这一节回顾两个重要的概念:紧致集和联通集。由于本书主要讨论
R
n
\mathbf R^n
Rn上的情况,因此紧致和有界闭集是等价的,但本节练习中也提到了非
R
n
\mathbf R^n
Rn上的情况。
Theorem 4.1其实是很多书上紧致集的定义。
Theorem 4.2很简单,
R
\mathbf R
R上闭区间是compact的。
Theorem 4.3说明在
R
n
\mathbf R^n
Rn上compact可以推出closed和bounded,推论是最大最小值定理。
Theorem 4.5是极值定理,说明compact区间上的连续函数的值域也是compact的(连续函数保留compact特性)
Theorem 4.6是
ϵ
\epsilon
ϵ邻域定理,如果
U
U
U是包含紧致集
X
X
X的一个开集,那么可以找到一个
X
X
X的
ϵ
\epsilon
ϵ邻域包含在
U
U
U中。
Theorem 4.7是一致连续定理,compact集上的连续函数是一致连续的。
Theorem 4.9说明在
R
n
\mathbf R^n
Rn上closed和bounded可以推出compact
Theorem 4.10说明
R
\mathbf R
R上闭区间是联通的。
Theorem 4.11是介值定理。
Exercises
Exercise 1. Let
R
+
\mathbf R_{+}
R+ denote the set of positive real numbers.
( a ) Show that the continuous function
f
:
R
+
→
R
f:\mathbf{R_+}\to\mathbf{R}
f:R+→R given by
f
(
x
)
=
1
/
(
1
+
x
)
f(x)=1/(1+x)
f(x)=1/(1+x) is bounded but has neither a maximum value nor a minimum value.
( b ) Show that the continuous function
g
:
R
+
→
R
g:\mathbf{R_+}\to\mathbf{R}
g:R+→R given by
g
(
x
)
=
sin
(
1
/
x
)
g(x)=\sin{(1/x)}
g(x)=sin(1/x) is bounded but does not satisfy the condition of uniform continuity on
R
+
\mathbf{R_+}
R+.
Solution:
(a)
x
>
0
x>0
x>0 means
1
+
x
>
1
1+x>1
1+x>1 and thus
f
(
x
)
<
1
f(x)<1
f(x)<1, also
f
(
x
)
>
0
f(x)>0
f(x)>0 is obvious, thus
0
<
f
(
x
)
<
1
0<f(x)<1
0<f(x)<1, and
f
f
f is bounded. Since
f
(
R
+
)
=
(
0
,
1
)
f(\mathbf{R}_+)=(0,1)
f(R+)=(0,1),
f
f
f has neither maximum value nor minimum value.
(b) It is obvious that
−
1
≤
g
(
x
)
≤
1
-1\leq g(x)\leq 1
−1≤g(x)≤1. To see
g
g
g is not uniform continuous, let
∀
δ
>
0
\forall\delta>0
∀δ>0, choose
k
∈
N
+
k\in\mathbf{N}^+
k∈N+ such that
4
k
δ
>
1
4k\delta>1
4kδ>1, if we let
x
1
=
1
/
2
k
π
,
x
2
=
1
/
(
2
k
π
+
π
/
2
)
x_1=1/2k\pi,x_2=1/(2k\pi+\pi/2)
x1=1/2kπ,x2=1/(2kπ+π/2), then
∣
x
1
−
x
2
∣
=
∣
1
2
k
π
−
1
2
k
π
+
π
/
2
∣
=
π
/
2
2
k
π
(
2
k
π
+
π
/
2
)
=
1
4
k
(
2
k
π
+
π
/
2
)
<
1
4
k
<
δ
|x_1-x_2|=\left|\frac{1}{2k\pi}-\frac{1}{2k\pi+\pi/2}\right|=\frac{\pi/2}{2k\pi(2k\pi+\pi/2)}=\frac{1}{4k(2k\pi+\pi/2)} <\frac{1}{4k}<\delta
∣x1−x2∣=∣∣∣∣2kπ1−2kπ+π/21∣∣∣∣=2kπ(2kπ+π/2)π/2=4k(2kπ+π/2)1<4k1<δ
but
∣
g
(
x
1
)
−
g
(
x
2
)
∣
=
∣
sin
(
2
k
π
)
−
sin
(
2
k
π
+
π
/
2
)
∣
=
1
|g(x_1)-g(x_2)|=|\sin(2k\pi)-\sin(2kπ+π/2)|=1
∣g(x1)−g(x2)∣=∣sin(2kπ)−sin(2kπ+π/2)∣=1.
