Chapter 1

Topology of Rn

1.1 Norms and distances on Rn

1.1.1 The space Rn

De…nition 1.1 We de…ne the space Rn = R

| R {z R} by
n times

Rn = fx = (x1 ; ; xn ) : x 1 ; ; xn 2 Rg ;
with the addition and the scalar multiplication
x + y = (x1 ; ; xn ) + (y1 ; ; yn ) = (x1 + y1 ; ; xn + yn ) ; 8x; y 2 Rn
and x = (x1 ; ; xn ) = ( x 1 ; ; xn ) ; 8 2 R; 8x 2 Rn :

Let e1 = (1; 0; ; 0), e2 = (0; 1; 0; ; 0); ; en = (0; ; 0; 1) be the canonical basis of Rn :

n
An element x = (x1 ; ; xn ) 2 R is therefore written in the form
X
n
x = x1 e1 + x2 e2 + xn en = xi ei :
i=1
0 1
x1
Matrix notation is often used x = @ ... A.
xn

For n = 2; R2 = fX = (x; y) : x; y 2 Rg ; representing the xy plane; with the orthonormal

! ! ! !
system O; i ; j and X = x i + y j ; 8X 2 R2 :

For n = 3; R3 = fX = (x; y; z) : x; y; z 2 Rg ; representing the xyz space; with the orthonormal

! ! ! ! ! !
system O; i ; j ; k and X = x i + y j + z k ; 8X 2 R3 :

De…nition 1.2 Rn is equipped with a scalar product de…ned, for two vectors x = (x1 ; ; xn ) and
y = (y1 ; ; yn ) of Rn ; by
X
n
x y = hx; yi = x1 y1 + + xn yn = xi yi :
i=1

3
M1105 CHAPTER 1. TOPOLOGY OF RN

Theorem 1.1 (Cauchy-Schwarz Inequality)

8x = (x1 ; ; xn ) 2 Rn ; 8y = (y1 ; ; yn ) 2 Rn ; we have
!1 !1
X
n X
n 2 X
n 2
xi yi x2i yi2 :
i=1 i=1 i=1

Proof : Suppose that x and y are not collinear.

Pn
We have (txi + yi )2 > 0; for all t 2 R: Then
i=1
P
n P
n P
n P
n P
n
(txi + yi )2 = (t2 x2i + 2txi yi + yi2 ) = x2i t2 + 2 xi yi t + yi2 > 0:
i=1 i=1 i=1 i=1 i=1
P
n P
n P
n
Let a = x2i ; b = xi yi and c = yi2 =) at2 + 2bt + c > 0;
i=1 i=1 i=1
0
as a > 0 =) = b2 ac < 0 =) b2 < ac
2
Pn P
n Pn
=) xi yi < x2i yi2 ; hence the inequality.
i=1 i=1 i=1
If x and y are collinear, then 9t0 2 R such that y = t0 x; therefore
1 1 1 1
P
n 2 P
n 2 P
n 2 P
n 2 P
n
x2i yi2 = x2i t20 x2i = jt0 j x2i
i=1 i=1 i=1 i=1 i=1
P
n P
n
and xi yi = jt0 j x2i ; hence the equality.
i=1 i=1

1.1.2 Norms and distances

De…nition 1.3 A norm on Rn is all mapping

N : Rn ! [0; 1[;
verifying the three properties :
(N1 ) 8x 2 Rn ; N (x) = 0 () x = 0Rn ; (Positivity)
(N2 ) 8 2 R; 8x 2 Rn ; N ( x) = j j N (x) ; (Homogeneity)
(N2 ) 8x; y 2 Rn ; N (x + y) N (x) + N (y) : (Triangular inequality)

De…nition 1.4 A distance on Rn is all mapping

d : Rn Rn ! [0; 1[;
that satis…es :
(D1 ) 8x; y 2 Rn ; d(x; y) = 0 () x = y;
(D2 ) 8x; y 2 Rn ; d(x; y) = d(y; x);
(D3 ) 8x; y; z 2 Rn ; d(x; z) d(x; y) + d(y; z):

Remark : For all norm k k on Rn ; we can associate a distance d( ; ) such that for x; y 2 Rn
d(x; y) = N (y x) :
The converse is not true, i.e., there are distances that are not deduced from a norm.

