Chapter 2

Vector Spaces
One of my favorite dictionaries (the one from Oxford) defines a vector as A
quantity having direction as well as magnitude, denoted by a line drawn from
its original to its final position. What is useful about this definition is that we
can draw pictures and use our spatial intuition. An objection to this being used
as the definition of a vector is: Its not very precise, and thus will be hard to
compute with to any degree of accuracy.
The first section of this chapter makes the phrase a quantity having direction and magnitude more precise and at the same time develops an algebraic
structure. That is, the addition of one vector to another and the multiplication
of vectors by numbers will be defined, and various properties of these operations
will be stated and proved.
In order to avoid confusing vectors with numbers, all vectors will be in
boldface type.

2.1

R2 through Rn

Fix some point (think of it as the origin in the Euclidean plane), draw two short
rays from this point, and put arrowheads at the tips of the rays (see Figure 2.1).
A

B
Figure 2.1

Youve just drawn two vectors. Now label one of them A and the other B . The
magnitudes of these vectors are their lengths.
In many applications, vectors are used to represent forces. Since several
forces may act on an object, and the result of this will be the same as if a single
55

56

CHAPTER 2. VECTOR SPACES

force, called the resultant or sum, acted on the object, we define the sum of
two vectors in order to model how forces combine. Thus, if we want the vector
A + B , we draw a dashed line segment starting at the tip of A , parallel to B ,
with the same length as B ; cf. Figure 2.2a and b. Label the end of this dashed
line c and now draw the vector C ; i.e., draw a line segment from the origin to c
and put an arrowhead there; cf. Figure 2.2a. The vector C is called A + B , and
this way of combining A and B is referred to as the parallelogram law of vector
addition. Note that the construction of B + A will be different from that of
A + B ; cf. Figure 2.2b. However, a little geometry convinces us that we get the
same two vectors. We repeat this. The fact that A + B = B + A is something
that has to be proved. Its truth is not self-evident.
There is another operation (scalar multiplication) that we can perform on a
As meaning
vector, and that is to multiply it by a number. If A is a vector, 2A
A is a vector twice as long, and with the
is clear. It is the same as A + A ; i.e., 2A
A (c 0) to be the vector pointing in
same direction, as A . Thus, we define cA
the same direction as A but whose magnitude is c times the magnitude of A . If
A points in the direction opposite to A .
c is negative, cA
Since we will be doing a lot of computing with vectors, we need a method
for doing so other than using a ruler, compass, and protractor. Following in the
footsteps of Descartes and others, we assign a pair of numbers to each vector and
then see how these pairs should be combined in order to model vector addition
and scalar multiplication.
A+B

B+A

(a)

(b)

Figure 2.2

Lets go back to Figure 2.1. This time we label our initial point (0,0),
the origin in the Euclidean plane. We also draw two perpendicular lines that
intersect at (0,0). We call the horizontal line the x1 axis and the vertical line
the x2 axis. We next associate a number with each point on these axes. This
number indicates the directed distance of the point from the origin; that is,
the point on the x1 axis labeled 2 is 2 units to the right of the origin while
the point labeled 2 is 2 units to the left. How large a distance the number
1 represents is arbitrary. For the x2 axis, a positive number indicates that the
point lies above the origin while a negative number means that the point lies
below the origin. We now associate an ordered pair of numbers with each point
in the plane. The first number tells us the directed distance of the point from
the x2 axis along a line parallel to the x1 axis and the second number gives us
the directed distance of the point from the x1 axis along a line parallel to the

2.1. R2 THROUGH RN

57

x2 axis; cf. Figure 2.5. Every vector, which we picture as an arrow emanating
from the origin, is uniquely determined once we know where its tip is located.
This means that every vector can be uniquely associated with an ordered pair
of numbers. In other words, two vectors are equal if and only if their respective
coordinates are the same.
2A

A
c A , c>0

A
c A, c>0

(a)

(b)

(c)

Figure 2.3

The preceding discussion may lead to some confusion as to what an ordered

pair of numbers represents: a point in the plane or a vector. In practice this
will not cause any problem, since one can just as easily think of a vector as a
point in the plane, that point where the tip of the vector is located.
Example 1. Sketch the following vectors:
a. (1,1)

b. (1, 3)

58

CHAPTER 2. VECTOR SPACES

c. (2, 3)


Our next task is to determine how the number pairs, i.e., the coordinates,
of two vectors should be combined to give their sum. Let A = (a1 , a2 ) and
B = (b1 , b2 ) be two vectors. Then a simple proof using congruent triangles
yields that A + B = (a1 + b1 , a2 + b2 ); cf. Figure 2.6a. Similar arguments show
A = (ca1 , ca2 ) if c is a rational number; cf. Figure 2.6b. We havent
us that cA
B should be equal
talked about subtracting one vector from another yet, but A B
A B ) + B = A . Thus if A = (a1 , a2 ) and B = (b1 , b2 ),
to a vector such that (A
then A B = (a1 b1 , a2 b2 ).
x2

(0,1)
(2, 0)

(2,0)
(0,0)

x1

Figure 2.4

x2
a<0
b>0

a>0
b>0

(0, b)

(a, b)
(a, 0)
x1

a<0
b<0

a>0
b<0

Figure 2.5

We now formally define R2 along with vector addition and scalar multiplication for these particular vectors. R2 is the classical example of a two-dimensional
vector space.

2.1. R2 THROUGH RN

59

(a1 + b1 , a2 + b2 )

a2 + b2
a2

(a1 , a2 )
(b1 , b2 )

b2
a1

b1 a1 + b1
(a)

(ca1 , ca2 )

(a1 , a2 )

(b)

Figure 2.6

Definition 2.1. R2 = {(x1 , x2 ) : x1 and x2 are arbitrary real numbers). If

A = (a1 , a2 ), B = (b1 , b2 ), and c is any real number, then
1. A + B = (a1 + b1 , a2 + b2 )
A = (ca1 , ca2 )
2. cA
Note that R2 can also be thought of as all possible 1 2 matrices, with vector
addition being matrix addition and scalar multiplication the same as multiplying
a matrix by a scalar; cf. Definitions 1.3 and 1.4.
A few words about the notation { : }. It is used to describe a set of elements.
The phrase after the colon describes the property or attributes that something
must possess in order for it to be in the set. For example, {x : x = 1, 2} is the
set {1, 2}, and {x : 1 x < 100} is the set of real numbers between 1 and 100
including 1 but excluding 100. Thus (1,23), (3.14159,53), and (0, 7) are in R2 ,
but (0,$), (#, f ), and !e, are not, since the last three objects are not pairs of
real numbers.
Example 2. Solve the following vector equation:
X = (3, 4)
2(6, 1) + 3X
X = (3, 4) 2(6, 1)
3X
= (3, 4) (12, 2)

= (3 12, 4 + 2) = (9, 2)

60

CHAPTER 2. VECTOR SPACES

Thus

1
X = (9, 2) =
3



2
3,
3

Example 3. Given any vector in R2 write it as the sum of two vectors that are
parallel to the coordinate axes.
Solution. Let A = (a1 , a2 ) be any vector in R2 . To say that a vector X is parallel
to the x1 axis is to say that the second component of X s representation as an
ordered pair of numbers is zero. A similar comment applies to vectors parallel
to the x2 axis. Thus,
A = (a1 , a2 ) = (a1 , 0) + (0, a2 )
The first vector, (a1 , 0), is parallel to the x1 axis and the second vector, (0, a2 ),
is parallel to the x2 axis. We could also write
A = (a1 , 0) + (0, a2 ) = a1 (1, 0) + a2 (0, 1)
The vectors (1,0) and (0,1) are often denoted by i and j , respectively, and in
this notation (a1 , a2 ) = a1i + a2j .


a2j = (0, a2 )

(a1 , a2 )

a1i = (a1 , 0)

Figure 2.7

The theorem below lists some of the algebraic properties that vector addition
and scalar multiplication in R2 satisfy.
Theorem 2.1. Let A = (a1 , a2 ), B = (b1 , b2 ), and C = (c1 , c2 ) be three arbitrary
vectors in R2 . Let a and b be any two numbers. Then the following equations
are true.
1. A + B = B + A
A + B ) + C = A + (B
B + C)
2. (A
3. Let 0 = (0, 0), then A + 0 = A [zero vector]
A) such that A + (A
A) = 0 [A
A = (a1 , a2 )]
4. For every A there is a (A
A + B ) = aA
A + aB
B
5. a(A

2.1. R2 THROUGH RN

61

A = aA
A + bA
A
6. (a + b)A
A = a(bA
A)
7. (ab)A
A=A
8. 1A
Proof. We verify equations 1, 4, 5, and 8, leaving the others for the reader.
1. A + B = (a1 , a2 ) + (b1 , b2 ) = (a1 + b1 , a2 + b2 ) = (b1 + a1 , b2 + a2 )
= (b1 , b2 ) + (a1 , a2 ) = B + A
A) = (a1 , a2 ) + (a1 , a2 ) = (a1 a1 , a2 a2 ) = (0, 0) = 0
4. A + (A
A + B ) = a[(a1 , a2 ) + (b1 , b2 )] = a(a1 + b1 , a2 + b2 )
5. a(A
= (a(a1 + b1 ), a(a2 + b2 )) = (aa1 + ab1 , aa2 + ab2 )
= (aa1 , aa2 ) + (ab1 , ab2 ) = a(a1 , a2 ) + a(b1 , b2 )
A + aB
B
= aA
A = 1(a1 , a2 ) = (1a1 , 1a2 ) = (a1 , a2 ) = A
8. 1A
We next discuss the standard three-dimensional space R3 . As with R2 , we
first picture arrows starting at some fixed point and ending at any other point
in three-dimensional space. We draw three mutually perpendicular coordinate
axes x1 , x2 , and x3 and impose a distance scale on each of the axes. To each
point P in three space we associate an ordered triple of numbers (a, b, c), where a
denotes the directed distance from P to the x2 , x3 plane, b the directed distance
from P to the x1 , x3 plane, and c the directed distance from P to the x1 , x2
plane. Just as we did for R2 , we now think of vectors in three space both as
arrows and as ordered triples of numbers.
Example 4. For each of the triples of numbers below sketch the vector they
represent.
a. (1,0,0), (0,1,0), (0,0,1): these three vectors are commonly denoted by i , j ,
and k , respectively.
x3

k = (0, 0, 1)

j = (0, 1, 0)

i = (1, 0, 0)
x1

x2

62

CHAPTER 2. VECTOR SPACES

This notation is somewhat ambiguous; e.g., does i represent (1,0) or
(1,0,0)? There should, however, be no confusion, since which vector space
we are talking about will be clear from the discussion, and this will determine the meaning of i .
b. A = (1, 2, 1)
x3

(1, 0, 0)

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

(0, 2, 0)

x2

x1


Definition 2.2 defines R3 and the algebraic operations of vector addition and
scalar multiplication for this vector space.
Definition 2.2. R3 = {(x1 , x2 , x3 ) : x1 , x2 , and x3 are any real numbers}.
Vector addition and scalar multiplication are defined as follows:
B = (a1 , a2 , a3 )(b1 , b2 , b3 ) = (a1 b1 , a2 b2 , a3 b3 ), for any vectors
1. A B
A and B in R3 .
A = c(a1 , a2 , a3 ) = (ca1 , ca2 , ca3 ), for any vector A and any number c.
2. cA
Example 5. Let A = (1, 1, 0), B = (0, 1, 2). Compute the following vectors:
a. A + B = (1, 1, 0) + (0, 1, 2) = (1, 0, 2)
A = 2(1, 1, 0) = (2, 2, 0)
b. 2A

Having defined R2 and R3 , we now define Rn , i.e., the set of ordered n-tuples
of real numbers.
Definition 2.3. Rn = {(x1 , x2 , . . . , xn ) : x1 , x2 , . . . , xn are arbitrary real number}. If A and B are any two vectors in Rn and a is any real number, we define
vector addition and scalar multiplication in Rn as follows:
1. A B = (a1 , a2 , . . . , an ) (b1 , b2 , . . . , bn ) =
(a1 b1 , a2 b2 , . . . , an bn )

2.1. R2 THROUGH RN

63

A = c(a1 , a2 , . . . , an ) = (ca1 , ca2 , . . . , can )

2. cA
We note that Rn may also be thought of as the set of 1 n matrices. Sometimes
we will want to think of Rn as the set of n 1 matrices, i.e., Rn consists of
x will be defined for
columns rather than rows. One reason for this is so that Ax
A an m n matrix and x in Rn .
There are time when complex numbers are used for scalars. We then define
Cn = {(z1 , z2 , . . . , zn ) : zk are arbitrary complex numbesr}. Vector addition
and scalar multiplication are defined as in Rn , except that the scalars may now
be complex numbers instead of just real numbers.
Example 6. Write the vector (1, 6, 4) as a sum of three vectors each of which
is parallel to one of the coordinate axes.
(1, 6, 4) = (1, 0, 0) + (0, 6, 0) + (0, 0, 4)
x3
(1, 6, 4)
(1, 6, 0)
(0, 6, 0)

x2

x1

We can also write (1, 6, 4) as a linear combination of the three vectors i , j , and
k.
(1, 6, 4) = (1, 0, 0) + 6(0, 1, 0) + 4(0, 0, 1) = ii + 6jj + 4kk .

