本文主要内容如下:
- 1. 范数的定义
- 2. 常见的范数举例
- 3. 范数的等价
- 4. 距离与度量空间的定义
1. 范数的定义
定义1-1:设
E
E
E 为向量空间,
R
\mathbb{R}
R 为实数域。若映射
∥
⋅
∥
:
E
→
R
:
x
↦
∥
x
∥
\begin{equation*} \lVert\cdot\rVert:~E\rightarrow\mathbb{R}:~x\mapsto\lVert{x}\rVert \end{equation*}
∥⋅∥: E→R: x↦∥x∥
对
∀
x
,
y
∈
E
;
λ
∈
R
\forall~x,y\in{E};~\lambda\in\mathbb{R}
∀ x,y∈E; λ∈R 满足
~~~ 1) 正定性: ∥ x ∥ ⩾ 0 ~\lVert{x}\rVert\geqslant0 ∥x∥⩾0 ,当且仅当 x = 0 x=0 x=0 时取等号;
~~~ 2) 正齐次性: ∥ λ x ∥ = ∣ λ ∣ ⋅ ∥ x ∥ ~\lVert{\lambda x}\rVert=\lvert{\lambda}\rvert\cdot\lVert{x}\rVert ∥λx∥=∣λ∣⋅∥x∥;
~~~ 3) 三角不等式: ∥ x + y ∥ ⩽ ∥ x ∥ + ∥ y ∥ ~\lVert{x+y}\rVert\leqslant\lVert{x}\rVert+\lVert{y}\rVert ∥x+y∥⩽∥x∥+∥y∥;
则将其称为范数,为区分定义在不同向量空间上的范数,也记作 ∥ ⋅ ∥ E \lVert\cdot\rVert_E ∥⋅∥E,而将 ( E , ∥ ⋅ ∥ ) (E,\lVert{\cdot}\rVert) (E,∥⋅∥) 称为赋范线性空间。
定理1-1:范数
∥
⋅
∥
\lVert{\cdot}\rVert
∥⋅∥ 是一个凸函数,即
∥
λ
x
+
(
1
−
λ
)
y
∥
⩽
λ
∥
x
∥
+
(
1
−
λ
)
∥
y
∥
,
∀
x
,
y
∈
E
;
λ
∈
[
0
,
1
]
\begin{equation} \lVert{\lambda x+(1-\lambda)y}\rVert\leqslant\lambda\lVert{x}\rVert+(1-\lambda)\lVert{y}\rVert ,\qquad\forall~x,y\in{E};~\lambda\in[0,1] \end{equation}
∥λx+(1−λ)y∥⩽λ∥x∥+(1−λ)∥y∥,∀ x,y∈E; λ∈[0,1]证明:由范数定义中的三角不等式:
∥
λ
x
+
(
1
−
λ
)
y
∥
⩽
∥
λ
x
∥
+
∥
(
1
−
λ
)
y
∥
=
∣
λ
∣
⋅
∥
x
∥
+
∣
1
−
λ
∣
⋅
∥
y
∥
=
λ
∥
x
∥
+
(
1
−
λ
)
∥
y
∥
(证毕)
\begin{align*} \lVert{\lambda x+(1-\lambda)y}\rVert &\leqslant\lVert{\lambda x}\rVert+\lVert(1-\lambda){y}\rVert \\[3mm] &=|\lambda|\cdot\lVert{x}\rVert+|1-\lambda|\cdot\lVert{y}\rVert \\[3mm] &=\lambda\lVert{x}\rVert+(1-\lambda)\lVert{y}\rVert \qquad\qquad\text{(证毕)} \end{align*}
∥λx+(1−λ)y∥⩽∥λx∥+∥(1−λ)y∥=∣λ∣⋅∥x∥+∣1−λ∣⋅∥y∥=λ∥x∥+(1−λ)∥y∥(证毕)
定理1-2:定义在向量空间
E
E
E 上的范数
∥
⋅
∥
\lVert{\cdot}\rVert
∥⋅∥ 满足不等式:
∣
∣
∣
x
∣
∣
−
∣
∣
y
∣
∣
∣
≤
∣
∣
x
±
y
∣
∣
≤
∣
∣
x
∣
∣
+
∣
∣
y
∣
∣
,
x
,
y
∈
E
\begin{equation} \big\lvert~||{x}||-||{y}||~\big\rvert\le||{x}\pm{y}||\le||{x}||+||{y}|| ,\qquad x,y\in E \end{equation}
∣∣x∣∣−∣∣y∣∣
≤∣∣x±y∣∣≤∣∣x∣∣+∣∣y∣∣,x,y∈E证明:由三角不等式:
{
∣
∣
x
−
y
∣
∣
+
∣
∣
y
∣
∣
≥
∣
∣
x
∣
∣
⟹
∣
∣
x
−
y
∣
∣
≥
∣
∣
x
∣
∣
−
∣
∣
y
∣
∣
∣
∣
x
−
y
∣
∣
+
∣
∣
x
∣
∣
=
∣
∣
y
−
x
∣
∣
+
∣
∣
x
∣
∣
≥
∣
∣
y
∣
∣
⟹
∣
∣
x
−
y
∣
∣
≥
∣
∣
y
∣
∣
−
∣
∣
x
∣
∣
\begin{cases} ||{x}-{y}||+||{y}||\ge||{x}||\Longrightarrow ||{x}-{y}||\ge||{x}||-||{y}|| \\[5mm] ||{x}-{y}||+||{x}||=||{y}-{x}||+||{x}||\ge||{y}||\Longrightarrow ||{x}-{y}||\ge||{y}||-||{x}|| \end{cases}
⎩
⎨
