abstract
- 高阶偏导数
- 二元复合函数高阶偏导
- 全微分形式不变性
高阶偏导数
二元函数二阶偏导数
- 设
y
=
f
(
x
,
y
)
y=f(x,y)
y=f(x,y)在区域
D
D
D内的偏导数存在,则
f
x
(
x
,
y
)
f_x(x,y)
fx(x,y),
f
y
(
x
,
y
)
f_y(x,y)
fy(x,y)在区域
D
D
D内仍然是关于
x
,
y
x,y
x,y的函数
- 若这两个函数的偏导数也存在,则称它们是 z = f ( x , y ) z=f(x,y) z=f(x,y)的二阶偏导数
- 按照对变量求导次序的不同,这样的二阶偏导数由4个
- ∂ ∂ x ( ∂ z ∂ x ) \frac{\partial}{\partial{x}}{(\frac{\partial{z}}{\partial{x}})} ∂x∂(∂x∂z)= ∂ 2 z ∂ x 2 \frac{\partial^2{z}}{\partial{x^2}} ∂x2∂2z= f x x ( x , y ) f_{xx}(x,y) fxx(x,y)
- ∂ ∂ y ( ∂ z ∂ y ) \frac{\partial}{\partial{y}}{(\frac{\partial{z}}{\partial{y}})} ∂y∂(∂y∂z)= ∂ 2 z ∂ y 2 \frac{\partial^2{z}}{\partial{y^2}} ∂y2∂2z= f y y ( x , y ) f_{yy}(x,y) fyy(x,y)
- ∂ ∂ y ( ∂ z ∂ x ) \frac{\partial}{\partial{y}}{(\frac{\partial{z}}{\partial{x}})} ∂y∂(∂x∂z)= ∂ 2 z ∂ x ∂ y \frac{\partial^2{z}}{\partial{x}\partial{y}} ∂x∂y∂2z= f x y ( x , y ) f_{xy}(x,y) fxy(x,y)
- ∂ ∂ x ( ∂ z ∂ y ) \frac{\partial}{\partial{x}}{(\frac{\partial{z}}{\partial{y}})} ∂x∂(∂y∂z)= ∂ 2 z ∂ y ∂ x \frac{\partial^2{z}}{\partial{y}\partial{x}} ∂y∂x∂2z= f y x ( x , y ) f_{yx}(x,y) fyx(x,y)
- 其中(3),(4)称为混合偏导数
- 类似地可以定义 n n n阶偏导数, n ⩾ 2 n\geqslant{2} n⩾2称为高阶偏导数
- 求 n n n阶偏导前要先求 n − 1 n-1 n−1阶导,即从一阶偏导求导 n n n阶偏导
混合偏导数相等定理
- 若二元函数 f ( x , y ) f(x,y) f(x,y)的两个混合偏导数在区域 D D D上连续,则它们必然相等
- n n n元函数也有类似结论:高阶混合偏导数在偏导数连续的条件下也与求导的次序无关
例
-
求 z = x 2 y e y z=x^2ye^{y} z=x2yey的所有二阶偏导数
-
先求一阶偏导:
- z x z_{x} zx= 2 y e y x 2ye^{y}x 2yeyx
- z y z_{y} zy= x 2 ( e y + y e y ) x^2(e^{y}+ye^{y}) x2(ey+yey)= x 2 ( 1 + y ) e y x^2(1+y)e^{y} x2(1+y)ey
-
再求二阶偏导
- z x x z_{xx} zxx= 2 y e y 2ye^{y} 2yey
- z x y z_{xy} zxy= 2 x ( e y + y e y ) 2x(e^{y}+ye^{y}) 2x(ey+yey)= 2 x ( 1 + y ) e y 2x(1+y)e^{y} 2x(1+y)ey
- z y x z_{yx} zyx= 2 ( 1 + y ) e y x 2(1+y)e^{y}x 2(1+y)eyx
- z y y z_{yy} zyy= x 2 ( e y + ( 1 + y ) e y ) x^2(e^{y}+(1+y)e^{y}) x2(ey+(1+y)ey)= x 2 e y ( 2 + y ) x^2e^{y}(2+y) x2ey(2+y)
-
可以发现 z x y = z y x z_{xy}=z_{yx} zxy=zyx
-
-
事实上,对混合偏导而言,自变量顺序的调整不影响偏导结果
二元复合函数高阶偏导数的计算
-
主要讨论二元函数的二阶偏导数
-
设 z = f ( x , y ) z=f(x,y) z=f(x,y)在区域D内有偏导数
-
