本文已参与「新人创作礼」活动,一起开启创作之路。
特征值、特征向量
定义:设AAA是nnn阶矩阵,\alpha是nnn维非000列向量,且
A=A \alpha=\lambda \alphaA=
则称\lambda是矩阵AAA的特征值,\alpha是矩阵AAA对应于特征值\lambda的特征向量
由A=,≠0⇒(E−A)=0,(E−A)x=0⇒A \alpha=\lambda \alpha,\alpha\ne0\Rightarrow (\lambda E-A)\alpha=0,(\lambda E-A)x=0\Rightarrow \alphaA=,=0⇒(E−A)=0,(E−A)x=0⇒是齐次方程组(E−A)x=0(\lambda E-A)x=0(E−A)x=0的非000解
-
由∣E−A∣=0|\lambda E-A|=0∣E−A∣=0求特征值i\lambda_{i}i,共nnn个(含重根)
-
对(iE−A)x=0(\lambda_{i}E-A)x=0(iE−A)x=0,求根底解系,即特征值i\lambda_{i}i的线性无关的特征向量,写通解得i\lambda_{i}i一切的特征向量
定理:如果1,2\alpha_{1},\alpha_{2}1,2都是矩阵AAA对应于特征值\lambda的特征向量,则当k11+k22≠0k_{1}\alpha_{1}+k_{2}\alpha_{2}\ne0k11+k22=0时,k11+k22k_{1}\alpha_{1}+k_{2}\alpha_{2}k11+k22仍是矩阵AAA关于特征值\lambda的特征向量
定理:如果1\lambda_{1}1与2\lambda_{2}2是AAA不同的特征值,对应的特征向量分别是1\alpha_{1}1和2\alpha_{2}2,则1,2\alpha_{1},\alpha_{2}1,2必定线性无关
定理:设AAA是nnn阶矩阵,特征值是1,2,⋯ ,n\lambda_{1},\lambda_{2},\cdots,\lambda_{n}1,2,⋯,n,则有
例:求A=(17−2−2−214−4−2−414)A=\begin{pmatrix}17 & -2 & -2 \\ -2 & 14 & -4 \\ -2 & -4 & 14\end{pmatrix}A=⎝⎛17−2−2−214−4−2−414⎠⎞特征值,特征向量
三阶队伍式主对角线元素都含有未知数,直接打开要解三次方程,一般考虑通过队伍加减,找出某一行或某一列一切的元素都含有含未知数的公因式或000
由AAA的特征多项式
∣E−A∣=∣−17222−14424−14∣=∣−17222−144018−−18∣=∣−17422−10400−18∣=(−18)∣−1742−10∣=(−18)(2−27+162)=(−18)2(−9)\begin{aligned} |\lambda E-A|&=\begin{vmatrix} \lambda-17 & 2 & 2 \\ 2 & \lambda-14 & 4 \\ 2 & 4 & \lambda-14 \end{vmatrix}=\begin{vmatrix} \lambda-17&2&2\\2&\lambda-14&4\\0&18-\lambda&\lambda-18 \end{vmatrix}\\ &=\begin{vmatrix} \lambda-17&4&2\\2&\lambda-10&4\\0&0&\lambda-18 \end{vmatrix}=(\lambda-18)\begin{vmatrix} \lambda-17&4\\2&\lambda-10 \end{vmatrix}\\ &=(\lambda-18)(\lambda^{2}-27\lambda+162)=(\lambda-18)^{2}(\lambda-9) \end{aligned}∣E−A∣=∣∣∣∣∣∣∣−17222−14424−14∣∣∣∣∣∣∣=∣∣∣∣∣∣∣−17202−1418−24−18∣∣∣∣∣∣∣=∣∣∣∣∣∣∣−17204−10024−18∣∣∣∣∣∣∣=(−18)∣∣∣∣∣−1724−10∣∣∣∣∣=(−18)(2−27+162)=(−18)2(−9)
因而1=2=18,3=9\lambda_{1}=\lambda_{2}=18,\lambda_{3}=91=2=18,3=9
当=18\lambda=18=18时,(18E−A)x=0(18E-A)x=0(18E−A)x=0
(122244244)→(122000000)\begin{aligned} \begin{pmatrix} 1 & 2 & 2 \\ 2 & 4 & 4 \\ 2 & 4 & 4 \end{pmatrix}\rightarrow \begin{pmatrix} 1 & 2 & 2 \\ 0 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix} \end{aligned}⎝⎛122244244⎠⎞→⎝⎛100200200⎠⎞