A common mistake is to think that R2 is a subset of R3 ; that is, W =
{(x1 , x2 , 0) : x1 and x2 arbitrary} is equated with R2 . Clearly W is not R2 ,
since W consists of triples of numbers, while R2 consists of pairs of numbers.
Example 7. Solve the following vector equation in R5 :
X = 3(2, 0, 6, 1, 1)
2(1, 4, 2, 0, 1) + 6X
X = (6, 0, 18, 3, 3) 2(1, 4, 2, 0, 1) = (8, 8, 14, 3, 5)
6X
Thus,
X =

1
(8, 8, 14, 3, 5) =
6

4 4 7 1 5
, , , ,
3 3 3 2 6

The algebraic operations we have defined in Rn have the same properties as

those in R2 , and we list them in the following theorem.
Theorem 2.2. Let A , B , and C be three arbitrary vectors in Rn . Let a and b
be any two numbers. Then the following are true:

64

CHAPTER 2. VECTOR SPACES

1. A + B = B + A
A + B ) + C = A + (B
B + C)
2. (A
3. Let 0 = (0, 0, . . . , 0), then 0 + A = A
A) in Rn such that A + (A
A) = 0 . If
4. For every A in Rn there is a (A
A = (a1 , . . . , an ), A
A = (a1 , . . . , an )
A + B ) = aA
A + aB
B
5. a(A
A = aA
A + bA
A
6. (a + b)A
A = a(bA
A)
7. (ab)A
A=A
8. 1A

We ask the reader to check that these identities are valid.

Problem Set 2.1

1. Sketch the following vectors in R2 :
a. (1, 2), (0, 6), 2(1, 4)

b. (1, 1) + (1, 1), 3(2, 1) 2(1, 1)

2. If A = (1, 6) and B = (3, 2), sketch the vectors
A
a. A , A

A + B,A B
b. A + B , A
A 3B
B
c. 2A

A, where t is an arbitrary
3. Let A = (1, 1). Sketch all vectors of the form tA
real number.
X + 2A
A = 3B
B.
4. Let A = (1, 2). Let B = (2, 6). Find X such that 6X
5. Let A and B be the same vectors as in problem 4. Sketch vectors of the
form X = c1A + c2B for various values of c1 and c2 . Which vectors in R2
can be written in this manner?
6. Prove Theorem 2.2 for n = 3.
7. Let A = (1, 1, 2), B = (0, 1, 0).
a. Find all vectors x in R3 such that x = t1A + t2B for some constants
t1 and t2 .
b. Is the vector (0,0,1) one of the vectors you found in part a?
8. Let x = (1, 0), y = (0, 1). Give a geometrical description of the following
sets of vectors.

2.2. DEFINITION OF A VECTOR SPACE

65

x + (1 t)yy : 0 t 1}
a. {tx

b. {t1x + t2y : 0 t1 1, 0 t2 1}
x + (1 t)yy : any real number}
c. {tx

d. {t1x + t2y : 0 t1 , 1 t2 }

e. {t1x + t2y : t1 and t2 arbitrary}

9. Let x = (1, 2), y = (1, 1). Describe the following sets of vectors.
x + (1 t)yy : 0 t 1}
a. {tx

b. {t1x + t2y : 0 t1 1, 0 t2 1}

x + (1 t)yy : t any real number}

c. {tx

d. {t1x + t2y : 0 t1 , 1 t2 }

e. {t1x + t2y : t1 and t2 arbitrary}

10. Let x = (1, 1, 0) and let y = (1, 1, 1). Describe the following sets of vectors:
x + (1 t)yy : 0 t 1}
a. {tx

b. {t1x + t2y : 0 t1 1, 0 t2 1}
x + (1 t)yy : t any real nubmer}
c. {tx

11. Given two sets A and B, their intersection, A B, is defined as everything

they have in common.
A B = {x : x A and x B}
Let
W1 = {(x1 , x2 , x3 ) : x1 = x2 , x3 = 0}

W2 = {(x1 , x2 , x3 ) : x1 = x2 }
W3 = {(x1 , x2 , x3 ) : x1 + x2 = 0}
Graph each of the following sets:
a. W1 W2

2.2

b. W2 W3

c. W1 W2 W3

Definition of a Vector Space

In the last section we defined the classical vector spaces Rn , for n = 2, 3, . . . . In

this section we extract from these sets their crucial properties, and then define
a vector space to be anything that has these properties. Thus, let V denote a
set or collection of objects. The elements of V will be called vectors and we
assume that there are two operations defined. The first (vector addition) will
be denoted by a + sign, and the second (scalar multiplication) by a or
by juxtaposition. Specifically if x and y are any two vectors, i.e., elements of
x is
V , then x + y is in V . If a is any real number and x any vector, then ax

66

CHAPTER 2. VECTOR SPACES

x be in V are expressed by saying that

in V . The conditions that x + y and ax
V is closed under vector addition and scalar multiplication. The term in V
has been underlined to underscore the idea that there may be times when an
addition or multiplication has been defined such that the result will not be back
in V . In such cases V is not a vector space.
In order for V to be a vector space, it is not enough that these two operations
have been defined. They must also satisfy certain laws: those needed so that a
reasonable arithmetic is possible. The requirements are listed in the following
definition.
Definition 2.4. Let V be a set on which two operations, vector addition and
scalar multiplication, have been defined. V will be called a real vector space if
the following properties are true:
1. For any x and y in V, x + y is in V .
x is in V .
2. For any x in V , and any real number a in R, ax
3. x + y = y + x , for any x , and z in V .
x + y ) + z , for any x , y , and z in V .
4. x + (yy + z ) = (x
5. There is a vector in V denoted by 0 such that x + 0 = x , for every vector
x in V .
x) in V such that x + (x
x) = 0 .
6. For any vector x in V , there is a vector (x
x + y ) = ax
x + ayy , for any two vectors x and y in V and any real number
7. a(x
a.
x = ax
x + bx
x, for any two real numbers a and b and any vector x .
8. (a + b)x
x = a(bx
x), for any two real numbers a and b and any vector x .
9. (ab)x
x = x , for every vector x in V .
10. 1x
What the above properties mean practically is that all the algebraic manipulations youve learned to do with real numbers and vectors in Rn can also be
done in this more abstract setting.
Remember, whenever someone says, V is a real vector space they are
merely stating that the elements of V along with V s version of vector addition
and scalar multiplication satisfy rules 1 through 10.
One more comment: The scalars are required to be real numbers and V is
called a real vector space. There will be times when the scalars are complex
numbers, in which case we call V a complex vector space. In the following we
will not explicitly use the adjective real, since all our vector spaces will be real,
unless the contrary is stated.

2.2. DEFINITION OF A VECTOR SPACE

67

Example 1. Rn (n 2) is a vector space. Definition 2.3 defined Rn as the set of

n tuples of real numbers as well as what vector addition and scalar multiplication
mean in this setting. Theorem 2.2 shows that 1 through 10 of Definition 2.4
hold.

Example 2. R1 = {x : x is a real number} with vector addition and scalar
multiplication being ordinary addition and multiplication of real numbers. 
Example 3. Mmn , the set of all m n matrices with real entries, with the
usual way of adding matrices and multiplying matrices by numbers, is a vector
space.

Example 4. Let V = {(x1 , 0, x3 ): such that x1 and x3 are real numbers}.
Define addition and scalar multiplication by
x + y = (x1 , 0, x3 ) + (y1 , 0, y3 ) = (x1 + y1 , 0, x1 + y3 )
x = a(x1 , 0, x3 ) = (ax1 , 0, x3 )!!
ax
Clearly V is closed under these versions of vector addition and scalar multiplication; i.e., 1 and 2 hold. As a matter of fact, one can quickly check that 3
through 7 hold, but 8 does not. For
x + bx
x = a(x1 , 0, x3 ) + b(x1 , 0, x3 ) = (ax1 , 0, x3 ) + (bx1 , 0, x3 )
ax
= [(a + b)x1 , 0, 2x3 ]
while
x = (a + b)(x1 , 0, x3 ) = [(a + b)x1 , 0, x3 ]
(a + b)x
Since these two expressions will be equal only if 2x3 = x3 or x3 = 0 it is clear
that 8 is not true for all the elements of V . Thus V is not a vector space.

Example 5. Let V = {x : x > 0}. That is, V is the set of all positive real
numbers. In this example will denote vector addition and will denote scalar
multiplication. Define these operations as follows:
x y = xy; that is, vector addition will be ordinary multiplication
a x = xa
Since the product of two positive numbers is positive, and a positive number
raised to any power is still positive, V is closed under these two operations; i.e.,
1 and 2 in the definition of a vector space hold. We now proceed to check, as
we must, each of the other required properties.
3. x y = xy = yx = y x
x y) z
4. x (yy z ) = x(yz) = (xy)z = (x
5. For the 0 vector we try the number 1 : 1 x = 1x = x = x

68

CHAPTER 2. VECTOR SPACES

x ) we try x1 : (x
x) x = x1 x = 1 = 0 . Note that if x > 0, so
6. For (x
1
is x .
x y ) = (xy)a = (xa )(y a ) = (a x ) (a y )
7. a (x
8. (a + b) x = xa+b = xa xb = (a x ) (b x )
9. (ab) x = xab = (xb )a = a (b x )

10. 1 x = x1 = x

Notice that in 5, the number 1 is treated as a vector in V , while in 10, it is

treated as a scalar. Since V , with these definitions of vector addition and scalar
multiplication, satisfies 1 through 10 of Definition 2.4, V is a vector space. 
The moral of this example, if there is one, is that neither addition nor
multiplication need always be what were used to. Notice that the zero vector in the above example is not the number zero but the number 1. What is
important about the zero vector is not that it be zero (the number) but that it
satisfy the algebraic property 5 of Definition 2.4.
Example 6. V = {(x1 , x2 , 0) : x1 , x2 are arbitrary real numbers}. The usual
definitions of addition and multiplication of vectors in R3 will be used.

The reader is asked to check that all 10 properties do indeed hold and conclude
that V is a vector space.
Example 7. V = {ff : f is a real-valued function defined on [0,1]}. For f and
g in V , f + g must be in V ; that is, it must be a function defined on [0,1].
Thus for each t, 0 t 1, define (ff + g )(t) = f (t) + g (t). We similarly define
agg , by (agg )(t) = agg (t). With these definitions of vector addition and scalar
multiplication, 1 and 2 are satisfied. The reader is asked to check that the other
eight properties are also true.

Example 8. Let Pn = {polynomials in t of degree n} =

{an tn + an1 tn1 + + a1 t + a0 : where the aj are real numbers}

Thus
P0 = {a : a is any real number} = R1

P1 = {a0 + a1 t : a0 and a1 are arbitrary real numbers}

If f = f0 + + fn tn and g = g0 + + gn tn are any two polynomials in Pn ,

we define (ff + g )(t) by
[ff + g ](t) = (f0 + g0 ) + (f1 + g1 )t + + (fn + gn )tn
Thus f + g is in Pn . We define aff by
aff (t) = (af0 ) + (af1 )t + + (afn )tn
which is also in Pn . To see that Pn satisfies the other conditions in Definition 2.4
is routine and is left to the reader. We note that Pn , for each value of n, is a
subset of the vector space in Example 7.


2.2. DEFINITION OF A VECTOR SPACE

69

We remind the reader that a vector space is not just a set, but a set with
two operations: vector addition and scalar multiplication. If we change any one
of these three, all ten properties must again be checked before we can say that
we still have a vector space.
In the future when we deal with any of the standard vector spaces Rn ,
Pn , Mmn , or the vector space in Example 7, the operations of vector addition
and scalar multiplication will not be explicitly stated. The reader, unless told
otherwise, should assume the standard operations in these spaces.

Problem Set 2.2

1. Verify the details of Example 6.
2. Let V = {(x1 , x2 ) : x1 and x2 are real numbers}. Define addition by
(x1 , x2 ) + (y1 , y2 ) = (x1 + y1 , x2 y2 )
and scalar multiplication by
a(x1 , x2 ) = (ax1 , ax2 )
Is V a vector space? If not, list all the axioms that V fails to satisfy.
3. Let V = {(x1 , x2 ) : x1 and x2 are real numbers}. Define addition and
scalar multiplication by
(x1 , x2 ) + (y1 , y2 ) = (x2 + y2 , x1 + y1 )
a(x1 , x2 ) = (ax1 , ax2 )
Is V a vector space? If not, list all the axioms that V fails to satisfy. What
happens if we define a(x1 , x2 ) = (ax2 , ax1 )?
4. Verify that Mmn (Example 3) is a vector space.
5. Verify that the set in Example 4 satisfies all the properties of Definition 2.4
except number 8.
6. Let C[0, 1] be the set of real-valued continuous functions defined on the
interval [0,1]. This set is a subset of the vector space defined in Example 7.
Define addition and multiplication as in that example, and show that
C[0, 1] is also a vector space.
7. Show that V in Example 7 is a vector space.
8. Show that Pn for n = 0, 1, . . . , is a vector space.
9. Let V0 be all 2 2 matrices for which the 1,1 entry is zero. Let V1 be all
2 2 matrices for which the 1,1 entry is nonzero. Determine if either of
these two sets is a vector space. Define addition and multiplication as we
did in Mmn .