⎧∣∣x−y∣∣+∣∣y∣∣≥∣∣x∣∣⟹∣∣x−y∣∣≥∣∣x∣∣−∣∣y∣∣∣∣x−y∣∣+∣∣x∣∣=∣∣y−x∣∣+∣∣x∣∣≥∣∣y∣∣⟹∣∣x−y∣∣≥∣∣y∣∣−∣∣x∣∣故有:
∣
∣
∣
x
∣
∣
−
∣
∣
y
∣
∣
∣
≤
∣
∣
x
−
y
∣
∣
\begin{equation*} \big\lvert~||{x}||-||{y}||~\big\rvert\le||{x}-{y}|| \end{equation*}
∣∣x∣∣−∣∣y∣∣
≤∣∣x−y∣∣又
∣
∣
x
−
y
∣
∣
≤
∣
∣
x
∣
∣
+
∣
∣
−
y
∣
∣
=
∣
∣
x
∣
∣
+
∣
∣
y
∣
∣
(证毕)
\begin{equation*} ||{x}-{y}||\le||x||+||-y||=||x||+||y||\qquad\text{(证毕)} \end{equation*}
∣∣x−y∣∣≤∣∣x∣∣+∣∣−y∣∣=∣∣x∣∣+∣∣y∣∣(证毕)
2. 常见的范数举例
例
1
1
1:定义在 实数空间
R
\mathbb{R}
R 上的范数 :
∥
x
∥
=
∣
x
∣
,
x
∈
R
\lVert{x}\rVert=|x|,\qquad x\in\mathbb{R}
∥x∥=∣x∣,x∈R容易验证上述定义是满足范数的三点要求的。
例
2
2
2:定义在
n
n
n 维 Euclidean 空间
R
n
\mathbb{R}^n
Rn
R
n
=
{
x
=
(
x
1
,
x
2
,
⋯
,
x
n
)
∣
x
i
∈
R
,
1
=
1
,
2
,
⋯
,
n
}
\begin{equation*} \mathbb{R}^n=\bigg\{x=(x_1,x_2,\cdots,x_n)\bigg|x_i\in\mathbb{R},~1=1,2,\cdots,n\bigg\} \end{equation*}
Rn={x=(x1,x2,⋯,xn)
xi∈R, 1=1,2,⋯,n}上的
p
p
p-范数:
∣
∣
x
∣
∣
p
≜
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
,
x
∈
R
n
;
p
∈
[
1
,
∞
)
\begin{equation} ||{x}||_{p}\triangleq\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}} ,\qquad x\in\mathbb{R}^n~;~p\in[1,\infty) \end{equation}
∣∣x∣∣p≜(i=1∑n∣xi∣p)p1,x∈Rn ; p∈[1,∞)
特比地,
{
1
-范数:
∣
∣
x
∣
∣
1
≜
∑
i
=
1
n
∣
x
i
∣
2
-范数(欧式范数):
∣
∣
x
∣
∣
2
≜
∑
i
=
1
n
x
i
2
∞
-范数(最大范数):
∣
∣
x
∣
∣
∞
≜
max
1
≤
i
≤
n
∣
x
i
∣
\begin{cases} \text{$1$-范数:} &||{x}||_{1}\triangleq \displaystyle{\sum_{i=1}^n}~|x_i| \\[6mm] \text{$2$-范数(欧式范数):} &||{x}||_{2}\triangleq \sqrt{\displaystyle{\sum_{i=1}^n}~x_i^2}\\[6mm] \text{$\infty$-范数(最大范数):}&||{x}||_{\infty}\triangleq \max\limits_{1\le i\le n}~|x_i| \end{cases}
⎩
⎨
⎧1-范数:2-范数(欧式范数):∞-范数(最大范数):∣∣x∣∣1≜i=1∑n ∣xi∣∣∣x∣∣2≜i=1∑n xi2∣∣x∣∣∞≜1≤i≤nmax ∣xi∣
证明:先验证上述“范数”的定义满足范数的三点要求 :
1) 正定性 :
∣
∣
x
∣
∣
p
=
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
≥
0
||{x}||_{p}=\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}\ge0
∣∣x∣∣p=(i=1∑n∣xi∣p)p1≥0上述不等式取等号时,当且仅当
∣
x
i
∣
=
0
(
i
=
1
,
2
,
…
,
n
)
⟺
x
=
0
|x_i|=0\ (i=1,2,\dots,n)\Longleftrightarrow {x}=0
∣xi∣=0 (i=1,2,…,n)⟺x=0 2)正齐次性 :
∣
∣
λ
x
∣
∣
p
=
(
∑
i
=
1
n
(
∣
λ
∣
⋅
∣
x
i
∣
)
p
)
1
p
=
∣
λ
∣
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
=
∣
λ
∣
⋅
∣
∣
x
∣
∣
p
||\lambda{x}||_{p}=\left(\sum_{i=1}^n (|\lambda|\cdot|x_i|)^p\right)^{\frac{1}{p}}=|\lambda|\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}=|\lambda|\cdot||{x}||_{p}
