本节主要说明链式法则,即多维空间中,可导函数的复合也是可导的。Theorem 7.1是本节的主要定理。Corollary 7.2说明 C r C^r Cr类函数的复合也是 C r C^r Cr类。Theorem 7.3是多维空间的均值定理。Theorem 7.4说明多维空间下反函数定理也成立,但是反函数的导数变为原函数导数的逆矩阵,由此可知,反函数可导需要原函数导数矩阵非奇异。
####Exercises
Exercise 1. Let
f
:
R
3
→
R
2
f:\mathbf{R}^3\to\mathbf{R}^2
f:R3→R2 satisfy the condition
f
(
0
)
=
(
1
,
2
)
f(\mathbf{0})=(1,2)
f(0)=(1,2) and
D
f
(
0
)
=
[
1
2
3
0
0
1
]
Df(\mathbf{0})=\begin{bmatrix}1&2&3\\0&0&1\end{bmatrix}
Df(0)=[102031]
Let
g
:
R
2
→
R
2
g:\mathbf{R}^2\to\mathbf{R}^2
g:R2→R2 be defined by the equation
g
(
x
,
y
)
=
(
x
+
2
y
+
1
,
3
x
y
)
g(x,y)=(x+2y+1,3xy)
g(x,y)=(x+2y+1,3xy)
Find
D
(
g
∘
f
)
(
0
)
D(g\circ f)(\mathbf{0})
D(g∘f)(0).
Solution: We first have
D
g
(
x
,
y
)
=
[
1
2
3
y
3
x
]
,
D
g
(
1
,
2
)
=
[
1
2
6
3
]
Dg(x,y)=\begin{bmatrix}1&2\\3y&3x\end{bmatrix},Dg(1,2)=\begin{bmatrix}1&2\\6&3\end{bmatrix}
Dg(x,y)=[13y23x],Dg(1,2)=[1623]
then by the chain rule
D
(
g
∘
f
)
(
0
)
=
D
g
(
f
(
0
)
)
⋅
D
f
(
0
)
=
D
g
(
1
,
2
)
⋅
D
f
(
0
)
=
[
1
2
6
3
]
[
1
2
3
0
0
1
]
=
[
1
2
5
6
12
21
]
\begin{aligned}D(g\circ f)(\mathbf{0})&=Dg(f(\mathbf{0})){\cdot}Df(\mathbf{0})=Dg(1,2){\cdot}Df(\mathbf{0})\\&=\begin{bmatrix}1&2\\6&3\end{bmatrix}\begin{bmatrix}1&2&3\\0&0&1\end{bmatrix}=\begin{bmatrix}1&2&5\\6&12&21\end{bmatrix}\end{aligned}
D(g∘f)(0)=Dg(f(0))⋅Df(0)=Dg(1,2)⋅Df(0)=[1623][102031]=[16212521]
Exercise 2. Let
f
:
R
2
→
R
3
f:\mathbf{R}^2\to\mathbf{R}^3
f:R2→R3 and
g
:
R
3
→
R
2
g:\mathbf{R}^3\to\mathbf{R}^2
g:R3→R2 be given by the equations
f
(
x
)
=
(
e
2
x
1
+
x
2
,
3
x
2
−
cos
x
1
,
x
1
2
+
x
2
+
2
)
f(x)=(e^{2x_1+x_2},3x_2-\cos{x_1},x_1^2+x_2+2)
f(x)=(e2x1+x2,3x2−cosx1,x12+x2+2)
g
(
y
)
=
(
3
y
1
+
2
y
2
+
y
3
2
,
y
1
2
−
y
3
+
1
)
g(y)=(3y_1+2y_2+y_3^2,y_1^2-y_3+1)
g(y)=(3y1+2y2+y32,y12−y3+1)
( a ) If
F
(
x
)
=
g
(
f
(
x
)
)
F(x)=g(f(x))
F(x)=g(f(x)), find
D
F
(
0
)
DF(\mathbf{0})
DF(0).
