主要内容

高等代数教学辅导

7.2 不变子空间

建设中!

子节 7.2.1 基础知识回顾

练习 7.2.2 练习

基础题.

1.
\(\phi:\mathbb{F}^2\rightarrow\mathbb{F}^2,\ (a,b)^T\mapsto (b,a)^T\),试求所有非平凡的\(\phi\)-不变子空间。
解答.
\(U\)是非平凡的\(\phi\)-不变子空间,则\(\dim U=1\)。设\((a,b)^T\)\(U\)的基,则\(\phi((a,b)^T)=(b,a)^T\in U\),即存在\(k\in\mathbb{F}\)使得\((b,a)^T=k(a,b)^T\),故\(b=\pm a\)。因此\(U=\langle (1,1)\rangle\)\(\langle (1,-1)\rangle\)
2.
\(V\)是4维线性空间,\(V\)上线性变换\(\phi\)在基\((\xi_1,\xi_2,\xi_3,\xi_4)\)下的矩阵为
\begin{equation*} \begin{pmatrix} 1&0&2&-1\\ 0&1&4&-2\\ 2&-1&0&1\\ 2&-1&-1&2 \end{pmatrix}, \end{equation*}
  1. 证明:\(U=\langle \xi_1+2 \xi_2, \xi_2+\xi_3+2 \xi_4\rangle\)\(\phi\)-不变子空间;
  2. \(\phi|_U\)在基\((\xi_1+2 \xi_2, \xi_2+\xi_3+2 \xi_4)\)下的矩阵。
解答.
  1. 根据已知条件,有
    \begin{equation*} \left\{\begin{array}{l} \phi (\xi_1)=\xi_1+2 \xi_3+2 \xi_4,\\ \phi (\xi_2)=\xi_2- \xi_3- \xi_4,\\ \phi (\xi_3)=2\xi_1+4 \xi_2- \xi_4,\\ \phi (\xi_4)=-\xi_1-2 \xi_2+\xi_3+2 \xi_4,\\ \end{array}\right. \end{equation*}
    \begin{equation*} \phi (\xi_1+2 \xi_2)=\phi (\xi_1)+2 \phi(\xi_2)=\xi_1+2 \xi_2\in U, \end{equation*}
    \begin{equation*} \phi (\xi_2+ \xi_3+2 \xi_4) =\phi (\xi_2)+\phi (\xi_3)+2 \phi(\xi_4) =\xi_2+\xi_3+2 \xi_4\in U, \end{equation*}
    因此\(U\)\(\phi\)-子空间。
  2. 由于
    \begin{equation*} \begin{array}{c} \phi|_U(\xi_1+2 \xi_2)=\phi(\xi_1+2 \xi_2)=\xi_1+2 \xi_2,\\ \phi|_U(\xi_2+ \xi_3+2 \xi_4)=\phi(\xi_2+\xi_3+2 \xi_4)=\xi_2+\xi_3+2 \xi_4, \end{array} \end{equation*}
    \begin{equation*} \phi(\xi_1+2 \xi_2,\xi_2+\xi_3+2 \xi_4)=(\xi_1+2 \xi_2,\xi_2+\xi_3+2 \xi_4)\begin{pmatrix} 1&0\\0&1 \end{pmatrix}, \end{equation*}
    所以\(\phi|_U\)在基\((\xi_1+2 \xi_2, \xi_2+\xi_3+2 \xi_4)\)下的矩阵为\(\begin{pmatrix} 1&0\\0&1 \end{pmatrix}\)

提高题.

3.
\(\phi\)是数域\(\mathbb{F}\)\(n\)维线性空间\(V\)上的线性变换,取定\(\lambda\in\mathbb{F}\),记
\begin{equation*} V_\lambda^{(\phi)}=\{\alpha\in V\ |\ \phi (\alpha)=\lambda \alpha\}, \end{equation*}
证明:\(V_\lambda^{(\phi)}\)\(\phi\)-不变子空间。
解答.
\begin{equation*} \phi(0_V)=0_V=\lambda 0_V, \end{equation*}
所以\(0_V\in V_\lambda\)\(V_\lambda\)\(V\)的非空子集。对任意\(\alpha,\beta\in V_\lambda,a,b\in\mathbb{F}\),有
\begin{equation*} \phi (\alpha)=\lambda \alpha,\phi(\beta)=\lambda \beta, \end{equation*}
\begin{equation*} \phi (a \alpha+b \beta)=a\phi(\alpha)+b\phi(\beta)=a(\lambda \alpha)+b(\lambda \beta)=\lambda(a \alpha+b \beta)\mbox{。} \end{equation*}
因此\(V_\lambda\)\(V\)的子空间。又
\begin{equation*} \phi (\alpha)=\lambda \alpha\in V_\lambda, \end{equation*}
\(V_ \lambda\)\(\phi\)-不变子空间。
4.
