Solutions to Problems in Introduction to Topology by Bert

Mendelson (Chapter 2)
Isaac Dobes
July 29, 2019

2
2.1
N/A

2.2
2.2.1
We need to show that dk : X × X → R defined as dk (x, y) = kd(x, y) satisfies the 4 conditions for metric spaces.
Observe that for any x, y, z ∈ X:
1. Since k > 0 and d : X × X → R is a metric, it follows that dk (x, y) = kd(x, y) ≥ 0

2. dk (x, y) = 0 ⇐⇒ kd(x, y) = 0 ⇐⇒ d(x, y) = 0 ⇐⇒ x = y

3. dk (x, y) = kd(x, y) = kd(y, x) = dk (y, x)
4. dk (x, z) = kd(x, z) ≤ k(d(x, y) + d(y, z)) = kd(x, y) + kd(y, z) = dk (x, y) + dk (y, z)

Thus, (X, d) is a metric space.

2.2.2
n
We are told that d00 : Rn × Rn → R is defined as d00 (x, y) = |xi − yi |. Observe that for any x, y, z ∈ Rn :
P
i=1

n
1. d00 (x, y) = |xi − yi | ≥ 0 ∀x, y ∈ Rn since |a − b| ≥ 0 ∀a, b ∈ R
P
i=1

n
2. d00 (x, y) = 0 ⇐⇒
P
|xi − yi | = 0 ⇐⇒ xi = yi ∀i ∈ [n] ⇐⇒ x = y
i=1

n n
3. d00 (x, y) = |yi − xi | = d00 (y, x)
P P
|xi − yi | =
i=1 i=1

n n n n n
4. d00 (x, z) =
P P P P P
|xi − zi | = |xi − yi + yi − zi | ≤ (|xi − yi | + |yi − zi |) = |xi − yi | + |yi − zi | =
i=1 i=1 i=1 i=1 i=1
d00 (x, y) + d (y, z)
00

Hence, (Rn , d00 ) is a metric space.

1
2.2.3
Observe that
 2

2
d(x, y) = max {|xi − yi |}
1≤i≤n

= max {|xi − yi |2 }
1≤i≤n
Xn
≤ |xi − yi |2
i=1
n
X
= (xi − yi )2
i=1

2
= d0 (x, y)
⇒ d(x, y) ≤ d0 (x, y). Moreover,
v
u n
uX
d0 (x, y) = t (xi − yi )2
i=1
v
u n
uX
=t |xi − yi |2
i=1
s  
≤ n max {|xi − yi |}
1≤i≤n

√ r
= n max {|xi − yi |}
1≤i≤n
√
= n · d(x, y)
√
Thus, d(x, y) ≤ d0 (x, y) ≤ n · d(x, y).
The next set of inequalities is easier to see, but note that
n
X
d(x, y) = max {|xi − yi |} ≤ |xi − yi | = d00 (x, y)
1≤i≤n
i=1

and
n
X  
00
d (x, y) = |xi − yi | ≤ n max {|xi − yi |} = n · d(x, y)
1≤i≤n
i=1
Hence, d(x, y) ≤ d00 (x, y) ≤ n · d(x, y).

2.2.4
Rb
We are told that d : C 0 ([a, b]) × c0 ([a, b]) → R is defined as d(f, g) = a
|f (t) − g(t)|dt. Observe that for any
f, g, h ∈ C 0 ([a, b]):
Rb
1. d(f, g) = a |f (t) − g(t)|dt ≥ 0 since |f (t) − g(t)| ≥ 0 ∀t ∈ [a, b]
2.
Z b
d(f, g) = 0 ⇐⇒ |f (t) − g(t)|dt = 0
a
⇐⇒ |f (t) − g(t)| = 0 ∀t ∈ [a, b]
⇐⇒ f (t) = g(t) ∀t ∈ [a, b]

2
 3. Z b Z b

Z b
d(f, g) = |f (t) − g(t)|dt = (−1) g(t) − f (t) dt = |g(t) − f (t)|dt = d(g, f )
a a a

4.
Z b
d(f, h) = |f (t) − h(t)|dt
a
Z b
= |f (t) − g(t) + g(t) − h(t)|dt
a
Z b

≤ |f (t) − g(t)| + |g(t) − h(t)| dt
a
Z b Z b
= |f (t) − g(t)|dt + |g(t) − h(t)|dt
a a
= d(f, g) + d(g, h)

Hence, (C 0 ([a, b]), d) is a metric space.

2.2.5
Note that C b (X) is the set of all bounded functions defined on the set X. We are told that d0 : C b ([a, b])×C b ([a, b]) →
R is defined as d0 (f, g) = sup {|f (x) − g(x)|}. Observe that for any f, g, h ∈ C b ([a, b]):
x∈[a,b]

1. d0 (f, g) = sup {|f (x) − g(x)|} ≥ 0 since |f (x) − g(x)| ≥ 0 ∀x ∈ [a, b]

x∈[a,b]

2.

d0 (f, g) = 0 ⇐⇒ sup {|f (x) − g(x)|} = 0

x∈[a,b]


⇐⇒ |f (x) − g(x)| = 0 ∀x ∈ [a, b] again, because |f (x) − g(x)| ≥ 0 ∀x ∈ [a, b]
⇐⇒ f (x) = g(x) ∀x ∈ [a, b]

3.

d0 (f, g) = sup {|f (x) − g(x)|} = sup = sup {|g(x) − f (x)|} = d0 (g, f )


(−1) g(x) − f (x)
x∈[a,b] x∈[a,b] x∈[a,b]

4.

d0 (f, h) = sup {|f (x) − h(x)|}

x∈[a,b]

= sup {|f (x) − g(x) + g(x) − h(x)|}

x∈[a,b]

≤ sup {|f (x) − g(x)| + |g(x) − h(x)|}

x∈[a,b]


since ∀x ∈ [a, b], |f (x) − h(x)| ≤ |f (x) − g(x)| + |g(x) − h(x)| by triangle inequality for real numbers
= sup {|f (x) − g(x)|} + sup {|g(x) − h(x)|}
x∈[a,b] x∈[a,b]
0 0
= d (f, g) + d (g, h)

Hence, C b ([a, b]), d0 is a metric space.

3
2.2.6
Observe that
Z b Z b Z b
d(f, g) = |f (t) − g(t)|dt ≤ sup {|f (t) − g(t)|}dt = d0 (f, g)dt = (b − a)d0 (f, g)
a a t∈[a,b] a

In particular, setting b := 1 and a := 0, we have d(f, g) ≤ d0 (f, g).