Exercise 2. Let
X
X
X denote the subset
(
−
1
,
1
)
×
0
(-1,1)\times 0
(−1,1)×0 of
R
2
\mathbf{R^2}
R2, and let
U
U
U be the open ball
B
(
0
;
1
)
B(\mathbf{0};1)
B(0;1) in
R
2
\mathbf{R^2}
R2, which contains
X
X
X. Show there is no
ϵ
>
0
\epsilon>0
ϵ>0 such that the
ϵ
\epsilon
ϵ-neighborhood of
X
X
X in
R
n
\mathbf{R^n}
Rn is contained in
U
U
U.
Solution: Given any
ϵ
>
0
\epsilon>0
ϵ>0, the point
a
=
(
1
−
ϵ
/
2
,
0
)
∈
X
\mathbf{a}=(1-\epsilon/2,0)\in X
a=(1−ϵ/2,0)∈X, and
B
(
a
,
ϵ
)
B(\mathbf{a},\epsilon)
B(a,ϵ) is a
ϵ
\epsilon
ϵ-neighborhood of
X
X
X, but for the point
b
=
(
1
+
ϵ
/
4
,
0
)
\mathbf{b}=(1+\epsilon/4,0)
b=(1+ϵ/4,0), we have
∥
b
−
a
∥
=
3
ϵ
/
4
<
ϵ
⟹
b
∈
B
(
a
,
ϵ
)
,
∥
b
−
0
∥
=
1
+
ϵ
/
4
>
1
⟹
b
∉
B
(
0
,
1
)
=
U
\|\mathbf{b}-\mathbf{a}\|=3\epsilon/4<\epsilon \implies b\in B(\mathbf{a},\epsilon),\quad \|\mathbf{b}-\mathbf{0}\|=1+\epsilon/4>1 \implies \mathbf{b}\notin B(\mathbf{0},1)=U
∥b−a∥=3ϵ/4<ϵ⟹b∈B(a,ϵ),∥b−0∥=1+ϵ/4>1⟹b∈/B(0,1)=U
thus
B
(
a
,
ϵ
)
⊈
U
B(\mathbf{a},\epsilon)\nsubseteq U
B(a,ϵ)⊈U.
Exercise 3. Let
R
∞
\mathbf{R}^{\infty}
R∞ be the set of all “infinite-tuples”
x
=
(
x
1
,
x
2
,
…
)
\mathbf{x}=(x_1,x_2,\dots)
x=(x1,x2,…) of real numbers that end in an infinite string of 0’s. Define an inner product on
R
∞
\mathbf{R}^{\infty}
R∞ by the rule
⟨
x
,
y
⟩
=
∑
x
i
y
i
\langle x,y\rangle=\sum{x_iy_i}
⟨x,y⟩=∑xiyi. (This is a funite sum, since all but finitely many terms vanish.) Let
∥
x
−
y
∥
\|\mathbf{x}-\mathbf{y}\|
∥x−y∥ be the corresponding metric on
R
∞
\mathbf{R}^{\infty}
R∞. Define
e
i
=
(
0
,
…
,
0
,
1
,
0
,
…
,
0
,
…
)
\mathbf{e}_i=(0,\dots,0,1,0,\dots,0,\dots)
ei=(0,…,0,1,0,…,0,…)
where 1 appears in the
i
t
h
i^{th}
ith place. Then the
e
i
\mathbf{e}_i
ei form a basis for
R
∞
\mathbf{R}^{\infty}
R∞. Let
X
X
X be the set of all the points
e
i
\mathbf{e}_i
ei. Show that
X
X
X is closed, bounded and non-compact.