70

CHAPTER 2. VECTOR SPACES

10. Let V = {A, B, . . . , Z}; that is, V is the set of capital letters. Define
A + B = C, B + C = D, A + E = F ; i.e., the sum of two letters is the first
letter larger than either of the summands, unless Z is one of the letters
to be added. In that case the sum will always be A. Can you define a
way to multiply the elements in V by real numbers in such a way that V
becomes a vector space?
11. Let V = {(x, 0, y) : x and y are arbitrary real numbers}. Define addition
and scalar multiplication as follows:
(x1 , 0, y1 ) + (x2 , 0, y2 ) = (x1 + x2 , y1 + y2 )
c(x, 0, y) = (cx, cy)
Is V a vector space?
12. Let Vc = {(x1 , x2 ) : x1 + 2x2 = c}.
a. For each value of c sketch Vc .
b. For what values of c, if any, is Vc a vector space?



a1 a2 a3
: a1 + a2 + a3 = c . For what values of c is Vc
13. Let Vc =
a4 a5 a6
a vector space?
14. Let Vc = {pp : p is in P4 and p (c) = 0}. For what values of c is Vc a vector
space?
15. Let V = {A : A is in M23 , and ij aij = 0}. That is, A is in V if A is a
2 3 matrix whose entries sum to zero. Is V a vector space?
16. Let V1 = {(x1 , x2 ) : x21 + x22 = 1}, V2 = {(x1 , x2 ) : x21 + x22 } 1, V3 =
{(x1 , x2 ) : x1 0, x2 0}, and V4 = {(x1 , x2 ) : x1 + x2 0}. Which of
these subsets of R2 is a vector space?
17. Let V = {pp : p is in P17 and p (1) + p (6) p (2) = 0}. Is V a vector
space? Does the subscript 17 have any bearing on the matter? What if
the constraint is p (1) + p (6) p (2) = 1?

2.3

Spanning Sets and Subspaces

The material covered so far has been relatively concrete and easy to absorb, but
we now have to start thinking about an abstract concept, vector spaces. The
only rules that we may use are those listed in Definition 2.4 and any others we
are able to deduce from them. This, especially for the novice, is not easy and
is somewhat tedious, but well worth the effort.
In the definition of a vector space, we have as axioms the existence of a
zero vector (axiom 5) and an additive inverse (axiom 6) for every vector. One
question that occurs is, how many zero vectors and inverses are there? Lets see

2.3. SPANNING SETS AND SUBSPACES

71

if we can find an answer to this, at least in R2 . Suppose k = (k1 , k2 ) is another

zero vector in R2 . Thus x = x +kk = (x1 , x2 ) + (k1 , k2 ) = (x1 + k1 , x2 + k2 ). This
implies that x1 = x1 + k1 and x2 = x2 + k2 , from which k1 = k2 = 0. Hence
k = (0, 0) = 0 , which shows that R2 has a unique zero vector. Suppose now
that y is an additive inverse for x in R2 . Then 0 = x + y = (x1 , x2 ) + (y1 , y2 ) =
(x1 + y1 , x2 + y2 ). Hence x1 + y1 = 0 = x2 + y2 . Thus y1 = x1 and y2 = x2 ,
and we have uniqueness of inverses, at least in R2 . What about other vector
spaces?
Theorem 2.3. Let V be a vector space. Then
1. If 0 and k are both zero vectors (i.e., both satisfy axiom 5), then 0 = k .
x and y are both additive inverses of x , then x
x = y.
2. If x
x
Thus, in any vector space the zero vector is unique, and the additive inverse x
of any vector x is uniquely determined by x
Proof.
1. Since 0 satisfies 5, we have k + 0 = k but k also satisfies 5; thus 0 + k = 0 .
These two equations plus commutativity of addition imply k = 0 . Thus,
there is only one zero vector.
x + (x
x)] = (yy + x ) + (x
x) = 0 + (x
x) = x
x. We remind
2. y = y + 0 = y + [x
the reader that y + x = 0 since y is assumed to be an additive inverse for
x.
Some additional properties of vector spaces are listed in the following theorem.
The reader should try to prove them for V equal to R2 ; then read the proof of
Theorem 2.4 and compare his or her version with ours.
Theorem 2.4. Let V be a vector space. Let X be any vector in V and a any
number. Then
X =0
1. 0X
2. a00 = 0
X = X
X
3. (1)X
X = 0 only if a = 0 or X = 0
4. aX
Proof.
1. Let x be any vector in V . Then,
x = 1x
x + 0x
x = (1 + 0)x
x = 1x
x=x
x + 0x
x to both sides of this equation, we get
Adding x
x + x = x
x + (x
x + 0x
x) = (x
x + x ) + 0x
x = 0 + 0x
x = 0x
x
0 = x

72

CHAPTER 2. VECTOR SPACES

2. If a = 0, then from 1 above we have a00 = 0 . Thus we may suppose a 6= 0.
Then,
x = a(00 + a1x ) = a(a1x ) = 1x
x=x
a00 + x = a00 + (aa1 )x
Hence from Theorem 2.3 a00 = 0 .
x = 1x
x + (1)x
x = [1 + (1)]x
x = 0x
x = 0 . Theorem 2.3 now implies
3. x + (1)x
x = x
x.
that (1)x
x = 0 and a 6= 0. Then
4. Suppose ax
x ) = (aa1 )x
x = 1x
x=x
0 = a10 = a1 (ax
x = 0 only when a = 0 or x = 0 .
Thus ax

Quite often when the statement of a theorem or definition is read for the first
time, its meaning is lost in a maze of strange words and concepts. The reader
is strongly urged to always go back to R2 or R3 and try to understand what the
statement means in this perhaps friendlier setting. For example, before proving
Theorem 2.3 we analyzed the special case V = R2 .
xk : k = 1, . . . , n}, we often wish to form
Given a collection of vectors {x
arbitrary sums of these vectors; that is, we wish to look at all vectors x of the
form
n
X
ck x k
(2.1)
x = c1x 1 + c2x 2 + + cnx n =
k=1

where the ck are arbitrary real numbers. Such sums are called linear combinations of the vectors x k .

Example 1. Let x1 = (1, 2, 0), x2 = (1, 1, 0). Determine all linear combinations of these two vectors.
Solution. Any linear combination of these two vectors must be of the form
x = c1x 1 + c2x 2
= c1 (1, 2, 0) + c2 (1, 1, 0) = (c1 , 2c1 , 0) + (c2 , c2 , 0)
= (c1 c2 , 2c1 + c2 , 0)
Thus, no matter what values the constants c1 and c2 are, the third component
of x will always be zero, and it seems likely that as c1 and c2 vary over all pairs
of real numbers, so too will the terms c1 c2 and 2c1 + c2 . We conjecture that
the set of all linear combinations of these two vectors will be the set of vectors
S in R3 whose third component is zero. To prove this, we only have to show
that if x is in S, then there are constants c1 and c2 such that x = c1x 1 + c2x 2 .
Thus suppose x is in S. Then x = (a, b, 0) for some numbers a and b. We want
to find constants c1 and c2 such that
(a, b, 0) = c1 (1, 2, 0) + c2 (1, 1, 0) = (c1 c2 , 2c1 + c2 , 0)

73

2.3. SPANNING SETS AND SUBSPACES

or
c1 c2 = a

and 2c1 + c2 = b

The solution is

2a + b
a+b
c2 =
3
3
Thus (a, b, 0) is a linear combination of (1,2,0) and (1, 1, 0).
c1 =

The set of all linear combinations of a set of vectors will occur frequently
enough that we give it a special name and notation.
xk : k = 1, 2, . . . , n} in some
Definition 2.5. Given a set of vectors A = {x
vector space V , the set of all linear combinations of these vectors will be called
their linear span, or span, and denoted by S[A].
The result of the previous example restated in this notation is
S[(1, 2, 0), (1, 1, 0)] = {(x1 , x2 , 0) : x1 and x2 are arbitrary real numbers}.
Example 2. Find the span of the vector (1, 1).
x2

x2

(1, 1)

x1

x1
x1 + x2 = 0

(a)

(b)
Figure 2.8

S[(1, 1)] = {c(1, 1) : c is any real number}

= {(c, c) : c is any real number}
Thus the span of (1, 1) may be thought of as the straight line x1 + x2 = 0. 
Example 3. Find the span of A = {(1, 1, 0), (1, 0, 0)}.
S[A] = {a(1, 1, 0) + b(1, 0, 0) : a and b arbitrary real numbers}
= {(a b, a, 0) : a, b arbitrary numbers}
= {(x, y, 0) : x, y arbitrary numbers}

74

CHAPTER 2. VECTOR SPACES

x3

x3

S[A] = x1 , x2 plane
(1, 0, 0)

x2

x2

(1,1,0)
x1

x1
(a)

(b)

Figure 2.9

The span of a nonempty set of vectors has some properties that we state and
prove in the next theorem.
xk : k = 1, 2, . . . , n}.
Theorem 2.5. Let S[A] be the linear span of the set A = {x
Then S[A] satisfies axioms 1 through 10 of the definition of a vector space. That
is, S[A] is a vector space.
Proof.
1. Let y and z be in S[A]. Then there are constants aj and bj , j = 1, 2, . . . , n
such that
n
n
X
X
bj x j
aj x j
z=
y=
j=1

j=1

Thus

y +z =
=

n
X

j=1
n
X

aj x j +

n
X

bj x j =

j=1

n
X

(aj x j + bj x j )

j=1

xj
(aj + bj )x

j=1

and we see that y + z is also in S[A].

2. Let x be in S[A] and k a real
Pn number. Then there are constants cj ,
j = 1, 2, . . . , n such that x = j=1 cj x j . Thus

n
n
X
X
x = k
kcj x j
kx
cj x j =
j=1

j=1

x is in S[A].
and we have that kx

3. and 4. Commutativity and associativity of vector addition are true because we are in a vector space to begin with

75

2.3. SPANNING SETS AND SUBSPACES

5. Clearly each of the constants in a linear combination of the vectors x j can

be set equal to zero. Thus we have
n
X

xj =
0x

j=1

n
X

0 =0

j=1

and S[A] contains the zero vector.

P
Pn
x = ( nj=1 cj x j ) =
cj x j and x
6. Let x be in S[A]. Then x =
j=1
Pn
xj . Thus, x
x is also in S[A].
j=1 (cj )x

7. 8, 9, and 10 are true because they are true for all the vectors in V , and
hence these properties are true for the vectors in S[A]. Remember everything in S[A] is automatically in V , since V is a vector space and is closed
under linear combinations.
This fact, that the span of a set of vectors is itself a vector space contained in
the original vector space, is expressed by saying that the span is a subspace of
V.
Definition 2.6. Let V be a vector space. Let W be a nonempty subset of V .
Then if W (using the same operations of vector addition and scalar multiplication as in V ) is also a vector space, we say that W is a subspace of V .
Example 4. V = R3 . Let W = {(r, 0, 0) : r is any real number}. Show that
W is a subspace of V .
x3

x2
W

x1

Figure 2.10

Solution. Since the operations of vector addition and scalar multiplication will
be unchanged, we know that W satisfies properties 3, 4, 7, 8, 9, and 10. Hence,
we merely need to verify that W satisfies 1, 2, 5, and 6 of Definition 2.4. Thus
suppose that x and y are in W . Then x = (r, 0, 0) and y = (s, 0, 0) for some
numbers r and s, and
x + y = (r, 0, 0) + (s, 0, 0) = (r + s, 0, 0)

76

CHAPTER 2. VECTOR SPACES

Thus x + y is in W . Now let a be any real number, then

x = a(r, 0, 0) = (ar, 0, 0)
ax
x are in W we note that
is also in W . To see that 0 and x
x = (0, 0, 0) and
0 = 0x

x = (r, 0, 0) = (r, 0, 0)

Thus W is a subspace of V = R3 .

Example 5. Let V be any vector space and let A be any nonempty subset of
V . Then S[A] is a subspace of V . This is merely Theorem 2.5 restated.

x], which
Example 6. Let x = (a, b, c) be any nonzero vector in R3 . Then S[x
is a subspace of R3 , is just the straight line passing through the origin and the
point with coordinates (a, b, c). See Figure 2.11. The reader should verify the
details of this example.

