本文目录
前言
作者进行矩阵分解章节复习总结的几种计算。
矩阵分解主要有LDU分解、Cholesky分解、QR分解、满秩分解、奇异值分解。
本文不讲理论知识，只是理清计算过程。
提示：以下是本篇文章正文内容，下面例题可供参考
一、LDU分解
三角分解，其中还有上三角和下三角形式。
求
A
=
[
2
−
1
3
1
2
1
2
4
2
]
A= \left[\begin{array}{lll} 2 & -1 & 3 \\ 1 & 2 & 1 \\ 2 & 4 & 2 \end{array}\right]
A=⎣⎡212−124312⎦⎤的LDU分解
解答：
(1)根据A构造第1列的单位下三角矩阵
L 1 = [ 1 1 2 1 1 0 1 ] {L_1} = \left[\begin{array}{lll} 1 \\ {\boldsymbol{\frac{1}{2}}}& 1 \\ 1 & 0 & 1 \end{array}\right] L1=⎣⎡1211101⎦⎤ L 1 − 1 = [ 1 − 1 2 1 − 1 0 1 ] {L_1}^{-1} = \left[\begin{array}{lll} 1 \\ {\boldsymbol{-{\frac{1}{2}}}}& 1 \\ -1 & 0 & 1 \end{array}\right] L1−1=⎣⎡1−21−1101⎦⎤
问：怎么得来的
1
2
和
1
？
\frac{1}{2}和1？
21和1？
a
21
a
11
=
1
2
a
31
a
a
11
=
2
2
=
1
\frac{a_{21}}{a_{11}} = \frac{1}{2}\\\frac{a_{31}}{a_{a11}} = \frac{2}{2} = 1
a11a21=21aa11a31=22=1
L 1 − 1 A = [ 2 − 1 3 0 5 2 − 1 2 0 5 − 1 ] = A ( 1 ) {L_1}^{-1}A = \left[\begin{array}{lll} 2 & -1 & 3 \\ 0& {\boldsymbol{\frac{5}{2}}}& {\boldsymbol{-{\frac{1}{2}}}} \\ 0 & 5 & -1 \end{array}\right] = A^{(1)} L1−1A=⎣⎡200−12553−21−1⎦⎤=A(1)
(2)根据 A ( 1 ) A^{(1)} A(1)第2列构造单位下三角矩阵
L 2 = [ 1 0 1 0 2 1 ] {L_2} = \left[\begin{array}{lll} 1 \\ 0 & 1 \\ 0 & 2 & 1 \end{array}\right] L2=⎣⎡100121⎦⎤ L 2 − 1 = [ 1 0 1 0 − 2 1 ] {L_2}^{-1} = \left[\begin{array}{lll} 1 \\ 0 & 1 \\ 0 & -2 & 1 \end{array}\right] L2−1=⎣⎡1001−21⎦⎤
L 2 − 1 A ( 1 ) = [ 2 − 1 3 0 5 2 − 1 2 0 0 0 ] = A ( 2 ) {L_2}^{-1}A^{(1)}= \left[\begin{array}{lll} 2 & -1 & 3 \\ 0& {\boldsymbol{\frac{5}{2}}}& {\boldsymbol{-{\frac{1}{2}}}} \\ 0 & 0 & 0 \end{array}\right] = A^{(2)} L2−1A(1)=⎣⎡200−12503−210⎦⎤=A(2)
此时可以看到矩阵为上三角形式，因此可以分解成如下形式，得出对角矩阵和单位上三角矩阵
A ( 2 ) = [ 2 5 2 0 ] [ 1 − 1 2 3 2 1 − 1 5 1 ] = D U A^{(2)} = { \left[\begin{array}{lll} 2 \\ &{\boldsymbol{\frac{5}{2}}} \\ & & 0 \end{array}\right] } { \left[\begin{array}{lll} 1&{\boldsymbol{-\frac{1}{2}}} &{\boldsymbol{\frac{3}{2}}} \\ & 1 &{\boldsymbol{-\frac{1}{5}}} \\ & & 1 \end{array}\right] }= DU A(2)=⎣⎡2250⎦⎤⎣⎡1−21123−511⎦⎤=DU