∂ z ∂ x = f x ′ ( x , y ) ∂ z ∂ x = f y ′ ( x , y ) \frac{\partial{z}}{\partial{x}}=f'_x(x,y) \\ \frac{\partial{z}}{\partial{x}}=f'_y(x,y) ∂x∂z=fx′(x,y)∂x∂z=fy′(x,y)
-
∂ 2 z ∂ x 2 = ∂ ∂ x ( ∂ z ∂ x ) = f x x ′ ′ ( x , y ) ∂ 2 z ∂ y 2 = ∂ ∂ y ( ∂ z ∂ y ) = f y y ′ ′ ( x , y ) ∂ 2 z ∂ y ∂ x = ∂ ∂ x ( ∂ z ∂ y ) = f y x ′ ′ ( x , y ) ∂ 2 z ∂ x ∂ y = ∂ ∂ y ( ∂ z ∂ x ) = f x y ′ ′ ( x , y ) \frac{\partial^2{z}}{\partial{x^2}} =\frac{\partial}{\partial{x}}\left(\frac{\partial{z}}{\partial{x}}\right) =f''_{xx}(x,y) \\ \frac{\partial^2{z}}{\partial{y^2}} =\frac{\partial}{\partial{y}}\left(\frac{\partial{z}}{\partial{y}}\right) =f''_{yy}(x,y) \\ \frac{\partial^2{z}}{\partial{y}\partial{x}} =\frac{\partial}{\partial{x}} \left(\frac{\partial{z}}{\partial{y}}\right) =f''_{yx}(x,y) \\ \frac{\partial^2{z}}{\partial{x}\partial{y}} =\frac{\partial}{\partial{y}} \left(\frac{\partial{z}}{\partial{x}}\right) =f''_{xy}(x,y) ∂x2∂2z=∂x∂(∂x∂z)=fxx′′(x,y)∂y2∂2z=∂y∂(∂y∂z)=fyy′′(x,y)∂y∂x∂2z=∂x∂(∂y∂z)=fyx′′(x,y)∂x∂y∂2z=∂y∂(∂x∂z)=fxy′′(x,y)
-
引入记号
-
设 z = f ( u , v ) z=f(u,v) z=f(u,v), u = u ( x , y ) u=u(x,y) u=u(x,y), v = v ( x , y ) v=v(x,y) v=v(x,y)
-
引入记号: f r 1 f_{r_1} fr1,其中 r 1 ∈ N + r_1\in{\mathbb{N}_+} r1∈N+表示对第 r 1 r_1 r1个变量求(偏)导
- 类似的,可以叠加,例如 f r 1 r 2 f_{r_1r_2} fr1r2表示先对第 r 1 r_1 r1个变量求导,在对第 r 2 r_2 r2个变量求导, r 1 , r 2 r_1,r_2 r1,r2独立
-
f
r
1
r
2
f_{r_1r_2}
fr1r2是
f
r
1
r
2
′
′
f_{r_1r_2}''
fr1r2′′简写,上角的
′
′
''
′′表示二阶导,其可以从下标看出阶数,因此有时不写
- f 1 f_{1} f1= ∂ f ( u , v ) ∂ u \frac{\partial{f(u,v)}}{\partial{u}} ∂u∂f(u,v)= f u f_{u} fu; f 2 = ∂ f ( u , v ) ∂ v f_2=\frac{\partial{f(u,v)}}{\partial{v}} f2=∂v∂f(u,v)= f v f_{v} fv
- f 12 f_{12} f12= ∂ f ( u , v ) ∂ u ∂ v \frac{\partial{f(u,v)}}{\partial{u}\partial{v}} ∂u∂v∂f(u,v)= f u v f_{uv} fuv: f 21 f_{21} f21= ∂ f ( u , v ) ∂ v ∂ u \frac{\partial{f(u,v)}}{\partial{v}\partial{u}} ∂v∂u∂f(u,v)= f v u f_{vu} fvu;
- f 11 f_{11} f11= ∂ ∂ u ( ∂ f ( u , v ) ∂ u ) \frac{\partial}{\partial{u}} (\frac{\partial{f(u,v)}}{\partial{u}}) ∂u∂(∂u∂f(u,v))= ∂ 2 ∂ u 2 f ( u , v ) \frac{\partial^2}{\partial{u^2}}{{f(u,v)}} ∂u2∂2f(u,v)= f u u f_{uu} fuu; f 22 f_{22} f22= f v v f_{vv} fvv