Note : For n = 1; the unique usual norm on R is the absolute value N (x) = jxj ; and the associated
distance is de…ned by d(x; y) = jy xj :

4
M1105 CHAPTER 1. TOPOLOGY OF RN

1.1.3 Usual Norms and associated distances

In what follows we will study the three usual norms of the space R2 :

First usual norm on R2 : Let x = (x1 ; x2 ) and y = (y1 ; y2 ) 2 R2 :

The …rst usual norm on R2 is de…ned by
kxk1 = jx1 j + jx2 j ;
and its associated distance is given by
d1 (x; y) = ky xk1 = jy1 x1 j + jy2 x2 j :

Proposition 1.1 k k1 is a norm on R2 and d1 is a distance on R2 :

Proof : Let x = (x1 ; x2 ) and y = (y1 ; y2 ) 2 R2 :
(N 1) kxk1 = 0 () jx1 j + jx2 j = 0 () jx1 j = jx2 j = 0 () x1 = x2 = 0 () x = (0; 0);
(N 2) k xk1 = j x1 j + j x2 j = j j jx1 j + j j jx2 j = j j (jx1 j + jx2 j) = j j kxk1 ; 8 2 R;
(N 3) kx + yk1 = jx1 + y1 j + jx2 + y2 j jx1 j + jy1 j + jx2 j + jy2 j kxk1 + kyk1 :
Let x; y; z 2 R2 :
(D1) d1 (x; y) = 0 () ky xk1 () y x = 0 () x = y;
(D2) d1 (x; y) = ky xk1 = k (x y)k1 = kx yk1 = d1 (y; x);
(D3) d1 (x; z) = kz xk1 = kz y + y xk1 kz yk1 + ky xk1 d1 (x; y) + d2 (y; z):

Second usual norm on R2 : Let x = (x1 ; x2 ) and y = (y1 ; y2 ) 2 R2 :

The second usual norm, called euclidean norm, on R2 is de…ned by
q
kxk2 = x21 + x22

and its associated distance is given by

q
d2 (x; y) = ky xk2 = (y1 x1 )2 + (y2 x2 )2 :

Proposition 1.2 k k2 is a norm on R2 and d2 is a distance on R2 :

Proof : Let x = (x1 ; x2 )pand y = (y1 ; y2 ) 2 R2 :
2 2
(N 1) kxk2 = 0 p() x21 + x22 = 0 () p x1 + x2 = 0 () p x1 = x2 = 0 () x = (0; 0);
2
(N 2) k xk2 = ( x1 ) + ( x2 ) = 2 2 (x1 + x2 ) = j j x21 + x22 = j j kxk2 ; 8 2 R;
2 2

(N 3) kx + yk22 = (x1 + y1 )2 + (x2 + y2 )2 = jx1 + y1 j2 + jx2 + y2 j2

(jx1 j + jy1 j)2 + (jx2 j + jy2 j)2
x21 + y12 + 2 jx1 j jy1 j + x22 + y22 + 2 jx2 j jy2 j
(x21 + x22 ) + (y12 + y22 ) + 2 (jx1 j jy1 j + jx2p
j jy2 j) : p
Using Cauchy-Schwarz inequality jx1 j jy1 j + jx2 j jy2 j x21 + x22 y12 + y22 ;
then kx + yk2 kxk2 + kyk2 + 2 kxk2 kyk2 (kxk2 + kyk2 )2
2 2 2

therefore kx + yk2 kxk2 + kyk2 :

For the distance, the proof is similar to the one of the previous proposition.

