Chapter 9

Metric Spaces
Therefore, either the reality on which our space is based must form a dis-
crete manifold or else the reason for the metric relationships must be sought
for, externally, in the binding forces acting on it.
Bernhard Riemann (18261866)
Let X be a set. A function
d : X X R
+
is called a metric on X if the following hold:
(M1) d(x, y) = 0 x = y,
(M2) d(x, y) = d(y, x) for all x, y X (Symmetry), and
(M3) d(x, y) d(x, z) + d(z, y) for all x, y, z X (Triangle Inequality).
If d is a metric on X, then (X, d) is called a metric space.
Let (X, d) be a metric space. A sequence (x
n
) X converges to an element x X if for all
> 0 there exists an N N such that d(x
n
, x) < whenever n N.
A sequence in a metric space (X, d) is a Cauchy sequence if for all > 0 there exists an N N
such that d(x
n
, x
m
) < whenever n, m N.
A metric space (X, d) is called complete if every Cauchy sequence in X converges to an
element of X.
181
A.G. Aksoy, M.A. Khamsi, A Problem Book in Real Analysis, Problem Books in Mathematics,
DOI 10.1007/978-1-4419-1296-1_9, Springer Science+Business Media, LLC 2010
182 CHAPTER 9. METRIC SPACES
Problem 9.1 Show that the following functions dene a metric on R.
(a) d(x, y) = |x y|
(b) d(x, y) =
_
|x y|
Problem 9.2 Show that the function
d(x, y) =
_
0 if x = y,
1 if x = y
is a metric on any nonempty set X. (This is called the discrete metric on X.)
Problem 9.3 Let X = R
2
be the set of points in the plane. Show that for x, y X with
x = (x
1
, x
2
) and y = (y
1
, y
2
),
d(x, y) =
_
(x
1
y
1
)
2
+ (x
2
y
2
)
2
is a metric on X, also known as the Euclidean metric on R
2
.
Problem 9.4 Let X = R
2
. Show that for x = (x
1
, x
2
), y = (y
1
, y
2
) X, the function
d(x, y) =
_
|x
1
y
1
| if x
2
= y
2
,
|x
1
| +|x
2
y
2
| +|y
1
| if x
2
= y
2
is a metric on X.
Problem 9.5 Let X be the set of continuous functions from [a, b] to R. For all x, y X dene
d(x, y) to be
d(x, y) = max{|x(t) y(t)| : t [a, b]}.
Show that (X, d) is a metric space.
Problem 9.6 Let X be the set of bounded functions from some set A to R. For x, y X dene
d(x, y) by
d(x, y) = sup{|x(t) y(t)| : t A}.
Show that (X, d) is a metric space.
183
Problem 9.7 Let X be the set of continuous functions from [a, b] into R. For x, y X dene
d(x, y) by
d(x, y) =
_
b
a
|x(t) y(t)| dt.
Prove that (X, d) is a metric space.
Problem 9.8 Let (X, d) be a metric space. Show for all x, y, z, w X we have

d(x, z) d(z, y)

d(x, y)
and

d(x, y) d(z, w)

d(x, z) + d(y, w).

Problem 9.9 Let (X, d) be a metric space. Suppose is dened by
(x, y) =
d(x, y)
1 + d(x, y)

Show that is also a metric on X.

(Note that the new metric is bounded because (x, y) < 1 for all x, y X.)
Problem 9.10 Let (X, d) be a metric space. Let p be a point in X. The SNCF metric d
p
is
dened by
d
p
(x, y) :=
_
0 if x = y,
d(x, p) + d(p, y) otherwise.
Show that d
p
is a metric.
CHAPTER 9. METRIC SPACES
184 CHAPTER 9. METRIC SPACES
Problem 9.11 Let (X, d) be a metric space. The metric d is called an ultrametric if
d(x, y) max
_
d(x, z), d(y, z)
_
for any x, y, z X.
(i) Show that the Euclidean metric (see Problem 9.3) on R
n
, for n 2, is not an ultrametric.
(ii) Let p be any positive prime number p 2. For any nonzero x Q, there exists a unique
n Z such that
x = p
n
u
v
with some integers u and v indivisible by p. Set |x|
p
= n. Show that
d
p
(x, y) :=
_