得根底解系:1=(−2,1,0)T,2=(−2,0,1)T\alpha_{1}=(-2,1,0)^{T},\alpha_{2}=(-2,0,1)^{T}1=(−2,1,0)T,2=(−2,0,1)T,因而特征向量k11+k22,k1,k2k_{1}\alpha_{1}+k_{2}\alpha_{2},k_{1},k_{2}k11+k22,k1,k2不一起为000
当=9\lambda=9=9时,(9E−A)x=0(9E-A)x=0(9E−A)x=0
(−8222−5424−5)→(20−101−1000)\begin{pmatrix} -8 & 2 & 2 \\ 2 & -5 & 4 \\ 2 & 4 & -5 \end{pmatrix}\rightarrow \begin{pmatrix} 2 & 0 & -1 \\ 0 & 1 & -1 \\ 0 & 0 & 0 \end{pmatrix}⎝⎛−8222−5424−5⎠⎞→⎝⎛200010−1−10⎠⎞
得根底解系3=(1,2,2)T\alpha_{3}=(1,2,2)^{T}3=(1,2,2)T,因而特征向量k33,k3≠0k_{3}\alpha_{3},k_{3}\ne0k33,k3=0
关于一个矩阵若,秩等于111即存在一行能表明其他一切行,秩等于222即存在两行能表明其他一切行,之后同理,类似最大线性无关组
(−8222−5424−5)\begin{pmatrix}-8 & 2 & 2 \\ 2 & -5 & 4 \\ 2 & 4 & -5\end{pmatrix}⎝⎛−8222−5424−5⎠⎞求根底解系,本题题解省略的化简步骤,其实化简步骤并不容易,但此处能够用其他思路。
因为已知∣E−A∣=0|\lambda E-A|=0∣E−A∣=0,即该矩阵的秩一定小于333;随意选两行,这里选二三行,发现二者线性无关(不成比例),因而该矩阵秩大于111,可得该矩阵秩为222。
因为这两行能线性表明另一行,因而能够结构新的矩阵
(2−5424−5000)\begin{pmatrix}2 & -5 & 4 \\ 2 & 4 & -5 \\ 0 & 0 & 0\end{pmatrix}⎝⎛220−5404−50⎠⎞
能够理解为因为这两行能线性表明另一行,因而另一行一定能被全消为000
此时只需对新的矩阵行化简即可
设A=,≠0A \alpha=\lambda \alpha,\alpha\ne0A=,=0,则有
例:AAA为333阶矩阵,特征值是−1,0,4-1,0,4−1,0,4,如果A+B=2EA+B=2EA+B=2E,则BBB的特征值为()
设A=,≠0A \alpha=\lambda \alpha,\alpha\ne0A=,=0
由A+B=2EA+B=2EA+B=2E,有B=2E−AB=2E-AB=2E−A,则
B=(2E−A)=2−A=(2−)B \alpha=(2E-A)\alpha=2\alpha-A \alpha=(2-\lambda)\alphaB=(2E−A)=2−A=(2−)
因而BBB的特征值为3,2,−23,2,-23,2,−2
例:AAA是333阶矩阵,A2+2A−3E=0A^{2}+2A-3E=0A2+2A−3E=0,证明矩阵AAA的特征值只能是111或−3-3−3
设\lambda是AAA的任一特征值,对应的特征向量是\alpha,即A=,≠0A \alpha=\lambda \alpha,\alpha\ne0A=,=0,那么
A2=2A^{2}\alpha=\lambda^{2} \alphaA2=2
由A2+2A−3E=0A^{2}+2A-3E=0A2+2A−3E=0
有
A2+2A−3=0(2+2−3)=0,≠02+2−3=0\begin{aligned} A^{2}\alpha+2A \alpha-3\alpha&=0\\ (\lambda^{2}+2\lambda-3)\alpha&=0,\alpha\ne0\\ \lambda^{2}+2\lambda-3&=0 \end{aligned}A2+2A−3(2+2−3)2+2−3=0=0,=0=0
因而=1\lambda=1=1或=−3\lambda=-3=−3
从标题条件只能推出矩阵特征值只能是111或−3-3−3
例如:(111),(333),(1000−1−20−2−1)\begin{pmatrix}1 & & \\ & 1 & \\ & & 1\end{pmatrix},\begin{pmatrix}3 & & \\ & 3 & \\ & & 3\end{pmatrix},\begin{pmatrix}1 & 0 & 0 \\ 0 & -1 & -2 \\ 0 & -2 & -1\end{pmatrix}⎝⎛111⎠⎞,⎝⎛333⎠⎞,⎝⎛1000−1−20−2−1⎠⎞
例:已知=(1,1,−1)T\alpha=(1,1,-1)^{T}=(1,1,−1)T是A=(2−125a3−1b−2)A=\begin{pmatrix}2 & -1 & 2 \\ 5 & a & 3 \\ -1 & b & -2\end{pmatrix}A=⎝⎛25−1−1ab23−2⎠⎞的一个特征向量,则a=(),b=()a=(),b=()a=(),b=()
