Linear Algebra and Differential Equations

Lecture 11
General Vector Spaces

July 5, 2016

Linear Algebra and Differential EquationsLecture 11General Vector

July
Spaces
5, 2016

1 / 21

General Vector Spaces

A vector space V is a collection of objects that satisfy
u+v V
u V
u+v=v+u
(u+v)+w=u+(v+w)
u+0=u
u+(-u)=0
(u+v)=u+v
( +)u=u+u
()u=(u)
1u=u
For u,v,w V and , R
Linear Algebra and Differential EquationsLecture 11General Vector
July
Spaces
5, 2016

2 / 21

Some Examples

(1) m n matrices: It is easy to check that the axioms are satisfied using
the usual properties of matrices, with the zero element being the m n
matrix with every element zero
(2) Polynomials in one variable with degree less than n
(3) Space of positive real numbers [x] with addition and scalar
multiplication defined by
[x] + [y ] = [xy ]

[x] = [x ]

(4) The set of continuous functions on [0,1] that satisfy f (0) = f (1) = 0

Linear Algebra and Differential EquationsLecture 11General Vector

July
Spaces
5, 2016

3 / 21

Subspaces
A subset U of a vector space V is a subspace, if it is a vector space in its
own right. Since U is contained in V it satisfies all the axioms with the
possible exception of the first two. So if the set U is closed under addition
and scalar multiplication it is a subspace.
Example: Subspaces of R3
Firstly to be a subspace it must contain 0. So one trivial subspace is just 0
by itself. The other possible subspaces are lines and planes through the
origin.
A line is generated by a vector u and consists of scalar multiples of u. It is
clear that addition and scalar multiplication will give vectors proportional
to u

Linear Algebra and Differential EquationsLecture 11General Vector

July
Spaces
5, 2016

4 / 21

The equation of a plane through the origin is

ax + by + cz = 0
with a, b, c not all zero. Let us assume c 6= 0.
1
z = (ax + by )
c
So the plane consists of vectors of the form

0
x
1
= x 0 + y 1 = xu1 + y u2
y
v=
1
ca
c (ax + by )
bc
Addition and scalar multiplication will give vectors that are also linear
combinations of u1 and u2

Linear Algebra and Differential EquationsLecture 11General Vector

July
Spaces
5, 2016

5 / 21

Another Example
Which of the following subsets of the vector space of 2 2 matrices is a
subspace?
a. The set of diagonal matrices
b. The set of non singular matrices
c. The set of singular matrices
a. Subspace
b. Not a subspace:


3 4
0 1

3 4
0 0

1 0
0 1

0 0
0 1

2 4
0 0

3 4
0 1

c. Not a subspace:



+

Linear Algebra and Differential EquationsLecture 11General Vector

July
Spaces
5, 2016

6 / 21

Linear Combinations of Vectors

A linear combination of a set of vectors v1 , v2 , , vr is an expression of
the form
1 v1 + 2 v2 + + r vr
If V is a set of vectors the span(V) is the set of all vectors expressible as
linear combinations of the vectors in V.
V is a set of vectors in the vector space U then span(V) is a subspace of U.
If v and w are in span(V) then
v = (1 v1 + + r vr ) = 1 v1 + + r vr
and
v+w = 1 v1 + +r vr +1 v1 + +r vr = (1 +1 )v1 + +(r +r )vr
are both in span(V).
Linear Algebra and Differential EquationsLecture 11General Vector
July
Spaces
5, 2016

7 / 21

Linear Independence
A set of vectors v1 , v2 , , vr are linearly independent if
1 v1 + 2 v2 + + r vr = 0
implies 1 = 2 = = r = 0.
If the vectors are not linearly independent they are linearly dependent. If a
set of vectors is linearly dependent then at least one of the vectors can be
expressed as a linear combination of the other vectors. Suppose the set
w1 , w2 , , wm is linearly dependent, then
1 w1 + 2 w2 + + m wm = 0
with at least one of the s non zero. If p 6= 0 then
wp =

1
(1 w1 + 2 w2 + + m wm )
p

Thus one of the vectors can be removed from the set without changing its
span.
Linear Algebra and Differential EquationsLecture 11General Vector
July
Spaces
5, 2016

8 / 21

Bases
A set of vectors B = {b1 , b2 , , br } is a basis for a vector space V if
every vector in V can be expressed as a linear combination of the vectors
in B uniquely.
Since 0 is in V then
1 b1 + 2 b2 + + r br = 0
since the representation is unique this implies 1 = 2 = = r = 0,
thus B is linearly independent.
A set of vectors B is a basis for a vector space V if B spans V and B is a
linearly independent set. Suppose given vector v in V has two different
representations as linear combinations of B
v = 1 b1 + 2 b2 + + r br
= 1 b1 + 2 b2 + + r br
Linear Algebra and Differential EquationsLecture 11General Vector
July
Spaces
5, 2016

9 / 21

Subtracting we get
(1 1 )b1 + (2 2 )b2 + + (r r )br = 0
but since B is linearly independent then 1 = 1 , 2 = 2 , , r = r
and the representation is unique. This means B is a basis.
The standard basis for Rn is

1
0
0
1

e1 = . e2 = .
..
..
0
0

en =

0
0
..
.

Linear Algebra and Differential EquationsLecture 11General Vector

JulySpaces
5, 2016

10 / 21

Coordinates
If B = {b1 , b2 , , bn } is a basis for a vector space V then v V can be
uniquely expressed in the form
v = v1 b1 + v2 b2 + + vr bn
The vi are called the coordinates of V with respect to the basis B. We can
represent v in terms of its components

v1
v2

[v]B = .
.
.
vn
This gives a map from V into Rn (in mathspeak an isomorphism) which
means that a a vector space with a basis of n vectors is essentially the
same as Rn .
Linear Algebra and Differential EquationsLecture 11General Vector
JulySpaces
5, 2016

11 / 21

Some examples
(a) A basis for the vector space of m n is given by the matrices Eij
which have a one in the ith row and jth column and zeros everywhere else.
There is evidently mn basis vectors.
(b) The set of polynomials of t with order less than 4 has a basis
{1, t, t 2 , t 3 }
(c) The set of smooth functions in an interval containing zero is a vector
space. The functions can be written in terms of the basis
{1, x, x 2 , , x n , }

X
f (x) =
an x n
i=0

This is the Taylor series of f about 0. The coefficients are given by

Taylors theorem
f (n) (0)
an =
n!
Linear Algebra and Differential EquationsLecture 11General Vector
JulySpaces
5, 2016

12 / 21

Theorem
If B = {b1 , b2 , , bn } is a basis for a vector space V then
1
2
3
4
5

A set of more than n vectors is linearly dependent.

A set of less than n vectors cannot span V.
Any basis for V must have exactly n vectors
Any set of n linearly independent vectors will be a basis for V
Any set of n vectors that span V is a basis.

The number of basis vectors for a vector space is called the dimension of
the space.
(1) A set of n+1 vectors is linearly independent
Let W = {w1 , , wn+1 }. We have to show
a1 w1 + + an+1 wn+1 = 0
with not all the ai equal to zero.
Linear Algebra and Differential EquationsLecture 11General Vector
JulySpaces
5, 2016

13 / 21

w1
..
.

=
..
.

w11 b1 + w12 b2 + + w1n bn

wn+1 = wn+1,1 b1 + wn+1,2 b2 + + wn+1,n bn

Substituting
a1 (w11 b1 + w12 b2 + + w1n bn +
+an+1 (wn+1,1 b1 + wn+1,2 b2 + + wn+1,n bn ) = 0
(a1 w11 + a2 w21 + + an+1 wn+1,1 )b1 + cdots
(a1 w1n + a2 w2n + + an+1 wn+1,n )bn = 0
Since B is a basis the coefficients must all vanish
a1 w11 + a2 w21 + + an+1 wn+1,1 = 0
..
..
.
.
a1 w1n + a2 w2n + + an+1 wn+1,n = 0
This is a system of n equations in n+1 unknowns so has non trivial
solutions thus W is linearly dependent.
Linear Algebra and Differential EquationsLecture 11General Vector
JulySpaces
5, 2016

14 / 21

(2) A set of less than n vectors cannot span V

Let W = {w1 , , wk } with k < n. Without loss of generality assume W
is linearly independent. Assume W spans V
b1 = 1 w1 + + k wk
One of the betas has to be non zero, say 1 then
w1 =

1
(b1 + 2 w2 + + k wk )
1

so span(W)=span({b1 , w2 , , wk }). Pick another basis vector b2 then

b2 = 1 b1 + 2 w2 + + wk
Not all of 2 , , k can be zero since B is a basis. If 2 6= 0 then
w2 =

1
(b2 + 1 b1 + 3 w3 + + k wk )
2

We can keep doing this until all the ws are replaced with b. Thus
span(W)=span({b1 , , bk }). But k < n and so some members of the
basis are not in span(W) this implies span(W) 6= V.
Linear Algebra and Differential EquationsLecture 11General Vector
JulySpaces
5, 2016

15 / 21

(3) Any basis must have exactly n vectors

If C = {c1 , , ck } is a basis it must span V so k n by part (2). But if
k > n then C is linearly dependent by part (1). Therefore k = n
(4) Any set of n linearly independent vectors is a basis Let
U = {u1 , , un } be linearly independent. Then for any vector v V the
set {v, u 1, , un } is linearly dependent by part (1). thus
av + 1 u 1 + + n un = 0
with not all of a, 1 , , n equal to zero. Since U is linearly independent
a 6= 0.
1
v = (1 bfu 1 + + n un )
a
Thus U spans V. a linearly independent set that spans V is a basis for V.

Linear Algebra and Differential EquationsLecture 11General Vector

JulySpaces
5, 2016

16 / 21

(5) Any set of n vectors that span V is a basis

Suppose W = {w1 , , wn } spans V. Then if W is linearly dependent
1 w1 + + n wn = 0
with not all the alphas zero. Say 1 6= 0 then
w1 =

1
(2 w2 + + n wn )
1

then span(W)=span({w2 , , wn }) but {w2 , , wn } has only n-1

vectors and by part (2) cannot span V. So the assumption that W was
linearly dependent is false and thus W is a basis.

Linear Algebra and Differential EquationsLecture 11General Vector

JulySpaces
5, 2016

17 / 21

Linear Transformations
A transformation, T, between two vector spaces U and V is linear if
T(v + w) = T(v) + T(w)
Note : T(0) = 0
An nonstandard example of a linear transformation on the vector space of
smooth functions is
D(f (x)) = f 0 (x)
It satisfies the requirement and so is a linear transformation. Another
linear transformation on the same space is
Z x
S(f (x)) =
f (t) dt
0

Linear Algebra and Differential EquationsLecture 11General Vector

JulySpaces
5, 2016

18 / 21

Linear transformations as matrices

Let B = {b1 , , bn } be a basis for the vector space V and let L be a
linear transformation from V to another vector space U. If v V then
v = v1 b1 + + vn bn
and
L(v) = v1 L(b1 ) + + vn L(bn )

v1

= [L(b1 )| |L(bn )]] ...

vn
so the action of L can be represented as a matrix, whose columns are the
images of the basis vectors under L, multiplied by a column vector of
coordinates with respect to the basis.
Linear Algebra and Differential EquationsLecture 11General Vector
JulySpaces
5, 2016

19 / 21

An example
Consider R3 (t) the set of polynomials in T with order less than 4. A basis
for R3 (t) is {b0 , b1 , b2 , b3 } = {1, t, t 2 , t 3 }. Now let D be the linear
transformation
D : R3 (t) R3 (t) Dp(t) = p 0 (t)
The action of D is determined by its effect on the basis vectors.
D(b0 ) = (1)0 = 0
D(b2 ) = (t 2 )0 = 2t = 2b1

D(b1 ) = (t)0 = 1 = b0
D(b3 ) = (t 3 )0 = 3t 2 = 3b2

so the matrix corresponding to D in this

0 1
0 0
[D] =
0 0
0 0

basis is

0 0
2 0

0 3
0 0

Linear Algebra and Differential EquationsLecture 11General Vector

JulySpaces
5, 2016

20 / 21

The action of D on
p(t) = 3 + 4t 3t 2 + t 3
is given by

0
0

0
0

1
0
0
0

0
2
0
0

3
0

0 4
3 3
1
0

4
6

=
3
0

corresponding to the action of D in R3 (t)

D(3 + 4t 3t 2 + t 3 ) = (3 + 4t 3t 2 + t 3 )0 = 4 6t + 3t 2

Linear Algebra and Differential EquationsLecture 11General Vector

JulySpaces
5, 2016

21 / 21