∣∣λx∣∣p=(i=1∑n(∣λ∣⋅∣xi∣)p)p1=∣λ∣(i=1∑n∣xi∣p)p1=∣λ∣⋅∣∣x∣∣p3)三角不等式:
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
=
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
−
1
∣
x
i
+
y
i
∣
≤
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
−
1
(
∣
x
i
∣
+
∣
y
i
∣
)
=
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
−
1
∣
x
i
∣
+
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
−
1
∣
y
i
∣
≤
[
∑
i
=
1
n
∣
x
i
+
y
i
∣
q
(
p
−
1
)
]
1
q
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
+
[
∑
i
=
1
n
∣
x
i
+
y
i
∣
q
(
p
−
1
)
]
1
q
(
∑
i
=
1
n
∣
y
i
∣
p
)
1
p
(
p
q
=
p
q
−
q
;
p
,
q
>
1
)
=
[
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
+
(
∑
i
=
1
n
∣
y
i
∣
p
)
1
p
]
[
∑
i
=
1
n
∣
x
i
+
y
i
∣
q
(
p
−
1
)
]
1
q
=
[
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
+
(
∑
i
=
1
n
∣
y
i
∣
p
)
1
p
]
[
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
]
1
q
\begin{aligned} & \quad\sum_{i=1}^n |x_i+y_i|^p \\\\ & =\sum_{i=1}^n |x_i+y_i|^{p-1}|x_i+y_i| \\\\ & \le\sum_{i=1}^n |x_i+y_i|^{p-1}(|x_i|+|y_i|) \\\\ & =\sum_{i=1}^n |x_i+y_i|^{p-1}|x_i|+\sum_{i=1}^n |x_i+y_i|^{p-1}|y_i| \\\\ & \le\left[\sum_{i=1}^n |x_i+y_i|^{q(p-1)}\right]^{\frac{1}{q}}\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}+\left[\sum_{i=1}^n |x_i+y_i|^{q(p-1)}\right]^{\frac{1}{q}}\left(\sum_{i=1}^n |y_i|^p\right)^{\frac{1}{p}} (pq=pq-q;p,q>1) \\\\ & =\left[\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}+\left(\sum_{i=1}^n |y_i|^p\right)^{\frac{1}{p}}\right]\left[\sum_{i=1}^n |x_i+y_i|^{q(p-1)}\right]^{\frac{1}{q}} \\\\ & =\left[\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}+\left(\sum_{i=1}^n |y_i|^p\right)^{\frac{1}{p}}\right]\left[\sum_{i=1}^n |x_i+y_i|^{p}\right]^{\frac{1}{q}} \end{aligned}
i=1∑n∣xi+yi∣p=i=1∑n∣xi+yi∣p−1∣xi+yi∣≤i=1∑n∣xi+yi∣p−1(∣xi∣+∣yi∣)=i=1∑n∣xi+yi∣p−1∣xi∣+i=1∑n∣xi+yi∣p−1∣yi∣≤[i=1∑n∣xi+yi∣q(p−1)]q1(i=1∑n∣xi∣p)p1+[i=1∑n∣xi+yi∣q(p−1)]q1(i=1∑n∣yi∣p)p1(pq=pq−q;p,q>1)=
(i=1∑n∣xi∣p)p1+(i=1∑n∣yi∣p)p1
[i=1∑n∣xi+yi∣q(p−1)]q1=
(i=1∑n∣xi∣p)p1+(i=1∑n∣yi∣p)p1
[i=1∑n∣xi+yi∣p]q1上述证明过程前后应用了绝对值不等式与Holder不等式,进一步:
[
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
]
1
−
1
q
=
[
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
]
1
p
≤
[
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
+
(
∑
i
=
1
n
∣
y
i
∣
p
)
1