设A=,≠0A \alpha=\lambda \alpha,\alpha\ne0A=,=0
(2−125a3−1b−2)(11−1)=(11−1)\begin{pmatrix}2 & -1 & 2 \\ 5 & a & 3 \\ -1 & b & -2\end{pmatrix}\begin{pmatrix} 1 \\ 1 \\ -1 \end{pmatrix}=\lambda \begin{pmatrix} 1 \\ 1 \\ -1 \end{pmatrix}⎝⎛25−1−1ab23−2⎠⎞⎝⎛11−1⎠⎞=⎝⎛11−1⎠⎞
有
{2−1−2=5+a−3=−1+b+2=−\begin{cases} 2-1-2=\lambda \\ 5+a-3=\lambda \\ -1+b+2=-\lambda \end{cases}⎩⎪⎪⎨⎪⎪⎧2−1−2=5+a−3=−1+b+2=−
可得=−1,a=−3,b=0\lambda=-1,a=-3,b=0=−1,a=−3,b=0
类似矩阵
设A,BA,BA,B都是nnn阶矩阵,如果存在可逆矩阵PPP使
P−1AP=BP^{-1}AP=BP−1AP=B
就称矩阵AAA类似与矩阵BBB,BBB是AAA的类似矩阵,记作A∼BA\sim BA∼B
类似的根本性质
-
A∼AA\sim AA∼A
-
如果A∼BA\sim BA∼B,则B∼AB\sim AB∼A
-
如果A∼B,B∼CA\sim B,B\sim CA∼B,B∼C,则A∼CA\sim CA∼C
如果A∼BA\sim BA∼B
-
∣E−A∣=∣E−B∣⇒A=B|\lambda E-A|=|\lambda E-B|\Rightarrow \lambda_{A}=\lambda_{B}∣E−A∣=∣E−B∣⇒A=B
证明:
∣E−B∣=∣E−P−1AP∣=∣P−1(E−A)P∣=∣P−1∣∣E−A∣∣P∣=∣E−A∣\begin{aligned} |\lambda E-B|&=|\lambda E-P^{-1}AP|\\&=|P^{-1}(\lambda E-A)P|\\&=|P^{-1}||\lambda E-A||P|\\&=|\lambda E-A| \end{aligned}∣E−B∣=∣E−P−1AP∣=∣P−1(E−A)P∣=∣P−1∣∣E−A∣∣P∣=∣E−A∣
-
r(A)=r(B)r(A)=r(B)r(A)=r(B)
证明:
如果AAA可逆,有r(AB)=r(B),r(BA)=r(B)r(AB)=r(B),r(BA)=r(B)r(AB)=r(B),r(BA)=r(B),则
r(B)=r(P−1AP)=r(AP)=r(A)\begin{aligned} r(B)=r(P^{-1}AP)=r(AP)=r(A) \end{aligned}r(B)=r(P−1AP)=r(AP)=r(A)
-
∣A∣=∣B∣|A|=|B|∣A∣=∣B∣
∣B∣=∣P−1AP∣=∣P−1∣∣A∣∣P∣=∣A∣|B|=|P^{-1}AP|=|P^{-1}||A||P|=|A|∣B∣=∣P−1AP∣=∣P−1∣∣A∣∣P∣=∣A∣
-
∑aii=∑bii\sum\limits a_{ii}=\sum\limits b_{ii}∑aii=∑bii
-
An∼BnA^{n}\sim B^{n}An∼Bn
A+kE∼B+kEA+kE\sim B+kEA+kE∼B+kE
A−1∼B−1A^{-1}\sim B^{-1}A−1∼B−1
类似对角化
如果A∼A\sim \LambdaA∼,则称矩阵AAA可类似对角化
定理:A∼⇔AA\sim \Lambda\Leftrightarrow AA∼⇔A有nnn个线性无关的特征向量
证明:
如果A1=11,A2=22,A3=33A \alpha_{1}=\lambda_{1}\alpha_{1},A \alpha_{2}=\lambda_{2}\alpha_{2},A \alpha_{3}=\lambda_{3}\alpha_{3}A1=11,A2=22,A3=33,且1,2,3\alpha_{1},\alpha_{2},\alpha_{3}1,2,3线性无关,则
A(1,2,3)=(A1,A2,A3)=(11,22,33)=(1,2,3)(123)\begin{aligned} A(\alpha_{1},\alpha_{2},\alpha_{3})&=(A \alpha_{1},A \alpha_{2},A \alpha_{3})\\ &=(\lambda_{1}\alpha_{1},\lambda_{2}\alpha_{2},\lambda_{3}\alpha_{3})\\ &=(\alpha_{1},\alpha_{2},\alpha_{3})\begin{pmatrix} \lambda_{1} & & \\ & \lambda_{2} & \\ & & \lambda_{3} \end{pmatrix} \end{aligned}A(1,2,3)=(A1,A2,A3)=(11,22,33)=(1,2,3)⎝⎛123⎠⎞