- 此外, f 1 x f_{1x} f1x= ( f 1 ) x ′ (f_{1})_{x}' (f1)x′= ∂ ∂ x f 1 \frac{\partial}{\partial{x}}f_1 ∂x∂f1,类似的有 f 1 y f_{1y} f1y, f 2 x f_{2x} f2x, f 2 y f_{2y} f2y
-
Notes
-
有混合偏导次序无关定理, f 12 f_{12} f12= f 21 f_{21} f21
-
需要明确一个事实是 f r 1 f_{r_1} fr1仍然是 u , v u,v u,v为中间变量的复合函数,若要对 f r 1 f_{r_1} fr1求对自变量 x r 1 x_{r_1} xr1的导数,也要以复合函数偏导法则计算
- 二元函数的偏导即便有些情形下只剩一个变量,也可以用二元的方法处理,
- 二元函数求导兼容一元函数求导,但是一元求导不能直接计算二元求导问题
-
f 1 x f_{1x} f1x= ( f 1 ( u , v ) ) x ′ (f_{1}(u,v))_{x}' (f1(u,v))x′= ∂ ∂ x f 1 \frac{\partial}{\partial{x}}f_1 ∂x∂f1= ∂ ∂ x [ ∂ f ( u , v ) ∂ u ] \frac{\partial}{\partial{x}} [\frac{\partial{f(u,v)}}{\partial{u}}] ∂x∂[∂u∂f(u,v)]= ∂ ∂ u [ ∂ f ( u , v ) ∂ u ] \frac{\partial}{\partial{u}} [\frac{\partial{f(u,v)}}{\partial{u}}] ∂u∂[∂u∂f(u,v)] ∂ u ∂ x \frac{\partial{u}}{\partial{x}} ∂x∂u+ ∂ ∂ v [ ∂ f ( u , v ) ∂ u ] \frac{\partial}{\partial{v}} [\frac{\partial{f(u,v)}}{\partial{u}}] ∂v∂[∂u∂f(u,v)] ∂ v ∂ x \frac{\partial{v}}{\partial{x}} ∂x∂v= ∂ 2 f ( u , v ) ∂ u 2 \frac{\partial^2{f(u,v)}}{\partial{u^2}} ∂u2∂2f(u,v) ∂ u ∂ x \frac{\partial{u}}{\partial{x}} ∂x∂u+ ∂ 2 f ( u , v ) ∂ u ∂ v \frac{\partial^2{f(u,v)}}{\partial{u}\partial{v}} ∂u∂v∂2f(u,v) ∂ v ∂ x \frac{\partial{v}}{\partial{x}} ∂x∂v
- 如果利用引入的记号,上式可以表示为 f 1 x f_{1x} f1x= f 11 u x + f 12 v x f_{11}u_{x}+f_{12}v_{x} f11ux+f12vx,显然简洁的多
-
这类记法一般应用在复合函数中间变量上,简化最值得简化的部分,尽管按照上面的规则, u 1 = u x u_1=u_{x} u1=ux; u 2 = u y u_2=u_{y} u2=uy,但一般不这么做
-
-
例如,取 u = x 2 y u=x^2y u=x2y, v = y x v=\frac{y}{x} v=xy,则
-
u x u_{x} ux= 2 x y 2xy 2xy; v x v_{x} vx= − y 1 x 2 -y\frac{1}{x^2} −yx21= − y x − 2 -yx^{-2} −yx−2
-
z x z_{x} zx= f 1 ( 2 x y ) + f 2 ( − y 1 x 2 ) f_1{(2xy)}+f_2(-y\frac{1}{x^2}) f1(2xy)+f2(−yx21); z y z_{y} zy= f 1 ( x 2 ) + f 2 ( 1 x ) f_1(x^2)+f_2(\frac{1}{x}) f1(x2)+f2(x1)
-
z x x z_{xx} zxx= ( 2 y x f 1 ) x ′ (2yxf_1)_{x}' (2yxf1)x′- ( y 1 x 2 f 2 ) x ′ (y\frac{1}{x^2}f_2)_{x}' (yx21f2)x′= 2 y ( f 1 + x f 1 x ) 2y(f_{1}+xf_{1x}) 2y(f1+xf1x)- y ( − 2 x − 3 f 2 + x − 2 f 2 x ) y(-2x^{-3}f_{2}+x^{-2}f_{2x}) y(−2x−3f2+x−2f2x)
