Vector Note
Uploaded by
abelgebrehiwot3Vector Note
Uploaded by
abelgebrehiwot3Linear Algebra
UNIT 3
VECTORS
Scalars: Physical quantities that have only magnitude.
Examples: mass, area, density, volume, temperature, energy.
Vectors: Physical quantities that have both magnitude and direction.
Examples: force, displacement, velocity, acceleration, momentum.
Importance of Vectors:
Vectors form the foundation of mechanics and physics.
They provide geometric motivation for understanding motion, forces, and other physical phenomena.
Both algebraic properties (addition, subtraction, scalar multiplication) and
geometric properties (direction, magnitude, representation by arrows) are essential.
Key Properties of Vectors:
Represented by directed line segments (arrows).
Can be added using the triangle law or parallelogram law.
Multiplication by a scalar changes magnitude but not direction (unless scalar is negative).
Equality of vectors: two vectors are equal if they have the same magnitude and direction,
regardless of position.
3.1 Definition of Points in n-space
A point in n-space is represented by an ordered set of n-real numbers (x1, x2, x3,…,xn).
Each coordinate specifies the position along one of the n-axes
In simple terms:
A number x locates a point on a line.
A pair (x, y) locates a point in the plane.
A triple (x, y, z) locates a point in space.
� In other words: one number for a line, two for a plane, three for space.
1
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
We cannot draw a picture to go further, we can say that a quadruple of numbers (x, y z, w) or (x1, x2, x3,
x4) represent a point in 4-sapce.
Activity 3.1.1: Define a point in n-space, where n is a positive integer.
The set of all points in n-space is represented by n .
For the point X in n , represented by n-tuple of real numbers (x1, x2, …, xn), the
numbers x1, x2;…, xn are called the coordinates of X.
Example 3.1.1 We live in 3-space, where a point is described by three coordinates once an origin and
axes are chosen. Extending to 4-space, a fourth coordinate (such as time) is added. If the
origin of time is set at the birth of Christ, negative time means events before that. If the
origin is the birth of the Earth, negative time means events before Earth’s existence.
If A = (a1, a2, ., an) and B = (b1, b2, …, bn) are points in the same space n , and if k is a real number then
i. A and B are equal (or represent the same point) if a1 = b1, a2 = b2, … and an = bn.
ii. A + B, A – B and kA are defined to the points whose coordinates are (a1 + b1, a2 + b2, …, an +
bn), (a1 - b1, a2 - b2, …,an - bn) and (ka1, ka2, …, kan), respectively.
Example 3.1.2 1) Let A = (1, 2), B = (-3, 4) , then A + B = (-2, 6), A - B = (4,-2), -3A = (-3, -6)
2) Let X = (1, 0, π, 4), Y = (2, 4,-2π,-6), then 2X + Y = (4, 4, 0, 2) and
X - (1/2) Y = (0,-2, 2π, 7).
2
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
3.2. Vectors in n-space, Geometric interpretation in 2 and 3- spaces
Every pair of distinct points A and B in n determines a directed line segment with initial point at A and
terminal point at B. We call such a directed line segment a vector and denote it by AB .
The length of the line segment is the magnitude of the vector. Although AA has zero length, and strictly
speaking, no direction, it is convenient to view it as a vector. It is called a zero or a null vector. It is often
denoted by O .
Two vectors AB and CD will be considered to be equal (or equivalence), AB = CD , if they have the
same magnitude and direction.
Notice that the definition of equality of two vectors does not require that the vectors have the same
initial and terminal points. Rather it suggests that we can move vectors freely provided we make no
change in magnitude and direction.
Activity 3.2.1
Let (a1, a2) be the coordinate representation of A and
let (b1, b2) be that of B. Let P be the point (b1-a1, b2 –a2).
If O is the origin, is OP in the direction of AB ?
Is length of OP equal to the length of AB ?
Is OP = AB ?
If your answer for the above questions is yes, then we can conclude that any vector V= AB in the plane is
a vector OP with initial point at the origin. This is the only vector whose initial point is the origin and
P = B – A, which is equal to AB . Moreover, V = OP is uniquely determined by its terminal point P.
If P = (x, y), then we shall write V = (x, y) and refer to it as the coordinate representation of V relative to
the chosen coordinate system. In view of this, we shall call (x, y) either a point or a vector, depending on
the interpretation which we have in mind. So if V = AB , then we can write V = B – A. In view of this
two vectors AB and CD are equal (or equivalence) if B – A = D – C.
3
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Example 3.2.1: If P = (1, 3), Q = (-1,0), R = (0, -1) and S = (-2, -4) then QP RS
As numbers can be added subtracted and multiplies, vectors can be combined in the following ways.
Let A = (a1, a2) and B = (b1, b2) vectors and t be a real number.
The sum A + B = (a1 + b1, a2 + b2)
The difference A – B = (a1 - b1, a2 – b2)
The scalar multiple tA = (t a1, t a2)
The geometric interpretation of the above vector operations is that A + B is a vector obtained by
placing the initial point of B on the terminal point of A.
If t > 0, then tA is a vector in the direction of A. What about if t < 0? A and tA are said to have
opposite direction.( see figure a and b)
tA
Fig. a for t > 0 Fig. b for t < 0
We can extend the above notions to vectors in n but the geometric interpretations for n > 3 are difficult.
Hence we focus on algebraic aspects of vectors.
If A = (a1, a2, …, an) and B = (b1, b2, …, bn) are vectors in n and if t is any real number, then
A B a1 b1 , a 2 b2 ,.., a n bn
tA = (ta1, ta2, …, tan)
A and B are said to be parallel vectors iff either A = tB for some real number t or B = tA.
In other words, A and B are parallel iff one is a scalar multiple of the other.
Using the arrow notation PQ and RS are parallel if and only if
Q – P = t(S – R) or S – R = t(Q – P) for some real number t.
Activity 3 .2.2: 1) Let A = (6, -2, 4). Find two vectors C and D which are parallel to A.
Are C and D also parallel to each other?
2) Let P = (3, 7), Q = (-4, 2), R = (5, 1), S = (-16, -14).
Is PQ parallel to RS ?
4
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Using the above definitions and applying the associative and commutative properties of real numbers, one
can prove the following theorem.
n
Theorem 3.2.1 Let A, B and C be any members of , and let m and n be any real numbers. Then
a) m(nA) (mn) A Associativity
b) ( A B) C A (B C)
c) A B B A Commutativity
d) (m n) A mA nA
Distributive property
e) m( A B ) mA mB
f) 0.A O
g) A O A
Proof c) A B (a1 , a 2 ,..., a n ) (b 1 , b 2 ,..., b n )
= (a1 b1 , a 2 b 2 ,..., a n b n )
= (b1 a1 , b 2 a 2 ,..., b n a n )
= (b 1 , b 2 , ..., b n ) (a 1 , a 2 ,..., a n )
= BA
Example 3.2.3
A boat captain wants the vessel’s actual velocity
relative to the ground to be due south at 40 knots.
The water current is flowing northwest at 16 knots.
Determine the magnitude and direction of the boat’s
engine velocity (relative to the water) that will result
in the boat traveling due south at 40 knots over the ground.
Solution:
we have u = v + w
where u corresponds to the engine’s vector and
w corresponds to the velocity of the current.
We have u = -40j and w = 8 2 i + 8 2 j
5
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Hence v = u – w = -40j – ( 8 2 i + 8 2 j) = 8 2 i – (40 + 8 2 )j. The magnitude is
(8 2) 2 (40 8 2 ) 2 = 52.5.
40 8 2
The direction is arctan = -1.35 radians
8 2
Exercise 3.2.1:
1. Given three vectors A = (1, 1, 1), B = (-1, 2, 3) and C = (0, 3, 4), find
a. A+B c. A+B – C
b. 2A – B d. A – 3B + 10C
2. Determine whether and can be found to satisfy the vector equations
a. (2, 1, 0) = (-2, 0, 2) + (1, 1, 1)
b. (-3, 1, 2) = (-2, 0, 2) + (1,1,1)
The Distance Formula
The distance formula is derived from the three dimensional version of the Pythagorean theorem, which is
displayed below,
x2 y 2 z 2
x z x2 y 2 z 2
y
The distance between two points (x1, y1, z1) and (x2, y2, z2) is given by:
D (x2 x1 ) 2 ( y2 y1 ) 2 (z2 z1 ) 2
6
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
3.3 Scalar product and norm of vector, orthogonal projection, and direction cosines
Let A = (a1, a2,…,an) and B = (b1, b2, …, bn) be two vectors. The scalar product of A and B is the number
A.B defined by
A.B a1b1 a 2 b 2 ... a n b n
Note: The scalar product is also called a dot product or inner product.
Example 3.3.1
1) Given A = (3, -1) and B = (2, 3), then A.B = 3
2) i.j O , where i = (1,0,0) and j = (0,1,0).
The scalar product satisfies many of the laws that hold for real numbers. The basic ones are:
a) A.B B.A
b) t ( A.B) (tA ).B A.(tB)
c) ( A B).C A.C B.C
d) If A O is the zero vector, then A.A 0 , and otherwise A.A 0
Activity 3.3.1: Proof the above properties.
Example 3.3.2: Given A = (3, 2,-1) and B = (2,0, 3), and C = (1,-1,1), then
a. A.B = B.A = 3
b. 2(A.B) = 6
c. (A + B).C = A.C + B.C = 5
Activity 3.3.2: Find A.A, B.B, and C.C. Are all positive values?
7
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Exercise 3.3.1: Given the vectors A = (1,-1,2), B = (-2,0,2), C = (3,2,1). Evaluate
a. A.C b. B.C c. (A + B).C d. A.(2B + 3C) e. (A.B)C
The length, or norm or magnitude of vector A = (a1, a2, …, an), denoted by ||A||, can be expressed in
terms of the scalar product. By definition
2
A a12 a 22 ... a n2 and A.A (a1 , a 2 ,..., a n ).(a1 , a 2 ..., a n ) a12 a 22 ... a n2
Thus, the norm of the vector A is the number:
||A|| = a12 a 22 ... a n2
Example 3.3.3: If A = (-1, 2, 3) then A 1 4 9 14
Note: 1. A unit vector is a vector having norm or length 1.
2. A 0 A0
1 1 1
Example 3.3.4: If A , , , then A (1 / 3) (1 / 3) (1 / 3) 1
3 3 3
Any non-zero vector can be fully represented by providing its magnitude and a unit vector along its
direction.
A
Let A be a unit vector in the direction of A. Then A
A
Example 3.3.5: Given a vector A = (1, 1, 1). Find a unit vector in the direction of A.
Solution:
A 1 1 1 3 , then the unit vector in the direction of A is:
A (1,1,1) 1 1 1 1
A , , (1,1,1)
A 3 3 3 3 3
Activity 3.3.3: 1. Given three vectors A = (1, 1, 1), B = (-1, 2, 3) and C = (0, 3, 4), find the unit vector in
the direction of A + B – C.
2. The vectors i (1, 0, 0), j = (0, 1, 0) and k = (0, 0, 1) are unit vectors in the direction of positive
x, y and z axis, respectively. Find a unit vector in the direction of A = (-1, 2, 3).
Let A, B be two n-tuples of vectors. We define the distance between A and B to be
A B ( A B ).( A B )
Note: Let A be any vector and k , then A A and kA k A
8
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
The following theorem gives us a geometric interpretation for the scalar product.
Theorem 3.3.1: Let A (a1 , a 2 , a 3 ) and B (b1 , b 2 , b 3 ) be non-zero vectors and let be the angle
between A and B (0 ) .
Then A.B A B cos
Proof: Consider a triangle formed by A, B and A B
2 2 2
A B A B 2 A B cos (Why?)
( a1 b1 ) 2 ( a 2 b2 ) 2 ( a 3 b3 ) 2 a 12 a 22 a 32 b12 b22 b32 2 A B cos
After cancellation, we get, a1b1 a 2 b 2 a 3b 3 A B cos
A.B A B cos
Activity 3.3.4: Given two non-zero vectors A and B, how do you find the angle between them? Take, for
example, A = (2, -1, 2), B = (1, -1, 0) and find the angle between them. Two non-zero
vectors are said to be orthogonal (Perpendicular) if the angle between them is .
2
Note: Two non-zero vectors A and B are said to be orthogonal (Perpendicular) if A.B = 0.
Activity 3.3.5: 1. Which of the following pairs of vectors are perpendicular?
a. (1,-1,1), and (2,1,5)
b. (-5,2,7), and (3,-1,2)
c. (1,-1,1), and (2,3,1)
d. ( ,2,1), and (2,- ,0)
3. Suppose A.B = A.C, what can you deduce about A, B and C?
9
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Exercise 3.3.2:
1. Show that A B A B if and only if A.B=0.
2. Let A1, A2, …, Ar be non-zero vectors such that [Link] = o if i j .
Let c1, c2, …., cr be numbers such that c1A1 + c2A2 +…..+ crAr = 0.
Show that ci = 0 for all i = 1,2, 3, …, r.
The following are two of the important inequalities
Theorem 3.3.2 Let A and B be vectors. Then
a) A.B A B (Schwarz inequality)
b) A B A B (Triangle inequality)
Proof a) If one of A or B is a zero vector then both sides of the inequality are equal to 0.
Suppose both A and B are non-zero.
From A.B A B cos ,
A .B A B | cos | which implies A.B A B (Why?)
2
b) From the fact that V V.V ,
2
AB ( A B) . ( A B)
A.A 2 A.B B.B
2 2
A 2 A.B B
2 2
A 2 A.B B (why ?)
2 2
A 2 A B B (why ?)
A B
2
By taking the square roots, AB A B
10
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Remark: The inequalities of Theorem 3.3.2 hold true also for any vectors A and B in Rn.
We now define the component of a vector in the direction of another vector.
Let A and B be two non-zero vectors.
Let be the angle between them.
From the terminal point of B drop a perpendicular to the
line containing A.
The vector OD has magnitude OD B cos , 0 , and its direction is either the same as that
of A or opposite to it depending on whether is acute or obtuse.
1
Since A is a unit vector in the direction of A and since OD has magnitude B cos and is in the
A
direction of A or opposite to A, we can write
1
OD B cos A
A
A.B
or OD A (why?)
A2
This vector is called the projection of B along A and is denoted by Proj A B .
A.B
That is, Pr oj AB 2 A
A
A.B
We call t 2
the component of B along A.
A
Note: 1. Pr oj AB is parallel to A. That is Pr oj AB tA , for some t
2. B - Pr oj AB is orthogonal (perpendicular) to A.
11
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
1
Example 3.3.6: Let A = (3, -1, -2) and B = 2, 3, , then
2
Proj A B = A.B2 A 6 3 1 A 12 , 4 , 8
9 1 4 7 7 7
A
Proj B A = A .B 6 3 1 B 64 , 96 , 16
B
2 1 53 53 53
B 49
4
Activity 3.3.4: One application of projections of vector arises in the definition of the work done by a force
on a moving body. Find another application.
Given a non-zero vector u (u1 , u 2 , u 3 ) in 3 . The direction cosines of the vector u are:
u1 u u
Cos , Cos 2 , Cos 3
u u u
Where the direction angles, , , and are the angles that the vector makes with the positive x, y, and z-
axes respectively.
12
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Remark: Cos Cos Cos 1
2 2 2
Example 3.3.7: Let u = (1, -2, 3). Find the direction cosines of u.
Solution: Since u u.u (1,2,3).(1,2,3) 1 4 9 14
Thus the direction cosines are: Cos u1 1 , Cos u 2 2 , and Cos u 3 3
u 14 u 14 u 14
Activity 3.3.5: Is Cos 2 Cos 2 Cos 2 1 ?
3.4 Cross Product of vector
The second type of product of two vectors is the cross product. Unlike the dot product, the cross product
of two vectors is a vector.
Definition 3.4.1 The cross product (or vector product) A x B of two vectors
A = (a1, a2, a3) and B = (b1, b2, b3) is given by the determinant:
� � �
AxB = 1
� �2 �3 , where i, j, k are the unit vectors along the x, y, z-axes.