_
0 when x = y,
_
1
p
_
|xy|
p
otherwise
is an ultrametric distance on Q, also known as the p-adic metric.
Problem 9.12 Let (X, d) be an ultrametric space (see Problem 9.11). Show that
(i) every triangle in X is an isosceles triangle;
(ii) every point inside a ball is its center.
Problem 9.13 Let (X, d) be a metric space. Show that if {x
n
} and {y
n
} are Cauchy sequences
of X, then {d(x
n
, y
n
)} is a Cauchy sequence in R, which implies that {d(x
n
, y
n
)} is convergent.
Problem 9.14 Show that every convergent sequence in a metric space (X, d) is a Cauchy
sequence.
Problem 9.15 Give an example of a metric space (X, d) and a Cauchy sequence {x
n
} X
such that {x
n
} does not converge in X.
Problem 9.16 Let (X, d) be a metric space. Show that X consists of one point i any bounded
sequence in X is convergent.
Problem 9.17 Let (X, d) be a metric space. Let {x
n
} be a sequence in X such that any
subsequence of {x
n
} has a subsequence which converges to some xed point x X. Show that
{x
n
} converges to x.
185
Problem 9.18 Let P be the set of all polynomials (of all degrees) dened on [0, 1]. Dene
d(x, y) by
d(x, y) = max
0 t 1
|x(t) y(t)|.
Prove that (P, d) is not a complete metric space.
Problem 9.19 The open ball B(x; r) in a metric space (X, d) is dened by
B(x; r) := {y X : d(x, y) < r}.
B(x; r) is called the unit ball if r = 1. Draw the unit balls centered at (0, 0) in R
2
with respect
to the metrics
(a) d
1
(x, y) =
_
(x
1
y
1
)
2
+ (x
2
y
2
)
2
(b) d
2
(x, y) = |x
1
y
1
| +|x
2
y
2
|
(c) d
3
(x, y) = max(|x
1
y
1
|, |x
2
y
2
|)
Problem 9.20 Show that B(x; r) in a Euclidean space is convex.
Problem 9.21 In a metric space (X, d), given a ball B(x
0
; r), show that for any x B(x
0
; r),
B(x; s) B(x
0
; r) for all 0 < s r d(x, x
0
).
Problem 9.22 Describe a closed ball, open ball, and sphere with center x
0
and radius r in a
metric space with the discrete metric.
Problem 9.23 Given a metric space (X, d) and a nonempty bounded subset A, the real number
(A) := sup{d(x, y) : x, y A} is called the diameter of A. It is clear that (S(x
0
; r)) 2r.
Show that equality is not always valid.
Problem 9.24 Let {I
n
} be a sequence of bounded nonempty closed subsets of a complete
metric space (X, d) such that
(a) I
n+1
I
n
, for all n 1;
(b) lim
n
(I
n
) = 0, where (A) = sup{d(x, y) : x, y A}.
Show that

n1
I
n
is not empty and reduced to a single point.
CHAPTER 9. METRIC SPACES
186 CHAPTER 9. METRIC SPACES
Solutions
Solution 9.1
(a) Since |x y| = |y x|, (M2) is satised. Note that
|x y| = 0 x y = 0 x = y,
(M1) is satised. To show (M3), we will make use of the inequality |a + b| |a| +|b| for any
a, b R. Indeed, we have
d(x, y) = |x y| = |(x z) + (z y)| |x z| +|z y| = d(x, z) + d(z, y)
for any x, y, z R.
(b) Since
_
|x y| =
_
|y x|, (M2) is satised. As in part (a),
_
|x y| = 0 |x y| = 0 x = y,
so (M1) is satised. Note that if a, b 0, then a b if and only if a
2
b
2
. Thus, showing
(M3) is equivalent to showing (
_
|x y|)
2
(
_
|x z| +
_
|z y|)
2
. So
(
_
|x y|)
2
= |x y|
|x z| +|z y|
|x z| +|z y| + 2
_
|x z|
_
|z y|
=
_
_
|x z| +
_
|z y|
_
2
which completes the proof of (M3).
Solution 9.2
(M1) and (M2) follow straight from the denition. For (M3), suppose x, y, z X.
If x = y, then d(x, y) d(x, z) + d(z, y) is clear.
If x = y, then either z = x or z = y:
If z = x, then d(x, y) = 1 = d(x, z) d(x, z) + d(z, y).
If z = y, then d(x, y) = 1 = d(y, z) d(x, z) + d(z, y).
Solution 9.3
We show that (M1), (M2), and (M3) hold:
(M1) d(x, y) = 0
_
(x
1
y
1
)
2
+ (x
2
y
2
)
2
= 0 x
1
= y
1
and x
2
= y
2
x = y.
(M2) d(x, y) =
_
(x
1
y
1
)
2
+ (x
2
y
2
)
2
=
_
(y
1
x
1
)
2
+ (y
2
x
2
)
2
= d(y, x).
187
(M3) Let x = (x
1
, x
2
), y = (y
1
, y
2
), and z = (z
1
, z
2
) be in R
2
. Note that the quadratic function
t
2
+t+ has a constant sign if and only if
2
4 0 which is equivalent to || 2()
1/2
.
Now observe that the following function (of t) is never negative:
F(t) = [(x
1
z
1
)t + (z
1
y
1
)]
2
+ [(x
2
z
2
)t + (z
2
y
2
)]
2
.
But F(t) = t
2
+ t + where
_