To verify that a subset of a vector space is a subspace can be tedious. The
following theorem shows that it is sufficient to verify axioms 1 and 2 of Definition 2.4.

x = (a, b, c)

Figure 2.11

Theorem 2.6. A nonempty subset W of a vector space V is a subspace of V

if and only if
1. For any x and y in W , x + y is also in W .
x is in W .
2. For any scalar c and any vector x in W , cx
Proof. If W is a subspace of V , clearly 1 and 2 must be true. Thus, suppose 1
and 2 are true. Then we need to verify that axioms 3 through 10 in the definition
of a vector space are also true. Since we have not changed how vectors are added
or multiplied by scalars, axioms 3, 4, and 7 through 10 are automatically true.
Thus, only axioms 5 and 6 need to be verified.

77

2.3. SPANNING SETS AND SUBSPACES

5. To show that zero is in W , we use the fact that W is assumed to be

nonempty. Let x be any vector in W . Then since W is closed under scalar
x and 0 must be in W .
multiplication we have 0 = 0x
x = x
x must also be in W .
6. Let x be any vector in W , then (1)x
In the following, whenever we wish to indicate that a subspace W equals
S[A], for some set A, we will say W is spanned by A or A is a spanning set of
W.
Example 7. Let W be that subspace of R4 spanned by the vectors (1, 1, 0, 2),
(3,0,1,6), and (0, 1, 1, 1). Is the vector x = (2, 3, 4, 0) in W ?

Solution. If x is in W , it must be a linear combination of the three vectors

whose span is W . Thus, there should exist constants c1 , c2 , and c3 such that
(2, 3, 4, 0) = c1 (1, 1, 0, 2) + c2 (3, 0, 1, 6) + c3 (0, 1, 1, 1)
This vector equation has a solution if and
solution.
c1 + 3c2
c1
+ c3
c2 + c3
2c1 + 6c2 c3

only if the following system has a

= 2
= 3
= 4
= 0

A few row operations on the augmented matrix of this system show that it is
row equivalent to the matrix

1 3 0
2
0 1 0
0

0 0 1
4
0 0 0 5

The nonhomogeneous system, which has the above matrix as its augmented
matrix, has no solution. This means that W does not contain the given vector.
Example 8. Let V = R2 . Let W = {(x, sin x) : x is a real number}. Show
x of every vector x in W , it
that while W contains 0 and the additive inverse x
still is not a subspace.
Solution. To see that 0 = (0, 0) is in W , we only need observe that sin 0 = 0.
Thus (0, sin 0) = (0, 0) is in W . If x is in W , then x = (r, sin r) for some number
r. But then
x = (r, sin r) = (r, sin r) = (r, sin(r))
x
must also be in W . To see that W is not a subspace, we note that if W were a
subspace, c(1, sin 1) must be in W for any choice of the scalar c. Since sin 1 > 0,
by making c large enough we will have c sin 1 > 1. Since 1 sin x 1 for all
x, we cannot have sin x = c sin 1 for any x. Thus, c(1, sin 1) = (c, csin 1) is not
in W . Hence, W is not closed under scalar multiplication, and it cannot be a
subspace.


78

CHAPTER 2. VECTOR SPACES

c(1, sin (1))

(1, sin (1))

4.0

Figure 2.12

x : x is in Rn and Ax
x = 0 }.
Example 9. Let A be an m n matrix. Let K = (x
Thus K is the solution set of the system of linear homogeneous equations whose
coefficient matrix is A. Show that K is a subspace of Rn . Notice that Rn is
being thought of as the set n 1 matrices.
Solution. We first note that 0 is in K, since A00 = 0 . Thus K is nonempty.
Suppose now that x 1 and x 2 are in K. Then
x 1 + x 2 ) = Ax
x 1 + Ax
x2 = 0 + 0 = 0
A(x
and K is closed under addition. To see that K is closed under scalar multiplication we have, if x is in K and c is any number:
x ) = cA(x
x ) = c00 = 0
A(cx
Hence, by Theorem 2.6, K is a subspace of Rn .

Example 10. Let V = R4 and let W = {(x1 , x2 , x3 , x4 ) : 2x1 x2 + x3 = 0}.

Show W is a subspace of R4 .
Solution. Note that W is the solution set of a system of homogeneous equations.
Thus, by the previous example W is a subspace.






1 1 1 0 0 0 0 0
Example 11. Show that F =
spans M22 .
0
0 0 0 1 1 1 0


a1 a2
be an arbitrary vector (matrix) in M22 . Then we
Solution. Let A =
a3 a4
have to find constants c1 , c2 , c3 , and c4 such that










0 0
0 0
1 0
1 1
a1 a2
+ c4
+ c3
+ c2
= c1
1 0
1 1
0 0
0
0
a3 a4
Thus, we have the equations
a1 = c1 + c2

a2 = c1

a3 = c3 + c4

a4 = c3

2.3. SPANNING SETS AND SUBSPACES

79

Hence,


a1
a3



a2
1
= (a2 )
0
a4

0
+ a4
1




1
1 0
+ (a1 + a2 )
0
0 0



0
0 0
+ (a3 a4 )
1
1 0

Thus, M22 = S[F ].

Problem Set 2.3

1. Show that S[(1, 0)(0, 1)] = R2 .
2. Show that S[(1, 0, 0), (0, 1, 0)] equals the x1 , x2 plane in R3 .
3. Show that S[(1, 0, 1)(1, 2, 1)] = {(x1 , x2 , x3 ) : x1 x2 x3 = 0}.
4. Show that S[(1, 1)] = {(x1 , x2 ) : x1 x2 = 0}.
5. Let A = {(1, 1, 0), (1, 0, 1)}. Determine which if any of the following
vectors is in S[A] : (1, 1, 1), (0, 1, 1), (2, 3, 1).
6. Let A = {(1, 2), (6, 3)}. Show that S[A] = R2 .
x] is a straight
7. Let x = (x1 , x2 ) be any nonzero vector in R2 . Show that S[x
line passing through the origin and the point with coordinates (x1 , x2 ).
8. What is S[(0, 0)]?
9. Let V be a vector space. Let W be that subset of V which consists of just
the zero vector. Prove the following:
a. W is a subspace of V .
b. Show that V is a subspace of itself.
10. Let A be any set of vectors in some vector space V . Show that A is
contained in S[A]. Show that if H is any subspace of V containing A,
then S[A] is also in H. Thus, S[A] is the smallest subspace of V which
contains A.
11. Let V = R3 . Let A1 = {(1, 1, 0)}, A2 = {(1, 1, 0), (0, 0, 1)}. Show that
S[A1 ] is contained in S[A2 ] and describe these two subspaces geometrically.
12. Let f (x) = ax for some constant a. Show that the graph of f = {(x, f (x)) :
x is any real number} is a subspace of R2 . Conversely suppose we have a
function f (x) whose graph is a subspace of R2 . Show that f (x) = ax for
some constant a. (Hint: What must a equal?)

80

CHAPTER 2. VECTOR SPACES




1 0 3
x = 0 , for x in R3 .
. Let K be the solution set of Ax
4 4 6
From Example 9 we know that K is a subspace of R3 . Find a vector x 0
x0 ] = K.
such that S[x

13. Let A =

14. Let W = {(x1 , x2 , x3 ) : x1 x2 + x3 = c}. For which values of c, if any,

is W a subspace of R3 ?
15. Example 11 exhibited a spanning set of M22 with four vectors. Can you
find a spanning set of M22 with five vectors? With three vectors?
16. Find spanning sets for each of the following vector spaces:
a. V = {(x1 , x2 , x3 ) : x1 + x2 6x3 = 0}

b. V = {pp : p is in P2 and p (1) = 0}

c. V = {A = [aij ] : A is in M23 and

P3

j=1

aij = 0 for i = 1, 2}

17. Let V be the vector space of all n n matrices. Show that the following
subsets of V are subspaces:
a. W = all n n scalar matrices = {cIn : c is any number}

b. W = all n n diagonal matrices = {[dj jk } : dj any number}

c. W = all n n upper triangular matrices

d. W = all n n lower triangular matrices

18. Let W be the set of all n n invertible matrices. Show that W is not a
subspace of the vector space of all n n matrices.
19. Let W be any subspace of the vector space Mmn of all m n matrices.
Show that W T , that subset of Mnm consisting of the transposes of all the
matrices in W , is a subspace of Mnm .
20. Let P2 be all polynomials of degree 2 or less. Let W = {a + bt : a and b
arbitrary real numbers}. Is W a subspace of P2 ?
21. Let V be a vector space. Let W1 and W2 be any two subspaces of V . Let
x : x belongs to both W1 and W2 }. Show that W is a
W = W1 W2 = {x
subspace of V .
x: x
22. Let V, W1 , and W2 be as in problem 21. Let W = W1 W2 = {x
belongs to W1 or W2 }. Show that W need not be a subspace of V .
23. Let V, W1 , and W2 be as in problem 21. Define W1 + W2 as that subset of
V which consists of all possible sums of the vectors in W1 and W2 , that is
u + v : u and v any vectors in W1 and W2 }
W1 + W2 = {u
Show W1 + W2 is a subspace of V and that any other subspace of V which
contains W1 and W2 must also contain W1 + W2 . Thus, W1 + W2 is the
smallest subspace of V containing W1 W2 .

2.4. LINEAR INDEPENDENCE

81

24. Let V = C[a, b]. Let W be that subset of V which consists of all polynomials. Is W a subspace of V ?
25. Let W = {. . . , 2, 1, 0, 1, 2, . . .}. Is W a subspace of R1 ?
26. Let W = {(x1 , x2 ) : x1 or x2 equals zero}. Let W1 = {(x1 , 0)} and
W2 = {(0, x2 )}. Thus, W = W1 W2 . Show that both W1 and W2 are
subspaces of R2 , and that W is not a subspace.
27. Let W = {(x1 , x2 ) : x21 + x22 c}. For which values of c is W a subspace
of R2 ?
28. Let W1 = {pp : p is in P2 and p(1) = 0}. Let W2 = {pp : p is in P2 and
p (2) = 0}.
a. Find a spanning set for W1 W2 .
b. Find spanning sets for W1 and W2 each of which contains the spanning set you found in part a.
29. Example 9 shows that the solution sets of homogeneous systems of equations are subspaces. Find spanning sets for each of the solution spaces of
the following systems of equations:
a. 2x1 + 6x2 x3 = 0
b. 2x1 + 6x2 x3 = 0, x2 + 4x3 = 0
c. 2x1 + 6x2 x3 = 0, x2 + 4x3 = 0, x1 + x2 = 0


2 6 1
30. Let A =
. Let V = {B : B M34 and AB = 024 }.
3 1 0
a. Show V is a subspace of M34 .
b. Find a spanning set for V .

2.4

Linear Independence

In this section we discuss what it means to say that a set of vectors is linearly independent. We encourage the reader to go over this material several times. The
ideas discussed here are important, but hard to digest, thus needing rumination.
Consider the two vectors x = (1, 0, 1) and y = (0, 1, 1) in R3 . They have
the following property: Any vector z in their span can be written in only one
way as a linear combination of these two vectors. Lets check the details of this.
Suppose there are constants c1 , c2 , c3 , and c4 such that
z = c1x + c2y = c3x + c4y

(2.2)

What we want to show is that c1 = c3 and c2 = c4 ; that is, the coefficients of x

must be equal and the coefficients of y must also be equal. The two expressions
for z lead to the equation
x + (c2 c4 )yy
0 = z z = (c1 c3 )x

(2.3)

82

CHAPTER 2. VECTOR SPACES

x + byy we must have a = 0 = b.

Thus, we need to show that anytime 0 = ax
Suppose now that
0 = (0, 0, 0) = a(1, 0, 1) + b(0, 1, 1) = (a, b, a + b)
Clearly this can occur only if a = 0 = b. This property of the two vectors x and
y is given a special name.
xk : k = 1, . . . , n} in a vector space V is said
Definition 2.7. A set of vectors {x
to be linearly independent if whenever
0 = c1x 1 + c2x 2 + + cnx n =

n
X

ck xk

k=1

then
c1 = c2 = = cn = 0
That is, the zero vector can be written in only one way as a linear combination
of these vectors.
A set of vectors is said to be linearly dependent if it is not linearly indepenxk : k = 1, . . . , n} is linearly dependent only if
dent. That is, a set of vectors {x
there are constants c1 , c2 , . . . , cn not all zero such that 0 = c1x 1 + c2x 2 + +
cnx n .
Example 1. Show that the vectors {(1, 0, 1), (2, 0, 3), (0, 0, 1)} are linearly dependent.
Solution. We need to find three constants c1 , c2 , c3 not all zero such that
c1 (1, 0, 1) + c2 (2, 0, 3) + c3 (0, 0, 1) = (0, 0, 0)
Thus we have to find a nontrivial solution to the following system of equations:
c1 + 2c2
=0
c1 + 3c2 + c3 = 0
This system has many nontrivial solutions. A particular one is c1 = 2, c2 = 1,
and c3 = 1. Since we have found a nontrivial linear combination of these
vectors, namely, 2(1, 0, 1)(2, 0, 3)+(0, 0, 1), which equals zero, they are linearly
dependent.