�1 �2 �3
Evaluating the determinant:
A x B = (a2 b3 – a3 b2)i - (a3 b1 – a1 b3)j + (a1 b2 - a2 b1)k
Or equivalently, as a vector:
A x B = (a2 b3 – a3 b2, a3 b1 – a1 b3, a1 b2 - a2 b1)
3
Note that the cross product is defined in
Example 3.4.1 Let A = (2, -1, 3) and B = (-1, -2, 4)
A x B = (-4 + 6, -3 – 8, -4 –1) = (2, -11, -5)
Activity 3.4.1: Find B x A.
The following are some of the basic properties of cross product
Theorem 3.4.1 For vectors A, B and C,
1) A x B = -(B x A)
2) A x A = O
3) tA x B = t(A x B) = A x (tB), t
13
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
B A.B
2 2 2 2
4) AxB A
5) C. (A x B) = B. (C x A) = A. (B x C)
6) (A + B) x C = (A x C) + (B x C)
7) C x (A + B) = (C x A) + (C x B)
8) A . (A x B) = 0 and B. (A x B) = 0 (that is, AxB is perpendicular to both A and B.)
9) (AxB) x C = (A. C) B – (B.C)A
Proof :The following is the proof for 1, 2 and 8. The rest are left as an exercise
1) From the definition of cross product,
A x B = (a2 b3 – a3 b2, a3 b1 – a1 b3, a1 b2 – a2 b1)
For B x A, interchange A and B to obtain
B x A = (b2 a3 – b3 a2, b3 a1 - b1 a3, b1 a2 - b2 a1)
= (a2 b3 – a3 b2, a3 b1 - a1 b3, a1 b2 - a2 b1)
= - (A x B)
2) A x A = (a2 a3 – a3 a2, a3 a1 - a1 a3, a1 a2 - a2 a1)
= (0, 0, 0)
8) Setting C = A in 5) yields
A . (A x B) = B . (A x A)
= B.0 (why?)
=0
By setting C = B in 5),
B .(A x B) = A . (B x B)
= A.0=0
This shows that for non zero vectors A and B, the cross product A x B is orthogonal to both A and B.
Activity 3.4.2: Are the usual commutative and associative laws valid?
i.e. for any vectors A, B and C in 3 , is A x B = B x A?
Is A x (B x C) = (A x B) x C?
Exercise 3.4.1:
Let A = (2,1,0), B = (2,-1,1) and C = (0,1,1). Find
14
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
a. AxB b. (AxB)xC c. (A.C)B – (B.C)A d. BxC e. Ax(BxC)
From 4) of theorem 3.4.1, we derive an important formula for the norm of the cross product.
B A.B
2 2 2 2
AxB A
2 2 2 2
A B A B cos 2 ( is the angle between A and B)
A
2 2
B 1 cos 2
2 2
A B sin 2
AxB A B sin (For 0 , sin is non- negative)
Activity 3.4.3:
- For the unit vectors i, j and k , find i j , j k and k i . What is j i ?
- If A and B are parallel, what is A B?
- If A and B are orthogonal, What is A B ?
Exercise 3.4.2:
1. Find a unit vector perpendicular to both A = (2,-3,1) and B = (1,2,-4).
2. Prove that (A – B)x(A + B) = 2(AxB).
Application on area and volume
Let u and v be vectors and consider the parallelogram that the two vectors make. Then
u v = Area of the Parallelogram
The direction of uxv is a right angle to the parallelogram that follows the right hand rule.
To find the volume of the parallelepiped spanned by three vectors u, v, and w, we find the triple product:
u.(vxw) = Volume
This can be found by computing the determinant of the three vectors:
15
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
u1 v1 w1
u2 v2 w2 = u1 (v 2 w3 v3 w2 ) v1 (u 2 w3 u 3 w2 ) w1 (u 2 v3 u 3 v 2 )
u3 v3 w3
Example 3.4.2:
1. Find the area of the parallelogram which is formed by the two vectors u = (1, 3, 2) and v = (-2, 1, 3).
2. Find the volume of the parallelepiped spanned by the vectors
u = (3, -2, -1), v = (1, 3, 2), and w = (-2, 1, 3).
Solution:
1. The area of the parallelogram is given by: u v (1, 3 , 2 ) ( 2 ,1, 3 ) ( 7 , 7 , 7 ) 147
2. The volume of the parallelepiped spanned by the three vectors is:
u .( vxw ) ( 3, 2 , 1).(1,3, 2 ) ( 2 ,1,3 ) ( 3, 2 , 1).( 7 , 7 , 7 ) 28
Exercise:
1. Find the area of the triangle having vertices at
u = (3, -2, -1), v = (1, 3, 2), and w = (-2, 1, 3).
2. If the area of an equilateral triangle with adjacent sides u and v is 50 3 cm2, then
find u.v
16
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
4.5 Lines and planes
Activity 4.5.1: Given two vectors A and B with B 0, consider vectors of the form A + tB, where t varies
over all real numbers. The vector A + 2B, for example is as shown in the figure.
1 3
Show the vectors A + tB for t = 0, 1, 3, , -1, -2, , what kind of figure is generated by
2 2
these vectors? i.e. what is the collection of all points which are terminal points of A + tB?
Definition 3.5.1: A line is any set of the form {p : p A tB, t } where B is assumed to be
a non-zero vector and A is a fixed point on the line.
Note that if (x, y, z) is on line and if A = (a1, a2, a3) and B = (b1, b2, b3 then
(x, y, z) = (a1, a2, a3) + t(b1, b2, b3) for some real number t.
Activity 3.5.2: Is point A on ? Is B parallel to
a vector formed by any two
points of ?
P = A + tB is a vector equation of a line through A
Example 3.5.1 Find equation of a line through P1 = (0, 1, 2) and P2 = (-1, 1, 1).
Soln: We need a point A on the line and a vector B parallel to the vector formed by two point of the line.
Take A = P1 and B = P2 – P1. Then
A + t B = (0, 1, 2) + t (-1, 0, -1)
17
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
(x, y, z) = (0, 1, 2) + t (-1, 0, -1) is equation of the line. By giving distinct values for t we
will obtain distinct points on the line. Find some of the points.
Note: The equation of a line passing through points A and B is given by:
P = A + t(B – A) or P = (1 – t)A + B, t
Exercise 3.5.1: Let the line L1 passes through the points (5,1,7) and (6,0,8) and the line L2 passes through
the points (3,1,3) and (1,3, ) . Find the value of for which the two lines intersect.
Suppose P = (x, y, z) is a point on line through A = (a1, a2 a3) in the direction of B = (b1, b2, b3).
Then p = A + tB (x, y, z) = (a1, a2, a3) + t(b1, b2, b3) or equivalently
x = a1 + b1t
y = a2 + b2t
z = a3 + b3t
These equations are parametric equation of a line and t is called a parameter.
Activity 3.5.3: 1) Find the parametric equation of a line that contains (2, -1, 1) and is parallel to the vector
1
(3, , 0).
2
2) From the parametric equation of a line in 3 , derive the equation
y a1 y a2 z a3
b1 b2 b3
It is called standard form of equation of a line.
If the line is on a plane show that the standard form reduces to an
equation of the form y = mx + c.
Two lines l1 and l2 given by A1 + tB1 and A2 + tB2 are said to be parallel if B1 and B2 are parallel.
That is the vectors P1 – Q1 and P2 – Q2 are parallel for any two points P1, Q1 of l1 and P2, Q2 of l2.
Let l1 be a line through A in the direction of B ( B 0) . Consider the distance between l1 and the origin.
This distance is the minimum of the lengths of all vectors with initial point the origin and terminal point on
2
l1. That is, minimum of A tB for any real number t. Now put f (t ) A tB
2 2
This is a quadratic function whose graph opens upward: f (t ) A 2t ( A.B ) t 2 B
18
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
2 A.B A.B
So it has minimum at: t 2
2
2B B
Let Po be a point and N be a non zero
vector. We define the plane passing
through Po to perpendicular to N to be
the collection of all points P such that
the vector p o p is perpendicular N.
According to our definition, if P is any point on the plane through P0 and perpendicular to
N, then N . p o p 0 or N . ( p p0 ) 0
Activity 3.5.4: Starting from the equation N . p o p 0 , show that equation of a plane
through point Po = (xo, yo, zo) perpendicular to N = (a, b, c) is
ax + by + cz = d where d = axo + byo + czo.
This equation can be written as N . P d . The vector N is said to be normal to the plane.
Hence a plane is any set of the form {P: N.P = d}. Where N is a given non-zero vector
and d is a given number.
Example 3.5.2: Find an Equation of the plane that contains point (-2, 4, 5) and that is
normal to (7, 0, -6).
Solution: The equation of the plane is given by 7(x+2)+0(y-4)-6(z-5) = 0 or 7x- 6z = -44
Activity 3.5.5: Does this plane intersect the y-axis?
Two planes in 3 spaces are said to be parallel if their normal vectors are parallel. They
are said to be perpendicular if their normal vectors are perpendicular.
The angle between two planes is defined to be the angle between their normal vectors.
Activity 3.5.6:
1. A plane passes through (-1, 2, 3) and is perpendicular to the y-axis. What is
the equation?
19
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
2. Consider the planes x + 2y - 3z = 2 and 15x - 9y – z = 2. Are they parallel or
perpendicular planes? Or neither parallel nor perpendicular?
Exercise 3.5.2: Find the equation of the plane passing through the three points
P1 = (2,1,1), P2 = (3,-1,1), P3 = (4,1,-1).
Let Q be a point outside a plane normal
to N. We define the distance from point
Q to the plane as follows. Let Po be the
point of intersection of the line through
Q, in the direction of N, and the plane
through P. The distance d from Q to the
plane is the distance between Q and Po.
20
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Now we find a formula for this distance. Clearly d Pr oj QP QP
= Pr oj QP
N
[Link]
However, Pr ojN QP N
N2
[Link] [Link] [Link]
Hence d N N =
N2 N
2
N
Therefore the distance d of a point Q from a plane through P which is normal to N is given by:
[Link]
d
N
Exercises 3.5.3
1. Let A = (0, 1, 5) ,B = 14 , 5, 1 . Find the angle between A and B.
2. Which of he following vectors are parallel or perpendicular to (1, 1, -1)?
a) (2, 2, -2) d) (1, 0, 1)
1 1 1
b) (2, -2, 0) e) , ,
2 2 2
c) (-2, 2, 2) f) (, , )
3. a) Find all vectors that are orthogonal to E1 = (1, 0, 0)
b) Find all vectors that are orthogonal to both E1 and E3 = (0, 0, 1)
c) Find all vectors that are orthogonal to E1, E2 and E3 = (0, 0, 1)
4. Find a non-zero vector orthogonal to (1, 2, -1)
5. Find a unit vector in the direction of (3, -1, 2, 4)
1 1 1 1 1 2 1 1
6. Let U1 , , , U 2 , , , U3 , ,0
3 3 3 6 6 6 2 2
a) Show that each u1, u2, u3 is orthogonal to the other two and that each is a unit vector
b) Find the projection of E1 on each of u1, u2, u3
c) Find the projection of A = (a1, a2, a3) on u1.
7. In the following cases compute (A x B).C
a) A = (1, 2, 0) B = (-3, 1, 0), C = (4, 9, -3)
b) A = (-3, 1, -2) B = (2, 0, 4), C = (1, 1, 1)
8. Prove that two non-zero vectors A and B are perpendicular if and only if A A tB for every
number t.
9. If A + B + C = 0. Show that A x B = B x C = C x A
21
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
10. Find a formula for the area of a parallelogram whose vertices, in order, are P, Q, R & S.
2 2
11. Show that A B A B A B
12. Find parametric equations of lines through
a) (-5, -6, 8) and (1, 3, 7) b. (10, 3, 1) and (6, -2, -3)
13. Find equation of a plane through points (0, 1, 0), (0, -1, -1) and (1, 2, 1)
14. Find a point of intersection of the lines {p:p = (1, -5, 2) + t(-1, 1, 0)} and
{p:p = (3, -3, 1) + t(4, 0, -1)}
15. Prove that the line {p:p = (1, 3, -1) + t(0, 3, 5)} lies entirely in the plane
{(x, y, z): 2x – 5y + 3z = -16}
16. Find the intersection point of the lines {p:p = (-2, 4, 6) + t(1, 2, 3)} and {p:p = (-2, 1, 6) + t(3, 2,
1)}. Find equation of the plane containing the two lines.
17. Find all points of intersection of the line {p:p = t(1, -3, 6)} and the plane {p:x + 3y + z = 2}
18. Prove that if B.N = 0 and A is on the plane {p:(p – po).N = 0} then the entire line {p:p = A
+ tB} lies in the plane.
19. Find a line through the point po = (-5, 2, 1) and normal to the plane {(x, y, z): x = y}
20. Find a line through (xo, yo, zo) and normal to the plane {(x, y, z): ax + by + cz = d}
21. Let be a line given by A + tB and let P be a point on different from A. For point Q not on ,
B x PQ
show that the distance d of Q form the line is given by: d
B
22. Let be the line x = 1 + 2t, y = -1 + 3t, z = -5 + 7t. Find the two points on at a distance 3 units
from the plane 2(x-1) + 2(y+3) –z = 0
23. The set of all points equidistant from (0, 1, 5) and 5, -1, 3) is a plane. Find the equation.
UNIT 4
VECTOR SPACES
22
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
4.1 The axioms of a vector space
Definition 4.1.1: Let K be a set of numbers. We shall say that K is field if it satisfies the following
conditions:
a) If x, y are elements of K then x + y and xy are also elements of K.
b) If x is an element of K, then –x is also an element of K. Furthermore, if x 0, then x-1
is also an element of K.
c) 0 and 1 are elements of K.
Example 4.1.1: The set of all real numbers and the set of all complex numbers ℂ are fields.
Activity 4.1.1: Are ℤ (The set of all integers) and Q (the set of all rational numbers fields?
Remark: The essential thing about a field is that its elements can be added and multiplied and the results are
also elements of the field. Moreover, every element can be divided by a non-zero element.
Definition 4.1.2: A vector space V over a field K is a set of objects which can be added and can be
multiplied by elements of K. It satisfies the following properties.
V1) For any u, v V and a K, we have
u+vV and au V
V2) For any u, v, w V,
(u + v) + w = u + (v + w)
V3) There is an element of V, denoted by O (called the zero element), such that
0 + u = u + 0 = u for all elements u of V.
V4) For u V, there exists –u V such that
u + (-u) = 0
V5) For u, v V, we have
u+v = v+u
V6) For u, v V and a k,
a(u + v) = au + av
V7) For u V and a, b k, (a + b) u = au + bu and (ab) u = a(bu)
V8) For u v,
1u = u
Activity 4.1.2:What is the name given for each of the above properties?
23
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Other properties of a vector space can be deduced from the above eight properties. For example, the property
0u = O can be proved as :
0u + u = 0u + 1.u (by V8)
= (0 + 1) u (by V7)
= 1. u
=u
By adding –u to both sides of ou + u = u, we have 0u = O
Examples of different models of a vector space
1) Consider sets 2 and
Clearly, for V = 2 and k = , properties V1 and V2 hold.
The element 0 of 2 that satisfied v3 is 0 = (0, 0).
The other 5 properties can be easily verified. Hence 2 is a vector space over.
2) Let V = 2 and K = ℂ
For any u, v 2, we have u + v 2.
But for a ℂ , au is not always in 2.
For example for a = 3i and u = (1, -2), au = (3i, -2i) 2 .
Hence 2 is not vector space over ℂ .
Thus when dealing with vector spaces, we shall always specify the field over which we take the
vector space.
3) Let F be the set of all functions form to , for any f and g in F, f + g is a function from to
defined by (f + g) (x) = f(x) + g(x).
For a , af = af(x) is in F.
The zero element O of F is the zero function f(x) = 0 for all x .
By verifying the other properties, we can see that F is a vector space over .
The algebraic properties of elements of an arbitrary vector space are very similar to those of elements of 2,
3, or n. Consequently, we call elements of a vector space as vectors
Activity 4.2.1:Which of the following are vector spaces?
a) C on 2
b) Cn over ℂ
c) Qn over Q
24
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
d) n over ℂ
4.3 Subspaces, Linear Combinations and generators
Definition4.3.1: Suppose V is a vector space over k and W is a subset of V. If, under the addition and scalar
multiplication that is defined on V, W is also a vector space then we call W a subspace of V.
