雅可比矩阵的意义
雅可比矩阵的应用场景是进行坐标基变换的时候,揭露微元之间的关系。
设一个在 x y xy xy 坐标下的向量 m → \overrightarrow{m} m ,用这个坐标系的基来表示: m → = m x x → + m y y → \overrightarrow{m}=m_x\overrightarrow{x}+m_y\overrightarrow{y} m=mxx+myy
设
m
→
\overrightarrow{m}
m 通过函数
f
f
f 变换到了
u
v
uv
uv 坐标系后成为
n
→
\overrightarrow{n}
n ,用新基表示:
n
→
=
n
u
u
→
+
n
v
v
→
\overrightarrow{n}=n_u\overrightarrow{u}+n_v\overrightarrow{v}
n=nuu+nvv ,并且有:
f
(
m
→
)
=
n
→
f(\overrightarrow{m})=\overrightarrow{n}
f(m)=n
对于二元函数泰勒公式,我们有:
f
(
x
+
△
x
,
y
+
△
y
)
=
f
(
x
,
y
)
+
△
x
⋅
∂
f
∂
x
+
△
y
⋅
∂
f
∂
y
+
o
2
f(x+\triangle x,y+\triangle y)=f(x,y)+\triangle x\cdot\frac{\partial f}{\partial x}+\triangle y\cdot\frac{\partial f}{\partial y}+o^2
f(x+△x,y+△y)=f(x,y)+△x⋅∂x∂f+△y⋅∂y∂f+o2
对于复杂的函数
f
f
f ,其映射的规律是难以讨论的。微积分的思想永远是用“直线”,“平面”来拟合"曲线",“曲面”,在这里也是这样,我们假设
f
f
f 对向量
m
→
\overrightarrow{m}
m 的向量空间邻域内的所有的向量的映射都是线性的,即泰勒公式忽略二次项以后的所有项:
f
(
x
+
△
x
,
y
+
△
y
)
=
f
(
x
,
y
)
+
△
x
⋅
∂
f
∂
x
+
△
y
⋅
∂
f
∂
y
f(x+\triangle x,y+\triangle y)=f(x,y)+\triangle x\cdot\frac{\partial f}{\partial x}+\triangle y\cdot\frac{\partial f}{\partial y}
f(x+△x,y+△y)=f(x,y)+△x⋅∂x∂f+△y⋅∂y∂f
我们设向量
m
→
\overrightarrow{m}
m 邻域内的另一个向量
m
′
→
\overrightarrow{m'}
m′ ,则有
m
′
→
=
(
m
x
+
△
x
)
x
→
+
(
m
y
+
△
y
)
y
→
\overrightarrow{m'}=(m_x+\triangle x)\overrightarrow{x}+(m_y+\triangle y)\overrightarrow{y}
m′=(mx+△x)x+(my+△y)y ,其经过
f
f
f 映射到向量
n
′
→
\overrightarrow{n'}
n′
将函数
f
f
f 拆分成两个函数:
U
,
V
U,V
U,V ,即有:
U
(
m
x
,
m
y
)
=
n
u
V
(
m
x
,
m
y
)
=
n
v
U(m_x,m_y)=n_u\\ V(m_x,m_y)=n_v
U(mx,my)=nuV(mx,my)=nv
这个拆分说明了向量
n
→
\overrightarrow{n}
n 的u和v坐标,和x,y都有关,而不是仅仅是简单的u和x对应,v和y对应。
对U,V做线性的泰勒展开:
U
(
m
x
+
△
x
x
→
,
m
y
+
△
y
y
→
)
=
U
(
m
x
,
m
y
)
+
△
x
x
→
⋅
∂
U
∂
x
+
△
y
y
→
⋅
∂
U
∂
y
=
n
u
+
△
x
x
→
⋅
∂
U
∂
x
+
△
y
y
→
⋅
∂
U
∂
y
U(m_x+\triangle x\overrightarrow{x},m_y+\triangle y\overrightarrow{y})=U(m_x,m_y)+\triangle x\overrightarrow{x}\cdot\frac{\partial U}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial U}{\partial y}=n_u+\triangle x\overrightarrow{x}\cdot\frac{\partial U}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial U}{\partial y}
U(mx+△xx,my+△yy)=U(mx,my)+△xx⋅∂x∂U+△yy⋅∂y∂U=nu+△xx⋅∂x∂U+△yy⋅∂y∂U
V ( m x + △ x x → , m y + △ y y → ) = V ( m x , m y ) + △ x x → ⋅ ∂ V ∂ x + △ y y → ⋅ ∂ V ∂ y = n v + △ x x → ⋅ ∂ V ∂ x + △ y y → ⋅ ∂ V ∂ y V(m_x+\triangle x\overrightarrow{x},m_y+\triangle y\overrightarrow{y})=V(m_x,m_y)+\triangle x\overrightarrow{x}\cdot\frac{\partial V}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial V}{\partial y}=n_v +\triangle x\overrightarrow{x}\cdot\frac{\partial V}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial V}{\partial y} V(mx+△xx,my+△yy)=V(mx,my)+△xx⋅∂x∂V+△yy⋅∂y∂V=nv+△xx⋅∂x∂V+△yy⋅∂y∂V
写成矩阵的形式:
