美文网首页高等代数
高等代数理论基础78:若尔当标准形的几何理论(1)

高等代数理论基础78:若尔当标准形的几何理论(1)

作者: 溺于恐 | 来源:发表于2019-04-29 07:53 被阅读3次

    若尔当标准形的几何理论(1)

    找一组基使线性变换\mathscr{A}在这组基下的矩阵称为若尔当标准形

    定义:对于线性空间V中的线性变换\mathscr{A}的多项式f(\mathscr{A})及任意向量\varepsilon,若有f(\mathscr{A})\varepsilon=0,则称f(\lambda)\varepsilon对于\mathscr{A}的零化多项式,若f(\lambda)\varepsilon对于\mathscr{A}的零化多项式中次数最低的首一多项式,则称f(\lambda)\varepsilon对于\mathscr{A}的最小多项式

    易证\varepsilon\mathscr{A}的最小多项式整除\varepsilon\mathscr{A}的任一零化多项式

    引理:对\C上有限维空间V上的线性变换\mathscr{A},下列结论等价

    1.\mathscr{A}在基\varepsilon_0,\varepsilon_1,\cdots,\varepsilon_{n-1}下的矩阵是若尔当块

    J(\lambda_0,n)=\begin{pmatrix}\lambda_0&0&0&\cdots&0&0&0\\ 1&\lambda_0&0&\cdots&0&0&0\\ \vdots&\vdots&\vdots& &\vdots&\vdots&\vdots\\ 0&0&0&\cdots&1&\lambda_0&0\\ 0&0&0&\cdots&0&1&\lambda_0\end{pmatrix}​

    2.\varepsilon_0,\varepsilon_1=(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0,\varepsilon_2=(\mathscr{A}-\lambda_0\mathscr{E})^2\varepsilon_0,\cdots,\varepsilon_{n-1}=(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0V的基且(\mathscr{A}-\lambda_0\mathscr{E})^n\varepsilon_0=(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_{n-1}=0

    3.V=P[\mathscr{A}]\varepsilon_0=\{f(\mathscr{A}\varepsilon_0|f(\lambda)\in\C[\lambda]\},且(\lambda-\lambda_0)^n\varepsilon_0的最小多项式

    证明:

    1\Leftrightarrow 2

    由线性变换矩阵的定义,显然成立

    2\Leftrightarrow 3

    必要性

    \forall v\in V,有

    v=l_0\varepsilon_0+l_1(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0

    =l_0\mathscr{E}+l_1(\mathscr{A}-\lambda_0\mathscr{E})+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0\in P[\mathscr{A}]\varepsilon_0

    此时(\lambda-\lambda_0)^n\varepsilon_0的一个零化多项式

    设为f(\lambda)=l_0+l_1(\lambda-\lambda_0)+\cdots+l_{n-1}(\lambda-\lambda_0)^{n-1}

    f(\mathscr{A})\varepsilon_0=0

    [l_0\mathscr{E}+l_1(\mathscr{A}-\lambda_0\mathscr{E})+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}]\varepsilon_0

    =l_0\varepsilon_0+l_1(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0=0

    \varepsilon_0,(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0,\cdots,(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0V的一组基,线性无关

    l_0=l_1=\cdots=l_{n-1}=0

    f(\lambda)=0

    (\lambda-\lambda_0)^n\varepsilon_0的最小多项式

    充分性

    首先(\lambda-\lambda_0)^n\varepsilon_0的零化多项式

    (\mathscr{A}-\lambda_0\mathscr{E})^n\varepsilon_0=0

    \forall v\in V=P[\mathscr{A}]\varepsilon_0

    f(\lambda)\in P[\lambda],v=f(\mathscr{A})\varepsilon_0

    作带余除法,f(\lambda)=(\lambda-\lambda_0)^ng(\lambda)+l_0+l_1(\lambda-\lambda_0)+\cdots+l_{n-1}(\lambda-\lambda_0)^{n-1}

    则有f(\mathscr{A})\varepsilon_0=g(\mathscr{A})(\mathscr{A}-\lambda_0\mathscr{E})^n\varepsilon_0+l_0\varepsilon_0+l_1(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0

    =l_0\varepsilon_0+l_1(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0

    \forall v\in V\varepsilon_0,(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0,\cdots,(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0​的线性组合

    l_0\varepsilon_0+l_1(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0=0

    [l_0\mathscr{E}+l_1(\mathscr{A}-\lambda_0\mathscr{E})+\cdots+l_{n-1}(\mathscr{A}-\lambda_0\mathscr{E})]\varepsilon_0=0

    r(\lambda)=l_0+l_1(\lambda-\lambda_0)+\cdots+l_{n-1}(\lambda-\lambda_0)^{n-1}

    r(\mathscr{A})\varepsilon_0=0

    r(\lambda)\neq 0,则\partial (r(\lambda))\le n-1

    (\lambda-\lambda_0)^n\varepsilon_0的最小多项式矛盾

    r(\lambda)=0

    l_0=l_1=\cdots=l_{n-1}=0

    即证\varepsilon_0,(\mathscr{A}-\lambda_0\mathscr{E})\varepsilon_0,\cdots,(\mathscr{A}-\lambda_0\mathscr{E})^{n-1}\varepsilon_0线性无关

    故为V的基\qquad\mathcal{Q.E.D}

    定理:V,\mathscr{A}如上

    \mathscr{A}在某基下的矩阵为若尔当形

    A=\begin{pmatrix}J(\lambda_1,k_1)\\&J(\lambda_2,k_2)\\& &\ddots\\& & &J(\lambda_s,k_s)\end{pmatrix}​

    的充要条件为V中存在\eta_1,\eta_2,\cdots,\eta_s,使

    V=P[\mathscr{A}]\eta_1\oplus P[\mathscr{A}]\eta_2\oplus \cdots\oplus P[\mathscr{A}]\eta_s

    且每个\eta_i的最小多项式是(\lambda-\lambda_i)^{k_i}

    证明:

    A=\begin{pmatrix}J(\lambda_1,k_1)\\&J(\lambda_2,k_2)\\& &\ddots\\& & &J(\lambda_s,k_s)\end{pmatrix}

    \Leftrightarrow V=W_1\oplus W_2\oplus \cdots\oplus W_s\mathscr{A}-不变子空间的直和

    且每个\mathscr{A}|_{W_i}W_i上有基使它的矩阵是J(\lambda_i,k_i)

    V=W_1\oplus \cdots\oplus W_s,对每个i,有\eta_i\in W_i使W_i=P[\mathscr{A}]\eta_i

    \eta_i\mathscr{A}的最小多项式为(\lambda-\lambda_i)^{k_i},i=1,2,\cdots,s\qquad\mathcal{Q.E.D}​

    注:定理说明,要证若尔当标准形存在,只需证存在不变子空间的直和分解

    相关文章

      网友评论

        本文标题:高等代数理论基础78:若尔当标准形的几何理论(1)

        本文链接:https://www.haomeiwen.com/subject/mxenwqtx.html