Advanced Mathematical Techniques

in Chemical Engineering
(CH61015)

Course notes and practice problems on Linear Algebra

P. A. Deshpande
Department of Chemical Engineering
Indian Institute of Technology Kharagpur
[email protected]
ii
Contents

0 Background: A review of set theory 1–10

1 Groups, rings and fields 11–17

2 Linear vector spaces 19–25

3 Linear independence, basis, dimension and span 27–33

4 Analysis of systems of simultaneous linear equations 35–45

5 Linear transformations 47–57

6 Inner product spaces 59–64

7 Norm and metric spaces 65–69

8 Adjoint operators 71–76

9 Eigenvalue problems 77–87

10 Sturm-Liouville theory 89–101

iii
iv CONTENTS
Chapter 0

Background: A review of set theory

A set is a collection of well-defined objects. The objects present in the set are called its
members or elements. We will generally represent a set by a capital letter and its members
by small letters.
A = {a, e, i, o, u} (1)

A is the set and a, e, i, o, u are its elements. To indicate that a is an element of A, the symbol
∈ is used.
a∈A (2)

The above is read as “a belongs to A” or “a is in A” or “a is an element of A”.

The number of elements present in a set is called its cardinality or cardinal number. A set
is said to be a finite set if its cardinality is finite. Else, the set is said to be an infinite set.
A set is said to be an empty set or a null set or a void set if it does not contain any element.
Such a set is generally denoted by φ. A few examples of empty sets are given below.

φ = {} (3)

φ = {x : x ∈ R ∧ x2 = −1} (4)

A set containing exactly one element is called a singleton set. Two sets A and B are said to
be equal if they have same elements.

A = B iff x ∈ A =⇒ x ∈ B ∧ x ∈ B =⇒ x ∈ A (5)

1
2 CHAPTER 0. BACKGROUND: A REVIEW OF SET THEORY

Two sets are said to be equivalent if they have the same number of elements in them i.e.
their cardinalities are equal.
A ∼ B iff |A| = |B| (6)

B is said to be a subset of A if all elements of B are also elements of A.

B ⊂ A iff x ∈ B =⇒ x ∈ A ∀x ∈ B (7)

A set of cardinality |B| has 2|B| number of subsets. Two sets A and B are said to be
comparable if either A ⊂ B or B ⊂ A.
Power set of a set A is a set of all possible subsets of A.

P(A) = {S : S ⊆ A} (8)

The union of two sets A and B is a set which has elements which are either in A or in B or
in both.
A ∪ B = {x : x ∈ A ∨ x ∈ B ∨ x ∈ A ∧ B} (9)

The intersection of two sets A and B is a set which has elements which are common to both
A and B.
A ∩ B = {x : x ∈ A ∧ x ∈ B} (10)

Laws of sets
Idempotent laws:

A∪A=A (11)

A∩A=A (12)

Identity laws:

A∪φ=A (13)

A∩φ=φ (14)

Commutative laws:

A∪B =B∪A (15)

A∩B =B∩A (16)

 3

Associative laws:

A ∪ (B ∪ C) = (A ∪ B) ∪ C (17)

A ∩ (B ∩ C) = (A ∩ B) ∩ C (18)

Distributive laws:

A ∪ (B ∩ C) = (A ∪ B) ∩ (A ∪ C) (19)

A ∩ (B ∪ C) = (A ∩ B) ∪ (A ∩ C) (20)

Complement laws:

A ∩ A0 = φ (21)

A ∪ A0 = U (22)

φ0 = U (23)

U0 = φ (24)

where U is the universal set.

Difference of two sets A and B is defined as

A − B = {x : x ∈ A ∧ x ∈
/ B} (25)

Symmetric difference of A and B is defined as

A∆B = (A − B) ∪ (B − A) (26)

Ordered pair: In any set, the order in which the elements are written is immaterial i.e.

A = {a1 , a2 , a3 } = {a1 , a3 , a2 } = {a3 , a2 , a1 } . . . (27)

An ordered pair is a collection of two elements described in a particular order i.e.

(a, b) = (b, a) iff a = b (28)

Two ordered pairs (a, b) and (c, d) are equal iff a = c and b = d. As an extension to an
ordered collection of n-elements, we can define ordered n-tuple.
4 CHAPTER 0. BACKGROUND: A REVIEW OF SET THEORY

Cartesian product of sets: Let A and B be two sets. The set of all ordered pairs (a, b) such
that a ∈ A and b ∈ B is called the cartesian product of the two sets. It is denoted by A × B.

A × B = {(a, b) : a ∈ A ∧ b ∈ B} (29)

A × B 6= B × A unless A = B (30)

As an extension, we can write

A1 × A2 × · · · × An = {(a1 , a2 . . . an ) : ai ∈ Ai ∀i} (31)

Relations:
A relation R from a set A to B is a subset of A × B.

R⊂A×B (32)

The domain of a relation R from A to B is the set of all first elements of the ordered pairs
which belong to R.
dom(R) = {a : a ∈ A ∧ (a, b) ∈ R} (33)

The range of a relation R from A to B is the set of all second elements of the ordered pairs
which belong to R.
range(R) = {b : b ∈ B ∧ (a, b) ∈ R} (34)

Consider A = {1, 2, 3} and B = {1, 2, 4}.

A × B = {(1, 1), (1, 2), (1, 4), (2, 1), (2, 2), (2, 4), (3, 1), (3, 2), (3, 4)} (35)

We choose a subset of A × B such that the sum of the elements of the ordered pairs is greater
than or equal to 4. Then

R = {(1, 4), (2, 2), (2, 4), (3, 1), (3, 2), (3, 4)} (36)

It can be seen from Figure 1 that there are more than one arrows emerging from A. Hence,
in relations, it is allowed to have more than one ordered pairs with the same first entry i.e.
more than one elements in A can have a corresponding image in B. Now we define a special
case of relations called a function.
 5

Figure 1: Representation of a relation given by Eq. (36)

Functions:

A function f from A to B, denoted as f : A → B, is a subset of A × B such that

(i) every element of A is the first entry of some ordered pair, and
(ii) no two ordered pairs have the same first entry.

The set A is called the domain of f while the set B is called the co-domain of f . Different
cases given in Figure 2 can be used to identify functions.

Figure 2: Different scenarios depicting a relation to be a function. (i) not a function, (ii)
not a function, (iii) function, (iv) function
6 CHAPTER 0. BACKGROUND: A REVIEW OF SET THEORY

Into function: There is atleast one element in B such that there is no pre-image correspond-
ing to it in A. q in Figure 3(i), for example, does not have any pre-image.

Onto or surjective function: Every element in B has a pre-image in A. In this case,

co − domain(f ) = range(f ) as can be seen from Figure 3(ii).

One-One or injective function: For every image, there is exactly one pre-image. But all
images need not have a pre-image. See Figure 3(iii) for example.

One-one onto or bijective function: This function is both one-one, i.e. all the images have
exactly one corresponding pre-image, and onto, i.e. every image has a pre-image. See Figure
3(iv) for example.

Figure 3: Different types of functions. (i) into function, (ii) onto or surjective function, (iii)
one-one or injective function, (iv) one-one onto or bijective function
 7

Problems
1. Write the following sets using set builder notation:
(i) A = { 12 , 32 , 34 . . . }
(ii) A = { 21 , 23 , 34 }
(iii) A = {0, 0, 7, 26, 63. . . }
(iv) A = {0, 2, 4, 6. . . }
(v) A = a set of all possible integers whose cube is an odd integer
(vi) A = a set of all real numbers which cannot be written as quotients of two integers

2. Show that a null set is unique.

3. Prove that
(i) every set is a subset of itself
(ii) φ is a subset of all sets
(iii) number of subsets of a given set equals 2N where N is the cardinality of the set
(iv) number of proper subsets of a given set equals 2N -2

4. If U is the universal set then identify the complement of A if

A = {x : x ∈ N ∧ x = 3n ∀ n ∈ N}.

5. Prove that the cardinality of the power set of A, P (A), is 2|A| .

6. If A = {x : x = 4n + 1, n ∈ N, n ≤ 5} and
B = {x : x = 3n, n ∈ N, n ≤ 8} then determine A∆B.

7. Prove the following set operations:

(i) Idempotent laws:

(a) A ∪ A = A (b) A ∩ A = A
(ii) Identity laws:
(a) A ∪ φ = A (b) A ∩ U = A
(iii) Commutative laws:
(a) A ∪ B = B ∪ A (b) A ∩ B = B ∩ A
(iv) Associative laws:
8 CHAPTER 0. BACKGROUND: A REVIEW OF SET THEORY

(a) A ∪ (B ∪ C) = (A ∪ B) ∪ C (b) A ∩ (B ∩ C) = (A ∩ B) ∩ C
(v) Distributive laws:
(a) A ∪ (B ∩ C) = (A ∪ B) ∩ (A ∪ C)
(b) A ∩ (B ∪ C) = (A ∩ B) ∪ (A ∩ C)
(vi) De Morgan’s laws:
(a) (A ∪ B)c = Ac ∩ B c (b) (A ∩ B)c = Ac ∪ B c
(vii) Complement laws:
(a) A ∩ Ac = φ (b) A ∪ Ac = U
(c) φc = U (d) Uc = φ
(viii) Involution law:
(Ac )c = A

8. The Cartesian product A × A has 9 elements with two of the elements being (-1,0)
and (0,1). Identify the set A and determine A × A.

9. Prove that |A × B| = |A| · |B|.

10. Check whether A × B and B × A form equivalent sets.

11. For two sets A = {ai }, i = 1 · · · n and B = {bi }, i = 1 · · · n, n ∈ Z, verify that < A, B >
⊂ PP(A ∪ B).

12. If a and b are elements of sets A and B, respectively, then verify the followings:

(a) a ∪ b ⊂ (∪A) ∪ (∪B)

(b) < a, b >⊂ PP((∪A) ∪ (∪B))
(c) < x, y >, x ∈ A, y ∈ B ⊂ PPP((∪A) ∪ (∪B))

13. Prove the following set operations:

(a) A × (B ∪ C) = (A × B) ∪ (A × C)
(b) A × (B ∩ C) = (A × B) ∩ (A × C)
(c) A × (B − C) = (A × B) − (A × C)

14. State whether each diagram in the following figures defines a mapping from A = {a, b, c}
into B = {x, y, z}.
 9

15. Let the mappings f : A → B, g : B → C, h : C → D be defined by the following

diagram. Determine whether or not each function is (a) injective, (b) surjective, (c)
bijective, (d) invertible.

16. Let f : A → B and g : B → C be defined by the diagram.

(a) Find the composition mapping (g ◦ f ) : A → C.

(b) Find the images of the mappings f , g, g ◦ f .

17. Consider the mapping F : R3 → R2 defined by F (x, y, z) = (yz, x2 ). Find

(a) F (2, 3, 4); (b) F (5, −2, 7); (c) F −1 (0, 0) that is, all v ∈ R3 such that F (v) = 0.

18. Consider the mapping F : R2 → R2 defined by F (x, y) = (3y, 2x). Let S be the unit
circle in R2 , that is, the solution set of x2 + y 2 = 1. (a) Describe F (S). (b) Find
F −1 (S).
10 CHAPTER 0. BACKGROUND: A REVIEW OF SET THEORY

19. Let f : R → R be defined by f (x) = 2x − 3. If f is one-to-one and onto, find a formula

for f −1 .

20. Suppose f : A → B and g : B → C. Hence, g ◦ f : A → C exists. Prove that

(a) If f and g are one-to-one, then g ◦ f is one-to-one.
(b) If f and g are onto mappings, then g ◦ f is an onto mapping.
(c) If g ◦ f is one-to-one, then f is one-to-one mapping.
(d) If g ◦ f is an onto mapping, then g is an onto mapping.
Chapter 1

Groups, rings and fields

Binary operation:
Let S be any non-empty set. A function f : S × S → S is called a binary operation on the
set S. The binary operation f on the set S associates every ordered pair (a, b) ∈ S × S to a
unique element f (a, b) ∈ S.
From the above, it can be easily seen that addition is a binary operation on N. Similarly,
multiplication is a binary operation on N. But subtration and division are not binary oper-
ations on N.

Commutative binary operation:

A binary operation ⊗ on a set S is said to be commutative if

a ⊗ b = b ⊗ a ∀a, b ∈ S (1.1)

Hence, multiplication of numbers is a commutative binary operation while multiplication of

matrices is not.

Associative binary operation:

A binary operation ⊗ is said to be associative if

(a ⊗ b) ⊗ c = a ⊗ (b ⊗ c) ∀a, b, c ∈ S (1.2)

Hence, multiplication and addition of numbers and matrices are associative binary opera-
tions.

11
12 CHAPTER 1. GROUPS, RINGS AND FIELDS

Distributivity:
Let two operations ⊗ and  be defined on S. The binary operation ⊗ is said to be
(a) left distributive over  if

a ⊗ (b  c) = (a ⊗ b)  (a ⊗ c) ∀a, b, c ∈ S (1.3)

(b) right distributive over  if

(b  c) ⊗ a = (b ⊗ a)  (c ⊗ a) ∀a, b, c ∈ S (1.4)

The binary operation ⊗ is said to be distributive over  if it is both left as well as right
distributive.

Closure property:
Let ⊗ be a binary operation on S. For T ⊆ S, T is said to be closed under ⊗ if a ⊗ b ∈
T ∀a, b ∈ T .