Example 2. Determine whether the set
{(1, 0, 1, 1, 2), (0, 1, 1, 2, 3), (1, 1, 0, 1, 0)}
is linearly dependent or independent.
Solution. We have to decide whether or not there are constants c1 , c2 , c3 not all
zero such that
(0, 0, 0, 0, 0) = c1 (1, 0, 1, 1, 2) + c2 (0, 1, 1, 2, 3) + c3 (1, 1, 0, 1, 0)

83

2.4. LINEAR INDEPENDENCE

The linear system that arises from this is
c1

+ c3
c2 + c3
c1 + c2
c1 + 2c2 + c3
2c1 + 3c2

=0
=0
=0
=0
=0

The only solution is the trivial one, c1 = c2 = c3 = 0. Thus the vectors are
linearly independent.

To say that a set of vectors is linearly independent is to say that the zero
vector can be written in only one way as a linear combination of these vectors;
that is, every coefficient ck must equal zero. The following theorem shows that
even more is true for a linearly independent set of vectors.
xk : k = 1, . . . , p} is a linearly independent set of
Theorem 2.7. If A = {x
vectors, then every vector in S[A] can be written in only one way as a linear
combination of the vectors in A.
Proof. Let y be any vector in S[A]. Suppose that we have
y=

p
X

ak x k =

p
X

bk x k

k=1

k=1

We wish to show that ak = bk for each k. Subtracting y from itself we have

0 =y y =

p
X

k=1

ak x k

p
X

bk x k =

k=1

p
X

k=1

xk
(ak bk )x

Since the set A is linearly independent we must have

a1 b 1 = 0

a2 b 2 = 0

ap b p = 0

Thus, we can write any vector in S[A] as a linear combination of the linearly
independent vectors x k in one and only one way.
The unique constants ak that are associated with any vector x in S[A] are
given a special name.
Definition 2.8. Given
Pn a linearly independent set of vectors A = {xk : k =
1, . . . , n}. If x =
k=1 akx k , the coefficients ak , 1 k n, are called the
coordinates of x with respect to A. Note that there is an explicit ordering of
the vectors in the set A.
Example 3. Show that the set A = {(1, 1, 1), (0, 1, 0), (1, 0, 2)} is linearly
independent, and then find the coordinates of (2, 4, 6) with respect to A.
Solution. To see that A is linearly independent suppose
c1 (1, 1, 1) + c2 (0, 1, 0) + c3 (1, 0, 2) = (0, 0, 0)

84

CHAPTER 2. VECTOR SPACES

The system of equations that is equivalent to this vector equation, namely,

c1
c3 = 0
c1 + c2
=0
c1
+ 2c3 = 0
has only the trivial solution. Thus A is a linearly independent set of vectors. We
next want to write, if possible, (2, 4, 6) as a linear combination of the vectors
in A. Thus we want constants ck , k = 1, 2, 3, such that
(2, 4, 6) = c1 (1, 1, 1) + c2 (0, 1, 0) + c3 (1, 0, 2)
The associate system is
c1
c3 = 2
c1 + c2
= 4
c1
+ 2c3 = 6
and its unique solution is c1 = 10, c2 = 14, c3 = 8. Thus (2, 4, 6) =
10(1, 1, 1) 14(0, 1, 0) + 8(1, 0, 2) and the coordinates of x = (2, 4, 6) with
respect to A are [10, 14, 8].

Example 4. Show that the set F is a linearly independent subset of M22 .





1 1 1 0 0 0 0 0
F =
0
0 0 0 1 1 1 0
Solution. Suppose there are constants c1 , . . . , c4 such







0
1 0
1 1
0 0
+ c3
+ c2
= c1
1
0 0
0
0
0 0

that


0
0
+ c4
1
1


0
0

Thus, since the equations

c1 + c2 = 0

c1 = 0 c3 + c4 = 0 c3 = 0

have only the trivial solution, F is linearly independent. From

Section 2.3 we know that S[F ] = M22 , and that









1 0
1 1
1 0
0 0
0
=0
+1
+0
+0
0 0
0
0
0 0
1 1
1


1 0
Thus, the coordinates of
with respect to F are [0,1,0,0].
0 0

Example 11 in
0
0




Example 5. Show that the set F = {1 + t, t2 , t 2} is a linearly independent

subset of P2 .
Solution. Suppose there are constants c1 , c2 , and c3 such that
0 = c1 (1 + t) + c2 (t2 ) + c3 (t 2)

= c2 t2 + (c1 + c3 )t + (c1 2c3 )

85

2.4. LINEAR INDEPENDENCE

Then these constants satisfy the equations
c1 2c3 = 0

c1 + c3 = 0

c2 = 0

Since the only solution to this system is the trivial one, F is a linearly independent subset of P2 .

xk : k = 1, . . . , n} is linearly dependent if and
Theorem 2.8. A set of vectors {x
only if one of its vectors can be written as a linear combination of the remaining
n 1 vectors.
Proof. Suppose one of the vectors, say x 1 , can be written as a linear combination
of the others. Then we have
x 1 = c2x 2 + c3x 3 + + cnx n
for some constants ck . But then
0 = x 1 c2x 2 cnx n
and we have found a nontrivial (the coefficient of x1 is 1) linear combination of
these n vectors that equals the zero vector. Thus, this set of vectors is linearly
dependent. Suppose now that there is a nontrivial linear combination of these
vectors that equals 0 ; that is
0 = c1x 1 + + cnx n
and not all the constants ck are zero. We may suppose without loss of generality
that c1 is not zero (merely rename the vectors). Solving the above equation for
x 1 , we have
x1 =

c2
cn
1
(c2x 2 cnx n ) = x2 xn .
c1
c1
c1

Hence, we have written one of the vectors in our linearly dependent set as a
linear combination of the remaining vectors.
If a linearly dependent set contains exactly two vectors, then one of them
must be a multiple of the other. In R3 , three nonzero vectors are linearly
dependent if and only if they are coplanar, i.e., one of them lies in the plane
spanned by the other two. For example, the three vectors (1,0,1), (2,0,3), and
(0,0,1), as we saw in Example 1, are linearly dependent and all three of them
lie in the plane generated by (1,0,1) and (2,0,3); that is, the plane x2 = 0.

Problem Set 2.4

1. Show that (1,0,1) is in the span of {(2, 0, 0), (0, 0, 1)}.
2. Show that S[(2, 0, 0), (0, 0, 1)] = S[(2, 0, 0), (0, 0, 1), (1, 0, 1)].

86

CHAPTER 2. VECTOR SPACES

3. Let A be a collection of vectors in a vector space V . Suppose that x is
any vector in S[A]. Show S[A, x ] = S[A].
4. Determine whether the following sets of vectors are linearly independent.
a. {(1, 3), (1, 4), (8, 12)}

b. {(1, 1, 4), (8, 3, 2), (0, 2, 1)}

c. {(1, 1, 2, 3), (2, 3, 0, 4), (8, 7, 4, 18)}

5. Which of the following sets of vectors is linearly dependent?

a. {1, 2, 3, 0), (6, 1, 0, 0), (1, 2, 0, 3)}

b. {(1, 1, 6), (4, 2, 0), (1, 1, 1), (1, 0, 1)}

c. {(2, 1, 4), (1, 1, 1), (1, 1, 1)}

6. Show that any set of vectors containing the zero vector must be linearly
dependent.
7. Suppose that A is a linearly dependent set of vectors and B is any set
containing A. Show that B must also be linearly dependent.
8. Show that any nonempty subset of a linearly independent set is linearly
independent.
9. Show that any set of four or more vectors in R3 must be linearly dependent.
10. Let A = {(0, 1, 1), (1, 0, 1), (1, 1, 0)}. Is A linearly dependent? S[A] =?
11. Let A = {1 + t, 1 t2 , t2 }. Is A a linearly independent subset of P2 ?
S[A] =?
12. Let A = {sin t, cos t, sin(t + )}. Is A a linearly independent subset of
C[0, 1]? Does A span C[0, 1]?
13. Let A = {(x1 , x2 ) : x21 + x22 = 1}. Is A a linearly independent subset of
R2 ? Is A a spanning subset of R2 ?
14. Show that F = {1 + t, t2 t, 2} is a linearly independent subset of P2 .
Show S[F ] = P2 .
15. Show that {sin t, sin 2t, cos t} is a linearly independent subset of C[0, 1].
Does it span C[0, 1]?
1
16. Let V = {pp(t) : p is in P2 and 0 p (t)dt = 0}. Find a spanning set of
V that has exactly two vectors in it. Show that your set is also linearly
independent.
17. Let V = {pp(t) : p is in P3 and p (0) = 0, p (1)pp(0) = 0}. Find a spanning
set of V that has exactly two vectors. Is the set also linearly independent?

2.5. BASES

87

18. Let V = {A : A is in M23 and AB = AC = 022 }, where

1
1
1 1
1
B = 1 1
C = 1
1
1
1 1

Find a linearly independent spanning set of V . How many vectors are in

your set?

19. Let A be a subset of a vector space V . We say that A is linearly indepenx1 , . . . , x p } is a finite subset of A and c1x 1 + +cpx p =
dent if, whenever {x
0, then each ck = 0.
a. Show that if A is a finite set of vectors to begin with, then Definition 2.7 and the above definition are equivalent.
b. Show Theorem 2.7 is also true using this definition of linear independence. Note: S[A] is the collection of all possible finite linear
combinations of the vectors in A.
20. Let V = C[0, 1], the space of real-valued continuous functions defined on
[0,1]. Let A = {1, t, . . . , tn , . . .}.
a. Show that A is a linear independent set as defined in problem 19.
b. What is S[A]?
21. Let V be the set of all polynomials with only even powers of t (constants
have even degree). Thus 1 + t is not in V , 1 + t + t2 is not in V , but 1 + t2
and t4 + 5t8 are in V . Let A = {1, t2 , t4 , . . . , t2n , . . .}.
a. Show that V is a vector space and that A is a linearly independent
subset of V .
b. S[A] =?

2.5

Bases

In the preceding two sections we discussed two different types of subsets of a

vector space V . In Section 2.3 we covered spanning sets, i.e., subsets A with
the property that any vector in V can be written as a linear combination of
vectors in A. Section 2.4 was concerned with sets that are linearly independent.
This section discusses properties of sets that are both linearly independent and
spanning. Such subsets are called bases of V . Besides their defining properties
all bases of a vector space share one other common feature: They have exactly
x1 , . . . , xp } and B =
the same number of vectors in them. That is, if A = {x
y
y
{y 1 , . . . , n } are linearly independent and they both span V , then n = p. Thus
the number of vectors in a basis of V depends only on V and not on which
linearly independent spanning set we happen to have chosen.
Before proving this useful and interesting fact, we prove two lemmas. These
lemmas aid us in the construction of bases.

88

CHAPTER 2. VECTOR SPACES

x1 , . . . , x n } is linearly dependent. Then it is posLemma 2.1. Suppose A = {x

sible to remove one of the vectors x k from A without changing S[A]; that is,
x1 , . . . , x k1 , x k+1 , . . . , x k ] = S[A].
S[x
Proof. Since A is linearly dependent, there are constants ck , 1 k n, not all
zero, such that
0 = c1 x 1 + + cn x n
(2.4)
Suppose, for convenience, that c1 6= 0. Then we may solve (2.4) for x1 and
obtain
1
(2.5)
x 1 = [c2x 2 + + cnx n ]
c1
x2 , . . . , x n ]. Let B = {x
x2 , . . . , x n }. Since B is a
We now claim that S[A] = S[x
subset of A, we have S[B] S[A]. Thus, to show these two subspaces of V are
equal, we only need show that any vector in S[A] is also in S[B]. To this end,
let y be in S[A]. Then there are constants ak , 1 k n, such that
y = a1x 1 + a2x 2 + + anx n


1
= a1
[c2 x2 + + cnx n ] + a2 x2 + + anx n
c1




a1 cn
a1 c2
x
xn
= a2
2 + + an
c1
c1
Since y can be written as a linear combination of the vectors in B, y is in
S[B].
Example 1. Let A = {(1, 1), (3, 2), (1 1)}. Its easy to see that the span of
A is R2 . Moreover since
(0, 0) = 5(1, 1) 2(3, 2) + (1, 1)
the set A is linearly dependent. Thus we may remove one of the three vectors
from A and still have a set with the same span, i.e., R2 . Which vector can we
delete? Since each of the vectors in the linear combination above has a nonzero
coefficient, we may delete any one of the three. Lets remove the second one.
Then we know that
S[(1, 1), (1, 1)] = R2
We note that these last two vectors are linearly independent. In fact this is
one way in which a linearly independent spanning set of a vector space can be
constructed. That is, from a linearly dependent spanning set remove, one by
one, those vectors that depend linearly on the remaining vectors, until a linearly
independent set is obtained.

x1 , . . . , x n } be a linearly independent set. Suppose that
Lemma 2.2. Let A = {x
the vector y is not in S[A]. Then the set obtained by adding y to A is linearly
independent.