( b ) If
G
(
y
)
=
f
(
g
(
y
)
)
G(y)=f(g(y))
G(y)=f(g(y)), find
D
G
(
0
)
DG(\mathbf{0})
DG(0).
Solution: We first have
f
(
0
,
0
)
=
(
1
,
−
1
,
2
)
,
g
(
0
,
0
,
0
)
=
(
0
,
1
)
f(0,0)=(1,-1,2),g(0,0,0)=(0,1)
f(0,0)=(1,−1,2),g(0,0,0)=(0,1) and
D
f
(
x
1
,
x
2
)
=
[
2
e
2
x
1
+
x
2
e
2
x
1
+
x
2
sin
x
1
3
2
x
1
1
]
,
D
g
(
y
1
,
y
2
,
y
3
)
=
[
3
2
2
y
3
2
y
1
0
−
1
]
Df(x_1,x_2 )=\begin{bmatrix}2e^{2x_1+x_2}&e^{2x_1+x_2}\\{\sin}x_1&3\\2x_1&1\end{bmatrix},Dg(y_1,y_2,y_3 )=\begin{bmatrix}3&2&2y_3\\2y_1&0&-1\end{bmatrix}
Df(x1,x2)=⎣⎡2e2x1+x2sinx12x1e2x1+x231⎦⎤,Dg(y1,y2,y3)=[32y1202y3−1]
( a ) By the chain rule we have
D
F
(
0
)
=
D
g
(
f
(
0
)
)
⋅
D
f
(
0
)
=
D
g
(
1
,
−
1
,
2
)
⋅
D
f
(
0
,
0
)
=
[
3
2
4
2
0
−
1
]
[
2
1
0
3
0
1
]
=
[
6
13
4
1
]
\begin{aligned}DF(\mathbf{0})&=Dg(f(\mathbf{0})){\cdot}Df(\mathbf{0})=Dg(1,-1,2){\cdot}Df(0,0)\\&=\begin{bmatrix}3&2&4\\2&0&-1\end{bmatrix}\begin{bmatrix}2&1\\0&3\\0&1\end{bmatrix}=\begin{bmatrix}6&13\\4&1\end{bmatrix}\end{aligned}
DF(0)=Dg(f(0))⋅Df(0)=Dg(1,−1,2)⋅Df(0,0)=[32204−1]⎣⎡200131⎦⎤=[64131]
( b ) By the chain rule we have
D
G
(
0
)
=
D
f
(
g
(
0
)
)
⋅
D
g
(
0
)
=
D
f
(
0
,
1
)
⋅
D
g
(
0
,
0
,
0
)
=
[
2
e
e
0
3
0
1
]
[
3
2
0
0
0
−
1
]
=
[
6
e
4
e
−
e
0
0
−
3
0
0
−
1
]
\begin{aligned}DG(\mathbf{0})&=Df(g(\mathbf{0})){\cdot}Dg(\mathbf{0})=Df(0,1){\cdot}Dg(0,0,0)\\&=\begin{bmatrix}2e&e\\0&3\\0&1\end{bmatrix}\begin{bmatrix}3&2&0\\0&0&-1\end{bmatrix}=\begin{bmatrix}6e&4e&-e\\0&0&-3\\0&0&-1\end{bmatrix}\end{aligned}
DG(0)=Df(g(0))⋅Dg(0)=Df(0,1)⋅Dg(0,0,0)=⎣⎡2e00e31⎦⎤[30200−1]=⎣⎡6e004e00−e−3−1⎦⎤
Exercise 3. Let
f
:
R
3
→
R
f:\mathbf{R}^3\to\mathbf{R}
f:R3→R and
g
:
R
2
→
R
g:\mathbf{R}^2\to\mathbf{R}
g:R2→R be differentiable. Let
F
:
R
2
→
R
F:\mathbf{R}^2\to\mathbf{R}
F:R2→R be defined by the equation
F
(
x
,
y
)
=
f
(
x
,
y
,
g
(
x
,
y
)
)
F(x,y)=f(x,y,g(x,y))
F(x,y)=f(x,y,g(x,y))
( a ) Find
D
F
DF
DF in terms of the partials of
f
f
f and
g
g
g.