Identity element:
Let ⊗ be a binary operation on S. An element e1 ∈ S is called a left identity if

e1 ⊗ a = a ∀a ∈ S (1.5)

An element e2 ∈ S is called a right identity if

a ⊗ e2 = a ∀a ∈ S (1.6)

An element e ∈ S is called an identity element if it is both left and right identity i.e.

e ⊗ a = a = a ⊗ e ∀a ∈ S (1.7)

Inverse of an element:
Let ⊗ be a binary operation on S. An element b1 ∈ S is called the left inverse of an element
a ∈ S if
b1 ⊗ a = e (1.8)
 13

where e is the identity element in S. An element b2 ∈ S is called the right inverse of an

element a ∈ S if
a ⊗ b2 = e (1.9)

An element b ∈ S is called an inverse element of a ∈ S if it is both left as well as right inverse

i.e.
b⊗a=e=a⊗b (1.10)

Groups:
A group is a non-empty set G equipped with a binary operation ⊗ : G × G → G that
associates an element a ⊗ b ∈ G ∀a, b ∈ G and having the following properties.

a ⊗ (b ⊗ c) = (a ⊗ b) ⊗ c (1.11)

a ⊗ e = e ⊗ a = a i.e. ∃e ∈ G (1.12)

∀a ∈ G∃a−1 ∈ G such that a ⊗ a−1 = a−1 ⊗ a = e (1.13)

The group is called abelian if it is commutative i.e.

a ⊗ b = b ⊗ a ∀a, b ∈ G (1.14)

A non-empty group M with a binary operation ⊗ : M × M → M is called a monoid if it

satisfies only the first two conditions for a group.
The identity element of a group is unique. To prove this, we assume there be two identity
elements of G.

eI ⊗ a = a ∀a ∈ G (1.15)

a ⊗ eII = a ∀a ∈ G (1.16)

eI , eII ∈ G. Therefore, for a = eII ,

eI ⊗ eII = eII (1.17)

Similarly, for a = eI ,
eI ⊗ eII = eI (1.18)

Hence, eI = eII . Therefore, a group has a unique identity element.

14 CHAPTER 1. GROUPS, RINGS AND FIELDS

Every element in a group has a unique inverse element. To prove this, we assume two inverse
elements of a ∈ G, aI , aII ∈ G.

aI ⊗ a = e (1.19)

a ⊗ aII = e (1.20)

(aI ⊗ a) ⊗ aII = e ⊗ aII = aII (1.21)

aI ⊗ (a ⊗ aII ) = aI ⊗ e = aI (1.22)

Since ⊗ is associative,

(aI ⊗ a) ⊗ aII = aI ⊗ (a ⊗ aII ) (1.23)

=⇒ aII = aI (1.24)

Subgroups:
Given a non-empty set G, a set H ⊆ G is called a subgroup of G iff

the identity element of e ∈ G also belongs to H (1.25)

h1 ⊗ h2 ∈ H ∀h1 , h2 ∈ H (1.26)

h−1 ∈ H ∀h ∈ H (1.27)

Rings:
A ring is a non-empty set R equipped with two binary operations

+ : R × R → R (addition) (1.28)

∗ : R × R → R (multiplication) (1.29)

having the following properties.

(a) R is an abelian group with respect to +
(b) * is associative and has an identity element 1 ∈ R
(c) * is distributive with respect to +
The identity element for addition is denoted by 0 (zero) and the additive inverse of a ∈ R is
 15

denoted by −a. ∀a, b, c ∈ R, the following properties hold.

(a + b) + c = a + (b + c) (1.30)

a+b=b+a (1.31)

a+0=0+a=a (1.32)

a + (−a) = (−a) + a = 0 (1.33)

a ∗ (b ∗ c) = (a ∗ b) ∗ c (1.34)

a∗1=1∗a=a (1.35)

(a + b) ∗ c = (a ∗ c) + (b ∗ c) (1.36)

a ∗ (b + c) = (a ∗ b) + (a ∗ c) (1.37)

Fields:
A set K is a field if it is a ring and the following properties hold.
(a) 06=1
(b) K ∗ = K − {0} is a group i.e. every a 6= 0 has an inverse with respect to *
(c) * is commutative
16 CHAPTER 1. GROUPS, RINGS AND FIELDS

Problems
1. Verify whether the set of integers Z is a group under addition? Identify whether
Z∗ = Z − {0} is a group under multiplication?

2. Verify whether the set Q of rational numbers is a group under addition? Identify
whether Q∗ = Q − {0} is a group under multiplication?

3. Verify whether a set of n × n invertible matrices is a group under (a) matrix addition
and (b) matrix multiplication?

4. Identify whether the following operations make a binary operation on the given sets.

(a) addition on N, Z, I+ , I− , Q, R, C
(b) subtration on N, Z, I+ , I− , Q, R, C
(c) multiplication on N, Z, I+ , I− , Q, R, C
(d) division on N, Z, I+ , I− , Q, R, C

5. For each of the above operations or combination of operations, determine whether the
operations are (a) commutative, (b) associative, (c) distributive.

6. For ⊗ : A × A → A, is A closed under ⊗?

7. Verify whether a universal set U is closed under every possible binary operation ⊗
defined on U ?

8. Verify whether N is a group under (a) addition, (b) multiplication? If yes, then check
if the group is abelian?

9. Verify whether N is a monoid under (a) addition, (b) multiplication?

10. Verify whether a set of non-singular matrices (n × n) forms a group under (a) matrix
addition, (b) matrix multiplication? Are the identity elements for the two operations
the same? Is the group abelian under both the operations?

11. Verify whether the additive groups Z, Q, R and C are commutative rings?
 17

12. Verify whether the rings Q, R and C are fields?

13. Verify whether the group R[X] of polynomials in one variable with real coefficients is
a ring under multiplication of polynomials? Check whether it is commutative?

14. Verify whether a group of n × n matrices is a ring under matrix multiplication? Check
whether it is commutative?

15. Verify whether a set of fractions of polynomials f (x)/g(x), f (x), g(x) ∈ R(x), g(x) 6= 0
is a field?
 
a b
16. Let G be a group with elements of the form   such that a, b, c, d ∈ {0, 1} and
c d
ad − bc 6= 0. Prove that G is a group under matrix multiplication with order 6.

17. If G is a group such that (ab)2 = a2 b2 ∀a, b ∈ G then prove that G is abelian.
18 CHAPTER 1. GROUPS, RINGS AND FIELDS
Chapter 2

Linear vector spaces

A vector space is an algebraic structure (V, F, ⊗, ) consisting of a non-empty set V , a field

F , a binary operation ⊗ : V × V → V and external mapping  : F × V → V associating
each a ∈ F, v ∈ V to a unique element a  v ∈ V and satisfying the following.
(i) (V, ⊗) is an abelian group
(ii) ∀u, v ∈ V, and a, b ∈ F

a  (u ⊗ v) = a  u ⊗ a  v (2.1)

(a ⊗ b)  v = (a  v) ⊗ (b  u) (2.2)

(a  b)  u = a  (b  u) (2.3)

1u=u (2.4)

The elements of V are called vectors and they will be denoted as ui from now onwards. The
elements of F are called scalars.

- The mapping  can be identified as scalar multiplication and the binary operation ⊗ can
be identified as vector addition.
- The vector space is called a real vector space if F = R and it is called a complex vector
space if F = C.
- Strictly speaking, the above properties define a linear vector space. But we will be using
the term vector space to signify the same thing.

19
20 CHAPTER 2. LINEAR VECTOR SPACES

Subspaces:
Let V be a vector space over a field F . A non-empty S ⊂ V is said to be a subspace of V if
S itself is a vector space over F under the operations on V restricted to S.
Restriction:
If S ⊂ V and a binary operation ⊗ is defined on S then ⊗ is said to be a restriction of S on
V if a ⊗ b ∈ S ∀a, b ∈ S.
Criteria for a subset to be a subspace:
Let V be a vector space over F . A non-empty set S ⊂ V is a subspace iff ∀u, v ∈ S and
∀α ∈ F

u+v∈S (2.5)

αu ∈ S (2.6)

Hence, S is an additive abelian group under vector addition and S is closed under scalar
multiplication. The above criteria can be alternatively posed as follows.
A non-empty S ⊂ V is a subspace of V iff αu + βv ∈ S, u, v ∈ S, α, β ∈ F .
 21

Problems
1. Show that a set of all ordred n-tuples of elements of any field F is a vector space over
F.

2. Show that a set of F m×n matrices over a field F is a vector space over F with respect
to the addition of matrices as the vector addition and multiplication of a matrix by a
scalar as the scalar multiplication.

3. Show that the set F [x] of all polynomials over a field F is a vector space over F .

4. Let V1 and V2 be two vector spaces over the same field F . Show that their Cartesian
product V1 × V2 = {(v1 , v2 ) : v1 ∈ V1 , v2 ∈ V2 } is a vector space over F .

5. Let F be a field and V be a set of all ordered pairs (a1 , a2 ), a1 , a2 ∈ F . The following
operations have been defined.

(a1 , a2 ) + (b1 , b2 ) = (a1 + b1 , a2 + b2 )

λ(a1 , a2 ) = (λa1 , a2 ), λ ∈ F

Verify whether the operations given above make V a vector space over F .

6. Show that a set V of all real valued continuous functions defined on a closed interval
[a, b] is a real vector space with vector addition and scalar multiplication defined as
follows.

(f + g)(x) = f (x) + g(x)

(λf )(x) = λf (x)

∀f, g ∈ V, λ ∈ R.

7. If X is a non-empty set and V is a vector space over F then show that

V X = {f : X → V } is a vector space over F under the vector addition and scalar
multiplication defined below.

(f + g)(x) = f (x) + g(x)

(λf )(x) = λf (x)

22 CHAPTER 2. LINEAR VECTOR SPACES

∀f, g ∈ V X , λ ∈ F .

8. Let V = {(x, y) : x, y ∈ R}. Show that V is not a vector space under the vector
addition and scalar multiplication defined below.

(ai , bi ) + (aj , bj ) = (3bi + 3bj , −ai − aj )

k(ai , bi ) = (3kbi , −kai )

∀(ai , bi ), (aj , bj ) ∈ V, k ∈ R.

9. Let V = {(x, y) : x, y ∈ R}. For any α = (xi , yi ), β = (xj , yj ) ∈ V, c ∈ R, the following

operations are defined.

α ⊗ β = (xi + xj + 1, yi + yj + 1)

c  α = (cxi , cyi )

(i) Prove that (V, ⊗) is an abelian group.

(ii) Prove that V is not a vector space over R under the two operations.

10. For any u = (xi , xj , xk ) and v = (yi , yj , yk ), u, v ∈ R3 , a ∈ R,

u ⊗ v = (xi + yi + 1, xj + yj + 1, xk + yk + 1)

a  u = (axi + a − 1, axj + a = 1, axk + a − 1)

Prove that R3 is a vector space over R under these two operations.

11. If V = R+ , u ⊗ v = uv, a  u = ua ∀u, v ∈ V, a ∈ R then prove that V is a vector space

over R under these two operations.

12. Identify whether R3 is a vector space over R under the following two operations.

(xi , xj , xk ) + (yi , yj , yk ) = (xi + yi , xj + yj , xk + yk )

a(xi , xj , xk ) = (axi , axj , axk )

∀(xi , xj , xk ), (yi , yj , yk ) ∈ R3 , a ∈ R.
 23

13. Let V = {(x, 1) : x ∈ R}. For any u = (x, 1), v = (y, 1) ∈ V, a ∈ R,

u ⊗ v = (x + y, 1)

a  u = (ax, 1)

Verify whether V is a vector space over R under these two operations.

14. Let V be a set of ordered pairs (a, b), a, b ∈ R. The following operations are defined.

(a, b) ⊗ (c, d) = (a + c, b + d)

k(a, b) = (ka, 0)

Show that V is not a vector space over R under these two operations.

15. Let V be the set of ordered pairs (a, b), a, b ∈ R and k ∈ R. Show that V is not a
vector space over R with the vector addition and scalar multiplication defined by:

(i)

(a, b) + (c, d) = (a + d, b + c)

k(a, b) =(ka, kb)

(ii)

(a, b) + (c, d) = (a + d, b + c)

k(a, b) =(a, b)

(iii)

(a, b) + (c, d) = (0, 0)

k(a, b) =(ka, kb)

(iv)

(a, b) + (c, d) = (ac, bd)

k(a, b) =(ka, kb)

24 CHAPTER 2. LINEAR VECTOR SPACES

16. Show that S = {(0, b, c) : b, c ∈ R} is a subspace of R3 .

17. If V = {(x, y, z) : x, y, z ∈ R} and S = {(x, x, x) : x ∈ R} then show that S is a

subspace of V . What is the geometrical interpretation of this?

18. Let ai , aj , ak ∈ F . Show that the set S of all triads (xi , xj , xk ) of elements of
F : ai xi + aj xj + ak xk = 0 is a subspace of F 3 . What is the geometrical interpretation
of this when F = R.

19. Show that a set S of all n × n symmetric matrices over a field F is a subspace of the
vector space F n×n matrices over F .

20. Let V be the vector space of all real valued continuous functions over R. Show that
the set S of solutions of the differential equation

d2 y dy
2 2
− 9 + 2y = 0
dx dx

is a subspace of V .

21. Let R be a field of real numbers and S be the set of all solutions of the equation
x + y + 2z = 0. Show that S is a subspace of R3 .

22. Let S be the set of all elements of the form (x + 2y, y, −x + 3y) ∈ R3 , x, y ∈ R. Show
that S is a subspace of R3 .

23. Let V be the vector space of all 2 × 2 matrices over R. Show that
(i) the set of all 2 × 2 singular matrices over R is not a subspace of V .
(ii) the set of all 2 × 2 matrices satisfying A × A = A, A ∈ V is not a subspace of A.

24. Let V be a vector space over R3 . Which of the following subsets of V are subspaces of
V?
(i) S1 = {(a, b, c) : a + b = 0}
(ii) S2 = {(a, b, c) : a = 2b + 1}
(iii) S3 = {(a, b, c) : a ≥ 0}
(iv) S4 = {(a, b, c) : a2 = b2 }
 25

(v) S5 = {(a, 2b, 3c) : a, b, c ∈ R}

(vi) S6 = {(a, a, a) : a ∈ R}
(vii) S7 = {(a, b, c) : a, b, c ∈ Q}

25. Let V = {a0 + a1 x + a2 x2 + a3 x3 : ai ∈ R} be the set of all polynomials of degree less

than or equal to 3 over R. V is a vector space over R. Prove that
(i) S1 = {a0 + a2 x2 : a0 , a2 ∈ R} is a subspace of V .
(ii) S2 = {a0 + a1 x + a2 x2 + a3 x3 : a0 = a2 + a3 , ai ∈ R} is a subspace of V .

26. Let V be the set of all 2 × 3 matrices. V is a vector space over R. Determine which of
the following subsets of V are subspaces of V ?
(i)
(  )
a b c
S1 =   : a, b, c, d ∈ R
d 0 0
(ii)
(  )
a b c
S2 =   : a + c = e + f, a, b, c, d, e, f ∈ R
d e f
(iii)
(  )
a b c
S3 =   : a > 0, b = c, a, b, c, d, e, f ∈ R
d e f
26 CHAPTER 2. LINEAR VECTOR SPACES
Chapter 3

Linear independence, basis, dimension

and span

Linear combination:
If V is a vector space over a field F , v1 , v2 , . . . , vn are n vectors in V and λ1 , λ2 , . . . λn are
scalars in F then
n
X
vn+1 = λi vi (3.1)
i=1

is called a linear combination of vi , vn+1 ∈ V .

