1.反对称矩阵
\qquad
假设矩阵
A
A
A为
n
n
n阶方阵,若有
A
T
=
−
A
A^T=-A
AT=−A,则称矩阵
A
A
A为反对称矩阵。
2.向量的反对称矩阵
\qquad
两个三维列向量
V
1
=
[
V
1
x
V
1
y
V
1
z
]
T
V_1=[V_{1x}\ V_{1y}\ V_{1z}\ ]^T
V1=[V1x V1y V1z ]T和
V
2
=
[
V
2
x
V
2
y
V
2
z
]
T
V_2=[V_{2x}\ V_{2y}\ V_{2z}\ ]^T
V2=[V2x V2y V2z ]T之间的叉积为:
V
1
×
V
2
=
∣
i
j
k
V
1
x
V
1
y
V
1
z
V
2
x
V
2
y
V
2
z
∣
=
∣
V
1
y
V
2
z
−
V
1
z
V
2
y
−
(
V
1
x
V
2
z
−
V
1
z
V
2
x
)
V
1
x
V
2
y
−
V
1
y
V
2
x
∣
(1)
V_1 \times V_2= \begin{vmatrix} i&j&k\\ V_{1x}&V_{1y}&V_{1z}\\ V_{2x}&V_{2y}&V_{2z} \end{vmatrix} =\begin{vmatrix} V_{1y}V_{2z}-V_{1z}V_{2y} \\-(V_{1x}V_{2z}-V_{1z}V_{2x}) \\V_{1x}V_{2y}-V_{1y}V_{2x}\end{vmatrix} \tag{1}
V1×V2=∣∣∣∣∣∣iV1xV2xjV1yV2ykV1zV2z∣∣∣∣∣∣=∣∣∣∣∣∣V1yV2z−V1zV2y−(V1xV2z−V1zV2x)V1xV2y−V1yV2x∣∣∣∣∣∣(1)
\qquad
构造由向量
V
1
V_1
V1中各元素构成的特殊矩阵,其与向量
V
2
V_2
V2的矩阵乘法为:
[
0
−
V
1
z
V
1
y
V
1
z
0
−
V
1
x
−
V
1
y
V
1
x
0
]
[
V
2
x
V
2
y
V
2
z
]
=
[
V
1
y
V
2
z
−
V
1
z
V
2
y
−
(
V
1
x
V
2
z
−
V
1
z
V
x
)
V
1
x
V
2
y
−
V
1
y
V
2
x
]
(2)
\begin{bmatrix} 0&-V_{1z}&V_{1y}\\ V_{1z}&0&-V_{1x}\\ -V_{1y}&V_{1x}&0\end{bmatrix}\begin{bmatrix}V_{2x}\\ V_{2y}\\ V_{2z} \end{bmatrix}=\begin{bmatrix} V_{1y}V_{2z}-V_{1z}V_{2y} \\-(V_{1x}V_{2z}-V_{1z}V_{x}) \\V_{1x}V_{2y}-V_{1y}V_{2x}\end{bmatrix} \tag{2}
⎣⎡0V1z−V1y−V1z0V1xV1y−V1x0⎦⎤⎣⎡V2xV2yV2z⎦⎤=⎣⎡V1yV2z−V1zV2y−(V1xV2z−V1zVx)V1xV2y−V1yV2x⎦⎤(2)
\qquad
可以看出此时式(1)和式(2)右侧结果形式完全相同。将式(2)左侧的特殊矩阵记作:
V
×
=
[
0
−
V
z
V
y
V
z
0
−
V
x
−
V
y
V
x
0
]
(3)
V\times=\begin{bmatrix} 0&-V_z&Vy\\V_z&0&-V_x\\-V_y&V_x&0\end{bmatrix} \tag{3}
V×=⎣⎡0Vz−Vy−Vz0VxVy−Vx0⎦⎤(3)
\qquad
并将其称为向量
V
=
[
V
x
V
y
V
z
]
T
V=[V_x\ V_y\ V_z]^T