( b ) If
F
(
x
,
y
)
=
0
F(x,y)=0
F(x,y)=0 for all
(
x
,
y
)
(x,y)
(x,y), find
D
1
g
D_1g
D1g and
D
2
g
D_2g
D2g in terms of the partials of
f
f
f.
Solution:
( a ) We have
D
F
(
x
,
y
)
=
[
D
1
F
(
x
,
y
)
D
2
F
(
x
,
y
)
]
DF(x,y)=\begin{bmatrix}D_1 F(x,y)&D_2 F(x,y) \end{bmatrix}
DF(x,y)=[D1F(x,y)D2F(x,y)], and
D
1
F
(
x
,
y
)
=
lim
t
→
0
f
(
x
+
t
,
y
,
g
(
x
+
t
,
y
)
)
−
f
(
x
,
y
,
g
(
x
,
y
)
)
t
=
lim
t
→
0
f
(
x
+
t
,
y
,
g
(
x
+
t
,
y
)
)
−
f
(
x
,
y
,
g
(
x
+
t
,
y
)
)
t
+
lim
t
→
0
f
(
x
,
y
,
g
(
x
+
t
,
y
)
)
−
f
(
x
,
y
,
g
(
x
,
y
)
)
t
=
D
1
f
(
x
,
y
,
g
(
x
,
y
)
)
+
D
3
f
(
x
,
y
,
g
(
x
,
y
)
)
⋅
D
1
g
(
x
,
y
)
\begin{aligned}D_1 F(x,y)&=\lim_{t→0}\frac{f(x+t,y,g(x+t,y))-f(x,y,g(x,y))}{t}\\&=\lim_{t→0}\frac{f(x+t,y,g(x+t,y))-f(x,y,g(x+t,y))}{t}+\lim_{t→0}\frac{f(x,y,g(x+t,y))-f(x,y,g(x,y))}{t}\\&=D_1 f(x,y,g(x,y))+D_3 f(x,y,g(x,y)){\cdot}D_1 g(x,y)\end{aligned}
D1F(x,y)=t→0limtf(x+t,y,g(x+t,y))−f(x,y,g(x,y))=t→0limtf(x+t,y,g(x+t,y))−f(x,y,g(x+t,y))+t→0limtf(x,y,g(x+t,y))−f(x,y,g(x,y))=D1f(x,y,g(x,y))+D3f(x,y,g(x,y))⋅D1g(x,y)
D
2
F
(
x
,
y
)
=
D
2
f
(
x
,
y
,
g
(
x
,
y
)
)
+
D
3
f
(
x
,
y
,
g
(
x
,
y
)
)
⋅
D
2
g
(
x
,
y
)
D_2 F(x,y)=D_2 f(x,y,g(x,y))+D_3 f(x,y,g(x,y)){\cdot}D_2 g(x,y)
D2F(x,y)=D2f(x,y,g(x,y))+D3f(x,y,g(x,y))⋅D2g(x,y)
( b ) In this case
D
1
F
(
x
,
y
)
=
D
2
F
(
x
,
y
)
=
0
D_1 F(x,y)=D_2 F(x,y)=0
D1F(x,y)=D2F(x,y)=0, thus
D
1
g
(
x
,
y
)
=
[
D
3
f
(
x
,
y
,
g
(
x
,
y
)
)
]
−
1
⋅
(
−
D
1
f
(
x
,
y
,
g
(
x
,
y
)
)
)
D_1 g(x,y)=[D_3 f(x,y,g(x,y))]^{-1}{\cdot}(-D_1 f(x,y,g(x,y)))
D1g(x,y)=[D3f(x,y,g(x,y))]−1⋅(−D1f(x,y,g(x,y)))
D
2
g
(
x
,
y
)
=
[
D
3
f
(
x
,
y
,
g
(
x
,
y
)
)
]
−
1
⋅
(
−
D
2
f
(
x
,
y
,
g
(
x
,
y
)
)
)
D_2 g(x,y)=[D_3 f(x,y,g(x,y))]^{-1}{\cdot}(-D_2 f(x,y,g(x,y)))
D2g(x,y)=[D3f(x,y,g(x,y))]−1⋅(−D2f(x,y,g(x,y)))