p
]
(
p
>
1
)
\left[\sum_{i=1}^n |x_i+y_i|^{p}\right]^{1-\frac{1}{q}} =\left[\sum_{i=1}^n |x_i+y_i|^{p}\right]^{\frac{1}{p}} \le\left[\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}+\left(\sum_{i=1}^n |y_i|^p\right)^{\frac{1}{p}}\right](p>1)
[i=1∑n∣xi+yi∣p]1−q1=[i=1∑n∣xi+yi∣p]p1≤
(i=1∑n∣xi∣p)p1+(i=1∑n∣yi∣p)p1
(p>1)当
p
=
1
p=1
p=1 时由绝对不等式知上述不等式同样成立,综上得证Minkowski不等式(闵可夫斯基不等式):
[
∑
i
=
1
n
∣
x
i
+
y
i
∣
p
]
1
p
≤
[
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
+
(
∑
i
=
1
n
∣
y
i
∣
p
)
1
p
]
(
p
≥
1
)
\left[\sum_{i=1}^n |x_i+y_i|^{p}\right]^{\frac{1}{p}} \le\left[\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}+\left(\sum_{i=1}^n |y_i|^p\right)^{\frac{1}{p}}\right](p\ge1)
[i=1∑n∣xi+yi∣p]p1≤
(i=1∑n∣xi∣p)p1+(i=1∑n∣yi∣p)p1
(p≥1)即,
∣
∣
x
+
y
∣
∣
p
≤
∣
∣
x
∣
∣
p
+
∣
∣
y
∣
∣
p
||{x}+{y}||_p\le||{x}||_p+||{y}||_p
∣∣x+y∣∣p≤∣∣x∣∣p+∣∣y∣∣p 最后说明,当
p
→
∞
p\rightarrow\infty
p→∞ 时,
p
−
p-
p−范数满足最大范数的定义。采用夹逼定理求极限,由于
0
≤
max
1
≤
i
≤
n
∣
x
i
∣
p
≤
∑
i
=
1
n
∣
x
i
∣
p
≤
n
(
max
1
≤
i
≤
n
∣
x
i
∣
p
)
0\le\max\limits_{1\le i\le n}|x_i|^p \le\sum_{i=1}^n |x_i|^p\le n\left(\max\limits_{1\le i\le n}|x_i|^p\right)
0≤1≤i≤nmax∣xi∣p≤i=1∑n∣xi∣p≤n(1≤i≤nmax∣xi∣p)又
{
lim
p
→
∞
(
max
1
≤
i
≤
n
∣
x
i
∣
p
)
1
p
=
max
1
≤
i
≤
n
∣
x
i
∣
lim
p
→
∞
[
n
(
max
1
≤
i
≤
n
∣
x
i
∣
p
)
]
1
p
=
max
1
≤
i
≤
n
∣
x
i
∣
lim
p
→
∞
n
1
p
=
max
1
≤
i
≤
n
∣
x
i
∣
\begin{cases} \displaystyle{\lim_{p\rightarrow\infty}}\left(\max\limits_{1\le i\le n}|x_i|^p\right)^{\frac{1}{p}}=\max\limits_{1\le i\le n}|x_i|\\\\ \displaystyle{\lim_{p\rightarrow\infty}}\left[n\left(\max\limits_{1\le i\le n}|x_i|^p\right)\right]^{\frac{1}{p}}=\max\limits_{1\le i\le n}|x_i|\lim_{p\rightarrow\infty}n^{\frac{1}{p}}=\max\limits_{1\le i\le n}|x_i| \end{cases}
⎩
⎨
⎧p→∞lim(1≤i≤nmax∣xi∣p)p1=1≤i≤nmax∣xi∣p→∞lim[n(1≤i≤nmax∣xi∣p)]p1=1≤i≤nmax∣xi∣p→∞limnp1=1≤i≤nmax∣xi∣故
∣
∣
x
∣
∣
∞
=
lim
p
→
∞
(
∑
i
=
1
n
∣
x
i
∣
p
)
1
p
=
max
1
≤
i
≤
n
∣
x
i
∣
(证毕)
||{x}||_{\infty}=\lim_{p\rightarrow\infty}\left(\sum_{i=1}^n |x_i|^p\right)^{\frac{1}{p}}=\max\limits_{1\le i\le n}|x_i| \qquad\qquad\text{(证毕)}
∣∣x∣∣∞=p→∞lim(i=1∑n∣xi∣p)p1=1≤i≤nmax∣xi∣(证毕)
3. 范数的等价
定义3-1:设
N
(
E
)
\mathcal{N}(E)
N(E) 为所有定义在向量空间
E
E
E 上的范数构成的集合,对于
∥
⋅
∥
,
∥
⋅
∥
×
∈
N
(
E
)
\lVert\cdot\rVert,~\lVert\cdot\rVert^\times\in\mathcal{N}(E)
∥⋅∥, ∥⋅∥×∈N(E),若
∃
α
,
β
∈
R
+
,
s
.