\(\phi ,\psi\)\(n\)维线性空间\(V\)上线性变换,
  1. \(\phi\psi=\psi\phi\),证明:\({\rm Ker}\phi\)\({\rm Im}\phi\)都是\(\psi\)-不变子空间;
  2. \(\phi^2=\phi\),证明:\({\rm Ker}\phi\)\({\rm Im}\phi\)都是\(\psi\)-不变子空间的充分必要条件是\(\phi\psi=\psi\phi\)
解答.
  1. 对任意\(\alpha\in{\rm Ker}\phi\), 有\(\phi(\alpha)=0\)。因为\(\phi\psi =\psi\phi\),所以
    \begin{equation*} \phi(\psi(\alpha))=\psi(\phi (\alpha))=\psi (0)=0, \end{equation*}
    \(\psi (\alpha)\in{\rm Ker}\phi\)。因此,\({\rm Ker}\phi\)\(\psi\)-子空间。
    对任意\(\beta\in{\rm Im}\phi\),存在\(\alpha\in V\),使得\(\beta=\phi(\alpha)\)。因为\(\phi\psi=\psi\phi\),所以
    \begin{equation*} \psi(\beta)=\psi\phi(\alpha)=\phi(\psi(\alpha))\in{\rm Im}\phi, \end{equation*}
    \({\rm Im}\phi\)\(\psi\)-子空间。
  2. \((1)\),充分性成立。下证必要性。
    对任意\(\alpha\in V\),有\(\phi^2(\alpha)-\phi(\alpha)=0\),即\(\phi(\alpha)-\alpha\in {\rm Ker}\phi\)。因为\({\rm Ker}\phi\)\(\psi\)-不变子空间,所以\(\psi(\phi(\alpha)-\alpha)\in{\rm Ker}\phi\),即\(\phi\left(\psi(\phi(\alpha)-\alpha)\right)=0\),故
    \begin{equation} \phi\psi(\alpha)=\phi\psi\phi(\alpha).\tag{7.2.1} \end{equation}
    注意到\({\rm Im}\phi\)\(\psi\)-不变子空间,且\(\phi(\alpha)\in{\rm Im}\phi\),所以\(\psi \left(\phi(\alpha)\right)\in {\rm Im}\phi\),即存在\(\beta\in V\)使得\(\psi \left(\phi(\alpha)\right)=\phi(\beta)\)。于是,
    \begin{equation} \phi\psi\phi(\alpha)=\phi\left(\phi(\beta)\right)=\phi^2(\beta)=\phi(\beta)=\psi \left(\phi(\alpha)\right).\tag{7.2.2} \end{equation}
    (7.2.1)(7.2.2)得,\(\phi\psi(\alpha)=\psi\phi(\alpha)\)
5.
\(\phi\)\(n\)维线性空间\(V\)上线性变换。证明:\(\phi\)是数乘变换当且仅当\(V\)的任意一维子空间均为\(\phi\)-不变子空间。
解答.
