(x^2+x^3+v)+x^4+(x^2+x^3+v)^4=-v+2x^4+o(v)+o(x^4)=0 \] 则 \(v=2x^4+o(x^4)\),从而在 \((0,0)\) 处四阶泰勒展开为 \(y=x^2+x^3+2x^4+o(x^4)\)
讲义3.1.5
设 \(f: U \times V \rightarrow \mathbb{R}\) 是 \(C^{2}\) 函数, \(\mathbf{x}_{0}\) 是 \(f\left(\cdot, \lambda_{0}\right)\) 的非退化临界点, 即 \(f\left(\mathbf{x}, \lambda_{0}\right)\) 关于 \(\mathbf{x}\) 在 \(\mathbf{x}_{0}\) 处的 \(Hesse\) 矩阵是可逆的。证明对足够接近 \(\lambda_{0}\) 的所有 \(\lambda\), \(f(\mathbf{x}, \lambda)\) 关于 \(\mathbf{x}\) 在 \(\mathbf{x}_{0}\) 附近有唯一的临界点, 并且该临界点的类型与 \(\mathbf{x}_{0}\) 作为 \(f\left(\cdot, \lambda_{0}\right)\) 的临界点的类型是一样的。
证明:由 \(\dfrac{\partial f}{\partial \mathbf{x}}=\vec{0},\dfrac{\partial f}{\partial \lambda}=0\) 有 \(\dfrac{\partial f_1}{\partial \mathbf{x}^{i}}(\mathbf{x},\lambda)=0,\dfrac{\partial f_1}{\partial \lambda}(\mathbf{x},\lambda)=0\),该组隐函数确定了 \(\lambda=\lambda(\mathbf{x})\)
这是由于 \(f(\mathbf{x}_0,\lambda)\) 在 \(\mathbf{x}=\mathbf{x}_0\) 处的 \(Hesse\) 矩阵是可逆的,从而在 \(\mathbf{x}_0\) 附近,都能找到唯一临界点 \(\lambda\)
又由于 \(\det(\mathbf{H})\big|_{\lambda_0,\mathbf{x}_0}\neq 0\) 且 \(\det\) 函数是连续的,从而总能找到 \(\lambda\) 附近的领域,使得 \(\det \mathbf{H}\) 在该
领域内是保号的,正定对称矩阵附近仍为正定对称矩阵,负定也为负定,从而临界点类型是一样的
讲义3.1.7
设 \(\lambda_{0}\) 是矩阵 \(A_{0}\) 的一个单重特征值 (即 \(p\left(\lambda_{0}\right)=0, p^{\prime}\left(\lambda_{0}\right) \neq 0\), 其中 \(p(\lambda)=\) \(\left.\operatorname{det}\left(\lambda I-A_{0}\right)\right), \mathbf{x}_{0}\) 是 \(A_{0}\) 对应于 \(\lambda_{0}\) 的一个单位特征向量。证明存在 \(\delta>0\) 使得对任意矩阵 \(A\), 只要 \(\left\|A-A_{0}\right\|<\delta, A\) 就有唯一的特征值 \(\lambda(A)\) 和相应的单位特征向量 \(\mathbf{x}(A)\) 使得 \(\lambda\left(A_{0}\right)=\lambda_{0}, \mathbf{x}\left(A_{0}\right)=\mathbf{x}_{0}\), 并且 \(\lambda(A)\) 和 \(\mathbf{x}(A)\) 关于 \(A\) 是 \(C^{\infty}\) 的。
证明:由 \(A\) 和 \(\lambda\) 由下列隐函数方程确定, \(f(A,\lambda)=\det(\lambda I-A)=p_{A}(\lambda)=0\)
而由题设条件,该方程满足 \(\dfrac{\partial f}{\partial \lambda}\big |_{\lambda=\lambda_0}=p'(\lambda_0)\neq 0\),且在该处只有 \(n\) 次求导后的结果是非零的,从而该方程在 \(A,\lambda_0\) 领域所确定的 \(\lambda(A)\) 关于 \(A\) 只能达到 \(C^{n}\)
而确定 \(\lambda\) 后,\(g(A,\lambda,\vec{\mathbf{x}})=g(A,\lambda(A),\vec{\mathbf{x}})=(A-\lambda I)\vec{\mathbf{x}}=0\),有 \(\dfrac{\partial g}{\partial \mathbf{x^i}}=(A-\lambda I)_{i}\)
为 \((A-\lambda I)\) 第 \(i\) 列,若其为 \(0\),由行列式性质 \(p_{A}(\lambda I-A)=0\),与题设矛盾,从而均不为 \(0\)
则确定了隐函数方程 \(\mathbf{x}=\mathbf{x}(A,\lambda(A))=\mathbf{x}(A)\) ,对高阶导由线性性质,只有对角元上有非零元素,
从而 \(\dfrac{\partial ^nf}{\partial \mathbf{x}^n}\) 确定的线性映射都是可逆的,从而 \(\mathbf{x}(A)\) 为 \(C^{\infty }\) 的
讲义3.2.1
设 \[ f\left(x_{1}, x_{2}\right)=\left(x_{1} x_{2},\left(1-x_{1}\right) x_{2}\right) . \] 证明 \(f\) 是 \((0,1) \times(0,1)\) 到 \(\left\{\left(y_{1}, y_{2}\right) \mid y_{1}>0, y_{2}>0, y_{1}+y_{2}<1\right\}\) 的微分同胚。
证明:对于变换 \(\begin{cases}y_1=x_1x_2\\y_2=(1-x_1)x_2\end{cases}\) 反解得 \(x_1=\dfrac{y_1}{y_1+y_2},x_2=y_1+y_2\)
则有 \(y_1+y_2<1\) 对应 \(x_1<1,x_2<1\),则 \(f\) 在 \((0,1)\longmapsto(0,1)\) 上是双射,满足一一对应,从而
\(f\) 微分同胚
讲义3.2.2
记 \(\mathcal{S}_{n}\) 为所有 \(n\) 阶对称矩阵组成的线性空间, \(\mathcal{P}_{n}\) 为所有 \(n\) 阶对称正定矩阵组成的集合。证明
\((a)\) \(\mathcal{P}_{n}\) 是 \(\mathcal{S}_{n}\) 的开子集;
\((b)\) 对任意 \(A \in \mathcal{P}_{n}\), 存在唯一的 \(A^{1 / 2} \in \mathcal{P}_{n}\) 使得 \(\left(A^{1 / 2}\right)^{2}=A\), 并且 \(A^{1 / 2}\) 关于 \(A\) 是 \(C^{\infty}\) 的。
证明:\((a)\) 显然 \(\mathcal{S}_{n}\) 是 \(\mathcal{P}_{n}\) 的真子集,而对于 \(A\in \mathcal{P}\) 为对称正定矩阵,由线性代数知识 \[ A\ 对称正定\Longleftrightarrow \forall \ \mathbf{x}\in \mathbb{R}^n,\mathbf{x}^TA\mathbf{x}>0 \] 考虑 \(f(A)=\mathbf{x^T}A\mathbf{x}\) 关于 \(A\) 是连续的,由于对所有 \(\mathbf{x}\) 该连续函数 \(f(A)>0\),
则总能找到 \(A\) 的领域使得所有的 \(f(A+E)>0\),即所有 \(n\) 阶对称正定矩阵组成的集合为开集
\((b)\) 先证存在性,令 \(A=L\mbox{diag}(\lambda_1,\lambda_2,\cdots,\lambda_n)L^{T},A^{\frac{1}{2}}:=L\mbox{diag}(\sqrt{\lambda_1},\sqrt{\lambda_2},\cdots,\sqrt{\lambda_n})L^{T}\)
此时有 \((A^{\frac{1}{2}})^2=L\mbox{diag}(\sqrt{\lambda_1},\sqrt{\lambda_2},\cdots,\sqrt{\lambda_n})L^{T}L\mbox{diag}(\sqrt{\lambda_1},\sqrt{\lambda_2},\cdots,\sqrt{\lambda_n})L^{T}=A\)
再证唯一性,若有 \(B_1^2=B_2^2=A\),且 \(B_1,B_2\in \mathcal{P}_n\),令 \(B_1=L_1D_1L_1^{T},B_2=L_2D_2L_2^{T}\)
代入有 \(L_1D_1^2L_1^{T}=L_2D_2^2L_2^{T}=A\Longrightarrow L_2^{-1}L_1=D_2^2L_2^{T}(L_1^{T})^{-1}(D_1^{-1})^2\)
左侧为上三角矩阵,右侧为下三角矩阵,从而 \(L_2^{-1}L_1=I\Longrightarrow L_1=L_2,D_1=D_2\)
即 \(B\) 存在且唯一。由该“求根”映射为双射,其逆映射 \(A^{\frac{1}{2}}\longmapsto A\) 无穷可微
这是因为 \(D(A^2)(B)=(A+B)^2-A^2=AB+BA+o(\|B\|)\),显然可以不断微分下去
由逆映射定理知,\(A^{\frac{1}{2}}\) 关于 \(A\) 也是 \(C^{\infty}\) 的
讲义3.3.1
设 \(f, g, h\) 都是自变量 \(x, y\) 的函数, 证明
\((a)\) \(\left(\dfrac{\partial f}{\partial g}\right)_{x}=\left(\dfrac{\partial f}{\partial y}\right)_{x} /\left(\dfrac{\partial g}{\partial y}\right)_{x}\);
\((b)\) \(\left(\dfrac{\partial y}{\partial x}\right)_{f}=-\left(\dfrac{\partial f}{\partial x}\right)_{y} /\left(\dfrac{\partial f}{\partial y}\right)_{x}\);
\((c)\) \(\left(\dfrac{\partial f}{\partial x}\right)_{g}=\left(\dfrac{\partial f}{\partial x}\right)_{y}+\left(\dfrac{\partial f}{\partial y}\right)_{x}\left(\dfrac{\partial y}{\partial x}\right)_{g}\) 。
证明:\((a)\) 由所证明式子保持 \(x\) 不变,则有 \(df=\dfrac{\partial f}{\partial y}dy,dg=\dfrac{\partial f}{\partial y}dy\),两者相比得 \[ \dfrac{df}{dg}_{x=\tiny \mbox{Const}}=(\dfrac{\partial f}{\partial g})_x=\dfrac{(\dfrac{\partial f}{\partial y})_{x}}{(\dfrac{\partial g}{\partial y})_x} \] \((b)\) 对 \(f=f(x,y)\) 全微分有 \(df=\dfrac{\partial f}{\partial x}dx+\dfrac{\partial f}{\partial y}dy\),当 \(f\) 不变时计算 \((\dfrac{\partial y}{\partial x})_f\) 得 \[ 0=\dfrac{\partial f}{\partial x}dx+\dfrac{\partial f}{\partial y}dy\Longrightarrow (\dfrac{\partial y}{\partial x})_f=-\dfrac{(\dfrac{\partial f}{\partial x})_{y}}{(\dfrac{\partial f}{\partial y})_x} \] \((c)\) 当 \(g\) 不变时有,\(0=\dfrac{\partial g}{\partial x}dx+\dfrac{\partial g}{\partial y}dy\),代入 \(df=\dfrac{\partial f}{\partial x}dx+\dfrac{\partial f}{\partial y}dy\) 消去 \(dy\) 后有 \[ df=\dfrac{\partial f}{\partial x}dx+\dfrac{\partial f}{\partial y}\cdot (-\dfrac{\dfrac{\partial g}{\partial x} }{\dfrac{\partial g}{\partial y}}dx) \] 其中 \(dg=\dfrac{\partial g}{\partial x}dx+\dfrac{\partial g}{\partial y}dy\),则固定 \(g\) 不变有,\((\dfrac{\partial y}{\partial x})_g=-\dfrac{\dfrac{\partial g}{\partial x} }{\dfrac{\partial g}{\partial y}}\) 代入有 \[ (\dfrac{\partial f}{\partial x})_g=(\dfrac{\partial f}{\partial x})_y+(\dfrac{\partial f}{\partial y})_x(\dfrac{\partial y}{\partial x})_g \]
讲义3.3.2
设 \(f_{1}, \ldots, f_{n} ; u_{1}, \ldots, u_{n}\) 都是独立变量 \(x_{1}, \ldots, x_{n}\) 的函数。证明
\((a)\) \(\operatorname{det} \dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(u_{1}, \ldots, u_{n}\right)}=\operatorname{det} \dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)} / \operatorname{det} \dfrac{\partial\left(u_{1}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}\);
\((b)\) \(\left(\dfrac{\partial f_{1}}{\partial x_{1}}\right)_{f_{2}, \ldots, f_{n}}=\operatorname{det} \dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)} / \operatorname{det} \dfrac{\partial\left(f_{2}, \ldots, f_{n}\right)}{\partial\left(x_{2}, \ldots, x_{n}\right)}\);
\((c)\) \(\left(\dfrac{\partial f_{1}}{\partial u_{1}}\right)_{u_{2}, \ldots, u_{n}}=\operatorname{det} \dfrac{\partial\left(f_{1}, u_{2}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)} / \operatorname{det} \dfrac{\partial\left(u_{1}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}\) 。
证明:\((a)\) 由各组微体积元之比为 \[ \begin{gathered} df_1df_2\cdots df_n=\det(\dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)})dx_1dx_2\cdots dx_n\\ df_1df_2\cdots df_n=\det(\dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(u_{1}, \ldots, u_{n}\right)})du_1du_2\cdots du_n\\ du_1du_2\cdots du_n=\det(\dfrac{\partial\left(u_{1}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)})dx_1dx_2\cdots dx_n\\ \end{gathered} \] 从而得到 \(\operatorname{det} \dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(u_{1}, \ldots, u_{n}\right)}=\dfrac{df_1\cdots df_n}{du_1\cdots du_n}=\operatorname{det} \dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)} / \operatorname{det} \dfrac{\partial\left(u_{1}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}\)