L = L 1 L 2 = [ 1 1 2 1 1 2 1 ] L = {L_1}{L_2}= \left[\begin{array}{lll} 1 \\ {\boldsymbol{\frac{1}{2}}} & 1 \\ 1 & 2 & 1 \end{array}\right] L=L1L2=⎣⎡1211121⎦⎤
因此A是经过L变换得到的DU形式
A = L D U A = LDU A=LDU
A = [ 1 1 2 1 1 2 1 ] [ 2 5 2 0 ] [ 1 − 1 2 3 2 1 − 1 5 1 ] A= \left[\begin{array}{lll} 1 \\ {\boldsymbol{\frac{1}{2}}} & 1 \\ 1 & 2 & 1 \end{array}\right] \left[\begin{array}{lll} 2 \\ &{\boldsymbol{\frac{5}{2}}} \\ & & 0 \end{array}\right] \left[\begin{array}{lll} 1&{\boldsymbol{-\frac{1}{2}}} &{\boldsymbol{\frac{3}{2}}} \\ & 1 &{\boldsymbol{-\frac{1}{5}}} \\ & & 1 \end{array}\right] A=⎣⎡1211121⎦⎤⎣⎡2250⎦⎤⎣⎡1−21123−511⎦⎤
二、Cholesky分解
求 A = [ 2 − 1 3 1 2 1 2 4 2 ] 的 C h o l e s k y 分 解 求A= \left[\begin{array}{lll} 2 & -1 & 3 \\ 1 & 2 & 1 \\ 2 & 4 & 2 \end{array}\right]的Cholesky分解 求A=⎣⎡212−124312⎦⎤的Cholesky分解
解答：过程与LDU一致，只是多了一个步骤
A = L ( D D ) U = L D ( L D ) T = G G T A = L{(\sqrt{D}\sqrt{D})}U = L\sqrt{D}(L\sqrt{D})^T = GG^T A=L(DD)U=LD(LD)T=GGT
三、QR分解
最熟悉的就是正交三角分解方法，但是某些时候运算比较复杂，所以了解一下后两种方法很有必要。
1. 正交三角分解
A = [ 1 2 1 2 1 2 1 2 1 ] ， Q R 分 解 A= \left[\begin{array}{lll} 1 & 2 & 1 \\ 2 & 1 & 2 \\ 1 & 2 & 1 \end{array}\right]，QR分解 A=⎣⎡121212121⎦⎤，QR分解
解答：
(1)取A的各列
a 1 = ( 1 , 2 , 1 ) T , a 2 = ( 2 , 1 , 2 ) T , a 3 = ( 2 , 2 , 1 ) T a_1 = (1,2,1)^T, a_2 =(2,1,2)^T, a_3=(2,2,1)^T a1=(1,2,1)T,a2=(2,1,2)T,a3=(2,2,1)T
(2)Schmide正交化
b 1 = a 1 = ( 1 , 2 , 1 ) T , ∣ b 1 ∣ = 6 b_1=a_1=(1,2,1)^T,|b_1|=\sqrt{6} b1=a1=(1,2,1)T,∣b1∣=6
b 2 = a 2 − k 21 b 1 = a 2 − b 1 ( 1 , − 1 , 1 ) T ∣ b 2 ∣ = 3 , k 21 = ( a 2 , b 1 ) ( b 1 , b 1 ) = 1 b_2=a_2-k_{21}b_1=a_2-b_1 (1,-1,1)^T \\ |b_2| =\sqrt{3},k_{21}=\frac{(a2,b1)}{(b1,b1)}=1 b2=a2−k21b1=a2−b1(1,−1,1)T∣b2∣=3,k21=(b1,b1)(a2,b1)=1
b 2 = a 2 − k 21 b 1 = a 2 − b 1 = ( 1 , − 1 , 1 ) T ∣ b 2 ∣ = 3 , k 21 = ( a 2 , b 1 ) ( b 1 , b 1 ) = 1 b_2=a_2-k_{21}b_1 =a_2-b_1=(1,-1,1)^T\\ |b_2| =\sqrt{3},k_{21} =\frac{(a2,b1)}{(b1,b1)}=1 b2=a2−k21b1=a2−b1=(1,−1,1)T∣b2∣=3,k21=(b1,b1)(a2,b1)=1
(3)正交矩阵