令P=(1,2,3)P=(\alpha_{1},\alpha_{2},\alpha_{3})P=(1,2,3),因为1,2,3\alpha_{1},\alpha_{2},\alpha_{3}1,2,3线性无关,则PPP可逆,有
AP=PP−1AP==(123)\begin{aligned} AP&=P \Lambda\\ P^{-1}AP&=\Lambda=\begin{pmatrix} \lambda_{1} & & \\ & \lambda_{2} & \\ & & \lambda_{3} \end{pmatrix} \end{aligned}APP−1AP=P==⎝⎛123⎠⎞
必要性得证
如果P−1AP=P^{-1}AP=\LambdaP−1AP=,有
AP=PA(1,2,3)=(1,2,3)(a1a2a3)=(a11,a22,a33)(A1,A2,A3)=(a11,a22,a33)\begin{aligned} AP&=P \Lambda\\ A(\alpha_{1},\alpha_{2},\alpha_{3})&=(\alpha_{1},\alpha_{2},\alpha_{3})\begin{pmatrix} a_{1} & & \\ & a_{2} & \\ & & a_{3} \end{pmatrix}\\ &=(a_{1}\alpha_{1},a_{2}\alpha_{2},a_{3}\alpha_{3})\\ (A \alpha_{1},A \alpha_{2},A \alpha_{3})&=(a_{1}\alpha_{1},a_{2}\alpha_{2},a_{3}\alpha_{3}) \end{aligned}APA(1,2,3)(A1,A2,A3)=P=(1,2,3)⎝⎛a1a2a3⎠⎞=(a11,a22,a33)=(a11,a22,a33)
因而有
A1=a11,A2=a22,A3=a33A \alpha_{1}=a_{1}\alpha_{1},A \alpha_{2}=a_{2}\alpha_{2},A \alpha_{3}=a_{3}\alpha_{3}A1=a11,A2=a22,A3=a33
又因为P=(1,2,3)P=(\alpha_{1},\alpha_{2},\alpha_{3})P=(1,2,3)可逆,则1,2,3\alpha_{1},\alpha_{2},\alpha_{3}1,2,3线性无关,充分性得证、
推论:如果AAA有nnn个不同的特征值,则A∼A\sim \LambdaA∼
定理:A∼⇔A\sim \Lambda\Leftrightarrow \lambdaA∼⇔是AAA的kkk重特征值,则\lambda有kkk个线性无关的特征向量
类似对角化解题步骤:
-
求特征值1,2,3\lambda_{1},\lambda_{2},\lambda_{3}1,2,3
-
求特征向量1,2,3\alpha_{1},\alpha_{2},\alpha_{3}1,2,3
-
结构可逆矩阵P=(1,2,3)P=(\alpha_{1},\alpha_{2},\alpha_{3})P=(1,2,3),则P−1AP=(123)P^{-1}AP=\begin{pmatrix}\lambda_{1} & & \\ & \lambda_{2} & \\ & & \lambda_{3}\end{pmatrix}P−1AP=⎝⎛123⎠⎞
例:已知A=(001010100)A=\begin{pmatrix}0 & 0 & 1 \\ 0 & 1 & 0 \\ 1 & 0 & 0\end{pmatrix}A=⎝⎛001010100⎠⎞,求可逆矩阵PPP,使P−1AP=P^{-1}AP=\LambdaP−1AP=
由特征多项式
∣E−A∣=∣0−10−10−10∣=(−1)2(+1)|\lambda E-A|=\begin{vmatrix} \lambda & 0 & -1 \\ 0 & \lambda-1 & 0 \\ -1 & 0 & \lambda \end{vmatrix}=(\lambda-1)^{2}(\lambda+1)∣E−A∣=∣∣∣∣∣∣∣0−10−10−10∣∣∣∣∣∣∣=(−1)2(+1)
则AAA的特征值1,1,−11,1,-11,1,−1
当=1\lambda=1=1时,由(E−A)x=0(E-A)x=0(E−A)x=0
(10−1000−101)→(10−1000000)\begin{pmatrix} 1 & 0 & -1 \\ 0 & 0 & 0 \\ -1 & 0 & 1 \end{pmatrix}\rightarrow\begin{pmatrix} 1 & 0 & -1 \\ 0 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}⎝⎛10−1000−101⎠⎞→⎝⎛100000−100⎠⎞
特征向量1=(0,1,0)T,2=(1,0,1)T\alpha_{1}=(0,1,0)^{T},\alpha_{2}=(1,0,1)^{T}1=(0,1,0)T,2=(1,0,1)T
当=−1\lambda=-1=−1时,由(−E−A)x=0(-E-A)x=0(−E−A)x=0