Linear span:
Let V be a vector space over a field F and S ⊂ V . A set of all possible finite linear
combinations of vectors in S in called the linear span of S.
( n
)
X
[S] = λi vi : λi ∈ F, n ∈ N, vi ∈ S (3.2)
i=1

Linear independence:
A set of vectors v1 , v2 , . . . , vn in a vector space V over F are said to be linearly independent
if
n
X
λi vi = 0 =⇒ λi = 0∀i, λi ∈ F (3.3)
i=1

27
28 CHAPTER 3. LINEAR INDEPENDENCE, BASIS, DIMENSION AND SPAN

Basis and dimension: A non-empty subset B ⊂ V is said to be a basis for a vector space V
if
(i) B spans V
(ii) all vectors in B are linearly independent.

A vector space is said to be finite dimensional if there exists a finite subset of V that
spans it. The followings are noteworthy.
(i) A null vector cannot be in the basis.
(ii) If the number of vectors in the basis is not finite then the vector space is called infinite
dimensional.

If the number of vectors present in the basis are n then the vector space is said to be n-
dimensional. If V (F ) is a vector space over F then for dimension n,
(i) any set of n + 1 or more vectors in V are linearly dependent.
(ii) no set of n − 1 or lesser vectors can span V .
 29

Problems
1. Express the following vectors in R3 as a linear combination of v1 , v2 , v3 .
(i) w1 = [1 −2 5]T ; (ii) w2 = [1 1 1]T ; (iii) w3 = [0 0 0]T

v1 = [1 1 1]T ; v2 = [1 2 3]T ; v3 = [2 −1 1]T

2. Express the polynomial f (x) as a linear combination of the polynomial functions

φ1 (x), φ2 (x), φ3 (x).

f (x) = x2 + 4x − 3

φ1 (x) = x2 − 2x + 5

φ2 (x) = 2x2 − 3x

φ3 (x) = x + 3

3. Let V = R2×2 be a vector space of all 2 × 2 matrices over R. Express M as a linear

combination of M 1 , M 2 , M 3 .
       
4 7 1 1 1 2 1 1
M = ; M1 =  ; M2 =  ; M3 =  
7 9 1 1 3 4 4 5

4. Consider v1 = [1 2 3]T and v2 = [2 3 1]T . Find conditions on a, b and c so that

v = [a b c]T is a linear combination of v1 and v2 .

5. Let V = P2 (t) be the vector space of all polynomials of degree lesser than or equal to
2. Express the polynomial f (t) = at2 + bt + c as a linear combination of the following
polynomials.

f1 (t) = (t − 1)2 ; f2 (t) = t − 1; f3 (t) = 1

6. Express the vector u = [2 4 5]T in R3 as a linear combination of v1 = [1 2 3]T

and v2 = [2 3 1]T .

7. Show that the vectors v1 = [1 1]T and v2 = [1 2]T span R2 over R.

30 CHAPTER 3. LINEAR INDEPENDENCE, BASIS, DIMENSION AND SPAN

8. Show that the polynomials 1, 1 + x and (1 + x)2 span the vector space V = P2 (x) of
all polynomials of degree at most 2 over R.

9. Show that v1 , v2 and v3 span R3 where v1 = [1 1 1]T , v2 = [1 2 3]T and

v3 = [1 5 8]T .

10. Determine the condition(s) on a, b and c so that v = [a b c]T in R3 over R belongs to

the subspace spanned by v1 = [1 2 0]T , v2 = [−1 1 2]T and v3 = [3 0 −4]T .

11. Let v1 = [−1 2 0]T , v2 = [3 2 − 1]T and v3 = [1 6 − 1]T be three vectors

in R3 over R. Show that [v1 v2 ] = [v1 v2 v3 ].

12. Let v1 = [1 2 − 1]T , v2 = [2 −3 2]T , v3 = [4 1 3]T and v4 = [−3 1 2]T

be four vectors in R3 over R. Show that [v1 v2 ] 6= [v3 v4 ].

13. Let R2×2 be the vector space of all 2 × 2 matrices. Show that the matrices
       
1 0 0 1 0 0 0 0
M1 =  ; M2 =  ; M3 =  ; M4 =  
0 0 0 0 1 0 0 1

span R2×2 .

14. Verify whether the following vectors span R4 .

v1 = [3 −1 0 − 1]T ; v2 = [2 −1 3 2]T
v3 = [−1 1 1 3]T ; v4 = [1 1 9 − 5]T

15. Find one vector in R3 over R that spans the intersection of subspaces of S and T where
S = {(a, b, 0) : a, b ∈ R} and T = [u1 , u2 ], u1 = [1 1 1]T , u2 = [1 2 3]T .

16. Consider the vector space V = Pn (t) consiting of all polynomials of degree ≤ n. Show
that the set of polynomials 1, t, t2 , t3 , . . . , tn span V .

17. Show that the following vectors are linearly independent in R4 over R.

v1 = [1 1 2 4]T ; v2 = [2 −1 −5 2]T
v3 = [2 1 1 6]T ; v4 = [1 −1 −4 0]T
 31

18. Determine whether the vectors

f (x) = 2x3 + x2 + x + 1

g(x) = x3 + 3x2 + x − 2

h(x) = x3 + 2x2 − x + 3

in the vector space R[x] over R are linearly independent?

19. Let V be the vector space of functions from R into R. Show that the functions f (t) =
sin(t), g(t) = et , h(t) = t2 are linearly independent in V .

20. Verify if the following matrices in R2×2 over R are linearly independent.
     
1 1 1 0 1 1
M1 =  ; M2 =  ; M3 =  
1 1 0 1 0 0

21. Which of the following subsets Si of R3 are linearly independent?

(i) S1 = {[1 2 1]T , [−1 1 0]T , [5 −1 1]T }
(ii) S1 = {[1 1 0]T , [0 0 1]T , [1 5 2]T }
(iii) S1 = {[1 3 2]T , [1 −7 8]T , [2 1 1]T }
(iv) S1 = {[1 5 2]T , [1 0 0]T , [0 1 0]T }

22. Which of the following subsets Si of R[x] are linearly independent?

(i) S1 = {1, x − x2 , x + x2 , 3x}
(ii) S2 = {x2 − 1, x2 + 1, x − 1}
(iii) S3 = {x, x − x3 , x2 + x4 , a + x2 + x4 + 21 }
(iv) S4 = {1, 1 + x, 1 + x + x2 , x4 }

23. Which of the following subsets of all continuous function space are linearly indepen-
dent?
(i) S1 = {sin(x), cos(x), sin(x + 1)}
(ii) S2 = {xex , x2 ex , (x2 + x − 1)ex }
(iii) S3 = {sin2 (x), cos(2x), 1}
(iv) S4 = {x, sin(x), cos(x)}
32 CHAPTER 3. LINEAR INDEPENDENCE, BASIS, DIMENSION AND SPAN

24. If the set {v1 , v2 , v3 } is linearly independent in a vector space V (F ) then prove that
the set {v1 + v2 , v2 + v3 , v3 + v1 } is also linearly independent.

25. If {v1 , v2 , . . . , vn } is a linearly independent set of vectors in a vector space V (F ) and

{v1 , v2 , . . . , vn , u} is a linearly dependent set then prove that u is a linear combination
of v1 , v2 , . . . , vn .

26. Find a maximal linearly independent subsystem of a system having the following vec-
tors.
v1 = [2 −2 − 4]T ; v2 = [1 9 3]T
v3 = [−2 −4 1]T ; v4 = [3 7 1]T

27. Determine whether or not each of the following sets forms a basis of R3 (R).
(i) B1 = {[1 1 1]T , [1 0 1]T }
(ii) B2 = {[1 1 1]T , [1 2 3]T , [2 −1 1]T }
(iii) B3 = {[1 1 2]T , [1 2 5]T , [5 3 4]T }
(iv) B4 = {[1 2 3]T , [1 3 5]T , [1 0 1]T , [2 3 0]T }

28. Let v1 = [1 i 0]T , v2 = [2i 1 1]T and v3 = [0 1+i 1 − i]T be three vectors
in C3 (C). Show that the set B = {v1 , v2 , v3 } is a basis of C3 (C).

29. Determine whether {[1 1 1 1]T , [1 2 3 2]T , [2 5 6 4]T , [2 6 8 5]T }

form a basis of R4 (R)? If not then determine the dimension of the subspace they span.

30. Extend the set {[1 1 1 1]T , [2 2 3 4]T } to a basis in R4 (R).

31. Let S be the set containing the following vectors in R5 (R).

{[1 2 1 3 2]T , [1 3 3 5 3]T , [3 8 7 13 8]T ,
[1 4 6 9 7]T , [5 13 13 25 19]T
Find a basis of [S] consisting of the originally given vectors. Also determine the
dimension of [S].

32. Let S be the set consisting of the following vectors in R4 .

[1 −2 5 3]T , [2 3 1 − 4]T , [3 8 −3 − 5]T
 33

(i) Determine a basis and dimension of the subspace spanned by S.

(ii) Extend the basis of [S] to a basis in R4 .
(iii) Find a basis of [S] consisting of the original vectors.

33. Let S = {u1 , u2 , u3 } and T = {v1 , v2 , v3 } be subsets of R3 where

u1 = [1 1 − 1]T ; u2 = [2 3 − 1]T ; u3 = [3 1 − 5]T
v1 = [1 −1 − 3]T ; v2 = [3 −2 − 8]T ; v3 = [2 1 − 3]T
Show that [S] = [T ].

34. Which of the following subsets Bi form a basis for the given vector space V .
(i) B1 = {[1 0]T , [i 0]T , [0 1]T }; V = C2 [R]
(ii) B2 = {[1 i i + 1]T , [1 i i − 1]T , [i −i 1]T }; V = C3 [C]
(iii) B3 = {1, sin(x), sin2 (x), cos2 (x)}; V = C3 [−π, π]

35. Show that the vectors v1 = [1 0 − 1]T , v2 = [1 2 1]T , v3 = [0 −3 2]T

form a basis in R3 . Express each of these vectors as a linear combination of the standard
basis in R3 .
34 CHAPTER 3. LINEAR INDEPENDENCE, BASIS, DIMENSION AND SPAN
Chapter 4

Analysis of systems of simultaneous

linear equations

Consider the following system of simultaneous linear equations.

a11 x1 + a12 x2 + · · · + a1n xn = b1

a21 x1 + a22 x2 + · · · + a2n xn = b2

..
.

an1 x1 + an2 x2 + · · · + ann xn = bn

The above set of equations can be cast as a matrix equation given as follows.
    
a a ... a1n x b
 11 12   1  1
    
 a21 a22 . . . a2n   x2   b2 
 .. ..   ..  =  ..  (4.1)
    
 . .  .   . 
    
an1 an2 . . . ann xn bn

4.1 can be written as A x = b where A is the coefficient matrix, x is the vector of unknowns
to be solved for, and b is the vector consisting of the entries of the RHS.

One of the methods to solve such a system of equations is the Gauss elimination which
involves converting A into its echelon form followed by back substitution. However, here we

35
36CHAPTER 4. ANALYSIS OF SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS

would be interested in the analysis of the nature of the solutions without actually having to
solve the system. We would be interested in answering the following questions.
(i) Does the solution exist?
(ii) If yes, then is the solution unique?
For this, we will make use of the concept of null space and range space. Let us first consider
the homogeneous case.

A x = 0 (zero vector) (4.2)

The first task is to know whether the null space of A is empty or it has elements in it. Null
space is a space which consists of the vectors which satisfy A x = 0 i.e. it is the space of
solutions of the homogeneous case. If the null space is empty then the only solution is a
trivial solution, x = [0 0 . . . 0]T .
If the null space is non-empty then it is required to determine the dimension and a basis of
the null space. All linear combinations of the basis in the null space satisfy A x = 0.
How to check if the null space is empty? In such a case, the only solution to the homoge-
neous system is a trivial solution. For an n × n system, if det(A)n×n = 0 then the only
solution is the trivial solution. In other words, the homogeneous system will have only a
trivial solution if the rank ρ(A) = n = number of unknowns. If this is not the case then
we need to determine a basis of the null space of A. This can be determined following the
procedure given below.

Consider a system of three simultaneous equations given below.

x1 + x2 + x3 = 0

2x1 + 3x2 + x3 = 0

5x1 + 6x2 + 4x3 = 0

    
1 1 1 x1 0
    
2 3 1 x2  = 0
    
    
5 6 4 x3 0
 37

R2 → R2 − 2R1
R3 → R3 − 5R1     
1 1 1 x 0
   1  
0 1 −1 x2  = 0
    
    
0 1 −1 x3 0
R3 → R3 − R2     
1 1 1 x 0
   1  
0 1 −1 x2  = 0
    
    
0 0 0 x3 0
No further operation will further reduce R2 to a zero row. Therefore, we stop here and recast
the matrix equation back to a system of simultaneous equations.

x1 + x2 + x 3 = 0

x2 − x 3 = 0

We have two equations in three unknowns. Hence we have a degree of freedom of one. From
the second equation, we assume x2 = α. Hence x2 = x3 = α and x1 = −2α. The solution to
the homogeneous equations can, hence, be written as
   
x1 −2
   
 x2  = α  1 
   
   
x3 1

Hence, the dimension of the null space of A is 1 since there is only one vector in its basis
and a possible basis is [−2 1 1]T . Every vector which is obtained by multiplying α to this
vector will be a solution to the homogeneous equation. It can be seen that this vector or any
multiple of it satisfy the original system of homogeneous equations. It can be noted that we
had three unknowns (x1 , x2 , x3 ) and two equations after row reduction. Hence the degree of
freedom of the system was one. This resulted in the appearance of one parameter α. The
dimension of the null space is also one. Hence, the dimension of the null space will be equal
to the number of parameters or the degree of freedom of the row reduced system.
A system of homogeneous equations always has a solution since a trivial solution always
exists for this case. Hence, for a homogeneous case, we have the following answers.
38CHAPTER 4. ANALYSIS OF SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS

(i) Does the solution exist? → Yes, always

(ii) Is the solution unique? → The only possible unique solution in case of a homogeneous
set of equations is a trivial solution. If the solution is not trivial then there are infinitely
many solutions.
Now we focus on the non-homogeneous case.