2.2.7
We are told that d : X × X → R is defined as d(x, x) = 0 and d(x, y) = 1 for any x 6= y. Observe that for any
x, y ∈ X:
1. d(x, y) ≥ 0 by definition

2. d(x, y) = 0 ⇐⇒ x = y by definition
3. x = y ⇐⇒ y = x ⇒ d(x, y) = 0 = d(y, x). On the other hand, x 6= y ⇐⇒ y 6= x ⇒ d(x, y) = 1 =
d(y, x).
4. If x = z, then d(x, z) = 0 ⇒ d(x, z) ≤ d(x, y) + d(y, z) since d(x, y), d(y, z) ≥ 0. If x 6= z, then d(x, z) = 1.
Let y ∈ X. Then exacly one of the following holds: (y = x ∧ y 6= z), (y = z ∧ y 6= x), or (y 6= x ∧ y 6= z);
i.e., we cannot have x = y = z because this would imply x = z. Hence, d(x, y) + d(y, z) ≥ 1 ⇒ d(x, z) ≤
d(x, y) + d(y, z).
Thus, (X, d) is a metric space.

2.2.8
Given p prime, we are told that d : Z × Z → R is defined as d(m, n) = 0 for m = n, and d(m, n) = p1t for m 6= n,
where t = t(m, n) is the unique integer such that m − n = pt · k (where k is not divisible by p). Obsere that for any
m, n, o ∈ Z:
1
1. d(m, n) ≥ 0 by definition (since 0 ≥ 0 and given p prime, for any integer t, pt > 0)
1
2. d(m, n) = 0 ⇐⇒ m = n by definition (again, since given p prime, for any integer t, pt > 0)

3. m = n ⇐⇒ n = m ⇒ d(m, n) = 0 = d(n, m). On the other hand, m 6= n implies that d(m, n) = p1r where
r = r(m, n) is the unique integer such that m − n = pr · a, where a ∈ Z such that a6 | p, and d(n, m) = p1s ,
where s = s(n, m) is the unique integer such that n − m = ps · b, where b ∈ Z such that b6 | p. Thus, it suffices
to show that r = s. Observe that pr a = m − n = −(n − m) ⇒ n − m = −pr a = pr (−a) ⇒ r = s. Hence,
d(m, n) = d(n, m).
4. We want to show that if m, n, o ∈ Z, then d(m, o) ≤ d(m, n) + d(n, o). ∃! r ∈ Z such that m − n = pr a,
where a ∈ Z such that a6 | p; similarly, ∃! s ∈ Z such that n − o = ps b, where b ∈ Z such that b6 | p. WLOG
suppose s ≤ r. Then m − o = (m − n) + (n − o) = pr a + ps b = ps (pr−s a + b) ⇒ m − o = pt c for some
integer t ≥ s and c ∈ Z such that c6 | p. Therefore, d(m, o) = p1t ≤ p1s = d(n, o) ≤ d(m, n) + d(n, o).

Thus, (Z, d) is a metric space.

4
2.3
2.3.1
Rb
We are told that X = C 0 ([a, b]), and we want to prove that I : (C 0 ([a, b]), d∗ ) → (R, d), with d∗ (f, g) = a |f (t) −
g(t)|dt, is continuous. Let  > 0 be given. Choose δ = . Then for any f, g ∈ C 0 ([a, b]) such that d∗ (f, g) < δ, we
have:
Z b Z b Z b Z b
|f (t) − g(t)|dt = d∗ (f, g) < δ = 


d(I(f ), I(g)) = f (t)dt − g(t)dt = f (t) − g(t) dt ≤
a a a a

⇒ I is continuous.

2.3.2
n n
We are told that for i = 1, ...n, (Xi , di ) and (Y, d0i ) are metric spaces, and that X =
Q Q
Xi and Y = Yi . X
i=1 i=1
and Y , equipped, respectively, with the metrics dX : X × X → R and dy : Y × Y → R, defined as dX (x, y) =
max {di (xi , yi )} and dY (x, y) = max {d0i (xi , yi )}, are metric spaces. Given that each fi : Xi → Yi are continuous,
1≤i≤n 1≤i≤n
we want to prove that F : X → Y defined as F (x) = F (x1 , ..., xn ) = (f1 (x1 ), ..., fn (xn )) is continuous.
Observe that for any F (x), F (y) ∈ Y , dY (F (x), F (y)) = max {d0i (fi (xi ), fi (yi ))} = d0j (fj (xj ), fj (yj )) for
1≤i≤n
some j ∈ [n]. Since each fi is continuous for i = 1, ..., n, this implies that given any  > 0, there exists a δ > 0
such that d0j (fj (xj ), fj (yj )) <  whenver dj (xj , yj ) < δ. Hence, given  > 0, we can always choose a δ > 0 so that
dY (F (x), F (y)) <  whenever dX (x, y) < δ ⇒ F : X → Y is continuous.

2.3.3
Given the metrics on R2 d and d0 , where d is defined as d((x1 , x2 ), (y1 , y2 )) = max {|xi − yi |} and d0 is the normal
1≤i≤2
Euclidean distance, we want to prove that f : R2 → R defined as f (x1 , x2 ) = x1 + x2 is continuous.
First we prove that f is continuous with the metric d on R2 . Let  > 0 be given. Choose δ = 2 . Then for any
(x1 , x2 ), (y1 , y2 ) ∈ R2 such that d((x1 , x2 ), (y1 , y2 )) < δ, we have:

|f (x1 , x2 ) − f (y1 , y2 )| = |x1 + x2 − (y1 + y2 )| = |(x1 − y1 ) + (x2 − y2 )| ≤ |x1 − y1 | + |x2 − y2 | < 2δ = 

⇒ f is continuous.
Now, we prove that f is continuous with the metric d0 on R2 . Let  > 0 be given. Choose δ = 2 . Then observe
that for any (x1 , x2 ), (y1 , y2 ) ∈ R2 such that d0 ((x1 , x2 ), (y1 , y2 )) < δ we have:
p
d0 ((x1 , x2 ), (y1 , y2 )) < δ ⇐⇒ (x1 − y1 )2 + (x2 − y2 )2 < δ
⇐⇒ (x1 − y1 )2 + (x2 − y2 )2 < δ 2
⇒ (x1 − y1 )2 < δ 2 − (x2 − y2 )2 ≤ δ 2 ∧ (x2 − y2 )2 < δ 2
p p
⇒ |x1 − y1 | = (x1 − y1 )2 < δ ∧ |x2 − y2 | = (x2 − y2 )2 < δ

⇒ |f (x1 , x2 ) − f (y1 , y2 )| = |x1 + x2 − (y1 + y2 )| ≤ |x1 − y1 | + |x2 − y2 | < 2δ =  ⇒ f is continuous.

2.3.4
I’m too lazy. Hopefully this omission does not come back to haunt me.

5
2.4
2.4.1
Recall that N ⊂ X is a neighborhood of a if N contains an open ball B(a; δ) ⊂ X centered at a with some radius
δ > 0. Let δ := 21 . Then since d(a, x) = 1 for any x ∈ X such x 6= a, B(a; δ) = {a} ⇒ B(a; δ) ⊆ {a} ⇒ {a} is a
neighborhood of a. Moreover, {a} constitutes a basis for the system of neighborhoods of a since for any neighborhood
N of a, N 3 a and a ∈ {a}. Now, let S be a subset of X. If p ∈ S, then {p} ⊆ S ⇒ S is a neighborhhod of p.