( U ( m x + △ x x → , m y + △ y y → ) V ( m x + △ x x → , m y + △ y y → ) ) = ( U ( m x , m y ) V ( m x , m y ) ) + ( △ x x → ⋅ ∂ U ∂ x + △ y y → ⋅ ∂ U ∂ y △ x x → ⋅ ∂ V ∂ x + △ y y → ⋅ ∂ V ∂ y ) \left( \begin{matrix} U(m_x+\triangle x\overrightarrow{x},m_y+\triangle y\overrightarrow{y})\\ V(m_x+\triangle x\overrightarrow{x},m_y+\triangle y\overrightarrow{y}) \end{matrix} \right)= \left( \begin{matrix} U(m_x,m_y)\\ V(m_x,m_y) \end{matrix} \right) + \left( \begin{matrix} \triangle x\overrightarrow{x}\cdot\frac{\partial U}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial U}{\partial y}\\ \triangle x\overrightarrow{x}\cdot\frac{\partial V}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial V}{\partial y} \end{matrix} \right) (U(mx+△xx,my+△yy)V(mx+△xx,my+△yy))=(U(mx,my)V(mx,my))+(△xx⋅∂x∂U+△yy⋅∂y∂U△xx⋅∂x∂V+△yy⋅∂y∂V)
即:
n
’
→
=
n
→
+
(
∂
U
∂
x
∂
U
∂
y
∂
V
∂
x
∂
V
∂
y
)
(
△
x
x
→
△
y
y
→
)
\overrightarrow{n’}=\overrightarrow{n}+ \left( \begin{matrix} \frac{\partial U}{\partial x}&\frac{\partial U}{\partial y}\\ \frac{\partial V}{\partial x}&\frac{\partial V}{\partial y} \end{matrix} \right) \left( \begin{matrix} \triangle x\overrightarrow{x}\\ \triangle y\overrightarrow{y} \end{matrix} \right)
n’=n+(∂x∂U∂x∂V∂y∂U∂y∂V)(△xx△yy)
我们将矩阵
(
∂
U
∂
x
∂
U
∂
y
∂
V
∂
x
∂
V
∂
y
)
\left( \begin{matrix} \frac{\partial U}{\partial x}&\frac{\partial U}{\partial y}\\ \frac{\partial V}{\partial x}&\frac{\partial V}{\partial y} \end{matrix} \right)
(∂x∂U∂x∂V∂y∂U∂y∂V) 记作雅可比矩阵
J
J
J
显然,我们由如下的结论:
( △ u u → △ v v → ) = J ⋅ ( △ x x → △ y y → ) \left( \begin{matrix} \triangle u\overrightarrow{u}\\ \triangle v\overrightarrow{v} \end{matrix} \right)= J\cdot \left( \begin{matrix} \triangle x\overrightarrow{x}\\ \triangle y\overrightarrow{y} \end{matrix} \right) (△uu△vv)=J⋅(△xx△yy)
雅可比矩阵联系起了不同微元之间的关系。
因
为
:
(
△
u
u
→
△
v
v
→
)
=
(
△
x
x
→
⋅
∂
U
∂
x
+
△
y
y
→
⋅
∂
U
∂
y
△
x
x
→
⋅
∂
V
∂
x
+
△
y
y
→
⋅
∂
V
∂
y
)
因为:\left( \begin{matrix} \triangle u\overrightarrow{u}\\ \triangle v\overrightarrow{v} \end{matrix} \right)= \left( \begin{matrix} \triangle x\overrightarrow{x}\cdot\frac{\partial U}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial U}{\partial y}\\ \triangle x\overrightarrow{x}\cdot\frac{\partial V}{\partial x}+\triangle y\overrightarrow{y}\cdot\frac{\partial V}{\partial y} \end{matrix} \right)
因为:(△uu△vv)=(△xx⋅∂x∂U+△yy⋅∂y∂U△xx⋅∂x∂V+△yy⋅∂y∂V)
我们将矩阵上下叉积:
△
u
△
v
=
∂
u
∂
x
∂
v
∂
y
△
x
△
y
−
∂
u
∂
y
∂
v
∂
x
△
x
△
y
=
∣
∂
U
∂
x
∂
U
∂
y
∂
V
∂
x
∂
V
∂
y
∣
△
x
△
y
\triangle u\triangle v = \frac{\partial u}{\partial x}\frac{\partial v}{\partial y}\triangle x\triangle y-\frac{\partial u}{\partial y}\frac{\partial v}{\partial x}\triangle x\triangle y= \left| \begin{matrix} \frac{\partial U}{\partial x}&\frac{\partial U}{\partial y}\\ \frac{\partial V}{\partial x}&\frac{\partial V}{\partial y} \end{matrix} \right|\triangle x\triangle y
△u△v=∂x∂u∂y∂v△x△y−∂y∂u∂x∂v△x△y=∣∣∣∣∣∂x∂U∂x∂V∂y∂U∂y∂V∣∣∣∣∣△x△y
可见,雅可比矩阵的行列式
∣
J
∣
|J|
∣J∣ 就是二维面积微元的比例。
一般的,对于三维,这个结论依然成立,雅可比行列式是体积微元的比例