A x = b; b 6= 0

If A x = 0 has only a trivial solution then A x = b has a unique solution. To prove this, let
us assume that A x = b has two distinct solutions v1 and v2 .

=⇒ A v1 = b

A v2 = b

=⇒ A (v1 − v2 ) = 0

Hence, v1 − v2 must be the solution to the homogeneous case. But the only solution to the
homogeneous case A x = 0 has been given as the trivial solution. Hence v1 − v2 = 0 or
v1 = v2 which means that the solution to A x = b must be unique.
From the foregoing discussion, it can be concluded that to know the nature of solutions of
A x=b, we must first solve for A x=0. If the only solution to A x=0 is a trivial solution then
A x=b must have a unique solution. If the solution to A x=0 is not unique then A x=b may
or may not be solvable. If it is solvable then it will have infinitely many solutions.
Now we focus on the question of determining whether A x=b has a solution or not.
A x=b will have a solution if ρ(A) = ρ(A|b). We make use of this fact to develop the range
space of A. Range space of A contains all the elements which satisfy A x=b. Consider the
previous case with b = [1 2 1]T . The corresponding matrix equation is given below.
    
1 1 1 x1 1
    
3 1 x2  = 2
    
2
    
5 6 4 x3 1
 
1 1 1 b1
 
A|b = 2 3 1 b2 
 
 
5 6 4 b3
 39

R2 → R2 − 2R1
R3 → R3 − 5R1  
1 1 1 b1
 
A|b = 0 1 −1 b2 − 2b1 
 
 
0 1 −1 b3 − 5b1
R3 → R3 − R2  
1 1 1 b1
 
A|b = 0 1 −1 b2 − 2b1 
 
 
0 1 −1 b3 − b2 − 3b1
It can be seen from the above rearrangements that ρ(A) = 2. The condition of solvability
dictates that b3 − b2 − 3b1 = 0. Note that despite the entries known in b, we continue with
bi ’s rather than the actual values given in b. We now do the degree of freedom analysis.
Number of unknowns = 3 (b1 , b2 , b3 )
Number of equations = 1 (b3 − b2 − 3b1 = 0)
Hence, we have two arbitarary parameters. Let b1 = α and b2 = β =⇒ b3 = 3α + β.
     
b 1 0
 1    
b2  = α 0 + β 1
     
     
b3 3 1

Therefore, the dimension of the range space of A = 2 and a possible basis is v1 = [1 0 3]T
and v2 = [0 1 1]T . All possible linear combinations of v1 and v2 satisfy A x=b. Any b
proposed must be expressed as a linear combination of v1 and v2 for the solution to exist.
Therefore, as given originally in the problem,
     
1 1 0
     
b = 2 = α 0 + β 1
     
     
1 3 1

=⇒ α = 1

β=2

3α + β = 1
40CHAPTER 4. ANALYSIS OF SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS

No values of α and β satisfy the above set of equations. Hence, the solution does not exist
for the system with b = [1 2 1]T .
 41

Problems
1. Determine whether the following set of simultaneous equations have a solution. If yes,
then comment upon the nature of solutions.
(i)

2u + 3v = 0

4u + 5v + w = 0

2u − 5v − 3w = 0

(ii)

x+y =0

x + 2y + z = 0

y + 2z + t = 0

z + 2t = 5

(iii)

2u − v = 0

−u + 2v − w = 0

−v + 2w − z = 0

−w + 2z = 5

(iv)

x1 + x2 + x3 = 6

x1 + 2x2 + 2x3 = 11

2x1 + 3x2 − 4x3 = 3

(v)

x1 + x2 + x3 = −2

3x1 + 3x2 − x3 = 6

x1 − x2 + x3 = −1
42CHAPTER 4. ANALYSIS OF SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS

2. Determine the value of ‘c’ which makes it possible to solve A x = b for

x1 + x2 + 2x3 = 2

2x1 + 3x2 − x3 = 5

3x1 + 4x2 + x3 = c

3. Consider the system of equations given below.

x1 + 2x2 + 3x3 + 5x4 = b1

2x1 + 4x2 + 8x3 + 12x4 = b2

3x1 + 6x2 + 7x3 + 13x4 = b3

(i) Determine the dimension and basis for the null space of the coefficient matrix.
(ii) Determine the dimension and basis for the range space.
(c) Find the conditions on b1 , b2 , b3 to have a solution.

4. For each of the following system of equations, find the conditions on bi ’s so as to have
a solution for the system.    
1 2 b
     1
   
2 4  x1 b 
    =  2
   
2 5  x2 b3 
   
3 9 b4

5. Construct matrices subject to the following conditions.

(i) The null space has a basis of [2 2 1 0]T and [3 1 0 1]T .
(ii) The column space has a basis of [1 1 1]T and the null space has a basis of
[2 2 1 0]T .
(iii) The null space has a basis of [4 3 2 1]T .
(iv) The column space has a basis of [1 1 0]T and [0 1 1]T and the null space has a
basis of [1 0 1 0]T and [0 0 1 1]T .
(v) The column space has a basis of [1 1 5]T and [0 3 1]T and the null space has a
basis of [1 1 2]T .
 43

6. Determine the corresponding ranges space for the following systems of equations.
(i)    
1 2 3   b
  x  1
   1  
2 4 6    b2 
  x  =  
  2  
2 5 7    b3 
  x3  
3 9 12 b4
(ii)  
  x1  
1 3 1 2  
  b
  x   1 
  2  
2 6 4 8   = b2 

   x3   
0 0 2 4   b3
x4
(iii)  
  x1  
1 3 3 2   b1 

  x
 2   

2

6 9 7   = b 
2
  x3 

  
−1 −3 3 4   b3
x4
(iv)  
x
   1  
 
1 2 0 3  x2  b
    =  1
 
2 4 0 7  x3  b2
 
x4
(v)

x + 2y − 2z = b1

2x + 5y − 4z = b2

4x + 9y − 8z = b3

(vi)

x1 + 2x2 + 2x3 + 4x4 + 6x5 = b1

x1 + 2x2 + 3x3 + 6x4 + 9x5 = b2

x3 + 2x4 + 3x5 = b3
44CHAPTER 4. ANALYSIS OF SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS

7. Examine the solvability of the following system.

x1 + 2x2 = b1

2x1 + 4x2 = b2

x 1 = b3

8. Consider the following matrix.

 
1 −1 2
 
A = 2 −1 6
 
 
1 2 4
(i) Determine the dimension and a basis for the null space of A.
(ii) Determine the dimension and a basis for the range space.
(iii) Which of the following vectors b will yield a solution to A x = b?
         
3 1 6 1 2
         
 2  , −1 ,  5  , 0 ,  9 
         
         
−1 2 12 0 13

9. For the following set of equations, determine the dimension and a basis for the null
space and the range space.

x1 + 2x2 + x3 + 2x4 − 3x5 = 2

3x1 + 6x2 + 4x3 − x4 − 2x5 = −1

4x1 + 8x2 + 5x3 + x4 − x5 = 1

−2x1 − 4x2 − 3x3 + 4x4 − 5x5 = 3

10. Consider the following matrix.

 
1 −1 2
 
 
2 1 2
A=
 
4 −1 9

 
2 1 1
Determine the dimension and a basis for the null and range spaces. Determine the
solutions to A x = b for b = [3 0 0 2]T .
 45

11. For the following systems of equations, determine the dimension and bases for the null
and range spaces. Determine the solutions to the systems on the basis of your analysis.
(i)

x1 − x2 + 3x3 + 2x4 = b1

3x1 + x2 − x3 + x4 = b2

−x1 − 3x2 + 7x3 + 3x4 = b3

(ii)

x1 + 2x2 − x3 + x5 = b1

3x1 + 2x2 + x4 = b2

x1 − 2x2 + 2x3 + x4 − 2x5 = b3

(iii)

x 1 + x 2 − x 3 = b1

−2x1 − x2 + x3 = b2

x1 + 2x2 − 2x3 = b3

(vi)

5x1 + 10x2 + x3 − 2x4 = 6

−x1 + x2 − 2x3 + x4 = 0

2x1 + 3x2 + x3 − x4 = 2

6x1 + 9x2 + 3x3 − 3x4 = 6

46CHAPTER 4. ANALYSIS OF SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS
Chapter 5

Linear transformations

Till now we considered how to analyze vectors in a given vector space. It is now desired
to study the relationships among different vector spaces over the same field. Let V1 and
V2 be two vector spaces over the same field F . A mapping t : V1 → V2 is called a linear
transformation or linear map or homomorphism if
(i) t(u + v) = t(u) + t(v)
(ii) t(αu) = αt(u)
∀u, v ∈ V1 , α ∈ F . A linear transformation from a vector space to itself is called a linear
operator.

The above two conditions can be merged together to get a single condition as follows.
t : V1 → V2 is called a linear transformation if t(αu + βv) = αu + βv ∀u, v ∈ V1 , α, β ∈ F .

Some other properties satisfied by linear transformations are given below.

(i) t(0V 1 ) = 0V 2 where 0V 1 and 0V 2 are the zero vectors in V1 and V2 , respectively.
(ii) t(−u) = −t(u)
(iii) t(u − v) = t(u) − t(v)
∀u, v ∈ V1

If V1 and V2 are two vector spaces over the same field F and B = {b1 , b2 , . . . , bn } is a basis
in V1 then there exists a unique transformation t : V1 → V2 such that t(bi ) = b0i where b0i are
the vectors in V2 .

47
48 CHAPTER 5. LINEAR TRANSFORMATIONS

To prove the above, the following steps are adopted. Since B forms the basis in V1 ,
n
X
v= λi bi , v ∈ V1 (5.1)
i=1

n
!
X
t(v) = t λi vi
i=1
n
X
= (λi t(bi ))
i=1
n
X
= λi b0i
i=1
Pn
Since λi ’s need to be unique scalars, i=1 λi b0i shall also be unique. If u, v ∈ V1 and
α, β, λi , µi ∈ F ,
n
X
u= λi bi
i=1
n
X
v= µi b i
i=1
n
! n
!!
X X
t(αu + βv) = t α λi bi +β µi bi
i=1 i=1
n
!
X
=t (αλi + βµi )bi
i=1
n
X
= (αλi + βµi )b0i
i=1
n
X
= (α(λi b0i ) + β(µi b0i ))
i=1
n
! n
!
X X
=α λi b0i +β µi b0i
i=1 i=1

= αt(u) + βt(v)

=⇒ t(αu + βv) = αt(u) + βt(v)∀u, v ∈ V1 , α, β ∈ F

Hence, t : V1 → V2 is a linear transformation.

Till now we have shown that the transformation t : V1 → V2 is linear and the vectors t(v) ∈ V2
are unique linear combinations of the basis in V2 . Now we show that the transformation itself
is unique.
 49

Let t0 : V1 → V2 be another linear transformation such that t(bi ) = b0i ∀i ∈ N.

n
X
v= λi bi ∈ V1
i=1
n
X
0 0


t (v) = t λi bi
i=1
n
X
= λi t0 (bi )
i=1
n
X
= λi b0i
i=1

= t(v)

=⇒ t0 (v) = t(v)

Hence, t : V1 → V2 is unique.
If A is an m × n matrix and a mapping tA is defined such that tA : F n×1 → F m×1

tA (X) = A X ∀X = [x1 x2 . . . xn ]T ∈ F n×1

then ∀X, Y ∈ F n×1 and λ ∈ F , we have

tA (X + Y ) = A(X + Y )

=A X +A Y

= tA (X) + tA (Y )

=⇒ tA (X + Y ) = tA (X) + tA (Y )

Similarly, it can be easily shown that

tA (λX) = A(λX) = λ(A X) = λt(X)

Thus, tA : F n×1 → F m×1 is a linear transformation. Hence, every m × n matrix can be

viewed as a linear transformation from F n×1 → F m×1 .

Kernel and image of a linear transformation:

If V1 and V2 are the vector spaces over the same field F and t : V1 → V2 then kernel of t,
ker(t), is defined as
ker(t) = {v ∈ V1 : t(v) = 0V2 } (5.2)
50 CHAPTER 5. LINEAR TRANSFORMATIONS

It is easy to recognize that ker(t) is nothing but the null space of t. Image of t, Im(t), is
defined as
Im(t) = {t(v) : v ∈ V1 } (5.3)

It is easy to prove that

(i) ker(t) is a subspace of V1
(ii) Im(t) is a subspace of V2
The following conclusions follow the previous results.
- If {v1 , v2 . . . vn } spans V1 then {t(v1 ), t(v2 ) . . . t(vn )} spans V2 .
- If {v1 , v2 . . . vn } is a set of linearly independent vectors in V1 then {t(v1 ), t(v2 ) . . . t(vn )} is
also linearly independent in V2 .
- As seen previously, A(m × n) is a linear transformation tA : F n×1 → F m×1 and the kernel
of tA consists of all vectors x for which t(x) = 0. Hence, ker(t) is {x : A x = 0}.

Rank and nullity:

If V1 and V2 are the vector spaces over the same field F and t : V1 → V2 is a linear
transformation then

rank(t) = dim(Im(t)) (5.4)

nullity(t) = dim(ker(t)) (5.5)

Sylvester’s law:

dim(Im(t)) + dim(ker(t)) = dim(V ) (5.6)

rank(t) + nullity(t) = dim(V ) (5.7)

 51

Problems
1. Let F be a field. Show that the mapping t : F 2 → F 3 given by

t(a, b) = (a, b, 0) ∀a, b ∈ F

is a linear transformation.

2. Let t : R3 → R3 be a mapping defined by

t(x, y, z) = (x, y, 0) ∀x, y, z ∈ R

Show that t is a linear transformation.

3. Let V = R[x] be the vector space of all polynomials over field R and let D : V → V be
d
the mapping associating each polynomial f (x) to its derivative dx
(f (x)). Show that D
is a linear transformation.

4. Let V be a real vector space of all continuous functions from R into itself. Show that
the mapping T : V → V given by
Z x
T [f (x)] = f (t)dt ∀f (x) ∈ V, x ∈ R
0

is a linear transformation from V to itself.

