al

d
Lecture Notes: Matrices (Semester 1 - CC1)
an
M
Dr. P. Mandal
P.

October 2021
.
Dr
©
 ii

© Dr
. P.
M
an
d al
Contents

Matrices v

1 Matrices 1

al
1.1 Definitions: Some Terminologies . . . . . . . . . . . . . . . . . 1

d
1.2 Matrix Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Special Square Matrices . . . . . . . . . . . . . . . . . . . . . 9
an
1.4 Eigenvalue Problems . . . . . . . . . . . . . . . . . . . . . .
1.4.1 Corollaries . . . . . . . . . . . . . . . . . . . . . . . .
.
.
13
18
M
1.5 Cayley-Hamilton Theorem . . . . . . . . . . . . . . . . . . . . 22
1.6 Diagonalization of Matrices . . . . . . . . . . . . . . . . . . . 24
1.6.1 Corollaries . . . . . . . . . . . . . . . . . . . . . . . . . 25
P.

1.7 Similarity Transformation . . . . . . . . . . . . . . . . . . . . 26

1.8 Unitary Transformation . . . . . . . . . . . . . . . . . . . . . 27
1.9 Evaluating Power of a Matrix . . . . . . . . . . . . . . . . . . 28
.
Dr

1.10 Solutions of Linear Coupled First Order Ordinary Differential

Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
©

iii
 iv

© Dr
. P.
M
an
d al
CONTENTS
 Matrices
© Dr

v
. P.
M
an
d al
vi CHAPTER 0. MATRICES

al
d
an
M
P.
.
Dr
©
Chapter 1

Matrices

al
Syllabus: (a) Addition and Multiplication of Matrices. Null Matrices. Di-
agonal, Scalar and Unit Matrices. Transpose of a Matrix. Symmetric and

d
Skew-Symmetric Matrices. Conjugate of a Matrix. Hermitian and Skew-
an
Hermitian Matrices. Singular and Non-Singular matrices. Orthogonal and
Unitary Matrices. Trace of a Matrix. Inner Product.
M
(b) Eigenvalues and Eigenvectors. Cayley-Hamiliton Theorem. Diago-
nalization of Matrices. Solutions of Coupled Linear Ordinary Differential
Equations. Functions of a Matrix.
P.

1.1 Definitions: Some Terminologies

.
Dr

1. Matrix: A matrix is an array of numbers which may be complex. The

array contains m × n numbers in m rows and n columns. m × n is the
©

order of the matrix.

 
a11 a12 ... ... a1n
 a21 a22 ... ... a2n 
 
A=  ... ... ... ... ... 

 ... ... ... ... ... 
am1 am2 ... ... amn

The element which belongs to the i-th row and j-th column is denoted
as aij .

2. Row matrix: A row matrix contains only one row i.e. the numbers
are arrayed in a single row. The order of such matrix is thus 1 × n.


R= a11 a12 ... ... a1n

1
2 CHAPTER 1. MATRICES

3. Column matrix: A column matrix contains only one column i.e. the
numbers are arrayed in a single column. The order of such matrix is
thus m × 1.  
a11
 a21 
 
C=  ... 

 ... 
am1

4. Null matrix: If all the elements of a matrix are zero, it is called a

null matrix i.e. for a null matrix aij = 0 for all i, j.

5. Square matrix: If the number of rows (m) and the number of columns

al
(n) of a matrix are equal, it is called a square matrix. The matrix S
below is a square matrix of order 3.

d
 
a11 a12 a13
an
S =  a21 a22 a23 
a31 a32 a33
M
The elements aii are called the diagonal elements.
P.

6. Diagonal matrix: If all the off-diagonal elements of a square matrix

are zero but at least one diagonal element is non-zero, it is called a
diagonal matrix i.e. for a diagonal matrix aij = 0 for all i ̸= j but
.

aii ̸= 0 for at least one i.

Dr

 
a11 0 0
©

D =  0 a22 0 
0 0 a33

D is a diagonal matrix of order 3.

7. Unit or Identity matrix: If all the diagonal elements of a diagonal

matrix are equal to identity, it is called a unit or identity matrix. For
a unit matrix aij = 0 for all i ̸= j and aii = 1 for all i i.e. aij = δij .
 
