高等代数

高等代数理论基础80:若尔当标准形的几何理论(3)

2019-05-01  本文已影响8人  溺于恐

若尔当标准形的几何理论(3)

易知复方阵的若尔当标准形的存在与唯一性

给定复n\times n方阵B,找矩阵T使T^{-1}BT称为若尔当标准形

等同于对给定线性变换\mathscr{A},找一组基使\mathscr{A}在这组基下矩阵成为若尔当标准形

计算步骤

1.取任意n维复线性空间V及它的一组基\varepsilon_1,\varepsilon_2,\cdots,\varepsilon_n,作线性变换\mathscr{A}使它在该基下矩阵为B,即有\mathscr{A}(\varepsilon_1,\varepsilon_2,\cdots,\varepsilon_n)=(\varepsilon_1,\varepsilon_2,\cdots,\varepsilon_n)B,令A=B',则有

\mathscr{A}\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\vdots\\\varepsilon_n\end{pmatrix}=A\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\vdots\\\varepsilon_n\end{pmatrix}

(\mathscr{A}E-\mathscr{E}A)\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\vdots\\\varepsilon_n\end{pmatrix}=O

2.找可逆\lambda-矩阵P(\lambda),Q(\lambda),使得

P(\lambda)(\lambda E-A)Q(\lambda)=\begin{pmatrix}h_1(\lambda)\\&h_2(\lambda)\\& &\ddots\\& & &h_n(\lambda)\end{pmatrix}
其中设h_1(\lambda),\cdots,h_s(\lambda)\{h_i(\lambda),1\le i\le n\}中全部非常数多项式,且不妨设所有的h_i(\lambda)首项系数为1

h_{s+1}(\lambda)=\cdots=h_n(\lambda)=1

\begin{pmatrix}\eta_1\\\eta_2\\\vdots\\\eta_n\end{pmatrix}=Q^{-1}(\mathscr{A})\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\vdots\\\varepsilon_n\end{pmatrix}

\eta_i的最小多项式为h_i(\lambda)

h_{s+1}(\mathscr{A})=\cdots=h_n(\mathscr{A})=\mathscr{E}

\mathscr{E}\eta_{s+1}=\mathscr{E}\eta_{s+2}=\cdots=\mathscr{E}\eta_n=0

h_1(\lambda),h_2(\lambda),\cdots,h_s(\lambda)非常数

分别为\eta_1,\eta_2,\cdots,\eta_s的最小多项式

\mathscr{E}\eta_1\neq 0,\cdots,\mathscr{E}\eta_s\neq 0

V=P[\mathscr{A}]\eta_1\oplus\cdots\oplus P[\mathscr{A}]\eta_s

3.分解各P[\mathscr{A}]\eta_i

\eta_i的最小多项式h_i(\lambda)有分解h_i(\lambda)=(\lambda-\lambda_1)^{l_1}\cdots(\lambda-\lambda_t)^{l_t}

\lambda_i为各不相同的复数

m_i(\lambda)={h_i(\lambda)\over (\lambda-\lambda_i)^{l_i}},\xi_{ij}=m_j(\mathscr{A})\eta_i,j=1,2,\cdots,t

P[\mathscr{A}]\eta_i=P[\mathscr{A}]\xi_{i1}\oplus\cdots\oplus P[\mathscr{A}]\xi_{it}

\xi_{it}的最小多项式为(\lambda-\lambda_j)^{l_j},j=1,2,\cdots,t

4.对每个\xi_{ij},作P[\mathscr{A}]\xi_{ij}的基\xi_{ij},(\mathscr{A}-\lambda_j\mathscr{E})\xi_{ij},\cdots,(\mathscr{A}-\lambda_j\mathscr{E})^{l_j-1}\xi_{ij}

\mathscr{A}|_{p[\mathscr{A}]\xi_{ij}}在这组基下矩阵为若尔当块J(\lambda_j,l_i)

对每个P[\mathscr{A}]\xi_{ij}这样做,则将所有P[\mathscr{A}]\xi_{ij}作出的基合起来即V的基

在该组基下,\mathscr{A}的矩阵J即它的若尔当标准形

5.以上步骤中各\eta_i可具体计算,\xi_{ij}也可计算

P[\mathscr{A}]\xi_{ij}的基\xi_{ij},(\mathscr{A}-\lambda_j\mathscr{E})\xi_{ij},\cdots,(\mathscr{A}-\lambda_j\mathscr{E})^{l_j-1}\xi_{ij}及V的使\mathscr{A}对角化的矩阵或若尔当标准形J的新基为(\varepsilon_1,\varepsilon_2,\cdots,\varepsilon_n)T