Using this definition and the axioms of a vector space, we can easily prove the following:
A subset W of a vector space V is called a subspace of V if:
i) W is closed under addition. That is, if u, w W, then u + w W
ii) W is closed under scalar multiplication. That is, if uW and a k, then auW.
iii) W contains the additive identity 0.
Then as W V, properties V1 – V8 are satisfied for the elements of W.
Hence W itself is a vector space over k. We call W a subspace of V.
Example 4.3.1: Consider H = {(x, y): x, y and x + 4y = 0} .
H is a subset of the vector space 2 over . To show that H is a subspace of V, it is enough to show the
above three properties hold in H.
Let u = (x1, y1) and w = (x2, y2) be in H. Then x1 + 4y1 = 0 and x2 + 4y2 = 0
u + w = (x1 + x2, y1 + y2) and (x1 + x2) + 4(y1 + y2) = x1 + 4y + x2 + 4y2 = 0 + 0 = 0
Which shows u w H .
For a , au ax1 , ax 2 and ( ax1 ) 4( ax 2 ) a ( x1 4 x 2 ) = 0 a = 0.
Hence au H. Now, the element O of 2 is (0, 0). 0 + 4(0) = 0. Hence O = (0, 0) is in H.
H is a subspace of 2
Activity 4.3.1:Take any vector A in 3. Let W be the set of all vectors B in 3 where B.A = 0. Discuss
whether W is a subspace of 3 or not.
Definition 4.3.2: Let v1, v2, …, vn be elements of a vector space V over k. Let x1, x2, …, xn be
elements of k. Then an expression of the form x1v1 + x2v2 +… + xn vn is called a linear combination of v1,
v2, …, vn..
25
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Example 4.3.2: The sum 2(3, 1) + 4(-1, 2) +(1, 0) is a linear combination of (3, 1), (-1, 2) and (1, 0). As this
sum is equal to (3, 10), we say that (3, 10) is a linear combination of the three ordered pairs.
Activity 4.3.2:
i) Take two elements v1 and v2 of 3. Let W be the set of all linear combinations of v1 & v2.
Show that W is a subspace of 3 . W is called the subspace generated by v1 and v2
ii) Generalize i) by showing that a set w generated by elements v1, v2, …, vn of a
vector space V is a subspace.
4.4. Linear dependence and independence of vectors
Definition 4.4.1: Let V be a vector space over k. Elements v1, v2, …, vn of V are said to be linearly
independent if and only if the following condition is satisfied:
whenever a1, a2, …, an are in k such that a1v1 + a2v2 + … + anvn = 0, then ai = 0 for all
i = 1, 2, …, n.
If the above condition does not hold, the vectors are called linearly dependent. In other words v1, v2,…, vn
are linearly dependent if and only if there are numbers a1, a2, …, an where a1v1 + a2v2 + … + anvn = 0 for at
least one non-zero ai.
Example 4.4.1: Consider v1 = (1, -1,1) , v2 = (2, 0, -1) and v3 = (2, -2, 2)
i) a1v1 + a2v2 = a1 (1, -1, 1) + a2 (2, 0, -1) = (a1 + 2a2, -a1, a1 – a2)
a1v1 + a2v2 = 0 a1 + 2a2 = 0, -a1 = 0 and a1 – a2 = 0
a1 = 0 and a2 = 0
Hence v1 & v2 are linearly independent.
ii) a1v1 + a2v3 = a1 (1, -1, 1) + a2 (2, -2, 2)
= (a1 + 2a2, -a1 – 2a2 , a1 +2 a2)
a1v1 + a2v3 = 0 a1 + 2a2 = 0, -a1 – 2a2 = 0 and a1 +2 a2 = 0
a1 = -2a2
Take a1 = 2 and a2 = -1, we get 2(1, -1, 1) + (-1) (2, -2, 2) = 0.
As the constants are not all equal to zero, v1 and v3 are linearly dependent.
26
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Activity 4.4.1: Show that v1, v2 and v3 are also linearly dependent.
Remark: If vectors are linearly dependent, at least one of them can be written as a linear
combination of the others.
Activity 4.4.2: Show that (1, 0, 0, …,0), (0, 1,0,…)…, (0,0,0, …, 1) are linearly independent vectors in n.
4.5 Bases and dimension of a vector space
Definition 4.5.1: If elements e1, e2, …, en of a vector space V are linearly independent and generate V, then
the set B = {e1, e2, …, en} is called a basis of V. we shall also say that the elements e1,
e2,…, en constitute or form a basis of V.
Example 4.5.1:
1) Show that e1 = (0, -1) and e2 = (2, 1) form a basis of 2.
Solution: we have to show that
i) e1 and e2 are linearly independent
ii) They generate 2 i.e every element (x,y) of 2 can be written as a linear combination of
e1and e2.
i) a1 e1 + a2 e2 = O a1(0, -1) + a2(2,1) = (0, 0)
2a2 = 0 and –a1 + a2 = 0
a2 = 0 and a1 = 0
e1 and e2 are linearly independent
ii) (x, y) = a1e2 + a2 e2 (x, y) = (0, -a1) + (2a2, a2)
x = 2a2 and y = -a1 + a2
x
a2 = and a1 = a2 – y … (*)
2
x 2y
=
2
Therefore, given any (x, y), we can find a1 and a2 given by (*) and (x, y) can be written as a linear
combination of e1 and e2 as
(x, y) = x 2 y (0, 1) x (2, 1)
2 2
4 6 4
For example, (4, 3) = (0,1) (2,1)
2 2
27
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Or (4, 3) = -(0, -1) + 2(2, 1)
Note that {(1, 0), (0, 1)} is also a basis of 2. Hence a vector space can have two or more basis. Find other
bases of 2.
2) Show that e1 = (2, 1, 0) and e2 = (1, 1, 0) form a basis of 3.
Solution: e1 = (2, 1, 0) and e2 = (1, 1, 0) are linearly independent but they do not generate 3. There are
no numbers a1 and a2 for which
(3, 4, 2) = a1 (2, 1, 0) + a2 (1, 1, 0).
Hence {(2, 1, 0), (1,1,0)} is not a basis of 3.
The vectors E1 = (1, 0, 0) , E2 = (0, 1, 0), E3 = (0, 0, 1) are linearly independent and every element (x, y, z) of
3 can be written as
(x, y, z) = x(1, 0, 0) + y(0, 1, 0) + z (0, 0, 1)
= xE1 + yE2 + zE3
Hence {E1, E2, E3} is a basis of 3.
Note that the set of elements E1 = (1, 0, 0,…,0), E2 = (0, 1, 0, … 0),…,En = (0, 0, 0, …,1) is a basis of n. It is
called a standard basis.
Let B = {e1 , e2, …, en} be a basis of V. since B generates V, any u in V can be represented as u = a1e1 + a2
e2 + … + an en. Since the ei are linearly independent, such a representation is unique. We call (a1, a2, …, an)
the coordinate vector of u with respect to the basis B, and we call ai the i – th coordinate.
Example 4.5.2
1) In 1) of example 3.3.1 The coordinate vector of (4,3) with respect to the basis
{(0, -1), (2,1)} is (-1, 2). But with respect to the standard basis it is (4, 3).
Find coordinates of (4,3) in some other basis of 2.
2) Consider the set V of all polynomial functions f: which are of degree less than or equal to 2.
Every element of V has the form f(x) = bx2 + cx + d, where b, c, d
V is a vector space over (show).
Clearly, e1 = x2, e2 = x and e3 = 1 are in V and a1e1 + a2 e2 + a3e3 = O
(0 is the zero function)
a1x2 + a2e2 + a3 e3 = 0 for all x
28
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
a1 = a2 = a3 = 0.
Which shows e1, e2 and e3 are linearly independent
bx2 + cx + d = a1e1 + a2e2 + a3e3 for all x
bx2 + cx + d = a1x2 + a2x + a3
b = a1, c = a2 and d = a3
Thus e1, e2 and e3 generate V.
{x2, x1 1} is a basis of V and the coordinate vector of an element
f(x) = bx2 + cx + d is (b, c, d)
The coordinate vector of x2 – 3x + 5 is (1, -3, 5)
Activity 4.5.1:Show that the polynomials
E1 = (x – 1)2 = x2 – 2x + 1
E2 = x – 1
and E3 = 1
form a basis of a vector space V defined in 2) of example 2.5.2. What is the coordinate of
f(x) = 2x2 – 5x + 6 with respect to the basis {E1, E2, E3}?
E = {(1, 0, 0), (0,1,0), (0,0,1)} and B = {(-1,1,0), (-2, 0, 2), (1, 1, 1)} are bases of 3 and each has three
elements. Can you find a basis of 3 having two elements? four elements?
The main result of this section is that any two bases of a vector space have the same number of elements. To
prove this, we use the following theorem.
Theorem 4.5.1: Let V be a vector space over the field K. Let {v1, v2,…,vn} be a basis of V. If w1, w2,…,wm
are elements of V, where m > n, then w1, w2, …, wm are linearly dependent.
Proof (reading assignment)
Theorem 4.5.2: Let V be a vector space and suppose that one basis B has n elements, and another basis W
has m elements. Them m = n.
Proof: As B is a basis, m > n is impossible. Otherwise by theorem 3.4.1, W will be a linearly dependent set.
Which contradicts the fact that W is a basis. Similarly, as W is a basis, n > m is also impossible.
Hence n = m.
29
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Definition 4.5.2: Let V be a vector space having a basis consisting of n elements. We shall say that n is the
dimension of V. It is denoted by dim V.
Remarks : 1. If V = {0}, then V doesn’t have a basis, and we shall say that dim v is
zero.
2. The zero vector space or a vector space which has a basis consisting of
a finite number of elements, is called finite dimensional. Other vector
spaces are called infinite dimensional.
Example 4.5.3:
1) 3 over has dimension 3. In general n over has dimension n.
2) over has dimension 1. In fact, {1} is a basis of , because
a .1 0 a 0 and
any number x has a unique expression x x.1 .
Definition 4.5.3: The set of elements {v1, v2, …,vn}of a vector space V is said to be a maximal set of
linearly independent elements if v1, v2, …,vn are linearly independent and if given any
element w of V, the elements w,v1, v2, …, vn are linearly dependent.
Example 4.5.4: In 3 {(1, 0, 0), (0, 1, 1), (0, 2, 1)} is a maximal set of linearly independent elements.
We now give criteria which allow us to tell when elements of a vector space constitute a basis.
Theorem 4.5.3: Let V be a vector space and {v1, v2, …,vn}be a maximal set of linearly independent elements
of V. Then {v1, v2, …,vn}is a basis of V.
Proof: It suffices to show that v1, v2, …,vn generate V. (Why?)
Let w v .
Then w, v1, v2, …,vn are linearly dependent (why?).
Hence there exist numbers ao, a1, a2, …, an not all 0 such that
ao w + a1v1 + a2v2 + …+ anvn = O
In particular a o 0 (why?
Therefore, by solving for w,
a1 a an
w V1 2 V2 ... Vn
ao ao ao
This proves that w is a linear combination of v1, v2, …,vn.
30
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Theorem 4.5.4: Let dim V = n, and let v1, v2, …,vn be linearly independent elements of v. Then
{v1, v2, …,vn} is a basis of v.
Proof: According to theorem 3.4.1, {v1, v2, …,vn} is a maximum set of linearly independent elements of V.
Hence it is a basis by theorem 2.5.3
Corollary 4.5.1: Let W be a subspace of V. If dim W = dimV, then V = W
Proof: Exercise
4.6. Direct sum and direct product of subspaces
Let V be a vector space over the field K. Let U, W be subspaces of V. We define the sum of U and W to be
the subset of V consisting of all sums u + w with u U and w W . We denote this sum by U +W and it is a
subspace of V. Indeed, if u1 , u 2 U and w1 , w2 W then
(u1 w1 ) (u 2 w2 ) u1 u 2 w1 w2 U W
If c K , then
c(u1 w1 ) cu1 cw1 U W
Finally, 0 0 U W . This proves that U + W is a subspace.
Definition 4.6.1: A vector space V is a direct sum of U and W if for every element v in V there exist
unique elements u U and w W such that v u w .
Theorem 4.6.1: Let V be a vector space over the field K, and let U, W be subspaces. If U + W = V, and
if U W 0 , then V is the direct sum of U and W.
Proof: Exercise
Note: When V is the direct sum of subspaces U, W we write:
V U W
Theorem 4.6.2: Let V be a finite dimensional vector space over the field K. Let W be a subspace. Then
there exists a subspace U such that V is the direct sum of W and U.
Proof: Exercise
Theorem 4.6.3: If V is a finite dimensional vector space over the field K, and is the direct sum of subspaces
U, W then
dim V = dim U + dim W
31
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Proof: Exercise
Remark: We can also define V as a direct sum of more than two subspaces. Let W1, W2, …., Wr be
subspaces of V. We shall say that V is their direct sum if every element of can be expressed in a unique way
as a sum
v w1 w2 ....... wr
With wi in Wi.
Suppose now that U, W are arbitrarily vector spaces over the field K(i.e. not necessarily subspaces of some
vector space). We let UXW be the set of all pairs (u, w) whose first component is an element u of U and
whose second component is an element w of W. We define the addition of such pairs component wise,
namely, if (u1 , w1 ) UXW and (u 2 , w2 ) UXW we define
(u1 , w1 ) (u 2 , w2 ) u1 u 2 , w1 w2 )
If c K , we define the product c(u1 , w1 ) by
c (u1 , w1 ) = (cu1 , cw1 )
It is then immediately verified that UXW is a vector space, called the direct product of U and W.
Note: If n is a positive integer, written as a sum of two positive integers, n r s , then we see that K n is
the direct product K r XK s and dim(UXW) dimU dimW .
Example 4.6.1: Let, V R , U (0,0, x3 ), x3 , and W ( x1 , x 2 ,0), x1 , x 2 . Show
3
that V is the direct sum of W and U.
Solution: Since V, U and W are vector spaces, and in addition to that U and W are subspaces of V.
The sum of U and W is:
U + W = ( x1 , x 2 , x 3 ), x1 , x 2 , x 3 R V
3
Thus; V = U + W
The intersection of U and W is: U W 0
Therefore, V is the direct sum of W and U.
Activity 4.6.1: 1. Let, V R , U ( x1 ,0, x 3 ), x1 , x 3 , and W (0, x 2 ,0), x 2 . Show that
3
V is the direct sum of W and U.
2. Let, V R , U ( x1 , x 2 ,0), x1 , x 2 , and W ( 0,0, x 3 ), x 3 . Show that V
3
is the direct sum of W and U.
32
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Exercise:
2 ab ab ab ba
1. Let, V R , U ( , ), a , b , and W ( , ), a , b .
2 2 2 2
Show whether V is the direct sum of W and U or not.
2. Let, V R 3 , U ( x1 , x 2 ,0), x1 , x 2 , and W (0, x 2 , x3 ), x 2 , x3 Show whether V is
the direct sum of W and U or not.
Exercise 4.1
1. Let k be the set of all numbers which can be written in the form a b 2 , where a, b are rational
numbers. Show that k is a field.
2. Show that the following sets form subspaces
a. The set of all (x, y) in 2 such that x = y
b. The set of all (x, y) in 2 such that x – y = 0
c. The set of all (x, y, z) in 3 such that x + y = 3z
d. The set of all (x, y, z) in 3 such that x = y and z = 2y
3. If U and W are subspaces of a vector space V, show that U W and U W are subspaces.
4. Decide whether the following vectors are linearly independent or not (on )
a) (, 0) and (0, 1)
b) (-1, 1, 0) and (0, 1, 2)
c) (0, 1, 1), (0, 2, 1), and (1, 5, 3)
5. Find the coordinates of X with respect to the vectors A, B and C
a. X = (1, 0, 0), A = (1, 1, 1), B = (-1, 1, 0), C = (1, 0, -1)
b. X = (1, 1, 1) , A = (0, 1, -1), B = (1, 1, 0), C = (1, 0, 2)
6. Prove: The vectors (a, b) and (c, d) in the plane are linearly dependent if and only if ad – bc = 0
7. Find a basis and the dimension of the subspace of 4 generated by
{(1, -4, -2, 1), (1, -3, -1, 2), (3, -8, -2, 7)}.