(−10−10−20−10−1)→(101010000)\begin{pmatrix} -1 & 0 & -1 \\ 0 & -2 & 0 \\ -1 & 0 & -1 \end{pmatrix}\rightarrow \begin{pmatrix} 1 & 0 & 1 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{pmatrix}⎝⎛−10−10−20−10−1⎠⎞→⎝⎛100010100⎠⎞
特征向量3=(−1,0,1)T\alpha_{3}=(-1,0,1)^{T}3=(−1,0,1)T
令
P=(1,2,3)=(01−1100011)P=(\alpha_{1},\alpha_{2},\alpha_{3})=\begin{pmatrix} 0 & 1 & -1 \\ 1 & 0 & 0 \\ 0 & 1 & 1 \end{pmatrix}P=(1,2,3)=⎝⎛010101−101⎠⎞
有
P−1AP=(11−1)P^{-1}AP=\begin{pmatrix} 1 & & \\ & 1 & \\ & & -1 \end{pmatrix}P−1AP=⎝⎛11−1⎠⎞
例:已知A=(20000101x),B=(2000y000−1)A=\begin{pmatrix}2 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & 1 & x\end{pmatrix},B=\begin{pmatrix}2 & 0 & 0 \\ 0 & y & 0 \\ 0 & 0 & -1\end{pmatrix}A=⎝⎛20000101x⎠⎞,B=⎝⎛2000y000−1⎠⎞类似
因为A∼BA\sim BA∼B,有
∑aii=∑bii⇒2+0+x=2+y+(−1)(1)∣A∣=∣B∣⇒−2=−2y(2)∣E−A∣=∣E−B∣⇒∣−2000−10−1−x∣=∣−2000−y000+1∣⇒(−2)(2−x−1)=(−2)[2+(1−y)−y]⇒{−x=1−y−1=−y(3)\begin{aligned} \sum\limits a_{ii}=\sum\limits b_{ii}&\Rightarrow 2+0+x=2+y+(-1)\quad\text{(1)}\\ |A|=|B|&\Rightarrow -2=-2y\quad\text{(2)}\\ |\lambda E-A|=|\lambda E-B|&\Rightarrow \begin{vmatrix} \lambda-2 & 0 & 0 \\ 0 & \lambda & -1 \\ 0 & -1 & \lambda-x \end{vmatrix}=\begin{vmatrix} \lambda-2 & 0 & 0 \\ 0 & \lambda-y & 0 \\ 0 & 0 & \lambda+1 \end{vmatrix}\tag{3}\\ &\Rightarrow (\lambda-2)(\lambda^{2}-x \lambda-1)=(\lambda-2)[\lambda^{2}+(1-y)\lambda-y]\\ &\Rightarrow \begin{cases} -x=1-y \\ -1=-y \end{cases} \end{aligned}∑aii=∑bii∣A∣=∣B∣∣E−A∣=∣E−B∣⇒2+0+x=2+y+(−1)(1)⇒−2=−2y(2)⇒∣∣∣∣∣∣∣−2000−10−1−x∣∣∣∣∣∣∣=∣∣∣∣∣∣∣−2000−y000+1∣∣∣∣∣∣∣⇒(−2)(2−x−1)=(−2)[2+(1−y)−y]⇒{−x=1−y−1=−y(3)
能够选(1),(2)(1),(2)(1),(2)或独自选(3)(3)(3),一般选(1),(2)(1),(2)(1),(2),计算较简单
或
BBB的特征值2,y,−12,y,-12,y,−1,则=−1\lambda=-1=−1是AAA的特征值,有
∣−E−A∣=∣−3000−1−10−1−1−x∣=−3∣−1−1−1−1−x∣=−3x=0|-E-A|= \begin{vmatrix} -3 & 0 & 0 \\ 0 & -1 & -1 \\0 & -1 & -1-x \end{vmatrix}=-3\begin{vmatrix} -1 & -1 \\ -1 & -1-x \end{vmatrix}=-3x=0∣−E−A∣=∣∣∣∣∣∣∣−3000−1−10−1−1−x∣∣∣∣∣∣∣=−3∣∣∣∣∣−1−1−1−1−x∣∣∣∣∣=−3x=0
若用=2\lambda=2=2,有
∣2E−A∣=∣00002−10−12−x∣=0|2E-A|=\begin{vmatrix} 0 & 0 & 0 \\ 0 & 2 & -1 \\ 0 & -1 & 2-x \end{vmatrix}=0∣2E−A∣=∣∣∣∣∣∣∣00002−10−12−x∣∣∣∣∣∣∣=0
若用=y\lambda=y=y,有
