张量的谱表示
基于特征多项式:
如果
T
T
T 是对称二阶张量,那么有三个实数特征值:
T
1
,
T
2
,
T
3
T_1, T_2, T_3
T1,T2,T3, 对应有特征向量:
主空间由特征向量构成的正交基
n
^
(
1
)
,
n
^
(
2
)
,
n
^
(
3
)
\hat n^{(1)}, \hat n^{(2)}, \hat n^{(3)}
n^(1),n^(2),n^(3) 所张成,张量的分量则由特征值表示:
特征向量形成的变换矩阵
A
A
A:
T
′
=
A
T
A
′
T' = ATA'
T′=ATA′
由于
A
−
1
=
A
T
A^{-1} = A^T
A−1=AT, 所以:
T
=
A
T
T
′
A
T = A^TT'A
T=ATT′A
其中:
将
T
=
A
T
T
′
A
T = A^TT'A
T=ATT′A,显式表示出来:
其中:
因此,可以将二阶张量的分量用特征值和特征向量的函数表示(谱表示):
T
i
j
=
T
1
n
^
i
(
1
)
n
^
j
(
1
)
+
T
2
n
^
i
(
2
)
n
^
j
(
2
)
+
T
3
n
^
i
(
3
)
n
^
j
(
3
)
T_{ij} = T_1 \hat n_i^{(1)} \hat n_j^{(1)} + T_2 \hat n_i^{(2)} \hat n_j^{(2)}+T_3 \hat n_i^{(3)} \hat n_j^{(3)}
Tij=T1n^i(1)n^j(1)+T2n^i(2)n^j(2)+T3n^i(3)n^j(3)
张量表示:
T
=
T
1
n
^
(
1
)
⨂
n
^
(
1
)
+
T
2
n
^
(
2
)
⨂
n
^
(
2
)
+
T
3
n
^
(
3
)
⨂
n
^
(
3
)
T = T_1 \hat n^{(1)}\bigotimes \hat n^{(1)} + T_2 \hat n^{(2)}\bigotimes \hat n^{(2)}+ T_3 \hat n^{(3)}\bigotimes \hat n^{(3)}
T=T1n^(1)⨂n^(1)+T2n^(2)⨂n^(2)+T3n^(3)⨂n^(3)
张量可以表示成并矢的线性组合:
T
=
∑
a
=
1
3
T
a
n
^
(
a
)
⨂
n
^
(
a
)
(
二阶张量的谱表示
)
\boxed{T = \sum_{a=1}^{3}T_a\hat n^{(a)}\bigotimes \hat n^{(a)}} \quad (二阶张量的谱表示)
T=a=1∑3Tan^(a)⨂n^(a)(二阶张量的谱表示)
NOTE: 也可以从定义得到:
因为
1
=
n
^
i
⨂
n
^
i
1 = \hat n_i \bigotimes \hat n_i
1=n^i⨂n^i, 所以表示成
1
=
∑
a
=
1
3
n
^
(
a
)
⨂
n
^
(
a
)
1 = \sum_{a=1}^{3}\hat n^{(a)}\bigotimes \hat n^{(a)}
1=∑a=13n^(a)⨂n^(a),那么:
T
=
T
⋅
1
=
T
⋅
(
∑
a
=
1
3
n
^
(
a
)
⨂
n
^
(
a
)
)
=
∑
a
=
1
3
T
⋅
n
^
(
a
)
⨂
n
^
(
a
)
=
∑
a
=
1
3
T
a
n
^
(
a
)
⨂
n
^
(
a
)
T = T \cdot 1 = T \cdot ( \sum_{a=1}^{3}\hat n^{(a)}\bigotimes \hat n^{(a)})\\= \sum_{a=1}^{3}T \cdot\hat n^{(a)}\bigotimes \hat n^{(a)}\\=\sum_{a=1}^{3}T_a \hat n^{(a)}\bigotimes \hat n^{(a)}
T=T⋅1=T⋅(a=1∑3n^(a)⨂n^(a))=a=1∑3T⋅n^(a)⨂n^(a)=a=1∑3Tan^(a)⨂n^(a)
其中,用到了特征值和特征向量的定义: T ⋅ n ^ ( a ) = T a n ^ ( a ) T \cdot\hat n^{(a)} = T_a \hat n^{(a)} T⋅n^(a)=Tan^(a)
现在,考虑一个正交张量
R
R
R,将正交变换作用到一个单位向量
N
^
\hat N
N^,得到一个单位向量
n
^
\hat n
n^,
n
^
=
R
⋅
N
^
\hat n = R \cdot \hat N
n^=R⋅N^,因此,可以将正交张量
R
R
R 表示成:
R
=
R
⋅
1
=
R
⋅
(
∑
a
=
1
3
N
^