V=[Vx Vy Vz]T的反对称矩阵。
\qquad
引入反对称矩阵的概念后,两个向量的叉乘运算即可表示为前一个向量的反对称矩阵与后一个向量的矩阵乘法运算。即:
V
1
×
V
2
=
(
V
1
×
)
V
2
(4)
V_1\times V_2=(V_1\times)V_2 \tag{4}
V1×V2=(V1×)V2(4)
\qquad
若
V
V
V是实向量,则有:
(
V
×
)
H
=
(
V
×
)
T
=
−
(
V
×
)
(5)
(V\times)^H=(V\times)^T=-(V\times) \tag{5}
(V×)H=(V×)T=−(V×)(5)
\qquad
其中
H
H
H表示共轭转置。可进一步推出:
(
V
×
)
H
(
V
×
)
=
(
V
×
)
(
V
×
)
H
=
[
V
y
2
+
V
z
2
−
V
x
V
y
−
V
x
V
z
−
V
x
V
y
V
x
2
+
V
z
2
−
V
y
V
z
−
V
x
V
z
−
V
y
V
z
V
x
2
+
V
y
2
]
(6)
(V\times)^H(V\times)=(V\times)(V\times)^H=\begin{bmatrix} V_y^2+V_z^2& -V_xV_y &-V_xV_z\\ -V_xV_y&V_x^2+V_z^2&-V_yV_z\\-V_xV_z&-V_yV_z&V_x^2+V_y^2 \end{bmatrix} \tag{6}
(V×)H(V×)=(V×)(V×)H=⎣⎡Vy2+Vz2−VxVy−VxVz−VxVyVx2+Vz2−VyVz−VxVz−VyVzVx2+Vy2⎦⎤(6)
\qquad
可知反对称矩阵
(
V
×
)
(V\times)
(V×)同时还是正规矩阵。由正规矩阵的性质可知,存在酉矩阵
U
U
U使得
(
V
×
)
(V\times)
(V×)酉相似于对角阵,且不同特征值对应的特征向量相互正交。
\qquad
反
对
称
矩
阵
(
V
×
)
反对称矩阵(V\times)
反对称矩阵(V×)的特征多项式为:
f
(
λ
)
=
d
e
t
(
λ
I
−
(
A
×
)
)
=
∣
λ
V
z
−
V
y
−
V
z
λ
V
x
V
y
−
V
x
λ
∣
=
λ
3
+
(
V
x
2
+
V
y
2
+
V
z
2
)
λ
=
λ
3
+
v
2
λ
(7)
\begin{aligned}f(\lambda)=det(\lambda I-(A\times)) &=\begin{vmatrix} \lambda &V_z& -V_y\\ -V_z &\lambda &V_x\\V_y&-V_x&\lambda\end{vmatrix} \\&=\lambda^3+(V_x^2+V_y^2+V_z^2)\lambda\\&=\lambda^3+v^2\lambda \end{aligned} \tag{7}
f(λ)=det(λI−(A×))=∣∣∣∣∣∣λ−VzVyVzλ−Vx−VyVxλ∣∣∣∣∣∣=λ3+(Vx2+Vy2+Vz2)λ=λ3+v2λ(7)
\qquad
其中
v
=
∣
V
∣
=
V
x
2
+
V
y
2
+
V
z
2
v=|V|=\sqrt{V_x^2+V_y^2+V_z^2}
v=∣V∣=Vx2+Vy2+Vz2为向量
V
V
V的模。
\qquad
令特征多项式
f
(
λ
)
=
0
f(\lambda)=0
f(λ)=0,可得矩阵
(
V
×
)
(V\times)
(V×)的特征值为:
λ
1
=
0
λ
2
,
3
=
±
j
v
(8)
\lambda_1=0 \qquad \lambda_{2,3}=\pm jv \tag{8}