5
M1105 CHAPTER 1. TOPOLOGY OF RN

Third usual norm on R2 : Let x = (x1 ; x2 ) and y = (y1 ; y2 ) 2 R2 :

The third usual norm, called in…nite norm, on R2 is de…ned by
kxk1 = max (jx1 j ; jx2 j) ;
and its associated distance is given by
d1 (x; y) = ky xk1 = max (jy1 x1 j ; jy2 x2 j) :

Proposition 1.3 k k1 is a norm on R2 and d1 is a distance on R2 :

Proof : Let x = (x1 ; x2 ) and y = (y1 ; y2 ) 2 R2 :
(N 1) kxk1 = 0 () max (jx1 j ; jx2 j) = 0 () jx1 j = jx2 j = 0 () x1 = x2 = 0 () x = (0; 0);
(N 2) k xk1 = max (j x1 j ; j x2 j) = max (j j jx1 j ; j j jx2 j) = j j max (jx1 j ; jx2 j) = j j kxk1 ;
8 2 R;
(N 3) kx + yk1 = max (jx1 + y2 j ; jx2 + y2 j)
we have jx1 + y1 j jx1 j + jy1 j max (jx1 j ; jx2 j) + max (jy1 j ; jy2 j) kxk1 + kyk1 ;
similarly jx2 + y2 j kxk1 + kyk1 ; then kx + yk1 kxk1 + kyk1
For the distance, the proof is similar to the one of the previous theorem.

Example : Let A(2; 3) and B( 1; 2) be two points of R2 : Calculate d(A; B) with respect to the
three usual distances.
Solution : d1 (A;
pB) = j 1 2j + j2 3j p= 4:
2 2
d2 (A; B) = ( 1 2) + (2 3) = 10;
d1 (A; B) = max fj 1 2j ; j2 3jg = 3:

Remarks : (1) In the same way, we can de…ne the three usual norms on Rn by
P
n
kxk1 = jx1 j + + jxn j = jxi j ;
r
i=1
p Pn
kxk2 = x21 + + x2n = x2i ;
i=1
kxk1 = max (jx1 j ; ; jxn j) = max jxi j ;
1 i n

8x = (x1 ; ; xn ) 2 R n :
(2) 8i = 1; ; n; jxi j kxk ; whatever the norm.
Pn
(3) The norm k k2 is associated to the inner product hx; yi = xi yi between the two vectors
p i=1
x; y 2 Rn ; with kxk2 = hx; xi:
(4) From Cauchy-Shwarz inequality, we can deduce that for all x; y 2 Rn
jhx; yij kxk2 kyk2 :

De…nition 1.5 For x = (x1 ; ; xn ) 2 Rn ; we de…ne the Hölder’s norm of order p (1 p < 1) by
!1
1 X
n p
kxkp = (jx1 jp + + jxn jp ) p = jxi jp :
i=1

6
M1105 CHAPTER 1. TOPOLOGY OF RN

1.2 Neighborhoods on Rn

1.2.1 Open balls, closed balls and spheres in Rn

Let k k be a norm on Rn ; d be the associated distance, r > 0 and a be a point of Rn :

De…nition 1.6 We call open ball of Rn of center a and radius r; associated to k k ; the set
B(a; r) = fx 2 Rn : d(a; x) < rg = fx 2 Rn : kx ak < rg :

De…nition 1.7 We call closed ball of Rn of center a and radius r; associated to k k ; the set

B(a; r) = fx 2 Rn : d(a; x) rg = fx 2 Rn : kx ak rg

De…nition 1.8 We call sphere of Rn of center a and radius r; associated to k k ; the set

S(a; r) = fx 2 Rn : d(a; x) = rg = fx 2 Rn : kx ak = rg :

Remark : If the center is the origin and r = 1; the closed balls respectively, spheres are called unit
balls respectively, unit spheres.

1.2.2 Balls associated to the usual norms of R2

Associated ball to k k1 : Let A(a; b) 2 R2 and r > 0:

B1 (A; r) = fM 2 R2 : d1 (A; M ) < rg = f(x; y) 2 R2 : k(x; y) (a; b)k1 < rg :

Let M (x; y) 2 B1 (A; r) =) d1 (A; M ) < r =) jx aj + jy bj < r:
p
Geometrically, B1 is the inside of the square of center A and side 2r rotated private of
4
its boundary drawn in dotted line.

Associated ball to k k2 : Let A(a; b) 2 R2 and r > 0:

B2 (A; r) = fM 2 R2 : d2 (A; M ) < rg = f(x; y)q

2 R2 : k(x; y) (a; b)k2 < rg :
Let M (x; y) 2 B2 (A; r) =) d2 (A; M ) < r =) (x a)2 + (y b)2 < r
=) (x a)2 + (y b)2 < r2 :
Geometrically, B2 is the disk D(A; r) of center A and radius r without the circumference of
radius r drawn in dotted line.