\((b)\) 由固定 \(f_2,f_3,\cdots,f_n\) 有 \(df_2=df_3=\cdots=df_n=0\) \[ df_2=\dfrac{\partial f_2}{\partial x_1}dx_1+\cdots+\dfrac{\partial f_2}{\partial x_n}dx_n,df_3=\dfrac{\partial f_3}{\partial x_1}dx_1+\cdots+\dfrac{\partial f_3}{\partial x_n}dx_n\cdots ,df_n=\cdots \] 移项写成矩阵的形式有 \(\dfrac{\partial (f_2,f_3\cdots,f_n)}{\partial (x_2,x_3\cdots,x_n)}\begin{pmatrix}dx_2\\dx_3\\\vdots\\dx_n\end{pmatrix}=A\begin{pmatrix}dx_2\\dx_3\\\vdots\\dx_n\end{pmatrix}=-\begin{pmatrix}\dfrac{\partial f_2}{\partial x_2}\\\dfrac{\partial f_3}{\partial x_3}\\\vdots\\\dfrac{\partial f_n}{\partial x_n}\end{pmatrix}dx_1\) 代入计算有
\(df_1=\dfrac{\partial f_1}{\partial x_1}dx_1+\cdots+\dfrac{\partial f_1}{\partial x_n}dx_n=dx_1(\dfrac{\partial f_1}{\partial x_1}-\begin{pmatrix}\dfrac{\partial f_1}{\partial x_2}\\\dfrac{\partial f_1}{\partial x_3}\\\vdots\\\dfrac{\partial f_1}{\partial x_n}\end{pmatrix}^{T}A^{-1}\begin{pmatrix}\dfrac{\partial f_2}{\partial x_2}\\\dfrac{\partial f_3}{\partial x_3}\\\vdots\\\dfrac{\partial f_n}{\partial x_n}\end{pmatrix})=dx_1(\dfrac{\partial f_1}{\partial x_1}-\vec{v}_1^{T}A^{-1}\vec{v}_2)\)
由矩阵分块 \(B=\operatorname{det} \dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}=\begin{pmatrix}\dfrac{\partial f_1}{\partial x_1}&\vec{v_1}^{T}\\\vec{v}_2&A\end{pmatrix}\) \[ \dfrac{\det B}{\det A}=\dfrac{\det \begin{pmatrix}\dfrac{\partial f_1}{\partial x_1}&\vec{v_1}^{T}\\\vec{v}_2&A\end{pmatrix}}{\det A}=\dfrac{\det \begin{pmatrix}\dfrac{\partial f_1}{\partial x_1}-\vec{v}_1^{T}A^{-1}\vec{v}_2&\vec{v_1}^{T}\\\vec{0}&A\end{pmatrix}}{\det A}=(\dfrac{\partial f_1}{\partial x_1})_{f_2,\cdots,f_n} \] 即有 \(\left(\dfrac{\partial f_{1}}{\partial x_{1}}\right)_{f_{2}, \ldots, f_{n}}=\operatorname{det} \dfrac{\partial\left(f_{1}, \ldots, f_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)} / \operatorname{det} \dfrac{\partial\left(f_{2}, \ldots, f_{n}\right)}{\partial\left(x_{2}, \ldots, x_{n}\right)}\)
\((3)\) 由固定 \(u_2,u_3,\cdots,u_n\) 有 \(du_2=du_3=\cdots=du_n=0\) \[ du_2=\dfrac{\partial u_2}{\partial x_1}dx_1+\cdots+\dfrac{\partial u_2}{\partial x_n}dx_n,du_3=\dfrac{\partial u_3}{\partial x_1}dx_1+\cdots+\dfrac{\partial u_3}{\partial x_n}dx_n\cdots ,du_n=\cdots \] 移项写成矩阵形式有 \(\dfrac{\partial (u_2,u_3\cdots,u_n)}{\partial (x_2,x_3\cdots,x_n)}\begin{pmatrix}dx_2\\dx_3\\\vdots\\dx_n\end{pmatrix}=A\begin{pmatrix}dx_2\\dx_3\\\vdots\\dx_n\end{pmatrix}=-\begin{pmatrix}\dfrac{\partial u_2}{\partial x_2}\\\dfrac{\partial u_3}{\partial x_3}\\\vdots\\\dfrac{\partial u_n}{\partial x_n}\end{pmatrix}dx_1\),代入
由上问结论可推知,\(du_1=\dfrac{\partial u_1}{\partial x_1}dx_1+\cdots+\dfrac{\partial u_1}{\partial x_n}dx_n=dx_1\dfrac{\operatorname{det} \dfrac{\partial\left(u_{1}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}}{\det A}\)
\(df_1=\dfrac{\partial f_1}{\partial x_1}dx_1+\cdots+\dfrac{\partial f_1}{\partial x_n}dx_n=dx_1(\dfrac{\partial f_1}{\partial x_1}-\begin{pmatrix}\dfrac{\partial f_1}{\partial x_2}\\\dfrac{\partial f_1}{\partial x_3}\\\vdots\\\dfrac{\partial f_1}{\partial x_n}\end{pmatrix}^{T}A^{-1}\begin{pmatrix}\dfrac{\partial u_2}{\partial x_2}\\\dfrac{\partial u_3}{\partial x_3}\\\vdots\\\dfrac{\partial u_n}{\partial x_n}\end{pmatrix})=dx_1(\dfrac{\partial f_1}{\partial x_1}-\vec{v}_1^{T}A^{-1}\vec{v}_2)\)
由矩阵分块 \(C=\operatorname{det} \dfrac{\partial\left(f_{1}, u_{2}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}=\begin{pmatrix}\dfrac{\partial f_1}{\partial x_1}&\vec{v_1}^{T}\\\vec{v}_2&A\end{pmatrix}\) \[ \dfrac{\det C}{\det A}=\dfrac{\det \begin{pmatrix}\dfrac{\partial f_1}{\partial x_1}&\vec{v_1}^{T}\\\vec{v}_2&A\end{pmatrix}}{\det A}=\dfrac{\det \begin{pmatrix}\dfrac{\partial f_1}{\partial x_1}-\vec{v}_1^{T}A^{-1}\vec{v}_2&\vec{v_1}^{T}\\\vec{0}&A\end{pmatrix}}{\det A}=(\dfrac{\partial f_1}{\partial x_1})_{u_2,\cdots,u_n} \]
\[ \left(\dfrac{\partial f_{1}}{\partial u_{1}}\right)_{u_{2}, \ldots, u_{n}}=\dfrac{\det C}{\det A}\cdot \dfrac{\det A}{\det \dfrac{\partial\left(u_{1}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}}=\dfrac{\operatorname{det} \dfrac{\partial\left(f_{1}, u_{2}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)} }{\operatorname{det} \dfrac{\partial\left(u_{1}, \ldots, u_{n}\right)}{\partial\left(x_{1}, \ldots, x_{n}\right)}} \]
讲义3.4.3
绘制平面地图有很多方法, \(Mercator\) 投影是其中之一。如图所示, 它把球面上的点 \(P(x, y, z)\) 映为从球心出发的射线 \(O P\) 与一个圆柱面的交点 \(P^{\prime}(\lambda x, \lambda y, \lambda z)\), 然后在柱面上再做一个变换得到点 \(Q(\lambda x, \lambda y, f(\lambda z))\), 该圆柱面与球面相切于赤道。这样除南北两极外, 地球上的点 \(P\) 与柱面上的点 \(Q\) 一一对应, 把圆柱面沿母线剪开就得到一张地图。求函数 \(f\) 的表达式使得 \(Mercator\) 地图上所画的两条相交道路的夹角与实际道路的夹角大小相等。两条相交曲线的夹角被定义为它们的切线所形成的夹角(锐角或直角)。(提示:线性变换 \(A: \mathbb{R}^{n} \rightarrow \mathbb{R}^{n}\) 保角当且仅当它把正交且长度相等的向量映为正交且长度相等的向量, 从而存在常数 \(C>0\) 使得内积 \(\left.\langle A \mathbf{u}, A \mathbf{v}\rangle=C\langle\mathbf{u}, \mathbf{v}\rangle, \forall \mathbf{u}, \mathbf{v} \in \mathbb{R}^{n}\right)\)
解:设球面为 \(x^2+y^2+z^2=R^2\),对其上一点 \((x_0,y_0,z_0)\) 与圆柱面交线 \((\lambda x_0,\lambda y_0,\lambda z_0)\)
满足 \(\lambda^2(x_0^2+y_0^2)=R^2\) 代入变换得 \(Q(\dfrac{Rx_0}{\sqrt{x_0^2+y_0^2}},\dfrac{Ry_0}{\sqrt{x_0^2+y_0^2}},f(\dfrac{Rz_0}{\sqrt{x_0^2+y_0^2}}))\)
考虑经度、纬度恒为定值的两条微元,取纬度 \(\varphi\sim \varphi+d\varphi\) 处 \(d\theta\) 的正交微圆弧,令其长度相同
\(d\theta\cdot R\cos\varphi=Rd\varphi\Longrightarrow d\theta=\dfrac{d\varphi}{\cos \varphi}\),该两条圆弧映射后的长度分别为 $Rd$ 和 \(dz\)
两者相等有 \(Rd\theta=dz=R\dfrac{d\varphi}{\cos \varphi}\) 两边积分得 \(z(\varphi)=\displaystyle \int R\dfrac{d\varphi}{\cos \varphi}=R\ln (\dfrac{1+\sin \varphi}{\cos \varphi})+C\)
代入 \(\varphi = 0\) 时 \(z=0\),有 \(C=0,z(\varphi)=R\ln (\dfrac{1+\sin \varphi}{\cos \varphi})\),\(\lambda z_0=\dfrac{Rz_0}{\sqrt{x_0^2+y_0^2}}=\dfrac{Rz_0}{\sqrt{R^2-z_0^2}}\)
令 \(x=\lambda z_0\) 反解得 \(z_0=\dfrac{Rx}{\sqrt{R^2+x^2}},\dfrac{1+\sin \varphi}{\cos \varphi }=\dfrac{R+z_0}{\sqrt{R^2-z_0^2}}=\dfrac{x+\sqrt{x^2+ R^2}}{R}\)
从而 \(f:x\longmapsto R\cdot\mbox{arcsinh}\ \dfrac{x}{R}\),为反双曲三角正弦函数
教材1.6.4
设方程 \(f\left(u^{2}-x^{2}, u^{2}-y^{2}, u^{2}-z^{2}\right)=0\) 确定了函数 \(u=u(x, y, z)\), 其中 \(f\) 可微,证明: \[ \frac{1}{x} \frac{\partial u}{\partial x}+\frac{1}{y} \frac{\partial u}{\partial y}+\frac{1}{z} \frac{\partial u}{\partial z}=\frac{1}{u} . \]
证明:由在方程 \(f\left(u^{2}-x^{2}, u^{2}-y^{2}, u^{2}-z^{2}\right)=0\) 下计算得 \[ \begin{gathered} \dfrac{\partial u}{\partial x}=-\dfrac{\dfrac{\partial f}{\partial x}}{\dfrac{\partial f}{\partial u}}=-\dfrac{-2xf_{1}'}{2u(f_1'+f_2'+f_3')}, \dfrac{\partial u}{\partial y}=-\dfrac{\dfrac{\partial f}{\partial y}}{\dfrac{\partial f}{\partial u}}=-\dfrac{-2yf_{1}'}{2u(f_1'+f_2'+f_3')},\\ \dfrac{\partial u}{\partial z}=-\dfrac{-2zf_{1}'}{2u(f_1'+f_2'+f_3')}\rightarrow \frac{1}{x} \frac{\partial u}{\partial x}+\frac{1}{y} \frac{\partial u}{\partial y}+\frac{1}{z} \frac{\partial u}{\partial z}=\dfrac{f_1'+f_2'+f_3'}{u(f_1'+f_2'+f_3')}=\dfrac{1}{u} \end{gathered} \]
教材1.7.1(2)
求下列曲面在给定点的切平面方程和法线方程.
\((2)\) \(z=\arctan \dfrac{y}{x}\), 点 \(P\left(1,1, \dfrac{\pi}{4}\right)\);
解:方程 \(f(x,y,z)=z-\arctan(\dfrac{y}{x})=0\) 在点 \((1,1,\dfrac{\pi}{4})\) 处的法向量为 \[ \vec{n}=(\dfrac{\partial f}{\partial x},\dfrac{\partial f}{\partial y},\dfrac{\partial f}{\partial z})=(\dfrac{y}{x^2+y^2},\dfrac{-x}{x^2+y^2},1)=(\dfrac{1}{2},-\dfrac{1}{2},1) \] 从而切平面方程为 \(\dfrac{1}{2}(x-1)-\dfrac{1}{2}(y-1)+(z-\dfrac{\pi}{4})=0\),即 \(x-y+2z=\dfrac{\pi}{2}\)
法线方程 \(\dfrac{x-1}{\dfrac{1}{2}}=\dfrac{y-1}{-\dfrac{1}{2}}=\dfrac{z-\dfrac{\pi}{4}}{1}\)
教材1.7.4(2)
证明下列各题
\((2)\) 曲面 \(\sqrt{x}+\sqrt{y}+\sqrt{z}=\sqrt{a}\) 的任意一点处的切平面在各坐标轴上的截距之和为 \(a\) ;
解:方程 \(f(x,y,z)=\sqrt{x}+\sqrt{y}+\sqrt{z}-\sqrt{a}=0\) 在任意一点 \((x_0,y_0,z_0)\) 处的法向量为 \[ \vec{n}=(\dfrac{\partial f}{\partial x},\dfrac{\partial f}{\partial y},\dfrac{\partial f}{\partial z})=(\dfrac{1}{2\sqrt{x}},\dfrac{1}{2\sqrt{y}},\dfrac{1}{2\sqrt{z}})=(\dfrac{1}{2\sqrt{x_0}},\dfrac{1}{2\sqrt{y_0}},\dfrac{1}{2\sqrt{z_0}}) \] 则其切平面方程 \(\dfrac{x-x_0}{2\sqrt{x_0}}+\dfrac{y-y_0}{2\sqrt{y_0}}+\dfrac{z-z_0}{2\sqrt{z_0}}=0\),则在 \(x,y,z\) 轴上的截距分别为
\(x_{inter}=x_0+({\sqrt{y_0}+\sqrt{z_0}}) \sqrt{x_0},y_{inter}=y_0+({\sqrt{x_0}+\sqrt{z_0}}) \sqrt{y_0},z_{inter}=z_0+({\sqrt{x_0}+\sqrt{y_0}})\sqrt{z_0}\)
从而 \(x_{inter}+y_{inter}+z_{inter}=x_0+y_0+z_0+2(\sqrt{x_0y_0}+\sqrt{x_0z_0}+\sqrt{y_0z_0})=\sqrt{a}^2=a\)
\(\large\textcolor{blue}{微积分第六次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.4.5}\)
习题1.9.8
求函数 \(u=x^{2}+2 y^{2}+z^{2}-2 x y-2 y z\) 在区域 \(x^{2}+y^{2}+z^{2} \leqslant 4\) 内的最值.