t
.
α
∥
x
∥
×
⩽
∥
x
∥
⩽
β
∥
x
∥
×
,
f
o
r
∀
x
∈
E
\begin{equation*} \exist~\alpha,\beta\in\mathbb{R}^+,~s.t.~ \alpha\lVert{x}\rVert^\times \leqslant \lVert{x}\rVert \leqslant \beta\lVert{x}\rVert^\times ,~for~\forall~x\in{E} \end{equation*}
∃ α,β∈R+, s.t. α∥x∥×⩽∥x∥⩽β∥x∥×, for ∀ x∈E则称 范数
∥
⋅
∥
\lVert\cdot\rVert
∥⋅∥与
∥
⋅
∥
×
\lVert\cdot\rVert^\times
∥⋅∥×等价,记作:
∥
⋅
∥
∼
∥
⋅
∥
×
\lVert\cdot\rVert\sim\lVert\cdot\rVert^\times
∥⋅∥∼∥⋅∥×。
定理3-1 :范数的等价具有传递性,即
∣
∣
x
⃗
∣
∼
∣
∣
x
⃗
∣
∣
×
,
∣
∣
x
⃗
∣
∣
∼
∣
∣
x
⃗
∣
∣
×
×
⟹
∣
∣
x
∣
∣
×
∼
∣
∣
x
∣
∣
×
×
,
∀
x
∈
E
\begin{equation*} ||\vec{x}|\sim||\vec{x}||^\times,~||\vec{x}||\sim||\vec{x}||^{\times\times} \Longrightarrow ||{x}||^\times\sim||{x}||^{\times\times} ,\qquad\forall~{x}\in{E} \end{equation*}
∣∣x∣∼∣∣x∣∣×, ∣∣x∣∣∼∣∣x∣∣××⟹∣∣x∣∣×∼∣∣x∣∣××,∀ x∈E
证明:若
∣
∣
x
⃗
∣
∣
||\vec{x}||
∣∣x∣∣ 与
∣
∣
x
⃗
∣
∣
×
||\vec{x}||^\times
∣∣x∣∣× 等价,且
∣
∣
x
⃗
∣
∣
||\vec{x}||
∣∣x∣∣ 与
∣
∣
x
⃗
∣
∣
×
×
||\vec{x}||^{\times\times}
∣∣x∣∣×× 等价,则
∃
c
i
∈
R
,
i
=
1
,
2
,
3
,
4
\exist~c_i\in\mathbb{R},~i=1,2,3,4
∃ ci∈R, i=1,2,3,4 使得 :
{
c
1
∣
∣
x
∣
∣
≤
∣
∣
x
∣
∣
×
≤
c
2
∣
∣
x
∣
∣
c
3
∣
∣
x
⃗
∣
∣
×
×
≤
∣
∣
x
∣
∣
≤
c
4
∣
∣
x
⃗
∣
×
×
,
∀
x
∈
E
\begin{cases} \ c_1||{x}||\le||{x}||^\times\le c_2||{x}||\\[4mm] \ c_3||\vec{x}||^{\times\times}\le||{x}||\le c_4||\vec{x}|^{\times\times} ,\qquad\forall~{x}\in{E} \end{cases}
⎩
⎨
⎧ c1∣∣x∣∣≤∣∣x∣∣×≤c2∣∣x∣∣ c3∣∣x∣∣××≤∣∣x∣∣≤c4∣∣x∣××,∀ x∈E那么,
c
1
c
3
∣
∣
x
∣
∣
×
×
≤
∣
∣
x
∣
∣
×
≤
c
2
c
4
∣
∣
x
⃗
∣
∣
×
×
,
∀
x
∈
E
c_1c_3||{x}||^{\times\times}\le||{x}||^\times\le c_2c_4||\vec{x}||^{\times\times} ,\qquad\forall~{x}\in{E}
c1c3∣∣x∣∣××≤∣∣x∣∣×≤c2c4∣∣x∣∣××,∀ x∈E故
∣
∣
x
∣
∣
×
||{x}||^\times
∣∣x∣∣× 与
∣
∣
x
∣
∣
×
×
||{x}||^{\times\times}
∣∣x∣∣××等价。
定理3-2:定义在
R
n
\mathbb{R}^n
Rn 上的
1
−
1-
1−范数、
2
−
2-
2−范数、
∞
−
\infty-
∞−范数间满足:
∥
x
∥
∞
⩽
∥
x
∥
2
⩽
∥
x
∥
1
⩽
n
∥
