Vector spaces and Euclidian vector spaces

Radu MICULESCU

Transilvania University of Braşov

octomber 2020

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 1 / 50
 The concept of vector space I

De…nition. A non-empty set V is called a K -vector space

(where K = R or where K = C)

if it is endowed with two operations

+:V V !V

and
:K V !V
such that the following conditions are satis…ed:

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 2 / 50
I. (V , +) is a commutative group i.e.

i)
x + (y + z ) = (x + y ) + z,
for every x, y , z 2 V ;
ii) there exists 0V 2 V such that

x + 0V = 0V + x = x,

for every x 2 V ;
iii) for every x 2 V there exists x 2 V such that

x + ( x ) = ( x ) + x = 0V ;

iv)
x + y = y + x,
for every x, y 2 V .

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 3 / 50
 The concept of vector space - continuation

II.
i)
α (x + y ) = α x + α y ,
for every α 2 K and every x, y 2 V ;
ii)
(α + β) x = α x + β x,
for every α, β 2 K and every x 2 V ;
iii)
α ( β x ) = (αβ)x,
for every α, β 2 K and every x 2 V ;
iv)
1 x = x,
for every every x 2 V .
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 4 / 50
 Terminology and notations

The elements of V are called vectors and the elements of K are called
scalars.

We shall denote α x by αx for every α 2 K and every x 2 V .

We shall denote the identity element of (V , +) by 0V (or by 0) and by

0K the zero scalar.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 5 / 50
 Some simple facts following from the de…nition of a vector space

For every K -vector space V we have:

i)
0K x = 0V ,
for every x 2 V ;
ii)
α0V = 0V ,
for every α 2 K ;
iii)
( 1)x = x,
for every x 2 V .

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 6 / 50
 Example

For every n 2 N ,

K n = f(x1 , ..., xn ) j x1 , ..., xn 2 K g

endowed with the operations described by:

(x1 , ..., xn ) + (y1 , ..., yn ) = (x1 + y1 , ..., xn + yn )

and
α(x1 , ..., xn ) = (αx1 , ..., αxn ),
for every (x1 , ..., xn ), (y1 , ..., yn ) 2 K n and every α 2 K , is a vector space
called the coordinate (or the n-tuple) space.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 7 / 50
 Example

For every m, n 2 N , Mm,n endowed with:

- the usual sum a matrices
and
- the operation described by:

αA = (αaij ),

for every A = (aij ) 2 Mm,n and every α 2 K , is a vector space.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 8 / 50
 Example

The set of the solutions (x1 , ..., xn ) 2 K n of the homogenous system of

linear equations

a11 x1 + ... + a1n xn = 0

f ... ,
am1 x1 + ... + amn xn = 0
where aij 2 K for every i 2 f1, ..., m g and every j 2 f1, ..., ng, endowed
with the operations described on the …rst example is a vector space.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 9 / 50
 The notion of a subspace of a vector space

Proposition. For a subset U of a K -vector space V , the following

statements are equivalent:
i) (U, +, ) is a K -vector space;
ii)
x + y 2 U and αx 2 U,
for every x, y 2 U and every α 2 K ;
iii)
αx + βy 2 U,
for every x, y 2 U and every α, β 2 K .

De…nition. A subset U of a K -vector space V is called a vector subspace

(or just a subspace) of the vector space V if it satis…es one of the
equivalent conditions from the above proposition.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 10 / 50
 Examples

1. The subset f0V g of a K -vector space V is a subspace of V called the

zero (or the trivial) subspace of V .

2. The subset
fA 2 Mn j A =t Ag
is a subspace of Mn .

3. The set of points of any line through the origin of R2 is a subspace of

R2 .
In particular

Rx = f(x, 0) j x 2 Rg and Ry = f(0, y ) j y 2 Rg

are subspaces of R2 .
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 11 / 50
 Operations with subspaces

Proposition. If V1 and V2 are subspaces of a K -vector space V , then:

i)
V1 \ V2
is a subspace of V ;
ii)
def
V1 + V2 = fv1 + v2 j v1 2 V1 and v2 2 V2 g
is a subspace of V .

