Linear Combinations and Span
Linear Combination
Definition: A linear combination of vectors x1 , x2 ,..., xk from vector space ( X , ) is a sum of the form
k
x
i 1
i i 1 x1 2 x2 ... k xk ,
where i for i 1,2,, k . The scalars i are called the weights or the coefficients of the linear
combination.
Is a linear combination of vectors from vector space ( X , ) also in X ?
Span
Definition: Let V be a set of vectors from vector space ( X , ) . The span of V , written span V , is
the set of all linear combinations of vectors in V .
k
span V x X x i xi with i , xi V , k
i 1
Examples:
1
V 2
1
1 0
V 0 , 0 3
0 1
V cos(t ),sin(t ),cos(2t ),sin(2t ) L2 ([ , ], )
Notice that span V is by definition a subset of X . Is span V also a subspace of ( X , ) ?
Linear Combination, Span, Linear Independence, Basis, Coordinates
ECE8010, R. Groff
Updated: 9/5/2018 1/7
Linear Dependence and Independence
Concept: In a set of linearly independent vectors, each vector provides some “new direction” in the
vector space that cannot be obtained from the other vectors. Linear independence is defined in terms of
linear dependence.
Let ( X , ) be a vector space. Let x1 , x2 ,..., xk be a set of k vectors from X , i.e. xi X for
i 1,2,, k .
Linear Dependence
Definition: A set x1 , x2 ,..., xk of k vectors is said to be linearly dependent if and only if
there exists a set of k scalars, 1 , , k , at least one of which is not zero, such that
Condition 1
1 x1 2 x2 k xk .
Claim: A set x1 , x2 ,..., xk of k vectors is said to be linearly dependent if and only if
one of the vectors can be expressed as a linear combination of the other k 1 vectors. Condition 2
Proof:
The claim is true if
Condition 1 Condition 2
or equivalently, if
(Condition 1 Condition 2) and (Condition 2 Condition 1)
We will prove these two cases separately.
Case: Condition 1 Condition 2
Condition 1 is true. Let 1 , , k be a set of scalars, at least one of which is nonzero, such
that 1 x1 2 x2 k xk .
Without loss of generality, assume 1 0 . Then, using only properties of vector spaces and
fields,
1 x1 2 x2 k xk
(VA) 1 x1 2 x2 k xk ( 2 ) x2 ( k ) xk ( 2 ) x2 ( k ) xk
(VA4,VA1,VSM3) 1 x1 ( 2 ( 2 )) x2 ( k ( k )) xk ( 2 ) x2 ( k ) xk
(VSM3, FA4) and 0
0
0
another property
(VSM)
1
1
1 x1 1
1
( 2 ) x2 ( k ) xk
(FM4, VSM4)
1
x1
(VSM2) 11 ( 2 ) x2 11 ( k ) xk
2 k
x1 2 x2 k xk
Thus Condition 2 holds.
Linear Combination, Span, Linear Independence, Basis, Coordinates
ECE8010, R. Groff
Updated: 9/5/2018 2/7
Case: Condition 2 Condition 1
Condition 2 is true. That is, one of the vectors can be expressed as a linear combination of the
others. Without loss of generality, assume x1 2 x2 k xk . Since the algebraic manipulations
from the first case work in both directions, we could reverse the operations to show that
x1 2 x2 k xk
1 x1 ( 2 ) x2 ( k ) xk
The coefficient for x1 is nonzero. Thus, Condition 1 is true. This completes the proof.
Because the proof only depends on properties of vector spaces and fields, this claim holds for ANY vector
space!
Linear Independence
Definition: A set x1 , x2 ,..., xk of k vectors is said to be linearly independent if and only if
the set is not linearly dependent. Condition 1
Claim: A set x1 , x2 ,..., xk of k vectors is said to be linearly independent if and only if
for any set of k scalars i , the linear combination Condition 2
1 x1 2 x2 k xk only for 1 2 k 0 .
Proof:
The claim is true if
Condition 1 Condition 2
Condition 1: the set x1 , x2 ,..., xk is not linearly dependent
NOT 1 , , k ,at least one of which is not zero, such that 1 x1 2 x2 k xk
1 ,, k ,at least one of which is not zero, 1 x1 2 x2 k xk (*)
Note that if 1 k 0 , then 1 x1 2 x2 k xk 0 x2 0 xk . Combining this
x1 0
fact with (*) yields
1 , k , 1 x1 2 x2 k xk 1 k 0 .
Thus Condition 2 is true. This completes the proof.
Linear Combination, Span, Linear Independence, Basis, Coordinates
ECE8010, R. Groff
Updated: 9/5/2018 3/7
Are the following sets of vectors linearly dependent or linearly independent?
1 0
0 , 1
( , )
3
0 0
1 0
1 , 1
1 0
1 1 1
1 , 2 , 0
0 1 1
1 1 1
1 , 1 , 0
0 1 1
L2 ([ , ], ) over , i.e., the set of all functions f :[ , ] satisfying f
2
(t ) dt
cos(t ),sin(t ),cos(2t ),sin(2t )
cos(t ),sin(t ),sin(t 4 )
1, t , t
2
1 t, 1 2t t 2
, 1 t 2
1 t , 1 t t 2
, 1 t 2
Linear Combination, Span, Linear Independence, Basis, Coordinates
ECE8010, R. Groff
Updated: 9/5/2018 4/7
Bases and Representations
Concept: A basis is a set of representative vectors from a vector space. Every vector in the
space can be represented as a unique linear combination of basis vectors.
Basis
Definition: Let V be a set of vectors from vector space ( X , ) . The set V is a basis for ( X , ) if and
only if
1) V is linearly independent, and
2) span V X
If V is a basis for ( X , ) , then the elements of V are called basis vectors.
Example: The standard basis for n (including n and n ) is given by
1 0 0 0
0 1 0 0
e1 0 , e2 0 , e3 1 , , en 0
0 0 0 1
Every field must contain elements called 0 and 1, so the vectors are well defined. Given a vector in n ,
how can you write it as a linear combination of the basis vectors?
The dimension of ( X , ) is defined as
dim( X ) : card(V ) where V is a basis for ( X , )
Also, by definition we say dim({}) 0 .
card(V ) is the cardinality of the set V , i.e. the number of elements in the set. Cardinality, and hence
dimension, is not necessarily finite.
Claim: The dimension of a vector space does not depend on the specific choice of basis, i.e. the
dimension of a vector space is well defined. Specifically, if two different bases are chosen for vector
space ( X , ) , then both bases will have the same cardinality.
(Proof omitted) Proof idea: If two bases for the same space have different dimensions, show that the
vectors in the bigger basis can be represented by linear combinations of vectors from the smaller
basis. Then show that this implies that the vectors in the bigger basis must be linearly dependent,
which contradicts the assumption that it is a basis.
Fact: Let (Y , ) be a subspace of ( X , ) . Then dim{Y } dim{ X } . Further, if dim{ X } is finite, then
dim{Y } dim{ X } only if Y X .
Proof: Included as Electronic Extra (included in pdf of notes, but not printed.)
Linear Combination, Span, Linear Independence, Basis, Coordinates
ECE8010, R. Groff
Updated: 9/5/2018 5/7
Representations (Coordinates) with Respect to a Basis
Let v1 ,, vn be a basis for vector space ( X , ) . Then for all x X , there exists a unique set of scalars
1 n
1
1 ,, n from , or equivalently , s.t. x i vi . The n scalars are called the
n
i 1
n n
components or the coordinates or the representation of x with respect to basis v1 ,, vn .
Proof:
Let basis v1 ,, vn for for vector space ( X , ) be given.