x
∥
∞
,
∀
x
∈
R
n
\begin{equation} \lVert{x}\rVert_\infty \leqslant \lVert{x}\rVert_2 \leqslant \lVert{x}\rVert_1 \leqslant n\lVert{x}\rVert_\infty ,~\forall~x\in\mathbb{R}^n \end{equation}
∥x∥∞⩽∥x∥2⩽∥x∥1⩽n∥x∥∞, ∀ x∈Rn故
∥
x
∥
1
∼
∥
x
∥
2
∼
∥
x
∥
∞
\lVert{x}\rVert_1\sim\lVert{x}\rVert_2\sim\lVert{x}\rVert_\infty
∥x∥1∼∥x∥2∼∥x∥∞
证明:显然
max
1
≤
i
≤
n
∣
x
i
∣
≤
∑
i
=
1
n
∣
x
i
∣
≤
n
(
max
1
≤
i
≤
n
∣
x
i
∣
)
⟹
∥
x
∥
∞
⩽
∥
x
∥
1
⩽
n
∥
x
∥
∞
\max\limits_{1\le i\le n}|x_i| \le\sum_{i=1}^n |x_i|\le n\left(\max\limits_{1\le i\le n}|x_i|\right) \Longrightarrow \lVert{x}\rVert_\infty\leqslant\lVert{x}\rVert_1\leqslant n\lVert{x}\rVert_\infty
1≤i≤nmax∣xi∣≤i=1∑n∣xi∣≤n(1≤i≤nmax∣xi∣)⟹∥x∥∞⩽∥x∥1⩽n∥x∥∞又
x
1
2
+
x
2
2
+
⋯
+
x
N
2
⩾
(
max
1
≤
i
≤
n
∣
x
i
∣
)
2
=
max
1
≤
i
≤
n
∣
x
i
∣
⟹
∥
x
∥
∞
⩽
∥
x
∥
2
\sqrt{x_1^2+x_2^2+\dots+x_N^2}\geqslant\sqrt{\left(\max\limits_{1\le i\le n}|x_i| \right)^2}=\max\limits_{1\le i\le n}|x_i| \Longrightarrow \lVert{x}\rVert_\infty\leqslant\lVert{x}\rVert_2
x12+x22+⋯+xN2⩾(1≤i≤nmax∣xi∣)2=1≤i≤nmax∣xi∣⟹∥x∥∞⩽∥x∥2且根据 Cauchy-Bunjakovski 不等式(Cauchy–Schwarz 不等式)可得
x
1
2
+
x
2
2
+
⋯
+
x
N
2
≤
(
∣
x
1
∣
+
∣
x
2
∣
+
⋯
+
∣
x
N
∣
)
2
⟹
∣
∣
x
∣
∣
2
⩽
∣
∣
x
∣
∣
1
(证毕)
x_1^2+x_2^2+\dots+x_N^2\le(|x_1|+|x_2|+\dots+|x_N|)^2 \Longrightarrow ||{x}||_2\leqslant||{x}||_1 \qquad\text{(证毕)}
x12+x22+⋯+xN2≤(∣x1∣+∣x2∣+⋯+∣xN∣)2⟹∣∣x∣∣2⩽∣∣x∣∣1(证毕)
4. 距离与度量空间的定义
定义4-1:设
S
S
S 是非空集合(不一定是向量空间),若映射
d
:
S
×
S
→
R
:
(
x
,
y
)
↦
d
(
x
,
y
)
\begin{equation*} d:~S\times S\rightarrow\mathbb{R}:~(x,y)\mapsto{d(x,y)} \end{equation*}
d: S×S→R: (x,y)↦d(x,y)满足:
~~~ 1) 正定性: d ( x , y ) ⩾ 0 ~d(x,y)\geqslant0 d(x,y)⩾0 ,当且仅当 x = y x=y x=y 时取等号;
~~~ 2) 对称性: d ( x , y ) = d ( y , x ) ~d(x,y)=d(y,x) d(x,y)=d(y,x);
~~~ 3) 三点不等式: d ( x , y ) ⩽ d ( x , z ) + d ( z , y ) ~d(x,y)\leqslant d(x,z)+d(z,y) d(x,y)⩽d(x,z)+d(z,y).