\mathscr{A}在基\varepsilon_1,\varepsilon_2,\cdots,\varepsilon_n及新基下的矩阵分别为BJ

J=T^{-1}BT,同时也得到所要的相似变换

例:对矩阵B=\begin{pmatrix}0&-1&2\\3&7&-14\\3&6&-10\end{pmatrix}

求T,使T^{-1}BT为若尔当标准形

解:

1.化成线性变换问题

取三维复线性空间V及它的一组基\varepsilon_1,\varepsilon_2,\varepsilon_3

\mathscr{A}使\mathscr{A}(\varepsilon_1,\varepsilon_2,\varepsilon_3)=(\varepsilon_1,\varepsilon_2,\varepsilon_3)B

A=B',则

\mathscr{A}\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\varepsilon_3\end{pmatrix}=A\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\varepsilon_3\end{pmatrix}=\begin{pmatrix}0&3&3\\-&8&6\\2&-14&-10\end{pmatrix}\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\varepsilon_3\end{pmatrix}

2.化\lambda E-A为对角形

可作多次初等变换

\lambda E-A=\begin{pmatrix}\lambda&-3&-3\\ 1&\lambda-8&-6\\ -2&14&\lambda+10\end{pmatrix}

\xrightarrow{行变换}\begin{pmatrix}1&\lambda-8&-6\\ \lambda&-3&-3\\ -2&14&\lambda+10\end{pmatrix}

\xrightarrow{行变换}\begin{pmatrix}1&\lambda-8&-6\\ 0&-3-(\lambda-8)\lambda&-3+6\lambda\\ 0&14+2(\lambda-8)&\lambda+10-12\end{pmatrix}

=\begin{pmatrix}1&\lambda-8&-6\\ 0&-\lambda^2+8\lambda-3&-3+6\lambda\\ 0&2\lambda-2&\lambda-2\end{pmatrix}

\xrightarrow[\begin{pmatrix}1&-(\lambda-8)&6\\0&1&0\\0&-1&1\end{pmatrix}]{右乘}\begin{pmatrix}1&0&0\\ 0&-\lambda^2+8\lambda-3&-3+6\lambda\\ 0&2\lambda-2&\lambda-2\end{pmatrix}

\xrightarrow[\begin{pmatrix}1&0&0\\0&1&0\\0&-1&1\end{pmatrix}]{右乘}\begin{pmatrix}1&0&0\\ 0&-\lambda(\lambda-2)&6\lambda-3\\ 0&\lambda&\lambda-2\end{pmatrix}

\xrightarrow{行变换}\begin{pmatrix}1&0&0\\ 0&\lambda&\lambda-2\\ 0&0&\lambda^2+2\lambda+1\end{pmatrix}

\xrightarrow{行变换}\begin{pmatrix}1&0&0\\ 0&{1\over 2}\lambda&{1\over 2}\lambda-1\\ 0&0&(\lambda+1)^2\end{pmatrix}

\xrightarrow[\begin{pmatrix}1&0&0\\ 0&1&0\\ 0&-1&1\end{pmatrix}]{右乘}\begin{pmatrix}1&0&0\\ 0&1&{1\over 2}\lambda-1\\ 0&-(\lambda+1)^2&(\lambda+1)^2\end{pmatrix}

\xrightarrow[\begin{pmatrix}1&0&0\\ 0&1&-({1\over 2}\lambda-1)\\ 0&0&1\end{pmatrix}]{右乘}\begin{pmatrix}1&0&0\\ 0&1&0\\ 0&-(\lambda+1)^2&{1\over 2}(\lambda+1)^2\lambda\end{pmatrix}

\xrightarrow{行变换}\begin{pmatrix}1&0&0\\ 0&1&0\\ 0&0&\lambda(\lambda+1)^2\end{pmatrix}

可得

Q(\lambda)=\begin{pmatrix}1&-(\lambda-8)&6\\0&1&0\\0&0&1\end{pmatrix}\begin{pmatrix}1&0&0\\0&1&0\\0&-1&1\end{pmatrix}^2\begin{pmatrix}1&0&0\\0&1&-({1\over 2}\lambda-1)\\0&0&1\end{pmatrix}

Q^{-1}(\lambda)=\begin{pmatrix}1&0&0\\0&1&{1\over 2}\lambda-1\\0&0&1\end{pmatrix}\begin{pmatrix}1&0&0\\0&1&0\\0&1&1\end{pmatrix}^2\begin{pmatrix}1&\lambda-8&-6\\0&1&0\\0&0&1\end{pmatrix}

=\begin{pmatrix}1&\lambda-8&-6\\0&\lambda-1&{1\over 2}\lambda-1\\0&2&1\end{pmatrix}

\begin{pmatrix}\eta_1\\\eta_2\\\eta_3\end{pmatrix}=Q^{-1}(\mathscr{A})\begin{pmatrix}\varepsilon_1\\\varepsilon_2\\\varepsilon_3\end{pmatrix}