Q = [ 1 1 1 2 2 − 1 0 1 1 − 1 2 ] Q = \left[\begin{array}{lll} 1 & 1 & \frac{1}{2} \\ 2 & -1 & 0 \\ 1 & 1 & -\frac{1}{2} \end{array}\right] Q=⎣⎡1211−11210−21⎦⎤
C = [ 1 k 21 k 31 1 k 32 1 ] = [ 1 1 7 6 1 1 3 1 ] C = \left[\begin{array}{lll} 1 & k_{21} & k_{31} \\ & 1 & k_{32} \\ & & 1 \end{array}\right]= \left[\begin{array}{lll} 1 & 1 & \frac{7}{6} \\ & 1 & \frac{1}{3} \\ & & 1 \end{array}\right] C=⎣⎡1k211k31k321⎦⎤=⎣⎡11167311⎦⎤
R = d i a g ( ∣ b 1 ∣ , ∣ b 2 ∣ , ∣ b 3 ∣ ) C R = diag(|b_1|,|b_2|,|b_3|)C R=diag(∣b1∣,∣b2∣,∣b3∣)C
R = [ 6 3 1 2 ] [ 1 1 7 6 1 1 3 1 ] R= \left[\begin{array}{lll} \sqrt{6} \\ & \sqrt{3} \\ & & \frac{1}{\sqrt{2}} \end{array}\right] \left[\begin{array}{lll} 1 & 1 & \frac{7}{6} \\ & 1 & \frac{1}{3} \\ & & 1 \end{array}\right] R=⎣⎡6321⎦⎤⎣⎡11167311⎦⎤
∴ A = Q R \therefore A=QR ∴A=QR
2. Givens变换方法（初等旋转）
A
=
[
0
1
1
1
1
0
1
0
1
]
，
Q
R
分
解
A= \left[\begin{array}{lll} 0 & 1 & 1 \\ 1 & 1 & 0 \\ 1 & 0 & 1 \end{array}\right]，QR分解
A=⎣⎡011110101⎦⎤，QR分解
解答：
(1) 对 A 的 第 1 列 a 1 = ( 0 , 1 , 1 ) T ， 构 造 T 12 对A的第1列a_1=(0,1,1)^T，构造T_{12} 对A的第1列a1=(0,1,1)T，构造T12
T 1 a 1 = ∣ a 1 ∣ e 1 = 2 e 1 T_1a_1=|a_1|e_1=\sqrt{2}e_1 T1a1=∣a1∣e1=2e1 T 12 ( c , s ) , c = 0 , s = 1 T_{12}(c,s),c=0,s=1 T12(c,s),c=0,s=1 T 12 = [ 0 1 0 − 1 0 0 0 0 1 ] , T 12 a 1 = [ 1 0 1 ] T_{12}= \left[\begin{array}{lll} 0 & 1 & 0 \\ -1 & 0 & 0 \\ 0 & 0 & 1 \end{array}\right],T_{12}a_1= \left[\begin{array}{l} 1 \\ 0 \\ 1 \end{array}\right] T12=⎣⎡0−10100001⎦⎤,T12a1=⎣⎡101⎦⎤
(2) 对 T 1 a 1 构 造 T 13 对T_1a_1构造T_{13} 对T1a1构造T13
T 13 ( c , s ) , c = 1 2 , s = 1 2 T_{13}(c,s),c=\frac{1}{\sqrt{2}},s=\frac{1}{\sqrt{2}} T13(c,s),c=21,s=21
T 13 = [ 1 2 0 1 2 0 1 0 − 1 2 0 1 2 ] T_{13}= \left[\begin{array}{lll} \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \\ 0 & 1 & 0 \\ -\frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \end{array}\right] T13=⎣⎡210−2101021021⎦⎤
得到变换矩阵 T 1 T_1 T1
T 1 = T 13 T 12 = [ 0 1 2 1 2 − 1 0 0 0 − 1 2 1 2 ] T_1=T_{13}T_{12}= \left[\begin{array}{lll} 0 & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \\ -1 & 0 & 0 \\ 0& -\frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \end{array}\right] T1=T13T12=⎣⎡0−10210−2121021⎦⎤
A 经 过 T 1 A经过T_1 A经过T1变换
T 1 A = [ 2 1 2 1 2 0 − 1 − 1 0 − 1 2 1 2 ] T_{1} A=\left[\begin{array}{ccc} \sqrt{2} & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \\ 0 & -1 & -1 \\ 0 & -\frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \end{array}\right] T1A=⎣⎢⎡20021−1−2121−121⎦⎥⎤
(3)得到新矩阵，继续构造变换矩阵
令 A ( 1 ) = [ − 1 − 1 − 1 2 1 2 ] ， b 1 = ( − 1 , − 1 2 ) T 构 造 T 2 令A^{(1)}=\left[\begin{array}{cc} -1 & -1 \\ -\frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \end{array}\right] ，b_{1}=\left(-1,-\frac{1}{\sqrt{2}}\right)^{T}构造T_2 令A(1)=[−1−21−121]，b1=(−1,−21)T构造T2
T 12 = [ − 2 3 − 1 3 1 3 − 2 3 ] T 12 b 1 = [ 3 2 0 ] T_{12}=\left[\begin{array}{cc} -\frac{\sqrt{2}}{3} & -\frac{1}{\sqrt{3}} \\ \frac{1}{\sqrt{3}} & -\sqrt{\frac{2}{3}} \end{array}\right] \quad T_{12} b_{1}=\left[\begin{array}{c} \frac{\sqrt{3}}{2} \\ 0 \end{array}\right] T12=⎣⎡−3231−31−32⎦⎤T12b1=[230]
得 到 变 换 矩 阵 T 2 得到变换矩阵T_2 得到变换矩阵T2
T 2 = T 12 = [ − 2 3 − 1 3 1 3 − 2 3 ] \begin{aligned} &T_{2}=T_{12}=\left[\begin{array}{cc} -\frac{\sqrt{2}}{3} & -\frac{1}{\sqrt{3}} \\ \frac{1}{\sqrt{3}} & -\frac{\sqrt{2}}{3} \end{array}\right]\end{aligned} T2=T12=[−3231−31−32]
A ( 1 ) 经 由 T 2 变 换 A_{(1)}经由T_2变换 A(1)经由T2变换
T 2 A ( 1 ) = [ − 2 3 1 6 0 − 2 3 ] \begin{aligned} T_{2} A^{(1)}=\left[\begin{array}{cc} -\sqrt{\frac{2}{3}} & \frac{1}{\sqrt{6}} \\ 0 & -\frac{2}{\sqrt{3}} \end{array}\right] \\ \end{aligned} T2A(1)=[−32061−32]
(3)已经得到了上三角矩阵，得到了最后的变换矩阵T，写出分解结果
T = [ 1 T 2 ] T 1 T=\left[\begin{array}{l} 1 \\ T_{2} \end{array}\right] T_{1} \\ T=[1T2]T1
Q = T T , R = T A Q=T^{T} ,R=T A Q=TT,R=TA
A = Q R A=Q R A=QR
3. Householder变换方法（初等反射）
A = [ 3 14 9 6 43 3 6 22 15 ] ， Q R 分 解 A=\left[\begin{array}{ccc} 3 & 14 & 9 \\ 6 & 43 & 3 \\ 6 & 22 & 15 \end{array}\right]，QR分解 A=⎣⎡3661443229315⎦⎤，QR分解
解答：
（1）对A的第1列 a 1 = ( 3 , 6 , 6 ) T a_1=(3,6,6)^{T} a1=(3,6,6)T构造H变换
a 1 − ∣ a 1 ∣ e 1 = 6 [ − 1 1 1 ] μ = a 1 − ∣ a 1 ∣ e 1 ∣ a 1 − ∣ a 1 ∣ e 1 ∣ = 1 3 [ − 1 1 1 ] H 1 = 1 − 2 μ μ ⊤ = 1 3 [ 1 2 2 2 1 − 2 2 − 2 1 ] H 1 A = [ 9 48 15 0 9 − 3 0 − 12 9 ] \begin{aligned} &a_{1}-\left|a_{1}\right| e_{1}=6\left[\begin{array}{c} -1 \\ 1 \\ 1 \end{array}\right]\\ &\mu=\frac{a_{1}-\left|a_{1}\right| e_{1}}{\left|a_{1}-\right| a_{1}\left|e_{1}\right|}=\frac{1}{\sqrt{3}}\left[\begin{array}{c} -1 \\ 1 \\ 1 \end{array}\right]\\ &H_{1}=1-2 \mu \mu^{\top}=\frac{1}{3}\left[\begin{array}{ccc} 1 & 2 & 2 \\ 2 & 1 & -2 \\ 2 & -2 & 1 \end{array}\right]\\ &H_{1} A=\left[\begin{array}{ccc} 9 & 48 & 15 \\ 0 & 9 & -3 \\ 0 & -12 & 9 \end{array}\right] \end{aligned} a1−∣a1∣e1=6⎣⎡−111⎦⎤μ=∣a1−∣a1∣e1∣a1−∣a1∣e1=31⎣⎡−111⎦⎤H1=1−2μμ⊤=31⎣⎡12221−22−21⎦⎤H1A=⎣⎡900489−1215−39⎦⎤
由 H 1 变 换 得 到 A ( 1 ) 由H_1变换得到A^{(1)} 由H1变换得到A(1)
A ( 1 ) = [ 9 − 3 − 12 9 ] A^{(1)}=\left[\begin{array}{cc} 9 & -3 \\ -12 & 9 \end{array}\right] A(1)=[9−12−39]
（2）对 A ( 1 ) A^{(1)} A(1)的第1列 b 1 = [ 9 − 12 ] b_{1}=\left[\begin{array}{c}9 \\-12\end{array}\right] b1=[9−12]构造H变换矩阵
b 1 − ∣ b 1 ∣ e 1 = 6 [ − 1 − 2 ] μ = b 1 − ∣ b 1 ∣ e 1 ∣ b 1 − ∣ b 1 ∣ e 1 ∣ = 1 5 [ − 1 − 2 ] H 2 = 1 − 2 μ μ ⊤ = 1 5 [ 3 − 4 − 4 − 3 ] H 2 A ( 1 ) = [ 15 − 9 0 − 3 ] \begin{aligned} &b_{1}-\left|b_{1}\right| e_{1}=6\left[\begin{array}{c} -1 \\ -2 \end{array}\right] \\ &\mu=\frac{b_{1}-\left|b_{1}\right| e_{1}}{\left|b_{1}-\right| b_{1}\left|e_{1}\right|}=\frac{1}{\sqrt{5}}\left[\begin{array}{l} -1 \\ -2 \end{array}\right] \\ &H_{2}= 1 -2 \mu \mu^{\top}=\frac{1}{5}\left[\begin{array}{cc} 3 & -4 \\ -4 & -3 \end{array}\right] \\ &H_{2} A^{(1)}=\left[\begin{array}{ll} 15 & -9 \\ 0 & -3 \end{array}\right] \end{aligned} b1−∣b1∣e1=6[−1−2]μ=∣b1−∣b1∣e1∣b1−∣b1∣e1=51[−1−2]H2=1−2μμ⊤=51[3−4−4−3]H2A(1)=[150−9−3]
(3)写出A的H变换矩阵S，得出分解结果