Remark. If V1 and V2 are subspaces of a K -vector space V , then, in

general,
V1 [ V2
is not a subspace of V .
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 12 / 50
 Examples

1.
V = f(x1 , x2 ) 2 R2 j 2x1 x2 = 0g
is a vector subspace of R2 .

2.
V = f(x1 , x2 , x3 ) 2 R3 j x3 = 1g
is not a vector subspace of R3 as

(1, 1, 1), (0, 0, 1) 2 V

but
(1, 1, 1) + (0, 0, 1) = (1, 1, 2) 2
/ V.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 13 / 50
 Homework

1. Is
V = f(x1 , x2 ) 2 R2 j 2x1 x2 + 1 = 0g
a vector subspace of R2 ?

2. Is
V = f(x1 , x2 , x3 ) 2 R3 j x1 x2 + 2x3 = 0g
a vector subspace of R3 ?

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 14 / 50
 Linear combinations of vectors

De…nition. Let V be a K -vector space and ∅ 6= S V . A vector of the

form
α1 x1 + ... + αp xp ,
where p 2 N , α1 ,...,αp 2 K and x1 ,...,xp 2 S, is called a linear
combination of elements of S.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 15 / 50
 The span of a subset of a vector space

Proposition. Let V be a K -vector space and ∅ 6= S V . The set of all

linear combinations of elements of S, which is denoted by L(S ), is a
subspace of V which is called the span of S.

De…nition. A subset S of a K -vector space V is called a spanning set for

V if
L(S ) = V .

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 16 / 50
 Linearly independent vectors

De…nition. A subset S = fx1 , ..., xp g of a K -vector space V is called

linearly independent (or we say that x1 , ..., xp are linearly independent) if
the equality
α1 x1 + ... + αp xp = 0V ,
where α1 ,...,αp 2 K , holds only for α1 = ... = αp = 0.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 17 / 50
 Linearly dependent vectors

De…nition. A subset S = fx1 , ..., xp g of a K -vector space V is called

linearly dependent (or we say that x1 , ..., xp are linearly dependent) if there
exist α1 ,...,αp 2 K such that:
i) at least one of the scalars α1 ,...,αp is non-zero;
ii)
α1 x1 + ... + αp xp = 0V .

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 18 / 50
 Example

The vectors

v1 = (2, 3, 1) 2 R3 and v2 = (1, 2, 2) 2 R3

are linearly independent.

Indeed, if α1 v1 + α2 v2 = 0R3 , then

2α1 + α2 = 0
f 3α1 + 2α2 = 0 .
α1 + 2α2 = 0

The above system of linear equations has the unique solution α1 = α2 = 0

2 1
since the matrix ( 3 2 ) -whose columns are given by v1 and v2 - has
1 2
rank 2.
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 19 / 50
 Examples

The vectors

v1 = (2, 3, 1) 2 R3 , v2 = (1, 2, 2) 2 R3 and v3 = (1, 1, 3) 2 R3

are linearly dependent.

Indeed, the relation α1 v1 + α2 v2 + α3 v3 = 0R3 implies

2α1 + α2 + α3 = 0
f 3α1 + 2α2 + α3 = 0 .
α1 + 2α2 3α3 = 0

The above homogenous system of linear equations has other solutions

2 1 1
besides the trivial one since the matrix ( 3 2 1 ) -whose columns
1 2 3
are given by v1 , v2 and v3 - has rank 2.
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 20 / 50
 Homework

1. Prove that the vectors

v1 = (1, 2, 1) 2 R3 and v2 = (1, 1, 1 ) 2 R3

are linearly independent.

2. Prove that the vectors

v1 = (1, 2, 1) 2 R3 , v2 = (1, 1, 1) 2 R3 and v3 = (1, 3, 3) 2 R3

are linearly dependent.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 21 / 50
 Basis of a vector space

De…nition. A …nite subset B of a vector space V is called a basis of V if:

i) B is linearly independent;

ii)
L(B ) = V ,
i.e. B spans the entire vector space V .