\(\phi=c{\rm id}_V\)是数乘变换,对\(V\)的任意一维子空间\(U=\langle\alpha\rangle\),由
\begin{equation*} \phi(\alpha)=c{\rm id}_V(\alpha)=c\alpha\in U \end{equation*}
可知\(U\)\(\phi\)-不变子空间。
反之,当\(V\)的任意一维子空间均为\(\phi\)-不变子空间时,设\(\xi_1,\dots,\xi_n\)\(V\)的一个基,则对任意\(1\leq k\leq n\)\(\langle\xi_k\rangle\)\(\phi\)-不变子空间。于是存在\(c_k\in\F\)使得
\begin{equation} \phi(\xi_k)=c_k\xi_k.\tag{7.2.3} \end{equation}
对任意\(1\leq j\neq k\leq n\),根据题设,\(\langle\xi_j+\xi_k\rangle\)也是\(\phi\)-不变子空间,故存在\(c_{jk}\in\F\)使得
\begin{equation} \phi(\xi_j+\xi_k)=c_{jk}(\xi_j+\xi_k).\tag{7.2.4} \end{equation}
而由(7.2.3)
\begin{equation} \phi(\xi_j+\xi_k)=\phi(\xi_j)+\phi(\xi_k)=c_j\xi_j+c_k\xi_k,\tag{7.2.5} \end{equation}
结合(7.2.4)(7.2.5)
\begin{equation*} (c_j-c_{jk})\xi_j+(c_k-c_{jk})\xi_k=0, \end{equation*}
注意到\(\xi_j,\xi_k\)线性无关,故\(c_j-c_{jk}=c_k-c_{jk}=0\),则\(c_j=c_k=c_{jk}\),因此存在\(c\in\F\)使得
\begin{equation*} \phi(\xi_k)=c\xi_k,\ k=1,\dots ,n. \end{equation*}
从而\(\phi=c{\rm id}_V\)是数乘变换。
6.
不借助表示矩阵,直接从线性变换的角度证明命题 7.2.1
解答.
  1. 因为\(\phi\)可逆,所以\(\phi\)是单射,进而可知\(\phi|_U\)是单射。根据 推论 7.1.1,可知\(\phi|_U\)可逆。
  2. 因为\(U\)\(\phi\)-子空间,所以\(\phi(U)\subseteq U\)。又因为\(\phi\)可逆,所以\(\phi(U)= U\)。于是\(\phi^{-1}(U)=\phi^{-1}(\phi(U))=U\),因此\(U\)\(\phi^{-1}\)-子空间。
  3. 首先,\(\phi^{-1}|_U\)\((\phi|_U)^{-1}\)都是从\(U\)\(U\)的线性变换。
    其次,对任意\(\alpha\in U\),由于\(\phi\)可逆且\(U\)\(\phi\)-子空间,所以存在唯一的\(\beta\in U\)使得\(\phi(\beta)=\alpha\),因此
    \begin{align*} (\phi|_U)^{-1}(\alpha)\amp =(\phi|_U)^{-1}(\phi(\beta)) \\ \amp =(\phi|_U)^{-1}(\phi|_U(\beta)) \\ \amp=\beta=\phi^{-1}(\alpha)=\phi^{-1}|_U(\alpha). \end{align*}
7.
\(\phi\)\(n\)维线性空间\(V\)上的线性变换,\(\phi\)在基\((\xi_1,\dots ,\xi_n)\)下的矩阵是
\begin{equation*} \begin{pmatrix} a&0&0&\cdots&0&0\\ 1&a&0&\cdots&0&0\\ 0&1&a&\cdots&0&0\\ \vdots&\vdots&\vdots&&\vdots&\vdots\\ 0&0&0&\cdots&a&0\\ 0&0&0&\cdots&1&a \end{pmatrix}, \end{equation*}
证明:
  1. \(U\)\(\phi\)-子空间,且\(\xi_1\in U\),则\(U=V\)
  2. 对于任意非零\(\phi\)-子空间\(U\),总有\(\xi_n\in U\)
  3. \(V\)不能分解为两个非平凡的\(\phi\)-子空间的直和。
解答.