则称 d ( x , y ) d(x,y) d(x,y) 为 x x x 到 y y y 的距离,并将 ( S , d ) (S,d) (S,d) 称为度量空间。
定理4-1:对于赋范向量空间
(
E
,
∥
⋅
∥
)
(E,\lVert\cdot\rVert)
(E,∥⋅∥),若取
d
(
x
,
y
)
≜
∥
x
−
y
∥
,
∀
x
,
y
∈
E
\begin{equation} d(x,y)\triangleq\lVert{x-y}\rVert,\qquad\forall~x,y\in{E} \end{equation}
d(x,y)≜∥x−y∥,∀ x,y∈E则在向量空间
E
E
E 上定义了距离
d
(
x
,
y
)
d(x,y)
d(x,y),即赋范线性空间可视为度量空间,并将上式定义的
d
(
x
,
y
)
d(x,y)
d(x,y) 称为由范数
∥
⋅
∥
\lVert\cdot\rVert
∥⋅∥ 导出的距离。由范数
∥
⋅
∥
\lVert\cdot\rVert
∥⋅∥ 导出的距离
d
(
x
,
y
)
d(x,y)
d(x,y) 满足:
{
d
(
x
−
y
,
0
)
=
∣
∣
(
x
−
y
)
−
0
∣
∣
=
∣
∣
x
−
y
∣
∣
=
d
(
x
,
y
)
d
(
λ
x
,
0
)
=
∣
∣
λ
x
−
0
∣
∣
=
∣
∣
λ
x
∣
∣
=
∣
λ
∣
⋅
∣
∣
x
−
0
∣
∣
=
∣
λ
∣
d
(
x
,
0
)
(
∀
x
,
y
∈
E
;
λ
∈
R
)
\begin{cases} d(x-y,0)=||(x-y)-0||=||x-y||=d(x,y)\\[4mm] d(\lambda x,0)=||\lambda x-0||=||\lambda x||=|\lambda|\cdot||x-0||=|\lambda|~d(x,0) \quad(~\forall~x,y\in E;~\lambda\in\mathbb{R}) \end{cases}
⎩
⎨
⎧d(x−y,0)=∣∣(x−y)−0∣∣=∣∣x−y∣∣=d(x,y)d(λx,0)=∣∣λx−0∣∣=∣∣λx∣∣=∣λ∣⋅∣∣x−0∣∣=∣λ∣ d(x,0)( ∀ x,y∈E; λ∈R)
证明:验证给出的距离取法是否满足距离的定义 ,对 ∀ x , y , z ∈ E \forall~x,y,z\in{E} ∀ x,y,z∈E
1)正定性:
d
(
x
,
y
)
=
∣
∣
x
−
y
∣
∣
≥
0
,
(当且仅当
x
−
y
=
0
⟹
x
=
y
时取等号)
\begin{equation*} d(x,y)=||x-y||\ge0,\quad\text{(当且仅当 $x-y=0\Longrightarrow x=y$ 时取等号)} \end{equation*}
d(x,y)=∣∣x−y∣∣≥0,(当且仅当 x−y=0⟹x=y 时取等号)2)对称性:
d
(
x
,
y
)
=
∣
∣
x
−
y
∣
∣
=
∣
∣
−
(
y
−
x
)
∣
∣
=
∣
∣
y
−
x
∣
∣
=
d
(
y
,
x
)
\begin{equation*} d(x,y)=||x-y||=||-(y-x)||=||y-x||=d(y,x) \end{equation*}
d(x,y)=∣∣x−y∣∣=∣∣−(y−x)∣∣=∣∣y−x∣∣=d(y,x)3)三点不等式:
d
(
x
,
z
)
+
d
(
z
,
y
)
=
∣
∣
x
−
z
∣
∣
+
∣
∣
z
−
y
∣
∣
≥
∣
∣
(
x
−
z
)
+
(
z
−
y
)
∣
∣
=
∣
∣
x
−
y
∣
∣
=
d
(
x
,
y
)
(证毕)
\begin{equation*} d(x,z)+d(z,y)=||x-z||+||z-y||\ge||(x-z)+(z-y)||=||x-y||=d(x,y) \qquad\text{(证毕)} \end{equation*}