A vector space having a basis is called …nite dimensional.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 22 / 50
 The dimension of a vector space

Proposition. If B1 and B2 are bases of the vector space V , then

#(B1 ) = #(B2 ).

De…nition. For a vector space V 6= f0g the number of elements of a

basis is called the dimension of V and it is denoted by

dim(V ).

The dimension of f0g is 0.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 23 / 50
 The dimension of a vector space - continuation

Proposition. If for the vector space V we have dim(V ) = n, then:

α) fv1 , ..., vn g V is a basis of V if and only if fv1 , ..., vn g is linearly

independent.

β) fv1 , ..., vn g V is a basis of V if and only if L(fv1 , ..., vn g) = V .

γ) Any fv1 , ..., vm g V , where m > n, is linearly dependent.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 24 / 50
 Example

The set
B = fe1 , ..., en g K n,
where
e1 = (1, 0, ..., 0)
e2 = (0, 1, 0, ..., 0)
...
en = (0, 0, ..., 1)
and K is R or C, is basis (called the standard or the canonical basis) of
K n.
So
dim(K n ) = n.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 25 / 50
 The coordinates of a vector in a basis

Proposition. If B = fe1 , ..., en g is a basis of V , then for every x 2 V

there exist in a unique manner the scalars λ1 , ...., λn such that
n
x= ∑ λi ei = λ1 e1 + ... + λn en .
i =1

λ1 , ...., λn are called the coordinates of x in basis B.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 26 / 50
 Grassmann dimension theorem

Theorem. If V1 and V2 are vector subspaces of a …nite dimensional

vector space V , then

dim(V1 + V2 ) = dim(V1 ) + dim(V2 ) dim(V1 \ V2 ).

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 27 / 50
 The matrix of change of basis - I

De…nition. Let us consider a …nite dimensional vector space V and

0
B = fe1 , ..., en g and B = ff1 , ..., fn g two bases of V .
Then there exists aij , i, j 2 f1, ..., ng, such that

f1 = a11 e1 + a21 e2 + ... + an1 en

f2 = a12 e1 + a22 e2 + ... + an2 en
.
...
fn = a1n e1 + a2n e2 + ... + ann en

The matrix
a11 ... a1n
. ... .
A=( . . ),
. .
an1 ... ann
(whose …rst column is given by the coordinates of f1 and so on), is called
0
the matrix of change of bases from B to B .
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 28 / 50
 The matrix of change of basis - II

Proposition. In the above framework, let us consider x 2 V and the

scalars x1 , x2 , ..., xn , y1 , y2 , ..., yn such that

x = x1 e1 + x2 e2 + ... + xn en = y1 f1 + y2 f2 + ... + yn fn .

Then
x1 y1
x y2
( 2 ) = A( ).
... ...
xn yn

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 29 / 50
 The matrix of change of basis - III

Proposition. Let us consider a …nite dimensional vector space V and

B = fe1 , ..., en g

a basis of V . Given the matrix

a11 ... a1n
. ... .
A=( . . ),
. .
an1 ... ann

we can consider the set

0
B = ff1 , ..., fn g,

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 30 / 50
where

f1 = a11 e1 + a21 e2 + ... + an1 en

f2 = a12 e1 + a22 e2 + ... + an2 en
.
...
fn = a1n e1 + a2n e2 + ... + ann en
0
Then B is a basis of V if and only if det(A) 6= 0.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 31 / 50
 Example

0
Prove that B = fe1 , e2 , e3 g and B = ff1 , f2 , f3 g are bases of R3 , where

e1 = (1, 1, 0), e2 = (1, 0, 0), e3 = (1, 2, 3)

and
f1 = (1, 3, 3), f2 = (2, 2, 3), f3 = (6, 7, 9).

0
Find the the matrix of change from B to B and the coordinates of the
vector
v = 2e1 + 5e2 + 7e3
0
with respect to the basis B .