5. Let C be the vector space of all complex numbers over the field of complex numbers
and let t : C → C be a mapping given by

t(x + iy) = x ∀x + iy ∈ C

Show that t is not a linear transformation.

6. Show that the mapping t : R2 → R2 given by

t(x, y) = (x + y, x) ∀x, y ∈ R

is a linear transformation.
52 CHAPTER 5. LINEAR TRANSFORMATIONS

7. Show that each of the following mappings is a linear transformation.

(i) t : R2 → R2 defined by

t(x, y) = (ax + by, cx + dy) ∀a, b, c, d ∈ R

(ii) t : R3 → R2 defined by

t(x, y, z) = (x + y + z, 2x − 3y + 4z)

(iii) t : R3 → R2 defined by

t(x, y, z) = (x + 2y − 3z, 4x − 5y + 6z)

8. Show that the following mappings are not linear transformations.

(i) t : R2 → R2 defined by
t(x, y) = (x + 1, y + 2)

(ii) t : R2 → R2 defined by
t(x, y) = (x2 , y 2 )

(iii) t : R3 → R2 defined by

t(x, y, z) = (x + 1, y + z)

(iv) t : R2 → R2 defined by
t(x, y) = (xy, y)

9. Let V = Rn×n be a vector space of n × n matrices and let M be a fixed non-null matrix
in V . Which of the followings is a linear transformation?
(i) t(A) = M A
(ii) t(A) = M A + A M
(ii) t(A) = A M − M A
(iv) t(A) = M + A
where t : V → V .
 53

10. Verify whether the operator L is linear where

(i)
∂ ∂2
L= −α 2 +β
∂t ∂z
(ii)
d2
L= 2
− m2
∂x

11. If L is a linear operator then show that Ln , n ∈ I+ is also a linear operator.

12. Write the nth order ordinary differential equation given below in operator form and
identify whether the operator is linear.

dn u dn−1 u
an + a n−1 + · · · + a0 = f (x)
dxx dxn−1

13. Let t : R2 → R2 be a linear transformation for which t(1, 2) = (2, 3) and t(0, 1) = (1, 4).
Find the formula for t and find t(x, y).

14. Let B = {(−1, 0, 1), (0, 1, −1), (1, −1, 1) be a basis for R3 (R) and t : R3 → R3 be a
linear transformation such that

t(−1, 0, 1) = (1, 0, 0)

t(0, 1, −1) = (0, 1, 0)

t(1, −1, 1) = (0, 0, 1)

Find the formula to compute t(1, −2, 3) from t(x, y, z).

15. Let t : R3 → R3 be such that t(1, 2, 3) = (1, 0, 0), t(1, 2, 0) = (0, 1, 0) and t(1, −1, 0) =
(0, 0, 1). Find t(a, b, c) ∀(a, b, c) ∈ R3 .

16. Let t : R4 → R3 be a linear transformation defined by

t(x1 , x2 , x3 , x4 ) = (x1 − x2 + x3 + x4 , x1 + 2x3 − x4 , x1 + x2 + 3x3 − 3x4 )

Find a basis and dimension of (i) image of t, (ii) kernel of t.

54 CHAPTER 5. LINEAR TRANSFORMATIONS

17. Let t : R3 → R3 be a linear transformation defined by

t(x, y, z) = (x + 2y − z, y + z, x + y − 2z)

Find a basis and dimension of (i) image of t, (ii) kernel of t.

18. Repeat the above for the followings.

(i) t : R3 → R3

t(x, y, z) = (x + 2y − 3z, 2x + 5y − 4z, x + 4y + z)

(ii) t : R4 → R3

t(x, y, z, u) = (x + 2y + 3z + 4u, 2x + 4y + 7z + 5u, x + 2y + 6z + 5u)

(iii) t : R3 → R2
t(x, y, z) = (x + y + z, 2x + 2y + 2z)

19. Let tA : R3×1 → R3×1 be a linear transformation given by tA (x) = A x where

 
1 1 2
 
A = 1 0 1
 
 
2 1 3

Determine
(i) basis for range of tA
(ii) basis for kernel of tA
(iii) rank and nullity of tA

20. Let tA : R4×1 → R3×1 be a linear transformation defined as

   
w   w
  2 1 −1 3  
   
x  x
tA   = 2 0 −1 5
     
 
y   y 
  3 0 −2 8  
z z

(i) Find the basis for the range of tA .

(ii) Find the basis for the kernel of tA .
 55

(iii) What are nullity(tA ) and rank(tA ).

(iv) Does [1 0 1 0]T ∈ ker(tA ).
(v) Does [3 5 8]T ∈ range(tA ).

21. Show that the mapping t : R2 → R3 given by t(a, b) = (a + b, a − b, b) ∀(a, b) ∈ R2 is a

linear transformation. Find the range, rank, kernel and nullity of t.

22. Let C be the field of complex numbers and let t : C3 → C3 be a mapping given by

t(a, b, c) = (a − b + 2c, 2a + b − c, −a − 2b)

Show that t is a linear transformation. Find its kernel.

23. Let t : R2 → R2 be a linear transformation given by t(a, b) = (2a − 3b, a + b) ∀(a, b) ∈

R2 . Find the matrix of t relative to the bases B = {b1 = (1, 0), b2 = (0, 1)} and
B 0 = {b01 = (2, 3), b02 = (1, 2)}.

24. Find the matrix of the linear transformation t : R3 → R3 given by

t(a, b, c) = (2b + c, a − 4b, 3a)

relative to the ordered bases given below.

B = {b1 = (1, 1, 1), b2 = (1, 1, 0), b3 = (1, 0, 0)}

25. B given above is an ordered basis for a linear transformation

t(x, y, z) = (3x + 2y − 4z, x − 5y + 3z)

Find the matrix of t relative to B.

26. The set B = {e3t , te3t , t2 e3t } is an ordered basis of the vector space V of all functions
f : R → R. Let D be the differential operator defined as d/dt. Find the matrix
representation of D relative to the basis B.

27. Let t : R2 → R2 be a linear transformation given by t(1, 1) = (3, 7) and t(1, 2) =

(5, −4). Find the matrix of t relative to the standard basis of R2 .
56 CHAPTER 5. LINEAR TRANSFORMATIONS

28. The matrix A on R defines a linear transformation tA : R3 → R3 by the rule tA (x) =

A x. Find the matrix representing tA relative to the basis B = {x1 , x2 , x3 } where
     
1 0 1
     
x1 = 1 , x2 = 1 , x3 = 2
     
     
1 1 3

29. Repeat question (26) with the following sets.

(i) B = {et , e2t , te2t }
(ii) B = {1, t, sin(3t), cos(3t)}
(iii) B = {e5t , αte5t , βt2 e5t }

30. Consider the following bases of R2 .

B = {b1 = (1, 0), b2 = (0, 1)}

C = {c1 = (1, 3), c2 = (1, 4)}

(i) Determine the change-of-basis matrix P from B to C.

(ii) Determine the change-of-basis matrix Q from C to B.

31. Repeat the above problem with

B = {b1 = (1, 0, 0), b2 = (0, 1, 0), b3 = (0, 0, 1)}

C = {c1 = (1, 0, 1), c2 = (2, 1, 2), c3 = (1, 2, 2)}

32. Consider the linear transformation t : R2 → R2 defined by t(x, y) = (5x − y, 2x + y)

and the following bases in R2 .

B = {b1 = (1, 0), b2 = (0, 1)}

C = {c1 = (1, 4), c2 = (2, 7)}

(i) Find the matrix P representing t relative to the basis B.

(ii) Find the matrix Q representing t relative to the basis C.
(iii) Find the change-of-basis matrix R from B to C.
(iv) Find the change-of-basis matrix S from C to B.
 57

33. Repeat the above problem with the linear transformation t : R3 → R3

t(x, y, z) = (3x + z, −2x + y, −x + 2y + 4z)

B = {b1 = (1, 0, 0), b2 = (0, 1, 0), b3 = (0, 0, 1)}

C = {c1 = (1, 0, 1), c2 = (−1, 2, 1), c3 = (2, 1, 1)}

58 CHAPTER 5. LINEAR TRANSFORMATIONS
Chapter 6

Inner product spaces

Let V be a real vector space. An inner product on V is a function <, >: V × V → R which
assigns each ordered pair (u, v) ∈ V × V to a real number < u, v > in such a way that the
following axioms are satisfied.
(i) Linearity:

< au1 + bu2 , v >= a <u1 , v > +b < u2 , v > (6.1)

∀u1 , u2 , v ∈ V, a, b ∈ R (6.2)

(ii) Symmetry:
< u, v >=< v, u > (6.3)

(iii) Positive definiteness:

< v, v > ≥ 0 ∀v ∈ V (6.4)

< v, v > = 0 iff v = 0 (6.5)

A vector space equipped with an inner product is called an inner product space. The above
conditions have been laid down for real vector spaces. As an extension, the most generalized
conditions can be written as given below.
(i) Linearity:

< u1 + u2 , v > =< u1 , v > + < u2 , v > (6.6)

< αu, v > = α < u, v > (6.7)

59
60 CHAPTER 6. INNER PRODUCT SPACES

(ii) Symmetry:
< u, v >= < v, u > (6.8)

(iii) Positive definiteness:

< v, v > ≥ 0 ∀v ∈ V (6.9)

< v, v > = 0 iff v = 0 (6.10)

where overbar represents the complex conjugate. From linearity and symmetry, the following
property follows.
< u, αv >= α < u, v > (6.11)

There can be several functions which can satisfy the conditions given above for an inner
product. A standard inner product is defined as follows.
n
X
< u, v > = ui v i (6.12)
i=1
Z b
< f, g > = f (x)g(x)dx (6.13)
a

Two vectors u and v are said to be orthogonal if their inner product is zero.

< u, v >= 0 (6.14)

A collection of vectors is said to be an orthogonal set if

< vi , vj >= 0 ∀i 6= j (6.15)

The above set is said to be orthonormal if


0, i 6= j
< vi , vj >= δij = (6.16)
1, i = j
 61

Problems
1. Consider the vector space Rn . Prove that Rn is an inner product space with the inner
product defined by
< u, v >= u1 v1 + u2 v2 + · · · + un vn

where u = (u1 , u2 , . . . , un ) and v = (v1 , v2 , . . . , vn ).

2. Consider the vector space C[a, b] of all continuous functions defined on the interval
[a, b]. Prove that the following operation makes it an inner product space.
Z b
< f, g >= f (t)g(t)dt, f (t), g(t) ∈ C[a, b]
a

3. Identify whether R2 is an inner product space with an inner product defined by

< u, v >= a1 b1 −a2 b1 − a1 b2 + 2a2 b2

u = (a1 , a2 ), v = (b1 , b2 ) ∈ R2

4. Let V be a real vector space. Show that the sum of two inner products on V is an
inner product on V . Is the same true for the difference of two inner product? What
about positive multiples of an inner product?

5. Find the value of k so that the following is an inner product on R2 .

< u, v >= a1 b1 −3a1 b2 − 3a2 b1 + ka2 b2

u = (a1 , a2 ), v = (b1 , b2 ) ∈ R2

6. Let <, > be the standard inner product on R2 . If u = [1 2]T and v = [−1 1]T ∈ R2
then find w ∈ R2 satisfying

< v, w >= 3 and < u, w >= −1

7. For each of the followings, determine whether the operation <, > makes the space an
inner product space.
(i) < u, v >= u1 v1 − 2u1 v2 − 2u2 v1 + 5u2 v2
62 CHAPTER 6. INNER PRODUCT SPACES

(ii) < u, v >= u1 v1 − u1 v2 − u2 v1 + 3u2 v2

(iii) < u, v >= u21 − 2u1 v2 − 2u2 v1 + v12
(iv) < u, v >= 2u1 v1 + 5u2 v2
(v) < u, v >= u1 v2 u3 + u2 v1 u3

8. Determine the value of λ so that the following is an inner product on R2 .

< u, v >= u1 v1 + 2u1 v2 + 2u2 v1 + λu2 v2

9. Let <, > be the standard inner product on R2 . If u = [1 3]T and v = [2 1]T ∈ R2
such that
< w, u >= 3 < w, v >= −1

then determine w.

10. For any u = [u1 u2 ]T and v = [v1 v2 ]T ∈ R2 , the following operation is defined.
  
h i 1 2 u
< u, v >= u1 v1    2
3 8 v2

Determine whether R2 is an inner product space with the inner product defined by the
above operation.

11. If f (x) = x and g(x) = e−ix then show that < f (x), g(x) >= < g(x), f (x) >.

12. Let A = [aij ] be a 2×2 matrix with real entries. For x, y ∈ R2×1 , let fA (x, y) = yT A x.
Show that fA is an inner product space on R2×1 iff AT = A, a11 > 0, a22 > 0 and |A| >
0.

13. Let C[−π, π] be the inner product space of all continuous functions defined on [−π, π]
with the inner product as the standard inner product. Verify whether sin(t) and cos(t)
are orthogonal in such a space.

14. Determine a non-zero vector in R3 that is orthogonal to the vectors v1 = [1 1 2]T ,

v2 = [2 1 3]T and v3 = [1 2 3]T .
 63

15. Let S be the set consisting of vectors v1 = [1 2 1]T , v2 = [2 1 − 4]T and v3 =

[3 − 2 1]T with standard inner product defined on R3 . Verify whether S is orthogonal
and a basis of R3 .

16. Let V be an inner product space and S = {v1 , v2 . . . , vn } be an orthogonal set of

vectors in V . Show that {λv1 , λv2 . . . , λvn } is also an orthogonal set for any scalar λ.

17. Let V be the vector space of all polynomials over R of degree ≤ 2 with the inner
R1
product defined as < f, g >= 0 f (t)g(t)dt. Find a basis of the subspace orthogonal
to the polynomial φ(t) = 2t + 1.

18. Consider the inner product space R4 with standard inner product defined. Let four
vectors in R4 be as given below.

v1 = [1 1 0 − 1]T , v2 = [1 2 1 3]T

v2 = [1 1 −9 2]T , v4 = [16 − 13 1 3]T

(i) Do the above vectors form a basis in R4 ?

(ii) Do they form an orthogonal basis?
(iii) Do they form an orthonormal basis?
(iv) Express an arbitarary vector v = [a b c d]T in terms of the above vectors.