8. Let W be the space generated by the polynomials x3 + 3x2 – x + 4, and
2x3 + x2 – 7x – 7. Find a basis and the dimension of W.
9. Let V = {(a, b, c, d) 4: b – 2c + d = 0}
W = {(a, b, c, d) 4: a = d, b = 2c}
33
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Find a basis and dimension of
a) V b) W c) V W
10. What is the dimension of the space of 2 x 2 matrices? Give a basis for this space. Answer the
same question for the space of n x m matrices.
11. Find the dimensions of the following
a) The space of n x n matrices all of whose elements are 0 except possibly the diagonal elements.
b) The space of n x n upper triangular matrices
c) The space of n x n symmetric matrices
d) The space of n x n diagonal matrices
12. Let V be a subspace of 3. What are the possible dimensions for V? Show that if
V 3, then either V = {0}, or V is a straight line passing through the origin, or V is a plane passing
through the origin.
34
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
LINEAR TRANSFORMATIONS
5.1. Definition of linear Transformations and examples
A linear transformation is a special type of function. Hence before the discussion about linear maps, it is
helpful to revise the concept of functions.
Activity 5.1.1: Recall about the meaning and properties of a function. Also try to recall about related
concepts like domain, range, one to one, on to, composition, and inverse.
Recall that a function (mapping) consists of the following:
i) a set X, each of whose element is mapped
ii) a set Y, to which each element of x is mapped
iii) a rule (correspondence) f, which associates with each element x of X a single element f (x) of Y.
A function (mapping) f from X to Y is written as f : X Y.
If f maps an element a of X in to element b of Y then we write f (a) = b. Here b is called the image of a
under f and a is called pre-image of b. Moreover, for any subset A of X, the set of all images of elements of
A is denoted by f [A] and we call it the image of A under f. On the other hand if B is a subset of Y the set of
all elements of X whose images are in B is denoted by f -1[B] and we call it the pre-image of B under f. Using
set notation the image of A and the pre-image of B under f can be defined respectively as
f [A] = {bY | b = f (a) for some a A} and f -1[B] = {aX | f (a) B}
Note: For any function f: X Y,
1. Domain of f = X and range of f = f[X] the image of set X under f.
2. f is one-to-one (injective) if and only if for any x1, x2 X, f (x1) = f (x2) implies x1 = x2.
3. f is on to (surjective) if and only if for every b in Y there exists an element a in X such that f
(a) = b, i.e f [X] = Y or range of f = Y.
4. We say that f is invertible (has an inverse) if and only if there is a function g from Y into X
such that
i) gof is the identity function on X, i.e (gof) (x) =x for all x X.
ii) fog is the identity function on Y i.e (fog) (y) = y for all y Y.
35
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
This function g:Y X is unique whenever it exists and we call it the inverse of f. It is denoted
by f -1. Moreover from (i) and (ii) it follows that f(x) = y if and only if f 1 (y ) x . Observe that
if there is a function
g:Y X satisfying (i) then f is one-to-one. Similarly if there is a function g:Y X satisfying
(ii) then f is on to. What do you conclude from this?
Further one can easily verify that if f:Y X is one-to one and on to then there exists a unique
function g:Y X which satisfies (i) and (ii). Thus f is invertible if and only if it is one-to-one
and on to.
Activity 5.1.2:
1. Let A = [2, ) and B = [-4, ). Define a function
f : A B by f (x) = x2 - 4x.
a) show that f is one-to-one
b) Show that f is on to
c) Find the inverse of f
2. Let g : 2 2 be given by g(a, b) = (a + b, b)
i) Is g one-to-one? Verify!
ii) Is g onto? Verify!
iii) Does g have inverse? If so find its inverse.
As it is already mentioned a linear transformation is a special type of function. What is/are the condition(s)
that a function must satisfy so as to be a linear transformation?
The following definition will enable us to give a complete answer for this question.
Definition 5.1.1: Let V and W be vector spaces over the same field K. A function
T: V W is called a linear transformation (or a linear mapping) of V in to W if it
satisfies the following conditions:
i) T (u + v) = T(u) + T(v) u,v V
ii) T(u) = T(u) K and u V
Note: 1. Using condition (ii) of the definition, one can show that T(Ov) = Ow
where Ov and Ow are zero vectors in V and W respectively.
T (Ov) = T(0.u) ( because 0.u = Ov for any u V)
36
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
= 0T(u) (by (ii),0 K (The zero element in the field K))
= OW (Why?)
This proves that a linear mapping maps a zero vector in to zero vector.
2) The two conditions in the definition are equivalent to
T(u + v) = T(u) + T(v) , K and u, vV
Let us prove that a function T from vector space V in to W over the same field K is a linear transformation
iff T(1v1 + 2v2) = 1T(v1) + 2T(v2) for any 1, 2 K and for any v1, v2 V.
Proof: First suppose T: V W is a linear transformation.
Since 1v1, 2v2 V 1, 2 K and v1, v2 V,
T(1v1 + 2v2) = T(1v1) + T(2v2) by condition (i) of definition 4.1.1
= T(v1) + 2 T(v2) by condition (ii) of definition 4.1.1
Thus T(1v1 + 2v2) = 1T(v1) + 2T(v2) 1, 2 K and v1, v2 V.
Next suppose that
T(1v1 + 2v2) = 1T(v1) + 2 T(v2) 1, 2 K and v1, v2 V *
We have to prove that T is a linear transformation.
T(v1 + v2) = T(1.v1) + 1.v2) , where 1 is the multiplicative identity in K.
= 1.T(v1) + 1.T(v2) From (*)
= T(v1) + T(v2).
T(1 v1) = T(1v1 + 0.v2) , where 0 is the zero element of K.
= 1T(v1) + 0.T(v2) , from (*)
= 1T(v1) + Ow, (0.T(v2) = Ow – zero vector in W)
= 1 T(v1)
Therefore T(v1 + v2) = T(v1) + T(v2)and T(1v1) = 1 T(v1)
1 k, v1, v2 V.
Consequently T: V W is a linear transformation if and only if
T(1v1 + 2v2) = 1T(v1) + 2T(v2) 1, 2 K and v1, v2 V.
Observe that by induction we also have a more general relation
T(1v1 + 2v2 + … + nvn) = 1T(v1) + 2T(v2) + …+T(vn) or
n n
T i v i i T(v i ) for any n vectors v1, v2, …, vn in V and for any n
i 1 i 1
scalars 1, 2, 3, …, n in K whenever T:VW is a linear transformation.
37
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
3) A linear mapping T: V V is often called a linear operator on V.
Now let us see some examples of a linear transformation (or a linear mapping).
Example 5.1.1 Let V be a vector space over the field K. Then the mapping
I: V V given by I(v) = v vV is a linear transformation. To prove this let u,v V and
K. Then u + v V and uV as V is a vector space. Since I(x) = x x V, we have
i) I (u + v) = u + v and I(u) + I(v) = u + v
Thus I (u + v) = I(u) + I(v)
ii) I (u) = u and I(u) = u
Thus I (u) = I(u)
Therefore I is a linear transformation. We call I identity transformation.
Example 5.1.2: Let T be a mapping from a vector space V over a field K into it self
given by T(v) = Ov v V where Ov is a zero vector in V.
Then T is a linear transformation. (Verify!). We call this linear transformation the zero
transformation.
Example 5.1.3: Let V be the vector space of all differentiable real valued functions
of real variables on an open interval (a, b). Then the mapping D: V V
given by D(f ) f ' (where f ' is the derivative of f) is a linear
transformation. This can be easily verified by using the properties of
derivative.
Example 5.1.4: Let T : Mn Mn( where Mn is the vector space of n x n matrices
over ) be given by T(A) = At . Show that T is a linear transformation.
To show this let A, B Mn and .Then using the properties of
transpose of matrices we have,
(A + B)t = At + Bt and (A)t = (At) = At
Thus T(A + B) = (A + B)t by definition of T
= At + Bt
= T(A) + T(B)
Also T(A) = (A)t = At = T(A)
38
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Therefore T is a linear transformation.
Example 5.1.5: Let a function L: 23 be given by L (x, y) = (x + y, y, x)
Is L a linear transformation?
Solution: Let (a, b) , (c, d) 2 and .
Then (i) T((a, b) + (c, d)) = T(a + c, b + d)
= (a + c + b + d, b + d, a + c)
= (a + b, b, a) + (c + d, d, c))
= T(a, b) + T(c, d)
ii) T((a, b)) = T(a, b)
= (a + b, b, a)
= (a + b, b, a)
= T(a, b)
Thus T is a linear transformation.
Example 5.1.6: Is the mapping N: 3 defined by N(x, y, z) = ( x , y, z )
a linear transformation?
Solution: No, take (0, 1, 2), (3, 0, 1) 3
N((0, 1,2) + (3, 0, 1)) = N(3, 1, 3)
= (3,1,3) 32 12 32
= 19
N(0, 1, 2) + N(3, 0, 1) = (0,1,2) 3, 0, 1
= 0 2 12 2 2 32 0 2 12
= 5 10
Since 19 5 10 , N ((0,1,2) + (3, 0, 1)) N (0,1,2) + (3,0,1)
Thus N is not a linear transformation as N(u + v) = N(u) + N(v) must hold true
for all vectors u, v in 3.
Remark: To show that a mapping T from a vector space V in to W over the same
field K is not a linear transformation it suffices to show that there exists two
39
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
vectors v1, v2 V such that T(v1 + v2) T(v1) + T(v2) or there exists scalar V
such that T(v ) T( v ) .
Activity 5.1.3:
1. Show that the mapping T : 32 defined by T(x, y, z) = (x-y, x – z) is a linear transformation.
2. Is the mapping L: 32 defined by L (a, b, c) = (| a |, 0) a linear transformation? Justify your answer!
Let us add one more example. Recall that vectors v1, v2, v3, … vm in a vector space V over a field K are
linearly independent iff 1 v1 + 2v2 +… + nvn = Ov (where 1,2, …, n K) implies 1 = 2 = ….
= n = 0
Example 5.1.7: Let T be a linear transformation from a vector space V in to W over the same field K. Prove
that the vectors v1, v2, v3, …, vn V are linearly independent if T(v1), T(v2), T(v3), …, T(vn)
are linearly independent vectors in W.
Solution: Suppose T(v1), T(v2), T(v3), …, T(vn) are linearly independent vectors in W where v1, v2, v3, …, vn
are vectors in V and T:V W is a linear transformation. To prove that v1, v2, v3,…,vn are
linearly independent.
Let 1 2, 3, … ,n K such that
1v1 + 2v2 + 3v3 + …+ nvn = Ov
Then T(1v1 + 2v2 + 3v3 + … + nvn) = T(Ov)
So 1T(v1) + 2T(v2) + 3T(v3) + … + nT(vn) = Ow
But T(v1), T(v2), T(v3), …, T(vn) are linearly independent.
Hence 1 = 2 = 3 = … = n = 0. Thus we have shown that
1v1 + 2v2 + 3v3 + … + nvn = Ov implies
1 = 2 = 3 = … = n = 0.
Consequently v1, v2, v3, …, vn are linearly independent.
Exercise
1. Determine whether or not each of the following mappings is linear transformation.
a) T: 22 given by T (x, y) = (x + y, x)
b) T: 2 given by T (x, y) = xy
c) T: 32 given by T (x1, x2, x3) = (1 + x1, x2)
40
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
d) L: 32 given by L (x, y, z) = (z + x, y)
e) L: 22 given by L (p, q) = (p3, q3)
2. Let M2 denote the vector space of 2 2 matrices over .
a b a b 2c
Let T: M2 M2 be given by T
c d 3a c d
Is Ta linear mapping? Justify your answer!
3. Let V be the vector space of m n matrices over . Let P be a fixed mm matrix and
Q a fixed n n matrix over . Show that the mapping L: V V defined by
T(A) = PAQ is a linear transformation.
4. Show that the mapping F: 2 defined by F(a, b) = |a – b| is not a linear transformation.
5. Let G : V W be a linear transformation where V and W are vector spaces over the same field K.
Prove that if u1, u2, u3, …, un are linearly dependent vectors in V then G(u1), G(u2), G(u3), …, G(un) are
linearly dependent vectors in W.
6. Let U, V, and W be vector spaces over the same field K.
If g: U V and f : V W are linear transformations show that fog is also a linear
transformation form U in to W.
7) i) Let A = (a, b, c) be a fixed given vector in 3. Define T: 3 by
T(X) = AX X 3.
(AX is the scalar (dot) product of A and X). Show that T is a linear transformation.
ii) Let A be as in (i) define T: 3 by T(X) = AX + 4. Show that T is not
a linear transformation.
8) Let V be the space of n x 1 matrices over and let W be the space of m 1 matrices over . Let A be
a fixed m n matrix over . Define T: V W by T(X) = AX, X V.
Prove that i) T is a linear transformation.
ii) T is a zero transformation if and only if A is the zero matrix.
9) Let V be the vector space of all n n matrices over and let B a fixed n n matrix.
If T: V V is defined by T(A) = AB – BA then verify that T is a linear transformation.
10) Let V be a vector space over , and f: V , g: V be two linear transformations. Let F: V
2 be the mapping defined by F(v) = (f(v), g(v)). Show that F is a linear transformation.
11) Let V, W be two vector spaces over the same field K and let F: V W be a linear transformation. Let
U be the subset of V consisting of all elements u such that F(u) = Ow. Prove that U is a subspace of V.
12) Let F: 3 4 be a linear transformation. Let P be a point of 3 and A is a non zero element of 3.
Describe F[S], where S = {X 3| X = P + t A, t }.
41
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
(Distinguish the cases when F(A) = 0 and F(A) 0).
Properties of Linear transformation
Recall that in the previous section, we have seen that if T: V W is a linear transformation from vector
space V in to W over the same field K then
i) T(Ov) = Ow
n
ii) T i v i
n
i 1
T( v
i 1
i
) i K and vi V, i = 1, 2, …, n.
We now state two other basic properties in the following theorem. The proof is left for you as an exercise.
Theorem 5.1.1: Let T be a linear transformation from a vector space V in to W over the
same field K. Then i) T(-v) = -T(v) v V
ii) T(v1 – v2) = T(v1) – T(v2) v1, v2 V
Our next theorem asserts that a linear transformation from a given finite dimensional vector space V in to any
vector space W is completely determined by its values on the elements of a given basis of V.
Theorem 5.1.2: Let V and W be vector spaces over the field K. Let {v1, v2, v3, …, vn}be
a basis of V. If {w1, w2, w3, …, wn}is a set of arbitrary vectors in W, then
there exists a unique linear transformation F: V W such that
F(vj) = wj for j = 1, 2, … n.
To prove the theorem, we need to
a) define a function F from V into W such that F(vi) = wi for all i = 1, 2, 3, … n.
b) show F is a linear transformation
c) show that F is unique.
Proof: Let V and W be vector spaces over the field K. Let {v1, v2, v3, …, vn} be a basis
of V and {w1, w2, w3, …, wn} be any set of n-vectors in W.
Since {v1, v2, v3, …, vn} is a basis of V, for any v V there exist unique scalars
a1, a2, a3, …, an K such that
n
v a
i 1
i
vi
a) Define F : V W by F(v) = a1w1 + a2 w2 + a3w3 + … + anwn
n n
i.e F( v) a i w i where v a
i 1
i
vi
i 1
42
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Every element v of V is mapped to only one element of W as the scalars ai ‘s are unique. As W is a
n
vector space, every linear combination of vectors in W is also in W. So aiwi W . Thus F is a
i 1
function from V in to W
Moreover, since vi = 0.v1 + 0.v2 + … + Ovi-1 + [Link] + [Link]+1 + …+ [Link]
for any i = 1, 2, 3, …, n, we have
F(vi) = 0.w1 + 0.w2 + … + 0wi-1 + [Link] + [Link]+1 + … + [Link] = wi
So F(vi) = wi for each i = 1, 2, 3, …, n
b) We show that F is a linear transformation,
i.e F(u + v) = F(u) + F(v) and F(v) = F(v) K and u, v V.