由于\eta_1,\eta_2的零化多项式为1

\eta_1=\eta_2=0,\eta_3=2\varepsilon_2+\varepsilon_3

\therefore V=P[\mathscr{A}\eta_3]=P[\mathscr{A}](2\varepsilon_2+\varepsilon_3)

3.继续分解P[\mathscr{A}](2\varepsilon_2+\varepsilon_3)

\eta_3的最小多项式为\lambda(\lambda+1)^2

P[\mathscr{A}]\eta_3=P[\mathscr{A}]\eta_1'\oplus P[\mathscr{A}]\eta_2'

\eta_1'=(\mathscr{A}+\mathscr{E})^2\eta_3,\eta_2'=\mathscr{A}\eta_3

它们的最小多项式分别为\lambda(\lambda+1)^2

\eta_1'=(\mathscr{A}+\mathscr{E})^2(2\varepsilon_2+\varepsilon_3)=2(\mathscr{A}+\mathscr{E})^2\varepsilon_2+(\mathscr{A}+\mathscr{E})^2\varepsilon_3

(\mathscr{A}+\mathscr{E})^2\varepsilon_1,\varepsilon_2,\varepsilon_3下的左矩阵为

\begin{pmatrix}1&3&3\\-1&9&6\\2&-14&-9\end{pmatrix}^2=\begin{pmatrix}*&*&*\\2&-6&-3\\-2&6&3\end{pmatrix}

\eta_1'=2(\mathscr{A}+\mathscr{E})^2\varepsilon_2+(\mathscr{A}+\mathscr{E})^2\varepsilon_3

=2(2\varepsilon_1+6\varepsilon_2-3\varepsilon_3)+(-2\varepsilon_2+6\varepsilon_2+3\varepsilon_3)

=2\varepsilon_1-6\varepsilon_2-3\varepsilon_3

P[\mathscr{A}]\eta_1'是1维的,基就是\eta_1'

\mathscr{A}\eta_1'=0

\mathscr{A}|_{P[\mathscr{A}]\eta_1'}\eta_1'下的矩阵是一级矩阵(0)

\eta_2'的最小多项式为(\lambda+1)^2

\eta_2'=\mathscr{A}\eta_3=\mathscr{A}(2\varepsilon_2+\varepsilon_3)

=2\mathscr{A}\varepsilon_2+\mathscr{A}\varepsilon_3

\eta_2'=2\mathscr{A}\varepsilon_2+\mathscr{A}\varepsilon_3

=2(-\varepsilon_2+8\varepsilon_2+6\varepsilon_3)+2\varepsilon_1-14\varepsilon_2-10\varepsilon_3

=2\varepsilon_2+2\varepsilon_3

P[\mathscr{A}]\eta_2'是2维的,\eta_2'的最小多项式为(\lambda+1)^2

则它的基为\eta_2'=2(\varepsilon_2+\varepsilon_3)

\eta_3'=(\mathscr{A}+\mathscr{E})\eta_2'=2(\mathscr{A}\varepsilon_2+\mathscr{A}\varepsilon_3+\varepsilon_2+\varepsilon_3)

\eta_3'=2\varepsilon_1-10\varepsilon_2-6\varepsilon_3

(\mathscr{A}+\mathscr{E})\eta_3'=(\mathscr{A}+\mathscr{E})^2\eta_2'=0

\mathscr{A}|_{P[\mathscr{A}\eta_2']}在基\eta_2',\eta_3'下的矩阵为J(-1,2)

最后\mathscr{A}在基\eta_1',\eta_2',\eta_3'下的矩阵为

J=\begin{pmatrix}0&0&0\\0&-1&0\\0&1&-1\end{pmatrix}

基变换

(\eta_1',\eta_2',\eta_3')=(\varepsilon_1,\varepsilon_2,\varepsilon_3)T=(\varepsilon_1,\varepsilon_2,\varepsilon_3)\begin{pmatrix}2&0&2\\-6&2&-10\\-3&2&-6\end{pmatrix}

相似变换为J=T^{-1}BT

\begin{pmatrix}2&0&2\\-6&2&-10\\-3&2&-6\end{pmatrix}^{-1}\begin{pmatrix}0^-1&2\\3&8&-14\\3&6&-10\end{pmatrix}\begin{pmatrix}2&0&2\\-6&2&-10\\-3&2&-6\end{pmatrix}=\begin{pmatrix}0&0&0\\0&-1&0\\0&1&-1\end{pmatrix}

上一篇下一篇

猜你喜欢

热点阅读