∣yE−A∣=∣y−2000y−10−1y−x∣=(y−2)(y2−xy−1)=0|yE-A|=\begin{vmatrix} y-2 & 0 & 0 \\ 0 & y & -1 \\ 0 & -1 & y-x \end{vmatrix}=(y-2)(y^{2}-xy-1)=0∣yE−A∣=∣∣∣∣∣∣∣y−2000y−10−1y−x∣∣∣∣∣∣∣=(y−2)(y2−xy−1)=0
该式需求结合上面的(1),(2),(3)(1),(2),(3)(1),(2),(3)运用,或许选择两个不相关的运用,例如=0,=y\lambda=0,\lambda=y=0,=y搭配
该办法当呈现本题=2\lambda=2=2时∣2E−A∣|2E-A|∣2E−A∣自身有一行就为000或两行成比例,则需求换\lambda或用(1),(2),(3)(1),(2),(3)(1),(2),(3)办法
解得x=0,y=1x=0,y=1x=0,y=1,代入题设有
A=(200001010)∼(21−1)=BA=\begin{pmatrix} 2 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & 1 & 0 \end{pmatrix}\sim \begin{pmatrix} 2 & & \\ & 1 & \\ & & -1 \end{pmatrix}=BA=⎝⎛200001010⎠⎞∼⎝⎛21−1⎠⎞=B
此处省略步骤,可得
P=(10001101−1)P=\begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 1 \\ 0 & 1 & -1 \end{pmatrix}P=⎝⎛10001101−1⎠⎞
实对称矩阵
向量的内积
定义:设=(a1,a2,⋯ ,an)T,=(b1,b2,⋯ ,bn)T\alpha=(a_{1},a_{2},\cdots,a_{n})^{T},\beta=(b_{1},b_{2},\cdots,b_{n})^{T}=(a1,a2,⋯,an)T,=(b1,b2,⋯,bn)T,有
(,)=a1b1+a2b2+⋯+anbn=T=T(\alpha,\beta)=a_{1}b_{1}+a_{2}b_{2}+\cdots+a_{n}b_{n}=\alpha^{T}\beta=\beta^{T}\alpha(,)=a1b1+a2b2+⋯+anbn=T=T
如果(,)=0(\alpha,\beta)=0(,)=0,称\alpha与\beta正交
(,)=a12+a22+⋯+an2(\alpha,\alpha)=a_{1}^{2}+a_{2}^{2}+\cdots+a_{n}^{2}(,)=a12+a22+⋯+an2,称a12+a22+⋯+an2\sqrt{a_{1}^{2}+a_{2}^{2}+\cdots+a_{n}^{2}}a12+a22+⋯+an2为向量\alpha的长度,记作∣∣∣∣||\alpha||∣∣∣∣
内积的性质
-
(,)=(,)(\alpha,\beta)=(\beta,\alpha)(,)=(,)
-
(k,)=(,k)=k(,)(k \alpha,\beta)=(\alpha,k \beta)=k(\lambda,\beta)(k,)=(,k)=k(,)
-
(+,)=(,)+(,)(\alpha+\beta,\gamma)=(\alpha,\gamma)+(\beta,\gamma)(+,)=(,)+(,)
-
(,)≥0(\alpha,\alpha)\geq0(,)≥0,等号当且仅当=0\alpha=0=0时建立
例:求与1=(−1,−1,1)T,2=(1,2,1)T\alpha_{1}=(-1,-1,1)^{T},\alpha_{2}=(1,2,1)^{T}1=(−1,−1,1)T,2=(1,2,1)T都正交的向量
设=(x1,x2,x3)T\alpha=(x_{1},x_{2},x_{3})^{T}=(x1,x2,x3)T与1,2\alpha_{1},\alpha_{2}1,2都正交,则T1=0,T2=0\alpha^{T}\alpha_{1}=0,\alpha^{T}\alpha_{2}=0T1=0,T2=0
{−x1−x2+x3=0x1−2×2−x3=0\begin{cases} -x_{1}-x_{2}+x_{3}=0 \\ x_{1}-2x_{2}-x_{3}=0 \end{cases}{−x1−x2+x3=0x1−2x2−x3=0
有
(−1−111−2−1)→(10−1010)\begin{pmatrix} -1 & -1 & 1 \\ 1 & -2 & -1 \end{pmatrix}\rightarrow \begin{pmatrix} 1 & 0 & -1 \\ 0 & 1 & 0 \end{pmatrix}(−11−1−21−1)→(1001−10)
根底解系(1,0,1)T(1,0,1)^{T}(1,0,1)T,因而=k(1,0,1)T\alpha=k(1,0,1)^{T}=k(1,0,1)T
正交矩阵
定义:AAA是nnn阶矩阵,若AAT=ATA=EAA^{T}=A^{T}A=EAAT=ATA=E,则称AAA是正交矩阵
如果AAA是正交矩阵
⇔AT=A−1\Leftrightarrow A^{T}=A^{-1}⇔AT=A−1