19. Let f (x) belong to a space of continuous functions with the standard inner product
defined. If f (x) is expressed as an infinite series as follows.
∞
X
f (x) = an sin(nπx)
n=1

then derive an expression for the coefficients of the expansion.

20. Determine whether the following functions

φn (x) = exp(2πinx), n ∈ I, 0 ≤ x ≤ 1

form an orthonormal set.

64 CHAPTER 6. INNER PRODUCT SPACES

21. Consider a piecewise continuous function f (x) defined on the interval [−c, c] with a
period 2c. The function is to be represented as
∞  
a0 X nπx nπx
f (x) = + an cos + bn sin
2 n=1
c c

Determine the expressions for an and bn .

22. Consider the following function.

∞
−αn2 π 2 t
   
X nπx
T (x, t) = an exp 2
sin
n=1
c c

If T (x, 0) = f (x), obtain an expression for an .

23. Apply Gram-Schmidt orthogonalization to the basis B = {[1 0 1]T , [1 0 −1]T , [0 3 4]T }
of an inner product space R3 to obtain an orthogonal and an orthonormal basis.

24. Let B be a set of vectors in R3 with standard inner product defined. For the follow-
ings, identify whether B is a basis. Is the basis orthonormal? If not then obtain an
orthonormal basis.
(i) B = {[2 0 1]T , [3 −1 5]T , [0 4 2]T }
(ii) B = {[1 0 0]T , [1 1 0]T , [1 1 1]T }

25. A space of polynomial functions of the form xn ∀n ∈ I+ ∪ 0 of dimension 3 needs to

be expanded using a suitable orthonormal basis. If the inner product in the space is
R1
defined as < p, q >= −1 p(x)q(x)dx then determine a suitable orthonormal basis for
the space.
Chapter 7

Norm and metric spaces

Consider a vector v in an inner product space. A scalar d, defined as

√
d= < v, v > (7.1)

gives the norm of the vector v. It is also denoted as ||v||.

< v, v >= ||v||2 (7.2)

Using the positive definiteness of inner products, we can write

< u + αv >< u + αv >≥ 0 (7.3)

for every scalar α ∈ R.

=⇒ < u, u > +2α < u, v > +α2 < v, v >≥ 0 (7.4)

The above inequality is quadratic in α. We want to know the value of α for which the above
equation attains a minima. From the derivative test,

− < u, v >
α= (7.5)
< v, v >

< u, v >2
=⇒ < u, u > − ≥0 (7.6)
< v, v >
=⇒ < u, u >< v, v >≥< u, v >2 (7.7)

65
66 CHAPTER 7. NORM AND METRIC SPACES

The inequality of (7.7) is called Cauchy-Bunyakowski-Schwarz (CBS) inequality. In terms

of norms, the inequality can be written as

||u||||v|| ≥ | < u, v > | (7.8)

Now consider

||u + v||2 =< u + v, u + v >

= ||u||2 + < u, v > + < v, u > +||v||2

= ||u||2 + 2Re < u, v > +||v||2

But following the CBS inequality,

Re < u, v >≤ ||u||||v||

=⇒ ||u + v||2 ≤ ||u||2 + 2||u||||v|| + ||v||2 (7.9)

=⇒ ||u + v||2 ≤ (||u|| + ||v||)2 (7.10)

=⇒ ||u + v|| ≤ ||u|| + ||v|| (7.11)

The inequality given by (7.11) is called the triangle inequality. This inequality can easily be
identified as a case of the definition of “distance” between vectors. Norm of a vector can be
identified as the distance of that vector from the zero vector. We can now define a space,
called the “metric space” if the followings are satisfied.

(i) Positivity
d(u, v) ≥ 0 (7.12)

(ii) Symmetry
d(u, v) = d(v, u) (7.13)

(iii) Triangle inequality

d(u, v) ≤ d(u, w) + d(w, v) (7.14)

A more general definition of a metric is the p-metric defined as

Xn  p1
p
dp (u, v) = |vi − ui | (7.15)
i=1
 67

For continuous functions,

Z b  p1
p
dp (f, g) = |f (x) − g(x)| dx (7.16)
a

L2 space:
Consider a space X[a, b] of functions defined in the interval [a, b] such that f : [a, b] → C and
Z b
|f (x)|2 dx < ∞ (7.17)
a

In such a case, the space is called L2 space.

Consider two functions f and g ∈ L2 (a, b). From triangle inequality,

||αf + βg|| ≤ ||αf || + ||βg||

||αf + βg|| ≤ |α|||f || + |β|||g||

α, β ∈ C

Therefore, if f, g ∈ L2 (a, b) then all linear combinations in the space also belong to the space.
68 CHAPTER 7. NORM AND METRIC SPACES

Problems
1. Prove the following (in)equalities.
(i) ||u + v||2 = ||u||2 + ||v||2
(ii) | ||u|| − ||v|| | ≤ ||u − v||
(iii) ||u + v||2 + ||u − v||2 = 2(||u||2 + ||v||2 )
(iv) ||w − u||2 + ||w − v||2 = 12 ||u − v||2 + 2||w||2 − 12 ||u + v||2
(v) ||u − v|| ≤ ||u − w|| + ||w − v||

2. For each of the followings, determine whether they belong to L2 and calculate the norm
if it is defined.

1

1, 0 ≤ x ≤

(i) f (x) = 2
1
0, ≤x≤1

2
1
(ii) f (x) = √x , 0 ≤ x ≤ 1
(iii) f (x) = √1 , 0 ≤ x ≤ 1
x3
1
(iv) f (x) = x
,1 ≤ x ≤ ∞

3. Show that the infinite set of functions {1, cos(x), cos(2x) . . . sin(x), sin(2x) . . . } is or-
thogonal in real inner product space L2 (−π, π).

4. Show that the infinite set of functions

{einx : n ∈ Z}

is orthogonal in complex space L2 (−π, π).

5. Verify the CBS inequality for the following functions on [0, 1].

f (x) = 1

g(x) = x

6. Determine which of the following functions belong to L2 (0, ∞) and calculate the norm
in cases it is defined.
 69

(i) f (x) = e−x

(ii) f (x) = sin(x)
(iii) f (x) = cos(x)
1
(iv) f (x) = 1+x

7. Determine the real values of α for which xα lies in (a) L2 (0, 1) and (b) L2 (1, ∞).

8. Define a function f (x) ∈ L2 (−1, 1) such that < f (x), x2 + 1 >= 0 and ||f (x)|| = 2.
70 CHAPTER 7. NORM AND METRIC SPACES
Chapter 8

Adjoint operators

Consider an operator L defined on an inner product space V and two vectors u, v ∈ V

satisfying the following identity.

< Lu, v >=< u, L∗ v > (8.1)

In such a case, L∗ is said to be the adjoint of L. If L∗ = L then the operator is said to be

self-adjoint.
Let us first consider a matrix as an operator.

 
a11 a12
A= 
a21 a22
 
u1
u= 
u2
 
v1
v= 
v2
 
a11 u1 + a12 u2
Au= 
a21 u1 + a22 u2

71
72 CHAPTER 8. ADJOINT OPERATORS

=⇒ < A u, v > = (a11 u1 + a12 u2 )v1 + (a21 u1 + a22 u2 )v2

= a11 u1 v1 + a21 u1 v2 + a12 u2 v1 + a22 u2 v2

= u1 (a11 v1 + a21 v2 ) + u2 (a12 v1 + a22 v2 )

 
a11 a21
=< u,  v >
a12 a22
Clearly, from the previous equations,
 
a 11 a 21
A∗ =  
a12 a22
It is clear that since in general, A 6= A∗ , the operator given by matrix A is not self-adjoint.
Now we extend the above discussion to the adjoint of a differential operator. Consider the
following differential equation with the associated boundary conditions.
d2 f
+ αf = 0
dx2
df
(0) = 0
dx
df
(1) = 0
dx
Let us denote the differential operator by D.
d2
D= 2
+α
Zdx1  2 
df
< Df, g > = + αf gdx
0 dx2
Z 1 2  Z 1
df
= g dx + αf gdx
0 dx2 0
1 1 Z 1 2 Z 1
df  dg  dg
= g  − f + f 2 dx + αf gdx
dx 0 dx 0 0 dx 0
Z 1  2  1 1
dg df  dg 
=⇒ < Df, g > = f + αg dx + g  − f 
0 dx2 dx 0 dx 0
1
df 

From the boundary conditions on f , we have dx  = 0. Now if we choose the boundary
0
conditions on g such that
dg
(0) = 0
dx
dg
(1) = 0
dx
 73

then
1
d2 g
Z  
< Df, g > = f + αg dx (8.2)
0 dx2
< Df, g > =< f, D∗ g > (8.3)

Hence, in this case, we have the differential operator given by D which is self-adjoint with
the specified boundary conditions.
We showed the above protocol for obtaining the adjoint of an operator when the operator
was a matrix or a differential operator. The adjoint of a matrix operator can be used to
check the existance and uniqueness of solutions of A x = b.

Fredholm’s alternative theorem:

Consider the system of linear equations A x = b. In order to determine the solvability of
this system, we need to examine the homogeneous adjoint problem A y = 0. According to
Fredholm’s alternative theorem, A x = b will have a solution iff

< b, y >= 0 ∀y : A∗ y = 0 (8.4)

74 CHAPTER 8. ADJOINT OPERATORS

Problems

1. Determine the adjoint of the following matrix operators. Check whether they are self-
adjoint.  
i 0
(i) A =  
i 1
 
1 −2i
(ii) A =  
3 i
 
3 2+i 4
 
(iii) A = 2 − i −i
 
2
 
2 i 1
 
−4 + 5i 2 + 2i 1 − 2i
 
(iv) A =  2 + 2i −1 + 8i −2 − 2i
 
 
4 + 4i −2 − 2i −4 + 5i

d
2. The momentum operator in Quantum Mechanics is given as p̂ = −i dx . Check if this
operator is self-adjoint for a region [a, b].

3. The time-independent Schrödinger equation in 1-D is given as follows.

−h2 d2 ψ
+ V̂ ψ = Eψ
8π 2 m dx2

As a background study, find out the meaning and significance of all the quantities that
appear in the above equation. Check whether the operator in the above equation is
(a) linear, (b) self-adjoint.

4. Consider the following operator with homogeneous boundary conditions as shown be-
low.

d2
L̂ = + 1, x ∈ [0, π]
dx2
u(0) = u(π) = 0

Verify whether L̂ is self-adjoint.

 75

5. Consider the following operator L̂.

d2 d
L̂ = p2 2
+ p1 + p0
dx dx

where pi ∈ R. Determine the adjoint of L̂ and suitable boundary conditions to make

it a self-adjoint operator.

6. Identify the solvability condition and determine the range space using the alternative
theorem for the following set of equations.

x 1 + x 2 + x 3 = b1

2x1 − x2 + x3 = b2

x1 − 2x2 = b3

7. Using the alternative theorem, determine the solvability condition for the following set
of equations.

x1 − x2 + 2x3 = 3

2x1 + x2 + 6x3 = 2

x1 + 2x2 + 4x3 = −1

8. Repeat the previous exercise with the following set.

x1 + 2x2 + x3 + 2x4 − 3x5 = 2

3x1 + 6x2 + 4x3 − x4 + 2x5 = −1

4x1 + 8x2 + 5x3 + x4 − x5 = 1

−2x1 − 4x2 − 3x3 + 3x4 − 5x5 = 3

9. Repeat the previous exercise with the following set.

x1 − x2 + 3x3 + 2x4 = 2

3x1 + x2 − x3 + x4 = −3

−x1 − 3x2 + 7x3 + 3x4 = 7

76 CHAPTER 8. ADJOINT OPERATORS

10. Solve all relevant problems of chapter 4 using Fredholm’s alternative theorem.

11. Determine the adjoint of the linear transformation t : R2 → R2 given by

t(x, y) = (x + 2y, x − y) ∀(x, y) ∈ R2

12. Determine the adjoint of the linear transformation t : R3 → R3 given by

(i) t(x, y, z) = (x + 2y, 3x − 3z, y)
(ii) t(x, y, z) = (3x + 4y − 5z, 2x − 6y + 7z, −5x + 7y + z)

13. Show that the product of two self-adjoint operators is also a self-adjoint operator iff
the two operators commute.

14. Show that for self-adjoint operators:

(i) eigenvalues are always real
(ii) eigenfunctions/eigenvectors are orthogonal
Chapter 9

Eigenvalue problems

Consider a family of two first order ordinary differential equations shown below.

dx
= ax + by (9.1)
dt
dy
= cx + dy (9.2)
dt

The family of equations shown above can be written as a single matrix equation as shown
below.     
d x a b  x
= (9.3)
dt y c d y

For the above equation, the zero vector [0 0]T is always a solution, which is a trivial
solution. In order to know whether non-trivial solutions exist for this system, one needs to
solve for the homogeneous system.
    
a b x 0
   =   (9.4)
c d y 0

If the above equation is represented as a matrix equation as A u = 0, and if det(A) 6= 0 then

the trivial solution is the only solution. If det(A) = 0 then we get a number of solutions, all
of which lie on a straight line. These are called straight line solutions.
It is desired to determine the general solution of u0 = A u where u0 shows the derivative
with respect to t. If v is an eigenvector of A with the corresponding eigenvalue as λ then by
definition, A v = λv. We propose that u(t) = eλt v is a solution to u0 = A u. To test this,

77
78 CHAPTER 9. EIGENVALUE PROBLEMS

we compute the derivative of the test solution.

d λt
u0 (t) = (e v)
dt
= λeλt v

= eλt (λv)

= eλt (A v)

= A(eλt v)

The above analysis shows that eλt v is a solution to u0 = A u.

We can now generalize the above analysis to an n-dimensional system. If the coefficient
matrix A has n distinct eigenvalues then following the principle of linearity, the general
solution of the system is given as follows.

n
X
u(t) = ci eλi t vi (9.5)
i=1

The system of equations considered previously was a homogeneous system of the form
u0 = A u. Now we consider the solutions of non-homogeneous initial value problems of
the form u0 = A u + b(t) with an initial condition u(0) = u0 . For solution of such systems,
we make use of similarity transformation.

Similar matrices:
If P is any non-singular matrix such that P −1 A P = B then A and B are said to be similar
matrices. For similar matrices A and B consider the following operations.