- = 2 y f 1 + 2 y x ( f 11 u x + f 12 v x ) 2yf_1+2yx(f_{11}u_{x}+f_{12}v_x) 2yf1+2yx(f11ux+f12vx)+ ( 2 x − 3 y f 2 ) (2x^{-3}yf_2) (2x−3yf2)- y x − 2 ( f 21 u x + f 22 v x ) yx^{-2}(f_{21}u_x+f_{22}v_x) yx−2(f21ux+f22vx)
- = 2 y f 1 + 2 x y f 11 2 x y + 2 x y f 12 ( − y 1 x 2 ) 2yf_1+2xyf_{11}2xy+2xyf_{12}(-y\frac{1}{x^2}) 2yf1+2xyf112xy+2xyf12(−yx21)+ 2 x − 3 y f 2 2x^{-3}{y}f_2 2x−3yf2- y x − 2 f 21 2 x y yx^{-2}f_{21}2xy yx−2f212xy- y x − 2 f 22 ( − y x − 2 ) yx^{-2}f_{22}(-yx^{-2}) yx−2f22(−yx−2)
- = 2 y f 1 + 2 x − 3 y f 2 + 4 x 2 y 2 f 11 2yf_1+2x^{-3}{y}f_2+4x^2y^2f_{11} 2yf1+2x−3yf2+4x2y2f11- 4 y 2 x f 12 4\frac{y^2}{x}f_{12} 4xy2f12+ y 2 x − 4 f 22 y^2x^{-4}f_{22} y2x−4f22
-
Note,由混合偏导次序无关定理, f 12 f_{12} f12= f 21 f_{21} f21
-
混合偏导与次序无关🎈
- f x y ′ ′ ( x , y ) = f y x ′ ′ ( x , y ) f''_{xy}(x,y)=f''_{yx}(x,y) fxy′′(x,y)=fyx′′(x,y)
记号补充
-
容易发现,上述的写法效率不高,大量的 ∂ \partial ∂符号使得公式变得冗长起来
-
约定如下写法,来化简中间变量微分的书写
-
二阶偏导
-
- 其中m1表示第一个中间变量,m2表示第二个中间变量…
- m i = m i ( x , y ) m_i=m_{i}(x,y) mi=mi(x,y), i = 1 , 2 , ⋯ , n i=1,2,\cdots,n i=1,2,⋯,n
-
∂ ∂ x i ⋅ f k ′ ( m 1 , m 2 ) \frac{\partial}{\partial{x_i}} \cdot f'_k(m_1,m_2) ∂xi∂⋅fk′(m1,m2)= f k 1 ′ ′ ⋅ ∂ m 1 ∂ x i f''_{k1}\cdot\frac{\partial{m_1}}{\partial{x_i}} fk1′′⋅∂xi∂m1+ f k 2 ′ ′ ⋅ ∂ m 2 ∂ x i f''_{k2}\cdot\frac{\partial{m_2}}{\partial{x_i}} fk2′′⋅∂xi∂m2, k ∈ { 1 , 2 } k\in\set{1,2} k∈{1,2}
-
k
=
1
,
i
=
1
k=1,i=1
k=1,i=1
- ∂ ∂ x f 1 ′ ( m 1 , m 2 ) \frac{\partial{}}{\partial{x}}f'_1(m_1,m_2) ∂x∂f1′(m1,m2)= f 11 ′ ′ ∂ m 1 ∂ x f''_{11}\frac{\partial{m_1}}{\partial{x}} f11′′∂x∂m1+ f 12 ′ ′ ∂ m 2 ∂ x f''_{12}\frac{\partial{m_2}}{\partial{x}} f12′′∂x∂m2
-
k
=
2
,
i
=
2
k=2,i=2
k=2,i=2
- ∂ ∂ y f 1 ′ ( m 1 , m 2 ) \frac{\partial{}}{\partial{y}}f'_1(m_1,m_2) ∂y∂f1′(m1,m2)= f 21 ′ ′ ∂ m 1 ∂ y f''_{21}\frac{\partial{m_1}}{\partial{y}} f21′′∂y∂m1+ f 22 ′ ′ ∂ m 2 ∂ y f''_{22}\frac{\partial{m_2}}{\partial{y}} f22′′∂y∂m2,
-
k
=
1
,
i
=
1
k=1,i=1
k=1,i=1
-
中间变量仅2个(函数 f f f是二元函数时)m m 1 , m 2 m_1,m_2 m1,m2分别通常用 u , v u,v u,v来表示
-
全导数和全微分对比
- 设
z
=
f
(
u
,
v
)
z=f(u,v)
z=f(u,v),
u
=
ϕ
(
t
)
u=\phi(t)
u=ϕ(t),
v
=
ψ
(
t