To do this, let x, y V and K
n n
Then x x i v i and y yv i i for some unique scalars x1, x2, x3, …,xn, y1, y2, y3, …, yn in
i 1 i 1
K.
n n
(i) F(x + y) = F x i v i yi vi
i 1 i 1
n
= F ( x i yi ) vi
i 1
n
= (x i yi ) w i by definition of F
i 1
n n
= xi wi yi w i
i 1 i 1
= F xi vi
n F y i v i
n
by definition of F
i 1 i 1
= F(x) + F(y)
F(x + y) = F(x) + F(y) x, y V
n
ii) F(x) = F x i v i
i 1
n
= F (x i ) v i
i 1
n
= (x i ) w i by definition of F
i 1
43
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
n
= xiwi
i 1
n
= F x i v i by definition of F
i 1
= F (x)
So F (x) = F(x) K and x V
Therefore from (i) and (ii) we conclude that F is a linear transformation from V in to W.
c) In (a) and (b) we have shown the existence of a linear transformation F : V W such that F(vi) =
wi for all i = 1, 2, 3, …, n.
To prove that F is unique, suppose that G: V W is a linear transformation such that G(vi) = wi
i {1, 2, …,n}
n
Let x be any vector in V then x xv
i 1
i i for some unique scalars x1, x2, x3, …, xn in K.
n
Thus G ( x ) G x i v i
i 1
n
= x i G ( v i ) as G is a linear transformation
i 1
n
= xiwi as G(vi) = wi for each i = 1, 2, 3, ..., n.
i 1
n
= F x i v i by definition of F.
i 1
= F(x)
Since G(x) = F(x) for any x V, we conclude that G = F.
This proves that F is unique. With this we complete the proof of the theorem.
Remark: 1) The vectors w1, w2,w3,…, wn in theorem 4.2.2 are completely arbitrary;
they may be linearly dependent, independent or they may even be equal to
each other. But the number of these vectors in W must be equal with that of
the number of basis vectors of V.
2) In determining the linear transformation from V in to W the assumption that
{v1, v2, …, vn} is a basis of V is essential.
44
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Example 5.1.8:
a) Is there a linear transformation T from 2 in to 2 such that T(2, 3) = (4, 5) and T(1, 0) = (0, 0)?
b) How many linear transformations satisfying the given conditions do we have?
Solution: a) Yes. The two vectors (2, 3) and (1, 0) are linearly independent and hence they form a basis for
2. Thus according to theorem 4.2.2, there is unique linear transformation from 2 in to 2
such that T(2,3) = (4, 5) and T(1, 0) = (0, 0).
b) As it is verified above, we have only one linear transformation satisfying the given conditions.
Example 5.1.9: Describe explicitly the linear transformation T: 2 2 such that
T(2, 3) = (4, 5) and T(1, 0) = (0, 0).
Solution: Let (x, y) 2. Then (x, y) can be expressed as a linear combination of
(2, 3) and (1, 0)as {(2,3), (1,0)} form a basis of 2.
So (x, y) = a(2, 3) + b(1, 0) for some a, b or (x, y) = (2a + b, 3a)
y 2y
Hence we have x = 2a + b and y = 3a, which inturn implies a and b x
3 3
y 2y
Thus (x, y) = (2, 3) x 1, 0
3 3
y 2
T ( x, y ) T (2,3) x y (1, 0)
3 3
y 2
= T (2,3) x y T (1, 0)
3 3
y 2
= (4,5) x y (0, 0)
3 3
4 5y
= y , (0, 0)
3 3
4 5y
Therefore T ( x, y ) y, (x, y) 2
3 3
Observe that the image of any vector (a, b) 2 under the linear transformation of
4 5
example 4.2.2 is b , . So the image of 2 under T is the line through (0, 0) with
3
3
4 5
direction vector , .
3 3
Activity 5.1.4:
45
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
4 5
Let T: 2 2 be as in example 4.2.2 above, i.e T(a, b ) b, b
3 3
i) Let A = {(x, y)| x2 + y2 = 1}. Find the image of A under T i.e T[A].
ii) Describe the set containing all elements in 2 whose images is (0,0)
Example 5.1.10: Find a linear transformation T : 2 3 such that
T(1, 2) = (3, -1, 5) and T(0, 1) = (2, 1, -1)
Solution: = {(1, 2), (0, 1)} is a basis of 2. (verify)
Thus there exist a unique linear transformation (by theorem 4.2.2) T: 2 3
such that T(1,2) = (3, -1, 5) and T(0, 1) = (2, 1, -1).
To describe this unique linear transformation explicitly, suppose (x, y) 2
Then (x, y) = t1 (1, 2) + t2 (0, 1) for some t1, t2 as is a basis of 2.
= (t1, 2t1 + t3)
x t 1
Thus we have
y 2t 1 t 2
Which in turn implies t1 = x and t2 = y – 2x. So (x, y) = x(1, 2) + (y – 2x) (0, 1)
and T(x, y) = x T(1, 2) + (y - 2x) T(0, 1)
= x(3, -1, 5) + (y – 2x) (2, 1, -1)
= (3x, -x, 5x) + (2y – 4x, y-2x, 2x – y)
= (2y – x, y – 3x, 7x – y)
Therefore the required linear transformation is given by
T(x, y) = (2y – x, y – 3x, -y + 7x).
Activity 5.1.5:
1 a) Find a linear transformation T: 22 such that T(1, 0) = (1, 1) and
T(0, 1) = (-1, 2)
b) Prove that T maps the square with vertices (0, 0), (1, 0), (1, 1) and
(0, 1) in to a parallelogram.
2. a) Is there a linear transformation T:33 such that
T(0, 1, 2) = (3, 1, 2) and T(1, 1, 1) = (2, 2, 2)?
b) If your answer in (a) is yes,
(i) find T.
(ii) is it unique? Why?
46
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Example 5.1.11: Is there a linear transformation L from 2 in to 2 such that
L(1, -1, 1) = (1, 0) and L(1, 1, 1) = (0, 1)?
Solution: Even though (1, -1, 1) = (1, 1,1) are linearly independent vectors in 3,
They do not from a basis of 3 . This is because a basis of 3 must contain three linearly
independent vectors as 3 a 3-dimensional vector space. So theorem 4.2.2 does not guarantee
us the existence of a linear transformation L : 3 2 such that L(1, -1, 1) = (1, 0) and
L(1, 1, 1) = (0, 1)? But to make use of the theorem we need to have a basis of 3 containing
the two given vectors v1 = (1, -1,1) and v2 = (1, 1, 1) and a vector v3 different from v1 and v2.
That is we need to find a vector v3 so that {v1, v2, v3} is a basis of 3. For this put v3 = (0, 0, 1).
Since (1, -1, 1) + (1, 1, 1) + (0, 0, 1) = (0,0,0) for some ,, implies = = = 0, (1,
-1, 1), (1,1,1) and (0, 0, 1) are three linearly independent vectors in 3. In addition to this let
us take one more vector in 2, say (0, 0). Now we have a basis {(1, -1, 1), (1, 1, 1), (0, 0, 1)}
of 3 and three vectors (1, 0), (0, 1), (0,0) in 2. Thus by theorem 4.2.2 there exists a unique
linear transformation L: 3 2 such that
L(1, -1, 1) = (1, 0), L(1, 1, 1) = (0, 1) and L(0, 0, 1) = (0, 0).
To find this linear transformation, first let us write any (x, y, z) 3 as a linear combination
of (1, -1, 1), (1,1,1) and (0, 0, 1). This is possible as they form basis of 3.
Set (x, y, z) = a(1, -1, 1) + b(1, 1,1) + c(0, 0, 1); a, b, c and solve for a, b, c interms of x, y,
and z.
(x, y, z) = (a, -a, a) + (b, b, b) + (0, 0, c)
(x, y, z) = (a + b, -a + b, a + b + c)
x ab
y a b
z a b c
xy xy
b , a , c z x
2 2
( x y) ( x y)
Thus (x, y, z) = (1,-1,1) + (1,1,1) + (z - x) (0,0,1)
2 2
( x y) ( x y)
L(x, y, z) = L(1,1,1) L(1,1,1) ( z x) L(0,0,1)
2 2
x y ( x y)
= (1, 0) (0, 1) ( z x) (0, 0)
2 2
47
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
x y x y
= ,
2 2
Therefore we have obtained a linear transformation L: 3 2 given by
x y x y
L(x, y, z) = ,
2 2
1 (1) 1 (1)
Moreover L(1,-1,1) = , = (1,0) and
2 2
1 1 1 1)
L(1, 1,1) = , (0, 1) .
2 2
Consequently we can say that there is a linear transformation L: 3 2 satisfying the two given
conditions L(1, -1, 1) = (1, 0) and L(1,1,1) = (0, 1).
Note: The linear transformation T: V W whose existence and uniqueness is guaranteed by theorem 4.2.2
depends on the given basis vectors of V and the given vectors of W whose number equals the number of basis
vectors in V.
x y x y
Is L: 3 2 given by L(x, y, z) = , the only linear mapping that can satisfy the
2 2
requirements of the question in the above example? Replace (0, 0, 1) by (1, 0, 0) in the solution of the above
example and find a linear transformation
L: 3 2 such that L(1, -1, 1) = (1, 0) and L(1, 1, 1) = (0, 1). Do the same by replacing (0, 0) by (1, 1) and
(0, 0,1) by (1, 0, -1).
Exercise:
1. Find a linear transformation
a) T: 2 2 such that T(1, 0) = (2, -1) and T(0,1) = (-2, 1)
b) T: 2 2 such that T(1, 2) = (3, 0) and T(2, 1) = (1, 2)
c) L: 2 3 such that L(2, -5) = (-1, 2, 3) and L(3, 4) = (0, 1, 5)
d) L: 3 2 such that L(3, 1,1) = (1, 0), L(2, -1, 5) = (0, 1)
and L(4, 0, -3) = (-1, 1)
e) F: 2 3 such that T(2, 1) = (3, -1, 5) and T(1, 0) = (2, 1, -1)
f) G: 3 2 such that G(2, 0, 1) = 1, 1) and G(0, 3, 0) = (-1, 1)
2. If v1 = (1, -1) w1 = (1, 0)
48
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
v2 = (2, -1) w2 = (0, 1)
v3 = (-3, 2) w3 = (1, 1)
Is there a linear transformation G from 2 into 2 such that G(vi) = wi
for i = 1, 2, 3 ? If so find G.
3. Let T: V W be a linear mapping . Let v be an element of V. Show that
T(-v) = -T(v)
4. Let L: 2 2 be a linear mapping, such that
L(3, 1) = (1, 2) and L(-1, 0) = (1, 1) compute L(1, 0)
5.2. The Rank and Nullity of a linear transformation and examples
In this section we will discuss in detail about two important sets related to a linear transformation T: V W
where V and W are vector spaces over the same field K. One of them is a subset of V and the other is a subset
of W. Is there an element v in V such that T(v) = Ow -the zero vector in W? We know that T(Ov) = Ow, so
there is at least one element v in V such that T(v) = Ow. Hence we have a non-empty subset
U = {u V| T(u) = Ow} of V. It is a subspace of V because
i) Ov U as T(Ov) = Ow
ii) If u1, u2 U then T(u1) = T(u2) = Ow
So T(u1 + u2) = T(u1) + T(u2) as T is a linear transformation
= Ow + Ow
= Ow
Thus u1 + u2 U for any u1, u2 U
iii) if u U and K, then T(u) = Ow
T(u) = T(u) as T is a linear transformation
= .Ow
= Ow
So u U for any K and for any u U
From (i), (ii) and (iii) it follows that U is a subspace of V.
On the other hand we know that, the range of T is a subset of W. It is also a subspace of W(verify!).
This section is concerned with these two special subspaces. Let us begin with the following definition.
Definition 5.2.1: Let V, W be vector spaces over K and T: V W be a linear
transformation.
i) The set of elements v V such that T(v) = Ow is called the Kernel of T or the null space of T.
49
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
ii) The set of elements w in W such that there exists an element v of V such that T(v) = w is
called the image or the range of T.
Notation: We shall denote the Kernel and image of a linear transformation
T: V W by KerT and ImT respectively. So KerT = {u V| T(u) = Ow}
and ImT = {w W| w = T(v) for some v V}
At the beginning of this section we have proved that kerT is a subspace of V. In the theorem below we prove
that ImT is a subspace of W.
Theorem 5.2.1: Let T be a linear transformation from a vector space V in to W over the some field K. Then
(a) Ker T is a subspace of V
(b) ImT is a subspace of W.
Proof :a) It is already proved
b) Clearly ImT is a subset of W
i) Since T(Ov) = Ow , Ow ImT.
ii) Let w1, w2 ImT. Then there exists u1, u2 V such that
T(u1) = w1 and T(u2) = w2. Since V is a vector space, u1 + u2 V. Moreover
T(u1 + u2) = T(u1) + T(u2) = w1 + w2. Thus w1 + w2 ImT as there exists a
vector v V such that T(v) = w1 + w2 (v = u1 + u2).
So we have w1 + w2 ImT w1, w2 W.
iii) Let K and w ImT. Then there exists v V such that T(v) = w
as w ImT. Since V is a vector space over K, v V. Moreover
T(v) = T(v) = w. Hence w ImT. From (i), (ii) and (iii) it
follows that ImT is a subspace of W.
Example 5.2.1: Let G : be defined by G ( x, y, z ) ( x y, y z )
3 2
a) Show that G is a linear transformation
b) Find the ker G and ImG.
Solution: a) Let ( x, y, z ), (u , v, w) 3 and
Then G (( x, y, z ) (u , v, w)) G ( x u , y v, z w)
= ( x u y v, y v z w)
= ( x y, y z ) (u v, v w)
= G ( x, y, z ) G (u , v, w)
50
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
and G ( ( x, y, z )) G (x, y, z )
= (x y, y z )
= ( x y, y z )
= G ( x, y , z )
Therefore G is a linear transformation.
b) ker G ( p, q, r ) 3 G ( p, q, r ) (0,0)
= ( p, q, r ) 3 ( p q, q r ) (0,0)
= ( p, q, r ) 3 p q 0 and q r 0
= ( p, q, r ) 3 pqr
= ( p, p, p ) P
= p (1,1,1) p
So kerG is the subspace of generated by (1,-1,-1). List at least four elements of ker G.
3
Im G (d , e) 2 G ( x, y , z ) (d , e) for some ( x, y , z ) 3
= (d , e) 2
( x y, y z ) (d , e) for some ( x, y, z ) 3
= (d , e) 2 x y d and y z e for some x, y, z
= ( x y, y z ) x, y, z
= ( x y,0) (0, y z ) x, y, z
= ( x y ) (1, 0) ( y z ) (0, 1) x, y , z
= t1 (1,0) t2 (0, 1) t1 x y and t2 y z
Thus ImG is the subspace of generated by (1, 0) and (0, 1).
2
That is Im G
2
(why?) observe that dim (ker G) = 1 and dim (ImG) = 2 .
In the following theorem we state an equivalent condition that help us to determine whether a given linear
transformation T is one-to-one (injective) or not, using kerT.
Theorem 5.2.2: Let T: V W be a linear transformation.
Then T is one-to-one if and only if ker T = {Ov}.
Proof: Suppose T is one-to-one. We need to show that kerT contains only the zero
51
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
vector. Let x kerT. Then T(x) = Ow.
But T(Ov) = Ow. So we have T(x) = T(Ov), which implies x = Ov as T is
one-to-one. Therefore kerT = {Ov}
Conversely suppose kerT = {Ov}. We wish to show that T is one-to-one.
Now let T(x1) = T(x2); x1, x2 V. Then T(x1) + -T(x2) = Ow or
T(x1) + T(-x2) = Ow. Thus we have T(x1 + -x2) = Ow, which implies
x1 + (-x2) ker T. Since ker T contains only the zero vector by assumption,
x1 + -x2 = Ov and hence x1 = x2. Consequently T is one-to-one.
Let us illustrate how theorem 4.3.2 can be used to check whether a given linear transformation is one-to-one
or not.
Example 5.2.2: Let T: 2 3 be a linear transformation given by
T(a, b) = (a + b, a – b, b).
i) Find ker T and ImT.
ii) Is T one-to-one? Why?
Solution: (i) Ker T = { (x, y) 2 | T (x, y) = (0,0,0)}
= {(x, y) 2 | T (x +y, x-y, y) = (0, 0, 0)}
= {(x, y) 2 | x + y = 0, x – y = 0 and y = 0}
= {(x, y) 2 | x = 0 and y = 0} = {(0, 0)}
From the definition of ImT, it follows that
(u , v, w) Im T (u , v, w) T (a, b) for some (a, b) 2
(u , v, w) (a b, a b, b) for some a, b
u a b, v a b and w b for some a, b
u v 2b and w b for some a, b
u v 2 w where u , v, w
u v 2 w where u , v, w
Thus Im T (u, v, w) u , v, w and u v 2 w
(v 2 w, v, w) v, w
(v, v,0) (2 w,0, w) v, w
v(1,1,0) w(2,0,1) v, w
Therefore ImT is a subspace of generated by (1, 1, 0) and (2, 0,1).
3
What is dim(kerT)? What is dim(ImT)?
52
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
ii) Since ker T = {(0, 0)} (contains only the zero vector of ), T is
2
one – to – one by theorem 4.3.2. But it is not on to why?
Notice that whenever kerT {Ov}, we can conclude that T is not one-to-one.
Activity 5.2.1:
Let T: 3 3 be given by T(x, y, z) = (x + y, 2y, 2y – x)
a) Show that T is a linear transformation
b) i) Find ker T
ii) Is T one-to-one? Verify
iii) Find ImT.
Does a linear transformation maps linearly independent vectors in to linearly independent vectors?
Consider the linear transformations L and T from 3 in to 3 given by
L(a, b, c) = (a, c, 0) and T(x, y, z) = (x, y + x, z + y). The vectors (1, 0, 0) and (2, 1, 0) are linearly
independent in 3. But L(1, 0, 0) = (1, 0, 0) and
L(2, 1, 0) = (2, 0, 0) are linearly dependent vectors in 3. On the other hand
T(1, 0, 0) = (1, 1, 0) and T(2, 1,0) = (2, 3, 1) linearly independent. Thus from this particular instance we
conclude that a linear transformation may or may not map linearly independent vectors in to linearly
independent vectors.
Under what condition does it map linearly independent vectors in to linearly independent vectors?
The following theorem provides a sufficient condition for this.
Theorem 5.2.3: Let T: V W be a linear transformation and v1, v2, v3 …, vn be linearly independent vectors
in V. If kerT = {0v} then T(v1), T(v2), T(v3),…, T(vn) are linearly independent vectors
in W.
Proof: Let x1, x2, x3 …, xn be scalars such that
x1T(v1) + x2T(v2) + x3 T(v3) + … + xnT(vn) = Ow
Then by linearity of T we get, T(x1v1 + x2v2 + x3v3 + …+ xnvn) = Ow.
Hence x1v1 + x2 v2 + x3v3 + … + xnvn ker T. But ker T = {Ov}. Thus
x1v1 + x2 v2 + x3v3 + … xnvn = Ov. Since v1, v2, v3, …, vn are linearly independent
vectors in V, it follows that x1 = x2 = x3 = … = xn = 0
This completes the proof.
53
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
In the theorem we have proved that if the kernel of a linear transformation T contains only the zero vector i.e
T is 1 – 1 then T maps linearly independent vectors in to linearly independent vectors.
The next theorem relates the dimension of the kernel and image of a linear transformation L: V W with the
dimension of V. Before going to it let us have the following definition.
Definition 5.2.2: Let L be a linear transformation from a vector space V in to W over the field K.
(a) The dimension of the Kernel (the null space) of L is called the nullity of L.
(b) The dimension of the Image (the range) of L is called the rank of L.
Example 5.2.3: The linear transformation G: V W with V = and W 2 of example 4.3.1 is given
3
by G ( x, y, z ) ( x y, y z ) .
We have seen that its kernel is a subspace of generated by
3
(1, -1, 1). So dim (ker G) =1 . That is Nullity of G = 1. Its image is a subspace of with
2
basis {(1, 0), (0, 1)}. So dim (ImG) = 2 = Rank of G. Since dim V= 3 and dim(ker G) +
dim (ImG) = 3, we have
dim V = Nullity of G + Rank of G.
Example 5.2.4: Consider the linear transformation T : given in example 4.3.2. We have seen that
2 3
its kernel contains only the zero vector, so
dim (ker T) = 0. The set {(1, 1, 0), (2, 0, 1)} is a basis of its image, so dim (ImT) = 2.
Again we have dim 2
Nullity of T + Rank of T
Theorem 5.2.4: (Rank-nullity theorem) Let V and W be vector spaces over the same field K. Let L: V
W be a linear transformation. If V is finite dimensional vector space then dim V = nullity
of L + rank of L
i.e dim V = dim (ker L) + dim (ImL)
Proof: Since V is finite dimensional vector space, it is obvious that Ker L and
ImL =L (V) are finite dimensional. Moreover dim (Ker L), dim (ImL) dim V
(Verify)
Let {u1, u2, …, up} and {w1, w2, …, wq} be basis of kerL and ImL respectively
(p, q dim V) Then there exist v1, v2, …, vq V such that L(vi) = wi for i = 1, 2,
3,…q as wi ImL.
Claim: = {u1, u2, …, up, v1, v2, …, vq} is a bais of V.
Now we show that
54
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
i) generates V
ii) is linearly independent.
i) Let v V. Then L(v) ImL and hence there exist unique scalars b1, b2, …, bqin K such that L(v) =
b1w1 + b2w2 + … + bqwq.
So L(v) = b1L(v1) + b2L(v2) + … + bqL(vq) as L(vi) = wi for i = 1, 2, ..q
= L(b1v1 + b2v2 + … + bqvq)
Thus L(v-b1v1 – b2v2 - … - bqvq) = Ow. Which implies v – b1v1 – b2v2- … - bqvq ker L.
Hence v – b1v1 – b2v2- … -bqvq = a1u1 + a2u2 + …+apup for some unique
scalars a1, a2, …, ap in K. Therefore v = a1u1 + a2u2 + …+apup + b1v1 + b2v2+…+bqvq
From the this we conclude that generates V.
ii) Suppose 1u1 + 2u2 + … + pup + r1v1 + r2v2+ … + rqvq = Ov ……… (1)
where 1, 2 …,pr1, r2, …, rq K
Then L(1u1 + 2u2 + … + pup + r1v1 + r2v2 + … + rqvq) = L(Ov) = Ow
So we have 1L(u1) + 2L(u2) + …+pL(up) + r1L(v1) + r2L(v2) + … + rqL(vq) = Ow
r1w1 + r2w2 + … + rqwq = Ow since L(uj) = Ow and
L(vi) = wi j = 1, 2, …, p and i = 1, 2, …,q
r1 = r2 = … = rq = 0, since {w1, w2, …, wq} is basis of ImL.
1u1 + 2u2 + … + pup = Ov (replace r1, r2 … rqby 0 in (1)
1 = 2 = … = p = 0, since {u1, u2, …, up} is a basis of Ker L.
Thus we have shown that
1u1 + 2 u2 + … + pup + r1v1 + r2v2 + … + rqvq = Ov
1 = 2 = … = p = r1 = v2 = … = rq = 0
That is is a linearly independent set in V. From (i) and (ii) it follows that
= {u1, u2, …, up, v1, v2, …, vq} is a basis of V.
Hence dimV = p + q = dim (kerL) + dim (ImL)
Therefore dim V = Nullity of L + Rank of L.
Example 5.2.5 : 1) Let T: 4 3 be a linear transformation defined by
T(a, b, c, d) = (a – b + c + d, a + 2c – d, a + b + 3c – 3d) a, b, c, d
i) Find the rank and nullity of T.
ii) Is T one-to-one? Is T on to?
Solution: i) First let us find nullity of T.
55
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Let (x, y, z, w) ker T
Then T(x, y, z, w) = (0, 0, 0)
(x – y + z + w, x + 2z – w, x + y + 3z – 3w) = (0,0,0)
x y z w 0 ... (1)
x 2z w 0 ...( 2)
x y 3z 3w 0 ... (3)
Adding (1) and (3) we get 2x + 4z – 2w = 0
By dividing both sides of this equation by 2, we get equation (2)
i.e. x + 2z – w = 0. So w = x + 2z, and y = x + z + w = 2x + 3z
Thus (x, y, z, w) = (x, 2x + 3z, z, x + 2z)
= (x, 2x, 0, x) + (0, 3z, z, 2z)
= x(1, 2, 0, 1) + z(0, 3, 1, 2)
Hence the vectors (1, 2, 0, 1) and (0, 3, 1, 2) generate ker T as every (x, y, z, w) in
ker T can be written as a linear combination of (1, 2, 0, 1) and (0, 3, 1, 2).
Moreover they are linearly independent vectors in ker T. Therefore
{(1, 2, 0, 1),(0, 3, 1, 2)} is a basis of ker T and hence dim ker T = 2. That is
nullity of T is 2. Now by using rank-nullity theorem, we can easily determine the
rank of T.
dim(4) = Nullity of T + rank of T
4 = 2 + Rank of T
Rank of T = 2
ii) T is not one-to-one because kerT {0}. Since Rank of T = 2 which is different from
dim(3), ImT 3. So T is not on to.
Example 5.2.6: Find a linear transformation T: 34 whose range (image) is spanned
by (1, 2, 0, -4), (2, 0, -1, -3).
Solution: Given ImT is spanned by {(1, 2, 0, -4), (2, 0, -1, -3)}
Let us include a vector (0,0,0,0) in this set which will not affect the given
spanning property. Now by using, the standard basis of 3 and theorem 4. 2.2,
one can get a linear transformation T: 3 4 such that
T(1, 0, 0) = (1, 2, 0, -4), T(0, 1, 0) = (2, 0, -1, -3) and T(0,0,1) = (0,0,0,0)
If (x, y, z) 3, T(x, y, z) = T(x(1, 0, 0) + y(0, 1, 0) + z(0, 0, 1))
= x T(1, 0, 0) + yT(0, 1, 0) + zT (0, 0, 1)
= (x, 2x, 0, -4x) + (2y, 0, -y, -3y) + z(0, 0, 0, 0)
56
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Therefore T(x, y, z) = (x + 2y, 2x, -y, -4x-3y) is the required linear
transformation.
Example 5.2.7: Let L: 3 3 be a linear transformation defined by
L(x, y, z) = (x + y+ z, 2x + y – z, 3x + 2y).
i) Find a basis for ker L
ii) Find a basis for ImL iii) Verify rank nullity theorem for L
Solution: i) From the definition of kerL we have,
(a, b, c) kerL L (a, b, c) = (0, 0, 0)
(a + b + c, 2a + b – c, 3a + 2b) = (0, 0, 0)
a b c 0
2 a b c 0
3a 2b 0
3
From 3a + 2b = 0, we get b a
2
a
Substituting this in a + b + c = 0 and 2a + b – c = 0 we get c .
2
Thus any element (a, b, c) of ker L, can be expressed as
(a, b, c) = a ,
3 a = 3 1
a , a 1, ,
2 2 2 2
That is 1,
3 1 generates ker L. So 3 1 is a basis of ker L
, 1, ,
2 2 2 2
ii) Again by using the definition for ImL, we have
(u, v, w) ImL (u, v, w) = L(x, y, z) for some (x, y, z) 3
(u, v, w) = (x + y + z, 2x + y – z, 3x + 2y), where x, y, z
xyzu
2 x y z v
3x 2 y w
By adding the first two equations we get, 3x + 2y = u + v. But 3x + 2y = w. So w = u + v
Thus any vector (u, v, w) in ImL can be written as
(u, v, w) = (u, v, u + v) ; u, v
= (u, o, u ) + (o, v, v) ; u, v
= u(1, 0, 1) + v(0, 1, 1) ; u, v
57
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Hence every vector in ImL is a linear combination of (1, 0, 1) and (0, 1, 1).
So {(1, 0, 1), (0, 1, 1)} generates ImL. Moreover (1, 0, 1) and (0, 1, 1) are linearly independent (verify).
Consequently {(1, 0, 1), (0, 1, 1)} is a basis of ImL.
iii) dim(kerL) = 1 as its basis contains only one non-zero vector, dim(ImL) = 2 as its basis contains two
vectors. dim (kerL) + dim (ImL) = 1 + 2 = 3
i.e. nullity of L + rank of L = dim(3) as stated in the theorem.
Recall that a linear transformation T: V W
i) is one-to-one iff ker T = {0v}
ii) is on to iff ImT = W
Now suppose V and W are respectively n and m dimensional vector spaces over the same field K. If n > m,
is there a one-to-one linear transformation from V in to W?
Suppose there is, then ker T = {0v} and hence dim kerT = 0. So dimV = dim (ImT) by using rank-nullity
theorem. Thus dim (ImT) > dim W. But this is not possible as ImT is a subspace of W. Therefore there is no
one-to-one linear transformation T from V in to W if dim V > dimW.
If n < m, does these exist an on to linear transformation T: V W?
If it exists ImT = W and dimW = dim (TmT). Thus form rank-nullity theorem it follows that n = dim ker T
+ m, which implies dim ker T = n – m < 0 which is not possible. Therefore there does not exist a linear
transformation T : V W which is on to if dim V is less than dim W.
Is there a one – to – one linear mapping form 3 to 2 ?
Is there an on to linear mapping from 3 to 4?
Exercise:
1 For each of the following linear transformation, find a basis and dimension of its image and kernel.
i) T : 3 3 defined by T(x, y, z) = (x, x + 2y, y)
ii) F : 3 2 defined by F(a, b, c) = (a, c)
iii) L : 4 3 defined by L(x, y, z, t) = (x – y + z + t, x + 2z – t, x + y + 3z- 3t)
iv) TA : 4 3 defined by TA(X) = A.X where
58
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
1 2 0 1
A 2 1 2 1 and X is a column vector in 4
1 4 4 2
2) Find a linear transformation T : 3 3 such that {(x, y, z): 4x – 3y + z = 0} is the i) Kernel of
T ii) image of T
3) Find a linear transformation F : 4 3 whose kernel is generated by
(1, 2, 3, 4) and (1, 1, 0, 1).
4) Find a linear transformation T : 3 4 whose image is generated by
(1, -1, 2, 3) and (2, 3, 0, 1).
5) Let L: V W be a liner transformation. Let w W and vo V such that
L(vo) = w. Show that L(x) = w iff x = vo + u where u is an element of the
kernel of L.
6) Let the linear transformation T : 3 3 be defined by
T(u, v, w) = (u + v – 2w, u + 2v – w, 2u + v)
Find the rank and nullity of T.
7. Show that the linear transformation T : 3 3 for which
T(e1) = e1 + e2, T(e2) = 2e2 + e3 and T(e3) = e1 + e2 + e3 is both one-to-one and on to.
8. Let the linear transformation from 3 in to 3 be defined by
T(x1, x2, x3) = (x1 – x2 + 2x3, 2x1 + x2, - x1 – 2x2 + 2x3)
i) If (a, b, c) is a vector in 3, what are the conditions on a, b, and c so that the vector be in the
range of T? What is the rank of T ?
ii) What are the conditions on a, b, and c so that (a, b, c) be in the null space of T? What is the
nullity of T?
iii) Is T one-to-one? Why?
Is T on-to? Why?
5.3. Algebra of Linear transformations
In the study of linear transformation from vector space V in to W, it is of fundamental importance that the set
of these transformations possesses a natural vector space structure.
59
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
The set of linear transformations from a vector space V into itself has even more algebraic structure because
of ordinary composition of function, provides a multiplication of such transformations. We shall explore
these ideas in this section.
Notation: We denote the set of all linear transformations from a vector space V in to W over the field K by
L(V, W). While using the notation L(V, W), it should be noticed that V and W are vector
spaces over the same field . Let T, S L(V, W) and K. Then T and S are linear
transformations from V in to W and hence they are functions from V in to W. Thus
i) the sum of T and S, T + S is a function from V in to W defined by (T + S) (v) = T(v) +
S(v) v V
ii) the scalar multiple of T by , T is a function from V in to W defined by (T) (v) =
(T(v)) v V.
We now state a theorem that asserts T + S and T are linear transformations form V in to W for any T, S
L(V, W) and K; and L (V, W) is a vector space with addition and multiplication by scalars defined as
above.