解:由 \(u=(x-y)^2+(y-z)^2\geq 0\),当 \(x=y=z=0\) 时能取到最小值 \(0\)
而若最大值在 \(x^2+y^2+z^2=r^2<4\) 时取到,为 \((x_0,y_0,z_0)\) 则线性扩大 \(\dfrac{2}{r}\) 倍,\(u\) 扩大 \(\dfrac{4}{r^2}\) 倍
从而最大值在 \(x^2+y^2+z^2=4\) 时取到,由拉格朗日乘子法 \[ f(x,y,z)=(x-y)^2+(y-z)^2+\lambda(x^2+y^2+z^2-4) \] 有 \(\begin{cases}\dfrac{\partial f}{\partial x}=2(x-y)+2\lambda x=0\\\dfrac{\partial f}{\partial y}=-2(x-y)+2(y-z)+2\lambda y=0\\\dfrac{\partial f}{\partial z}=-2(y-z)+2\lambda z=0\end{cases}\),则有 \(\lambda =\dfrac{y}{x}-1=\dfrac{y}{z}-1=\dfrac{x+z}{y}-2\)
若 \(y=0\),则 \(\lambda=-1,x=-z\),代入 \(u=x^2+z^2=4\)
若 \(y\neq 0\),得 \(x=z,2x^2=xy+y^2\) \(,x^2+y^2+z^2=2x^2+y^2=4=xy+2y^2,x=\dfrac{4}{y}-2y\)
有方程 \(4-y^2=2(\dfrac{4}{y}-2y)^2=\dfrac{32}{y^2}-32+8y^2\Longrightarrow 9y^4-36y^2+32=0,y^2=\dfrac{8}{3},\dfrac{4}{3}\) \[ u=2(x-y)^2=6(y^2-x^2)=6(-3y^2-\dfrac{16}{y^2}+16) \] 代入计算最大为 \(6(-6-8+16)=12\) 此时 \(y=\dfrac{2\sqrt{6}}{3},x=-\dfrac{\sqrt{6}}{3}=z\)
综上所述,\(u\) 在 \(x^2+y^2+z^2\leq 4\) 的最小值为 \(0\),最大值为 \(12\)
习题1.9.9(2)
求旋转抛物面 \(z=x^{2}+y^{2}\) 与平面 \(x+y-z=1\) 的最短距离;
解:显然从几何上,固定旋转抛物面上的点,该点到平面 \(x+y-z=1\) 的最小距离为垂线段 \[ d=\dfrac{|x_0+y_0-z_0-1|}{\sqrt{1^2+1^2+1^2}}=\dfrac{\sqrt{3}}{3}|x_0+y_0-z_0-1|=\dfrac{\sqrt{3}}{3}|x_0+y_0-x_0^2-y_0^2-1| \] 而由配方法 \(x_0-x_0^2=-(x_0-\dfrac{1}{2})^2+\dfrac{1}{4}\leq \dfrac{1}{4}\),则 \(x_0+y_0-x_0^2-y_0^2-1\leq -\dfrac{1}{2}\)
从而取等时对应绝对值最小,则 \(d_{\min }=\dfrac{\sqrt{3}}{6}\)
习题1.9.10(1)
水渠的断面为等腰梯形, 在渠道表面抺上水泥, 当断面面积一定时, 梯形的上、下底及腰的长度比例为多少时, 所用水泥最省?
解:显然所用水泥质量正比于 \(C_{\mbox{trapesoid}}=l_1+2l_{\mbox{waist}}\)
面积为定值 \(S=\dfrac{1}{2}(l_1+l_2)\cdot h\),勾股定理 \(l_{\mbox{waist}}=\sqrt{h^2+(\dfrac{l_2-l_1}{2})^2}\) \[ C=l_1+l_2+\sqrt{4h^2+(l_2-l_1)^2}=l_1+\sqrt{\dfrac{16S^2}{(l_1+l_2)^2}+(l_2-l_1)^2} \] 对 \(l_2\) 求导得 \(-\dfrac{32S^2}{(l_1+l_2)^3}+2(l_2-l_1)=0,8h^2=2(l_2-l_1)(l_1+l_2)\)
即有 \(h^2=\dfrac{1}{4}(l_2^2-l_1^2),l_{\mbox{waist}}=\dfrac{1}{2}\sqrt{(l_2-l_1)\cdot 2l_2},C=l_1+\sqrt{2l_2(l_2-l_1)}\)
在约束条件 \(16S^2=(l_2-l_1)(l_2+l_1)^3\) 下求 \(C(l_1,l_2)=l_1+\sqrt{2l_2(l_2-l_1)}\) 的最小值
作拉格朗日函数 \(f(l_1,l_2,\lambda )=l_1+\sqrt{2l_2(l_2-l_1)}+\lambda(l_2-l_1)(l_2+l_1)^3\),极值点对应
\(\begin{cases}\dfrac{\partial f}{\partial l_1}=1+\dfrac{-2l_2}{2\sqrt{2l_2(l_2-l_1)}}+\lambda(l_2+l_1)^2(2l_2-4l_1)=0\\\dfrac{\partial f}{\partial l_2}=\dfrac{4l_2-2l_1}{2\sqrt{2l_2(l_2-l_1)}}+\lambda(l_2+l_1)^2(4l_2-2l_1)=0\end{cases}\)
令 \(\eta=\dfrac{l_2}{l_1}\) 有 \(\dfrac{\sqrt{2\eta(\eta-1)}-\eta}{2\eta-1}=\dfrac{\eta-2}{2\eta-1}\),解得 \(\eta=2\),代入 \(l_1:l_2:l_{\mbox{waist}}=1:2:1\)
习题2.2.2(2)
求下列函数的导函数。 \(F(y)=\displaystyle \int_{a+y}^{b+y} \frac{\sin y x}{x} \mathrm{~d} x\);
解:对 \(F(y)=\displaystyle\int_{a+y}^{b+y}\dfrac{sin(yx)}{x}dx\),其对 \(x\) 的导函数分为两部分,分类讨论 \[ F'(y)=\dfrac{sin(y(b+y))}{b+y}-\dfrac{sin(y(a+y))}{a+y}+\int_{a+y}^{b+y}cos(xy)dx \] ## 习题2.2.5.(1)
\(\displaystyle \int_{0}^{1} \frac{\arctan x}{x} \frac{1}{\sqrt{1-x^{2}}} \mathrm{~d} x\left(\right.\) 提示: \(\displaystyle \frac{\arctan x}{x}=\int_{0}^{1} \frac{1}{1+x^{2} y^{2}} \mathrm{~d} y\) )
解:由提示变换 \(\dfrac{\arctan x}{x}=\displaystyle\int_0^1\dfrac{1}{1+x^2y^2}dy\),化为二重积分得 \[ \int_0^1\dfrac{\arctan x}{x}\dfrac{1}{\sqrt{1-x^2}}dx=\int_0^1\displaystyle\int_0^1\dfrac{1}{1+x^2y^2}\dfrac{1}{\sqrt{1-x^2}}dxdy \] 代换 \(x=\sin\varphi,\varphi\in[0,\dfrac{\pi}{2}]\),有 \(\dfrac{1}{1+x^2y^2}\dfrac{1}{\sqrt{1-x^2}}dx=\dfrac{d\varphi}{1+y^2\sin^2\varphi}=\dfrac{\cos^2\varphi \cdot d(\tan \varphi)}{(1+y^2)\sin^2\varphi+\cos^2 \varphi}\) \[ \begin{gathered} \int_0^1\dfrac{\arctan x}{x}\dfrac{1}{\sqrt{1-x^2}}dx=\int_0^1(\int_0^{\frac{\pi}{2}}\dfrac{d(\tan \varphi)}{1+(1+y^2)\tan^2\varphi})dy=\\\int_0^1(\dfrac{\arctan (\sqrt{1+y^2}\tan \varphi)}{\sqrt{1+y^2}})\big |_0^{\frac{\pi}{2}}dy =\dfrac{\pi}{2}\int_0^1\dfrac{dy}{\sqrt{1+y^2}}=\\\dfrac{\pi}{2}\ln(y+\sqrt{1+y^2})\big |_0^1=\dfrac{\pi}{2}\ln(1+\sqrt{2}) \end{gathered} \]
习题2.3.1
计算下列积分.
\((1)\) \(\displaystyle \int_{0}^{+\infty} \frac{\mathrm{e}^{-a x^{2}}-\mathrm{e}^{-b x^{2}}}{x} \mathrm{~d} x(a, b>0)\);
\((3)\) \(\displaystyle \int_{0}^{+\infty} \frac{\cos a x-\cos b x}{x^{2}} \mathrm{~d} x(a, b>0)\)
解:\((1)\) 解:\(\displaystyle \int_0^{+\infty}\dfrac{e^{-ax^2}-e^{-bx^2}}{x}dx=\int_0^{+\infty}\dfrac{1}{x}dx\int_a^bx^2e^{-x^2y}dy=\int_0^{+\infty}\int_a^bxe^{-x^2y}dxdy\) \[ \int_0^{+\infty}xe^{-x^2y}dx=\dfrac{1}{2}\int_0^{+\infty}e^{-x^2y}d(x^2)=-\dfrac{1}{2y}\cdot e^{-x^2y}\big |_0^{+\infty}=\dfrac{1}{2y}\\ \int_0^{+\infty}\dfrac{e^{-ax^2}-e^{-bx^2}}{x}dx=\int_a^{b}\dfrac{1}{2y}dy=\dfrac{1}{2}\ln (\dfrac{b}{a}) \] \((3)\) \(\displaystyle \int_0^{+\infty}\dfrac{\cos ax-\cos bx}{x^2}dx\ =-\int_0^{+\infty}\dfrac{1}{x^2}dx\int_b^{a}x\sin(xy)dy=\int_0^{+\infty}\int_a^b\dfrac{\sin (xy)}{x}dxdy\) \[ \mbox{set}\ \ t=xy\in[0,+\infty),\int_0^{+\infty}\dfrac{\sin (xy)}{x}dx=\int_0^{+\infty}\dfrac{y\sin t}{t}d(\dfrac{t}{y})=\int_0^{+\infty}\dfrac{\sin t}{t}dt=\dfrac{\pi}{2} \] 代入得 \(\displaystyle \int_0^{+\infty}\dfrac{\cos ax-\cos bx}{x^2}dx=\int_a^b\dfrac{\pi}{2}dy=\dfrac{\pi}{2}(b-a)\)
习题2.1.4
讨论下列积分在所给区间上的一致收敛性.
\((2)\) \(\displaystyle \int_{-\infty}^{+\infty} \frac{\cos y x}{1+x^{2}} \mathrm{~d} x(-\infty<y<+\infty)\);
\((9)\) \(\displaystyle \int_{1}^{+\infty} x^{1-y} \mathrm{~d} x(0<y<+\infty)\);
\((10)\) \(\displaystyle \int_{0}^{+\infty} \frac{\sin x^{2}}{x^{p}} \mathrm{~d} x(0 \leqslant p<+\infty)\).
解:\((2)\) 由 \(|\dfrac{\cos yx}{1+x^2}|\leq \dfrac{1}{1+x^2}\),且有 ${-}^{+}dx=x|{-}^{+}=$
对于 \(\displaystyle \int_{-\infty}^{+\infty}|\dfrac{\cos yx}{1+x^2}|dx=2\int_0^{+\infty}|\dfrac{\cos yx}{1+x^2}|dx\) 收敛,由柯西收敛准则 \[ \forall\ \epsilon >0,\exists\ A>0,s.t.\forall \ A_1>A_2>A,\int_{A_1}^{A_2}|\dfrac{\cos yx}{1+x^2}|dx<\epsilon \] 放缩绝对值有 \(\forall\ \epsilon >0,\exists\ A>0,s.t.\forall\displaystyle \ A_1>A_2>A,|\int_{A_1}^{A_2}\dfrac{\cos yx}{1+x^2}dx|\leq \int_{A_1}^{A_2}|\dfrac{\cos yx}{1+x^2}|dx<\epsilon\)
由柯西收敛准则得原广义含参积分收敛
\((9)\) 当 \(0<y\leq 2\) 时 \(x^{1-y}\geq x^{-1}=\dfrac{1}{x}\),则 \[ \exists\ \epsilon_0= \ln 2,\forall \ A>1,s.t.\exists \ A_2>2A_1>A,\displaystyle \int _{A_1}^{A_2}x^{1-y}dx>\int_{A_1}^{A_2}\dfrac{1}{x}dx=\ln x\big |_{A_1}^{A_2}>\epsilon_0 \] 此时广义含参积分不收敛;若当 \(2<y\) 时,直接计算有 \(\dfrac{1}{2-y}\cdot (0-1)=\dfrac{1}{y-2}\) 收敛
\((10)\) 代换 \(t=x^2\in [0,\infty)\),\(\displaystyle \int_{0}^{+\infty}\dfrac{\sin x^2}{x^p}dx=\int_{0}^{+\infty}\dfrac{\sin t}{t^{\frac{p}{2}}}d(\sqrt{t})=\int_{0}^{+\infty}\dfrac{\sin t}{t^{\frac{p+1}{2}}}dt\)
其中 \(\alpha=\dfrac{p+1}{2}\geq \dfrac{1}{2},\displaystyle \int_{1}^{+\infty}\dfrac{\sin t}{t^{\alpha}}dt=\int_{1}^{+\infty}\dfrac{-d(\cos t)}{t^{\alpha }}=-\dfrac{\cos t}{t^{\alpha }}\big |_{1}^{+\infty}-\int_{1}^{+\infty}\dfrac{\alpha dt}{t^{\alpha+1}}\)
\(=\cos 1-\alpha \displaystyle \int_{1}^{+\infty}\dfrac{\cos t dt}{t^{\alpha +1}}\),而放缩 \(\displaystyle \int_{1}^{+\infty}|\dfrac{\cos t}{t^{\alpha +1}}|dt\leq \int_{1}^{+\infty}\dfrac{dt}{t^{\alpha +1}}=\dfrac{1}{\alpha }\),后者绝对收敛
从而 \(\displaystyle \int_{1}^{+\infty}\dfrac{\sin t}{t^{\alpha}}dt\) 一致收敛,只需考虑 \(\displaystyle \int_{0}^{1}\dfrac{\sin t}{t^{\alpha}}dt\),这与 \(\lim\limits_{t\to 0}\dfrac{\sin t }{t^{\alpha }}\) 的收敛性相同
当且仅当 \(\alpha =\dfrac{p+1}{2}\leq 1\) 即 \(0\leq p\leq 1\) 时原广义含参积分收敛,\(1<p\) 时不收敛
\(\large\textcolor{blue}{微积分第七次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.4.24}\)
习题3.2.3
比较下列各组积分值的大小.
\((2)\) \(\displaystyle \iint_{D} \ln (x+y) \mathrm{d} x \mathrm{~d} y ,\displaystyle \iint_{D} x y \mathrm{~d} x \mathrm{~d} y\), \(D\) 由直线 \(x=0, y=0, x+y=\dfrac{1}{2}\), \(x+y=1\)
围成.
解:去掉边界积分结果不变,由于 \(x+y< 1\),则 \(\ln (x+y)< 0\) 而 \(xy> 0\),且积分区域 \(x,y>0\)
从而 \(\ln(x+y)dxdy<xydxdy\) 则 \(\displaystyle \iint_{D} \ln (x+y) \mathrm{d} x \mathrm{~d} y<\displaystyle \iint_{D} x y \mathrm{~d} x \mathrm{~d} y\)
习题3.3.6
计算下列二重积分.