)
v=\psi(t)
v=ψ(t);则全导数公式为
z
t
z_{t}
zt=
z
u
u
t
+
z
v
v
t
z_{u}u_{t}+z_{v}v_{t}
zuut+zvvt
(1) - 设
z
=
f
(
x
,
y
)
z=f(x,y)
z=f(x,y),全微分公式为
d
z
=
z
x
d
x
+
z
y
d
y
\mathrm{d}z=z_{x}{\mathrm{d}x}+z_{y}\mathrm{d}y
dz=zxdx+zydy
(2)
全微分形式不变性
- 设
z
=
f
(
u
,
v
)
z=f(u,v)
z=f(u,v),其中
u
,
v
u,v
u,v是函数
f
f
f的自变量则:
d
z
\mathrm{d}z
dz=
z
u
d
u
+
z
v
d
v
z_{u}\mathrm{d}{u}+z_{v}\mathrm{d}v
zudu+zvdv
(3)- 若
u
,
v
u,v
u,v又是中间变量,设
u
=
ϕ
(
x
,
y
)
u=\phi(x,y)
u=ϕ(x,y),
v
=
ψ
(
x
,
y
)
v=\psi(x,y)
v=ψ(x,y),且两个函数具有偏导数,则复合函数
z
=
z
(
x
,
y
)
=
f
(
ϕ
(
x
,
y
)
,
ψ
(
x
,
y
)
)
z=z(x,y)=f(\phi(x,y),\psi(x,y))
z=z(x,y)=f(ϕ(x,y),ψ(x,y))的全微分为:
d
z
\mathrm{d}z
dz=
z
x
d
x
+
z
y
d
y
z_{x}\mathrm{d}x+z_{y}\mathrm{d}y
zxdx+zydy
(4)- 而
d
u
\mathrm{d}u
du=
u
x
d
x
+
v
y
d
y
u_{x}\mathrm{d}x+v_{y}\mathrm{d}y
uxdx+vydy
(4-1); d v \mathrm{d}v dv= u x d x + v x d y u_{x}\mathrm{d}x+v_{x}\mathrm{d}y uxdx+vxdy(4-2)
- 而
d
u
\mathrm{d}u
du=
u
x
d
x
+
v
y
d
y
u_{x}\mathrm{d}x+v_{y}\mathrm{d}y
uxdx+vydy
- 利用多元复合函数偏导法则,
z
x
z_{x}
zx=
z
u
u
x
+
z
v
v
x
z_{u}u_{x}+z_{v}v_{x}
zuux+zvvx
(5-1); z y z_y zy= z u u y + z v v y z_{u}u_{y}+z_{v}v_{y} zuuy+zvvy(5-2) - 将(5-1),(5-2)分别代入(4)得 d z \mathrm{d}z dz= ( z u u x + z v v x ) d x (z_{u}u_{x}+z_{v}v_{x})\mathrm{d}x (zuux+zvvx)dx+ ( z u u y + z v v y ) d y (z_{u}u_{y}+z_{v}v_{y})\mathrm{d}y (zuuy+zvvy)dy= z u ( u x d x + u y d y ) z_{u}(u_{x}\mathrm{d}x+u_{y}\mathrm{d}y) zu(uxdx+uydy)+ z v ( v x d x + v y d y ) z_{v}(v_{x}\mathrm{d}x+v_{y}\mathrm{d}y) zv(vxdx+vydy)
- 在代入(4-1,4-2)得: d z \mathrm{d}z dz= z u d u + z v d v z_{u}\mathrm{d}u+z_{v}\mathrm{d}v zudu+zvdv,这和式(3)一致
- 若
u
,
v
u,v
u,v又是中间变量,设
u
=
ϕ
(
x
,
y
)
u=\phi(x,y)
u=ϕ(x,y),
v
=
ψ
(
x
,
y
)
v=\psi(x,y)
v=ψ(x,y),且两个函数具有偏导数,则复合函数
z
=
z
(
x
,
y
)
=
f
(
ϕ
(
x
,
y
)
,
ψ
(
x
,
y
)
)
z=z(x,y)=f(\phi(x,y),\psi(x,y))
z=z(x,y)=f(ϕ(x,y),ψ(x,y))的全微分为:
d
z
\mathrm{d}z
dz=
z
x
d
x
+
z
y
d
y
z_{x}\mathrm{d}x+z_{y}\mathrm{d}y
zxdx+zydy
- 由此可见,无论 u , v u,v u,v是自变量还是中间变量,函数 z = f ( u , v ) z=f(u,v) z=f(u,v)的全微分形式是一样的,称为全微分形式不变性
扫一扫
146
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