2.4.2
To show that f : R → R defined as (
0 x≤a
f (x) =
1 x>a
is discontinuous at a, we need to show that there exists 0 > 0
such that for every δ > 0 there is x ∈ B(a; δ) but
f (x) ∈/ B(f (a); 0 ). Let 0 := 21 . Then observe that a + 2δ − a = 2δ = 12 |δ| < δ, but |f (a + 2δ ) − f (a)| =
|1 − 0| = 1 ≥ 12 . Hence, ∀ δ > 0 ∃x ∈ B(a; δ) such that f (x) ∈ / B(a; 0 ); namely, x := a + 2δ .
Now, for any x ∈ R \ {0}, f is locally constant. Thus, it is clear that at any other point besides a, f is continuous.

2.4.3
(⇒) Suppose f is continuous. Then for each neighborhood M of a, f −1 (M ) is a neighborhood of a. If N ∈ Bf (a) ,
then N is a neighborhood of f (a); hence, it follows immediatley that f −1 (N ) is a neighborhood of a.
(⇐) Conversely, suppose that for every N ∈ Bf (a) , f −1 (N ) is a neighborhood of a. Then for any neighborhood M
of f (a), M contains an element B ∈ Bf (a) , which is a neighborhood of f (a). Hence, f −1 (M ) contains f −1 (B), a
neighborhood of a, which implies that f −1 (M ) is a neighborhood of a. Thus, f is continuous.

2.4.4
S S
(i) Observe that [a − , a + ] ⊇ (a − , a + ). Therefore, for any neighborhood N of a, N contains, for
>0 >0
S S S
some 0 > 0, the interval B(a; 0 ) ⊂ (a − , a + ) ⊆ [a − , a + ] ⇒ [a − , a + ] is a basis for the
>0 >0 >0
system of neighborhoods at a.

(ii) Let Ba := {B(a; , ) :  > 0 ∧  ∈ Q}. Then for any neighborhood N of a, N contains, for some 0 > 0, the
interval B(a; 0 ). Since 0 > 0, by the density of Q in R, there exists a rational number 1 > 0 so that 1 < 0 .
Hence, N ⊃ B(a; 1 ) and B(a; 1 ) ∈ Ba ⇒ Ba is a basis for the system of neighborhoods at a.
(iii) Let Ba := {B(a; n1 ) : n ∈ N}. Then for any neighborhood N of a, N contains, for some 0 > 0, the interval
B(a; 0 ). Since the sequence { n1 } converges to 0, there exists n0 ∈ N so that n ≥ n0 ⇒ n1 < 0 . Hence, for
any n ≥ n0 , N ⊃ B(a; n1 ) and B(a; n) ∈ Ba ⇒ Ba is a basis for the system of neighborhoods at a.

(iv) The reasoning in this subproblem is analogous to that in the previous subproblem (iii). The only difference is
that in this subproblem we require that n ≥ max{n0 , k}.

Now, assume for the sake of contradiction that in R there exists a finite collection of sets B˜a which forms a basis for the
system of neighborhoods at a. Since B˜a is finite, we may explicitly list its elements: suppose B˜a = {B1 , B2 , ..., Bn }.
Tn
Let B := Bi . Then B is a neighborhood of a and B ⊆ Bi for 1 ≤ i ≤ n. Moreover, there exists δ > 0 such that
i=1
the real interval B(a; δ) = (a − δ, a + δ) ⊆ B. Now, let δ ∗ := 2δ ; then N := B(a; δ ∗ ) $ B is a neighborhood of a
and there does not exists a Bi , 1 ≤ i ≤ n, so that Bi ⊆ B(a; δ ∗ ). Thus, we have a contradiction. Consequentially,
there does not exist a finite collection of subsets of R that can be a basis for the system of neighborhoods of a.

6
2.4.5
Given a ∈ X, we want to show that there exists a collection of neighborhoods {Bn }n∈N which constitutes a basis for
the system of neighborhoods at a. Let Bn := B(a; n1 ). Then for any neighborhood N of a, there exists  > 0 such
that N ⊇ B(a; ). Since n1 → 0 as n → ∞, there exists n0 ∈ N so that for any n ≥ n0 , n1 < , which implies that for
any n ≥ n0 , N ⊃ B(a; n1 ) = Bn ⇒ {Bn }n∈N constitutes a basis for the system of neighborhoods at a.

2.4.6
a, b ∈ X such that a 6= b ⇒ d(a, b) > 0; suppose d(a, b) = δ. Then let Na := B(a; 2δ ) and Nb := B(a; 2δ ). I claim
that Na ∩ Nb . To prove this claim, it suffices to show that d(a, x) < 2δ ⇒ d(b, x) > 2δ (because this is equivalent to
proving that x ∈ Na ⇒ x ∈ / Nb , and by symmetry we may conclude that x ∈ Nb ⇒ x ∈ / Na ).
Observe that d(a, x) < 2δ implies that:

d(a, b) ≤ d(a, x) + d(x, b) ⇐⇒ δ − d(a, x) ≤ d(x, b)

δ
⇒ δ − < δ − d(a, x) ≤ d(x, b)
2
δ
⇒ d(x, b) >
2

2.4.7
a ∈ X is a point a = (a1 , ..., an ) where ai ∈ Xi for i = 1, ..., n. Let B(a; δ) ⊂ X. Then,

B(a; δ) = {x ∈ X : d(a, x) < δ}

= {(x1 , ..., xn ) : xi ∈ Xi ∀i ∈ [n] ∧ max {di (ai , xi )} < δ}
1≤i≤n

= {(x1 , ..., xn ) : xi ∈ Xi ∀i ∈ [n] ∧ di (ai , xi ) < δ ∀i ∈ [n]}

Yn
= {xi ∈ Xi : di (ai , xi ) < δ}
i=1
Yn
= Bi (ai ; δ)
i=1

S n
Q
Given that Bai is a basis for the system of neighborhoods at ai , and that Ba = Bi , we want to show that
Bi ∈Bai i=1
Ba is a basis for the system of neighborhoods at a. Suppose N ⊂ X is a neighborhood of a. Then there exists δ > 0
Qn
such that N ⊇ B(a; δ) = Bi (ai ; δ). For each i ∈ [n], Bi (ai ; δ) is a neighborhood of Xi ⇒ for each i ∈ [n],
i=1
n
Q n
Q
Bi (ai ; δ) ⊇ Bi for some Bi ∈ Bai ⇒ B(a; δ) ⊇ Bi where Bi ∈ Bai ∀i ∈ [n] ⇒ B(a; δ) ⊇ B = Bi ∈ Ba .
i=1 i=1
Hence, Ba is a basis for the system of neighborhoods of a.
Now, for each i ∈ [n], let pi : X → Xi be the projection that maps pi (a) = ai . We want to show that for each
i ∈ [n], pi is continuous; i.e., we want to show that for every neighborhood M of pi (a), p−1 i (M )
is a neighborhood
of a. Let Ni be a neighborhood of pi (a) = ai . Then Ni ⊇ Bi (ai ; δ) ⇒ p−1 i (Ni ) ⊇ p −1
i B i (ai ; δ) = {(x1 , ..., xn ) :
xi ∈ Xi ∀i ∈ [n] ∧ di (ai , xi ) < δ} ⊃ B(a; δ) ⇒ pi is continuous.
Now, suppose f : Y → X is a continuous function. Then since for each i ∈ [n], pi is continuous, it follows
immediately that pi ◦ f is continuous. Conversely, suppose for each i ∈ [n], pi
◦ f is continuous. Then given b ∈ Y ,
for every  > 0 there exists δ > 0 such that (pi ◦ f )(B(b; δ)) ⊆ B (pi ◦ f )(b);  , for every