\((1)\) \(\displaystyle \iint_{D} x y^{2} \mathrm{~d} x \mathrm{~d} y, D=\left\{(x, y) \mid 4 x \geqslant y^{2}, x \leqslant 1\right\}\);
\((2)\) \(\displaystyle \iint_{D} \frac{1}{\sqrt{2 a-x}} \mathrm{~d} x \mathrm{~d} y, D=\left\{(x, y) \mid(x-a)^{2}+(y-a)^{2} \leqslant 1,0 \leqslant x, y \leqslant a\right\}\);
\((4)\) \(\displaystyle \iint_{D} x \cos (x y) \mathrm{d} x \mathrm{~d} y, D=\left\{(x, y) \mid x^{2}+y^{2} \leqslant R^{2}\right\}\);
\((9)\) \(\displaystyle \iint_{D} y^{2} \mathrm{~d} x \mathrm{~d} y, D\) 由 \(\left\{\begin{array}{l}x=a(t-\sin t), \\ y=a(1-\cos t),\end{array} \quad 0 \leqslant t \leqslant 2 \pi\right.\) 以及 \(x\) 轴围成;
解:\((1)\) 积分区域图像如右 \(\begin{cases}x\geq \dfrac{1}{4}y^2\\x\leq 1\end{cases}\) \[ \displaystyle \iint_{D} x y^{2} \mathrm{~d} x \mathrm{~d} y=\int_{0}^1dx\int_{-2\sqrt{x}}^{2\sqrt{x}}xy^2dy=\int_0^1x\cdot \dfrac{2}{3}(2\sqrt{x})^3dx=\dfrac{16}{3}\int_0^1x^{\frac{5}{2}}dx=\dfrac{32}{21} \] \((2)\) 积分区域 \(\begin{cases}(x-a)^2+(y-a)^2\leq a^2\\0\leq x,y\leq a\end{cases}\) 图像如右 \[ \begin{gathered}\underset{D}{\iint}\dfrac{1}{\sqrt{2a-x}}dxdy=\int_0^{a}dx\int_{a-\sqrt{x(2a-x)}}^a\dfrac{1}{\sqrt{2a-x}}dy\\=\int_0^{a}\sqrt{x(2a-x)}\cdot \dfrac{1}{\sqrt{2a-x}}dx=\dfrac{2}{3}a^{\frac{3}{2}} \end{gathered} \]
\((4)\) 积分区域 \(D:x^2+y^2\leq R^2\) 图像如右
拆成两部分 \(\displaystyle \underset{D}{\iint}x\cos (xy)dxdy=\underset{D\ \cap\ \{x|x\leq 0\}}{\iint}x\cos (xy)dxdy+\underset{D\ \cap\ \{x|x>0\}}{\iint}x\cos (xy)dxdy\)
对 \(f(x,y)=x\cos (xy)\) 有 \(f(x,y)=-f(-x,y)\) 为关于 \(x\) 的奇函数,则 \(\displaystyle \underset{D}{\iint}x\cos (xy)dxdy=0\)
\[ \underset{D}{\iint}|\cos (x+y)|dxdy=\underset{D'}{\iint}|\cos (x+y)|\dfrac{D (x,y)}{D(v,u)}dvdu=\underset{D}{\iint}|\cos (x+y)|\begin{vmatrix}\dfrac{1}{2}&\dfrac{1}{2}\\{-}\dfrac{1}{2}&\dfrac{1}{2}\end{vmatrix}dudv\\ =\dfrac{1}{2}\underset{D}{\iint}|\cos (x+y)|dvdu=\dfrac{1}{2}\cdot 4\int_0^{\pi}|\cos u|du\int_0^{u}dv=2\cdot 2\int_0^{\frac{\pi}{2}}ud(\sin u)=2\pi-4 \]
\((9)\) 解:滚轮线 \(\begin{cases}x=a(t-\sin t)\\y=a(1-\cos t)\end{cases}\ \small t\in[0,2\pi],\normalsize D\scriptsize:\normalsize \begin{cases}y\leq f(x)\\0\leq y\end{cases}\) 图像
\(\displaystyle\underset{D}{\iint}y^2dxdy=\int_0^{2\pi a} dx\int_{0}^{f(x)} y^2dy=\dfrac{1}{3}\int_0^{2\pi a} f(x)^3dx\),换成参数 \(t\) 积分得 \[ \dfrac{1}{3}\int_0^{2\pi} a^{4}(1-\cos t)^{3}d(t-\sin t)=\dfrac{a^4}{3}\int_0^{2\pi}(1-\cos t)^4dt=\dfrac{a^4}{3}\int_{-\pi}^{\pi}(1+\cos x)^4dx=\\ \dfrac{a^4}{3}\int_{-\pi}^{\pi}(1+4\cos x+6\cos ^2x+4\cos ^3x+\cos^4x)dx=\dfrac{a^4}{3}\int_{-\pi}^{\pi}(1+6\cos ^2x+\cos^4x)dx\\ =\dfrac{a^4}{3}(2\pi+\int_{-\pi}^{\pi}3(1+\cos 2x)dx+\int_{-\pi}^{\pi}\dfrac{(1+\cos 2x)^2}{4}dx=\dfrac{a^4}{3}(8\pi+\int_{-\pi}^\pi{\dfrac{1+\dfrac{1}{2}}{4}}dx+0)\\ =\dfrac{a^4}{3}(8\pi+\dfrac{3\pi}{4})=\dfrac{35\pi}{12}a^4\ \ \ \ (答案有误,分母缺少3) \]
习题3.3.12
\((3)\) \(\displaystyle \iint_{D}(x+y) \mathrm{d} x \mathrm{~d} y, D\) 是由 \(x^{2}+y^{2}=x+y\) 围成的平面区域
解:该区域为一个圆 \((x-\dfrac{1}{2})^2+(y-\dfrac{1}{2})^2=(\dfrac{\sqrt{2}}{2})^2\) 换元 \(\begin{cases}x=\dfrac{1}{2}+r\cos \theta\\y=\dfrac{1}{2}+r\sin \theta\end{cases}\)
得到 \(dxdy=rd\theta dr\) 计算有 \(\displaystyle \iint _{D}(x+y)dxdy=\iint _{D}(1+r\cos \theta+r\sin \theta)rd\theta dr\) \[ \begin{gathered} \iint _{D}(1+r\cos \theta+r\sin \theta)rd\theta dr=\int_0^{\frac{\sqrt{2}}{2}}dr\int_{0}^{2 \pi}(1+r\cos \theta+r\sin \theta)\cdot rd\theta\\ =\int_0^{\frac{\sqrt{2}}{2}}dr(2\pi+0+0)\cdot r=\pi r^2\big |_{0}^{\frac{\sqrt{2}}{2}}=\dfrac{\pi}{2} \end{gathered} \]
习题3.3.16
设函数 \(f(t)\) 连续, 证明:
\((2)\) \(\displaystyle \iint_{|x|+|y| \leqslant 1} f(x y) \mathrm{d} x \mathrm{~d} y=\ln 2 \int_{1}^{2} f(t) \mathrm{d} t, D\) 是由 \(x y=1, x y=2, y=x, y=4 x\) 所围成
第一象限的区域.
解:代换 \(xy=u,\dfrac{y}{x}=v\),逆变换 \(x=\sqrt{\dfrac{u}{v}},y=\sqrt{uv}\) ,则雅克比行列式 \[ \dfrac{D(x,y)}{D(u,v)}=\begin{vmatrix}\dfrac{\partial x}{\partial u}&\dfrac{\partial x}{\partial v}\\\dfrac{\partial y}{\partial u}&\dfrac{\partial x}{\partial v}\end{vmatrix}=\begin{vmatrix}\dfrac{1}{2\sqrt{uv}}&-\dfrac{1}{2}\sqrt{\dfrac{u}{v^3}}\\\dfrac{1}{2}\sqrt{\dfrac{v}{u}}&\dfrac{1}{2}\sqrt{\dfrac{u}{v}}\end{vmatrix}=\dfrac{1}{2v} \] 原区域 \(D:\begin{cases}1\leq xy\leq 2\\1\leq \dfrac{y}{x}\leq 4\end{cases}\) 变换后对应 \(D':\begin{cases}1\leq u\leq 2\\1\leq v\leq 4\end{cases}\) 图像如右 \[ \underset{D}{\iint}f(xy)dxdy=\underset{D'}{\iint}f(u)\cdot \dfrac{1}{2v}dudv=\int_1^2f(u)du\int_1^4\dfrac{dv}{2v}=\ln 2\int_1^2f(t)dt \]
习题3.4.5
计算下列三重积分的值.
\((1)\) \(\displaystyle \iiint_{\Omega} x y^{2} z^{3} \mathrm{~d} x \mathrm{~d} y \mathrm{~d} z, \Omega\) 是由马鞍面 \(z=x y\) 与平面 \(y=x, x=1, z=0\) 所围成空间区域;
\((3)\) \(\displaystyle \iiint_{\Omega} x \cos (y+z) \mathrm{d} x \mathrm{~d} y \mathrm{~d} z, \Omega\) 是由曲面 \(x=\sqrt{y}\) 与平面 \(x=0, z=0, y+z\) \(=\dfrac{\pi}{2}\) 围成区域;
解:\((1)\) 从 \(z=0\) 连续变化到 \(z=1\),图中画出 \(z=\dfrac{1}{2}\) 的情况
对每个 \(z\) 二重积分得到 \[ \begin{gathered} \displaystyle \iiint_{\Omega} x y^{2} z^{3} \mathrm{~d} x \mathrm{~d} y \mathrm{~d} z=\int_0^1dz\int_{\sqrt{z}}^1dx\int_{\frac{z}{x}}^xxy^2z^3dy=\int_0^1z^3dz\int_{\sqrt{z}}^1xdx\dfrac{y^3}{3}\big|_{\frac{z}{x}}^x\\ =\dfrac{1}{3}\int_0^1z^3dz\int_{\sqrt{z}}^1xdx(x^3-\dfrac{z^3}{x^3})=\dfrac{1}{3}\int_0^1z^3dz(\dfrac{1}{5}x^5+\dfrac{z^3}{x})\big|_{\sqrt{z}}^1 \\=\dfrac{1}{3}\int_0^1z^3(\dfrac{1}{5}-\dfrac{1}{5}z^{\frac{5}{2}}+z^3-z^{\frac{5}{2}})dz=\dfrac{1}{3}(\dfrac{1}{20}-\dfrac{6}{5}\cdot \dfrac{2}{13}+\dfrac{1}{7})=\dfrac{1}{364} \end{gathered} \] \((3)\) 对 \(z=0\) 连续变化到 \(z=\dfrac{\pi}{2}\) ,图像如右 \[ \begin{gathered} \displaystyle \iiint_{\Omega} x \cos (y+z) \mathrm{d} x \mathrm{~d} y \mathrm{~d} z=\int_0^{\frac{\pi}{2}}dz\int_0^{\frac{\pi}{2}-z}dy\int_0^{\sqrt{y}}x\cos (y+z)dx\\ =\dfrac{1}{2}\int_0^{\frac{\pi}{2}}dz\int_0^{\frac{\pi}{2}-z}dy\cdot y\cos (y+z)=\dfrac{1}{2}\int_0^{\frac{\pi}{2}}dz(y\sin (y+z)\big |_{0}^{\frac{\pi}{2}-z}-\int_0^{\frac{\pi}{2}-z}\sin (y+z)dy)\\ =\dfrac{1}{2}\int_0^{\frac{\pi}{2}}dz((\dfrac{\pi}{2}-z)+\cos (y+z)\big |_{0}^{\frac{\pi}{2}-z})=\dfrac{1}{2}\int_0^{\frac{\pi}{2}}((\dfrac{\pi}{2}-z)-\cos z)dz\\ =\dfrac{1}{2}(\dfrac{\pi^2}{4}-\dfrac{\pi^2}{8})-\dfrac{1}{2}=\dfrac{\pi^2-8}{16} \end{gathered} \]
习题3.4.6
计算累次积分 \(\displaystyle I=\int_{0}^{1} \mathrm{~d} x \int_{0}^{x} \mathrm{~d} y \int_{0}^{y} \frac{\cos z}{(1-z)^{2}} \mathrm{~d} z\) 的值.
解:三维图形为三棱锥,图像如右(使用 \(\mbox{meshlab}\) 绘图)
在 \(z=z_0\) 切面上由面积相似比等于高之比的平方 \(\eta=(1-z)^2\),从而有 \[ \int_0^{1}dx\int_0^xdy\int_0^y\dfrac{\cos z}{(1-z)^2}dz=\int_0^1\dfrac{\cos z}{(1-z)^2}\cdot Sdz=\int_0^1\dfrac{\cos z}{(1-z)^2}\dfrac{1}{2}(1-z)^2dz=\dfrac{\sin 1}{2} \]
习题3.4.7
计算下列三重积分的值.
\((4)\) \(\displaystyle \iiint_{\Omega} x \mathrm{e}^{\frac{x^{2}+y^{2}+z^{2}}{a^{2}}} \mathrm{~d} x \mathrm{~d} y \mathrm{~d} z, \Omega=\left\{(x, y, z) \mid x^{2}+y^{2}+z^{2} \leqslant a^{2}, x, y, z \geqslant 0\right\}\)
解:使用球坐标 \((x,y,z)=(r,\theta,\varphi)\) 有 \(dxdydz=r^2\sin \theta dr d\theta d\varphi\) 计算积分有 \[ \begin{gathered} \displaystyle \iiint_{\Omega} x \mathrm{e}^{\frac{x^{2}+y^{2}+z^{2}}{a^{2}}} \mathrm{~d} x \mathrm{~d} y \mathrm{~d} z=\displaystyle \iiint_{\Omega} r\sin \theta\sin \varphi e^{\frac{r^2}{a^2}}r^2\sin \theta dr d\theta d\varphi\\ = \int_{0}^{\frac{\pi}{2}}\dfrac{1-\cos 2\theta}{2} d\theta \int_{0}^{a}r^{3}e^{\frac{r^2}{a^2}}dr=\int_0^{a^2}\dfrac{\pi}{8}te^{\frac{t}{a^2}}dt=\dfrac{\pi}{8}(a^2t-a^4)e^{\frac{t}{a^2}}\big |_{0}^{a^2}=\dfrac{ \pi a^4}{8} \end{gathered} \]
习题3.4.10
\(f(t)\) 在 \((-\infty,+\infty)\) 连续, \(f(t)=3\displaystyle \iiint_{x^{2}+y^{2}+z^{2} \leqslant t^{2}} f\left(\sqrt{x^{2}+y^{2}+z^{2}}\right) \mathrm{d} x \mathrm{~d} y \mathrm{~d} z+\) \(\left|t^{3}\right|\), 求 \(f(t)\)
解:积分区域为球体 \(x^2+y^2+z^2\leq t^2\),换取球坐标 \((x,y,z)\longrightarrow(\rho,\theta,\varphi)\),有体积元变换 \[ dxdydz=\rho^2\sin \theta d\rho d\theta d\varphi \]
\[ f(t)=3\int_0^{2\pi}\int_0^{\pi}\int_0^tf(\rho)\rho^2\sin \theta d\rho d\theta d\varphi+|t^3|\\ =3\cdot2\cdot2\pi\int_0^tf(\rho)\rho^2d\rho+|t^3|=12\pi\int_0^tf(\rho)\rho^2d\rho+|t^3| \]
两边对 \(t\) 求导得 \[ f'(t)=12\pi t^2 f(t)+\mbox{sign}(t)\cdot 3t^2,\mbox{where}\ \mbox{sign}(x)=\begin{cases}1,x>0\\0,x=0\\-1,x<0\end{cases} \] 其特解与 \(t\) 的符号有关 \(\begin{cases}t>0,f_0=-\dfrac{1}{4\pi}\\t<0,f_0=\dfrac{1}{4\pi}\end{cases}\),其通解满足方程 \[ \dfrac{df}{dt}=12\pi t^2f\Longrightarrow \int\dfrac{df}{f}=\int12\pi t^2dt\ ,\ \ln f=4\pi t^3+C \ \therefore f(t)=\begin{cases}C_1e^{4\pi t^3}-\dfrac{1}{4\pi},t>0\\C_2e^{4\pi t^3}+\dfrac{1}{4\pi},t<0\end{cases} \] 由 \(f(t)\) 在 \((-\infty,+\infty)\) 连续得 \(C_1-\dfrac{1}{4\pi}=C_2+\dfrac{1}{4\pi}\),有 \(C_1=C_2+\dfrac{1}{2\pi}\),对 \(t>0\) 代入原积分 \[ \begin{gathered} f(t)=C_1e^{4\pi t^2}-\dfrac{1}{4\pi}=12\pi\int_0^t(C_1e^{4\pi \rho^3}-\dfrac{1}{4\pi})\rho^2d\rho+t^3\\=\int_0^tC_1e^{4\pi\rho^3}d(4\pi \rho^3)+0=C_1(e^{4\pi t^2}-1) \end{gathered} \] 比对得 \(C_1=\dfrac{1}{4\pi}\),\(C_2=-\dfrac{1}{4\pi}\),从而 \(f(t)=\dfrac{|e^{4\pi t^3}-1|}{4\pi}\)
习题3.5.1
求下列曲面的面积.