(
a
)
⨂
N
^
(
a
)
)
=
∑
a
=
1
3
R
⋅
N
^
(
a
)
⨂
N
^
(
a
)
=
∑
a
=
1
3
n
^
(
a
)
⨂
n
^
(
a
)
R = R \cdot 1 = R \cdot ( \sum_{a=1}^{3}\hat N^{(a)}\bigotimes \hat N^{(a)})\\= \sum_{a=1}^{3}R \cdot\hat N^{(a)}\bigotimes \hat N^{(a)}\\=\sum_{a=1}^{3}\hat n^{(a)}\bigotimes \hat n^{(a)}
R=R⋅1=R⋅(a=1∑3N^(a)⨂N^(a))=a=1∑3R⋅N^(a)⨂N^(a)=a=1∑3n^(a)⨂n^(a)
张量的谱表示在张量代数操作种是非常有用的,比如在主空间的张量的幂可以表示成:
所以
T
n
T^n
Tn 的谱表示为:
T
n
=
∑
a
=
1
3
T
a
n
n
^
(
a
)
⨂
n
^
(
a
)
T^n = \sum_{a= 1}^3T_a^n \hat n^{(a)} \bigotimes \hat n^{(a)}
Tn=a=1∑3Tann^(a)⨂n^(a)
同样,
T
\sqrt{T}
T的谱表示为:
T
=
∑
a
=
1
3
T
a
n
^
(
a
)
⨂
n
^
(
a
)
\sqrt{T}= \sum_{a= 1}^3\sqrt{T_a} \hat n^{(a)} \bigotimes \hat n^{(a)}
T=a=1∑3Tan^(a)⨂n^(a)
然后,我们考虑一个正定张量,有正的特征值,条件是: x ^ ⋅ T ⋅ x ^ ≥ 0 \hat x \cdot T \cdot \hat x \geq 0 x^⋅T⋅x^≥0,将此条件转化成谱表示:
x ^ ⋅ T ⋅ x ^ ≥ 0 ⟹ x ^ ⋅ ( ∑ a = 1 3 T a n ^ ( a ) ⨂ n ^ ( a ) ) ⋅ x ^ ≥ 0 ⟹ ∑ a = 1 3 T a x ^ ⋅ n ^ ( a ) ⨂ n ^ ( a ) ⋅ x ^ ≥ 0 ⟹ ∑ a = 1 3 T a [ x ^ ⋅ n ^ ( a ) ] 2 ≥ 0 ⟹ T 1 ( x ^ ⋅ n ^ ( 1 ) ) 2 + T 2 ( x ^ ⋅ n ^ ( 2 ) ) 2 + T 3 ( x ^ ⋅ n ^ ( 3 ) ) 2 ≥ 0 \hat x \cdot T \cdot \hat x \geq 0 \\ \implies \hat x \cdot (\sum_{a= 1}^3T_a \hat n^{(a)} \bigotimes \hat n^{(a)}) \cdot \hat x \geq 0\\ \implies \sum_{a= 1}^3T_a \hat x \cdot \hat n^{(a)} \bigotimes \hat n^{(a)} \cdot \hat x \geq 0 \\ \implies \sum_{a= 1}^3T_a [\hat x \cdot \hat n^{(a)} ]^2\geq 0 \\ \implies T_1(\hat x \cdot \hat n^{(1)})^2+T_2(\hat x \cdot \hat n^{(2)})^2+T_3(\hat x \cdot \hat n^{(3)})^2\geq 0 x^⋅T⋅x^≥0⟹x^⋅(a=1∑3Tan^(a)⨂n^(a))⋅x^≥0⟹a=1∑3Tax^⋅n^(a)⨂n^(a)⋅x^≥0⟹a=1∑3Ta[x^⋅n^(a)]2≥0⟹T1(x^⋅n^(1))2+T2(x^⋅n^(2))2+T3(x^⋅n^(3))2≥0
以上表达式在 x ^ ≠ 0 \hat x \neq 0 x^=0才成立
当 x ^ = n ^ \hat x = \hat n x^=n^,上式变成 T 1 ( n ^ ( 1 ) ⋅ n ^ ( 1 ) ) 2 = T 1 ≥ 0 T_1(\hat n^{(1)} \cdot \hat n^{(1)})^2 = T_1 \geq 0 T1(n^(1)⋅n^(1))2=T1≥0, 表示当张量是半正定,特征值大于等于0。因此张量是正定的,当且仅当特征值都为正的,不等于0
小结:正定张量的迹大于0,如果正定张量的迹为0,表示张量是个零张量
四阶张量的谱表示:
四阶张量
I
I
II
II的定义:
由于
I
I
II
II 是各向同性张量,所以也可以把它表示成任何正交基
n
^
(
a
)
\hat n^{(a)}
n^(a):
I
I
=
∑
a
=
1
3
∑
b
=
1
3
n