B = P −1 A P

=⇒ B P −1 = P −1 A (P P −1 )

=⇒ B P −1 = P −1 A

=⇒ B P −1 u = P −1 A u
 79

If u is an eigenvector of A then

B P −1 u = P −1 (λu)

=⇒ B (P −1 u) = λ(P −1 u)

=⇒ B v = λv

where v = P −1 u. The above equation means that if λ and u are the eigenvalue and eigen-
vector of A, respectively, then λ will also be the eigenvalue of the similar matrix B with the
corresponding eigenvector as P −1 u.

Diagonalization of matrices:
For similarity transformation, B = P −1 A P . Consider a matrix P whose columns are made
up of eigenvectors of A.

A P = A[v1 | v2 | . . . | vn ]

= [A v1 | A v2 | . . . | A vn ]

= [λ v1 | λ v2 | . . . | λ vn ]

=P Λ

where
 
λ1 0 0 . . . 0
 
 0 λ2 0 . . . 0 
 
 
Λ=. (9.6)
 

 
 
. 
 
0 0 0 . . . λn

Hence, an n × n matrix can be diagonalized if it has n linearly independent eigenvectors.

This property is of interest, as is shown below.

Power of matrices: nth power of a matrix can be easily determined using similarity transfor-
80 CHAPTER 9. EIGENVALUE PROBLEMS

mation. As seen before, when the matrix P is made of the eigenvectors of A then

P −1 A P = Λ

=⇒ A = P Λ P −1

=⇒ An = (P Λ P −1 )n

=⇒ An = (P Λ P −1 )(P Λ P −1 ) . . . (P Λ P −1 )

=⇒ An = P Λn P −1

Inverse of a matrix:
Following the above method, it is easy to determine the inverse of a matrix as shown below.

A = P Λ P −1

=⇒ A−1 = (P Λ P −1 )−1

=⇒ A−1 = P Λ−1 P −1

If an n × n matrix does not have n linearly independent eigenvectors then there exists a
non-singular matrix P such that
P −1 A P = J (9.7)

where J is called the Jordan matrix and has the following form.
 
J 0 0 ... 0
 1 
 0 J2 0 ... 0 
 
 
J = . (9.8)
 

 
 
. 
 
0 0 0 . . . Jn

where J i ’s are called Jordan blocks in which eigenvalues appear on the diagonal, 1’s are on
the first superdiagonal and the rest of the elements are all zero. As an example, a 3 × 3
Jordan block can be written as follows.
 
λ 1 0
 
J = 0 λ 1
 
 
0 0 λ
 81

If an n × n matrix has k linearly independent eigenvectors then P is constructed from k

eigenvectors and n−k generalized eigenvectors. For a 3×3 matrix, the following possibilities
exist.
Only one eigenvector =⇒ one Jordan block
 
λ 1 0
 
J = 0 λ 1
 
 
0 0 λ

Two eigenvectors =⇒ two Jordan blocks

 
λ 1 0
 
J = 0 λ 0
 
 
0 0 λ
or  
λ 0 0
 
J = 0 λ 1
 
 
0 0 λ
Three eigenvectors =⇒ three Jordan blocks
 
λ 0 0
 
J = 0 λ 0
 
 
0 0 λ

Hence, the number of Jordan blocks equals the number of linearly independent eigenvectors
of the matrix.
Now we look into the procedure of determining the generalized eigenvectors. We take the
case of a 3 × 3 matrix. Let us first consider the case of a system with only one eigenvector
available. The following holds true.

P = [v | q1 | q2 ]

where v is the eigenvector and qi ’s are the generalized eigenvectors.

A P = [A v | A q1 | A q2 ]

=⇒ A P = [λ v | A q1 | A q2 ]
82 CHAPTER 9. EIGENVALUE PROBLEMS
 
λ 1 0
 
P J = [v | q1 | q2 ]  0 λ 1 
 
 
0 0 λ

=⇒ P J = [λ v | v + λ q1 | q1 + λ q2 ]

For J = P −1 A P , we equate A P = P J.

[λ v | A q1 | A q2 ] = [λ v | v + λ q1 | q1 + λ q2 ]

=⇒ A q1 = v + λq1 and A q2 = q1 + λq2

=⇒ (A − λI)q1 = v and (A − λI)q2 = q1

Hence, q1 and q2 can be obtained as the solutions of the above non-homogeneous problem.
Let us now consider the case where two eigenvectors are available.

A P = [A v1 | A v2 | A q]

=⇒ A P = [λ v1 | λ v2 | A q]

 
λ 0 0
 
P J = [v1 | v2 | q]  0 λ 1 
 
 
0 0 λ

=⇒ P J = [λ v1 | λv2 | v2 + λ q]

=⇒ [λ v1 | λ v2 | A q] = [λ v1 | λv2 | v2 + λ q]

=⇒ (A − λI)q = v2

The generalized vector can be obtained by solving the above equation. By exchanging v1
and v2 , another generalized vector can be obtained by solving the following equation.

(A − λI)q = v1

and further by linearity, the following can be written.

(A − λI)q = αv1 + βv2 (9.9)

 83

Now we use all the concepts developed till now for solving non-homogeneous initial value
problems. Given

d
u(t) = A u(t) + b(t)
dt
d −1
=⇒ (P u) = P −1 A u + P −1 b
dt
d −1
=⇒ (P u) = (P −1 A P ) (P −1 u) + P −1 b
dt

We identify P −1 A P = Λ in the above equation. Denoting P −1 u = v and P −1 b = g in the

above equation, we can write
d
v(t) = Λ v(t) + g(t) (9.10)
dt
The above equation can be solved using the method of integrating factor.
 
d −Λ t
e v = e−Λ t Λ v(t) + e−Λ t g(t) (9.11)
dt

The above equation can be solved using usual method except that we do not know the
exponential of a matrix. This can be determined as follows.

(Λt)2 (Λt)3
e−Λ t = I + Λt + + ... (9.12)
 2! 3! 
eλ1 t 0 0 ... 0
 
 0 e λ2 t 0 ... 0 
 
 
−Λ t
=⇒ e = . (9.13)
 

 
 
 . 
 
λn t
0 0 0 ... e
84 CHAPTER 9. EIGENVALUE PROBLEMS

Problems
1. Find thegeneral
 solutions for each of the following problems.
1 2
(i) u0 =  u
0 3
 
1 2
(ii) u0 =  u
3 6
 
1 2
(ii) u0 =  u
1 0
 
1 2
(iv) u0 =  u
3 −3

2. Find the general solution of the system

 
a b
u0 =  u
c d

where bc > 0

3. For a harmonic oscillator governed by the following equation

d2 x dx
2
+ b + kx = 0
dt dt

determine all the values of b and k for which the system has real and distinct eigen-
values. Find the general solution and the solution which satisfies the initial condition
x(0) = 1.
 
a 1
4. Consider A =  . Determine the value of the parameter a for which A has
0 1
repeated real eigenvalues.

5. For eachof the following systems, determine

 the
 general solutions.
0 1 1 1
(i) u0 =  u (ii) u0 =  u
1 0 1 0
   
1 1 1 1
(iii) u0 =  u (iv) u0 =  u
−1 0 −1 3
 85
   
1 1 1 1
(v) u0 =  u (vi) u0 =  u
−1 −3 1 −1

6. Determine the general solutions of the following harmonic oscillators.

d2 x dx
+ +x=0
dt2 dt
d2 x dx
+ 2 +x=0
dt2 dt

 
a b
7. Consider u0 =   u, where a + d 6= 0 and ad − bc 6= 0. Determine the general
c d
solution of the system.

8. Consider the harmonic oscillator described by the following equation.

 
0 1
x0 =  x
−k −b

where b ≥ 0 and k ≥ 0. For which values of k and b does the system have complex
eigenvalues? real and distinct eigenvalues? repeated eigenvalues?

9. Solve all the above relevant problems using similarity tranformation.

10. Consider
 the following matrices.
7 −16 −8
 
A = −16 7
 
8
 
−8 8 −5
 
−2 0 −2i
 
A= 0 1 0 
 
 
2i 0 −2
(i) Diagonalize the above matrices.
(ii) Determine the square of the above matrice using similarity transformation. Verify
using multiplication.
86 CHAPTER 9. EIGENVALUE PROBLEMS

11. Solve the following problem using similarity transformation.

du
=Au
dt
u(0) = [1 1 1]T
 
5 −3 −2
 
A =  8 −5 4 
 
 
−4 3 3

12. Consider the following matrix.

 
−2 1
A= 
−1 −2

Using similarity transformation, obtain the solution of

d
u(t) = A u(t) + b(t)
dt

where b = [1 1]T and u(0) = [0 0]T .

13. Using similarity transform, solve

d
u(t) = A u(t) + b(t)
dt

where
 
−i i 0
 
A =  i −i 0 
 
 
0 0 −i
√ 
2t
√ 
b =  2t
 
 
−t
e
 
√1
 2
u(0) =  √1 
 
 2
1
 87

14. Solve the initial value problem

d2 u du
2
+ 5 + 6u = e−t
dt dt

with  
du
= (u)0 = 1
dt 0

15. Convert the following initial value problem to a matrix equation and solve using simi-
larity transformation and otherwise.

d2 u du du
+ 3 + 2u = 0; u(0) = 1; (0) = 3
dt2 dt dt

16. Verify that {[e2t e2t e2t ]T , [e−t 0 e−t ]T , [−e−t e−t 0]T } is a solution set to the
system  
0 1 1
dx  
= 1 0 1 x
 
dt  
1 1 0

17. Find the general solution of the following system.

   
1 −2 2 2et
dx   
 t 

= −2 1 2 x +  4e 
 
dt    
2 2 1 −2et

18. Find the general solution of the following system.

   
dx 1 1 −t − 1
= x+ 
dt 4 1 −4t − 2

19. Find the general solution of the following system.

   
dx  2 2 −4cos(t)
= x +  
dt 2 2 −sin(t)
88 CHAPTER 9. EIGENVALUE PROBLEMS
Chapter 10

Sturm-Liouville theory

Consider an operator of the form given below.

 
d d
L̂ = p(x) + r(x) (10.1)
dx dx
The eigenvalue problem
L̂u + λρ(x)u = 0, x ∈ (a, b) (10.2)

subject to homogeneous boundary conditions

du
α1 u(a) + α2 (a) = 0; |α1 | + |α2 | > 0 (10.3)
dx
du
β1 u(b) + β2 (b) = 0; |β1 | + |β2 | > 0 (10.4)
dx
with αi and βi as real coefficients constitute a Sturm-Liouville problem. It can be shown
with the method shown previously that the Sturm-Liouville operator is a self-adjoint oper-
ator. Hence, under the boundary conditions stated above, the eigenvalues are real and the
eigenfunctions are orthogonal. When the interval (a, b) is bounded and p(x) does not vanish
on [a, b] then the problem is called a regular Sturm-Liouville problem. Else, the problem is
called singular Sturm-Liouville problem. The set of eigenvalues of a regular Sturm-Liouville
problem are countably infinite and is a monotonically increasing sequence

λ0 < λ1 < λ2 < . . . λn < λn+1 < . . .

Numerous classes of Sturm-Liouville problems are identified and solutions derived in the
following sections

89
90 CHAPTER 10. STURM-LIOUVILLE THEORY

Bessel equation
The Bessel equation is given as follows.

d2 y dy
x2 2
+ x + (x2 − n2 )y = 0 (10.5)
dx dx

where n is a non-negative parameter. The solutions to the Bessel equation are called Bessel
functions. It can be seen that the equation is an ordinary differential equation with variable
coefficients. Such equations can be solved using a technique called series solution due to
Frobenius. We first recast the equation to the following form.

d2 y 1 dy n2
+ + (1 − )y = 0 (10.6)
dx2 x dx x2

The solution to the equation is of the following form.

∞
X
y= ar xk+r (10.7)
r=0
∞
dy X
=⇒ = (k + r)ar xk+r−1 (10.8)
dx r=0
2 ∞
dy X
=⇒ = (k + r − 1)(k + r)ar xk+r−2 (10.9)
dx2 r=0

Substitution of the assumed solution and the derivatives gives

∞
X ∞
X ∞
X ∞
X
k+r−2 k+r−2 k+r
(k + r − 1)(k + r)ar x + (k + r)ar x + ar x − ar n2 xk+r−2 = 0
r=0 r=0 r=0 r=0

∞ 
X 
k+r−2 k+r
=⇒ (k + r + n)(k + r − n)x +x =0 (10.10)
r=0

The above equation is an important result and a major landmark in obtaining solutions of
the type discussed here. We compare the coefficients of the above polynomial equation.

r = 0, a0 (k + n)(k − n) = 0 =⇒ k = ±n

r = 1, a1 [(k + 1 + n)(k + 1 − n)] = 0 =⇒ a1 [(k + 1)2 − n2 ] = 0 =⇒ a1 = 0

We compute the coefficient of a random term, xk+r , for r = r.

ar+2 [(k + r + 2 + n)(k + r + 2 − n)] + ar = 0

 91

−ar
=⇒ ar+2 = (10.11)
(k + r + 2 + n)(k + r + 2 − n)
The above equation is one of the most important results of this solution procedure and is
called the recurrence relation. This relation relates different coefficients appearing in the
solution. From a1 = 0 and the above recurrence relation, we can write,

a1 = a3 = a5 = · · · = 0

For k = +n,
−ar
ar+2 =
(2n + r + 2)(r + 2)
−a0
=⇒ a2 =
(2)(2n + 2)
−a0
=⇒ a2 = 2
2 (n + 1)
−a2 −a2
=⇒ a4 = = 2
(4)(2n + 4) 2 · 2 (n + 2)
a0
=⇒ a4 = 4
2! · 2 (n + 2)(n + 1)
−a4 −a4
=⇒ a6 = =
(6)(2n + 6) 3 · 22 (n + 3)
−a0
=⇒ a6 = 6
3! · 2 (n + 3)(n + 2)(n + 1)
From the above expressions of the coefficients, we can write the solution to the Bessel equa-
tion as follows.
x2 x4 x6
 
k
y = a0 x 1 − + − − ...
22 (n + 1) 2!24 (n + 2)(n + 1) 3! · 26 (n + 3)(n + 2)(n + 1)
Therefore, for k = +n, the solution to the Bessel equation can be written as follows.
∞
n
X (−1)r x2r
y = a0 x (10.12)
r=0
r!22r (n + 1)(n + 2) . . . (n + r)

We have determined the solution of the Bessel equation. The only unknown is a0 . For a
specific choice of a0 , we get a clean solution of the equation, as given below.
1 1
a0 = = (10.13)
2n Γ(n + 1) 2n n!
∞
1 n
X (−1)r x2r
=⇒ y = x (10.14)
2n Γ(n + 1) r=0 r!22r (n + 1)(n + 2) . . . (n + r)
92 CHAPTER 10. STURM-LIOUVILLE THEORY

∞
X (−1)r x2r+n
=⇒ y = (10.15)
r=0
22r+n r!(r + n)!
This series is called Bessel’s function of the first kind of order n, and is conventionally denoted
s Jn (x). With k = +n and − n, the Bessel’s functions can be written as follows.
∞  2r+1
r x 1
X
Jn (x) = (−1) (10.16)
r=0
2 r!(r + n)!
∞  2r+1
r x 1
X
J−n (x) = (−1) (10.17)
r=0
2 r!(r − n)!
The following plots show the first two Bessel functions J0 (x) and J1 (x).

Figure 10.1: Bessel functions of the first kind J0 (x) and J1 (x)

In the previous discussion, we used the concept of gamma function. Let us remind ourselves
of the gamma function which is defined as follows.
Z ∞
Γ(x) = e−t tx−1 dt (10.18)
0

As seen before, gamma function is related to the factorial as follows.

Γ(x + 1) = x! (10.19)

We discussion only about the Bessel functions of the first kind. Bessel functions of the second
kind can be derived from Bessel functions of the first kind. This is left as a background study
and an exercise.
 93

Legendre equation
The Legendre equation is given as

d2 y
2 dy
(1 − x ) 2 − 2x + λy = 0; − 1 < x < 1 (10.20)
dx dx

Following the Frobenius method, the series solution to the above equation can be represented
as an infinite series given below.
∞
X
y= ar xk+r (10.21)
r=0
∞
dy X
=⇒ = (k + r)ar xk+r−1 (10.22)
dx r=0
2 ∞
dy X
=⇒ = (k + r − 1)(k + r)ar xk+r−2 (10.23)
dx2 r=0

Hence, the original differential equation can be cast as follows.

∞
X X∞ ∞
X ∞
X
k+r−2 k+r k+r
ar (k+r−1)(k+r)x − ar (k+r−1)(k+r)x −2 ar (k+r)x +λ ar xk+r = 0
r=0 r=0 r=0 r=0

∞
X    
k+r−2 k+r
=⇒ ar (k + r − 1)(k + r)x + λ − (k + r)(k + r + 1) x =0 (10.24)
r=0

The above equation puts us in a position to write the recurrence relation as given below.
 
(k + r)(k + r + 1) − λ
ar+2 = ar (10.25)
(k + r + 1)(k + r + 2)

Therefore, a knowledge of a0 and a1 will yield the complete series through the recurrence
relation. When k = 0 and λ = n(n + 1), n ∈ I+ ∪ 0, For such a case

an+2 = an+4 = an+6 = · · · = 0

Therefore, the infinite series solution reduces to a polynomial for the above choice of param-
eters. In such a case,  
r(r + 1) − n(n + 1)
ar+2 = ar
(r + 1)(r + 2)
 
(r − n)(r + n + 1)
ar+2 = ar (10.26)
(r + 1)(r + 2)
94 CHAPTER 10. STURM-LIOUVILLE THEORY

With arbitrary a0 and a1 , we get,

   
−n(n + 1) (n − 2)n(n + 1)(n + 3)
a2 = a0 a4 = a0
2! 4!
.
   
−(n − 1)(n + 2) (n − 3)(n − 1)(n + 2)(n + 4)
a3 = a1 a5 = a1
3! 5!
.

Hence, the solution to the Legendre equation takes the following form.
 
n(n + 1) 2 (n − 2)n(n + 1)(n + 3) 4
y = a0 1 − x + x + ...
2! 4!
 
(n − 1)(n + 2) 3 (n − 3)(n − 1)(n + 2)(n + 4) 5
+ a1 x − x + x + ...
3! 5!

If u0 (x) and u1 (x) are the power series with even only and odd only powers, respectively,
then the above solution can be written as follows.

y = a0 u0 (x) + a1 u1 (x) (10.27)

It can be seen from the above analysis that the solution depends upon n. For each n, a pair
of linearly independent solutions is obtained.

n = 0, u0 (x) = 1
1 1
u1 (x) = x + x3 + x5 + . . .
3 5
1
n = 1, u0 (x) = 1 − x2 − x4 + . . .
3
u1 (x) = x

n = 2, u0 (x) = 1 − 3x2
2 1
u1 (x) = x − x3 − x5 + . . .
3 5
n = 3, u0 (x) = 1 − 6x2 + 3x4 + . . .
5
u1 (x) = x − x3
3
.
 95

It can be observed that for every n, one of the solutions is a polynomial solution while the
other is an infinite series. One is generally interested in the polynomial solutions. Such
solutions can be written in the following form.

y = an xn + an−2 xn−2 + an−4 xn−4 . . . (10.28)

One specific case is when an is given as follows.

(2n)! 1 · 3 · 5 · · · (2n − 1)
an = n 2
= (10.29)
2 (n!) n!

The resulting polynomial is called Legendre polynomial of degree n, denoted as Pn (x). The
general expression for Legendre polynomial of degree n can be written as follows.
[n/2]  
1 X k (2n − 2k)!
Pn (x) = n (−1) xn−2k (10.30)
2 k=0 k!(n − k)!(n − 2k)!

where 
n/2, if n is even
[n/2] =
(n − 1)/2, if n is odd

P0 (x) = 1 P1 (x) = x
1 1
P2 (x) = (3x2 − 1) P3 (x) = (5x3 − 3x)
2 2
1 1
P4 (x) = (35x4 − 30x2 + 3) P5 (x) = (63x5 − 70x3 + 15x)
8 8

The following figure shows the first five Legendre polynomials.

96 CHAPTER 10. STURM-LIOUVILLE THEORY

Laguerre equation
Laguerre equation is given as follows.
d2 y dy
x 2 + (1 − x) + λy = 0 (10.31)
dx dx
Using the series solution technique, we can write the followings.
∞
X
y= ar xk+r (10.32)
r=0
∞
dy X
=⇒ = (k + r)ar xk+r−1 (10.33)
dx r=0
∞
d2 y X
=⇒ = (k + r − 1)(k + r)ar xk+r−2 (10.34)
dx2 r=0

Substitution of the above quantities in the original equation gives

∞
X ∞
X ∞
X
k+r−2 k+r−1
x (k + r − 1)(k + r)ar x + (1 − x) (k + r)ar x +λ ar xk+r = 0
r=0 r=0 r=0
∞
X ∞
X
2 k+r−1
=⇒ (k + r) ar x − (k + r − λ)ar xk+r = 0 (10.35)
r=0 r=0
The recurrence relation can be obtained from the above equation by collecting the terms of
xk+r−1 .
(k + r)2 ar − [(r − 1) + k − λ]ar−1 = 0
 
k+r−λ−1
=⇒ ar = ar−1 (10.36)
(k + r)2
Hence, with arbitrary a0 6= 0, we get the complete series. Further,

(k + r)2 ar = (k + r − λ − 1)ar−1 = 0

with a−1 = 0. Therefore, a0 = 0 or k = 0. Since a0 6= 0, k = 0.

 
r−λ−1
=⇒ ar = ar−1 (10.37)
r2
 
r+λ
=⇒ ar+1 = − ar (10.38)
(r + 1)2
Hence, the solution of the Laguerre equation is of the following form.
∞
X
y= ar x r (10.39)
r=0
 97

with the coefficients obtained by the recurrence relation obtained above.

Now consider a case when r = λ. In such a case, the series terminates to a polynomial
solution. This means that further coefficients all become zero (ar+1 , ar+2 = · · · = 0). Hence,
the solution looks like the one given below.
λ
X
y= ar xr = ar xr + ar−1 xr−1 + · · · + a1 x + a0 (10.40)
r=0

This polynomial solution is called the Laguerre polynomial. From the previous recurrence
relation, we can write
r2
 
ar−1 = ar
r−λ−1
We can substitute r = λ, λ−1, λ−2 . . . in the above recurrence relation to obtain the general
coefficient of the polynomial solution. The expression for aλ−n can be written as follows.

(λ!)2
 
n
aλ−n = (−1) aλ (10.41)
((λ − n)!)2 (n!)

Hence, the solution of the equation as the Laguerre polynomial can be written as
λ 
(λ!)2
X 
n
Lλ (x) = (−1) aλ xλ−n (10.42)
n=0
((λ − n)!)2 (n!)

First four Laguerre polynomials are given below.

L0 (x) = 1 L1 (x) = 1 − x

L2 (x) = 2 − 4x − x2 L3 (x) = 6 − 18x + 9x2 − x3

98 CHAPTER 10. STURM-LIOUVILLE THEORY

Hermite equation
The following differential equation is called the Hermite equation.
d2 y dy
2
− 2x + 2λy = 0 (10.43)
dx dx
Using the series solution with k = 0, we get the followings.
∞
X
y= ar x r (10.44)
r=0
∞
dy X
=⇒ = rar xr−1 (10.45)
dx r=0
∞
d2 y X
=⇒ = (r − 1)rar xr−2 (10.46)
dx2 r=0
∞
X ∞
X ∞
X
=⇒ (r − 1)rar xr−2 − 2x rar xr−1 + 2λ ar x r = 0 (10.47)
r=0 r=0 r=0

Proceeding like before, we get the recurrence relation as shown below.

 
2(r − λ − 2)
ar = ar−2 (10.48)
r(r − 1)
Hence, in the polynomial solutions, only odd and only even powers are observed. With
r = 2n (even coefficients) and r = 2n + 1 (odd coefficients) we get the followings.
 
2(−λ + 2n − 2)
a2n = a2n−2
2n(2n − 1)
  
2(−λ + 2n − 2) 2(−λ + 2n − 4)
=⇒ a2n = a2n−4
2n(2n − 1) (2n − 2)(2n − 3)
    
2(−λ + 2n − 2) 2(−λ + 2n − 4) 2(−λ + 2 − 2)
=⇒ a2n = ... a0
2n(2n − 1) (2n − 2)(2n − 3) 2(2 − 1)
2n
=⇒ a2n = (2n − λ − 2)(2n − λ − 4) . . . (−λ)a0
(2n)!
Similarly,
 
2(−λ + 2n − 1)
a2n+1 = a2n−1
(2n + 1)2n
  
2(−λ + 2n + 1) 2(−λ + 2n + 3)
=⇒ a2n = − − a2n−3
(2n + 1)2n (2n − 1)(2n − 2)
    
2(−λ + 2n − 1) 2(−λ + 2n − 3) 2(−λ + 2 − 1)
=⇒ a2n = ... a1
(2n + 1)2n (2n − 1)(2n − 2) (2 + 1)2
2n
=⇒ a2n = (2n − λ − 1)(2n − λ − 3) . . . (1 − λ)a1
(2n + 1)!
 99

Hence, the solution to the Hermite equation can be written as the summation of the even
and odd series.
y = a0 yeven + a1 yodd (10.49)

Like before, we can see that the infinite series truncates to polynomials when λ is an integer
(say N ) in which case, the recurrence relation can be written as follows.
 
2(r − N − 2)
ar = ar−2
r(r − 1)

It can be seen that the series will terminate when r reaches a values of N + 2. Hence,
we get polynomial solutions. Following the previous analysis, we list out first few Hermite
polynomials below and leave it for you to plot and see their nature.

H0 (x) = 1

H1 (x) = 2x

H2 (x) = 4x2 − 2

H3 (x) = 8x3 − 12x

H4 (x) = 16x4 − 48x2 + 12

H5 (x) = 32x5 − 160x3 + 60x

100 CHAPTER 10. STURM-LIOUVILLE THEORY

Problems
1. Show that the Sturm-Liouville operator is:
(i) a linear operator
(ii) self-adjoint

2. Show that every second order ordinary differential equation

d2 y dy
a2 (x) 2
+ a1 (x) + a0 (x)y = f (x)
dx dx
can be cast as a Sturm-Liouville problem
 
d dy
p(x) + r(x)y = F (x)
dx dx
using the following transformations.
Z 
a1 (x)
p(x) = exp dx
a2 (x)
a0 (x)
r(x) = p(x)
a2 (x)
f (x)
F (x) = p(x)
a2 (x)

3. Verify whether Legendre, Laguerre, Hermite and Bessel equations

(i) can be cast as Sturm-Liouville problems?
(ii) are linear?
(iii) are self-adjoint?

4. Determine the eigenvalues and eigenfunctions of the boundary value problem

d2 y
+ λy = 0; a ≤ x ≤ b
dx2
y(a) = y(b) = 0

5. Use Gram-Schmidt method to construct an orthogonal set of polynomials out of the

independent set
{1, x, x2 , x3 , x4 , x5 : −1 ≤ x ≤ 1}

Compare the result with the Legendre polynomials Pi (x), i = 1 − 5. Show that there
is a linear relationship between the two set of polynomials.
 101

6. Verify that Pn (x) satisfies the Legendre’s equation for n = 3 and n = 4.

7. Show that the substitutions

x = cosθ

y(cosθ) = u(θ)

transform the Legendre equation to

d2 u du
sinθ 2
+ cosθ + [n(n + 1)sinθ]u = 0, 0 ≤ θ ≤ π
dx dx

8. Verify Rodrigues generating function for Legendre polynomial as

1 dn 2
Pn (x) = (x − 1)n
2n n! dxn

9. Show that Legendre polynomials are orthogonal in L2 (−1, 1).

10. Verify whether the following is a correct generating function for Hermite polynomials.
2 dn −x2
Hn (x) = (−1)n ex (e )
dxn

11. Determine whether the set {Hn (x) : n ∈ N0 } is orthogonal in L2 (R).

√
12. Show that ||Hn (x)||2 = 2n n! π.

13. Verify whether the following is a correct generating function for Laguerre polynomials.
n
1 n k
X  
k
Ln (x) = (−1) x
k=0
k! k

where
n
 
n!
=
k k!(n − k)!

14. Verify whether the following is a correct generating function for Laguerre polynomials.
dn n −x
Ln (x) = ex (x e )
dxn

15. Show that Bessel functions Jn and J−n are not linearly independent if n ∈ N0 . Check
the same for the case of n ∈ I.