_
= (x
1
z
1
)
2
+ (x
2
z
2
)
2
,
= 2
_
(x
1
z
1
)(z
1
y
1
) + (x
2
z
2
)(z
2
y
2
)
_
,
= (z
1
y
1
)
2
+ (z
2
y
2
)
2
.
So, we must have || 2()
1/2
. But
()
1/2
= 2
_
[(x
1
z
1
)
2
+ (x
2
z
2
)
2
][(z
1
y
1
)
2
+ (z
2
y
2
)
2
]
_
1/2
= 2d(x, z)d(z, y),
so

(x
1
z
1
)(z
1
y
1
) + (x
2
z
2
)(z
2
y
2
)

d(x, z)d(z, y).

Therefore,
[d(x, y)]
2
= (x
1
y
1
)
2
+ (x
2
y
2
)
2
= [(x
1
z
1
) + (z
1
y
1
)]
2
+ [(x
2
z
2
) + (z
2
y
2
)]
2
(x
1
z
1
)
2
+ (x
2
z
2
)
2
+ 2d(x, z)d(z, y) + (z
1
y
1
)
2
+ (z
2
y
2
)
2
= [d(x, z)]
2
+ 2d(x, z)d(z, y) + [d(z, y)]
2
= [d(x, z) + d(z, y)]
2
.
Thus, d(x, y) d(x, z) + d(z, y), which completes the proof of (M3).
Note that the same proof will show that the function d dened on R
n
by
d
_
(x
i
), (y
i
)
_
=
_
(x
1
y
1
)
2
+ + (x
n
y
n
)
2
is a distance on R
n
also known as the Euclidean distance.
Solution 9.4
Before we check that (M1)(M3) hold, note that if x = (x
1
, x
2
) and y = (y
1
, y
2
), then we have the
inequality |x
1
y
1
| d(x, y). This inequality will be useful in the proof of (M1)(M3).
(M1) Assume d(x, y) = 0. If x
2
= y
2
, then we have
d(x, y) = |x
1
| +|x
2
y
2
| +|y
1
| = 0
which implies |x
2
y
2
| = 0 or x
2
= y
2
, which is a clear violation with our assumption. So we
must have x
2
= y
2
. In this case, we have
d(x, y) = |x
1
y
1
| = 0
which implies x
1
= y
1
. In other words, we have x = y.
CHAPTER 9. METRIC SPACES
188 CHAPTER 9. METRIC SPACES
(M2) We observe that
d(x, y) =
_
|x
1
y
1
| if x
2
= y
2
|x
1
| +|x
2
y
2
| +|y
1
| if x
2
= y
2
=
_
|y
1
x
1
| if y
2
= x
2
|y
1
| +|y
2
x
2
| +|x
1
| if y
2
= x
2
= d(y, x).
(M3) Let x = (x
1
, x
2
), y = (y
1
, y
2
), and z = (z
1
, z
2
). We examine two cases:
(1) If x
2
= y
2
, then
d(x, y) = |x
1
y
1
| |x
1
z
1
| +|z
1
y
1
| d(x, z) + d(z, y).
(2) If x
2
= y
2
, then z
2
= x
2
or z
2
= y
2
. Without loss of generality, assume z
2
= x
2
. Then
d(x, y) = |x
1
| +|x
2
y
2
| +|y
1
|
|x
1
| +|x
2
z
2
| +|z
2
y
2
| +|y
1
|