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 32 / 50
 0
One can easily check that the sets B and B are linearly independent, so,
as
0
#(B ) = #(B ) = 3,
we conclude that they are bases.

As one can easily check that

f 1 = e1 e2 + e3 ,

f 2 = e2 + e3
and
f3 = e1 + 2e2 + 3e3 ,
0
we conclude that the matrix of change from B to B is

1 0 1
A=( 1 1 2 ).
1 1 3

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 33 / 50
If
v = 2e1 + 5e2 + 7e3 = y1 f1 + y2 f2 + y3 f3 ,
then
2 y1
( 5 ) = A( y2 ),
7 y3
so
y1 2 1 1 1 2 0
1
( y2 ) = A ( 5 )=( 5 2 3 )( 5 ) = ( 1 ).
y3 7 2 1 1 7 2
Therefore
v = f2 + 2f3 .

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 34 / 50
 Homework

1. Prove that Mm,n with the usual addition and scalar multiplication is a
vector space of dimension mn.

2. Find α 2 R such that

f(1, α, 0), (α, 1, 1), (1, 0, α)g R3

is a basis.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 35 / 50
 Homework

3. Prove that
0
B = fe1 , e2 , e3 g and B = ff1 , f2 , f3 g
are bases of R3 , where

e1 = (1, 1, 0), e2 = (1, 0, 1), e3 = (1, 0, 1)

and
f1 = (1, 0, 0), f2 = (1, 1, 0), f3 = (1, 1, 1).

Find:
0
- the matrix of change from B to B
0
- the coordinates, with respect to the basis B and to the basis B , of the
vector v having the coordinates 2, 1, 1 with respect to the canonical
basis.
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 36 / 50
 Homework

4. Let us consider a > 0. Prove that:

i)
V = ff : ( a, a) ! Rg
with the usual addition and scalar multiplication is a vector space;
ii)
V1 = ff : ( a, a) ! R jf (x ) = f ( x ) for all x 2 ( a, a)g
is a subspace of V ;
iii)

V2 = ff : ( a, a) ! R jf (x ) = f ( x ) for all x 2 ( a, a)g

is a subspace of V .
iv)
V = V1 + V2 .
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 37 / 50
 The concept of scalar product

De…nition. Given V a R-vector space, a function <, >: V V ! R is

called a scalar product on V if it satis…es the following conditions:
i)
< x, x > 0 and (< x, x >= 0 if and only if x = 0),
for every x 2 V ;
ii)
< x, y >=< y , x > ,
for every x, y 2 V .
iii)
< x, y + z >=< x, y > + < x, z > ,
for every x, y , z 2 V ;
iv)
< λx, y >= λ < x, y > ,
for every x, y 2 V and every λ 2 R.
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 38 / 50
 The concept of scalar product - continuation

Remarks. Given V a scalar product <, >: V V ! R on a R-vector

space, we have the following properties:
i)
< x + y , z >=< x, z > + < y , z > ,
for every x, y , z 2 V ;
ii)
< x, λy >= λ < x, y > ,
for every x, y 2 V and every λ 2 R.

De…nition. A vector space on which it is de…ned a scalar product is called

Euclidian vector space and we denote it by (V , <, >).

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 39 / 50
 Example

<, >: Rn Rn ! R, given by

< x, y >= x1 y1 + ... + xn yn ,

for every x = (x1 , ..., xn ), y = (y1 , ..., yn ) 2 Rn , de…nes a scalar product.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 40 / 50
 A norm on an Euclidian vector space

Proposition. Given an Euclidian vector space (V , <, >), the function

k.k : V ! R given by p
kx k = < x, x >,
for every x 2 V is a norm, i.e. it satis…es the following properties:
i)
kx k 0 and (kx k = 0 if and only if x = 0),
for every x 2 V ;
ii)
kλx k = jλj kx k ,
for every x 2 V and every λ 2 R;
iii)
kx + y k kx k + ky k ,
for every x, y 2 V .
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 41 / 50
The above mentioned norm is called Euclidian norm.