\((2)\) 锥面 \(z=\sqrt{x^{2}+y^{2}}\) 在柱面 \(z^{2}=2 x\) 内的部分;
解:\(z=\sqrt{x^2+y^2}\) 和 \(z^2=2x\) 在 \(y=0\) 处截图如下
取柱坐标 \((x,y,z)\longmapsto(\rho,\theta,z)\) 则锥面 \(z=\rho,z^2=2\rho \cos \theta\) 其要求 \(\rho \leq 2\cos \theta\) 为圆
\(\sqrt{1+\left(\dfrac{\partial z}{\partial x}\right)^{2}+\left(\dfrac{\partial z}{\partial y}\right)^{2}}=\sqrt{2}\) 则 \(S=\displaystyle \iint_D\sqrt{2}dxdy=\sqrt{2}\cdot S_{圆}=\sqrt{2}\pi\)
习题3.5.2
求下列曲面所包围的均匀物体的质心.
\((3)\) \(z=4-x^{2}(x \geqslant 0), x=0, y=0, z=0, y=6\).
解:由对称性该物体的质心 \(y\) 坐标为 \(3\) ,且质心的 \(x,z\) 坐标与平面内抛物三角形相同 \[ x_c=\dfrac{\displaystyle \int_0^2x(4-x^2)dx}{\displaystyle \int_0^2(4-x^2)dx}=\dfrac{4}{\frac{16}{3}}=\dfrac{3}{4},z_c=\dfrac{\displaystyle \int_0^2\dfrac{4-x^2}{2}(4-x^2)dx}{\displaystyle \int_0^2(4-x^2)dx}=\dfrac{\frac{128}{15}}{\frac{16}{3}}=\dfrac{8}{5} \] 则质心坐标为 \((\dfrac{3}{4},3,\dfrac{8}{5})\)
习题4.2.1
计算下列曲线积分.
\((4)\) \(\displaystyle \int_{L}\left(x^{\frac{4}{3}}+y^{\frac{4}{3}}\right) \mathrm{d} l\), 其中 \(L\) 为星形线 \(\left\{\begin{array}{l}x=a \cos ^{3} t, \\ y=a \sin ^{3} t,\end{array} \quad 0 \leqslant t \leqslant 2 \pi\right.\).
解:由曲线微元 \(dl=\sqrt{dx^2+dy^2}=3a|\cos t\sin t|\) 代入有 \[ \begin{gathered} \int_{L}(x^{\frac{4}{3}}+y^{\frac{4}{3}})3a|\cos t\sin t |dt=3\int_0^{2\pi}a^{\frac{7}{3}}(\cos^4 t+\sin ^4 t)|\cos t\sin t|dt\\ =12a^{\frac{7}{3}}\int_0^{\frac{\pi}{2}}(1-\dfrac{1}{4}\sin^2 2t)\dfrac{1}{2}\sin 2tdt=12a^{\frac{7}{3}}(\dfrac{1}{2}-\int_0^{\frac{\pi}{2}}\dfrac{1}{8}(1-\cos 4t )\sin 2tdt)\\ =12a^{\frac{7}{3}}(\dfrac{3}{8}-\dfrac{1}{8}\int_0^{\frac{\pi}{2}}\dfrac{\sin 6t-\sin 2t}{2}dt=12a^{\frac{7}{3}}(\dfrac{3}{8}-\dfrac{1}{8}(-\dfrac{1}{12}\cos 6t+\dfrac{1}{4}\cos 2t)\big |_0^{\frac{\pi}{2}})\\ =12a^{\frac{7}{3}}(\dfrac{3}{8}-\dfrac{1}{8}(\dfrac{1}{6}-\dfrac{1}{2}))=4a^{\frac{7}{3}} \end{gathered} \]
习题4.2.2
计算下列曲线积分.
\((4)\) \(\displaystyle \int_{L} x \mathrm{~d} l\), 其中 \(L\) 为球面 \(x^{2}+y^{2}+z^{2}=4\) 在第一象限部分的边界.
解:由题设求和即为 \(y=\sqrt{4-x^2}\) 位于 \(xy\) 象限和 \(xz\) 象限的部分,有 \(\begin{cases}x=2\cos \theta\\y=2\sin \theta\end{cases},\theta\in [0,\dfrac{\pi}{2}]\) \[ \int_{L}xdl=2\cdot \int_0^{\frac{\pi}{2}}2\cos \theta\cdot 2d\theta=2\cdot 4=8 \]
习题4.3.3
求抛物面 \(2 z=x^{2}+y^{2}\) 在 \(z \in[0,1]\) 部分的质量, 其中质量面密度为 \(\sigma=z\).
解:由面积微元,\[\sqrt{1+\left(\dfrac{\partial z}{\partial x}\right)^{2}+\left(\dfrac{\partial z}{\partial y}\right)^{2}}=\sqrt{1+x^2+y^2}\] 代入质量面密度 \[ \begin{gathered} m=\iint \sqrt{1+x^2+y^2}\cdot zdxdy=\int_{0}^{2\pi}d\theta\int_0^{\sqrt{2}}\sqrt{1+r^2}\dfrac{r^2}{2}rdr\\ =\dfrac{\pi}{2}\int_0^{\sqrt{2}}r^{2}\sqrt{1+r^2}d(r^2)=\dfrac{\pi}{2}\int_0^2t\sqrt{1+t}dt=\dfrac{\pi}{2}(\dfrac{2}{3}t(1+t)^{\frac{3}{2}}-\dfrac{4}{15}(1+t)^{\frac{5}{2}})\big|_0^{2}\\ =\dfrac{\pi}{2}(\dfrac{4}{3}\cdot 3\sqrt{3}-\dfrac{4}{15}(9\sqrt{3}-1))=\dfrac{(12\sqrt{3}+2)\pi}{15} \end{gathered} \]
习题4.3.11
证明: (\(\mbox{Poisson}\) 公式) \(\displaystyle \iint_{S} f(a x+b y+c z) \mathrm{d} S=2 \pi \int_{-1}^{1} f\left(\sqrt{a^{2}+b^{2}+c^{2}} t\right) \mathrm{d} t\), 其中
\(S=\left\{(x, y, z) \mid x^{2}+y^{2}+z^{2}=1\right\}, f\) 是连续函数.
证明:对球面的面积微元 \(z=\pm \sqrt{1-x^2-y^2}\) \(\sqrt{1+\left(\dfrac{\partial z}{\partial x}\right)^{2}+\left(\dfrac{\partial z}{\partial y}\right)^{2}}=\dfrac{1}{\sqrt{1-x^2-y^2}}\)
由球面的对称性,旋转球面的基向量到 \(\dfrac{1}{\sqrt{a^2+b^2+c^2}}\) 方向,积分结果为定值 \[ \begin{gathered} \displaystyle \iint_{S} f(a x+b y+c z) \mathrm{d} S=\iint_{S} f(\sqrt{a^2+b^2+c^2}z) \mathrm{d} S=2\iint_{S_{xy}}\dfrac{f(\sqrt{a^2+b^2+c^2}z)}{\sqrt{1-x^2-y^2}}dxdy \\=2\int_0^{2\pi}d\theta\int_0^{1}\dfrac{f(\sqrt{a^2+b^2+c^2}\sqrt{1-r^2})}{\sqrt{1-r^2}}rdr\\=4\pi\int_{0}^{1}f(\sqrt{a^2+b^2+c^2}\sqrt{1-r^2})d(\sqrt{1-r^2})=2 \pi \int_{-1}^{1} f\left(\sqrt{a^{2}+b^{2}+c^{2}} t\right) \mathrm{d} t \end{gathered} \]
\(\large\textcolor{blue}{微积分第八次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.5.15}\)
习题4.5.3
\((5)\) \(\displaystyle \iint_{S^{+}} z^{2} \mathrm{~d} x \wedge \mathrm{d} y\), 其中 \(S^{+}\)是 \(z=\sqrt{R^{2}-x^{2}-y^{2}}\) 被柱面 \(x^{2}+y^{2}=R x\) 所截部分的上侧.
解:原积分化为二重积分,代入条件 \(z^2=R^2-x^2-y^2\) \[ \begin{gathered} \displaystyle \iint_{S^{+}} z^{2} \mathrm{~d} x \wedge \mathrm{d} y=\iint_{D}(R^2-x^2-y^2)dxdy=2\pi \int_{0}^{\frac{R}{2}}(R^2-(\dfrac{R}{2}+r\cos \varphi)^2-(\dfrac{R}{2}+r\sin \varphi)^2)rdr\\ 2\pi \int_{0}^{\frac{R}{2}}(\dfrac{3}{4}R^2-r^2)rdr=2\pi(\dfrac{3}{32}-\dfrac{1}{4}\cdot \dfrac{1}{16})R^4=\dfrac{5\pi}{32}R^4 \end{gathered} \]
习题4.5.5
求流速场 \(\boldsymbol{V}=x y \boldsymbol{i}+y z \boldsymbol{j}+z x \boldsymbol{k}\) 由里往外穿过球面 \(x^{2}+y^{2}+z^{2}=1\) 在第一象限部分的流量.
解:记平面区域 \(D\) 为圆心在原点,半径为 \(1\) 的圆在第一象限的部分,由对称性有 \[ Q=\iint_{D} x y \mathrm{~d} y \wedge \mathrm{d} z+y z \mathrm{~d} z \wedge \mathrm{d} x+z x \mathrm{~d} x \wedge \mathrm{d} y=3 \iint_{D} z x \mathrm{~d} x \wedge \mathrm{d} y=3\iint_{D}\sqrt{1-x^2-y^2}xdxdy \] 化为极坐标进行积分(中间使用三角换元)有 \[ Q=3\iint_{D}\sqrt{1-r^2}r\cos \varphi rdrd\varphi=3\int_{0}^{\frac{\pi}{2}}\cos \varphi d\varphi \int_0^1\sqrt{1-r^2}r^2dr=3\int_{0}^{\frac{\pi}{2}}\sin ^2\theta \cos ^2\theta d\theta=\dfrac{3\pi}{16} \]
习题4.5.7
\(\displaystyle \iint_{S^{+}}\left(x^{2}+y^{2}\right) \mathrm{d} x \wedge \mathrm{d} y+y^{2} \mathrm{~d} y \wedge \mathrm{d} z+z^{2} \mathrm{~d} z \wedge \mathrm{d} x\), 其中 \(S\) 是螺旋面 \(x=\) \(u \cos v, y=u \sin v, z=a v\) 在 \[ D_{u v}=\{(u, v) \mid 0 \leqslant u \leqslant 1,0 \leqslant v \leqslant 2 \pi\} \] 的部分, 上侧为正.
解:代入微分式,\(\mathrm{d} x \mathrm{~d} y=u \mathrm{~d} u \mathrm{~d} v, \mathrm{~d} y \mathrm{~d} z=a \sin v \mathrm{~d} u \mathrm{~d} v, \mathrm{~d} z \mathrm{~d} x=-a \cos v \mathrm{~d} u \mathrm{~d} v\) 化为二重积分
\[ \begin{gathered} \iint_{S^{+}}\left(x^{2}+y^{2}\right) \mathrm{d} x \wedge \mathrm{d} y+y^{2} \mathrm{~d} y \wedge \mathrm{d} z+z^{2} \mathrm{~d} z \wedge \mathrm{d} x=\iint_{D uv}\left(u^{3}+a u^{2} \sin ^{3} v-a^{3} v^{2} \cos v\right) \mathrm{d} u \mathrm{~d} v\\ =\iint_{D_{uv}}(u^3-a^3v^2\cos v)\mathrm{d} u \mathrm{~d} v+0=\int_0^1u^3\mathrm{d} u \int_0^{2\pi}\mathrm{~d} v-a^3\int_0^{2\pi}v^2\cos v\mathrm{~d} v=\dfrac{\pi}{2}-4\pi a^3 \end{gathered} \]
习题4.7.2
计算曲面积分 \(\displaystyle {\int\kern{-8pt}\int \kern{-23mu} \bigcirc}_{S^{+}} \frac{x \mathrm{~d} y \wedge \mathrm{d} z+y \mathrm{~d} z \wedge \mathrm{d} x+z \mathrm{~d} x \wedge \mathrm{d} y}{\left(x^{2}+y^{2}+z^{2}\right)^{\frac{3}{2}}}\), 其中 \(S^{+}\)为:
\((1)\) 不包含也不经过原点的半径为 \(R\) 的球面外侧;
\((2)\) 不包含也不经过原点的任意封闭曲面的外侧;
\((3)\) 球面 \(x^{2}+y^{2}+z^{2}=\varepsilon^{2}(\varepsilon>0)\);
\((4)\) 包含原点在其内部的任意封闭曲面的外侧.