89

2.5. BASES
Proof. Suppose there are constants c, c1 , . . . , cn such that
0 = cyy + c1x 1 + + cnx n

(2.6)

We wish to show that each of these n + 1 constants is zero. Suppose that c is

not zero. Then we may solve (2.6) for y as a linear combination of the vectors
in A, contradicting our assumption that y is not in the span of A. Thus c = 0.
But then (2.6) reduces to a linear combination, of the vectors in A, which equals
the zero vector. Since A is linearly independent, each of the constants ck = 0.
Thus, the set {yy , x 1 , . . . , x n } is linearly independent.
The last lemma is also used to construct linearly independent spanning subsets of a vector space. We start out with a linearly independent subset. If its
span is not V , we add to it a vector y not in its span. If the new set spans
V , were done. If not, another vector is added. This process continues until we
reach a spanning set.
For most of the vector spaces weve seen either of the above procedures
is successful in constructing bases. There are, however, some vector spaces
for which this does not work. Those which are not finite dimensional. See
Definition 2.10
Example 2. Let A = {(1, 1, 1), (0, 1, 1)}. Find a vector y such that when y
is added to A, we have a linearly independent spanning subset of R3 .
S[A] = {c1 (1, 1, 1) + c2 (0, 1, 1)}
= {(c1 , c1 + c2 , c2 c1 )}

= {(x1 , x2 , x3 ) : x2 2x1 = x3 }

From this last description of S[A], it is easy to find vectors that are not in
the span of A. For example, (1,0,0) is one such vector. Since A is linearly
independent, Lemma 2.2 now tells us that the set
{(1, 0, 0), (1, 1, 1), (0, 1, 1)}
is also linearly independent. The reader may easily verify, and should do so,
that this set does indeed span R3 .

Definition 2.9. Let V be a vector space. A subset B of V is called a basis of
V if it satisfies:
1. B is linearly independent.
2. B is a spanning set for V ; that is, every vector in V can be written as a
linear combination of the vectors in B.
It can be shown that every vector space containing more than the zero vector
has a basis. We shall not prove this theorem for arbitrary spaces but instead
will exhibit a basis for most of the vector spaces discussed in this text.
Example 3.

90

CHAPTER 2. VECTOR SPACES

a. For R1 , the vector space of real numbers, any set consisting of just one
nonzero number is a basis.
b. In R2 let e 1 = (1, 0), e 2 = (0, 1). The set {ee1 , e 2 } is a basis of R2 and is
the standard basis of R2 .
c. If V = Rn , define e j , 1 j n, by
e 1 = (1, 0, . . . , 0)
e 2 = (0, 1, 0, . . . , 0)
..
.
e j = (0, . . . , 0, 1, 0, . . . , 0), a 1 is in the jth slot.

The set {eej : 1 j n} is a basis for Rn , as the reader can easily show. This
particular set is the standard basis of Rn .

Theorem 2.9. Let V be a vector space that has a basis consisting of n vectors.
Then:
a. Any set with more than n vectors is linearly dependent.
b. Any spanning set has at least n vectors.
c. Any linearly independent set has at most n vectors.
d. Every basis of V contains exactly n vectors.
Proof. Part a. Suppose that A = {ff 1 , . . . , f n } is the given basis of V with n
x1 , . . . , x n , x n+1 } be any subset of V with n + 1 vectors.
vectors. Let B = {x
We want to show that B is linearly dependent. Since A is a basis of V , it is a
spanning set. Hence, there are constants ajk such that
x k = a1kf 1 + + ank f n
We want to find constants ck , 1 k n + 1, not all zero such that
0 = c1x 1 + + cn+1x n+1 =
=

n+1
X
k=1

ck

n
X
j=1

ajk f j =

n
X
j=1

n+1
X

k=1
n+1
X

ck xk
!

ck ajk f j

k=1

If we pick the constants ck in such a manner that each of the coefficients multiplying f j is zero, this linear combination of the vectors in B will equal the zero
vector. In other words, we wish to find a nontrivial solution of the following
system of equations:
aj1 c1 + aj2 c2 + + aj(n+1) cn+1 = 0

91

2.5. BASES

for j = 1, 2, . . . , n. This is a homogeneous system of equations with more

unknowns (n + 1) than equations (n). Therefore, there is a nontrivial solution,
and the set B is linearly dependent. Notice that we did not use the fact that
A is also linearly independent. The only hypothesis needed to show the linear
dependence of any set with n + 1 vectors was the existence of a spanning set
with fewer than n + 1 vectors.
x1 , . . . , x m } be any spanning set of V . We wish to show
Part b. Let B = {x
that B must contain at least n elements. Suppose it doesnt, i.e., m is less
than n. But then by part a, since B is a spanning set, any set with more
than m vectors must be linearly dependent. In particular the set A would have
to be linearly dependent. This contradicts the assumption that A is linearly
independent.
x1 , . . . , x p } is linearly independent; then clearly part
Part c. Suppose B = {x
a implies that p n. That is, any linearly independent set of vectors cannot
have more than n vectors in it.
Part d. Let A1 = {gg 1 , . . . , g p } be any other basis of V . By part b, since A1
is spanning, we must have p n. But reversing the roles of A and A1 , we have
by the same reasoning that p n. Thus every basis of V must have the same
number of vectors.
Definition 2.10. Let V be a vector space. We say that the dimension of V
is n, dim(V ) = n, if V has a basis of n vectors. If V is the vector space that
consists of the zero vector only, we define the dimension of V to be 0. If for any
n, V has a set of n linearly independent vectors, we say that the dimension of
V is infinite.
Most of the vector spaces we discuss in this book will have a basis consisting
of a finite number of vectors. There are, however, a large number of vector
spaces that do not have such a basis. Two examples of such spaces are the set
of all polynomials, and the set of all real-valued continuous functions defined
on some interval of real numbers, i.e., C[a, b]. Both of these spaces are infinitedimensional.
Example 3 exhibited a basis of Rn that contained n vectors. Theorem 2.9
now tells us that any subset of R2 with more than two vectors must be linearly
dependent. We also have that dim(Rn ) = n, n = 1, 2, . . . .
Example 4.
a. Show that dim(Pn ) = n + 1. Since Pn consists of all polynomials with real
coefficients of degree less than or equal to n, we have
Pn = {an + a1 t + + an tn : ak arbitrary real numbers}
We claim that the set A = {1, t, t2 , . . . , tn } is a basis for Pn . It is clear that
any polynomial of degree less than or equal to n can be written as a linear
combination of the vectors in A. We leave it to the reader to show that A
is also linearly independent. Since A has n + 1 vectors, dim Pn = n + 1.

92

CHAPTER 2. VECTOR SPACES

b. The vector space Mmn , consisting of all m n matrices, has dimension
equal to mn. The reader is asked in the problems to show that the matrices
Ejk , 1 j m, 1 k n, which contain a 1 in the j, k position and
zeros everywhere else, form a basis for Mmn .


Example 5. Consider the following system of homogeneous equations:

2x1 x3 + x4 = 0

(2.7)

x2 + x4 = 0

Find a basis for the solution space.



2 0 1 1
Solution. The coefficient matrix
is row equivalent to
0 1
0 1


1 0 21 12
. Thus every solution to (2.7) must satisfy
0 1
0 1
x3 x4
2
x2 = x4
x1 =

We have two free parameters x3 and x4 which we set equal to 1 and 0 to get
one solution. To get a second solution, which is linearly independent of the
first, we set x3 and x4 equal to 0 and 1, respectively. Thus we have the set
S = {( 12 , 0, 1, 0), ( 21 , 1, 0, 1)} which is contained in the solution space of
(2.7). Clearly S is linearly independent. [The last two slots look like (1,0) and
(0,1).] Moreover, every solution of (2.7) can be written as a linear combination
of these two vectors; for suppose (x1 , x2 , x3 , x4 ) satisfies (2.7). Then we have


x3 x4
, x4 , x3 , x4
(x1 , x2 , x3 , x4 ) =
2




1
1
= x3
, 0, 1, 0 + x4 , 1, 0, 1
2
2
Thus, S is a basis for the solution space of (2.7), and the dimension of this
subspace of R4 is 2.

Example 6. Consider the system of linear equations
x1 2x2 + x3 = 0

x2 x3 = 0
2x1 + x2 + x3 = 0

Find the dimensional of the solution space of this system.

Solution. The coefficient matrix

1
0
2

2
1
1 1
1
1

93

2.5. BASES
of this system is row equivalent to

1 2
1
0
1 1
0
0
4

This second system has only the trivial solution. Therefore, the solution space
K consists of just the zero vector and we have dim(K) = 0.


Example 7. Let V = R2 . Let W = {(x1 , x2 , x3 ) : x3 = 2x1 + x2 }. Find a

basis for W and then extend it to a basis of V ; i.e., add enough vectors to the
basis of W to get a basis for V .
Solution. Since W = {(x1 , x2 , 2x1 + x2 ) : x1 and x2 arbitrary}, there are two
free parameters x1 and x2 . Lets alternately set them equal to 0 and 1 to get a
basis for W . This gives us
f 1 = (1, 0, 2) : x1 = 1, x2 = 0
f 2 = (0, 1, 1) : x1 = 0, x2 = 1
Both f 1 and f 2 belong to W , and they are linearly independent. To see that
they span W , let x be in W . Then for some x1 and x2
x = (x1 , x2 , 2x1 + x2 ) = (x1 , 0, 2x1 ) + (0, x2 , x2 )
= x1 (1, 0, 2) + x2 (0, 1, 1) = x1f 1 + x2f 2
Since the set {ff 1 , f 2 } is linearly independent and spans W , it is a basis. We
now face the problem of extending this basis to one for R3 . Since dim(R3 ) = 3,
we need one additional vector. Thus, we want to find a vector y such that the
set {ff 1 , f 2 , y } is linearly independent. Let y = (1, 1, 0). One easily sees that y
is not in W . Hence, by Lemma 2.2, we may conclude that {ff 1 , f 2 , y } is linearly
independent. We leave it for the reader to show that this set spans R3 . (Hint:
Suppose not. Then we should be able to find a linearly independent subset of
R3 with four vectors in it! See c. of Theorem 2.9.)

Figure 2.13 shows f 1 , f 2 , and y . It is clear that W = S[ff 1 , f 2 ] is the plane
in R3 passing through the three points (1,0,2), (0,1,1), and (0,0,0). The vector
y = (1, 1, 0) clearly does not lie in this plane.
The preceding example used Theorem 2.10 to infer that a certain set is a
basis.
Theorem 2.10. Let V be a vector space with dim(V ) = n. Then
a. Any linearly independent set with n vectors is a basis of V .
b. Any spanning set with n vectors is a basis of V .
Proof. Suppose first that A is a linearly independent set of n vectors such that
S[A] is not all of V . Then there must be a vector y not in S[A]. But then

94

CHAPTER 2. VECTOR SPACES

x3
f1

f2

x2
y
x1

Figure 2.13

Lemma 2.2 implies that the set consisting of A and y must be linearly independent. This contradicts a. of Theorem 2.9. Therefore, any linearly independent
set with n vectors must also span the vector space. Hence it is a basis. Suppose
next that A is a spanning set of n vectors. Then, by Lemma 2.1, if A is not linearly independent, we may discard a vector from A, obtaining a set with fewer
than n vectors, and which still spans V . This contradicts b. of Theorem 2.9.


3 2 16 23
Example 8. Let A =
. Find a basis for the solution space of
2 1 10 13
X = 0. If this is not a basis for R4 , find a basis of R4 that contains the basis
AX
of the solution space.
Solution. The matrix A is row equivalent to the matrix


1 0 4 3
0 1 2 7

T
X = 0, we must have x2 = 2x3
Thus, if X = x1 , x2 , x3 x4 solves AX
7x4 and x1 = 4x3 3x4 . Setting x3 = 1, x4 = 0, and then x3 = 0, x4 = 1, we
have two linearly independent solutions to our system:
f 1 = (4, 2, 1, 0) and f 2 = (3, 7, 0, 1)
Clearly the set (ff 1 , f 2 } is a basis for the solution space. To extend this set to a
basis of R4 we have to find two more vectors g 1 and g 2 such that {ff 1 , f 2 , g 1 , g 2 }
is linearly independent. R4 has as a basis {ee1 , e2 , e3 , e4 } where e1 = (1, 0, 0, 0),
etc. Two of these ek s will not depend linearly on f 1 and f 2 . We need to find
such a pair. A quick inspection of f 1 and f 2 shows us that e 1 and e 2 will work.
We leave the details of showing that {ff 1 , f 2 , e 1 , e 2 } is linearly independent to
the reader.


95

2.5. BASES

We next outline a systematic procedure that can be used to extend any

x1 , . . . , x r } is a linearly
linearly independent set to a basis. Suppose A = {x
independent subset of a vector space V , where dim(V ) = n and r < n. We need
to find n r vectors to adjoin to A to form a basis. Suppose {ff 1 , . . . , f n } is any
x1 , . . . , x r , f 1 , . . . , f n }. It is linearly dependent
basis of V . Consider the set {x
and its span is all of V . Now just discard those f k which depend on the x j and
the remaining f p , until r vectors have been discarded. The vectors that are left
form a basis containing A.
Example 9. Extend the linearly independent set of vectors B = {(2, 0, 1, 0, 0),
(1, 1, 0, 1, 1)} to a basis of R5 .
Solution. Let A be the 7 5 matrix whose first two rows are the vectors in B
and whose last five rows are the standard basis vectors of R5 .