_
(|x
1
| +|x
2
z
2
| +|z
1
|) +|z
1
y
1
| if y
2
= z
2
(|x
1
| +|x
2
z
2
| +|z
1
|) + (|z
1
| +|z
2
y
2
| +|y
1
|) if y
2
= z
2
= d(x, z) + d(z, y).
Solution 9.5
We check that (M1)(M3) hold:
(M1) First note that d(x, y) = max{|x(t) y(t)| : t [a, b]} = 0 if and only if for any t [a, b] we
have |x(t) y(t)| = 0, which implies x(t) = y(t). So d(x, y) = 0 if and only if x(t) = y(t) for
all t [a, b], i.e., x = y.
(M2) d(x, y) = max{|x(t) y(t)| : t [a, b]} = max{|y(t) x(t)| : t [a, b]} = d(y, x).
(M3) Let x, y, z X. For any s [a, b] we have
|x(s) y(s)| |x(s) z(s)| +|z(s) y(s)| max
t[a,b]
|x(t) z(t)| + max
t[a,b]
|z(t) y(t)|.
Since s was arbitrarily taken in [a, b], we have
max
t[a,b]
|x(t) y(t)| max
t[a,b]
|x(t) z(t)| + max
t[a,b]
|z(t) y(t)|,
or d(x, y) d(x, z) + d(z, y).
Solution 9.6
The proofs of (M1), (M2), and (M3) follow exactly the proofs given in the previous problem.
Solution 9.7
We check that (M1) through (M3) hold:
189
(M1) The key behind the proof of (M1) is the following property
_
b
a
|f(t)|dt = 0 if and only if
f(t) = 0 for all t [a, b] provided f(x) is a continuous function. Hence for any x, y X, we
have
d(x, y) =
_
b
a
|x(t) y(t)| dt = 0 |x(t) y(t)| = 0 for all t [a, b],
which easily implies x = y.
(M2) For any x, y X, we have
d(x, y) =
_
b
a
|x(t) y(t)| dt =
_
b
a
|y(t) x(t)| dt = d(y, x).
(M3) For x, y, z X, we have
d(x, y) =
_
b
a
|x(t) y(t)| dt
=
_
b
a
_
|x(t) z(t)| +|y(t) z(t)|
_
dt

_
b
a
|x(t) z(t)| dt +
_
b
a
|y(t) z(t)| dt
= d(x, z) + d(y, z).
Solution 9.8
From (M3) we get the inequalities
d(x, z) d(x, y) + d(y, z) d(x, z) d(y, z) d(x, y)
and
d(y, z) d(y, x) + d(x, z) d(y, z) d(x, z) d(y, x).
Together, these inequalities imply

d(x, z) d(z, y)

d(x, y).
For the other inequality note that we have

d(x, y) d(y, z)

d(x, z)
and

d(y, z) d(z, w)

d(y, w).
Hence

d(x, y) d(z, w)

d(x, y) d(y, z)

d(y, z) d(z, w)

d(x, z) + d(y, w).

Solution 9.9
Before we check that (M1), (M2) and (M3) hold, note that is well dened since 1 + d(x, y) = 0.
CHAPTER 9. METRIC SPACES
190 CHAPTER 9. METRIC SPACES
(M1) (x, y) = 0 i the numerator d(x, y) = 0, which is true i x = y.
(M2) It is clear from the symmetry of d that (x, y) = (y, x).
(M3) Let a, b, and c be positive numbers such that a b + c. Then we have
a b + c b + c + 2bc + abc
which implies
a + a(b + c) + abc a(b + c) + b + c + 2bc + 2abc.
But
a + a(b + c) + abc = a[1 + b][1 + c]
and
a(b + c) + b + c + 2bc + 2abc = b[1 + a][1 + c] + c[1 + a][1 + b].
Hence
a[1 + b][1 + c] b[1 + a][1 + c] + c[1 + a][1 + b]
which implies
a
1 + a

b
1 + b
+
c
1 + c

Therefore, if we set a = d(x, y), b = d(x, z), and c = d(z, y), we know that a b + c since d
obeys the triangle inequality. Hence
a
1 + a

b
1 + b
+
c
1 + c
,
or
(x, y) (x, z) + (z, y).
Solution 9.10
Let us check that (M1), (M2) and (M3) hold.
(M1) From the denition of d
p
we know that d
p
(x, y) = 0 i x = y.
(M2) Since d(x, p) + d(p, y) = d(y, p) + d(p, x), we get d
p
(x, y) = d
p
(y, x).
(M3) Let x, y, z X. Without loss of generality, we may assume that the three points x, y, and z
are dierent. Then
d
p
(x, y) = d(x, p) + d(p, y) d(x, p) + d(p, z) + d(z, p) + d(p, y) = d
p
(x, z) + d
p
(z, y).
Solution 9.11
(i) Take x = (1, 0, . . . , 0), y = (0, 1, 0, . . . , 0), and O = (0, . . . , 0). Then we have d(x, y) =

2,
d(x, O) = 1, and d(y, O) = 1. Hence
d(x, y) max
_
d(x, O), d(y, O)
_
,
which implies that d is not an ultrametric on R
n
.
191
(ii) First let us show that the p-adic function d
p
is a metric on Q. Note that (M1) and (M2) hold
directly from the denition of d
p
. In order to prove (M3), we will prove
d
p
(x, y) max
_
d
p
(x, z), d
p
(y, z)
_
for any x, y, z Q. This will prove (M3) and the ultrametric property at the same time.
Indeed, this will follow from the inequality
max
_
d
p
(x, z), d
p
(y, z)
_
d
p
(x, z) + d
p
(y, z).
Let x, y, z Q be three dierent rationals. Set n = |x z|
p
, and m = |z y|
p
. Without loss
of generality, assume n m. By denition of | |
p
, we have x z = p
n
u
v
, and z y = p
m
u

,
which implies
x y = p
n
_
u
v
+ p
mn
u

_
= p
n
_
uv

+ p
mn
vu

vv

By denition of | |
p
, we get n |x y|
p
since vv

is indivisible by p. Since
_
1
p
_
|xy|
p

_
1
p
_
n
= max
_
_
1
p
_
n
,
_
1
p
_
m _
or
d
p
(x, y) max
_
d
p
(x, z), d
p
(y, z)
_
.
Solution 9.12
(i) Let x, y, z X. Let us prove that
d(x, y) = d(y, z), or d(x, z) = d(y, z), or d(x, y) = d(z, x).
Assume not. Without loss of generality assume d(x, y) < d(y, z) < d(x, z). But these inequal-
ities will contradict the fact
d(x, z) max{d(x, y), d(y, z)}.
(ii) Let r > 0 and x, y X. Assume that d(x, y) < r. Then B(x; r) = B(y; r), where B(x; r) =
{z X; d(x, z) < r}. We will only prove that B(y; r) B(x; r). Let z B(y; r). Since
d(x, z) max{d(x, y), d(y, z)} < r,
we get z B(x; r).
Solution 9.13
From Problem 9.8, we know that for any n, m N we have

d(x
n
, y
n
) d(x
m
, y
m
)

d(x
n
, x
m
) + d(y
n
, y
m
).
CHAPTER 9. METRIC SPACES
192 CHAPTER 9. METRIC SPACES
So, given > 0, there exist N
1
N and N
2
N such that for any n, m N
1
we have
d(x
n
, x
m
) <

2
,
and for any n, m N
2
we have
d(y
n
, y
m
) <

2

Set N = max{N
1
, N
2
}. Then for any n, m N we have

d(x
n
, y
n
) d(x
m
, y
m
)

d(x
n
, x
m
) + d(y
n
, y
m
) <

2
+

2
= .
This implies that {d(x
n
, y
n
)} is a Cauchy sequence in R.
Solution 9.14
Let {x
n
} be a convergent sequence in X with limit x. Then for each > 0 there exists some N
such that d(x
n
, x) < /2 for all n N. Using the triangle inequality (M3), it follows that
d(x
n
, x
m
) d(x
n
, x) + d(x, x
m
)
<

2
+

2
=
for all n, m N. Hence, {x
n
} is a Cauchy sequence.
Solution 9.15
Take X = Q with the metric d(x, y) = |x y|. Consider the sequence {x
n
} dened in Problem
3.19 by x
0
= 1 and
x
n+1
=
1
2
_
x
n
+
2
x
n
_
for all n 1. Then x
n
Q for all n N and limx
n
=