解:\((1)\) 令 \(r=\sqrt{x^2+y^2+z^2}\) 有 \(\boldsymbol V=(\dfrac{x}{r^3},\dfrac{y}{r^3},\dfrac{z}{r^3})\) 且由高斯公式 \[ \displaystyle \iint_{\partial \Omega^{+}} \boldsymbol{V} \cdot \mathrm{d} \boldsymbol{S}=\iiint_{\Omega}\left(\frac{\partial X}{\partial x}+\frac{\partial Y}{\partial y}+\frac{\partial Z}{\partial z}\right) \mathrm{d} x \mathrm{~d} y \mathrm{~d} z=\iiint_{\Omega}\left(\dfrac{3r^2-3x^2-3y^2-3z^2}{r^5}\right) \mathrm{d} x \mathrm{~d} y \mathrm{~d} z \] 可得,无论是否为球面,只要不包含原点,积分结果为 \(0\)
\((2)\) 由上问知,积分结果为 \(0\)
\((3)\) 直接化为二重积分,注意球面上下积分贡献为 \(2\) 倍 \[ \displaystyle I=\frac{3}{\epsilon^{3}} {\int\kern{-8pt}\int \kern{-23mu} \bigcirc}_{S^{+}} z \mathrm{~d} x \wedge \mathrm{d} y=\frac{3}{\epsilon^{3}} \cdot 2\int_{0}^{\frac{\pi}{2}} \int_{0}^{2 \pi} \epsilon^{3} \sin \theta \cos ^{2} \theta \mathrm{d} \varphi \mathrm{d} \theta=4 \pi \] \((4)\) 取两个封闭曲面,一个为以原点为圆心的球,此球总能完全属于该封闭曲面,另一个封闭曲面挖掉洞
\(I=0+4\pi=4\pi\),这与静电场点电荷对应的高斯定理 \(\displaystyle 4\pi KQ_{in}={\int\kern{-8pt}\int \kern{-23mu} \bigcirc}\boldsymbol{E}\cdot dS\) 等价
习题4.7.5
利用 \(\mbox{Stokes}\) 公式计算下列积分.
\((1)\) \(\displaystyle \oint_{L^{+}} y \mathrm{~d} x+z \mathrm{~d} y+x \mathrm{~d} z\), 其中 \(L\) 是球面 \(x^{2}+y^{2}+z^{2}=R^{2}\) 与平面 \(x+y+\) \(z=0\) 的交线, 从 \(z\) 轴正向
看上去为逆时针方向;
解:由 \(\mbox{Stokes}\) 公式计算将曲线积分化成二重积分结合椭圆面积二次型知识有 \[ \begin{gathered} I=-3 \iint_{D} 1 \mathrm{~d} x \mathrm{~d} y=-3 \cdot\left(\pi ab\right)\quad 2x^2+2y^2+2xy=R^2\Longrightarrow \lambda_{1,2}=3,1,a=1,b=\dfrac{1}{\sqrt{3}} \\\therefore I =-\sqrt{3} \pi R^{2} \end{gathered} \]
第四章总复习题 7
证明: 由闭曲面 \(S\) 包围的空间体的体积为 \(V=\displaystyle \frac{1}{3} {\int\kern{-8pt}\int \kern{-23mu} \bigcirc}_{S}(x \cos \alpha+y \cos \beta+\) \(z \cos \gamma) \mathrm{d} S\), 其中
\(\boldsymbol{n}=(\cos \alpha, \cos \beta, \cos \gamma)\) 为 \(S\) 的单位外法向量.
解:由高斯公式可得 \[ \displaystyle V=\iiint_D\mathrm dx\mathrm dy\mathrm dz=\dfrac{1}{3}\iint_Sx\mathrm dy\wedge\mathrm dz+y\mathrm dz\wedge\mathrm dx+z\mathrm dx\wedge\mathrm dy=\frac{1}{3}{\int\kern{-8pt}\int \kern{-23mu} \bigcirc}_S(x\cos\alpha+y\cos\beta+z\cos\gamma)\mathrm dS \]
第四章总复习题 10
设 \(u\) 为有界开集 \(\Omega \subset \mathbb{R}^{3}\) 上的调和函数即 \(\displaystyle \left(\Delta u \equiv \frac{\partial^{2} u}{\partial x^{2}}+\frac{\partial^{2} u}{\partial y^{2}}+\frac{\partial^{2} u}{\partial z^{2}}=0\right)\), 证明:
\((1)\) \(\displaystyle u\left(\boldsymbol{r}_{0}\right)=\frac{1}{4 \pi} \iint_{\partial \Omega}\left(u \frac{\cos \langle\boldsymbol{r}, \boldsymbol{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \boldsymbol{n}}\right) \mathrm{d} S\), 其中 \(\boldsymbol{r}_{0}\) 为 \(\Omega\) 内任意一点, \(\boldsymbol{r}\) 为 \(\boldsymbol{r}_{0}\) 到 \(\partial \Omega\) 上点的向量,
\(r=\|\boldsymbol{r}\|, \boldsymbol{n}\) 为 \(\Omega\) 的单位外法向;
\((2)\) \(\displaystyle \iint_{\partial \Omega}\left(\frac{1}{r} \frac{\partial u}{\partial \boldsymbol{n}}-u \frac{\partial \frac{1}{r}}{\partial \boldsymbol{n}}\right) \mathrm{d} S=\iint_{\partial \Omega_{0}}\left(\frac{1}{r} \frac{\partial u}{\partial \boldsymbol{n}}-u \frac{\partial \frac{1}{r}}{\partial \boldsymbol{n}}\right) \mathrm{d} S\),
其中 \(\boldsymbol{M}\left(x_{0}, y_{0}, z_{0}\right) \in \Omega, \Omega_{0}=B(\boldsymbol{M}, \rho) \subset \Omega, \boldsymbol{r}=\left(x-x_{0}\right) \boldsymbol{i}+\left(y-y_{0}\right) \boldsymbol{j}+(z-\) \(\left.z_{0}\right) \boldsymbol{k}, r=\|\boldsymbol{r}\|\).
证明:\((1)\) 由于 \(\dfrac{\partial }{\partial x}(\dfrac{1}{r})=-\dfrac{x-x_0}{r^3}\),\(\dfrac{\partial^2}{\partial x^2}(\dfrac{1}{r})=\dfrac{3(x-x_0)^2-r^2}{r^5}\) 可以得到
\(\Delta(\dfrac{1}{r})=\dfrac{3(x-x_0)^2+3(y-y_0)^2+3(z-z_0)^2-3r^2}{r^5}=0\) 取 \(\Omega\) 内部的一个以 \(\boldsymbol r_0\) 为球心,\(\epsilon>0\) 为半径
的小球,球面记为\(S_{\epsilon}\). \(S_{\epsilon}\)和\(\Omega\)围成的区域记为\(D\). 将表达式化为梯度形式(其中 \(\nabla(\dfrac{1}{r})=-\dfrac{\boldsymbol{r}}{r^3}\)) \[ u \frac{\cos \langle\mathbf{r}, \mathbf{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \mathbf{n}}=u \frac{\mathbf{r}}{r^{3}} \cdot \mathbf{n}+\frac{1}{r} \nabla u \cdot \mathbf{n}=\left(\frac{1}{r} \nabla u-u \nabla\left(\frac{1}{r}\right)\right) \cdot \mathbf{n} \] 从而利用 \(\mbox{Guass}\) 公式对不包含该瑕点 \(\boldsymbol{r}_0\) 的区域 \(D\) 有(从而证明了 \((2)\) 问 ) \[ \begin{gathered} \iint_{\partial \Omega}\left(u \frac{\cos \langle\mathbf{r}, \mathbf{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \mathbf{n}}\right) \mathrm{d} S-\iint_{S_{\epsilon}}\left(u \frac{\cos \langle\mathbf{r}, \mathbf{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \mathbf{n}}\right) \mathrm{d} S \\=\iiint_{D}\left(\frac{1}{r} \Delta u-u \Delta\left(\frac{1}{r}\right)\right) \mathrm{d} x \mathrm{~d} y \mathrm{~d} z=0 \end{gathered} \] 从而转换为求 \(S_{\epsilon}\) 上的积分问题 \[ \begin{gathered} \iint_{\partial \Omega}\left(u \frac{\cos \langle\mathbf{r}, \mathbf{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \mathbf{n}}\right) \mathrm{d} S =\iint_{S_{\epsilon}}\left(u \frac{\cos \langle\mathbf{r}, \mathbf{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \mathbf{n}}\right) \mathrm{d} S =\iint_{S_{\epsilon}} \frac{u}{\epsilon^{2}} \mathrm{~d} S+\frac{1}{\epsilon} \iint_{S_{\epsilon}} \frac{\partial u}{\partial \mathbf{n}} \mathrm{d} S \\ =4 \pi u\left(r_{1}\right)+\frac{1}{\epsilon} \iint_{S_{\epsilon}} \nabla u \cdot \mathbf{n d} S =4 \pi u\left(r_{1}\right)+\frac{1}{\epsilon} \iiint_{B_{\epsilon}} \Delta u \mathrm{~d} x \mathrm{~d} y \mathrm{~d} z =4 \pi u\left(r_{1}\right) \end{gathered} \] \((2)\) 由上问知,\(\displaystyle \iint_{\partial \Omega}\left(u \frac{\cos \langle\mathbf{r}, \mathbf{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \mathbf{n}}\right) \mathrm{d} S-\iint_{S_{\epsilon}}\left(u \frac{\cos \langle\mathbf{r}, \mathbf{n}\rangle}{r^{2}}+\frac{1}{r} \frac{\partial u}{\partial \mathbf{n}}\right) \mathrm{d} S =0\) 从而 \[ \displaystyle \iint_{\partial \Omega}\left(\frac{1}{r} \frac{\partial u}{\partial \boldsymbol{n}}-u \frac{\partial \frac{1}{r}}{\partial \boldsymbol{n}}\right) \mathrm{d} S=\iint_{\partial \Omega_{0}}\left(\frac{1}{r} \frac{\partial u}{\partial \boldsymbol{n}}-u \frac{\partial \frac{1}{r}}{\partial \boldsymbol{n}}\right) \mathrm{d} S \] # \(\large\textcolor{blue}{微积分第九次作业\ \small(W\times f)}\ \ \ \ \ \ _\textcolor{blue}{2022.5.6}\)
习题4.4.2
计算下列第二类曲线积分.
\((2)\) \(\displaystyle \oint_{L^{+}} \dfrac{(x+y) \mathrm{d} x+(y-x) \mathrm{d} y}{x^{2}+y^{2}}\), 其中 \(L^{+}\)是 \(x^{2}+y^{2}=a^{2}\), 逆时针为正向;
\((3)\) \(\displaystyle \oint_{L^{+}} \dfrac{\mathrm{d} x+\mathrm{d} y}{|x|+|y|}\), 其中 \(L^{+}\)是以 \((1,0),(0,1),(-1,0),(0,-1)\) 为顶点的正方形,逆时针为正向;
\((4)\) \(\displaystyle \int_{L^{+}}\left(y^{2}-z^{2}\right) \mathrm{d} x+\left(z^{2}-x^{2}\right) \mathrm{d} y+\left(x^{2}-y^{2}\right) \mathrm{d} z\), 其中 \(L^{+}\) 是球面 \(x^{2}+y^{2}+\) \(z^{2}=1\) 在第一象限的部分与三个坐标平面的交线,其正向为从点 \((1,0,0)\) 出发, 经过点 \((0,1,0)\), 到点 \((0,0,1)\), 再回到 \((1,0,0)\);
解:\((2)\) 令 \(x=a\cos \theta,y=a\sin \theta\) 代入有 \[ \displaystyle \oint_{L^{+}} \dfrac{(x+y) \mathrm{d} x+(y-x) \mathrm{d} y}{x^{2}+y^{2}}=\oint_{L^{+}}((\cos \theta+\sin \theta )(-\sin \theta)+(\sin \theta-\cos \theta)(\cos \theta))d\theta=-2\pi \] \((3)\) 对 \(x+y=\pm 1\) 第一象限和第三象限的所有点均满足 \(dx+dy=0\),而对第二象限和第四象限
满足通过原点对称的两个点 \((x,y)\longmapsto(-x,-y)\) \((\dfrac{dx+dy}{|x|+|y|})_{L_{2}}=-(\dfrac{dx+dy}{|x|+|y|})_{L_{4}}\) 从而原式 \(=0\)
\((4)\) 第一条线 \(\displaystyle \int_{L^{+}}y^2dx-x^2dy=\displaystyle \int_{L^{+}}(-\sin ^3\theta-\cos ^3\theta )d\theta=-\dfrac{4}{3}\) 后两条线相同,原式 \(=-4\)
习题4.4.3
计算 \(\displaystyle \int_{L^{+}} \boldsymbol{F} \cdot \mathrm{d} \boldsymbol{r}\).
\((4)\) \(\boldsymbol{F}\) 为由原点处质量为 \(m\) 的质点所形成的空间引力场 (对单位质量质点的引力 \(), L\) 是由点 \((1,1,1)\) 到点
\(\left(x_{0}, y_{0}, z_{0}\right)\) 的直线段. \(\left(x_{0}, y_{0}, z_{0}\right.\) 不同时为零. \()\)
解:由题意计算\(W=\displaystyle \int_{L_{1}(\boldsymbol{A})}^{(\boldsymbol{B})} \frac{Gm x \mathrm{~d} x}{\left(x^{2}+y^{2}+z^{2}\right)^{\frac{3}{2}}}+\frac{Gm y \mathrm{~d} y}{\left(x^{2}+y^{2}+z^{2}\right)^{\frac{3}{2}}}+\frac{Gm z \mathrm{~d} z}{\left(x^{2}+y^{2}+z^{2}\right)^{\frac{3}{2}}}\)
取直线上 \(t\) 比例的点 \(\boldsymbol{r}=(1-t)(1,1,1)+t(x_0,y_0,z_0)\) 解得 \(W=Gm(\dfrac{1}{\sqrt{3}}-\dfrac{1}{\sqrt{x_0^2+y_0^2+z_0^2}})\)
习题4.4.4
今有一平面力场 \(\boldsymbol{F}\), 大小等于点 \((x, y)\) 到坐标原点的距离, 方向指向坐标原点.
\((2)\) 计算质点 \(P\) 沿上述椭圆逆时针绕一圈时,力 \(\boldsymbol{F}\) 所做的功.
解:该力为 \(\boldsymbol F=(x,y)\) 有势能函数 \(\boldsymbol U=\dfrac{x^2+y^2}{2}\) 从而绕一圈做功 \(W=0\)
习题4.6.1
利用 \(\mbox{Green}\) 公式计算下列曲线积分.
\((4)\) \(\displaystyle \oint_{L^{+}} \mathrm{e}^{x}[(1-\cos y) \mathrm{d} x-(y-\sin y) \mathrm{d} y]\), 其中 \(L\) 是区域 \(\{(x, y) \mid 0 \leqslant x \leqslant\) \(\pi, 0 \leqslant y \leqslant \sin x\}\) 的正向边界.