i = 1, ..., n. Given b ∈ Y ,
f (b) = a for some a ∈ X;
consequentially, for each
i ∈ [n], (p i ◦ f )(b) = p i f (b) = pi (a) = ai ⇒ ∀i ∈ [n],
B(ai ; ) = B (pi ◦ f )(b);  ⊇
(pi ◦ f ) B(b;
δ) . Hence, B(b; δ) = {y ∈ Y : d(b, y) < δ} ⊆ {y ∈ Y : f (y) =
x ∧ d(x, a) < } ⇒ f B(b; δ) ⊆ B f (b);  . Thus, f : Y → X is continuous.

7
2.4.8
We are told that f : R → R is continuous and that there exists a ∈ R such that f (a) > 0. We want to show that there
exists k > 0 and a closed interval F = [a − δ, a + δ] such that f (x) ≥ k ∀x ∈
F .

Recall that f is continuous at a iff ∀  > 0, ∃ δ > 0 such that f B(a; δ) ⊂ B f (a);  . Now, f (a) > 0 ⇒ ∃
k > 0 such that f (a) > k > 0 by the density of
R. Choose 
> 0 so that  < f (a) − k. Then by continuity
of
f at a, there exists δ
 > 0 such that f
B(a; δ ) ⊂ B f (a);  ⇐⇒ ∃ δ > 0 such that f (a − δ , a + δ ) ⊂
f (a) − , f (a) +  = k, 2f (a) − k ⇒ f (x) ≥ k ∀ x ∈ (a − δ , a + δ ). Choose δ > 0 so that δ < δ , and set
F := [a − δ, a + δ]. Then f (x) ≥ k ∀ x ∈ F .

2.5
2.5.1
 k

Q
We are given the metric space Xi , d where d(x, y) = max {di (xi , yi )}. a1 , a2 , ... are points in X where
i=1 1≤i≤k
an = (an1 , an2 , ..., ank ) and c = (c1 , c2 , ..., ck ) ∈ X. We want to show that lim an = c ⇐⇒ lim ani = ci for each
n→∞ n→∞
i ∈ [k].
(⇒) Suppose lim an = c. Then for every neighborhood V of c, there exists N ∈ N such that an ∈ V for n ≥ N .
n→∞
1
Therefore, for any m ∈ N, there exists N ∈ N such that an ∈ B(c; m ) for n ≥ N ; this implies that for any m ∈ N
1 1
there exists N ∈ N so that for any n ≥ N , d(an , c) < m ⇐⇒ max {di (ani , ci )} < m ⇒ lim d(ani , ci ) = 0 for
1≤i≤k n→∞
i = 1, 2, ..., k ⇒ lim ani = ci .
n→∞
(⇐) Suppose that for i = 1, 2, ..., k, lim ani = ci . Then for i = 1, 2, ..., k, for every neighborhood Vi of ci ,
n→∞
there exists Ni ∈ N such that ani ∈ Vi for n ≥ Ni ⇒ for any m ∈ N, there exists Ni (for i = 1, 2, ..., k) such that
1 1 1
ani ∈ B(ci , m ) for n ≥ Ni ⇒ for i = 1, 2, ..., k, di (ani , ci ) < m ⇒ max {di (ani , ci )} < m ⇒ lim d(an , c) =
1≤i≤k n→∞
0 ⇒ lim an = c.
n→∞

2.5.2
s
k k
Recall that d(x, y) = max {|xi − yi |}, d0 (x, y) = (xi − yi )2 , and d00 (x, y) =
P P
|xi − yi |; also, recall from
1≤i≤k i=1 i=1
√
exercise 2.2.2, d0 (x, y) ≤ n · d(x, y) and d00 (x, y) ≤ n · d(x, y).
Therefore, if {ai }i∈N is a sequence in Rk and lim an = a, then lim d(an , a) = 0 ⇒ lim d0 (an , a) ≤
√ √ n→∞ n→∞ n→∞
k · lim d(an , a) = k · 0 = 0 and lim d00 (an , a) ≤ k · lim d(an , a) = k · 0 = 0. Therefore, lim d(an , a) =
n→∞ n→∞ n→∞ n→∞
0 ⇒ lim d0 (an , a) = 0 and lim d00 (an , a) = 0. Moreover, from exercise 2.2.2, d(x, y) ≤ d0 (x, y) and d(x, y) ≤
n→∞ n→∞
d00 (x, y); therefore, d0 (an , a) = 0 or d00 (an , a) = 0 implies that d(an , a) = 0. Thus, lim d(an , a) = 0 ⇐⇒
n→∞
lim d0 (an , a) = 0 ⇐⇒ lim d00 (an , a) = 0.
n→∞ n→∞

2.5.3
Suppose the sequence {an }n∈N of points in the metric space (X, d) converges to the point a. Then lim d(an , a) =
n→∞
0 ⇐⇒ ∀, ∃ N ∈ N such that d(an , a) <  for all n ≥ N . If {ank }∞ k=1 is a subsequence of {an }, then recall that
nk is a strictly increasing sequence from N to N; thus, d(ank , a) <  for all nk ≥ N ⇒ lim d(ank , a) = 0 ⇐⇒
k→∞
lim ank = a.
k→∞

8
2.5.4
Let {ai }i∈N be a convergent sequence of real numbers that converges to a ∈ R. We want to show that {ai } is
bounded; i.e., we want to show that ∀  > 0, ∃ M > 0 such that |ai | ≤ M ∀ i ∈ N. {ai } converges to a ⇒ ∀
 > 0, ∃ N ∈ N such that |an − a| <  for all n ≥ N . |an − a| <  ⇒ |an | < |a| + ; therefore, given  > 0, let
M = max{|a1 |, |a2 |, ..., |an−1 |, |a| + }. Then M ≥ |ai | ∀ i ∈ N; i.e., {ai } is bounded.
Let {ai }i∈N be a non-decreasing sequence bounded above. Then for every i ∈ N, ai ≤ ai+1 and there exists
M ∈ R such that ai ≤ M . Since {ai } ⊂ R is bounded above, by the completeness axiom, there exists a least upper
bound of {ai }, which we denote as a. a = l.u.b.{ai } implies that ∀  > 0, ∃ an ∈ {ai } such that a − an <  ⇒ ∀
 > 0, ∃ N ∈ N such that |a − an | <  for n ≥ N ⇒ {ai } converges to a. The proof showing that a non-increasing
sequence bounded below converges to its greatest lower bound is analogous.