d(x,z)+d(z,y)=∣∣x−z∣∣+∣∣z−y∣∣≥∣∣(x−z)+(z−y)∣∣=∣∣x−y∣∣=d(x,y)(证毕)
定理4-2:设
(
S
,
d
)
(S,d)
(S,d) 为定义了距离的线性空间,若距离
d
d
d 满足:
{
d
(
x
−
y
,
0
)
=
d
(
x
,
y
)
d
(
λ
x
,
0
)
=
∣
λ
∣
d
(
x
,
0
)
(
∀
x
,
y
∈
S
;
λ
∈
R
)
\begin{cases} d(x-y,0)=d(x,y)\\[4mm] d(\lambda x,0)=|\lambda|~d(x,0) \quad(~\forall~x,y\in S;~\lambda\in\mathbb{R}) \end{cases}
⎩
⎨
⎧d(x−y,0)=d(x,y)d(λx,0)=∣λ∣ d(x,0)( ∀ x,y∈S; λ∈R)则可以取
∣
∣
x
∣
∣
≜
d
(
x
,
0
)
,
x
∈
S
\begin{equation*} ||x||\triangleq d(x,0),\qquad x\in S \end{equation*}
∣∣x∣∣≜d(x,0),x∈S使得定义出的
∥
⋅
∥
\lVert\cdot\rVert
∥⋅∥ 为范数,并且使得
d
d
d 是由
∥
⋅
∥
\lVert\cdot\rVert
∥⋅∥ 导出的距离。
证明:验证给出的范数取法是否满足范数的定义 ,对 ∀ x , y ∈ S , λ ∈ R \forall~x,y\in{S},~\lambda\in\mathbb{R} ∀ x,y∈S, λ∈R
1)正定性:
∣
∣
x
∣
∣
=
d
(
x
,
0
)
≥
0
,
(当且仅当
x
=
0
时取等号)
\begin{equation*} ||x||=d(x,0)\ge0,\quad\text{(当且仅当 $x=0$ 时取等号)} \end{equation*}
∣∣x∣∣=d(x,0)≥0,(当且仅当 x=0 时取等号)2)正齐次性:
∣
∣
λ
x
∣
∣
=
d
(
λ
x
,
0
)
=
∣
λ
∣
d
(
x
,
0
)
=
∣
λ
∣
⋅
∣
∣
x
∣
∣
\begin{equation*} ||\lambda x||=d(\lambda x,0)=|\lambda|~d(x,0)=|\lambda|\cdot||x|| \end{equation*}
∣∣λx∣∣=d(λx,0)=∣λ∣ d(x,0)=∣λ∣⋅∣∣x∣∣3)三角不等式:
∣
∣
x
∣
∣
+
∣
∣
y
∣
∣
=
d
(
x
,
0
)
+
d
(
y
,
0
)
=
d
(
x
,
0
)
+
d
(
−
y
,
0
)
=
d
(
x
,
0
)
+
d
(
0
,
−
y
)
≥
d
(
x
,
−
y
)
=
d
(
x
+
y
,
0
)
=
∣
∣
x
+
y
∣
∣
\begin{align*} ||x||+||y||&=d(x,0)+d(y,0)=d(x,0)+d(-y,0) \\[3mm] &=d(x,0)+d(0,-y)\ge d(x,-y)=d(x+y,0)=||x+y|| \end{align*}
∣∣x∣∣+∣∣y∣∣=d(x,0)+d(y,0)=d(x,0)+d(−y,0)=d(x,0)+d(0,−y)≥d(x,−y)=d(x+y,0)=∣∣x+y∣∣故所取的范数形式满足范数的定义,即
S
S
S 可视为赋范线性空间。最后说明
d
d
d 是由
∥
⋅
∥
\lVert\cdot\rVert
∥⋅∥ 导出的距离:
∀
x
,
y
∈
S
,
d
(
x
,
y
)
=
∣
∣
x
−
y
∣
∣
=
d
(
x
−
y
,
0
)
\forall~x,y\in{S},\qquad d(x,y)=||x-y||=d(x-y,0)
∀ x,y∈S,d(x,y)=∣∣x−y∣∣=d(x−y,0)
结合上述两条定理,可得如下包含关系: