ESE 501 Mathematics of Modern Engineering I

Lecture 2

”Vector spaces and their subspaces. Dimension and a basis of a vector

space.”

The concept of the Euclidean space IRn can be generalized to the concept
of a general vector space. Vector spaces have a number of useful proper-
ties. We are going to discuss some elements from the general theory of
vector spaces that are of practical importance, also for some constructions
discussed in this course when considering some elements of the PDEs. In
particular, it applies to the notion of orthogonality in a vector space. The
latter is based on concept of an inner product and concept of a norm.

Definition 1. A collection of elements, called V , is said to be a vector

space if one has two operations defined on elements of V , called ”addition
of vectors”: u + v, and ”scalar multiplication”: αu so that the following
properties are satisfied:
1) u + v ∈ V for all u, v ∈ V ;
2) αv ∈ V for all u ∈ V and α ∈ IR;
3) There is a ”zero” vector 0 ∈ V such that: 0 + u = u for all u ∈ V ;
4) u + v = v + u;
5) (u + v) + w = u + (v + w);
6) For any u ∈ V there is −u such that u + (−u) = 0;
7) α(u + v) = αu + αv;
8) (α + β)u = αu + βu for all α, β ∈ IR and u ∈ IR;
9) (αβ)u = α(βu);
10) 1u = u;

Example 1. a) Let V = IRn be the Euclidean space of n-dimensional

vectors where n = 1, 2, ... is fixed. Obviously, V is a vector space. Here
the addition ” + ” on elements from IRn is defined as usual addition of vec-
tors (component-wise) and the scalar multiplication αx is defined as usual

1
scalar multiplication of a vector x ∈ IRn by a scalar α ∈ IR (component-
wise). All properties follow just directly from properties of addition and
multiplication of real numbers.

b) V = {f : IR → IR} - set of all real-valued functions defined on IR

with values in IR. The addition ” + ” on V is then defined as addition of
functions at any fixed value of x: f (x) + g(x) := h(x) ∈ V and the scalar
multiplication is understood as usual multiplication of a scalar α ∈ IR by
a function f (x): αf (x) := h(x) ∈ V .

c) V = {Pn (t) = ni=0 ai ti , ai ∈ IR} - set of all real-valued polynomials of

P

degree n where n = 1, 2, ... is fixed.

d) V = {X : S → IR} - set of all real-valued random variables X defined

on a sample space S with values in IR (from probability).

Often there is the need to consider a subset of elements from V called a

subspace.

Definition 2. Any H ⊂ V such that

1) 0 ∈ H (0 is the ”zero” from V );
2) u + v ∈ H for all u, v ∈ H;
3) αu ∈ H for all u ∈ H and α ∈ IR
is called a subspace of V .

A subspace of a vector space is obviously a vector space as well.

Example 2. a) H = {0} is a subspace of any vector space V . Here H

consists of the ”zero” vector only.

b) Let V be a given vector space and u1 , u2 , ..., up ∈ V . Define

Xp
Span{u1 , u2 , ..., up } := { αi ui , αi ∈ IR}
i=1

as the set of all linear combinations of the vectors u1 , u2 , ..., up .

2
Then H is a subspace of V and is often referred to as a subspace spanned
by the vectors u1 , u2 , ..., up .

c) H := {P (t) = a + t2 , a ∈ IR} is not a subspace of P2 (t) since the ”zero”

vector (zero polynomial) does not belong to H.

d) Let H := {[2b + 3c, −b, 2c], b, c ∈ IR} - a set of all 3-dimensional vectors
of the specified form where the numbers b and c are arbitrary.
Is H a subspace of IR3 ?

It follows that
H := [2b + 3c, −b, 2c] = b[2, −1, 0] + c[3, 0, 2] = Span{u, v}
where u = [2, −1, 0] and v = [3, 0, 2]. Therefore, H is a subspace of IR3 .

e) Let H ⊂ V and K ⊂ V be subspaces of a vector space V . Define

H + K := {w = u + v for some u ∈ H and some v ∈ K}.
It can then easily be seen (verify it as an exercise) that
a) H + K is a subspace of V ;
b) H ⊂ H + K, K ⊂ H + K;

◦ Some other important examples of subspaces are associated with a given

matrix A of the size m × n.

Definition 3. N ul(A) := {x ∈ IRn |Ax = 0} is called the null space of the

matrix A.

It is directly seen that N ul(A) ⊂ IRn is a subspace of IRn .