  1. 依题意,
    \begin{equation*} \phi(\xi_1)=a\xi_1+\xi_2, \end{equation*}
    \begin{equation*} \phi(\xi_2)=a\xi_2+\xi_3, \end{equation*}
    \begin{equation*} \cdots\cdots \end{equation*}
    \begin{equation*} \phi(\xi_{n-1})=a\xi_{n-1}+\xi_n, \end{equation*}
    \begin{equation*} \phi(\xi_n)=a\xi_n. \end{equation*}
    \(U\)\(V\)\(\phi\)-不变子空间,且\(\xi_1\in U\),所以
    \begin{equation*} \phi(\xi_1)-a\xi_1\in U, \end{equation*}
    \(\xi_2\in U\)。于是
    \begin{equation*} \xi_3=\phi(\xi_2)-a\xi_2\in U. \end{equation*}
    依此类推,得\(\xi_1,\xi_2,\dots ,\xi_n\in U\)。因此\(U=V\)
  2. 因为\(U\)\(V\)的非零\(\phi\)-子空间,所以存在\(\alpha\in U\),满足\(\alpha\neq 0\)。设\(\alpha=a_1\xi_1+\dots +a_n\xi_n\),这里\(a_1,\dots ,a_n\)不全为零。假设\(a_i\)\(a_1,\dots ,a_n\)中第一个不为零的数,此时
    \begin{equation*} \alpha=a_i\xi_i+a_{i+1}\xi_{i+1}+\dots +a_n\xi_n\in U. \end{equation*}
    根据\(\phi(\alpha)-a \alpha\in U\)
    \begin{equation*} a_i \xi_{i+1}+a_{i+1}\xi_{i+2}+\dots +a_{n-1}\xi_n\in U. \end{equation*}
    \(\beta=a_i \xi_{i+1}+a_{i+1}+\xi_{i+2}+\cdots +a_{n-1}\xi_n\),根据\(\phi(\beta)-a \beta\in U\)
    \begin{equation*} a_i\xi_{i+2}+a_{i+1}\xi_{i+3}+\cdots +a_{n-2}\xi_n\in U. \end{equation*}
    依此类推,我们有\(a_i\xi_n\in U\),由\(a_i\neq 0\)\(\xi_n=\frac{1}{a_i}(a_i \xi_n)\in U\)
  3. 根据上题,任意两个非平凡\(\phi\)-不变子空间\(V_1,V_2\)必含\(\xi_n\),故\(V_1\bigcap V_2\neq \{0\}\)。从而\(V\)不能分解成两个非平凡\(\phi\)-不变子空间的直和。
8.
\(\phi\)\(n\)维线性空间\(V\)上的线性变换,证明:
  1. \({\rm Ker}\phi\subseteq{\rm Ker}\phi^2\subseteq{\rm Ker}\phi^3\subseteq\cdots\subseteq{\rm Ker}\phi^n\subseteq\cdots\)
  2. \({\rm Im}\phi\supseteq{\rm Im}\phi^2\supseteq{\rm Im}\phi^3\supseteq\cdots\supseteq{\rm Im}\phi^n\supseteq\cdots\)
  3. 存在正整数\(s\),使得\({\rm Ker}\phi^s={\rm Ker}\phi^{s+1}\)
  4. 存在正整数\(t\),使得\({\rm Im}\phi^t={\rm Im}\phi^{t+1}\)
  5. \({\rm Ker}\phi^s={\rm Ker}\phi^{s+1}\),则对于任意正整数\(i\),有\({\rm Ker}\phi^s={\rm Ker}\phi^{s+i}\)
  6. \({\rm Im}\phi^t={\rm Im}\phi^{t+1}\),则对于任意正整数\(i\),有\({\rm Im}\phi^t={\rm Im}\phi^{t+i}\)
  7. \({\rm Ker}\phi^s={\rm Ker}\phi^{s+1}\)的充分必要条件是\({\rm Im}\phi^s={\rm Im}\phi^{s+1}\)
  8. \({\rm Ker}\phi^s={\rm Ker}\phi^{s+1}\),那么\(V={\rm Ker}\phi^s\oplus{\rm Im}\phi^s\)
解答.