2.5.5
Omitted.

2.5.6
Recall that d(x, A) = g.l.b.{d(x, a) : a ∈ A} and d(y, A) = g.l.b.{d(y, a) : a ∈ A}. We want to show that
d(x, A) ≤ d(x, y) + d(y, A). Consider the following cases:
• Suppose x ∈ A. Then d(x, A) = 0 ≤ d(x, y) + d(y, A)

• Suppose x ∈
/ A but y ∈ A. Then d(x, A) = g.l.b.{d(x, a) : a ∈ A} ≤ d(x, y) ⇒ d(x, A) ≤ d(x, y) + d(y, A)
/ A. Then there exists x0 ∈ A and y 0 ∈ A such that d(x, A) = d(x, x0 ) and d(y, A) = d(y, y 0 ).
• Suppose x, y ∈
Hence,
d(x, A) = d(x, x0 ) ≤ d(x, y) + d(y, y 0 ) + d(y 0 , x0 ) = d(x, y) + d(y, A)

Therefore, after exhausting all cases, we have d(x, A) ≤ d(x, y) + d(y, A).

2.5.7
Given a nonempty subset A of the metric space (X, d), we want to prove that the function f : X → R defined by
f (x) = d(x, A) is continuous; i.e., we want to show that ∀  > 0, ∃ δ > 0 such that d(x, y) < δ ⇒ |f (x) − f (y)| < .
Recall from the previous exercise, for any x, y ∈ X, d(x, A) ≤ d(x, y) + d(y, A) ⇒ d(x, A) − d(y, A) ≤ d(x, y);
since x and y are arbitrary, we also have: d(y, A) − d(x, A) ≤ d(x, y). Therefore, |f (x) − f (y)| = |d(x, A) −
d(y, A)| ≤ d(x, y); thus, given  > 0, letting δ :=  we have d(x, y) < δ ⇒ |f (x) − f (y)| < . That is, f is
continuous.

2.5.8
Give a nonempty subset A of the metric space (X, d) and a point x ∈ X, we want to show that d(x, A) = 0 if and
only if every neighborhood of x contains a point y ∈ A.
(⇒) d(x, A) = 0 ⇐⇒ glb{d(x, A)} = 0 ⇒ ∀  > 0, ∃ y ∈ A such that d(x, y) < . Given a neighborhood M
of x, ∃ δ > 0 such that B(a; δ) ⊂ M ; therefore, for any neighborhood M of a, ∃ y ∈ A such that y ∈ B(a; δ) ⊂ M .
(⇐) Suppose every neighborhood M of x contains a point y ∈ A. Then for every n ∈ N, there exists y ∈ A
such that y ∈ B(a; n1 ). Since ∀  > 0, ∃ m ∈ N such that m 1
< , this implies that ∀ > 0, ∃ y ∈ A such that
d(x, y) <  ⇒ d(x, A) = 0.

2.5.9
Omitted.

9
2.6
2.6.1
n
Q
Given that (Xi , di ), i = 1, 2, ..., n, are metric spaces, we form the set X = Xi equipped with the metric d :
i=1
X × X → R≥0 defined as d(x, y) = max {di (xi , yi )}. We want to prove that for i = 1, 2, ..., n, if Oi is an open
1≤i≤n
× n
subset of Xi , then i=1 Oi is an open subset of X.
× n
i=1
= {(x1 , x2 , ..., xn ) : xi ∈ Oi , i = 1, 2, ..., n}. Since xi ∈ Oi , ∃ δi > 0 such that B(xi ; δi ) ⊂ Oi
for i = 1, 2, ..., n. Let δ := 21 min {δi }. Then, δ > 0 and B(xi ; δ) ⊂ Oi for i = 1, 2, ..., n ⇒ B(x; δ) =
1≤i≤n
×n


B (x1 , x2 , ..., xn ); δ ⊂ O1 × O2 × ... × On ⇒ i=1 Oi is open.
n
×


Oα , where Oiα are open sets for
S
Now, suppose O is an open subset of X; we want to show that O = i=1 i
α∈I
i = 1, 2..., n and every α ∈ I. Since O ⊂ X, there are sets Ai ⊂ Xi , i = 1, 2, ..., n, such that O = A1 ×A2 ×...×An .
If x = (x1 , x2 , ..., xn ) ∈ O, then there exists δ > 0 so that B(x; δ) ⊂ O. Hence, for i = 1, 2, ..., n, B(xi ; δ) ⊂ Ai ⇒
each set Ai , i = 1, 2, ..., n, is open. Since the arbitrary
Sunionn of open
sets is open, this implies that there exists open
subsets Oiα , α ∈ I and i = 1, 2, ..., n, such that O =
α∈I
× O
i=1 i
α
.

2.6.2
Given the metric space (X, d) with metric d : X × X → R≥0 defined as
(
0 x=y
d(x, y) =
1 x 6= y

we want to prove that every subset of X is open.

Let A ⊂ X. Then if a ∈ A, ∀ x ∈ X \ {a}, d(a, x) = 1 ⇒ B(a; 21 ) = {a} ⊂ A ⇒ A is open. QED.

2.6.3
We are told that (X, d1 ) and (Y, d2 ) are metric spaces, and we form the metric space (X × Y, d) where d : (X × Y ) ×
(X × Y ) → R≥0 is defined as d(a, b) = max {di (ai , bi )}. Given that f : X → Y is continuous, we want to show
1≤i≤2
that the graph of f , Γf = {(x, f (x)) : x ∈ X}, is closed.
Let {(xn , f (xn ))}∞
n=1 be a sequence of points in Γf which converges to the point (x, y) ∈ X×Y . Then lim xn =
n→∞
x and lim f (xn ) = y. x ∈ X, hence by continuity of f , lim f (xn ) = f (x) ⇒ f (x) = y ⇒ (x, y) ∈ Γf . Thus, Γf
n→∞ n→∞
is closed.

2.6.4
We are told that f : R → R is defined as: (
1
x x>0
f (x) =
0 x 6= 0

and we want to show that Γf is a closed subset of (R2 , d), but that f is not continuous.
f |(−∞,0) (x) = 0 and f |(0,∞) (x) = x1 , thus it is clear that f is continuous on (−∞, 0) and (0, ∞) ⇒ Γf |(−∞,0] and


Γf |(0,∞) are closed. Moreover, {0, f (0)} = {(0, 0)} ⇒ CR2 ((0, f (0)) = (−∞, 0) ∪ (0, ∞), (−∞, 0) ∪ (0, ∞) =
R \ {0} × R \ {0}, which is open since the finite product of open sets is also open. Hence, {0, f (0)} is closed. Since
the finite union of closed sets is closed, Γf |(−∞,0) ∪ {(0, f (0))} ∪ Γf |(0,∞) = Γf is closed.
Note, however, that f is not continuous; in particular, f is discontinuous at x = 0. Observe that the sequence
{xn = n1 }∞ 1 1
n=1 converges to 0, but, f (xn ) = f ( n ) = 1 = n → ∞ = 6 f (0) = 0 ⇒ f is not continuous at 0.
n

10
2.6.5
We are told that A is a non-empty, closed subset of R, and that A is bounded below. By the completeness axiom, there
exists a greatest lower bound of A, α ∈ R, where α ≤ a ∀a ∈ A. We want to show that A 3 α.
Since α is the greateast lower bound of A, for any  > 0, there exists a ∈ A such that a < α + . Hence, ∀ n ∈ N,
n→∞
∃ an ∈ A such that an < α + n1 ⇒ an −−−−→ α. A closed ⇐⇒ any sequence in A which converges to a point
x ∈ R implies that x ∈ A; thus, A 3 α.