^
(
a
)
⨂
n
^
(
b
)
⨂
n
^
(
a
)
⨂
n
^
(
b
)
II = \sum_{a=1}^3 \sum_{b = 1}^3\hat n^{(a)} \bigotimes \hat n^{(b)} \bigotimes \hat n^{(a)} \bigotimes \hat n^{(b)}
II=a=1∑3b=1∑3n^(a)⨂n^(b)⨂n^(a)⨂n^(b)
同理,可以得到
I
I
‾
和
I
I
‾
\overline {II}和 \underline {II}
II和II的谱表示:
问题1.34 W W W是反对称二阶张量, V V V 是正定对称张量,其谱表示为: V = ∑ a = 1 3 λ a n ^ ( a ) ⨂ n ^ ( a ) V = \sum_{a= 1}^3 \lambda_a \hat n^{(a)}\bigotimes \hat n^{(a)} V=∑a=13λan^(a)⨂n^(a),证明: W W W 可以表示成: W = ∑ a , b = 1 , a ≠ b 3 W a b n ^ ( a ) ⨂ n ^ ( b ) W = \sum_{a, b =1, a \neq b}^3 W_{ab}\hat n^{(a)}\bigotimes \hat n^{(b)} W=∑a,b=1,a=b3Wabn^(a)⨂n^(b); 且证明: W ⋅ V − V ⋅ W = ∑ a , b = 1 , a ≠ b 3 W a b ( λ b − λ a ) n ^ ( a ) ⨂ n ^ ( b ) W \cdot V - V \cdot W = \sum_{a, b =1, a \neq b}^3 W_{ab}(\lambda_b -\lambda_a)\hat n^{(a)}\bigotimes \hat n^{(b)} W⋅V−V⋅W=∑a,b=1,a=b3Wab(λb−λa)n^(a)⨂n^(b)
Cayley-Hamilton 定理
Cayley-Hamilton 定理: 对于任意张量
T
T
T,满足特征方程,特征值满足
λ
3
−
λ
2
I
T
+
λ
I
I
T
−
I
I
I
T
=
0
\lambda^3 - \lambda^2 I_T + \lambda II_T - III_T=0
λ3−λ2IT+λIIT−IIIT=0,所以张量
T
T
T:
T
3
−
T
2
I
T
+
T
I
I
T
−
I
I
I
T
=
0
T^3 - T^2 I_T +T II_T - III_T=0
T3−T2IT+TIIT−IIIT=0
Cayley-Hamilton 定理的应用:表示张量的幂
T
3
⋅
T
−
T
2
⋅
T
I
T
+
T
⋅
T
I
I
T
−
I
I
I
T
1
⋅
T
=
0
⟹
T
4
=
T
3
I
T
−
T
2
I
I
T
+
T
I
I
I
T
T^3\cdot T - T^2 \cdot TI_T +T \cdot TII_T - III_T 1 \cdot T=0 \\ \implies T^4 = T^3 I_T -T^2 II_T +T III_T
T3⋅T−T2⋅TIT+T⋅TIIT−IIIT1⋅T=0⟹T4=T3IT−T2IIT+TIIIT
应用Cayley-Hamilton 定理, 可以将第三不变量表示成迹的函数:
T
3
:
1
−
T
2
:
1
I
T
+
T
:
1
I
I
T
−
I
I
I
T
1
:
1
=
0
:
1
T^3 :1 - T^2:1I_T +T : 1II_T - III_T 1:1=0:1
T3:1−T2:1IT+T:1IIT−IIIT1:1=0:1
由于:
T
3
:
1
=
T
r
(
T
3
)
;
T
2
:
1
=
T
r
(
T
2
)
;
T
:
1
=
T
r
(
T
)
;
1
:
1
=
T
r
(
1
)
=
3
;
0
:
1
=
T
r
(
0
)
=
0
T^3 : 1 = Tr(T^3); \\ T^2: 1=Tr(T^2); \\ T:1=Tr(T);\\ 1:1=Tr(1)=3;\\ 0:1=Tr(0) = 0
T3:1=Tr(T3);T2:1=Tr(T2);T:1=Tr(T);1:1=Tr(1)=3;0:1=Tr(0)=0
所以:
T
r
(
T
3
)
−
I
T
T
r
(
T
2
)
+
I
I
T
T
r
(
T
)
−
3
I
I
I
T
=
0
⟹
I
I
I
T
=
1
3
(
T
r
(
T
3
)
−
I
T
T
r
(
T
2
)
+
I
I
T
T
r
(
T
)
)