We shall prove the first two assertions and leave the last one as an exercise.
Theorem 5.3.1: Let V and W be vector spaces over the field K. Let T and S be linear transformations from V
in to W and K. Then
i) the function T + S is a linear transformation from V in to W.
i.e T + S L (V, W).
ii) the function T is a linear transformation from V in to W.
i.e T L (V, W).
iii) L (V, W) the set of all linear transformations from V in to W with
respect to the operation of vector addition and scalar multiplication
defined as (T + S) (v) = T(v) + S(v) and (T (v) = T(v) for all v V,
is a vector space over K.
Proof: i) We need to show that T + S is a linear transformation from V in to W for
any linear transformation T and S from V in to W. For this let
v1 , v2 V and K.
Then (T + S) (v1 + v2) = T(v1 + v2) + S(v1 + v2) why?
= T(v1) + T(v2) + S(v1) + Sv2) why?
= T(v1) + S(v1) + T(v2) + S(v2) why?
60
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
= (T + S) (v1) + (T + S) (v2) why?
Moreover (T + S) (v1) = T(v1) + S(v1)
= T(v1) + S(v1) why?
= (T(v1) + S(v1)) why?
= (T + S) (v1)
Therefore T + S is a linear transformation.
ii) Let T be a linear transformation form V into W and K
Then for any v1, v2 V and K.
We have (T) (v1 + v2) = (T(v1 + v2))
= (T(v1) + T(v2) ) why?
= (T(v1)) + (T(v2)) why?
= (T) (v1) + (T) (v2)
and (T) (v1) = (T(v1))
= (T(v1)) why?
= () (T(v1))
= () (T(v1))
= ((T(v1)))
= (T)(v1)
Therefore T is a linear transformation.
iii) Here we need to verify that all the conditions in the definition of a vector
space are satisfied. We leave this to the students.
Note: In the above theorem, we have asserted that L(V,W) is a vector space over K. With this we can
consider every linear transformation in L(V, W) as a vector. The zero vector in this vector space will
be the zero transformation that sends every vector of V in to the zero vector in W.
Example 5.3.1: Let T : 32 and H : 32 be linear transformations given by
T(x, y, z) = (3x, y + z) and H(x, y, z) = (2x – z, y)
a. Are T + H and 3T – 4H linear transformations form 3 in to 2
b. Find (i) T + H (c) 3T – 4H
Solution: a) Since T and H are linear transformation from 3 in to 2, we have
T, H L(3, 2) and hence T + H, 3T – 4H L (3, 2) as
61
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
L( 3, 2) is a vector space over . So T + H and 3T – 4H are linear transformations
from 3 in to 2.
b) i) (T + H) (x, y, z) = T (x, y, z) + H(x, y, z)
= (3x, y + z) + (2x-z, y)
= (5x – z, 2y + z)
ii) (3T – 4H) (x, y, z) = (3T + (-4H)) (x, y, z)
= (3T) (x, y, z) + (-4H) (x, y, z)
= 3.T(x, y, z) + (-4) H(x, y, z)
= 3(3x, y + z) + (-4) (2x – z, y)
= (9x, 3y + 3z) + (-8x + 4z, -4y)
= (x + 4z, 3z – y)
Theorem 5.3.2: Let V, W and Z be vector spaces over the field K. Let T and S be linear transformations
from V in to W and from W into Z respectively. Then the compose function SoT defined
by (SoT) (v) = S(T(v)) for all v V is a linear transformation.
Proof: Let T and S be as in the hypothesis of the theorem Let v1, v2 V and r K
Then (SoT) (v1 + v2) = S(T(v1 + v2))
= S(T(v1) + T(v2)) why?
= S(T(v1)) + S(T(v2)) why?
= (SoT) (v1) + (SoT) (v2)
and (SoT) (rv1) = S(T(rv1))
= S(rT(v1)) because T is a linear transformation.
= rS(T(v1)) why?
= r(SoT) (v1)
Therefore SoT is a linear transformation.
Notation: For the sake of brevity we shall simply denote the composition SoT of S and T by ST.
Activity 5.3.1:
1) Give two linear transformation S and T from 3 in to it self such that ST TS. From this you may
conclude that composition of linear transformations is not commutative.
2) Is composition of linear transformations associative? Justify your answer!
3) Let U, V and W be vector spaces over the field K.
Let S, S' L (U,V) and T, T' L (V, W)
Verify each of the following.
62
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
i) T(S + S') = TS + TS'
ii) (T + T')S = TS + T'S
iii) (T S) = ( T)S = T( S) for all K
Remember that by a linear operator on vector space V we mean a linear transformation from a vector space V
in to itself.
If T is a linear operator on vector space V, we can compose T with T to get TT which is again a linear
operator on V. We shall use the notation T2 = TT, T3 = TTT and in general
Tn = T.T … T (the composite of T with itself n times) for n = 1, 2, 3, …. We define To = I (identity mapping)
if T 0 (the zero mapping).
Example 5.3.2: Let T be a linear operator on a vector space V over the field F.
(i) If T2 = O the zero mapping then what can you say about the
relation of the range of T to the kernel of T?
ii) Give an example of a linear operator on 2 such that
T2 = O but T O.
Solution: i) Method 1: T2 = O T2(v) = O(v), v V
T(T(v)) = Ov – zero vector in V.
T(v) ker T v V
But T(v) is an arbitrary element of Range of T for any v V.
Therefore Range of T ker T.
Method 2: Let w Range of T. Then T(v) = w for some v V.
So T(w) = T(T(v)) = T2(v) = Ov as
T2(v) = Ov v V. Hence T(v) = wkerT
Therefore, Range of T is a subset of kerT
ii) Consider the linear map T: 2 2 such that
T(a, b) = (0, a) (a, b) 2
Then T O as T(1, 1) = (0, 1) (0, 0)
Now T2 (a, b) = (T.T) (a, b)
= T(T(a, b))
= T(0, a)
= (0, 0)
So, T2(a, b) = (0, 0) (a, b) 2
63
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Hence, T2 = O
Example 5.3.3: Let T:22 be a linear operator on 2 given by
T(a, b) = (a + b, a + b). Show that T(T – 2I) = O, where I is the identity
mapping on 2 and O is the zero mapping on 2.
Solution: (T – 2I) (x, y) = T(x, y) - 2I (x, y)
= (x + y, x + y) – 2(x, y) = (y – x, x – y)
So, [T(T – 2I)] (x, y) = T((T – 2I) (x, y))
= T(y – x, x – y)
= (y – x + x – y, y – x + x – y)
= (0, 0)
Thus T(T – 2I) =O
Activity 5.3.2:
1. Let T and S be linear operators on 2. Does TS = O imply either T = O or S = O? Explain! Give a
counter example if your answer is No.
2. Let V be finite dimensional vector space over the field F and T be a linear operator on V. Suppose
that rank (T2) = rank T. Prove that the range and null space of T have only the zero vector in common.
Now let us discuss about inverse of a linear transformation. Recall that a function T from V in to W is called
invertible if there exists a functions S from W in to V such that ST is the identity function on V and TS is the
identity function on W. If T is invertible the function S is unique and is denoted by T-1and is called the
inverse of T.
T-1(w) = v T(v) = w whenever T-1 exists. Furthermore we know that T is invertible iff T is one- to- one
and on to.
Theorem 5.3.3: Let V and W be vector spaces over the field F and Let T be a linear transformation from V in
to W. If T is invertible then the inverse function T-1 is a linear transformation.
Proof: Suppose T : V W is invertible then there exists a unique function
T-1: W V such that T-1(w) = v T(v) = w for all w W. Moreover
T is 1-1 and on to. We need to show that T-1 is linear. Let w1, w2 W and F
Then there exist unique vectors v1, v2 V such that T(v1) = w1 and T(v2) = w2
as T is 1-1 and on to. So T-1(w1) = v1 and T-1(w2) = v2.
T(v1 + v2) = T(v1) + T(v2) because T is linear
= w1 + w2
64
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Thus T-1(w1 + w2) = v1 + v2 = T-1(w1) + T-1(w2) for any w1, w2 W.
Since T( v1) = w1, T-1 (w1) = v1 = T-1(w1). Therefore T-1 is a linear transformation.
Remark: Let T : V W and S : W Z be invertible linear transformations. Then
i) T-1 : W V is also invertible and T
1 1
T
ii) ST : V Z is also invertible and (ST)-1 = T-1S-1
Example 5.3.4: Let T : 3 3 be a linear transformation given by
T(x, y, z) = (x, 3y, z). Is T invertible? If so, find a rule for T-1 like the
one which defines T.
Solution: We know that if kerT = {} then T is one to-one
KerT = (a, b, c) 3
| T (a, b, c) (0,0,0)
= (a, b, c) 3
| (a,3b, c) (0,0,0)
= (a, b, c) 3
| a 0, 3b 0 and c 0
= (0,0,0).
Therefore T is one to one.
Moreover from rank-nullity theorem, dim(3) = dim (kerT) + dim (ImT).
3 = 0 + dim(ImT). So dim(ImT) = 3. Since ImT is a subspace of 3
and dim(ImT) = 3, we have ImT = 3. Hence T is on to as ImT = 3.
Therefore T is invertible as it one to- one and on to.
To find T-1, let T (x, y, z) = (u, v, w).
v
Then (x, 3y, z) = (u, v, w). So x = u, y , z w.
3
But T(x, y, z) = (u, v, w) T-1 (u, v, w) = (x, y, z).
v
Thus T-1 (u, v, w) = u, , w .
3
Whenever V is a finite dimensional vector space, one – to – oneness can be related to linear independence
and rank as stated in the theorem below.
Theorem 5.3.4: Let T : V W be a linear transformation and V be finite
dimensional vector space with dimV = n. Then the following
65
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
statements are equivalent.
i) T is 1 – 1
ii) If v1, v2, v3, …, vk are linearly independent vectors in V then T(v1), T(v2), T(v3), …, T(vk) are
linearly independent vectors in ImT
iii) If {v1, v2, v3, …, vn} is a basis of V then {T(v1), T(v2), T(v3), …, T(vn)} is a basis of ImT.
iv) Dim (ImT) = n
Proof: Left for students
Activity 5.3.3:
Prove theorem 4.4.4. Follow the following steps
1. Show that (iv) (i)
2. Show that (i) (ii)
3. Show that (ii) (iii)
4. Show that (iii) (iv)
5. Form 1, 2, 3 and 4, can we conclude that (i) (ii) (iii) (iv)? How?
Remark: For finite dimensional vector spaces V and W over the field K with
dimV = dimW, we have the following results about any linear transformation T : V W.
a) T is 1 – 1 T is invertible
b) T is on to T is invertible.
Example 5.3.5: Let L : V V be a linear operator such that L2 = O. Show that I – L is invertible. (I is the
identity mapping on V.)
Solution: Clearly I – L is a linear operator on V.
Let x ker(I-L). Then (I - L) (x) = Ov
That I(x) – L(x) = Ov. So we have L(x) = x and L2(x) = L(x).
But L2 = O and hence L2 (x) = Ov. Thus x = L(x) = Ov.
Therefore ker (I – L) = {Ov}. Consequently I – L is one to one. In this case it is not sufficient
to have I – L is 1 – 1 to conclude that it is invertible as V can be infinite dimensional. (Nothing
is assumed about V whether it is finite or infinite dimensional). So we need to show that I – L
is on to. For this let v V. Then L(v) V and L(v) + v V. Moreover
(I – L) (L(v) + v) = I(L(v)+v) – L(L(v) + v)
= L(v) + v – L2(v) – L(v)
=v (L2(v) = Ov )
Thus L(v) + v is the pre-image of v under I – L. So for any vector v V
66
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
there exists a vector x = L(v) + v in V such that (I – L) (x) =v. From this
it follows that (I – L): V V is on to. Therefore I – L is invertible as it is
1 – 1 and on to.
Exercise:
1. Let T and S be linear operators on 2 defined by T(x, y) = (y, x) and S(a, b) = (a, 0)
i) How do you describe T and S geometrically?
ii) Give rules like the one defining T and S for each of the linear transformations S – 2T, ST,
TS, T2, S2.
2. Let T be the linear operator on 3 defined by
T(x1, x2, x3) = (3x1, x1 – x2, 2x1 + x2 + x3). Is T invertible? If so, find a rule for T-1.
3. For the linear operator T of exercise 2, Show that (T2 – I) (T – 3I) = O
(I – identity aping and O zero mapping)
4. Let T be a linear transformation from 3 in to 2 and Let U be a linear transformation from 2 in to 3.
Prove that the linear transformation UT is not
invertible.
5. L : 3 3 be a linear transformation. Show that L is invertible and find L-1 for:
a) L( x, y, z ) ( x y, x z , y 2 z)
b) L( x, y, z ) (2 x y z , x y, 3x y z )
6. a) Let S : V V be a linear operator such that S2 – S + I = (where I is the identity mapping on V and
is the zero mapping on V) Show that S-1 exists and is equal to I – S.
b) Let T be a linear operator on a vector space V, and assume that L3 (v) = Ov for all v V. Show that I
– L is invertible.
7. Let F and G be linear operators on a vector space V over the set of real numbers.. Assume that FG = GF.
Show that
i) (F + G)2 = F2 + 2FG + G2
ii) (F + G) (F – G) = F2 – G2
[Link] representation of a linear transformation
In this section we shall investigate the strong relationship that exists between linear transformation and
matrices.
67
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Linear transformation associated with a given matrix
Let us consider a matrix A a ij m n over a field K ( i .e a ij K for each i = 1,2,…,m and for each j = 1,
2, …,n). We can associate a mapping T A : K n K m by defining T A ( X ) AX for any column vector X in
Kn.
Activity 5.4.1: Show that TA is a linear transformation
Definition 5.4.1: For any matrix A a ij
mn
over a field K. The mapping T A : K n K m defined by
T A ( X ) AX X K n is called the linear transformation associated with the
matrix A.
Note: In this definition, each element X of Kn should be considered as column vector otherwise AX is not
defined.
1 0
Example 5.4.1: a) Let A . Then T A : 2 2 is an identity linear transformation.
0 1
1 0 1 0 x
x x
b) Let B 2 1 . Then T B y 2 1 y 2 x y
3 0 3 0 3x
x
x
Thus T B : given by T B 2 x
2 3
y is the linear
y 3x
transformation associated with matrix B.
1 3 0 2 2
c) Let A , u , v , w and TA be the linear
0 1 2 0 2
transformation associated with matrix A.
Find i) T A ( u ), T A ( v ) and T A ( w )
ii) The image of the square with vertices
0 2 2 0
, , and .
0 0 2 2
Solution:
1 3 0 6
i) T A ( u ) Au
0 1 2 2
68
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
1 3 2 2
T A ( v ) AV
0 1 0 0
1 3 2 8
T A ( w ) Aw
0 1 2 2
ii) T A deforms the given square as if the top of the square were
pushed to the right while the base is held fixed (see the figure below).
We call such transformation shear transformation.
Activity 5.4.2:
1. Let A be a 7 5 matrix. What must a and b be inorder to define T : a b by T ( x ) AX
1 0
2. Let A . Give a geometrical description of the linear transformation TA associated with A.
0 1
1 3 1
3. Let B , b
3 1 7
i) Find the linear transformation T associated with B.
ii) Find X in 2 whose image under TB is b
iii) Determine whether the system BX b has no solution, unique
solution or many solutions
In view of definition 4.5.1 we can study system of linear equations with the help of linear transformation
associated with the coefficient matrix of the system. Consider the system AX b where A is an m n real
matrix. Then AX b iff T A ( X ) b where TA is the linear transformation associated with matrix A. Thus
the system AX = b has a solution iff b is in the range of TA. If there is exactly one element X n whose
image is b under TA then the system AX b has exactly one solution. But if b has more than one pre-images
69
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
n
under TA then the system has more than one solution. If there is no X such that T A ( X ) b , (i.e b is
not in the range of TA), then the system has no solution.
Activity 5.4.3:
Prove that if ker TA {} , then the system AX b has at most one solution.
Further if b the zero column vector in m then the homogeneous system AX has at least one
solution. What is this solution?