2.6.6
Recall that A0 = {x ∈ X : ∀  > 0, ∃ y ∈ A s.t. y 6= x ∧ y ∈ B(x; )}, and Ai = {a ∈ A : ∃ δ > 0 s.t. B(a; δ)∩A =
a}. Thus, it immediately follows that A0 ∩ Ai = ∅. If x ∈ A, then given any  > 0, either B(x; ) ⊆ {x}, or there
exists y ∈ A such that B(x; ) ⊇ {x, y}. If no such y exists for any  > 0, then x ∈ Ai , otherwise x ∈ A0 . That is,
A ⊆ A0 ∪ Ai .
Now, let A = A0 ∪ Ai . Then we want to show that x ∈ A if and only if there exists {an }∞ n=1 ⊂ A such that
n→∞
an −−−−→ x. So, suppose x ∈ A. Then x ∈ A0 or x ∈ Ai , but x ∈ / A0 ∩ Ai . If x ∈ A0 then ∀  > 0, there exists y ∈ A
such that y 6= x and y ∈ B(x; ); equivalently, ∀ n ∈ N, ∃ A 3 y := an such that an 6= x and an ∈ B(x; n1 ) ⇒ there
exists {an } ⊂ A such that lim an = x. Alternatively, if x ∈ Ai , then ∃ δ > 0 such that B(x; δ) ∩ A = {x}. Observe
n→∞
that the sequence {x}∞ ∞
n=1 = x, x, ... ⊂ A and lim x = x. Conversley, if {an }n=1 ⊂ A such that lim an = x, then
n→∞ n→∞
∀  > 0, ∃ N ∈ N such that n ≥ N implies that an ∈ B(x; ) ⇒ x ∈ A0 ⊂ A.
Now, let F be a closed set such that F ⊃ A. Then F closed ⇐⇒ F contains all of its limits points. Since
A ⊂ F , this implies that F contains all limit points of A ⇒ A0 ⊂ F . Furthermore, since Ai ⊂ A ⊂ F , this implies
that F ⊃ ATi . Hence, A = A0 ∪ Ai ⊂ F . Since F is an arbitrary closed set containing A, and A ⊂ F , this implies that
A⊆ F ; moreover, since A ⊃ A and A is closed (since we showed that A contains all limit points of A),
F ⊃A, F closed
T T
this implies that A ⊇ F ; thus, A = F.
F ⊃A, F closed F ⊃A, F closed

2.7
2.7.1


Given a, b ∈ Rn , define the functions f
: Rn → Rn as f (x) = f (x1 , ..., xn ) = (x1 + b1 − a1 , ..., xn + bn − an )
and
g : Rn → Rn as g(x) = g (x1 , ..., xn ) = (x1 −b1 +a1 , ..., xn −bn +an ). Then observe that f (a) = f a1 , ..., an =
(a1 + b1 − a1 , ..., an + bn − an ) = (b1 , ..., bn ) = b. Moreover, we have:


(f ◦ g)(x) = f g(x1 , ..., xn )
= f (x1 − b1 + a1 , ..., xn − bn + an )


= (x1 − b1 + an ) + b1 − a1 , ..., (xn − bn + an ) + bn − an
= (x1 , ..., xn )

⇒ f ◦ g = idRn . Similarly, we have:

(g ◦ f )(x) = g f (x1 , ..., xn )
= f (x1 + b1 − a1 , ..., xn + bn − an )


= (x1 + b1 − an ) − b1 + a1 , ..., (xn + bn − an ) − bn + an
= (x1 , ..., xn )

⇒ g ◦ f = idRn . Thus, f and g are inverses.

11
 Now, we want to show that f and g are continuous. Given  > 0, let δ = . Then observe that for any x, y ∈ Rn ,
when d(x, y) < δ we have:


d f (x), f (y) = max {|(xi + bi − ai ) − (yi + bi − ai )|}
1≤i≤n

= max {|xi − yi |}
1≤i≤n

= d(x, y)
<δ=

⇒ f is continuous. The proof that g is continuous is analogous. Therefore, we conclude that there is an equivalence
between Rn and itself such that f (a) = b.

2.7.2
Let φ(x) = tan(x). Then φ : (− π2 , π2 ) → R is continuous, one-to-one and onto, with inverse function φ−1 : R →
(− π2 , π2 ) defined as φ−1 (x) = arctan(x), which is also continuous, one-to-one, and onto. Therefore, (− π2 , π2 ) is
topologically equivalent to R.
Now, we want to show that any two open intervals, considered as subspaces of the real number system, are topo-
logically equivalent. To do so, we first establish the following lemma:
Lemma 1. If (X, dX ) and (Y, dY ) are both topologically equivalent to (Z, dZ ), then (X, dX ) and (Y, dY ) are topo-
logically equivalent.

Proof. (X, dX ) topologically equivalent to (Z, dZ ) means that there exists a continuous inverse functions f : X → Z
and f −1 : Z → X, and similarly (Y, dY ) topologically equivalent to (Z, dZ ) means that there exists continuous
inverse functions g : Y → Z and g −1 : Z → Y . The composition of bijective functions is a bijection, and the
composition of continuous functions is a continuous function; therefore, g −1 ◦ f : X → Y is a continuous bijection,
as well as its inverse f −1 ◦ g : Y → X. Hence, (X, dX ) is topologically equivalent to (Y, dY ).

Let X ⊂ R be an open interval. Define h := φ−1 |X . Then h is continuous, one-to-one, and onto; moreover, there
exists inverse function h−1 := φ|φ−1 (X) , which is also continuous, one-to-one, and onto. Therefore, X and (− π2 , π2 )
are topologically equivalent. Analogously, if Y ⊂ R is an open interveal, then similar reasoning implies that Y and
(− π2 , π2 ) are topoligically equivalent. Therefore, by the above lemma, we conclude that X and Y are topologically
equivalent. Furthermore, we also conclude that any open interval is topologically equivalent to R.

2.7.3
We are told that for i = 1, 2, ..., n, (Xi , di ) is topologically equivalent to (Yi , d0i ); that is, for i = 1, ..., n, there exists
n
continuous inverse functions fi : Xi → Yi and fi−1 : Yi → Xi . X :=
Q
Xi is equipped with the metric dX : X ×
i=1

n
Q
X → R≥0 defined as dX (x, y) = dX (x1 , ..., xn ), (y1 , ..., yn ) = max {di (xi , yi )}, and Y := Yi is equipped
1≤i≤n i=1
with the metric dY : Y × Y → R≥0 is defined as dY (x, y) = dY x1 , ..., xn ), (y1 , ..., yn ) = max {d0 −i (xi , yi )}.