2 R\Q, as shown in Problem 3.19. Thus,
by the previous problem we have that {x
n
} is a Cauchy sequence in R, and hence in Q too, but
limx
n
Q.
Solution 9.16
Obviously, if X consists of one point, then any sequence in X is constant and therefore is convergent.
Conversely, assume that any bounded sequence in X is convergent. Let us prove that X consists
of one point. Assume not. Let x, y X with x = y. Consider the sequence {x
n
} dened by
x
2n
= x and x
2n+1
= y.
It is clear that {x
n
} is bounded and is not convergent. Contradiction.
Solution 9.17
Assume not. Then there exists
0
> 0 such that for any N 1, there exists n N with
|x
n
x|
0
. By induction, we construct a subsequence {x
n
k
} of {x
n
} such that for any k 1 we
have |x
n
k
x|
0
. It is clear no subsequence of {x
n
k
} will converge to x. Contradiction.
193
Solution 9.18
It is clear that (P, d) is a metric space (see Problem 9.5). To show (P, d) is not complete, consider
the following sequence:
x
n
(t) =
n

k=0
_
t
2
_
k
= 1 +
t
2
+ +
t
n
2
n
0 t 1.
Clearly, x
n
(t) P for each n N. Next we show that the sequence {x
n
} is a Cauchy sequence.
Taking m < n we observe
d(x
n
, x
m
) = max
0 t 1
|x
n
(t) x
m
(t)|
= max
0 t 1

k=0
_
t
2
_
k

k=0
_
t
2
_
k

= max
0 t 1

k=m+1
_
t
2
_
k

max
0 t 1

k=m+1
1
2
k

=
1
2
m

1
2
n
.
This dierence is arbitrarily small for large enough m and n, which implies that {x
n
} is a Cauchy
sequence in P. However, this sequence does not converge in (P, d), because the only candidate for
limx
n
(t) =
2
2 t
, for 0 t 1, and this is not a polynomial function. Since not every Cauchy
sequence converges in (P, d), we have that (P, d) is not complete.
Solution 9.19
See the following gure.
Figure 9.1
CHAPTER 9. METRIC SPACES
194 CHAPTER 9. METRIC SPACES
Solution 9.20
Let y
1
, y
2
B(x; r) and set z = y
1
+ (1 )y
2
, where 0 1. Then we have
d(x, z) = d(x, y
1
+ (1 )y
2
)
d(x, y
1
) + d(x, (1 )y
2
)
= d(x, y
1
) + (1 )d(x, y
2
).
Since y
1
, y
2
B(x; r), we have that d(x, y
1
) < r and d(x, y
2
) < r, which implies d(x, z) < r, and
therefore z B(x; r).
Solution 9.21
Let y B(x; s), then d(x, y) < s and by the triangle inequality we have
d(y, x
0
) d(y, x) + d(x, x
0
) < s + d(x, x
0
) r.
Solution 9.22
Let x
0
X, then the sphere centered at x
0
with radius 1 is
S(x
0
; 1) = {x X : d(x
0
, x) = 1} = X\{x
0
}.
For r > 0 and r = 1 , S(x
0
; r) = {x X : d(x
0
, x) = r} = . Let B[x
0
; r] and B(x
0
; r) denote the
closed and open ball with center x
0
and radius r (respectively). Then we have that
B[x
0
; r] = B(x
0
; r) = B[x
0
; 1] =
_
X for r > 1,
x
0
for 0 < r < 1.
Solution 9.23
Let (X, d) be a discrete metric space where X has more than one element. Then
S(x
0
; 1) = X\{x
0
} and (S(x
0
; 1)) = 1 < 2.
For r > 0 and r = 1, S(x
0
; r) = and (S(x
0
; r)) is not dened.
Solution 9.24
Since {I
n
} is a sequence of nonempty sets, form a sequence {x
n
} with x
n
I
n
for all n 1. We
claim that {x
n
} is a Cauchy sequence in X. Indeed, let > 0 and since lim
n
(I
n
) = 0, there exists
n
0
1 such that for all n n
0
we have (I
n
) < . Let m n n
0
, then x
n
, x
m
I
m
because
{I
n
} is decreasing. Then
d(x
n
, x
m
) (I
n
) < .
This proves our claim. Since X is complete we conclude that {x
n
} is convergent. Let x X be
its limit. We claim that

n1
I
n
= {x}. Indeed, for any n 1 and any m n we have x
m
I
n
.
195
Since the subsequence {x
m
}
mn
also converges to x and I
n
is closed, we conclude that x I
n
.
Since n was arbitrary, we get x

n1
I
n
. Hence

n1
I
n
is not empty. Let y

n1
I
n
. By denition
of the diameter, we get d(x, y) (I
n
) for all n 1. Since lim
n
(I
n
) = 0, we get d(x, y) 0 or
d(x, y) = 0 which implies y = x. This completes the proof of our statement.
CHAPTER 9. METRIC SPACES