  1. 对任意\(m\in\mathbb{Z}^+,\alpha\in{ \rm Ker}\phi^m\),有\(\phi^m(\alpha)=0\),则
    \begin{equation*} \phi^{m+1}(\alpha)=\phi(\phi^m(\alpha))=\phi(0)=0. \end{equation*}
    \({ \rm Ker}\phi^m\subseteq{ \rm Ker}\phi^{m+1}\)
  2. 对任意\(m\in\mathbb{Z}^+\),下证\({\rm Im}\phi^m\supseteq{\rm Im}\phi^{m+1}\)。对任意\(\beta\in{\rm Im}\phi^{m+1}\),存在\(\alpha\in V\),使得\(\beta=\phi^{m+1}(\alpha)\),则
    \begin{equation*} \beta=\phi^m(\phi(\alpha))\in{\rm Im}\phi^m. \end{equation*}
    从而\({\rm Im}\phi^m\supseteq{\rm Im}\phi^{m+1}\)
  3. \({\rm Ker}\phi^m\neq{\rm Ker}\phi^{m+1}\)的充要条件是\(\dim{\rm Ker}\phi^m<\dim{\rm Ker}\phi^{m+1}\)。因为\(V\)是有限维线性空间,所以项 7.2.2.8.a的包含关系不可能全部是真包含。故存在正整数\(s\),使得\({\rm Ker}\phi^s={\rm Ker}\phi^{s+1}\)
  4. \({\rm Im}\phi^m\neq{\rm Im}\phi^{m+1}\)的充要条件是\(\dim{\rm Im}\phi^m>\dim{\rm Im}\phi^{m+1}\)。因为\(V\)是有限维线性空间,所以项 7.2.2.8.b中的包含关系不可能全部是真包含。故存在正整数\(t\),使得\({\rm Im}\phi^t={\rm Im}\phi^{t+1}\)
  5. 我们只需证明对任意\(j>s\),总有\({\rm Ker}\phi^j={\rm Ker}\phi^{j+1}\)。事实上,对任意\(\alpha\in{\rm Ker}\phi^{j+1}\),有\(\phi^{j+1}(\alpha)=0\),即\(\phi^{s+1}(\phi^{j-s}(\alpha))=0\),则\(\phi^{j-s}(\alpha)\in{\rm Ker}\phi^{s+1}\)。注意到\({\rm Ker}\phi^s={\rm Ker}\phi^{s+1}\),所以\(\phi^{j-s}(\alpha)\in{\rm Ker}\phi^s\)。因而
    \begin{equation*} 0=\phi^s(\phi^{j-s}(\alpha))=\phi^j(\alpha), \end{equation*}
    这说明\(\alpha\in\phi^j\),即\({\rm Ker}\phi^{j+1}\subseteq{\rm Ker}\phi^j\)。由项 7.2.2.8.a,有\({\rm Ker}\phi^{j}\subseteq{\rm Ker}\phi^{j+1}\),因此\({\rm Ker}\phi^j={\rm Ker}\phi^{j+1}\)。从而,
    \begin{equation*} {\rm Ker}\phi^s={\rm Ker}\phi^{s+1}={\rm Ker}\phi^{s+2}=\dots={\rm Ker}\phi^{s+i}=\cdots . \end{equation*}
  6. 我们只需证明对任意\(j>t\),总有\({\rm Im}\phi^j={\rm Im}\phi^{j+1}\)。事实上,对任意\(\alpha\in{\rm Im}\phi^{j}\),存在\(\beta\in V\),使得\(\alpha=\phi^{j}(\beta)\),即\(\alpha=\phi^{j-t}(\phi^t(\beta))\)。注意到\(\phi^t(\beta)\in{\rm Im}\phi^t\)\({\rm Im}\phi^{t}={\rm Im}\phi^{t+1}\),所以存在\(\gamma\in V\),使得\(\phi^t(\beta)=\phi^{t+1}(\gamma)\)。因而
    \begin{equation*} \alpha=\phi^{j-t}(\phi^{t+1}(\gamma))=\phi^{j+1}(\gamma)\in{\rm Im}\phi^{j+1}, \end{equation*}
    这说明\({\rm Im}\phi^{j}\subseteq{\rm Im}\phi^{j+1}\)。由项 7.2.2.8.b,有\({\rm Im}\phi^{j+1}\subseteq{\rm Im}\phi^{j}\),因此\({\rm Im}\phi^j={\rm Im}\phi^{j+1}\)。从而,
    \begin{equation*} {\rm Im}\phi^t={\rm Im}\phi^{t+1}={\rm Im}\phi^{t+2}=\cdots={\rm Im}\phi^{t+i}=\cdots . \end{equation*}
  7. 根据维数公式
