Metric Spaces

September 26, 2021

1 Metric Spaces
1.1 Metric
De…nition 1 Let X be a set. A function d : X X ! R is a metric on X if
d satis…es the following conditions for all x; y; z 2 X.
(M 1) d (x; y) 0 and d (x; y) = 0 () x = y,
(M 2) d (x; y) = d (y; x),
(M 3) d (x; y) d (x; z) + d (z; y) k y k, (Triangle inequality).

De…nition 2 If d is a metric on X, then we call (X; d) a metric space.

Example 3 The set R of real numbers is a metric space with the metric

d (x; y) = jx yj

It is easy to see that d satis…es the …rst two conditions of the metric. If x; y; z 2
R, then

d (x; y) = jx yj = j(x z) + (z y)j jx zj + jz yj = d (x; z) + d (z; y)

Thus the triangle inequality also holds.

Example 4 The plane R2 is a metric space. We denote the point x 2 R2 by

x = (x1 ; x2 ). The usual metric onR2 is de…ned by
q
2 2
d (x; y) = (x1 y1 ) + (x2 y2 )

It is easy to see that d satis…es the …rst two conditions of the metric. We will
prove shortly d also satis…es triangle inequality.

Example 5 There are other metrics on the set R2 . As an example we de…ne

dt (x; y) = jx1 y1 j + jx2 y2 j

It is easy to see that dt is also a metric on R2 . This called the taxi cab metric.

1
Example 6 The n-dimensional real vector space Rn is a metric space. We
denote the point x 2 R2 by x = (x1 ; x2 ; : : : ; xn ). The usual metric on Rn is
de…ned by
q
2 2 2
d (x; y) = (x1 y1 ) + (x2 y2 ) + + (xn yn )

It is easy to see that d satis…es the …rst two conditions of the metric. Triangle in-
equality follows from Cauchy-Scwartz inequality. It states that if a1 ; a2 ; :::; an ; b1 ; b2 ; :::; bn 2
R, then
n
!2 n
! n !
X X X
2 2
ai bi ai bi
i=1 i=1 i=1
n
X n
X
To see this we let A = a2i and B = b2i . If A = 0, then ai = 0 for
i=1 i=1
i = 1; 2; : : : ; n. Thus equality holds. So we assume without loss of generality
that A > 0. Let
Xn
C
C= ai bi and t =
i=1
A
Then
n
X n
X
2
0 (bi tai ) = t2 a2i 2tai bi + b2i
i=1 i=1
= t2 A2tC + B
2
C AB C 2
= A t +
A A
2
AB C
=
A
Thus 0 AB C 2 and hence C 2 AB as required.
Now let x; y; z 2 Rn . Let ai = xi zi and bi = zi yi . Then
n
X
2 2
d (x; y) = (ai + bi ) = A + 2C + B
i=1

where A; B; C have the same meaning as above. Applying Cauchy-Scwartz in-

equality we obtain

2
p p p 2
d (x; y) A + 2 AB + B = A+ B

Thus p p
d (x; y) A+ B = d (x; z) + d (z; y)

2
1.2 Normed-Space
A vector space by convention is a vector space over R or C. We will use the
notation F to mean R or C. If X is a vector space over F and F = R we say X
is a real vector space, if F = C we say X is a complex vector space.

De…nition 7 Let X be a vector space over F. A norm is a real-valued function

on X and satis…es the following conditions for all x; y 2 X and 2 F:
(N 1) kxk 0, ( positive) and kxk = 0 () x = 0,
(N 2) k xk = j j kxk, ( homogeneous)
(N 3) kx + yk kxk + kyk, (Triangle inequality).

De…nition 8 A normed space is a pair (X; k:k) where X is a vector space

over F and k:k is a norm on X.

A normed space (X; k:k) is also a metric space. If we de…ne

d (x; y) = kx yk

for x; y 2 X, then d is a metric on X making X a metric space. This metric is

said to be induced by norm.

Example 9 Let S be any non-empty set and X = Fb (S; R) be the vector space
of bounded functions from S into R. For f 2 X de…ne

kf k = sup jf (x)j
x2S

It is easy to see that k k satis…es the …rst two conditions of the norm. Let
f; g 2 X. For xS we have

jf (x) + g (x)j jf (x)j + jg (x)j kf k + kgk

Thus
kf + gk = sup jf (x) + g (x)j kf k + kgk
x2S

Thus k k satis…es the triangle inequality. This norm is called the sup norm.

Example 10 A special case of the above example is C ([a; b] ; R) where [a; b] is a

closed interval in R and C ([a; b] ; R) is the vector space of continuous functions
from [a; b] into R. We know that a real valued continuous function de…ned on
a closed interval is bounded. So C ([a; b] ; R) is actually is vector subspace of
Fb ([a; b] ; R). So the same norm, i.e. sup norm is also a norm for C ([a; b] ; R).
If we recall the fact that a real valued continuous function de…ned on a closed
interval attains its maximum we may express the norm as

kf k = max jf (x)j
x2[a;b]

3
Example 11 Let X be a real inner product space with inner product ( ; ). Re-
call that in this case we have the Cauchy-Scwartz inequality

j(x; y)j kxk kyk

p
where kxk = (x; x). We also recall that we may prove using Cauchy-Scwartz
inequality that kxk is a norm on X. So every inner product space is a metric
space where the metric is the one induced by norm. Give an inner product space
we assume that it is endowed with this metric unless otherwise stated.