λ1=0λ2,3=±jv(8)
\qquad
当
λ
=
0
\lambda=0
λ=0时,对应的特征向量满足:
(
0
V
z
−
V
y
−
V
z
0
V
x
V
y
−
V
x
0
)
(
x
1
x
2
x
3
)
=
(
0
0
0
)
(9)
\begin{pmatrix} 0 &V_z& -V_y\\ -V_z &0 &V_x\\V_y&-V_x&0 \end{pmatrix} \begin{pmatrix} x_1\\x_2\\x_3\end{pmatrix}=\begin{pmatrix}0\\0\\0\end{pmatrix} \tag{9}
⎝⎛0−VzVyVz0−Vx−VyVx0⎠⎞⎝⎛x1x2x3⎠⎞=⎝⎛000⎠⎞(9)
\qquad
对应的特征向量可取:
u
1
=
(
V
x
V
y
V
z
)
(10)
u_1=\begin{pmatrix} V_x\\V_y\\V_z\end{pmatrix} \tag{10}
u1=⎝⎛VxVyVz⎠⎞(10)
\qquad
当
λ
=
j
v
\lambda= jv
λ=jv时,对应的特征向量满足:
(
j
v
−
V
z
V
y
V
z
j
v
−
V
x
−
V
y
V
x
j
v
)
(
x
1
x
2
x
3
)
=
(
0
0
0
)
⇒
(
j
v
−
V
z
V
y
0
V
x
2
+
V
y
2
V
y
V
z
−
j
v
V
x
0
V
y
V
z
+
j
v
V
x
V
x
2
+
V
z
2
)
(
x
1
x
2
x
3
)
=
(
0
0
0
)
(11)
\begin{aligned} &\begin{pmatrix} jv &-V_z& V_y\\ V_z &jv &-V_x\\-V_y&V_x&jv \end{pmatrix} \begin{pmatrix} x_1\\x_2\\x_3\end{pmatrix}=\begin{pmatrix}0\\0\\0\end{pmatrix} \\ \rArr&\begin{pmatrix} jv &-V_z& V_y\\ 0 &V_x^2+V_y^2 &V_yV_z-jvV_x\\0&V_yV_z+jvV_x&V_x^2+V_z^2 \end{pmatrix} \begin{pmatrix} x_1\\x_2\\x_3\end{pmatrix}=\begin{pmatrix}0\\0\\0\end{pmatrix} \end{aligned} \tag{11}
⇒⎝⎛jvVz−Vy−VzjvVxVy−Vxjv⎠⎞⎝⎛x1x2x3⎠⎞=⎝⎛000⎠⎞⎝⎛jv00−VzVx2+Vy2VyVz+jvVxVyVyVz−jvVxVx2+Vz2⎠⎞⎝⎛x1x2x3⎠⎞=⎝⎛000⎠⎞(11)
\qquad
对应的特征向量可取:
u
2
=
(
−
V
x
V
z
−
j
v
V
y
−
V
y
V
z
+
j
v
V
x
V
x
2
+
V
y
2
)
(12)
u_2=\begin{pmatrix} -V_xV_z-jvV_y\\-V_yV_z+jvV_x\\V_x^2+V_y^2\end{pmatrix} \tag{12}
u2=⎝⎛−VxVz−jvVy−VyVz+jvVxVx2+Vy2⎠⎞(12)
\qquad
同理,当
λ
=
−
j
v
\lambda= -jv
λ=−jv时,对应特征向量可取:
u
3
=
(
−
V
x
V
z
+
j
v
V
y
−
V
y
V
z
−
j
v
V
x
V
x
2
+
V
y
2
)
(13)
u_3=\begin{pmatrix} -V_xV_z+jvV_y\\-V_yV_z-jvV_x\\V_x^2+V_y^2\end{pmatrix} \tag{13}
u3=⎝⎛−VxVz+jvVy−VyVz−jvVxVx2+Vy2⎠⎞(13)
\qquad
显然,若
u
i
(
i
=
1
,
2
,
3
)
u_i(i=1,2,3)
ui(i=1,2,3)是反对称矩阵
(
V
×
)
(V\times)
(V×)对应特征值
λ
i
\lambda_i
λi的特征向量,那么
k
u
i
(
k
≠
0
)
ku_i(k\ne0)
kui(k=0)也应该是对应于
λ
i
\lambda_i
λi的特征向量。因此,反对称矩阵
(
V
×
)
(V\times)
(V×)的特征向量并不唯一。
\qquad
记:
U
=
[
u
1
u
2
u
3
]
Λ
=
d
i
a
g
(
λ
1
λ
2
λ
3
)
(14)
U=[u_1\ u_2\ u_3] \qquad \varLambda=diag(\lambda_1\ \lambda_2\ \lambda_3) \tag{14}
U=[u1 u2 u3]Λ=diag(λ1 λ2 λ3)(14)
\qquad
若
V
x
=
V
y
=
0
V_x=V_y=0
Vx=Vy=0,取反对称矩阵
(
V
×
)
(V\times)
(V×)的单位正交特征向量:
u
1
=
[
0
0
1
]
u
2
=
1
2
[
1
−
j
1
]
u
3
=
1
2
[
1
j
1
]
(15)
u_1=\begin{bmatrix} 0\\0\\1\end{bmatrix} \qquad u_2=\frac{1}{\sqrt{2}}\begin{bmatrix} 1\\-j\\1\end{bmatrix} \qquad u_3=\frac{1}{\sqrt{2}}\begin{bmatrix} 1\\j\\1\end{bmatrix} \tag{15}
u1=⎣⎡001⎦⎤u2=21⎣⎡1−j1⎦⎤u3=21⎣⎡1j1⎦⎤(15)
\qquad
可验证
U
H
U
=
I
U^HU=I
UHU=I,可知
U
U
U是酉矩阵。
\qquad
由矩阵特征值与特征向量的关系知:
(
V
×
)
U
=
U
Λ
(16)
(V\times)U=U\varLambda \tag{16}
(V×)U=UΛ(16)
\qquad
式(16)两遍同时左乘
U
−
1
U^{-1}
U−1的:
Λ
=
U
−
1
(
V
×
)
U
(17)
\varLambda=U^{-1}(V\times)U \tag{17}
Λ=U−1(V×)U(17)
\qquad
验证了反对称矩阵
(
V
×
)
(V\times)
(V×)可酉相似于对角阵,矩阵
U
U
U为相应的酉相似变换矩阵。
\qquad
反对称矩阵的幂运算如下:
(
V
×
)
1
=
v
0
(
V
×
)
(
V
×
)
2
=
V
V
T
−
v
2
I
=
v
0
(
V
×
)
2
(
V
×
)
3
=
(
V
V
T
−
v
2
I
)
(
V
×
)
=
V
V
T
(
V
×
)
−
v
2
(
V
×
)
=
V
⋅
0
1
×
3
−
v
2
(
V
×
)
2
=
−
v
2
(
V
×
)
2
(
V
×
)
4
=
(
V
×
)
3
(
V
×
)
=
−
v
2
(
V
×
)
2
(
V
×
)
5
=
(
V
×
)
2
(
V
×
)
3
=
(
V
V
T
−
v
2
I
)
(
−
v
2
(
V
×
)
)
=
v
4
(
V
×
)
(
V
×
)
6
=
(
V
×
)
3
(
V
×
)
3
=
(
−
v
2
(
V
×
)
)
(
−
v
2
(
V
×
)
)
=
v
4
(
V
×
)
2
⋮
(18)
\begin{aligned} (V\times)^1 &=v^0(V\times) \\ (V\times)^2 &= VV^T-v^2I=v^0(V\times)^2 \\ (V\times)^3 &= (VV^T-v^2I)(V\times)=VV^T(V\times)-v^2(V\times)=V\cdot0_{1\times 3}-v^2(V\times)^2 =-v^2(V\times)^2\\(V\times)^4 &= (V\times)^3(V\times)=-v^2(V\times)^2 \\ (V\times)^5 &=(V\times)^2(V\times)^3= (VV^T-v^2I)(-v^2(V\times))=v^4(V\times)\\ (V\times)^6 &=(V\times)^3(V\times)^3= (-v^2(V\times)) (-v^2(V\times))=v^4(V\times)^2\\ \vdots \end{aligned} \tag{18}
(V×)1(V×)2(V×)3(V×)4(V×)5(V×)6⋮=v0(V×)=VVT−v2I=v0(V×)2=(VVT−v2I)(V×)=VVT(V×)−v2(V×)=V⋅01×3−v2(V×)2=−v2(V×)2=(V×)3(V×)=−v2(V×)2=(V×)2(V×)3=(VVT−v2I)(−v2(V×))=v4(V×)=(V×)3(V×)3=(−v2(V×))(−v2(V×))=v4(V×)2(18)
\qquad
由式(18)可写出通项公式:
(
V
×
)
i
=
{
(
−
1
)
i
−
1
2
v
i
−
1
(
V
×
)
2
i=1,3,5,
…
(
−
1
)
i
−
2
2
v
i
−
2
(
V
×
)
2
i=2,4,6,
…
(19)
\begin{aligned}(V\times)^i= \begin{cases} (-1)^{\frac{i-1}{2}}v^{i-1}(V\times)^2 &\text{i=1,3,5,…}\\ (-1)^{\frac{i-2}{2}}v^{i-2}(V\times)^2 &\text{i=2,4,6,…}\end{cases} \end{aligned} \tag{19}
(V×)i={(−1)2i−1vi−1(V×)2(−1)2i−2vi−2(V×)2i=1,3,5,…i=2,4,6,…(19)
3.反对称矩阵的矩阵指数函数
\qquad
根据哈密顿-凯莱定理,矩阵的指数函数
e
(
V
×
)
e^{(V\times)}
e(V×)可以展开成
(
V
×
)
(V\times)
(V×)的有限项级数展开式,即:
e
(
V
×
)
=
∑
i
=
0
∞
(
V
×
)
i
i
!
=
k
0
I
+
k
1
(
V
×
)
+
k
2
(
V
×
)
2
(20)
e^{(V\times)}=\sum_{i=0}^{\infty} \frac{(V\times)^i}{i!}=k_0I+k_1(V\times)+k_2(V\times)^2 \tag{20}
e(V×)=i=0∑∞i!(V×)i=k0I+k1(V×)+k2(V×)2(20)
\qquad
其中
k
0
,
k
1
,
k
2
k_0,k_1,k_2
k0,k1,k2为待定系数。
\qquad
由式(17)和式(20)可得:
e
Λ
=
e
U
−
1
(
V
×
)
U
=
∑
i
=
0
∞
(
U
−
1
(
V
×
)
U
)
i
i
!
=
U
−
1
[
∑
i
=
0
∞
(
V
×
)
i
i
!
]
U
=
U
−
1
e
(
V
×
)
U
=
U
−
1
[
k
0
I
+
k
1
(
V
×
)
+
k
2
(
V
×
)
2
]
U
=
k
0
U
−
1
U
+
k
1
U
−
1
(
V
×
)
U
+
k
2
U
−
1
(
V
×
)
2
U
=
k
o
I
+
k
1
Λ
+
k
2
Λ
2
(21)
\begin{aligned} e^{\varLambda} &=e^{U^{-1}(V\times)U} =\sum_{i=0}^{\infty} \frac{(U^{-1}(V\times)U)^i}{i!} =U^{-1}[\sum_{i=0}^{\infty} \frac{(V\times)^i}{i!}]U \\ &=U^{-1}e^{(V\times)}U= U^{-1}[k_0I+k_1(V\times)+k_2(V\times)^2 ]U \\&=k_0U^{-1}U+k_1U^{-1}(V\times)U+k_2U^{-1}(V\times)^2U \\ &=k_oI+k_1\varLambda+k_2\varLambda^2\end{aligned} \tag{21}
eΛ=eU−1(V×)U=i=0∑∞i!(U−1(V×)U)i=U−1[i=0∑∞i!(V×)i]U=U−1e(V×)U=U−1[k0I+k1(V×)+k2(V×)2]U=k0U−1U+k1U−1(V×)U+k2U−1(V×)2U=koI+k1Λ+k2Λ2(21)
\qquad
将式(21)两边展开成矩阵形式:
[
e
λ
1
0
0
0
e
λ
2
0
0
0
e
λ
3
]
=
[
k
0
+
k
1
λ
1
+
k
2
λ
1
2
0
0
0
k
0
+
k
1
λ
2
+
k
2
λ
2
2
0
0
0
k
0
+
k
1
λ
3
+
k
2
λ
3
2
]
(22)
\begin{bmatrix} e^{\lambda_1}&0&0 \\ 0&e^{\lambda_2}&0 \\ 0&0&e^{\lambda_3} \end{bmatrix}=\begin{bmatrix}k_0+k_1\lambda_1+k_2\lambda_1^2&0&0 \\ 0&k_0+k_1\lambda_2+k_2\lambda_2^2 &0\\ 0&0&k_0+k_1\lambda_3+k_2\lambda_3^2 \\\end{bmatrix} \tag{22}
⎣⎡eλ1000eλ2000eλ3⎦⎤=⎣⎡k0+k1λ1+k2λ12000k0+k1λ2+k2λ22000k0+k1λ3+k2λ32⎦⎤(22)
\qquad
将式(8)带入式(22)得:
{
e
0
=
k
0
e
j
v
=
k
0
+
k
1
(
j
v
)
+
k
2
(
j
v
)
2
e
−
j
v
=
k
0
+
k
1
(
−
j
v
)
+
k
2
(
−
j
v
)
2
即
{
k
0
=
1
k
0
+
k
1
(
j
v
)
−
k
2
v
2
=
c
o
s
v
+
j
s
i
n
v
k
0
−
k
1
(
j
v
)
−
k
2
v
2
=
c
o
s
v
−
j
s
i
n
v
(23)
\begin{cases}e^0 =k_0\\e^{jv}=k_0+k_1(jv)+k_2(jv)^2 \\ e^{-jv}=k_0+k_1(-jv)+k_2(-jv)^2\end{cases} 即\begin{cases} k_0 =1 \\k_0+k_1(jv)-k_2v^2=cosv+jsinv \\ k_0-k_1(jv)-k_2v^2=cosv-jsinv \end{cases} \tag{23}
⎩⎪⎨⎪⎧e0=k0ejv=k0+k1(jv)+k2(jv)2e−jv=k0+k1(−jv)+k2(−jv)2即⎩⎪⎨⎪⎧k0=1k0+k1(jv)−k2v2=cosv+jsinvk0−k1(jv)−k2v2=cosv−jsinv(23)
\qquad
由式(23)解得待定系数:
k
0
=
1
,
k
1
=
s
i
n
v
v
,
k
2
=
1
−
c
o
s
v
v
2
(24)
k_0=1,k_1=\frac{sinv}{v},k_2=\frac{1-cosv}{v^2} \tag{24}
k0=1,k1=vsinv,k2=v21−cosv(24)
\qquad
将式(24)回带入式(23),则有反对称矩阵的矩阵函数公式为:
e
(
V
×
)
=
I
+
s
i
n
v
v
(
V
×
)
+
1
−
c
o
s
v
v
2
(
V
×
)
2
(25)
e^{(V\times)}=I+\frac{sinv}{v}(V\times)+\frac{1-cosv}{v^2}(V\times)^2 \tag{25}
e(V×)=I+vsinv(V×)+v21−cosv(V×)2(25)
\qquad
直接将式(19)带入式(20),亦可求得:
e
(
V
×
)
=
∑
i
=
0
∞
(
V
×
)
i
i
!