Solution:
X
X
X contains only isolated points, thus is closed. Since
∥
e
i
∥
=
⟨
e
i
,
e
i
⟩
=
1
,
∀
i
\|\mathbf{e}_i\|=\sqrt{\langle \mathbf{e}_i,\mathbf{e}_i\rangle}=1,\forall i
∥ei∥=⟨ei,ei⟩=1,∀i,
X
X
X is bounded.
To see
X
X
X is not compact, notice that
∥
e
i
−
e
j
∥
=
2
,
∀
i
≠
j
\|\mathbf{e}_i-\mathbf{e}_j\|=\sqrt{2},\forall i\neq j
∥ei−ej∥=2,∀i=j, then
B
(
e
i
,
1
)
∩
X
=
e
i
,
∀
i
B(\mathbf{e}_i,1)\cap X=\mathbf{e}_i,\forall i
B(ei,1)∩X=ei,∀i, the open sets
{
B
(
e
i
,
1
)
:
i
∈
N
+
}
\{B(\mathbf{e}_i,1):i\in \mathbf{N}^+ \}
{B(ei,1):i∈N+} covers
X
X
X, but no finite subcover of
X
X
X exists.
Exercise 4.
( a ) Show that open balls and open cubes in
R
n
\mathbf{R}^n
Rn are convex.
( b ) Show that (open and closed) rectangles in
R
n
\mathbf{R}^n
Rn are convex.
Solution:
( a ) Let
B
(
y
,
ϵ
)
B(\mathbf{y},{\epsilon})
B(y,ϵ) be an open ball in
R
n
\mathbf{R}^n
Rn,
C
(
y
,
ϵ
)
C(\mathbf{y},{\epsilon})
C(y,ϵ) be an open cube in
R
n
\mathbf{R}^n
Rn. First for any
a
,
b
∈
B
(
y
,
ϵ
)
\mathbf{a},\mathbf{b}\in B(\mathbf{y},{\epsilon})
a,b∈B(y,ϵ), there is the relation
∥
a
−
y
∥
<
ϵ
,
∥
b
−
y
∥
<
ϵ
\|\mathbf{a}-\mathbf{y}\|<{\epsilon},\|\mathbf{b}-\mathbf{y}\|<{\epsilon}
∥a−y∥<ϵ,∥b−y∥<ϵ, thus if
x
=
a
+
t
(
b
−
a
)
,
0
≤
t
≤
1
\mathbf{x}=\mathbf{a}+t(\mathbf{b}-\mathbf{a}),0\leq t\leq 1
x=a+t(b−a),0≤t≤1, then
∥
x
−
y
∥
=
∥
a
+
t
(
b
−
a
)
−
y
∥
=
∥
(
1
−
t
)
(
a
−
y
)
+
t
(
b
−
y
)
∥
≤
(
1
−
t
)
∥
a
−
y
∥
+
t
∥
b
−
y
∥
<
ϵ
\begin{aligned}\|\mathbf{x}-\mathbf{y}\|&=\|\mathbf{a}+t(\mathbf{b}-\mathbf{a})-\mathbf{y}\|=\|(1-t)(\mathbf{a}-\mathbf{y})+t(\mathbf{b}-\mathbf{y})\|\\&\leq (1-t)\|\mathbf{a}-\mathbf{y}\|+t\|\mathbf{b}-\mathbf{y}\|<{\epsilon}\end{aligned}
∥x−y∥=∥a+t(b−a)−y∥=∥(1−t)(a−y)+t(b−y)∥≤(1−t)∥a−y∥+t∥b−y∥<ϵ
which means
x
∈
B
(
y
,
ϵ
)
\mathbf{x}\in B(\mathbf{y},{\epsilon})
x∈B(y,ϵ), thus
B
(
y
,
ϵ
)
B(\mathbf{y},{\epsilon})
B(y,ϵ) is convex.
