单位矩阵为 I I I,矩阵 A A A的行列式记作 det ( A ) \det \left( A \right) det(A),伴随矩阵记作 a d j ( A ) \mathrm{adj} \left(A\right) adj(A).
矩阵 A A A的特征多项式定义为:
χ A ( s ) ≜ det ( s I − A ) = s n + d 1 s n − 1 + ⋯ + d n , \chi _A\left( s \right) \triangleq \det \left( sI-A \right) =s^n+d_1s^{n-1}+\cdots +d_n, χA(s)≜det(sI−A)=sn+d1sn−1+⋯+dn,
χ A ( A ) = A n + d 1 A n − 1 + ⋯ + d n = 0 \chi _A\left( A \right) =A^n+d_1A^{n-1}+\cdots +d_n=0 χA(A)=An+d1An−1+⋯+dn=0
考虑矩阵 ( s I − A ) \left( sI-A \right) (sI−A)的逆,可以表示为:
( s I − A ) − 1 = a d j ( s I − A ) det ( s I − A ) ⋯ ( ∗ ) , \left( sI-A \right) ^{-1}=\frac{\,\,\mathrm{adj}\left( sI-A \right)}{\det \left( sI-A \right)}\cdots \left( * \right) , (sI−A)−1=det(sI−A)adj(sI−A)⋯(∗),
其中 ( s I − A ) \left( sI-A \right) (sI−A)的行列式可以表示为 { 1 , s , s 2 , ⋯ , s n } \left\{ 1,s,s^2,\cdots ,s^n \right\} {1,s,s2,⋯,sn}的线性组合,即
det ( s I − A ) = s n + d 1 s n − 1 + ⋯ + d n . \det \left( sI-A \right) =s^n+d_1s^{n-1}+\cdots +d_n. det(sI−A)=sn+d1sn−1+⋯+dn.
而 ( s I − A ) \left( sI-A \right) (sI−A)的伴随矩阵可以表示为 { 1 , s , s 2 , ⋯ , s n − 1 } \left\{ 1,s,s^2,\cdots ,s^{n-1} \right\} {1,s,s2,⋯,sn−1}的线性组合,即
a d j ( s I − A ) = B 0 s n − 1 + B 1 s n − 2 + ⋯ + B n − 1 . \mathrm{adj}\left( sI-A \right) =B_0s^{n-1}+B_1s^{n-2}+\cdots +B_{n-1}. adj(sI−A)=B0sn−1+B1sn−2+⋯+Bn−1.(注:根据伴随矩阵的定义,可以知道多项式最高阶数为 ( n − 1 ) (n-1) (n−1))
下证:
χ A ( A ) = A n + d 1 A n − 1 + ⋯ + d n = 0. \chi _A\left( A \right) =A^n+d_1A^{n-1}+\cdots +d_n=0. χA(A)=An+d1An−1+⋯+dn=0.
对 ( ∗ ) (*) (∗)式,在等号两边右乘 ( s I − A ) \left( sI-A \right) (sI−A),左乘
det ( s I − A ) I \det \left( sI-A \right) I det(sI−A)I,可以得到
a d j ( s I − A ) ( s I − A ) = det ( s I − A ) I . \mathrm{adj}\left( sI-A \right) \left( sI-A \right) =\det \left( sI-A \right) I. adj(sI−A)(sI−A)=det(sI−A)I.
对等号左边进行展开
L H S = ( B 0 s n − 1 + B 1 s n − 2 + ⋯ + B n − 1 ) ( s I − A ) = ( B 0 s n + B 1 s n − 1 + ⋯ + B n − 1 s ) − ( B 0 A s n − 1 + B 1 A s n − 2 + ⋯ + B n − 1 A ) = B 0 s n + ( B 1 − B 0 A ) s n − 1 + ⋯ + ( B n − 1 − B n − 2 A ) s − B n − 1 A . \begin{aligned} \mathrm{LHS}&=\left( B_0s^{n-1}+B_1s^{n-2}+\cdots +B_{n-1} \right) \left( sI-A \right) \\ &=\left( B_0s^n+B_1s^{n-1}+\cdots +B_{n-1}s \right) -\left( B_0As^{n-1}+B_1As^{n-2}+\cdots +B_{n-1}A \right) \\ &=B_0s^n+\left( B_1-B_0A \right) s^{n-1}+\cdots +\left( B_{n-1}-B_{n-2}A \right) s-B_{n-1}A. \end{aligned} LHS=(B0sn−1+B1sn−2+⋯+Bn−1)(sI−A)=(B0sn+B1sn−1+⋯+Bn−1s)−(B0Asn−1+B1Asn−2+⋯+Bn−1A)=B0sn+(B1−B0A)sn−1+⋯+(Bn−1−Bn−2A)s−Bn−1A.
而等式右边为
R H S = s n I + d 1 s n − 1 I + ⋯ + d n I . \mathrm{RHS}=s^nI+d_1s^{n-1}I+\cdots +d_nI. RHS=snI+d1sn−1I+⋯+dnI.
对照系数,有
{ B 0 = I B 1 − B 0 A = d 1 I ⋮ B n − 1 − B n − 2 A = d n − 1 I O − B n − 1 A = d n I ⇒ { B 0 A n = A n B 1 A n − 1 − B 0 A n − 2 = d 1 A n − 1 ⋮ B n − 1 A − B n − 2 A 2 = d n − 1 A − B n − 1 A = d n I \begin{cases} B_0=I\\ B_1-B_0A=d_1I\\ \vdots\\ B_{n-1}-B_{n-2}A=d_{n-1}I\\ O-B_{n-1}A=d_nI\\ \end{cases}\Rightarrow \begin{cases} B_0A^n=A^n\\ B_1A^{n-1}-B_0A^{n-2}=d_1A^{n-1}\\ \vdots\\ B_{n-1}A-B_{n-2}A^2=d_{n-1}A\\ -B_{n-1}A=d_nI\\ \end{cases} ⎩ ⎨ ⎧B0=IB1−B0A=d1I⋮Bn−1−Bn−2A=dn−1IO−Bn−1A=dnI⇒⎩ ⎨ ⎧B0An=AnB1An−1−B0An−2=d1An−1⋮Bn−1A−Bn−2A2=dn−1A−Bn−1A=dnI
上式等号左边累加结果为 O O O(零矩阵),而右边累加为 A A A的特征多项式 χ A ( A ) \chi _A\left( A \right) χA(A),得证.
TODO