If
ke k = 1,
then we call e a unit vector.

For example, if x 6= 0, then

1
e= x
kx k
is a unit vector and
x = kx k e.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 42 / 50
 Orthogonal vectors

De…nition. Given an Euclidian vector space (V , <, >), the vectors

x, y 2 V are called orthogonal if
< x, y >= 0.

De…nition. Given an Euclidian vector space (V , <, >), the subset S of V

is called orthogonal if
< x, y >= 0,
for every x, y 2 S, x 6= y .

De…nition. Given an Euclidian vector space (V , <, >), the subset S of V

is called orthonormal if
< x, y >= 0,
for every x, y 2 S, x 6= y and
kx k = 1,
for every x 2 X .
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 43 / 50
 Orthogonal basis

Let B = fe1 , ..., en g an orthogonal basis of the Euclidian vector space V .

Then we have
< x, e1 > < x, en >
x= e1 + ... + en ,
< e1 , e1 > < en , en >
for every x 2 V .

In particular, if B is an orthonormal basis, then

x =< x, e1 > e1 + ...+ < x, en > en ,

for every x 2 V and < x, ei > are called the Euclidian coordinates of x.

Remark. The matrix of change from an orthonormal basis to another

orthonormal basis of an Euclidian vector space is orthogonal.
Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 44 / 50
 The orthogonal projection

De…nition. Given an Euclidian vector space (V , <, >) and x, y 2 V ,

y 6= 0, the vector
< x, y >
y
< y, y >
is called the orthogonal projection of x onto the vector y and it is denoted
by pr y x.

So
< x, y >
pr y x = y.
< y, y >

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 45 / 50
 Gram-Schmidt theorem

Theorem. If fv1 , ..., vn g is a basis of an Euclidian vector space (V , <, >),

then there exists an orthonormal basis fe1 , ..., en g of V such that

L(fv1 , ..., vp g) = L(fe1 , ..., ep g),

for every p 2 f1, ..., ng.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 46 / 50
Proof. One can check that f kww 11 k , ..., kww nn k g, where

w1 = v1 ,
< v2 , w1 >
w2 = v2 w1 = v2 pr w 1 v2 ,
< w1 , w1 >
< v3 , w1 > < v3 , w2 >
w3 = v3 ( w1 + w2 ) = v3 (pr w 1 v3 + pr w 2 v3 ),
< w 1 , w1 > < w 2 , w2 >
...
< vj , w1 > < vj , wj 1 >
wj = vj ( w1 + ... + wj 1 ) =
< w1 , w 1 > < wj 1 , wj 1 >
= vj (pr w1 vj + ... + pr wj 1 vj )
is the orthonormal basis fe1 , ..., en g that we are looking for.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 47 / 50
 Example

Using the Gram-Schmidt procedure, orthonormalize the system of vectors

fv1 , v2 , v3 g R3 ,

where
v1 = (1, 2, 2)
v2 = ( 2, 1, 2)
and
v3 = (5, 3, 5)
and R3 is endowed with the usual scalar product.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 48 / 50
We have

w1 = v1 = (1, 2, 2),
< v2 , w1 >
w2 = v2 w1 = ( 2, 1, 2),
< w1 , w 1 >
< v3 , w1 > < v3 , w2 > 14 14 7
w 3 = v3 ( w1 + w2 ) = ( , , ) ,
< w1 , w 1 > < w2 , w2 > 3 3 3
so
1 2 2 2 1 2 2 2 1
e1 = ( , , ) , e2 = ( , , ) and e3 = ( , , ).
3 3 3 3 3 3 3 3 3

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 49 / 50
 Homework

Using the Gram-Schmidt procedure, orthonormalize the system of vectors

fv1 , v2 , v3 g R3 ,

where
v1 = (1, 2, 2)
v2 = ( 1, 0, 1)
and
v3 = (5, 3, 7)
and R3 is endowed with the usual scalar product.

Radu Miculescu (Institute) Vector spaces and Euclidian vector spaces octomber 2020 50 / 50