解:由格林公式 \(\displaystyle \oint_{L^{+}} \mathrm{e}^{x}[(1-\cos y) \mathrm{d} x-(y-\sin y) \mathrm{d} y]=\iint -e^x(y-\sin y)-e^x\sin ydxdy\)
\(\displaystyle =-\iint e^xydxdy=-\int_0^{\pi}e^{x}dx\int_0^{\sin x}ydy=-\dfrac{1}{2}\int_0^{\pi}\sin ^2xe^xdx=-\dfrac{1}{4}\int_0^{\pi}(1-\cos 2x)e^xdx\)
而 \(\displaystyle \int\cos 2xe^xdx=\dfrac{1}{2}\int e^xd(\sin 2x)=\dfrac{1}{2}e^x\sin 2x-\dfrac{1}{2}\int e^x \sin 2xdx=\dfrac{1}{2}e^x\sin 2x+\dfrac{1}{4}\int e^x d(\cos 2x)\)
\(=\displaystyle \dfrac{1}{2}e^x\sin 2x+\dfrac{1}{4}e^x\cos 2x-\dfrac{1}{4}\int e^x\cos 2xdx\Longrightarrow \int e^x\cos 2xdx=\dfrac{2\sin 2x+\cos 2x}{5}e^x+C\)
从而原式 \(=(-\dfrac{1}{4}e^{x}+\dfrac{2\sin 2x+\cos 2x}{20}e^x)\big |_{0}^{\pi}=-\dfrac{1}{4}(e^{\pi}-1)+\dfrac{1}{20}(e^{\pi}-1)=-\dfrac{e^{\pi}-1}{5}\)
答案有误,\(-\dfrac{e^{\pi}-1}{5}\) 与 \(\mbox{Mathematica}\) 软件计算结果相同(计算耗时 \(47\mbox{s}\) )
习题4.6.4
计算下列区域的面积.
\((1)\) 星形线 \(\left\{\begin{array}{l}x=a \cos ^{3} t, \\ y=a \sin ^{3} t,\end{array}\ \ 0 \leqslant t \leqslant 2 \pi\right.\) 所围区域 \((a>0)\);
解:由 \(\mbox{Green}\) 公式计算将二重积分化成曲线积分有 \[ \begin{gathered} S=\displaystyle \iint_{D} dxdy=\dfrac{1}{2}\int_{\partial D}xdy-ydx=\dfrac{a^2}{2}\int_0^{2\pi}\cos ^3t\cdot 3\sin ^2t\cos tdt+\sin ^3t\cdot 3\cos ^2t\sin tdt\\ =\dfrac{3a^2}{2}\int_0^{2\pi}\sin ^2t\cos ^2 tdt=\dfrac{3a^2}{8}\int_0^{2\pi}\dfrac{1-\cos 4t}{2}dt=\dfrac{3}{8}a^2\cdot \pi=\dfrac{3\pi}{8}a^2 \end{gathered} \]
习题4.6.7
设 \(D\) 是平面区域, \(\partial D\) 为逐段光滑曲线, \(f \in C^{2}(D)\), 证明: \(\displaystyle \oint_{\partial D} \frac{\partial f}{\partial \boldsymbol{n}} \mathrm{d} l=\) \(\displaystyle \iint_{D}\left(\frac{\partial^{2} f}{\partial x^{2}}+\frac{\partial^{2} f}{\partial y^{2}}\right) \mathrm{d} x \mathrm{~d} y\)
证明:由沿梯度方向的导数 \(\dfrac{\partial f}{\partial\boldsymbol{n}}=\dfrac{\partial f}{\partial x}\hat{n}_{x}+\dfrac{\partial f}{\partial y}\hat{n}_{y}=\dfrac{\partial f}{\partial x}(\dfrac{dy}{dl})+\dfrac{\partial f}{\partial y}(-\dfrac{dx}{dl})\)
其中 \(\vec{n}\) 方向与 \((dx,dy)\) 方向垂直且满足为单位向量,从而为 \(\vec{n}=(\dfrac{dy}{dl},-\dfrac{dx}{dl})\) 再代入格林公式即得证 \[ \oint_{\partial D} \frac{\partial f}{\partial \boldsymbol{n}} \mathrm{d} l=\oint_{\partial D}(\dfrac{\partial f}{\partial x}(\dfrac{dy}{dl})+\dfrac{\partial f}{\partial y}(-\dfrac{dx}{dl}))dl=\oint_{\partial D}(\dfrac{\partial f}{\partial x}dy-\dfrac{\partial f}{\partial y}dx)=\displaystyle \iint_{D}\left(\frac{\partial^{2} f}{\partial x^{2}}+\frac{\partial^{2} f}{\partial y^{2}}\right) \mathrm{d} x \mathrm{~d} y \]
习题4.6.8
设 \(D\) 是平面区域, \(\partial D\) 为逐段光滑曲线, \(n\) 为 \(D\) 的单位外法向, \(u, v \in\) \(C^{2}(D)\), 证明:
\((1)\) \(\displaystyle \oint_{\partial D} \frac{\partial u}{\partial \boldsymbol{n}} \mathrm{d} l=\iint_{D} \Delta u \mathrm{~d} x \mathrm{~d} y\);
\((2)\) \(\displaystyle \oint_{\partial D} v \frac{\partial u}{\partial \boldsymbol{n}} \mathrm{d} l=\iint_{D} v \Delta u \mathrm{~d} x \mathrm{~d} y-\iint_{D} \nabla u \cdot \nabla v \mathrm{~d} x \mathrm{~d} y\);
\((3)\) \(\displaystyle \oint_{\partial D}\left|\begin{array}{cc}\dfrac{\partial u}{\partial \boldsymbol{n}} & \dfrac{\partial v}{\partial \boldsymbol{n}} \\ u & v\end{array}\right| \mathrm{d} l=\iint_{D}\left|\begin{array}{cc}\Delta u & \Delta v \\ u & v\end{array}\right| \mathrm{d} x \mathrm{~d} y\).
其中 \(\Delta=\dfrac{\partial^{2}}{\partial x^{2}}+\dfrac{\partial^{2}}{\partial y^{2}}, \nabla=\dfrac{\partial}{\partial x} i+\dfrac{\partial}{\partial y} j\).
证明:\((1)\) 由 \(4.6.7\) 可知,对 \(u\in C^2(D)\),\(\displaystyle \oint_{\partial D} \frac{\partial u}{\partial \boldsymbol{n}} \mathrm{d} l=\displaystyle \iint_{D}\left(\frac{\partial^{2} u}{\partial x^{2}}+\frac{\partial^{2} u}{\partial y^{2}}\right) \mathrm{d} x \mathrm{~d}y=\iint_{D} \Delta u \mathrm{~d} x \mathrm{~d} y\)
\((2)\) 代入 \(\dfrac{\partial u}{\partial \boldsymbol{n}}dl=\dfrac{\partial u}{\partial x}dy-\dfrac{\partial u}{\partial y}dx\) 有 \(\displaystyle \oint_{\partial D} v \frac{\partial u}{\partial \boldsymbol{n}} \mathrm{d} l=\oint_{\partial D} v (\dfrac{\partial u}{\partial x}dy-\dfrac{\partial u}{\partial y}dx)\) 由格林公式 \[ \begin{gathered} \oint_{\partial D} v (\dfrac{\partial u}{\partial x}dy-\dfrac{\partial u}{\partial y}dx)=\iint_{D}(\dfrac{\partial v}{\partial x}\dfrac{\partial u}{\partial x}+v\dfrac{\partial ^2u}{\partial x^2})-(-\dfrac{\partial v}{\partial y}\dfrac{\partial u}{\partial y}-v\dfrac{\partial ^2u}{\partial y^2})dxdy\\ =\iint_{D} (v(\dfrac{\partial ^2u}{\partial x^2}+\dfrac{\partial ^2u}{\partial y^2})+(\dfrac{\partial v}{\partial x}\dfrac{\partial u}{\partial x}+\dfrac{\partial v}{\partial y}\dfrac{\partial u}{\partial y}))dxdy=\iint_{D} v \Delta u \mathrm{~d} x \mathrm{~d} y+\iint_{D} \nabla u \cdot \nabla v \mathrm{~d} x \mathrm{~d} y \end{gathered} \] \((3)\) 由行列式公式 \(\displaystyle \oint_{\partial D}\left|\begin{array}{cc}\dfrac{\partial u}{\partial \boldsymbol{n}} & \dfrac{\partial v}{\partial \boldsymbol{n}} \\ u & v\end{array}\right| \mathrm{d} l=\displaystyle \oint_{\partial D}(v\dfrac{\partial u}{\partial \boldsymbol{n}}-u\dfrac{\partial v}{\partial \boldsymbol{n}}) \mathrm{d} l\) 代入 \(\small (2)\) 问结论 \[ \displaystyle \oint_{\partial D}(v\dfrac{\partial u}{\partial \boldsymbol{n}}-u\dfrac{\partial v}{\partial \boldsymbol{n}}) \mathrm{d} l=\iint_{D}((v\Delta u-\nabla u\cdot \nabla v)-(u\Delta v-\nabla u\cdot \nabla v))dxdy=\iint_{D}\left|\begin{array}{cc}\Delta u & \Delta v \\ u & v\end{array}\right| \mathrm{d} x \mathrm{~d} y \]
习题4.6.10
求解下列常微分方程.
\((1)\) \(\displaystyle \left(x^{2}-y\right) \mathrm{d} x-\left(x+\sin ^{2} y\right) \mathrm{d} y=0\);
\((2)\) \(\displaystyle \mathrm{e}^{y} \mathrm{~d} x+\left(x \mathrm{e}^{y}-2 y\right) \mathrm{d} y=0\);
\((3)\) \(\displaystyle \frac{x \mathrm{~d} x+y \mathrm{~d} y}{\sqrt{x^{2}+y^{2}}}=\frac{y \mathrm{~d} x-x \mathrm{~d} y}{x^{2}}\);
\((4)\) \(\displaystyle \left(\cos x+\frac{1}{y}\right) \mathrm{d} x+\left(\frac{1}{y}-\frac{x}{y^{2}}\right) \mathrm{d} y=0\).
解:\((1)\) 由 \(\dfrac{\partial (x^2-y)}{\partial y}=\dfrac{\partial (-x-\sin ^2y)}{\partial x}=-1\) 为全微分方程,并注意到 \[ d(\dfrac{1}{3}x^3-xy-(\dfrac{1}{2}y-\dfrac{1}{4}\sin 2y))=\left(x^{2}-y\right) \mathrm{d} x-\left(x+\sin ^{2} y\right) \mathrm{d} y=0 \] 则原方程解为 \(\dfrac{1}{3}x^3-xy-(\dfrac{1}{2}y-\dfrac{1}{4}\sin 2y)+C=0,C\in R\)
\((2)\) 由 \(\dfrac{\partial (e^y)}{\partial y}=\dfrac{\partial (xe^y-2y)}{\partial x}=e^y\) 为全微分方程,并注意到 \[ d(xe^y-y^2)=\mathrm{e}^{y} \mathrm{~d} x+\left(x \mathrm{e}^{y}-2 y\right) \mathrm{d} y=0 \] 则原方程解为 \(xe^y-y^2+C=0,C\in R\)
\((3)\) 由 \(\dfrac{\partial (\dfrac{x}{\sqrt{x^2+y^2}}-\dfrac{y}{x^2})}{\partial y}=\dfrac{\partial (\dfrac{y}{\sqrt{x^2+y^2}}+\dfrac{1}{x})}{\partial x}=-\dfrac{xy}{(x^2+y^2)^{\frac{3}{2}}}-\dfrac{1}{x^2}\) 为全微分方程,并注意到 \[ d(\sqrt{x^2+y^2}-\dfrac{y}{x})=\frac{x \mathrm{~d} x+y \mathrm{~d} y}{\sqrt{x^{2}+y^{2}}}-\frac{y \mathrm{~d} x-x \mathrm{~d} y}{x^{2}}=0 \] 则原方程解为 \(\sqrt{x^2+y^2}-\dfrac{y}{x}+C=0,C\in R\)
\((4)\) 由 \(\dfrac{\partial (\cos x+\dfrac{1}{y})}{\partial y}=\dfrac{\partial (\dfrac{1}{y}-\dfrac{x}{y^2})}{\partial x}=-\dfrac{1}{y^2}\) 为全微分方程,并注意到 \[ d(\sin x+\dfrac{x}{y}+\ln y)=\left(\cos x+\frac{1}{y}\right) \mathrm{d} x+\left(\frac{1}{y}-\frac{x}{y^{2}}\right) \mathrm{d} y=0 \] 则原方程解为 \(\sin x+\dfrac{x}{y}+\ln y+C=0,C\in R\)
习题4.6.11
解下列方程.
\((1)\) \((y \cos x-x \sin x) \mathrm{d} x+(y \sin x+x \cos x) \mathrm{d} y=0\);
\((2)\) \((x+y) \mathrm{d} x+(y-x) \mathrm{d} y=0\);
\((3)\) \(\left(3 x^{3}+y\right) \mathrm{d} x+\left(2 x^{2} y-x\right) \mathrm{d} y=0\);
\((4)\) \((x+y)(\mathrm{d} x-\mathrm{d} y)=\mathrm{d} x+\mathrm{d} y\);
\((5)\) \(\left(x^{2}-\sin ^{2} y\right) \mathrm{d} x+x \sin 2 y \mathrm{~d} y=0\).