⇔A\Leftrightarrow A⇔A的列向量都是单位向量且两两正交
如果AAA是正交矩阵⇒∣A∣\Rightarrow |A|⇒∣A∣为111或−1-1−1
证明:
AAT=E⇒∣AAT∣=∣E∣∣A∣⋅∣AT∣=1∣A∣2=1\begin{aligned} AA^{T}=E\Rightarrow |AA^{T}|&=|E|\\ |A|\cdot|A^{T}|&=1\\ |A|^{2}&=1 \end{aligned}AAT=E⇒∣AAT∣∣A∣⋅∣AT∣∣A∣2=∣E∣=1=1
因而∣A∣|A|∣A∣为111或−1-1−1
实对称矩阵
定理:实对称矩阵必可类似对角化
定理:实对称矩阵不同特征值所对应的特征向量相互正交
定理:实对称矩阵必存在正交矩阵QQQ,使Q−1AQ=QTAQ=Q^{-1}AQ=Q^{T}AQ=\LambdaQ−1AQ=QTAQ=
例:AAA为333阶实对称矩阵,满意A2=AA^{2}=AA2=A,如果r(A−E)=2r(A-E)=2r(A−E)=2,则AAA的特征值()
设A=,≠0A \alpha =\lambda \alpha,\alpha \ne 0A=,=0,则
A2=2A ^{2}\alpha=\lambda ^{2}\alphaA2=2
由A2=AA^{2}=AA2=A,有
2=(2−)=0⇒2−=0\begin{aligned} \lambda ^{2}\alpha&=\lambda \alpha\\ (\lambda ^{2}-\lambda )\alpha&=0\\ &\Rightarrow \lambda ^{2}-\lambda =0 \end{aligned}2(2−)==0⇒2−=0
因而\lambda是111或000,又因为AAA是实对称矩阵,r(A−E)=2r(A-E)=2r(A−E)=2,有
A∼⇒A−E∼−E⇒r(−E)=2A \sim \Lambda \Rightarrow A-E \sim \Lambda-E \Rightarrow r(\Lambda-E)=2A∼⇒A−E∼−E⇒r(−E)=2
\Lambda与AAA类似,即特征值相同,又有r(−E)=2r(\Lambda-E)=2r(−E)=2,则可结构
=(100)\Lambda=\begin{pmatrix} 1 & & \\ & 0 & \\ & & 0 \end{pmatrix}=⎝⎛100⎠⎞
因而AAA的特征值为1,0,01,0,01,0,0
求Q−1AQ=QTAQ=Q^{-1}AQ=Q^{T}AQ=\LambdaQ−1AQ=QTAQ=
-
求出AAA的特征值1,2,3\lambda_{1},\lambda_{2},\lambda_{3}1,2,3
-
求出对应的特征向量1,2,3\alpha_{1},\alpha_{2},\alpha_{3}1,2,3
-
该特征向量为1,2,3\gamma _{1},\gamma _{2},\gamma _{3}1,2,3
1. 如果特征值不同,只需单位化(实对称矩阵不同特征值所对应的特征向量相互正交)
2. 若特征值有重根
1. 如果特征向量已正交,只需单位化
2. 如果特征向量不正交,需施密特(Schmidt)正交化
- 结构正交矩阵Q=(1,2,3),Q−1AQ==(123)Q=(\gamma_{1},\gamma_{2},\gamma_{3}),Q^{-1}AQ=\Lambda=\begin{pmatrix}\lambda_{1} & & \\ & \lambda_{2} & \\ & & \lambda_{3}\end{pmatrix}Q=(1,2,3),Q−1AQ==⎝⎛123⎠⎞
例:已知A=(1a−1a31−111)A=\begin{pmatrix}1 & a & -1 \\ a & 3 & 1 \\ -1 & 1 & 1\end{pmatrix}A=⎝⎛1a−1a31−111⎠⎞,若r(A)=2r(A)=2r(A)=2,求aaa,求正交矩阵QQQ和对角矩阵\Lambda,使Q−1AQ=Q^{-1}AQ=\LambdaQ−1AQ=
求aaa能够用行改换,这里用秩的定义,即队伍式不为000。因为AAA中有二阶子式
∣3111∣≠0\begin{vmatrix} 3 & 1 \\ 1 & 1 \end{vmatrix}\ne 0∣∣∣∣∣3111∣∣∣∣∣=0
所以r(A)=0⇔∣A∣=0r(A)=0 \Leftrightarrow |A|=0r(A)=0⇔∣A∣=0
∣1a−1a31−111∣=∣0a−1a+131011∣=−(a+1)2\begin{vmatrix} 1 & a & -1 \\ a & 3 & 1 \\ -1 & 1 & 1 \end{vmatrix}=\begin{vmatrix} 0 & a & -1 \\ a+1 & 3 & 1 \\ 0 & 1 & 1 \end{vmatrix}=-(a+1)^{2}∣∣∣∣∣∣∣1a−1a31−111∣∣∣∣∣∣∣=∣∣∣∣∣∣∣0a+10a31−111∣∣∣∣∣∣∣=−(a+1)2
因而a=−1a=-1a=−1,有
A=(1−1−1−131−111)A=\begin{pmatrix}1 & -1 & -1 \\ -1 & 3 & 1 \\ -1 & 1 & 1\end{pmatrix}A=⎝⎛1−1−1−131−111⎠⎞
由AAA的特征多项式
∣E−A∣=∣−1111−3−11−1−1∣找除主对角线外对应方位两数字和为0=∣01−3−11−1−1∣=(−1)(−4)\begin{aligned} |\lambda E-A|&=\begin{vmatrix} \lambda-1 & 1 & 1 \\ 1 & \lambda-3 & -1 \\ 1 & -1 & \lambda-1 \end{vmatrix}\\ &找除主对角线外对应方位两数字和为0\\ &=\begin{vmatrix} \lambda & 0 & \lambda \\ 1 & \lambda-3 & -1 \\ 1 & -1 & \lambda-1 \end{vmatrix}=\lambda(\lambda-1)(\lambda-4) \end{aligned}∣E−A∣=∣∣∣∣∣∣∣−1111−3−11−1−1∣∣∣∣∣∣∣找除主对角线外对应位置两数字和为0=∣∣∣∣∣∣∣110−3−1−1−1∣∣∣∣∣∣∣=(−1)(−4)
因而AAA的特征值为0,1,40,1,40,1,4
对=1\lambda=1=1
(E−A)=(0111−2−11−10)→(101011000)⇒1=(−1,−1,1)T(E-A)=\begin{pmatrix} 0 & 1 & 1 \\ 1 & -2 & -1 \\ 1 & -1 & 0 \end{pmatrix}\rightarrow \begin{pmatrix} 1 & 0 & 1 \\ 0 & 1 & 1 \\ 0 & 0 & 0 \end{pmatrix}\Rightarrow \alpha_{1}=(-1,-1,1)^{T}(E−A)=⎝⎛0111−2−11−10⎠⎞→⎝⎛100010110⎠⎞⇒1=(−1,−1,1)T
对=4\lambda=4=4
(4E−)=(31111−11−13)→(10101−2000)⇒2=(−1,2,1)T(4E-\lambda)=\begin{pmatrix} 3 & 1 & 1 \\ 1 & 1 & -1 \\ 1 & -1 & 3 \end{pmatrix}\rightarrow \begin{pmatrix} 1 & 0 & 1 \\ 0 & 1 & -2 \\ 0 & 0 & 0 \end{pmatrix}\Rightarrow \alpha_{2}=(-1,2,1)^{T}(4E−)=⎝⎛31111−11−13⎠⎞→⎝⎛1000101−20⎠⎞⇒2=(−1,2,1)T
对=0\lambda=0=0
(0E−A)=(−1111−3−11−1−1)→(10−1010000)⇒3=(1,0,1)T(0E-A)=\begin{pmatrix} -1 & 1 & 1 \\ 1 & -3 & -1 \\ 1 & -1 & -1 \end{pmatrix}\rightarrow \begin{pmatrix} 1 & 0 & -1 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{pmatrix}\Rightarrow \alpha_{3}=(1,0,1)^{T}(0E−A)=⎝⎛−1111−3−11−1−1⎠⎞→⎝⎛100010−100⎠⎞⇒3=(1,0,1)T
实对称矩阵,特征值不同,特征向量已经正交,只需单位化
1=13(−1−11),2=16(−121),3=12(101)\gamma _{1}=\frac{1}{\sqrt{3}}\begin{pmatrix}-1 \\ -1 \\ 1\end{pmatrix},\gamma _{2}=\frac{1}{\sqrt{6}}\begin{pmatrix} -1 \\ 2 \\ 1 \end{pmatrix},\gamma _{3}=\frac{1}{\sqrt{2}}\begin{pmatrix} 1 \\ 0 \\ 1 \end{pmatrix}1=31⎝⎛−1−11⎠⎞,2=61⎝⎛−121⎠⎞,3=21⎝⎛101⎠⎞
令
Q=(1,2,3)=(−13−1612−13260131612)Q=(\gamma_{1},\gamma_{2},\gamma_{3})=\begin{pmatrix} -\frac{1}{\sqrt{3}} & -\frac{1}{\sqrt{6}} & \frac{1}{\sqrt{2}} \\ – \frac{1}{\sqrt{3}} & \frac{2}{\sqrt{6}} & 0 \\ \frac{1}{\sqrt{3}} & \frac{1}{\sqrt{6}} & \frac{1}{\sqrt{2}} \end{pmatrix}Q=(1,2,3)=⎝⎜⎜⎛−31−3131−61626121021⎠⎟⎟⎞
则
Q−1AQ==(140)Q^{-1}AQ=\Lambda=\begin{pmatrix} 1 & & \\ & 4 & \\ & & 0 \end{pmatrix}Q−1AQ==⎝⎛140⎠⎞