1≤i≤n
We want to show that X and Y are topologically
equivalent.

Define f : X → Y as f (x) = f (x1 , ..., xn ) = f1 (x1 ), ..., fn (xn ) . Then, since each fi is bijective and has an
inverse, we can define g : Y → X as g(y) = g (y1 , ..., yn ) = (f1−1 (y1 ), ..., fn−1 (yn ) . Then observe that

(f ◦ g)(y) = f g(y1 , ..., yn )
= f (f1−1 (y1 ), ..., fn−1 (yn )

= f1 (f1−1 (y1 )), ..., fn (f −1 (yn )

= (y1 , ..., yn )

12
⇒ f ◦ g = idY . Similarly,


(g ◦ f )(x) = g f (x1 , ..., xn )


= g (f1 (x1 ), ..., fn (xn )
= f1−1 (f1 (x1 )), ..., fn−1 (f (xn )

= (x1 , ..., xn )
⇒ g ◦ f = idX . Hence, f and g are inverses; i.e., g = f −1 .
Now, since for i = 1, ..., n, each fi : Xi → Yi is continuous, this means that ∀  > 0, ∃ δi > 0 such that for
all xi , yi ∈ Xi , di (xi , yi ) < δi implies that d0i fi (xi ), fi (yi ) < . Since there are only finitley



many δi , define
δ := max {di (xi , yi )}. Then, given  > 0, for x, y ∈ X, dX (x, y) = dX (x1 , ..., xn ), (y1 , ..., yn ) < δ implies that
1≤i≤n



dY f (x), f (y) = dY (f1 (x1 ), ..., fn (xn )), (f1 (y1 ), ..., fn (yn ))
= max {d0i fi (xi ), fi (yi ) }


1≤i≤n

<
⇒ f is continuous. Thus, X and Y are topologically equivalent.

2.7.4
Let Xi = (0, 1) ⊂ R. Then by exercise 2.7.2, we know that Xi is topologically equivalent to R. Now, by exercise
n
Xi = {(x1 , ..., xn ) ∈ Rn : 0 < xi < 1, i = 1, ..., n} is topologically equivalent to
Q
2.7.3, we conclude that
i=1
n
R = Rn .
Q
i=1

2.7.5
We want to show that metric equivalence, or isometry, is an equivalence relation.


1. Given a metric space (X, d), the identity function id : X → X is a bijection and ∀ x, y ∈ X, d id(x), id(y) =
d(x, y) ⇒ XRX.
2. Suppose (X, dX ) and (Y, d
Y ) are metric spaces, and XRY . Then there exists bijective function f : X → Y
such that dY f (x1 ), f (x2 ) = dX (x1 , x2 ). Since f is a bijection, it has an inverse function f −1 : Y → X,
which is also a bijection; thus, if y1 , y2 ∈ Y , then there exists x1 , x2 ∈ X such that f (x1 ) = y1 and f (x2 ) = y2 .
Therefore,
dY (y1 , y2 ) = dY f (x1 ), f (x2 ) = dX (x1 , x2 ) = dX f −1 (y1 )f −1 (y2 )

Thus, Y RX.
3. Let (X, dX ), (Y, dY ), and (Z, dZ ) be metric spaces, and suppose XRZ
and Y RZ. Then there exist bijections

f : X → Z and g : Y → Z such that dX (x1 , x2 ) = dZ f (x1 ), f (x2 ) and dY (y1 , y2 ) = dZ g(y1 ), g(y2 ) ,
for any x1 , x2 ∈ X and for any y1
, y2 ∈ Y . Then there exists an inverse function g −1 : Z → Y such that
dZ (z1 , z2 ) = dY g −1 (z1 ), g −1 (z2 ) ∀ z1 , z2 ∈ Z; furthermore, g −1 ◦ f : X → Y is a bijective function (since
it is the composition of bijections). Therefore, if x1 , x2 ∈ X and f (x1 ) = z1 , f (x2 ) = z2 , for some z1 , z2 ∈ Z,
then


dX (x1 , x2 ) = dZ f (x1 ), f (x2 )
= dZ (z1 , z2 )
= dY g −1 (f (x1 )), g −1 (f (x2 ))

= dY (g −1 ◦ f )(x1 ), (g −1 ◦ f )(x2 )

⇒ XRY .

13
Thus, metric equivalence is an equivalence relation. Now, since metric equivalence implies topological equivalence,
we conclude that topological equivalence is also an equivalence relation.

2.7.6
We are told that (Y, d0 ) be a subspace of the metric space (X, d). First we want to show that subset O0 of Y is open
⇐⇒ there exists an open subset O of X such that O0 = Y ∩ O.
(⇒) Suppose O0 ⊂ Y is open. Since X ⊃ Y , this implies that X ⊃ O0 . Therefore, there exists an open subset O
of X such that O0 = Y ∩ O; namely, O0 .
(⇐) Let O be an open subset of X and suppose O0 = Y ∩ O. Then since Y is a metric space, Y is open
⇒ O0 = Y ∩ O is open, since the finite intersection of open sets is open.
Now, we want to prove that a subset F 0 of Y is closed ⇐⇒ there exists a closed subset F of X such that
F = Y ∩ F . Suppose F 0 ⊂ Y is closed. Then CY (F 0 ) is open ⇐⇒ ∃ O ⊂ X closed such that CY (F 0 ) =
0

Y ∩ O ⇐⇒ CY (CY (F 0 )) = CY (Y ∩ O) ⇐⇒ F 0 = CY (Y ) ∪ CY (O) = CY (O) = Y ∩ CX (O).

Lastly, we want to show that N 0 ⊂ Y is a neighborhood of a ∈ Y ⇐⇒ there exists a neighborhood N ⊂ X of a
such that N 0 = Y ∩ N .
(⇒) Suppose N 0 ⊂ Y is a neighborhood of a ∈ Y . Then since X ⊃ Y , this implies that X ⊃ N 0 . Therefore,
there exists a neighborhood N ⊂ X such that N 0 = Y ∩ N ; namely N 0 .
(⇐) Conversely, suppose there exists a neighborhood N ⊂ X such that N 0 = Y ∩ N . Then since Y is a
metric space, Y is open, and since a ∈ Y , this implies that there exists a neighborhood M ⊂ Y of a; hence, Y is a
neighborhood of a. Therefore, N 0 = Y ∩ N is a neighborhood of a.