2 0 1 0 0
1 1
0 1 1

1 0
0 0 0

0 0 0
A=
0 1

0 0
1 0 0

0 0
0 1 0
0 0
0 0 1

Now use elementary row operations but not row interchanges to find a matrix
that is row equivalent to A and that has two rows of zeros. Zeroing out rows
corresponds to discarding those vectors in the standard basis which depend
linearly on the remaining vectors. After a few row operations, we have A row
equivalent to the matrix

2 0 1 0 0
1 1
0 1 1

1 0
0 0 0

0 1
0 0 0

0 0
0 0 0

0 0
0 1 0
0 0
0 0 0
Thus the set {(2, 0, 1, 0, 0), (1, 1, 0, 1, 1), e1 , e 2 , e 4 } is a basis of R5 that contains our original set A.

Example 10. Find a basis for V = {pp : p is in P2 , p (1) = 0}. If p is in P2 ,
then p (t) = a0 + a1 t + a2 t2 , and p (1) = 0 implies
p (1) = 0 = a0 + a1 + a2
Thus, F = {1 t2 , t t2 } is a subset of V . F is easily shown to be linearly
independent. To see that F spans V , let p be in V . Then
p (t) = a0 + a1 t + a2 t2
= a0 + a1 t + (a0 a1 )t2
= a0 (1 t2 ) + a1 (t t2 )

96

CHAPTER 2. VECTOR SPACES

Thus, p (t) is in S[F ]; F is a basis of V and dim(V ) equals 2.

Problem Set 2.5

1. Find a basis for S[(2, 6, 0, 1, 3), (1, 1, 0, 1, 4), (0, 0, 0, 1, 1)].
2. Let W = {(x1 , x2 , x3 , x4 ) : x1 x2 = x3 }. Show that W is a subspace of
R4 and find a basis of W .
3. Let K be the solution space of the following homogeneous system of equations:
2x1 6xx + 3x3
=0
x2 + x3 + x4 = 0
Find any two linearly independent vectors in K. Show that every other
vector in K can be written as a linear combination of your two vectors,
and conclude that these two vectors are a basis of K. Extend your basis
to a basis of R4 .
4. Show that any subset of R3 containing at most two vectors will not span
all of R3 .
5. Find a basis of R3 that is not the standard one given in the text.
6. Find a basis of P2 that is different from the standard basis {1, t, t2 }.
7. Let K = {(x1 , x2 , x3 ) : 2x1 + x2 = 0, x1 + x2 x3 = 0}.
a. Show that every vector in K can be written as a scalar multiple of
(1, 2, 1). Thus, dim(K) = 1.
b. Find two vectors y 1 and y 2
R3 .


1 0
8. Show that the matrices
,
0 0
M22 , the set of 2 2 matrices.

such that {(1, 2, 1), y 1 , y 2 } is a basis of



0
0

 
1
0
,
0
1

 
0
0
,
0
0


0
form a basis of
1

9. Show that the m n matrices Eij described in Example 4b form a basis

of Mmn , the set of m n matrices. Since there are mn such matrices, the
dimension of the set of m n matrices is mn.
10. Let K = {(x1 , x2 , x3 ) : x1 x2 + 3x3 = 0}.
a. Find a basis for K.
b. dim(K) =?
c. Extend your basis of K to a basis of R3 .
11. Consider the complex numbers as a real vector space. Remember complex
numbers are things of the form a + bi, where i2 = 1 and a and b are
arbitrary real numbers. Find a basis for the complex numbers.

97

2.5. BASES
12. Let K = {(x1 , x2 , x3 ) : x1 x2 = x3 , x2 + 2x3 = 0)}.
a. Find a basis B for K.
b. dim(K) =?
c. How many vectors have to be added to B to get a basis for R3 ?

13. Let V be the vector space of all n n matrices. Then we know that
dim(V ) = n2 . Find a basis and determine the dimension of each of the
following subspaces of V .
a. Scalar matrices
c. Upper triangular matrices

b. Diagonal matrices
d. Lower triangular matrices

14. Verify that the subset A = {eek : k = 1, . . . , n} is a basis of Rn ; cf. Example 3.

15. Show that the set A = {(1, 0, 0), (1, 1, 1), (0, 1, 1)} spans R3 ; cf. Example 2. Is A a basis of R3 ?
16. Show that the subset {1, t, . . . , tn } is a basis of Pn .
17. In problem 19 of Section 2.4 we defined what it means to say that an
infinite set of vectors is linearly independent. Using that definition, we say
that B is a basis of a vector space V , if B is linearly independentand if S[B]
equals V . Again, the span of a set is all possible finite linear combinations
of vectors from that set. Let V = C[0, 1]. Let B = {1, et , e2t , . . . , ent , . . .}.
If B is a basis for V ?
18. Let V be the set of all polynomials with real coefficients.
{1, t, . . . , tn , . . .}. Is B a basis for V ?

Let B =

19. Let V be a three-dimensional space. Let x be any nonzero vector in V .

x, y 1 , y 2 } is a basis
a. Show that there are vectors y 1 and y 2 such that {x
of V .

20.

21.
22.
23.

b. Show that any linearly independent subset A of V can be extended

to a basis of V .



a b
Let V =
: a + b + c = 0 . Find a basis for V and thus deterc d
mine dim(V ).
1
Let V = {pp : p is in P4 , 0 p (t)dt = 0, p (2) = 0}. Find a basis for V .





0 1 1 0 1 1 1 1
Show that
is a basis for M22 .
1 1 1 1 0 1 1 0





1 1 0 1 0 0 1 0
Is
a basis of M22 ?
0 0 1 0 1 1 0 1

98

CHAPTER 2. VECTOR SPACES

24. Let V = {pp : p is in P4 ,

1
0

p (t)dt = 0 =

2
1

p (t)dt}. Find a basis for V .

25. Let A = {(1, 0, 0), (0, 1, 1), (0, 2, 3), (0, 3, 4)(1, 1, 3)}.
a. Show that A is linearly dependent.
b. What is the largest number of vectors in A that can form a linearly
independent set?
x}]?
c. For which vectors x in A is it true that S[A] = S[A\x

d. Find a subset B of A such that B is linearly independent and for

which S[B] = S[A].

2.6

Coordinates of a Vector

We have seen in an earlier section (Theorem 2.7) that given a basis for a vector
space, every vector can be uniquely written as a linear combination of the basis
vectors. The constants that are used in this sum are called the coordinates of the
vector with respect to the basis. In this section we show how these coordinates
change when we change our bases. For convenience, we again state the definition
of the coordinates of a vector.
Definition 2.11. Let B = {ff 1 , f 2 , . . . , f n } be a basis of a vector space V ,
dim(V ) = n. Then the coordinates of a vector x in V with respect to the
basis B are the constants needed to write x as a linear combination of the basis
vectors. Thus if
x = c1f 1 + c2f 2 + + cnf n

then the coordinates of x with respect to B are the scalars c1 , c2 , . . . , cn . We

write this as
x]B = [c1 , c2 , . . . , cn ]
[x

Note that the ordering of the vectors in B is important. If we change this order,
we change the order in which we list the coordinates.
Example 1. Let V = R3 .
a. Let S = {ee1 , e 2 , e 3 }. Find the coordinates of x = (1, 1, 2) with respect
to S. Clearly,
x = (1, 1, 2) = (1, 0, 0) (0, 1, 0) + 2(0, 0, 1)
Thus,
x ]S = [1, 1, 2]
[x

Note that since S is the standard basis, the coordinates of x with respect
to S are the same three numbers used to define x .
b. Let G = {(0, 0, 1), (0, 1, 0), (1, 0, 0)}. Thus, G is just a reordering of the
standard basis of R3 . Then
x ]G = [(1, 1, 2)]G = [2, 1, 1]
[x

99

2.6. COORDINATES OF A VECTOR

c. Let G = {(2, 1, 3), (0, 1, 1), (1, 1, 0)}. Find the coordinates of x =
(1, 1, 2) with respect to G. We need to find constants c1 , c2 , and c3
such that
(1, 1, 2) = c1 (2, 1, 3) + c2 (0, 1, 1) + c3 (1, 1, 0)
The system of equations derived from this vector equation is
1 = 2c1
+ c3
1 = c1 + c2 c3
2 = 3c1 + c2
The solution is c1 = 1, c2 = 1, c3 = 1. Hence
x]G = [1, 1, 1]
[x

We now discuss the general situation. Let F = {ff 1 , . . . , f n } and G = {gg 1 , . . . , g n }

be any two bases of a vector space V . Then each vector in F can be written
uniquely in terms of the vectors in G and each vector in G can be written
uniquely in terms of the basis F . Let [pjk ], 1 j, k n, be such that
f j = p1j g 1 + p2j g 2 + + pnj g n

(2.8)

Thus [ff j ]G = [p1j , p2j , . . . , pnj ] for 1 j n. Let [qjk ], 1 j, k n, be such

that
g j = q1j f 1 + q2j f 2 + + qnj f n
(2.9)
That is, [gg j ]F = [q1j , q2j , . . . , qnj ].
Thus, the first column of the n n matrix P = [pjk ] consists of the coordinates of f 1 and the jth column of P consists of the coordinates of f j , both
with respect to the basis G. Similarly, the kth column of Q = [qjk ] is formed
using the coordinates of g k with respect to the basis F .
Since P gives the coordinates of the vectors in F with respect to G and Q
gives the coordinates of the vectors in G with respect to F , these two matrices
should be related to each other. In fact, we have
Theorem 2.11. Let F and G be two bases of a vector space V . Let the matrices
P and Q be as defined above. Then.
Q = P 1
Proof. From (2.8) and (2.9) we have
fj =
=

n
X

pkj g k =

i=1

k=1

k=1
n
X

n
X

pkj

i=1

k=1
n
X

qik pkj

n
X

fi

qik f i

100

CHAPTER 2. VECTOR SPACES

Thus the terms

clearly

Pn

k=1 qik pkj

are the coordinates of f j with respect to F . But

f j = 0ff 1 + 0ff 2 + + (1)ff j + + 0ff n

Thus

n
X

for

j = 1, 2, . . . , n

qik pkj = ij

(2.10)

k=1

But this summand is the i, j entry of the matrix QP . Hence we have QP = In .

In a similar manner we can show that P Q = In . Therefore, Q = P 1 .
In the following we refer to P or P 1 as the change of basis matrix. Which one
is meant will be clear from the context.
Example 2. Let F = {(1, 0, 0), (0, 1, 0)(0, 0, 1)}. Let G = {(2, 1, 3), (0, 1, 1),
(1, 1, 0)}. Thus f 1 , f 2 , and f 3 are our standard basis vectors, and g 1 =
(2, 1, 3), g 2 = (0, 1, 1), and g 3 = (1, 1, 0).
We clearly have
[gg 1 ]F = [2, 1, 3] [gg 2 ]F = [0, 1, 1] [gg 3 ]F = [1, 1, 0]
Thus, the matrix Q = P 1 must equal

2 0
1 1
3 1

1
1
0

We next want to find the coordinates of f j with respect to the basis G. There are
two ways to go; one is to actually find the pij such that f 1 = p11g 1 +p21g 2 +p31g 3 ,
etc., or we can just compute Q1 . The first way is instructive; so we will find
the first column of P by computing the coordinates of f 1 with respect to G. We
want to find constants ck , k = 1, 2, 3, such that
(1, 0, 0) = c1 (2, 1, 3) + c2 (0, 1, 1) + c3 (1, 1, 0)
The solutions are c1 = 21 ,c2 = 23 , c3 = 2. Thus (1, 0, 0) = ( 12 )gg 1 +( 23 )gg 2 +2gg 3 .
If we write out the above equations for the cj as a matrix equation we have

or

2 0
1 c1
1
1 1 1 c2 = 0
3 1
0 c3
0

c1
1
Q c2 = 0
c3
0

101

2.6. COORDINATES OF A VECTOR

T
Thus, c1 c2 c3 = Q1 1 0
be the first column of Q1 , and

Q1

T
0 , and clearly the first column of P must

1
2
3
=
2
2

21
3
2

1
2

12

Reading down the columns of Q1 = P we have


 

3
1
g1 +
g 2 + 2gg 3
f1 =
2
2


 
1
3
f2 =
g1 +
g 2 + (1)gg 3
2
2


 
1
1
g1 +
g 2 + (1)gg 3
f3 =
2
2

We next discuss how to use the change of basis matrix P to express the
coordinates of x with respect to G in terms of the coordinates of x with respect
x ]F =
to F . Thus let F = {ff 1 , . . . , f n } and G = {gg 1 , . . . , g n }. Suppose [x
[x1 , . . . , xn ]. Then
x = x1f 1 + x2f 2 + + xnf n

n
n
n
X
X
X
pjng j
pj2g j + + xn
pj1g j + x2
= x1
=

n
X

k=1

xk

j=1

j=1

j=1

n
X
j=1

pjk g j =

n
n
X
X
j=1

pjk xk g j

k=1

Thus,
x ]G =
[x

"

n
X

k=1

p1k xk ,

n
X

p2k xk , . . . ,

k=1

n
X

k=1

pnk xk

(2.11)