The solution set of AX is the kernel of TA. So the solution set of AX is a subspace of n . Suppose
dim ker T A k and v 1 , v 2 , v 3 ,..., v k is a basis of ker TA. Then any solution v of AX 0 can be expressed
as
v 1v 1 2 v 2 3 v 3 ... k v k where 1 , 2 , 3 ,..., k are scalars.
Now let Xo be one particular solution of the non-homogeneous system AX b and w be any solution of
AX b .
Then AX o b Aw so we have, T A ( X o ) T A ( w ) . This in turn implies T A ( X o w ) 0 and hence
w X o ker T A . Thus w x o 1v 1 2 v 2 ... k v k
or w xo 1v1 2v2 ... k vk where 1 , 2 ,..., k are scalars.
Therefore if Xo is one particular solution of the non-homogeneous system AX b then every solution w of
AX b is given by w xo 1v1 2v2 3v3 ... k vk where v 1 , v 2 , v 3 , ..., v k is a basis ker TA and
1 , 2 , 3 ,..., k are scalars.
This called the general solution of the non homogeneous system AX b .
Example 5.4.2
1 0 3 4
Let A , b and T A be the linear transformation associated with matrix A.
2 1 3 9
i) Find ker TA ?
ii) Is b Im TA ?
iii) Is there more than one X whose image under TA is b?
iv) Describe the solution set of AX b
Solution: T A : 3 2 is given by
70
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
x1 x
1 0 3 1 x1 3 x 3
T A x 2 x 2
x 2 1 3 x 2 x1 x 2 3 x 3
3 3
i) ker TA X 3 TA ( X ) 0
a
a 3c 0 a 3c 0
So b ker T A
c 2 a b 3c 0 2a b 3c 0
a 3c b
3c 3
Thus ker T A 3c c = c 3 c
c 1
ii) Since dim(ker TA) = 1 and 3 = dim (ker TA) + dim(ImTA),
dim(ImTA) = 2. Hence TA is on to. So every element in 2 has a pre-
image under TA . Consequently b Im T A as b 2 .
iii) Since b has a pre-image under TA , there exists x o 3 such that T A ( x o ) b . Moreover for any
3 3 3
, w x o 3 is a pre-image of b, as T A ( w ) T A ( x o 3 T A ( x o ) T A 3 =
1 1 1
b . = b
Thus we have more than one element in 3 whose image under TA is b.
4
iv) 1 is a particular solution of AX = b
0
4
1 0 3 4
as 1 .
2 1 3 0 9
4 3
Therefore, 1 q 3 q is the solution set of AX = b
0 1
71
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Exercise:
1 0 1
1) Let A 3 1 5 and TA be the linear transformation associated with matrix A. Find X
4 2 1
such that TA ( X )
1 3 4 3 1
2) Let A 0 1 3 2 , b 1 and TA be the linear transformation associated with
3 7 6 5 7
matrix A.
i) Find ket TA ii) Is b in the range of TA
iii) Describe the solution set of AX = b
3) Suppose T : 5 2 and T(X) = AX for some matrix A and each X in 5 . How many
rows and columns does A have?
4 5
4) Use a rectangular coordinate system to plot , , and their images under the linear
2 2
transformation TA associated with matrix A. (Make a separate sketch for each exercise). Give a
x
geometric description of what TA does to a vector X 1 in 2 where
x2
0 .5 0 1 0
i) A ii) A
0 0 . 5 0 1
0 1 2 0
iii) A iv) A
1 0 0 1
The matrix of a linear transformation
In the previous subsection we have seen that associated to any given m n matrix A there is a linear
transformation T A : n m defined by T A ( X ) AX . In this subsection we shall see the reverse process,
that is to find a matrix associated to a given linear transformation from a finite dimensional vector space V in
to finite dimensional vector space W over the same field K.
Before going further let us recall about the coordinates of an element V of a finite dimensional vector space
V with respect to a given ordered basis of V. What do we mean by ordered basis? Suppose
b1 , b2 ,..., bn is an ordered basis for a finite dimensional vector space V over a field K and v is in V. The
72
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
coordinates of v relative to the basis (or the coordinates of v) are the scalars c1 , c 2 ,..., c n in K such
that v c1b1 c 2 b2 ... c n bn .
If c1 , c 2 ,..., c n are the -coordinates of v then the vector in Kn,
c1
c2
.
v
.
.
cn
is the coordinate vector of v relative to or the -coordinate vector of v.
Example 5.4.3: (i) The coordinates of (x, y, z) relative the standard basis
1,0,0, 0,1,0, 0,0,1 of 3 are simply x, y and z, since
(x, y, z) = x(1, 0, 0)+ y(0, 1, 0+ z(0, 0, 1).
1 1
ii) Consider a basis = {b1, b2} of 2 , where b1 and b2 . The coordinate
0 2
1 2
vector of X relative to is X ,
6 3
since X = (-2)b1 + 3b2.
Activity 5.4.4:
5 3 4
1) Find the vector X determined by the coordinate vector X where ,
3 5 6
3 1 2 1
2) Find the coordinate vector X of X 5 relative to the basis 0 , 1 , 1 of
3
4 3 8 2
3) Let P2 be the space of polynomial functions from in to of degree two or less and P be an
element of P2 defined by P(t) = 1 + 4t + 7t2
a) Find the coordinate vector of P relative to the standard basis f o , f 1 , f 2 where
f o ( t ) 1, f 1 ( t ) t and f 2 ( t ) t 2 .
73
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
b) Find the coordinate vector of P relative to the basis ' g o , g1 , g 2 where
g o ( t ) 1 t 2 , g1 ( t ) t t 2 and g 2 ( t ) 1 2 t t 2
Now let us deal with the matrix of a given linear transformation.
Let V be an n-dimensional vector space over the field K and let W be an m-dimensional vector space over K.
Let v 1 , v 2 ,..., v n and ' w1 , w2 ,..., wm be ordered bases of V and W respectively. Suppose
T : V W is a linear transformation.
Then for any x in V, T ( x ) W and T(x) can be expressed as a linear combination of elements of the basis
. So we have,
T (v1 ) a11 w1 a21w2 am1wm
T (v2 ) a12 w1 a22 w2 am 2 wm
(1)
T (v j ) a1 j w1 a2 j w2 amj wm
T (vn ) a1n w1 a2 n w2 amn wm
where aij’s are scalars in K.
Writing the coordinates of T(v1),T(v2),…,T(vn) successively as columns of a matrix we get
a11 a12 ... a1 j ... a1n
a a 22 ... a 2 j ... a 2 n
21
. . . .
. . . .
. . . .
M …. (2)
a i1 ai2 a ij a in
. . . .
. . . .
. . . .
a m1 a m2 a mj a mn
m n
That is M T (v1 ) ' T (v2 ) ' ... T (v j ) ' ...T (vn ) '
74
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
The matrix M in (2) is called a matrix representation of T or the matrix for T relative to the bases and
' . If and ' are standard bases of V and W respectively, we call the matrix M in (2) the standard
matrix for the linear transformation T.
Example 5.4.4: Define T :3 2 by
T(x1, x2, x3) = (3x1 - 2x2 + x3, - x1 + x2 +5x3)
a) Find the standard matrix for T.
b) Find the matrix of T relative to the ordered bases
= {(1,0,1), (0,1,1), (0,0,1)} and ’ = {(1,0), (1,1)} of 3 and 2
respectively.
Solution: a) Here we use the standard basis of 3 and 2 .
T(1,0,0) = (3, -1) = 3(1,0) + -1(0.1)
T(0,1,0) = (-2,1) = -2(1,0) + 1(0,1)
T(0,0,1) = (1,5) = 1(1,0) + 5(0,1)
Writing the coordinates of T(1,0,0), T(0,1,0), T(0,0,1) as first, second and
third columns of a matrix we get the standard matrix M for T as
3 2 1
M
1 1 5
b) T(1,0,1) = (4,4) = 0(1,0) + 4(1,1)
T(0,1,1) = (-1,6) = -7(1,0) + 6(1,1)
T(0,0,1) = (1,5) = -4(1,0) + 5(1,1)
0 7 4
The matrix of T relative to and ’ is M
4 6 5
Our next task is to examine how the matrix M in (2) determines the linear transformation T. If x = x1v1 + x2v2
x1
x2
.
+ …+xnvn is a vector in V, then the coordinate vector of x relative to X
.
.
x n
and T(x) = T(x1v1 + x2v2 + …+xnvn) = x1T(v1) + x2T(v2) + …+xnT(vn) ….. (3)
75
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Using the basis ' in W, we can rewrite (3) in terms of coordinate vectors relative to ' as
T ( x ) ' x1 T ( v 1 ) ' x 2 T ( v 2 ) ' ... x n T ( v n ) ' … (4)
Further the vector equation (4) can be written as a matrix equation?
T ( X ) ' M X …. (5)
Thus if X is the coordinate vector of x relative to , then the equation in (5) shows that M X is the
coordinate vector of the vector T(X) relative to ' .
Note: In case when W is the same as V and the basis ' is the same as , the matrix M in (2) is called the
matrix for T relative to and is denoted by T .
Activity 5.4.5: Using equation (3), verify equations (4) and (5).
Example 5.4.5: Let T : 3 2 be the linear transformation defined by
T(x, y, z) = (3x + 2y - 4z, x -5y + 3z). Find the matrix of T relative to the bases
B1 = {(1, 1, 1), (1, 1, 0), (1, 0, 0) of 3 and B2 = {(1, 3), (2, 5)} of 2 .
Solution: T(1, 1, 1) = (1, -1) = -7(1, 3) + 4(2, 5)
T(1, 1, 0) = (5, -4) = -33(1, 3) + 19(2, 5)
T(1, 0, 0) = (3, 1) = -13(1, 3) + 8(2, 5)
The matrix M of T relative to the bases B1 and B2 is:
7 33 13
M
4 19 8
Example 5.4.6: Let b1 , b2 , b3 be a basis for a vector space V over the set of real numbers. Find T(3b1 –
4b2), where T is a linear transformation form V in to V whose matrix relative to is
0 6 1
T 0 5 1
1 2 7
76
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
3
Solution: Let x = 3b1 - 4b2. Then the coordinate vector of x relative to X 4 and the coordinate
0
vector of T(X) relative to is
0 6 1 3 24
T ( X ) T X = 0 5 1 4 = 20
1 2 7 0 11
Hence T(x) = 24b1 - 20b2 + 11b3.
Exercise:
1. Let F : 3 2 be defined by F(x, y, z) = (z - x, x + y). Find the matrix associated with F with
respect to the standard bases of 2 and 3 .
2. Let T : 2 3 defined by T(a, b) = (a, b, a+2b). Find the matrix of T relative to the bases B1=
{(1,1), (2,0)} and B2 = {(1,1,1), (1,1,0), (0,1,1)}.
1 2 1
3. Let A and TA be a linear mapping from 3 to 2 defined by TA ( v ) Av where v is
3 4 0
1 0 0
a column vector in 3. Find the matrix of TA relative to the bases B 1 0 , 1 , 0 and
0 0 1
1 2
B 2 , of 3 and 2 respectively.
3 5
4. Suppose that b1 , b2 , b3 and ' d 1 , d 2 be a basis for real vector spaces V and W,
respectively. Let T : V W be a linear transformation with the property that T(b1) = 3d1 - 5d2, T(b2)
= -d1 + 6d2 , T(b3) = 4d2. Find the matrix M for T relative to and ' .
5.5. Eigenvalues and Eigenvectors of a linear Transformation
In this section we look for vectors that are sent by a linear operator into scalar multiple of themselves. Recall
that a linear operator is a linear transformation from a vector space to it self.
Definition 5.5.1:
Let T: V V be a linear operator on a vector space V over a field K. An eigenvalue of T is a scalar in K
such that there is a non-zero vector v in V with T(v) = v. If is an eigenvalue of T, then
77
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
a) any vector v in V such that T(v) = v is called an eigenvector of T associated with the eigenvalue ;
b) the collection of all vectors v of V such that T(v) = v is called the eigenspace associated with .
Note:
1. One of the meanings of the word “eigen” in German is “Proper”. Thus eigen values are also called
proper values or characteristic values or latent roots.
2. If and w are eigenvectors associated with eigenvalue . Then:
i) + w is also an eigenvector with eigenvalue , because:
T( + w) = T() + T(w) = + w = ( + w)
ii) each scalar multiple k, k 0, is also an eigenvector with eigenvalue , because: T(k)
= kT() = k() = (k).
Activity 5.5.1: Let T: V V be a linear operator with ker T {0}. Prove that every non
zero vector in ker T is an eigenvector of T with eigenvalue 0.
Example 5.5.1:
a) Let id: V V be the identity operator.
Every non-zero vector in V is an eigenvector of id with an eigenvalue 1, since:
Id() = = 1.
b) Let T: 2 2 be a linear operator which rotates each 2 by an angle of 90o.
T has no eigenvalues and hence no eigenvectors.
c) Let D: V V be the differential operator on the vector space of differentiable functions. We have
D(e5t) = 5e5t.
Hence, e5t is an eigenvector of D corresponding eigenvalue 5.
5.6. Eigenspace of a linear transformation
Let T: V V a linear operator over a field K.
78
SMU Faculty of Informatics Department of Computer Science
Linear Algebra
Let K. Let V = The set of all eigenvectors of T with eigenvalue .
Claim V is a subspace of V.
Proof:
i) T(0) = 0 = . 0, Showing that 0 V.
ii) 1, 2 V T(1 + 2) = T(1) + T(2)
= 1 + 2
= (1 + 2)
1 + 2 V
iii) V, K T() = T()= ()= ()
V
Thus we have proved that the eigenspace associated with is
79
SMU Faculty of Informatics Department of Computer Science
You might also like
- Understanding Vectors in Linear AlgebraNo ratings yetUnderstanding Vectors in Linear Algebra144 pages
- Introduction to Vectors in Linear AlgebraNo ratings yetIntroduction to Vectors in Linear Algebra146 pages
- Introduction to Vectors in Linear AlgebraNo ratings yetIntroduction to Vectors in Linear Algebra67 pages
- Chapter 2 Real Vector Spaces All of ThemNo ratings yetChapter 2 Real Vector Spaces All of Them79 pages
- Understanding Vectors in Linear AlgebraNo ratings yetUnderstanding Vectors in Linear Algebra21 pages
- Vectors, Lines, and Planes in Calculus 3No ratings yetVectors, Lines, and Planes in Calculus 334 pages
- Understanding Vectors in Linear AlgebraNo ratings yetUnderstanding Vectors in Linear Algebra98 pages
- Understanding Vectors in Linear AlgebraNo ratings yetUnderstanding Vectors in Linear Algebra98 pages
- Vectors and Spaces in Applied Math 231BNo ratings yetVectors and Spaces in Applied Math 231B27 pages
- Introduction to Vector Calculus ConceptsNo ratings yetIntroduction to Vector Calculus Concepts20 pages
- Applications of Vector Algebra ExplainedNo ratings yetApplications of Vector Algebra Explained71 pages
- Understanding Series and Vectors in CalculusNo ratings yetUnderstanding Series and Vectors in Calculus2 pages
- Mathematical Concepts in Coordinate SystemsNo ratings yetMathematical Concepts in Coordinate Systems9 pages
- Algebraic Groups and Discontinuous SubgroupsNo ratings yetAlgebraic Groups and Discontinuous Subgroups424 pages
- Statistical Analysis of Cointegration Vectors100% (2)Statistical Analysis of Cointegration Vectors24 pages
- Derivation Transfer Function From State Space ModelNo ratings yetDerivation Transfer Function From State Space Model4 pages
- Statistical Mechanics in Online LearningNo ratings yetStatistical Mechanics in Online Learning23 pages
- (Gauss) Lin, Kuan-Pin - 2001 - Computational Eco No Metrics. GAUSS Programming For Econometricians and Financial AnalystsNo ratings yet(Gauss) Lin, Kuan-Pin - 2001 - Computational Eco No Metrics. GAUSS Programming For Econometricians and Financial Analysts375 pages
- Hilbert Space of Mean-Zero Random VectorsNo ratings yetHilbert Space of Mean-Zero Random Vectors9 pages
- Compressive Sensing for Face RecognitionNo ratings yetCompressive Sensing for Face Recognition14 pages