    \begin{equation*} \dim{\rm Ker}\phi^s+\dim{\rm Im}\phi^s=\dim V=\dim{\rm Ker}\phi^{s+1}+\dim{\rm Im}\phi^{s+1}, \end{equation*}
    结合\({\rm Ker}\phi^s\subseteq{\rm Ker}\phi^{s+1}\)\({\rm Im}\phi^s\supseteq{\rm Im}\phi^{s+1}\),得到
    \begin{equation*} \begin{array}{ccl}{\rm Ker}\phi^s={\rm Ker}\phi^{s+1}&\Leftrightarrow&\dim{\rm Ker}\phi^s=\dim{\rm Ker}\phi^{s+1}\\ &\Leftrightarrow&\dim{\rm Im}\phi^s=\dim{\rm Im}\phi^{s+1}\\ &\Leftrightarrow&{\rm Im}\phi^s={\rm Im}\phi^{s+1}. \end{array} \end{equation*}
  8. 因为\({\rm Ker}\phi^s={\rm Ker}\phi^{s+1}\),所以由项 7.2.2.8.e项 7.2.2.8.f项 7.2.2.8.g
    \begin{equation*} {\rm Ker}\phi^s={\rm Ker}\phi^{2s},{\rm Im}\phi^s={\rm Im}\phi^{2s}. \end{equation*}
    对任意\(\alpha\in V\)\(\phi ^s(\alpha)\in{\rm Im}\phi^s\),所以存在\(\beta\in V\)使得\(\phi ^s(\alpha)=\phi^{2s}(\beta)\),则
    \begin{equation*} \alpha=\phi^s(\beta)+(\alpha-\phi^s(\beta)), \end{equation*}
    这里
    \begin{equation*} \phi^s(\alpha-\phi^s(\beta))=\phi^s(\alpha)-\phi^{2s}(\beta)=0, \end{equation*}
    \(\alpha-\phi^s(\beta)\in{\rm Ker}\phi^s\)。而\(\phi^s(\beta)\in{\rm Im}\phi^s\),故\(V={\rm Ker}\phi^s+{\rm Im}\phi^s\)
    对任意\(\alpha\in{\rm Ker}\phi^s\bigcap{\rm Im}\phi^s\),存在\(\beta\in V\)使得\(\alpha=\phi^s(\beta)\)。又\(\phi^s(\alpha)=0\),所以
    \begin{equation*} 0=\phi^s(\phi^s(\beta))=\phi^{2s}(\beta), \end{equation*}
    \(\beta\in{\rm Ker}\phi^{2s}={\rm Ker}\phi^{s}\)。于是\(\alpha=\phi^s(\beta)=0\),即\({\rm Ker}\phi^s\bigcap{\rm Im}\phi^s=0\)
    综上,\(V={\rm Ker}\phi^s\oplus{\rm Im}\phi^s\)
9.
\(\phi\)\(n\)维线性空间\(V\)上的线性变换,满足\(\dim {\rm Im}\phi^2=\dim{\rm Im}\phi\),证明:\({\rm Im}\phi\bigcap{\rm Ker}\phi=0\)
解答.
对任意\(\alpha\in{\rm Im}\phi\bigcap{\rm Ker}\phi\),存在\(\beta\in V\)使得\(\alpha=\phi(\beta)\)。因\(\phi(\alpha)=0\),所以\(\phi^2(\beta)=0\),即\(\beta\in{\rm Ker}\phi^2\)。由条件\(\dim {\rm Im}\phi^2=\dim{\rm Im}\phi\)\({\rm Im}\phi^2\subseteq{\rm Im}\phi\),得\({\rm Im}\phi^2={\rm Im}\phi\)。根据 项 7.2.2.8.g\({\rm Ker}\phi^2={\rm Ker}\phi\)。故\(\beta\in{\rm Ker}\phi\),即
\begin{equation*} \alpha=\phi(\beta)=0. \end{equation*}
因此\({\rm Im}\phi\bigcap{\rm Ker}\phi=0\)

挑战题.

10.
\(\phi\)\(n\)维线性空间\(V\)上的线性变换,\(\phi\)在基\((\xi_1,\dots ,\xi_n)\)下的矩阵是
\begin{equation*} \begin{pmatrix} a&0&0&\cdots&0&0\\ 1&a&0&\cdots&0&0\\ 0&1&a&\cdots&0&0\\ \vdots&\vdots&\vdots&&\vdots&\vdots\\ 0&0&0&\cdots&a&0\\ 0&0&0&\cdots&1&a \end{pmatrix}, \end{equation*}
\(\phi\)的所有不变子空间。
解答.