Next for any
a
,
b
∈
C
(
y
,
ϵ
)
a,b\in C(\mathbf{y},{\epsilon})
a,b∈C(y,ϵ), there is the relation
∣
a
−
y
∣
<
ϵ
,
∣
b
−
y
∣
<
ϵ
|\mathbf{a}-\mathbf{y}|<{\epsilon},|\mathbf{b}-\mathbf{y}|<{\epsilon}
∣a−y∣<ϵ,∣b−y∣<ϵ, thus if
x
=
a
+
t
(
b
−
a
)
,
0
≤
t
≤
1
\mathbf{x}=\mathbf{a}+t(\mathbf{b}-\mathbf{a}),0\leq t\leq 1
x=a+t(b−a),0≤t≤1, then
∣
x
−
y
∣
=
∣
a
+
t
(
b
−
a
)
−
y
∣
=
∣
(
1
−
t
)
(
a
−
y
)
+
t
(
b
−
y
)
∣
≤
(
1
−
t
)
∣
a
−
y
∣
+
t
∣
b
−
y
∣
<
ϵ
\begin{aligned}|\mathbf{x}-\mathbf{y}|&=|\mathbf{a}+t(\mathbf{b}-\mathbf{a})-\mathbf{y}|=|(1-t)(\mathbf{a}-\mathbf{y})+t(\mathbf{b}-\mathbf{y})|\\&\leq (1-t)|\mathbf{a}-\mathbf{y}|+t|\mathbf{b}-\mathbf{y}|<{\epsilon}\end{aligned}
∣x−y∣=∣a+t(b−a)−y∣=∣(1−t)(a−y)+t(b−y)∣≤(1−t)∣a−y∣+t∣b−y∣<ϵ
which means
x
∈
C
(
y
,
ϵ
)
\mathbf{x}\in C(\mathbf{y},{\epsilon})
x∈C(y,ϵ), thus
C
(
y
,
ϵ
)
C(\mathbf{y},{\epsilon})
C(y,ϵ) is convex.
( b ) If
Q
Q
Q is a rectangle in
R
n
\mathbf{R}^n
Rn, we can write
Q
=
[
c
1
,
d
1
]
×
⋯
×
[
c
n
,
d
n
]
Q=[c_1,d_1 ]\times{\cdots}\times[c_n,d_n]
Q=[c1,d1]×⋯×[cn,dn], for any
a
,
b
∈
Q
a,b\in Q
a,b∈Q, we write
a
=
(
a
1
,
…
,
a
n
)
,
b
=
(
b
1
,
…
,
b
n
)
a=(a_1,{\dots},a_n ),b=(b_1,{\dots},b_n)
a=(a1,…,an),b=(b1,…,bn), then there is the relation
c
i
≤
a
i
,
b
i
≤
d
i
,
1
≤
i
≤
n
c_i\leq a_i,b_i\leq d_i,\quad 1\leq i\leq n
ci≤ai,bi≤di,1≤i≤n
thus if
x
=
a
+
t
(
b
−
a
)
,
0
≤
t
≤
1
\mathbf{x}=\mathbf{a}+t(\mathbf{b}-\mathbf{a}),0\leq t\leq 1
x=a+t(b−a),0≤t≤1, then
x
i
=
a
i
+
t
(
b
i
−
a
i
)
=
(
1
−
t
)
a
i
+
t
b
i
∈
[
c
i
,
d
i
]
x_i=a_i+t(b_i-a_i )=(1-t) a_i+tb_i\in [c_i,d_i ]
xi=ai+t(bi−ai)=(1−t)ai+tbi∈[ci,di], so
x
∈
Q
\mathbf{x}\in Q
x∈Q.
If
Q
Q
Q is an open rectangle in
R
n
\mathbf{R}^n
Rn, the proof is very similar.