S = [ 1 H 2 ] H 1 = 1 15 [ 5 10 10 − 1 11 − 10 − 14 2 5 ] R = S A = [ 9 48 15 15 − 9 15 − 3 ] Q = S T = 1 15 [ 5 − 2 − 14 10 11 2 10 − 10 5 ] \begin{aligned} &S=\left[\begin{array}{ll} 1 & \\ & H_{2} \end{array}\right] H_{1}=\frac{1}{15}\left[\begin{array}{ccc} 5 & 10 & 10 \\ -1 & 11 & -10 \\ -14 & 2 & 5 \end{array}\right] \\ &R=S A=\left[\begin{array}{ccc} 9 & 48 & 15 \\ 15 & -9 \\ 15 & -3 \end{array}\right] \\ &Q=S^{T}=\frac{1}{15}\left[\begin{array}{ccc} 5 & -2 & -14 \\ 10 & 11 & 2 \\ 10 & -10 & 5 \end{array}\right]\end{aligned} S=[1H2]H1=151⎣⎡5−1−141011210−105⎦⎤R=SA=⎣⎡9151548−9−315⎦⎤Q=ST=151⎣⎡51010−211−10−1425⎦⎤
∴ A = Q R \begin{aligned}\therefore A=Q R \end{aligned} ∴A=QR
四、 满秩分解
两种方法：初等行变换和Hermite标准形
A
=
[
−
1
0
1
2
1
2
−
1
1
2
2
−
2
−
1
]
，
对
矩
阵
A
满
秩
分
解
F
G
A=\left[\begin{array}{cccc} -1 & 0 & 1 & 2 \\ 1 & 2 & -1 & 1 \\ 2 & 2 & -2 & -1 \end{array}\right]，对矩阵A满秩分解FG
A=⎣⎡−1120221−1−221−1⎦⎤，对矩阵A满秩分解FG
方法（1）初等行变换
[ A ∣ I ] = [ − 1 0 1 2 ⋮ 1 0 0 1 2 − 1 1 ⋮ 0 1 0 2 2 − 2 − 1 ⋮ 0 0 1 ] ⟶ 行 变 换 [ − 1 0 1 2 ⋮ 1 0 0 0 2 0 3 ⋮ 0 1 0 0 0 0 0 ⋮ 1 − 1 1 ] \begin{gathered} \left.[A|{I}\right]=\left[\begin{array}{ccccccc} -1 & 0 & 1 & 2 & \vdots & 1 & 0 & 0 \\ 1 & 2 & -1 & 1 & \vdots & 0 & 1 & 0 \\ 2 & 2 & -2 & -1& \vdots & 0 & 0 & 1 \end{array}\right] \\ \stackrel{行变换}{\longrightarrow}\left[\begin{array}{ccccccc} -1 & 0 & 1 & 2 & \vdots & 1 & 0 & 0 \\ 0 & 2 & 0 & 3 & \vdots &0 & 1 & 0 \\ 0 & 0 & 0 & 0 & \vdots &1 & -1 & 1 \end{array}\right] \end{gathered} [A∣I]=⎣⎢⎢⎢⎡−1120221−1−221−1⋮⋮⋮100010001⎦⎥⎥⎥⎤⟶行变换⎣⎢⎢⎢⎡−100020100230⋮⋮⋮10101−1001⎦⎥⎥⎥⎤
左边是一个熟悉的行最简，前两行就是G。右边为矩阵P，算出P的逆矩阵，前两列就是F
P = [ 1 0 0 0 1 0 1 − 1 1 ] ， P − 1 = [ 1 0 0 − 1 1 0 − 2 1 1 ] P=\left[\begin{array}{ccc} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 1 & -1 & 1 \end{array}\right] ，P^{-1}=\left[\begin{array}{ccc} 1 & 0 & 0 \\ -1 & 1 & 0 \\ -2 & 1 & 1 \end{array}\right] P=⎣⎡10101−1001⎦⎤，P−1=⎣⎡1−1−2011001⎦⎤
F = [ − 1 0 − 1 1 − 2 1 ] , G = [ − 1 0 1 2 0 2 0 3 ] F=\left[\begin{array}{cc} -1 & 0 \\ -1 & 1 \\ -2 & 1 \end{array}\right], G=\left[\begin{array}{cccc} -1 & 0 & 1 & 2 \\ 0 & 2 & 0 & 3 \end{array}\right] F=⎣⎡−1−1−2011⎦⎤,G=[−10021023]