Tr(T^3) - I_T Tr(T^2) +II_T Tr(T) - 3III_T =0 \\ \implies III_T = \frac{1}{3}(Tr(T^3) - I_T Tr(T^2) +II_T Tr(T) )
Tr(T3)−ITTr(T2)+IITTr(T)−3IIIT=0⟹IIIT=31(Tr(T3)−ITTr(T2)+IITTr(T))
将不变量代入:
得到:
I
I
I
T
=
1
3
(
T
r
(
T
3
)
−
3
2
T
r
(
T
2
)
T
r
(
T
)
+
1
2
[
T
r
(
T
)
]
3
)
III_T = \frac{1}{3}(Tr(T^3)-\frac{3}{2}Tr(T^2)Tr(T)+\frac{1}{2}[Tr(T)]^3)
IIIT=31(Tr(T3)−23Tr(T2)Tr(T)+21[Tr(T)]3)
指标形式:
I
I
I
T
=
1
3
(
T
i
j
T
j
k
T
k
i
−
3
2
T
i
j
T
j
i
T
k
k
+
1
2
T
i
i
T
j
j
T
k
k
)
III_T=\frac{1}{3} (T_{ij}T_{jk}T_{ki}-\frac{3}{2}T_{ij}T_{ji}T_{kk}+\frac{1}{2}T_{ii}T_{jj}T_{kk})
IIIT=31(TijTjkTki−23TijTjiTkk+21TiiTjjTkk)
问题1.35 基于Cayley-Hamilton 定理,用张量幂求出张量的逆
n×n矩阵
A
n
×
n
A_{n \times n}
An×n,那么特征行列式为:
∣
λ
1
n
×
n
−
A
∣
=
0
|\lambda 1_{n\times n}-A|=0
∣λ1n×n−A∣=0
展开,得:
λ
n
−
I
1
λ
n
−
1
+
I
2
λ
n
−
2
−
.
.
.
(
−
1
)
n
I
n
=
0
\lambda^n -I_1\lambda^{n-1}+I_2\lambda^{n-2}- ... (-1)^nI_n=0
λn−I1λn−1+I2λn−2−...(−1)nIn=0
其中
I
1
,
I
2
,
.
.
.
,
I
n
I_1, I_2, ..., I_n
I1,I2,...,In是A的不变量,n=3是二阶张量的情况
应用Cayley-Hamilton 定理,
A
n
−
I
1
A
n
−
1
+
I
2
A
n
−
2
−
.
.
.
+
(
−
1
)
n
I
n
1
=
0
A^n-I_1A^{n-1}+I_2A^{n-2}-...+(-1)^nI_n 1=0
An−I1An−1+I2An−2−...+(−1)nIn1=0
通过乘以
A
−
1
A^{-1}
A−1,可以得到:
A
n
A
−
1
−
I
1
A
n
−
1
A
−
1
+
I
2
A
n
−
2
A
−
1
−
.
.
.
+
(
−
1
)
n
I
n
1
A
−
1
=
0
⟹
A
n
−
1
−
I
1
A
n
−
2
+
I
2
A
n
−
3
−
.
.
.
+
(
−
1
)
n
−
1
I
n
−
1
1
+
(
−
1
)
n
I
n
A
−
1
=
0
A^nA^{-1}-I_1A^{n-1}A^{-1}+I_2A^{n-2}A^{-1}-...+(-1)^nI_n 1A^{-1}=0 \\ \implies A^{n-1}-I_1A^{n-2}+I_2A^{n-3}-...+(-1)^{n-1}I_{n-1} 1 +(-1)^nI_nA^{-1}=0
AnA−1−I1An−1A−1+I2An−2A−1−...+(−1)nIn1A−1=0⟹An−1−I1An−2+I2An−3−...+(−1)n−1In−11+(−1)nInA−1=0
那么:
A
−
1
=
(
−
1
)
n
−
1
I
n
(
A
n
−
1
−
I
1
A
n
−
2
+
I
2
A
n
−
3
−
.
.
.
+
(
−
1
)
n
−
1
I
n
−
1
1
)
A^{-1}= \frac{(-1)^{n-1}}{I_n}(A^{n-1}-I_1A^{n-2}+I_2A^{n-3}-...+(-1)^{n-1}I_{n-1} 1 )
A−1=In(−1)n−1(An−1−I1An−2+I2An−3−...+(−1)n−1In−11)
只有当
I
n
=
det
A
≠
0
I_n = \det A \neq 0
In=detA=0,
A
−
1
A^{-1}
A−1才存在
问题1.36 验证Cayley-Hamilton 定理
参考教材:
Eduardo W.V. Chaves, Notes On Continuum Mechanics