Note that the jth coordinate of x with respect to G is the jth row of the matrix
product P ([x1 , . . . , xn ])T . Thus, (2.11) implies the following equation:
x]TG = P [x
x]TF
[x

(2.12)

In the sequel we use equation (2.12) to specify P ; that is, we ask for the change
x]TG = P [x
x]TF . The following summarizes these
of basis matrix P such that [x

102

CHAPTER 2. VECTOR SPACES

relationships:
F = {ff 1 , . . . , f n }
n
X
fj =
pkj g k

G = {gg 1 , . . . , g n }
n
X
gj =
qkj f k

k=1

k=1

x]F = [x1 , . . . , xn ]
[x

yj =

x]G = [y1 , . . . , yn ]
[x
n
X

pjk xk

k=1

x ]TG = P [x
x ]TF
[x

x ]TF = P 1 [x
x]TG
[x

One way of remembering how to compute a change of basis matrix is to

realize that we really want a matrix P such that, given two bases F and G, the
coordinates of any vector x with respect to F and G are related by equation
(2.12):
x ]TF
x ]TG = P [x
(2.12)
[x
Let us start from this equation and see what P must equal. Let F = {ff 1 , . . . , f n }.
If (2.12) holds for all x in V , it must also hold for x = f 1 . What are the coordinates of f 1 with respect to F ? Clearly we have
[ff 1 ]F = [1, 0, . . . , 0]
But then P [ff 1 ]TF is just the first column of P ; that is, the first column of P
equals [ff 1 ]TG . This is, as we would expect, equation (2.8) for j = 1. Similarly
we see that the kth column of P must equal [ff k ]TG .
Example 3. Let V be a vector space. Let F1 , F2 , and F3 each denote a basis
of V . Suppose P and Q denote change of basis matrices between F1 and F2 ,
and between F3 and F2 , respectively, such that
x]TF2
x ]TF1 = P [x
[x

(2.13)

x]TF2
x]TF3 = Q[x
[x

(2.14)

and
Find the change of basis matrix R, relating F1 and F3 , such that
x]TF1
x]TF3 = R[x
[x
x ]TF3 = Q[x
x]TF2 = QP 1 [x
x ]TF1 . By problem 11
From (2.13) and (2.14) we have [x
at the end of this section we must have
R = QP 1

The reader might find this fact useful when computing change of basis matrices,
when neither basis is the standard basis. We illustrate this below.

2.6. COORDINATES OF A VECTOR

103

Example 4. Let V = R2 . Let F = {(7, 8), (9, 20)}. Let G = {(6, 5), (1, 1)}.
Find the change of basis matrix R such that
x ]TG = R[x
x]TF
[x
x ]TS = P [x
x]TF . Let Q be
Let S be the standard basis of R2 . Let P be such that [x
T
T
x ]S = Q[x
x]G . Thus, the columns of P and Q are the coordinates of
such that [x
the vectors in F and G, respectively, with respect to the standard basis. Hence,




7 9
6 1
P =
and Q =
8 20
5 1
and we have
x ]TS
x ]TG = Q1 [x
[x
x ]TF )
= Q1 (P [x
x ]TF
= (Q1 P )[x
Thus,
R=Q




1
1 1 7 9
P =
5
6 8 20
11

 
1
1 29
=
83
75
11


To check our computations, we verify that the first column of R consists of the
coordinates of f 1 with respect to G



  
77 88
1
83
(1, 1) =
= (7, 8) = f 1 .


(6, 5) +
,
11
11
11 11

Problem Set 2.6

1. Find the coordinates of x = (1, 1, 2) with respect to each of the following
bases:
a. {(1, 0, 0), (0, 1, 0), (0, 0, 1)}

b. {(1, 1, 1), (2, 1, 4), (1, 1, 1)}

c. {(2, 0, 6), (4, 2, 0), (0, 3, 2)}

2. Find the coordinates of (1, 2, 0, 4) with respect to each of the following

bases:
a. The standard basis of R4
b. {(0, 1, 1, 1), (1, 0, 1, 1), (1, 1, 0, 1), (1, 1, 1, 0)}

104

CHAPTER 2. VECTOR SPACES

3. Find the coordinates of x = (6, 4) with respect to each of the following

bases of R2 :
a. {(1, 0), (0, 1)}
c. {(1, 1), (1, 0)}

b. {(6, 4), (9, 17)}

d. {(9, 17), (6, 4)}

4. Let F = {(1, 2), (8, 3)}

x ]F = [1, 0], then x =?
a. If [x
x ]F = [1, 1], then x =?
b. If [x
x ]F = [0, 0], then x =?
c. If [x
x]F =?
d. If x = (1, 1), then [x
5. Let V = M22 . Let F be the set:




0 1 1 0 1 1 1
1 1 1 1 0 1 1


1
0

a. Show that F is a basis of V .



6 4
x]F =?
b. Let x =
; [x
3
2

x ]F = [0, 1, 0, 0], then x equals?

c. If [x

6. Let V = P2 . Let F = {1, 1 + t, 1 + t + t2 }.

a. Show that F is a basis of V .
x ]F = [2, 3, 7], then x equals?
b. If [x
c. [6 t2 ]F =
7. Let S be the standard basis of R2 , and let G = {(1, 6), (2, 3)}.
a. Show that G is a basis.
x]TS = P [x
x ]TG .
b. Find the change of basis matrix P such that [x
x]TG = Q[x
x]TS .
c. Find the change of basis matrix Q such that [x
d. Compute P Q and QP .
8. Let x = (8, 4). Let S and G be the bases in problem 7.
x ]S and [x
x ]G .
a. Find [x
x ]TG = P [x
x ]TS .
b. Show that [x
9. Let F = {(1, 7), (2, 3)} and G = {(1, 2), (1, 3)}.
a. Show that both F and G are bases of R2 .
x]TG = P [x
x]TF .
b. Find the change of basis matrix P such that [x
10. Let x = (3, 7). Let F and G be the bases in problem 9.

2.6. COORDINATES OF A VECTOR

105

x]F and [x
x ]G .
a. Find [x
x]TG = P [x
x]TF .
b. Show that [x
x = Bx
x for every
11. Let A and B be the two m n matrices. Suppose that Ax
x in Rn . Show A = B. Note that it will be sufficient to show that if
x = 0 for every x in Rn , then A must be the m n zero matrix. Why?
Ax
12. Let F = {(1, 1), (1, 2)}, G = {(1, 2), (1, 1)}.
x ]TG = P [x
x]TF .
a. Find the change of basis matrix P such that [x
x ]F and [x
x ]G .
b. If x = (1, 1), find [x
x]TF = P 1 [x
x ]TG .
c. Show [x
13. Let S be the standard basis of R3 and let G = {(6, 0, 1), (1, 1, 1), (0, 3, 1)}.
a. Show that G is a basis of R3 .
x ]TG = P [x
x]TS .
b. Find the change of basis matrix P such that [x
c. Wite e k , k = 1, 2, 3, as linear combinations of the vectors in G.
x]G in
14. Let x = (5, 3, 4). Let S and G be the bases in problem 13. Find [x
two different ways.
15. Let F = {ff 1 , f 2 } be a basis of R2 . Let G = {ff 2 , f 1 }. Find the change of
x]TG = P [x
x]TF .
basis matrix P , such that [x
16. Let F be any basis of some n-dimensional vector space. Let G consist of
the same vectors that belong to F , but in perhaps a different order. Show
that the matrix P relating these two bases is a permutation matrix; cf.
problem 13 in Section 1.5.





1 0 0 1 0 0 0 0
17. Let V = M22 . Let F =
. Let G =
0 0 0 0 1 0 0 1





0 1 1 0 1 1 1 1
.
1 1 1 1 0 1 1 0


2 6
x]F =? [x
x ]G =?
a. Let x =
[x
7
5
x ]TG = P [x
x]TF .
b. Find the change of basis matrix P such that [x
x]TG = P [x
x]TF for the vector x of part a.
c. Check that [x
x]TG = [1, 3, 2, 4], then x equals?
d. If [x
18. Let V = P3 . Let F = {2, t + t2 , t2 1, t3 + 1}. Let S denote the standard
basis of V .
a. Verify that F is a basis of V .
x]TF .
x ]TS = P [x
b. Find the change of basis matrix P such that [x

106

CHAPTER 2. VECTOR SPACES

d. Compute the coordinates of t3 , t2 , t, and 1 with respect to the basis
F.

19. Let S = {(1, 0), (0, 1)} and let B = {(2, 3), (1, 6)}.
a. Show that both S and B are bases of R2 .
b. Find the coordinates of (6, 3) with respect to S.
c. Find the coordinates of (6, 3) with respect to B.
20. Let S and B be the same sets as in problem 19. The coordinates of (2, 3)
and (1,6) with respect to S are [2, 3] and [1,6], respectively. Show that
1
2
, 15
],
the coordinates of (1,0) and (0,1) with respect to B are [ 25 , 15 ] and [ 15
respectively. Let
#
"2


1
15
2 1
5
and Q =
P =
2
1
3 6
5

15

a. Identify the columns of P and Q with the above coordinates.

b. Show that P Q = QP = I2
21. Let S and B be the sets defined in problem 19. In that problem, we saw
4
that the coordinates of (6, 3) with respect to B are [ 13
5 , 5 ]. Show
that
" 13 #
 
5
6
=P
4
3
5
22. Find the coordinates of (2,6) with respect to each of the following bases
of R2 :
a. {(1, 0), (0, 1)}

c. {(2, 6), (3, 0)}

b. {(0, 1), (1, 0)}

d. {(3, 0), (2, 6)}

Supplementary Problems
1. Define each of the following terms and then give at least two examples of
each one:
a. A vector space of dimension 5
b. Linearly independent set
c. Basis
d. A vector space of dimension 100 = 102
e. Spanning set
2. Let V be the vector space of all polynomials in t. Show that there is no
finite set of vectors in V that spans V .

107

2.6. COORDINATES OF A VECTOR

3. Let F = {(1, 1, 2), (0, 1, 1), (1, 1, 1)}.
a. Show that F is a basis of R3 .
x ]F .
b. Let x = (1, 0, 0). Find [x

c. Find the change of basis matrix P , relating F to the standard basis,

x ]TF = P [x
x]TS .
for which [x
d. Explain why the first column of P is your answer to part b.
4. Show that any nontrivial subspace W of a finite-dimensional vector space
V has a basis and dim(W ) dim(V ).
5. Let V be the vector space consisting of all real-valued functions defined
on [0,1]. Which of the following subsets are subspaces, and which of these
are finite-dimensional?
a. All polynomials of degree 4

b. All f (t) for which f ( 21 ) = 0

c. All f (t) such that f (0) = f (1)

d. All f (t) such that f (0) + f (1) = 1
e. All functions of the form cert for some constants c and r
6. Let A be an m m matrix. Show that A is invertible if and only if its rows
form a linearly independent set of vectors in Rm . Hint: A is invertible if
and only if A is row equivalent to Im .
7. Suppose X 1 , . . . , X m are linearly independent vectors. Let A = [ajk ] be
an m m matrix. Define
Yj =

m
X

ak X k

j = 1, . . . , m

k=1

Y 1 , . . . , Y m } is linearly independent if and

Show that the set of vectors {Y
only if the matrix A is invertible.
8. If a, b, and c are any three distinct constants, show that sin(a+t), sin(b+t),
and sin(c + t) are linearly dependent on [0,1].



a b
9. Let V =
: ad bc = 1 . Is V a vector space?
c d
10. Let W be that subspace of M22 spanned by




1 1
1 1
and
0 1
0
1
a. Show that dim(W ) = 2 and that W =



a
c



b
: a = d and c = 0 .
d

108

CHAPTER 2. VECTOR SPACES

b. Show that there is no real matrix A such that




1 1 1
1 1
A
A =
0 1
0
1

11. Let x 1 and x 2 be any two vectors. Let y 1 , y 2 , and y 3 be any three vectors
x1 , x 2 ]. Show that these three vectors form a linearly dependent set.
in S[x
x1 , . . . , x r } be any set of vectors. Let F be any set of m vectors
12. Let {x
contained in the span of the given set of r vectors. Show that if m > r,
then F must be a linearly dependent set of vectors.
13. Let V = M22 . Let C = {A : AB = BA for every B in V }. That is, C
consists of all 2 2 matrices that commute with every other 2 2 matrix.
a. Show that C is a subspace of V .
b. Find a basis of C and hence determine dim(C).
c. Let V = Mnn , define C as above, and repeat parts a and b.
14. For which numbers x are the vectors (2,3) and (1, x) linearly independent?