=
(
V
×
)
0
+
1
1
!
(
V
×
)
1
+
1
2
!
(
V
×
)
2
+
1
3
!
(
V
×
)
3
+
1
4
!
(
V
×
)
4
…
=
(
V
×
)
0
+
[
1
1
!
(
V
×
)
1
+
1
3
!
(
V
×
)
3
+
1
5
!
(
V
×
)
5
+
…
]
+
[
1
2
!
(
V
×
)
2
+
1
4
!
(
V
×
)
4
+
1
6
!
(
V
×
)
6
+
…
]
=
(
V
×
)
0
+
[
1
1
!
(
V
×
)
−
v
2
3
!
(
V
×
)
+
v
4
5
!
(
V
×
)
+
…
]
+
[
1
2
!
(
V
×
)
2
−
v
2
4
!
(
V
×
)
2
+
v
4
6
!
(
V
×
)
2
+
…
]
=
I
+
s
i
n
v
v
(
V
×
)
+
1
−
c
o
s
v
v
2
(
V
×
)
2
(26)
\begin{aligned} e^{(V\times)}&=\sum_{i=0}^{\infty} \frac{(V\times)^i}{i!}=(V\times)^0+\frac{1}{1!}(V\times)^1+\frac{1}{2!}(V\times)^2+\frac{1}{3!}(V\times)^3+\frac{1}{4!}(V\times)^4…\\ &=(V\times)^0+[\frac{1}{1!}(V\times)^1+\frac{1}{3!}(V\times)^3+\frac{1}{5!}(V\times)^5+…]+[\frac{1}{2!}(V\times)^2+\frac{1}{4!}(V\times)^4+\frac{1}{6!}(V\times)^6+…] \\&=(V\times)^0+[\frac{1}{1!}(V\times)-\frac{v^2}{3!}(V\times)+\frac{v^4}{5!}(V\times)+…]+[\frac{1}{2!}(V\times)^2-\frac{v^2}{4!}(V\times)^2+\frac{v^4}{6!}(V\times)^2+…] \\ &=I+\frac{sinv}{v}(V\times)+\frac{1-cosv}{v^2}(V\times)^2\end{aligned} \tag{26}
e(V×)=i=0∑∞i!(V×)i=(V×)0+1!1(V×)1+2!1(V×)2+3!1(V×)3+4!1(V×)4…=(V×)0+[1!1(V×)1+3!1(V×)3+5!1(V×)5+…]+[2!1(V×)2+4!1(V×)4+6!1(V×)6+…]=(V×)0+[1!1(V×)−3!v2(V×)+5!v4(V×)+…]+[2!1(V×)2−4!v2(V×)2+6!v4(V×)2+…]=I+vsinv(V×)+v21−cosv(V×)2(26)
\qquad
根据式(21)可得:
e
(
V
×
)
U
=
U
e
Λ
=
[
e
Λ
1
u
1
e
Λ
2
u
2
e
Λ
3
u
3
]
(27)
e^{(V\times)}U=Ue^{\varLambda}=\begin{bmatrix} e^{\varLambda_1}u_1& e^{\varLambda_2}u_2& e^{\varLambda_3}u_3 \end{bmatrix} \tag{27}
e(V×)U=UeΛ=[eΛ1u1eΛ2u2eΛ3u3](27)
\qquad
对比式(27)与式(16),可知
e
(
V
×
)
e^{(V\times)}
e(V×)与反对称矩阵
(
V
×
)
(V\times)
(V×)具有相同的特征向量,均为矩阵
U
U
U的列向量:
{
λ
1
′
=
e
λ
1
=
e
0
=
1
λ