解:\((1)\) 记 \(M=y \cos x-x \sin x, N=y \sin x+x \cos x\), 有 \[ \frac{\partial M}{\partial y}=\cos x, \frac{\partial N}{\partial x}=y \cos x+\cos x-x \sin x, \] 则方程不是恰当的. 但观察到 \(\dfrac{\dfrac{\partial M}{\partial y}-\dfrac{\partial N}{\partial x}}{-M}=1\), 对应原方程有积分因子 \(\mu=\mathrm{e}^{\int1 d y}=\mathrm{e}^{y}\). 两边乘以 \(\mu=e^{y}\) \[ \begin{gathered} \left(y \mathrm{e}^{y} \cos x \mathrm{~d} x+y \mathrm{e}^{y} \sin x \mathrm{~d} y\right)+\left(\mathrm{e}^{y} x \cos x \mathrm{~d} y-\mathrm{e}^{y} x \sin x \mathrm{~d} x\right)=0\\\mathrm{d}\left[(y-1) \mathrm{e}^{y} \sin x+\mathrm{e}^{y} x \cos x\right]=0 . \end{gathered} \] 则原方程的通解为 \(\mathrm{e}^{y}(y-1) \sin x+\mathrm{e}^{y} x \cos x=C, C\in R\)
\((2)\) 两边除以 \(x^2+y^2\) 有 \(\dfrac{x+y}{x^2+y^2}dx+\dfrac{y-x}{x^2+y^2}dy=0\) 且 \(\dfrac{\partial (\dfrac{x+y}{x^2+y^2})}{\partial y}=\dfrac{\partial (\dfrac{y-x}{x^2+y^2})}{\partial x}=\dfrac{x^2-2xy-y^2}{(x^2+y^2)^2}\)
从而变换后的方程为全微分方程,并注意到 \[ d(\dfrac{1}{2}\ln(x^2+y^2)-\arctan (\dfrac{y}{x}))=\dfrac{x+y}{x^2+y^2}dx+\dfrac{y-x}{x^2+y^2}dy=0 \] 则原方程的解为 \(\ln(\sqrt{x^2+y^2})-\arctan (\dfrac{y}{x})=C,C\in R\)
\((3)\) 两边同时除以 \(x^2\)(消去 \(x^2y\) 中含 \(x\) 项 ),得到 \((3+\dfrac{y}{x^2})dx+(2y-\dfrac{1}{x})dy=0\) 且有
\(\dfrac{\partial (3+\dfrac{y}{x^2})}{\partial y}=\dfrac{\partial (2y-\dfrac{1}{x})}{\partial x}=\dfrac{1}{x^2}\) 为全微分方程,注意到 \[ d(3x-\dfrac{y}{x}+y^2)=0 \] 则原方程的解为 \(3x-\dfrac{y}{x}+y^2=C,C\in R\)
\((4)\) 两边除以 \(x+y\) 有 \(dx-dy=\dfrac{dx+dy}{x+y}\) 后者有 \(\dfrac{\partial (\dfrac{1}{x+y})}{\partial y}=\dfrac{\partial (\dfrac{1}{x+y})}{\partial x}=-\dfrac{1}{(x+y)^2}\) 为全微分
\(d(x-y)=\dfrac{dx+dy}{x+y}=d(\ln(|x+y|))\),则原方程的解为 \(x-y=\ln|x+y|+C,C\in R\)
\((5)\) 记 \(M=x^2-\sin ^2y, N=x\sin 2y\), 有 \[ \frac{\partial M}{\partial y}=-\sin 2y, \frac{\partial N}{\partial x}=\sin 2y \] 则方程不是恰当的. 观察到 \(\dfrac{\dfrac{\partial M}{\partial y}-\dfrac{\partial N}{\partial x}}{-N}=-\dfrac{2}{x}=f(x)\), 对应原方程有积分因子 \(\mu=\mathrm{e}^{-\int\frac{2}{x} dx}=\dfrac{1}{x^2}\) \[ \begin{gathered} (1-\dfrac{\sin ^2y}{x^2})dx+\dfrac{\sin 2y}{x}dy=0\Longrightarrow d(x+\dfrac{\sin ^2y}{x})=0 \end{gathered} \] 则原方程的通解为 \(x+\dfrac{\sin ^2y}{x}=C, C\in R\)
第四章总复习题 4
设 \(L \subset \mathbb{R}^{2}\) 为光滑闭曲线, 逆时针为正向, \(\boldsymbol{n}\) 为 \(L\) 的外法线单位向量, \(\boldsymbol{a}\) 为一固定向量, 求证: \[ \oint_{L} \cos \langle\boldsymbol{n}, \boldsymbol{a}\rangle \mathrm{d} l=0 \]
证明:由 \(\boldsymbol{n}=\dfrac{1}{dl}(dy,-dx)\) 则代入有 \(\displaystyle \oint_{L} \cos \langle\boldsymbol{n}, \boldsymbol{a}\rangle \mathrm{d} l=\oint_{L} a_xdy-a_ydx\) 由格林公式 \[ \oint_{L} a_xdy-a_ydx=\iint_{D}(0-0)dxdy=0 \]
第四章总复习题 6
设 \(L\) 为一单连通区域 \(D \subset \mathbb{R}^{2}\) 的边界, \(P_{0}\) 为一定点, \(P \in L, \boldsymbol v=\overrightarrow{P_{0} P}, v=\) \(\|\boldsymbol v\|, \boldsymbol{n}\) 为 \(L\) 的外法线单位向量, 证明 \[ \oint_{L} \frac{\cos \langle\boldsymbol{n}, v\rangle}{v} \mathrm{~d} l= \begin{cases}2\pi, & P_{0} \in D, \\ 0, & P_{0} \notin D .\end{cases} \]
证明:代入 \(\boldsymbol v =\boldsymbol p-\boldsymbol p_0=(x,y)-\boldsymbol p_0,\boldsymbol{n}=\dfrac{1}{dl}(dy,-dx)\) 令 \(\boldsymbol p_0=(x_0,y_0)\)有 \[ \displaystyle \oint_{L} \frac{\cos \langle\boldsymbol{n}, v\rangle}{v} \mathrm{~d} l=\oint_{L} \dfrac{(x-x_0)dy-(y-y_0)dx}{(\sqrt{(x-x_0)^2+(y-y_0)^2})^2}=\oint_{L} \dfrac{(x-x_0)dy-(y-y_0)dx}{(x-x_0)^2+(y-y_0)^2} \] 由格林公式化为二重积分 \[ \begin{gathered} \oint_{L} \dfrac{(x-x_0)dy-(y-y_0)dx}{(x-x_0)^2+(y-y_0)^2}=\iint_{D}(\dfrac{(x-x_0)^2+(y-y_0)^2-(x-x_0)\cdot 2(x-x_0)}{((x-x_0)^2+(y-y_0)^2)^2}\\ -(-\dfrac{(x-x_0)^2+(y-y_0)^2-(y-y_0)\cdot 2(y-y_0)}{((x-x_0)^2+(y-y_0)^2)^2}))dxdy=\iint_{D}\dfrac{0\cdot dxdy}{(x-x_0)^2+(y-y_0)^2} \end{gathered} \] 当 \((x,y)\) 保持不等于 \((x_0,y_0)\) 时,上述二重积分恒为 \(0\),只需满足 \(P_0\) 不在二重积分区域内,此时 \(P_0\not\in D\)
而当 \(P_0\not\in D\) 时,令 \(x-x_0=r\cos \theta,y-y_0=r\sin \theta\),二重积分有瑕点,对线积分计算 \[ \oint_{L} \dfrac{(x-x_0)dy-(y-y_0)dx}{(x-x_0)^2+(y-y_0)^2}=\oint_{L} \dfrac{r\cos \theta d(r\sin \theta)-r\sin \theta d(r\cos \theta)}{r^2}=\oint_{L}\dfrac{r^2}{r^2}d\theta \] 由于瑕点 \(r=0\) 为零测度集,从而 \(\displaystyle \int_{0}^{2\pi}d\theta=2\pi\),综上有 \(\displaystyle \oint_{L} \frac{\cos \langle\boldsymbol{n}, v\rangle}{v} \mathrm{~d} l= \begin{cases}2\pi, & P_{0} \in D, \\ 0, & P_{0} \notin D .\end{cases}\)
第四章总复习题 9
设 \(u\) 为开集 \(D \subset R^{2}\) 上的调和函数. 即 \(\left(\Delta u \equiv \dfrac{\partial^{2} u}{\partial x^{2}}+\dfrac{\partial^{2} u}{\partial y^{2}}=0\right)\), 证明:
\((1)\) \(\displaystyle u\left(x_{0}, y_{0}\right)=\frac{1}{2 \pi} \oint_{\partial D}\left(u \frac{\partial \ln r}{\partial \boldsymbol{n}}-\ln r \frac{\partial u}{\partial \boldsymbol{n}}\right) \mathrm{d} l\), 其中 \(\left(x_{0}, y_{0}\right) \in D, \boldsymbol{r}\) 为 \(\left(x_{0}, y_{0}\right)\) 到 \(\partial D\) 上点的向量,
\(r=\|\boldsymbol{r}\|, \boldsymbol{n}\) 为 \(D\) 的单位外法向量;
\((2)\) \(\displaystyle u\left(x_{0}, y_{0}\right)=\frac{1}{2 \pi R} \oint_{L} u(x, y) \mathrm{d} l\), 其中 \(L\) 是以 \(\left(x_{0}, y_{0}\right)\) 为中心, \(R\) 为半径位于 \(D\) 中的任意一个圆周.
证明:\((1)\) 当 \((x, y) \neq\left(x_{0}, y_{0}\right)\) 时, \(v=\ln r=\ln \sqrt{\left(x-x_{0}\right)^{2}+\left(y-y_{0}\right)^{2}}\) 为调和函数,
(可以计算拉普拉斯 \(\Delta v=\dfrac{\partial^{2} v}{\partial x^{2}}+\dfrac{\partial^{2} v}{\partial y^{2}}=\dfrac{1}{r^{4}}\left[\left(y-y_{0}\right)^{2}-\left(x-x_{0}\right)^{2}\right]+\dfrac{1}{r^{4}}\left[\left(x-x_{0}\right)^{2}-\right.\) \(\left.\left(y-y_{0}\right)^{2}\right]=0\) )
因此, 以 \(\left(x_{0}, y_{0}\right)\) 为中心挖去半径为 \(\varepsilon>0\) (充分小)的圆形区域 \(S_{r}\), 则在剩下的区域(记作 \(D_{1}\) ), 对 \(u, v\) 可应用
\(\small 4.6.8(3)\) 中结论:\(\displaystyle \oint_{\partial D}\left|\begin{array}{cc}\dfrac{\partial u}{\partial \boldsymbol{n}} & \dfrac{\partial v}{\partial \boldsymbol{n}} \\ u & v\end{array}\right| \mathrm{d} l=\iint_{D}\left|\begin{array}{cc}\Delta u & \Delta v \\ u & v\end{array}\right| \mathrm{d} x \mathrm{~d} y\) \[ \oint_{L \uparrow+c_{r}\downarrow } \left(u \frac{\partial \ln r}{\partial n}-\ln r \frac{\partial u}{\partial n}\right) \mathrm{d} l=\iint_{D}\left|\begin{array}{cc} \Delta u & \Delta \ln r \\ u & \ln r \end{array}\right| \mathrm{d} x \mathrm{~d} y=\iint_{D}\left|\begin{array}{cc} 0 & 0 \\ u & \ln r \end{array}\right| \mathrm{d} x \mathrm{~d} y=0 \] 从而拆开左边环路积分并移项有 \[ \begin{gathered} \oint_{L\uparrow }\left(u \frac{\partial \ln r}{\partial n}-\ln r \frac{\partial u}{\partial n}\right) \mathrm{d} l=\oint_{c_{g}\uparrow } \left(u \frac{\partial \ln r}{\partial n}-\ln r \frac{\partial u}{\partial n}\right) \mathrm{d} l=\oint_{c_{g}\uparrow } u \frac{\partial \ln r}{\partial n}- \oint_{c_{g}\uparrow } \ln r \frac{\partial u}{\partial n} \mathrm{~d} s \stackrel{\triangle}{=}I_{1}+I_{2} \end{gathered} \] 利用 \(\mbox{Green}\) 公式计算 \(I_2\) \[ \begin{aligned} I_{2} =\oint_{c_{g}\uparrow } \ln r \frac{\partial u}{\partial n} \mathrm{~d} l =\ln \varepsilon \oint_{c_{g}\uparrow }\left[\frac{\partial u}{\partial x} \cos (n, x)+\frac{\partial u}{\partial y} \cos (n, y)\right] \mathrm{d} l =\ln \varepsilon \iint_{S_{\epsilon}} \Delta u \mathrm{~d} x \mathrm{~d} y=0 . \end{aligned} \] 注意在 \(C_{g}\) 上 \[ \left.\frac{\partial \ln r}{\partial n}\right|_{r=e}=\left.\frac{\partial \ln r}{\partial r}\right|_{r=e}=\left.\frac{1}{r}\right|_{r=e}=\frac{1}{\varepsilon} \] 因此 \[ I_{1}=\oint_{C_{\varepsilon}} u \frac{\partial \ln r}{\partial n} \mathrm{~d} s=\frac{1}{\varepsilon} \oint_{c_{\varepsilon}} u \mathrm{~d} s \] 将 \(I_2,I_1\) 结果代入并利用中值定理得到 \[ \begin{gathered} \frac{1}{2 \pi} \oint_{L}\left(u \frac{\partial \ln r}{\partial n}+\ln r \frac{\partial u}{\partial n}\right) \mathrm{d} s=\frac{1}{2 \pi \varepsilon} \oint_{C_{\varepsilon}} u \mathrm{~d} s =\frac{1}{2 \pi \varepsilon} u(Q) \oint_{c_{\varepsilon}} \mathrm{d} s \end{gathered} \] 取 \(\epsilon \to 0\) 便得到 \(u(Q)\to u(x_0,y_0)\)
\((2)\) 在 \(\displaystyle u\left(x_{0}, y_{0}\right)=\frac{1}{2 \pi} \oint_{\partial D}\left(u \frac{\partial \ln r}{\partial \boldsymbol{n}}-\ln r \frac{\partial u}{\partial \boldsymbol{n}}\right) \mathrm{d} l\) 中取 \(L=C_{r}\),为半径为 \(R\) 的正圆,有 \[ \oint_{\partial D}\ln r \frac{\partial u}{\partial \boldsymbol{n}}\mathrm{d} l=\ln R\oint_{\partial D}\frac{\partial u}{\partial \boldsymbol{n}}\mathrm{d} l=\ln R \oint_{\partial D}(\frac{\partial u}{\partial x}dy-\frac{\partial u}{\partial y}dx)=0 \] 前者有 \(R=r=\sqrt{x^2+y^2}\) 代入 \(\dfrac{\partial \ln r}{\partial \boldsymbol{n}}=\dfrac{x}{(\sqrt{x^2+y^2})^2}\dfrac{dy}{dl}-\dfrac{y}{(\sqrt{x^2+y^2})^2}\dfrac{dx}{dl}=\dfrac{x}{R^2}\dfrac{dy}{dl}-\dfrac{y}{R^2}\dfrac{dx}{dl}\)
有 \(x(x_0,y_0)=\dfrac{1}{2\pi R}\displaystyle \oint u\dfrac{xdy-ydx}{R}=\dfrac{1}{2\pi R}\displaystyle \oint u dl\)
在物理中,该公式代表在一个无电荷分布静电场 \(\nabla^2 \varphi=0\) 中,任意平面圆的平均电势等于圆心的电势
物理证明:在平面圆心中央放置一个带电量为 \(\mathcal{q_1}\) 的点电荷,其周围半径为 \(R\) 的一固定圆环均匀放置
总电量为 \(\mathcal{q}_2\) 的电荷,两者电荷量 \(\to 0\),对原始电场分布几乎无影响(以任意 $$ 精度保持解析性)
由静电场中的格林互逆定理 \[ \sum\varphi(q_1)q_2=\sum\varphi(q_2)q_1 \] 代入静电能 $(q_1)q_2=(_0+K)q_2,(q_2)q_2=Kq_1+() $
令 \(q_1=q_2\to 0\),便可以得到 \(\varphi_0=\dfrac{1}{2\pi}\displaystyle \int\varphi(\theta)d\theta\)