相乘得到结果
∴ A = F G = [ − 1 0 − 1 1 − 2 1 ] [ − 1 0 1 2 0 2 0 3 ] \therefore A=FG=\left[\begin{array}{cc} -1 & 0 \\ -1 & 1 \\ -2 & 1 \end{array}\right]\left[\begin{array}{cccc} -1 & 0 & 1 & 2 \\ 0 & 2 & 0 & 3 \end{array}\right] ∴A=FG=⎣⎡−1−1−2011⎦⎤[−10021023]
方法（2）Hermite标准形：
通过行变换得到标准形
A = [ 0 0 1 2 1 1 2 j j 0 ] ( j = − 1 ) → 行变换 [ 1 1 2 0 0 0 1 0 0 0 ] = G (Hermite) \begin{aligned} &A=\left[\begin{array}{lll} 0 & 0 & 1 \\ 2 & 1 & 1 \\ 2 j & j & 0 \end{array}\right](j=\sqrt{-1}) &\stackrel{\text { 行变换 } }{\rightarrow}\left[\begin{array}{lll} 1 & \frac{1}{2} & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{array}\right]=G \text { (Hermite) } \end{aligned} A=⎣⎡022j01j110⎦⎤(j=−1)→ 行变换 ⎣⎡1002100010⎦⎤=G (Hermite)
∴ c 1 = 1 , c 2 = 3 , F 取 第 1 和 第 3 列 \begin{aligned}&\therefore {c_1=1 , c_2=3},F取第1和第3列\end{aligned} ∴c1=1,c2=3,F取第1和第3列
F = [ 0 1 2 1 2 j 0 ] \begin{aligned}F=\left[\begin{array}{cc} 0 & 1 \\ 2 & 1 \\ 2 j & 0\end{array}\right]\end{aligned} F=⎣⎡022j110⎦⎤
∴ A = F G = [ 0 1 2 1 2 j 0 ] [ 1 1 2 0 0 0 1 ] \begin{aligned}&\therefore A=FG=\left[\begin{array}{cc} 0 & 1 \\ 2 & 1 \\ 2 j & 0 \end{array}\right] \left[\begin{array}{lll} 1 & \frac{1}{2} & 0 \\ 0 & 0 & 1 \end{array}\right] \end{aligned} ∴A=FG=⎣⎡022j110⎦⎤[1021001]
五、奇异值分解
奇
异
值
分
解
即
A
=
U
[
∑
0
0
0
]
V
H
奇异值分解即A=U\left[\begin{array}{ll} \sum & 0 \\ 0 & 0 \end{array}\right] V^{H}
奇异值分解即A=U[∑000]VH
A
=
[
1
0
1
0
1
1
0
0
0
]
，
对
矩
阵
A
进
行
奇
异
值
分
解
。
A=\left[\begin{array}{lll} 1 & 0 & 1 \\ 0 & 1 & 1 \\ 0 & 0 & 0 \end{array}\right]，对矩阵A进行奇异值分解。
A=⎣⎡100010110⎦⎤，对矩阵A进行奇异值分解。
解答：
（1）首先求
A
T
A
A^TA
ATA特征值和特征向量
B = A T A = [ 1 0 1 0 1 1 1 1 2 ] ⇒ λ 1 = 3 λ 2 = 1 λ 3 = 0 B=A^{T} A=\left[\begin{array}{lll} 1 & 0 & 1 \\ 0 & 1 & 1 \\ 1 & 1 & 2 \end{array}\right] \Rightarrow \begin{aligned} &\lambda_{1}=3 \\ &\lambda_{2}=1 \atop \lambda_{3}=0 \end{aligned} B=ATA=⎣⎡101011112⎦⎤⇒λ1=3λ3=0λ2=1