2.7.7
We are told that (Y, d0 ) is a subspace of (X, d); i.e., Y ⊂ X and d0 = d|Y ×Y . We want to show that if {an }∞
n=1 ⊂ Y ,
a ∈ Y , and lim an = a in (Y, d0 ), then lim an = a in (X, d).
n→∞ n→∞
Since lim an = a in (Y, d0 ), this means that ∀  > 0, ∃ N1 ∈ N such that d0 (an , a) < 
2 for n ≥ N1 . Now,
n→∞

suppose there exists b ∈ X such that lim an = b in (X, d). Then ∀  > 0, ∃ N2 ∈ N such that d(an , a) < 2 for
n→∞
n ≥ N2 . Now, let N := max{N1 , N2 }. Then given  > 0, n ≥ N implies that:

d(a, b) ≤ d(a, an ) + d(an , b)

= d0 (an , a) + d(an , b)
 
< + =
2 2
⇒ a = b.

2.7.8
√
We are told that there exists a sequence of points {an }∞ n=1 ⊂ R such that lim an = 2; that is, given  > 0, ∃
√ n→∞ √ √
N ∈ N such that n ≥ N implies that |an − 2| < 2 . Hence, if n, m ≥ N , then |an − am | = |an − 2 + 2 − am | ≤
√ √ √ √
|an − 2| + | 2 − am | = |an − 2| + |am − 2| < 2 + 2 = ; thus, {an } is a Cauchy sequence. Note that {an }
√
does not converge in (Q, d|Q×Q ) since limits of real-valued sequences are unique and 2 ∈
/ Q.

2.8
2.8.1
It is straightforward veryifying that H is a vector space over R with the usual operations of componentwise addition
and componentwise scalar multiplication.

14
 ∞
P
Now, let u, v, w ∈ H, and define A : H × H → R as A(u, v) = ui vi . Then if α, β, γ ∈ H, we have:
i=1

∞
X ∞
X ∞
X ∞
X
A(αu + βv, w) = (αui + βvi )wi = (αui wi + βvi wi ) = α ui w i + β vi wi = αA(u, w) + βA(v, w)
i=1 i=1 i=1 i=1

and similarly, we have:

∞
X ∞
X ∞
X ∞
X
A(u, βv + γw) = ui (βvi + γwi ) = (βui vi + γui wi ) = β ui vi + γ vi wi = βA(u, v) + γA(u, w)
i=1 i=1 i=1 i=1

∞
u2i > 0 since ui ≥ 0 ∀i ∈ N
P
Thus, A is of bilinear form. Moreover, observe that for any u ∈ H \ {0}, A(u, u) =
i=1
and there exists atleast one j ∈ N such that uj 6= 0 ⇒ u2j > 0. Therefore, A is positive definite.

2.8.2
1
Let A : V × V → R be a positive definite bilinear form on V and define N : V → R as N (v) = [A(v, v)] 2 ; we want
to show that N defines a norm on V . Observe that if v, w ∈ V and α ∈ R, then we have:
1 p
1. v 6= 0 ⇒ N (v) = [A(v, v)] 2 = A(v, v) > 0 since A(v, v) > 0.

√ form, A(0, 0) = A(0 · 0, 0) = 0 · A(0, 0) = 0. Therefore, v = 0 ⇒ N (v) =

2. Note that sincepA is a bilinear
1
[A(0, 0)] = A(0, 0) = 0 = 0. Thus, we conclude that v = 0 ⇐⇒ N (v) = 0.
2

3. Observe that

[N (v + w)]2 = A(v + w, v + w)
= A(v, v) + A(v, w) + A(w, v) + A(w, w)
= N (v)2 + N (w)2 + A(v, w) + A(w, v)
≤ N (v)2 + N (w)2 + 2A(v, v)A(w, w) b/c by Schwarz inequality A(v, w) ≤ A(v, v)A(w, w)
= N (v)2 + N (w)2 + 2N (v)N (w)
= [N (v) + N (w)]2 − 2N (v)N (w) + 2N (v)N (w)
= [N (v) + N (w)]2

⇒ N (v + w) ≤ N (v) + N (w)
4. Observe that
1 1 1
N (αv) = [A(αv, αv)] 2 = [αA(v, αv)] 2 = [α2 A(v, v)] 2 = |α|N (v)

Therefore, N defines a norm on V .

2.8.3
First we want to show that d : V ×V → R defined as d(u, v) = N (u−v) is a metric. Observe that for any u, v, w ∈ V ,
we have:
1. d(u, v) = N (u − v) ≥ 0 since N is a norm on the vector space V .
2. u = v ⇒ d(u, v) = d(u, u) = N (u − u) = N (0) = 0, and conversely d(u, v) = N (u − v) = 0 ⇒ u − v = 0.
Hence, d(u, v) = 0 ⇐⇒ u = v.


3. d(u, w) = N (u−w) = N (u−v +v −w) = N (u−v)+(v −w) ≤ N (u−v)+N (v −w) = d(u, v)+d(v, w)

15
⇒ d is a metric.
Now, we want to show that the function a : V × V → V defined as a(u, v) = u + v is continuous. Equipping
V
with the metric above and V × V with the metric d0 : (V × V ) × (V × V ) → R defined as d0 (u1 , u2 ), (v1 , v2 ) =
max{d(u1 , v1 ), d(v1 , v2 )} = max{N (u1 − v1 ), N (u
2 − v2 )}, we want to show that ∀ 
> 0, ∃ δ > 0 s.t. ∀
(u1 , u2 ), (v1 , v2 ) ∈ V × V with d0 (u1 , u2 ), (v1 , v2 ) < δ, we have d a(u1 , u2 ), a(v1 , v2
) < . Given  > 0, let
δ := 2 . Then observe that for all (u1 , u2 ), (v1 , v2 ) ∈ V × V such that d0 (u1 , u2 ), (v1 , v2 ) , we have:


d a(u1 , u2 ), a(v1 , v2 ) = d(u1 + u2 , v1 + v2 )


= N u1 + u2 − (v1 + v2 )
= N (u1 − v1 ) + N (u2 − v2 )
≤ 2 max{N (u1 − v1 ), N (u2 − v2 )}
= 2d0 (u1 , u2 ), (v1 , v2 )

< 2δ
=

⇒ a is continuous.
Now, we want to show that the function b : V → V defined as b(v) = −v is continuous;

that is, we want

to prove
that ∀  > 0, ∃ δ > 0 such that d(u, v) = N (u − v) < δ implies that d b(u), b(v) = N b(u) − b(v) < . Given
 > 0, let δ := . Then for all u, v ∈ V such that d(u, v) < δ, we have:


d (b(u), b(v) = d(−u, −v)


= N − u − (−v)
= N (v − u)
= N (−1(u − v))
= | − 1|N (u − v)
= d(u, v)
<δ
=

⇒ b is continuous.
Lastly, we want to show that the function c : R × V → V defined as c(α, v) = αv is continuous;

that is, we
want
to prove that ∀  > 0, ∃ δ > 0 such that d(u, v) = N (u − v) < δ implies that d c(u), c(v) = N c(u) − c(v) < .

Given  > 0 let δ := |α| . Then for all u, v ∈ V such that d(u, v) < δ, we have:


d (c(u), c(v) = d(αu, αv)
= N (αu − αv)


= N α(u − v)
= |α|N (u − v)
<δ
=

⇒ c is continuous. And we are done folks!

16