Definition 4. Col(A) := { ni=1 αi ai } = Span{a1 , a2 , ..., an } is called the

P

column space of the matrix A.

Here the vectors a1 , a2 , ..., an denote the column vectors of the matrix
A.

We see that Col(A) ⊂ IRm is a subspace of IRm because it is spanned by

3
the vectors a1 , a2 , ..., an .

Similarly, one defines the space Row(A), the row space of a matrix A, as
a space spanned by the row vectors of A. By definition, Row(A) is then a
subspace of IRm .

As noticed before, the notion of linear independence of vectors plays a

fundamental role when studying the properties of a vector space. The
definition of linear independent vectors in an arbitrary vector space V is
the same as the one we provided earlier for the case of V = IRn .

The notion of linear independence leads to the following important concept

of a basis in a vector space V .

Definition 5. A set of vectors {v1 , v2 , ..., vr } from a vector space V is said

to be a basis in V if
1) vectors {v1 , v2 , ..., vr } are linearly independent;
2) V = Span{v1 , v2 , ..., vr }.

Example 3. a) Let V = IR2 . The vectors e1 = [0, 1] and e2 = [1, 0] form

a basis in IR2 .

b) Let V = Pn (t). The set of vectors {1, t, t2 , ..., tn } is a basis in Pn (t).

c) Let V = C[0,1] - set of all continuous real-valued functions defined on the

interval [0, 1]. We can easily see that the set of functions {sin t, cos t} is a
linearly independent set. However, to find a basis in C[0,1] becomes a more
difficult task as C[0,1] is an example of so-called infinite-dimensional vector
space.

Remark 1. A basis of a vector space V is not uniquelly determined as any

vector space can have in fact infinitely many different bases. For example,
in V = IR2 , a set of any two non-zero vectors such that they are not
multiples of each other, will be a different basis in IR2 .

4
Remark 2. It follows from the definition of a basis that a basis in V is a
maximal set of linearly independent vectors that spans V . Practically, if
V = Span{v1 , v2 , ..., vm }, then to find a basis in V , one has to find among
the vectors {v1 , v2 , ..., vm } the maximal set of linearly independent vectors.

One can show that all bases of a any vector space V have the same
number of vectors in them. This leads to the following

Definition 6. The number of vectors in a basis of the vector space V is

called the dimension of the vector space V .

Example 3. Consider the set of vectors W = {[3p−5q, 4q, p, q +1]T , p, q ∈

IR}. Is it a subspace of IR4 ?
Solution: We see that [3p−5q, 4q, p, q +1]T = p[3, 0, 1, 0]T +q[−5, 4, 0, 1]T +
[0, 0, 0, 1]T so that W ⊂ IR4 . However, 0 ∈
/ W so that W is not a subspace
4
of IR .

Example 4. Are the vectors v1 = [2, −1, 1]T , v2 = [2, −3, 2]T , and v3 =
[−8, 5, 4]T a basis in IR3 ?
Solution: To check the vectors for linear independency, we have to solve
the system
Ax = 0
where A = [v1 , v2 , v3 ].
It is not hard to see that the above system will have only the trivial
solution so that the vectors are linearly independent.
To see if IR3 = Span{v1 , v2 , v3 }, it is enough to verify that for any vector
b ∈ IR3 the system
Ax = b
has a solution. It can be done in the same way as solving the corresponding
homogeneous equation. In summary, indeed the given vectors build a basis
in IR3 .

Example 5. a) Consider W = {v = (v1 , v2 )|v1 ≥ v2 } ⊂ IR2 .

5
Clearly, W is not a subspace of IR2 . To see that, choose v = (2, 1) ∈ W .
However, then −v = (−2, −1) ∈ / W.
As a particular case, consider W = {v = (v1 , v2 )|v1 = v2 }. In this
case, any vector from W can be written as v = v1 (1, 1), v1 ∈ IR so that
W = Span{(1, 1)} and W is a vector space where (1, 1) is a basis and the
dimension of W is one.

b) Consider W = {(v1 , v2 , v3 )|3v1 − v3 = 0, 2v1 + 3v2 − 4v3 = 0}. Is W a

vector space?
Solving the system of two equations

3v1 − v3 = 0, 2v1 + 3v2 − 4v3 = 0,

we find that any solution can be expressed as

10 1
v2 = v3 , v1 = v3
9 3
so that
v = (v1 , v2 , v3 ) = v3 (3, 10, 9) = Span{(3, 10, 9)}.
Thus, W is a vector space spanned by the vector (3, 10, 9).