Exercise 4. Let
g
:
R
2
→
R
2
g:\mathbf{R}^2\to\mathbf{R}^2
g:R2→R2 be defined by the equation
g
(
x
,
y
)
=
(
x
,
y
+
x
2
)
g(x,y)=(x,y+x^2)
g(x,y)=(x,y+x2). Let
f
:
R
2
→
R
f:\mathbf{R}^2\to\mathbf{R}
f:R2→R be the function defined in Example 3 of Section 5. Let
h
=
f
∘
g
h=f\circ g
h=f∘g. Show that the directional derivatives of
f
f
f and
g
g
g exist everywhere, but there is a
u
≠
0
\mathbf{u}\neq\mathbf{0}
u=0 for which
h
′
(
0
;
u
)
h'(\mathbf{0};\mathbf{u})
h′(0;u) does not exist.
Solution: The directional derivatives of f exists everywhere is clear from the text. Since
D
g
(
x
,
y
)
Dg(x,y)
Dg(x,y) exists everywhere and
D
g
(
x
,
y
)
=
[
1
2
x
0
1
]
Dg(x,y)=\begin{bmatrix}1&2x\\0&1\end{bmatrix}
Dg(x,y)=[102x1], the directional derivatives of g exists everywhere since
g
′
(
a
;
u
)
=
D
g
(
a
)
⋅
u
g'(\mathbf{a};\mathbf{u})=Dg(\mathbf{a}){\cdot}\mathbf{u}
g′(a;u)=Dg(a)⋅u
To show
∃
u
≠
0
\exists\mathbf{u}\neq 0
∃u=0, s.t.
h
′
(
0
;
u
)
h'(\mathbf{0};\mathbf{u})
h′(0;u) does not exist, first notice
h
(
0
)
=
f
(
g
(
0
)
)
=
f
(
0
,
0
)
=
0
h(\mathbf{0})=f(g(\mathbf{0}))=f(0,0)=0
h(0)=f(g(0))=f(0,0)=0, let
u
=
(
1
,
0
)
T
\mathbf{u}=(1,0)^T
u=(1,0)T, then when
t
→
0
,
t
≠
0
t→0,t\neq 0
t→0,t=0, we have
h
(
0
+
t
u
)
−
h
(
0
)
t
=
h
(
t
,
0
)
t
=
f
(
t
,
t
2
)
t
=
1
t
t
2
t
2
(
t
4
+
t
4
)
=
1
2
t
\frac{h(\mathbf{0}+t\mathbf{u})-h(\mathbf{0})}{t}=\frac{h(t,0)}{t}=\frac{f(t,t^2)}{t}=\frac{1}{t} \frac{t^2 t^2}{(t^4+t^4)}=\frac{1}{2t}
th(0+tu)−h(0)=th(t,0)=tf(t,t2)=t1(t4+t4)t2t2=2t1
thus the limit
h
′
(
0
;
u
)
=
lim
t
→
0
h
(
0
+
t
u
)
−
h
(
0
)
t
=
lim
t
→
0
1
2
t
h'(\mathbf{0};\mathbf{u})=\lim_{t→0}\frac{h(\mathbf{0}+t\mathbf{u})-h(\mathbf{0})}{t}=\lim_{t→0}\frac{1}{2t}
h′(0;u)=t→0limth(0+tu)−h(0)=t→0lim2t1
does not exist.