Let x X be given. (Note that x is arbitrary, so the proof will apply to any x X )
First we will show that a representation for x with respect to the basis exists, then we will show that
the representation is unique.
Existence:
Since v1 ,, vn is a basis, span v1 ,, vn X . Since x X , by definition of span,
n
1 , 2 ,, n such that x i vi . Thus a representation of x w.r.t. to basis v1 ,, vn exists.
i 1
Uniqueness:
To show uniqueness, we will assume that there are two representations of x with respect to basis
v1 ,, vn and then show that these representations must be exactly the same.
n
Assume there is another set of scalars 1 , 2 ,, n that also satisfies x i vi . Then,
i 1
n n n n
x i vi i vi v ( )v
i i i i
i 1 i 1 i 1 i 1
n
(
( )) v
i i i
i 1
i
n
v i i
i 1
Since v1 ,, vn is a basis, the vectors v1 ,, vn must be linearly independent. Thus, by our test for
linear independence,
n
v i i 1 n 0 1 (1 ) n ( n ) 0
i 1
1 1 , , n n .
Thus, the representations are actually the same! In other words, the representation is unique.
Since representations are unique, there is a correspondence between vectors and representations. We can
describe vectors in terms of their representation with respect to a specific basis. The representation is in
n , which might be significantly easier to work with compared to the vector space X .
Linear Combination, Span, Linear Independence, Basis, Coordinates
ECE8010, R. Groff
Updated: 9/5/2018 6/7
Electronic Extra: Proof of Fact about Dimension of Subspace
Fact: Let (Y , ) be a subspace of ( X , F) . Then dim{Y } dim{ X } . Further, if dim{ X } is finite, then
dim{Y } dim{ X } if and only if Y X .
Proof (for case when dim{Y },dim{ X } are both finite):
Let dim{ X } m and dim{Y } n . Then there exists a basis for X with m vectors, say u1 ,, um and
there exists a basis for Y with n vectors, say {v1 , , vn } . Since Y is a subspace of X and the basis for X
must span X , it follows that we can write the basis vectors for Y as linear combinations of basis vectors of X .
That is,
m m m
v1 ai1ui , v2 ai 2 ui , , vn ain ui
i 1 i 1 i 1
with coefficients aij . We can write the coefficients as a matrix
a11 a12 a1n
a a22
21 .
am1 am 2 amn
We will perform a proof by contradiction. Assume n m , that is, dim{Y } dim{ X } . Then by properties of
matrices (we’ll see this formally later in the semester), there exists a vector of scalars satisfying
a11 a12 a1n 1 0
a a22 0
21 2
am1 am 2 amn n 0
Then
m m m
1v1 2 v2 n vn 1 ai1ui 2 ai 2ui n ai1ui
i 1 i 1 i 1
n n n
a1 j j u1 a2 j j u2 amj j um
j 1
j 1
j 1
0 0 0
0 u1 0 u2 0 um
But this implies that {v1 , , v n } are not linearly independent, so they are not a basis for Y . This is a
contradiction. Thus, it must be that m n , i.e. dim{Y } dim{ X } . This establishes the first portion of the
Fact. The second portion of the Fact is left as an exercise.
Note: This is a proof by contradiction. In this approach to prove P Q , statement P is given as true and we assume Q
is false. We proceed to make logical deductions until we get a contradiction, which implies Q must actually be true,
establishing P Q . This is a dangerous technique, since faulty reasoning or miscalculation may falsely appear as a
contradiction when there is none. A direct proof of P Q or of the contrapositive Q P is very much preferred
to a proof by contradiction. In a direct proof, errors are generally much easier to detect because they lead to a result other
than the desired outcome ( Q in the case of P Q ). In proof by contradiction, the proof ends when you reach a
contradiction, so it is more difficult to know whether the contradiction resulted from an error.
Linear Combination, Span, Linear Independence, Basis, Coordinates
ECE8010, R. Groff
Updated: 9/5/2018 7/7