2
′
=
e
λ
2
=
e
j
v
=
c
o
s
v
+
j
s
i
n
v
λ
3
′
=
e
λ
3
=
e
−
j
v
=
c
o
s
v
−
j
s
i
n
v
(28)
\begin{cases}\lambda'_1=e^{\lambda_1}= e^0=1 \\ \lambda'_2=e^{\lambda_2}= e^{jv}=cosv+jsinv\\ \lambda'_3=e^{\lambda_3}= e^{-jv}=cosv-jsinv\end{cases}\tag{28}
⎩⎪⎨⎪⎧λ1′=eλ1=e0=1λ2′=eλ2=ejv=cosv+jsinvλ3′=eλ3=e−jv=cosv−jsinv(28)
\qquad
由式(28)可知
(
e
Λ
)
H
e
Λ
=
I
(e^{\varLambda})^He^{\varLambda}=I
(eΛ)HeΛ=I成立,则
e
Λ
e^{\varLambda}
eΛ是酉矩阵。
\qquad
若
V
V
V是实向量,则
e
V
×
e^{V\times}
eV×是实矩阵。同时
e
V
×
e^{V\times}
eV×还是单位正交阵,证明如下:
[
e
(
V
×
)
]
T
e
(
V
×
)
=
[
I
+
s
i
n
v
v
(
V
×
)
+
1
−
c
o
s
v
v
2
(
V
×
)
2
]
T
e
(
V
×
)
=
[
I
+
s
i
n
v
v
(
V
×
)
T
+
1
−
c
o
s
v
v
2
(
(
V
×
)
2
)
T
]
e
(
V
×
)
=
[
I
+
s
i
n
v
v
(
−
V
×
)
T
+
1
−
c
o
s
v
v
2
(
(
−
V
×
)
2
)
T
]
e
(
V
×
)
=
e
(
−
V
×
)
e
(
V
×
)
=
I
(29)
\begin{aligned}[e^{(V\times)}]^T e^{(V\times)} &=[I+\frac{sinv}{v}(V\times)+\frac{1-cosv}{v^2}(V\times)^2]^Te^{(V\times)} \\ &=[I+\frac{sinv}{v}(V\times)^T+\frac{1-cosv}{v^2}((V\times)^2)^T]e^{(V\times)} \\&=[I+\frac{sinv}{v}(-V\times)^T+\frac{1-cosv}{v^2}((-V\times)^2)^T]e^{(V\times)} \\ &= e^{(-V\times)}e^{(V\times)} \\&=I \end{aligned} \tag{29}
[e(V×)]Te(V×)=[I+vsinv(V×)+v21−cosv(V×)2]Te(V×)=[I+vsinv(V×)T+v21−cosv((V×)2)T]e(V×)=[I+vsinv(−V×)T+v21−cosv((−V×)2)T]e(V×)=e(−V×)e(V×)=I(29)
\qquad
故
d
e
t
(
e
(
V
×
)
)
=
e
t
r
(
V
×
)
=
e
0
=
I
det(e^{(V\times)})=e^{tr(V\times)}=e^0=I
det(e(V×))=etr(V×)=e0=I,可以看出三阶单位正交阵只有行列式为1时才可以表示为
e
(
V
×
)
e^{(V\times)}
e(V×)的形式。对于行列式为1的单位正交阵我们一般称之为右手直角坐标变换矩阵;反之,行列式为-1的单位正交阵称为左手直角坐标变换矩阵。
参考文献:
[1] 同济大学数学系. 工程数学, 线性代数(第6版)[M]. 北京:高等教育出版社, 2014.
[2] 严恭敏, 翁浚. 捷联惯导算法与组合导航原理[M]. 西安:西北工业大学出版社, 2020.