Associated ball to k k1 : Let A(a; b) 2 R2 and r > 0:

B1 (A; r) = fM 2 R2 : d1 (A; M ) < rg = f(x; y) 2 R2 : k(x; y) (a; b)k1 < rg :

Let M (x; y) 2 B1 (A; r) =) d1 (A; M ) < r =) max (jx aj ; jy bj) < r
=) jx aj < r and jy bj < r:
Geometrically, B1 is the inside of the square of center A and side 2r private of its boundary
drawn in dotted line.

7
M1105 CHAPTER 1. TOPOLOGY OF RN

y y y
b+r b+r b+r

b * b * b *
A A A

b−r b−r b−r

O a−r a a+r x O a−r a a+r x O a−r a a+r x

Example : The (open) unit balls associated to the three usual norms are
2
B1 (O; 1) = f(x;
n y) 2 R : k(x; y)k1 = jxj + jyj < 1g ;o
p
2
B2 (O; 1) = (x; y) 2 R : k(x; y)k2 = x2 + y 2 < 1
and
B1 (O; 1) = f(x; y) 2 R2 : k(x; y)k1 = max (jxj ; jyj) < 1g :
y y y
1 1 1

O 1 x O 1 x O 1 x

Remarks : (1) In R, we obtain the interval ]a r; a + r[:

(2) In R3 , we well obtain full regular octahedron, full spheres (balls) and bull cubes respectively.

1.2.3 Equivalent norms

De…nition 1.9 Two norms N1 and N2 on Rn are said to be equivalent if there exist > 0 and
> 0 such that
8x 2 Rn ; N2 (x) N1 (x) N2 (x):

Proposition 1.4 Let N1 and N2 be two norms on Rn :

The following statements are equivalent
(i) There exist > 0 and > 0 such that
8x 2 Rn ; N2 (x) N1 (x) N2 (x):
(ii) There exist > 0 and > 0 such that
BN1 (0; ) BN2 (0; 1) BN1 (0; ):
(iii) There exist > 0 and > 0 such that
1 1
BN2 0; BN1 (0; 1) BN2 0; :

8
M1105 CHAPTER 1. TOPOLOGY OF RN

Proposition 1.5 For all x = (x1 ; ; xn ) 2 Rn ; we have

kxk1 kxk2 kxk1 n kxk1 :
Proof : 8i = 1; ; n; jxi j kxk2 =) max jxi j kxk2 =) kxk1 kxk2 ;
1 i n
X
n X
n
kxk22 = x21 + + x2n x2i + 2 jxi j jxj j (jx1 j + + jxn j)2 kxk21 =) kxk2 kxk1 ;
i=1 i=1; i<j
kxk1 = jx1 j + + jxn j kxk1 + + kxk1 n kxk1 :

Remark : All the norms on Rn are equivalent.

1.2.4 Neighborhood

De…nition 1.10 Let a 2 Rn and V Rn : We say that V is a neighborhood of a; if there exists a

real r > 0 such that B(a; r) V:

r
a*
B ( a, r )
V

De…nition 1.11 We call pointed neighborhood of a, noted Vb all neighborhood of a not containing
a:

Proposition 1.6 The intersection of two neighborhoods of a is a neighborhood of a:

Proof : Consider two neighborhoods V and W of a; then

9r1 > 0 = B(a; r1 ) V and 9r2 > 0 = B2 (a; r2 ) W:

Let r = inf(r1 ; r2 ) =) B(a; r) B(a; r1 ) V and B(a; r) B(a; r2 ) W =) B(a; r) V \W;
then V \ W is a neighborhood of a:

1.3 Convergence in Rn

1.3.1 Convergence of a vector sequence

De…nition 1.12 A vector sequence of Rn is all sequence (xk )k 0 such that xk = (x1k ; ; xnk ) with
xik 2 R; 8i = 1; ; n:

9
10
 at a

at a .
+
+

+ +
+