特 征 值 向 量 为 特征值向量为 特征值向量为
ξ 1 = [ 1 1 2 ] ξ 2 = [ 1 − 1 0 ] ξ 3 = [ 1 1 − 1 ] \xi_{1}=\left[\begin{array}{l} 1 \\ 1 \\ 2 \end{array}\right] \xi_{2}=\left[\begin{array}{c} 1 \\ -1 \\ 0 \end{array}\right] \xi_{3}=\left[\begin{array}{c} 1 \\ 1 \\ -1 \end{array}\right] ξ1=⎣⎡112⎦⎤ξ2=⎣⎡1−10⎦⎤ξ3=⎣⎡11−1⎦⎤
rank A = 2 , Σ = [ 3 0 0 1 ] \operatorname{rank} A=2, \Sigma=\left[\begin{array}{cc} \sqrt{3} & 0 \\ 0 & 1 \end{array}\right] rankA=2,Σ=[3001]
（2）得到正交矩阵
∴ V = [ 1 6 1 2 1 3 1 6 − 1 2 1 3 2 6 0 1 3 ] （ 正 交 矩 阵 ) \therefore V=\left[\begin{array}{ccc} \frac{1}{\sqrt{6}} & \frac{1}{\sqrt{2}} & \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}} \end{array}\right]（正交矩阵) ∴V=⎣⎢⎡61616221−210313131⎦⎥⎤（正交矩阵)
V 1 = [ 1 3 1 2 1 6 − 1 2 2 6 0 ] \begin{aligned} &V_{1}=\left[\begin{array}{cc} \frac{1}{\sqrt{3}} & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{6}} & -\frac{1}{\sqrt{2}} \\ \frac{2}{\sqrt{6}} & 0 \end{array}\right] \end{aligned} V1=⎣⎢⎡31616221−210⎦⎥⎤
（3）根据公式计算出U
U 1 = A V 1 Σ − 1 = [ 1 2 1 2 1 2 1 2 0 0 ] U 2 = [ 0 0 1 ] U = [ U 1 ⋮ U 2 ] = [ 1 2 1 2 0 1 2 − 1 2 0 0 0 1 ] \begin{gathered}U_{1}=A V_{1} \Sigma^{-1}=\left[\begin{array}{cc} \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \\ 0 & 0 \end{array}\right]\\ U_{2}=\left[\begin{array}{l} 0 \\ 0 \\ 1 \end{array}\right] \\ U=\left[U_{1} \vdots U_{2}\right]=\left[\begin{array}{ccc} \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} & 0 \\ \frac{1}{\sqrt{2}} & -\frac{1}{\sqrt{2}} & 0 \\ 0 & 0 & 1 \end{array}\right] \end{gathered} U1=AV1Σ−1=⎣⎡2121021210⎦⎤U2=⎣⎡001⎦⎤U=[U1⋮U2]=⎣⎡2121021−210001⎦⎤
∴ A = U [ 3 0 0 0 1 0 0 0 0 ] V T \therefore A=U\left[\begin{array}{ccc} \sqrt{3} & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{array}\right] V^{T} ∴A=U⎣⎡300010000⎦⎤VT
总结
建议有时间的可以好好学学这一部分，这部分在编程的时候用处比较大，便于计算机计算分析矩阵。
参考资料
本文章题目以及知识点部分源于
1、《矩阵论》—张凯院，西北工业大学出版社第4版
2、《矩阵论—讲稿》—张凯院
3、《矩阵论》——国科大2021秋季叶世伟上课PPT
278
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