依题意知,对任意\(1\leq i\leq n\),子空间\(0,U_i=\langle \xi_i,\xi_{i+1},\dots ,\xi_n\rangle\)\(\phi\)-不变子空间。下证\(V\)\(\phi\)-不变子空间有且只有以上这些。设\(U\)\(V\)的非零\(\phi\)-不变子空间,记
\begin{equation*} i_0=\min\{i\ |\ a_i\neq 0\mbox{且}a_i \xi_i+a_{i+1} \xi_{i+1}+\dots +a_n \xi_n\in U\}, \end{equation*}
\(U\subseteq\langle \xi_{i_0},\xi_{i_0+1},\dots ,\xi_n\rangle\)。我们断言\(U=\langle \xi_{i_0},\xi_{i_0+1},\dots ,\xi_n\rangle\)。事实上,由\(i_0\)的定义知:存在\(a_{i_0},a_{i_0+1},\dots ,a_n\in\mathbb{F}\)使得\(a_{i_0}\xi_{i_0}+a_{i_0+1}\xi_{i_0+1}+\dots +a_n \xi_n\in U\),其中\(a_{i_0}\neq 0\)。由项 7.2.2.7.b\(\xi_n\in U\),所以
\begin{equation*} (a_{i_0}\xi_{i_0}+a_{i_0+1}\xi_{i_0+1}+\cdots +a_n \xi_n)-a_n \xi_n\in U. \end{equation*}
\(\beta_1\triangleq a_{i_0}\xi_{i_0}+a_{i_0+1}\xi_{i_0+1}+\cdots +a_{n-1} \xi_{n-1}\),则\(\beta_1\in U\),于是 \(\phi(\beta_1)-a \beta_1-a_{n-1}\xi_n\in U\), 即
\begin{equation*} \gamma_1\triangleq a_{i_0}\xi_{i_0+1}+a_{i_0+1}\xi_{i_0+2}+\cdots +a_{n-2} \xi_{n-1}\in U, \end{equation*}
再根据\(\phi(\gamma_1)-a \gamma_1-a_{n-2}\xi_n\in U\)
\begin{equation*} \gamma_2\triangleq a_{i_0}\xi_{i_0+2}+a_{i_0+1}\xi_{i_0+3}+\cdots +a_{n-3} \xi_{n-1}\in U, \end{equation*}
依此类推,有\(a_{i_0}\xi_{n-1}\in U\)。由\(a_{i_0}\neq 0\)可知\(\xi_{n-1}\in U\)。于是,
\begin{equation*} \beta_2\triangleq\beta_1-a_{n-1}\xi_{n-1}= a_{i_0} \xi_{i_0}+a_{i_0+1}\xi_{i_0+1}+\cdots +a_{n-2}\xi_{n-2}\in U. \end{equation*}
同理,我们有
\begin{equation*} \phi(\beta_2)-a\beta_2-a_{n-2}\xi_{n-1}\in U, \end{equation*}
\begin{equation*} \eta_1\triangleq a_{i_0}\xi_{i_0+1}+a_{i_0+1}\xi_{i_0+2}+\cdots +a_{n-3}\xi_{n-2}\in U. \end{equation*}
再由\(\phi(\eta_1)-a\eta_1-a_{n-3}\xi_{n-1}\in U\),得
\begin{equation*} \eta_2\triangleq a_{i_0}\xi_{i_0+2}+a_{i_0+1}\xi_{i_0+3}+\cdots +a_{n-4}\xi_{n-3}\in U. \end{equation*}
依此类推,\(a_{i_0}\xi_{n-2}\in U\)。由\(a_{i_0}\neq 0\)可知\(\xi_{n-2}\in U\)。于是,
\begin{equation*} \beta_3\triangleq\beta_2-a_{n-2}\xi_{n-2}= a_{i_0} \xi_{i_0}+a_{i_0+1}\xi_{i_0+1}+\cdots +a_{n-3}\xi_{n-3}\in U. \end{equation*}
重复上述步骤,最后有\(a_{i_0}\xi_{i_0}\in U\)。由\(a_{i_0}\neq 0\)\(\xi_{i_0}\in U\)。因此
\begin{equation*} U=\langle \xi_{i_0},\xi_{i_0+1},\cdots ,\xi_n\rangle . \end{equation*}