1 0 0
I =  0 1 0  is a diagonal matrix of order 3.
0 0 1
8. Equal matrices: Two matrices of same order are said to be equal if
the elements of one matrix are equal to the corresponding elements of
other matrix i.e. matrices A and B are equal if aij = bij for all i, j.
1.2. MATRIX ALGEBRA 3

1.2 Matrix Algebra

1. Addition & Subtraction: Addition or subtraction of two matrices
A and B of same order is defined as C = A ± B where cij = aij ± bij
for all i and j. For example,
   
a11 a12 a13 b11 b12 b13
A= and B =
a21 a22 a23 b21 b22 b23

 
a11 ± b11 a12 ± b12 a13 ± b13
∴A±B =
a21 ± b21 a22 ± b22 a23 ± b23

al
Properties:

d
ˆ Matrix addition is commutative i.e. A + B = B + A.
an
ˆ Matrix addition is associative i.e. A + (B + C) = (A + B) + C
M
2. Multiplication:

(a) Multiplication by a scalar: Multiplying a matrix by a number or

P.

scalar (say k) means multiplying each element by the number i.e.

kA = Ak = k(aij ).
.

(b) Matrix product: The product of two matrices A and B is defined

Dr

as C = AB, where
Xn
cij = aik bkj
©

k=1

For the product to be defined, matrices A and B must be con-

formable i.e. if A = (a)ij is an m × n matrix, then B = (b)ij must
be an n × s matrix. The product matrix C is an m × s matrix.
Consider, for example,
 
  b11 b12
a11 a12 a13
A= and B =  b21 b22 
a21 a22 a23
b31 b32

Therefore, by definition, the product matrix

 
a11 b11 + a12 b21 + a13 b31 a11 b12 + a12 b22 + a13 b32
C=
a21 b11 + a22 b21 + a23 b31 a21 b12 + a22 b22 + a23 b32
4 CHAPTER 1. MATRICES

Consider the following system of equations:

3x − y + 2z = −3
2x + 3y − z = 2 (1.1)
x − 2y + z = 5

The set of equations can be written in matrix form as

    
3 −1 2 x −3
 2 3 −1   y  =  2 
1 −2 1 z 5

Properties:

al
ˆ Matrix multiplication is, in general, not commutative i.e. AB ̸=
BA.

d
ˆ Matrix multiplication is associative i.e. A(BC) = (AB)C
an
ˆ Matrix multiplication follows the distributive law i.e. A(B +
C) = AB + AC.
M
(c) Direct product: The direct product is defined for general matrices.
Given an n×n matrix A and an m×m matrix B, the direct product
of A and B is an nm × nm matrix, and is defined by C = A ⊗ B
P.

where cik,jl = aij bkl . If

   
a11 a12 b11 b12
.

A= and B =
Dr

a21 a22 b21 b22

©

 
  a11 b11 a11 b12 a12 b11 a12 b12
a11 B a12 B  a11 b21 a11 b22 a12 b21 a12 b22 
C = A⊗B = = 
a21 B a22 B  a21 b11 a21 b12 a22 b11 a22 b12 
a21 b21 a21 b22 a22 b21 a22 b22

3. Commutator of matrices: Commutator of two square matrices A

and B of same order is defined as [A, B] = AB − BA. In general,
AB ̸= BA and hence [A, B] ̸= 0. For example,
   
1 2 1 0
A= and B =
1 3 1 2
Note that
         
1 2 1 0 3 4 1 0 1 2 1 2
AB = = but BA = =
1 3 1 2 4 6 1 2 1 3 3 8
1.2. MATRIX ALGEBRA 5

If the commutator [A, B] = AB − BA = 0 i.e. AB = BA, the matrices

are said to be commutative.
Properties:

ˆ [A, B] = −[B, A]
ˆ [A, B + C] = [A, B] + [A, C]
ˆ [A, BC] = [A, B]C + B[A, C]
ˆ [A, [B, C]] + [B, [C, A]] + [C, [A, B]] = 0

The anticommutator of two matrices is defined as {A, B} = AB + BA.

al
4. Power of a matrix: For any positive integer n, the power of a square
matrix A is defined as An = AA...A (n times) i.e. A2 = AA, A3 =

d
AAA. In particular, A0 = I, the unit matrix1 .
an
5. Function of a matrix: The function of a matrix maps a matrix
M
to another matrix. For example, consider a matrix function f (A) =
3A2 − 2A + I wherePI is the unit matrix of same order. A more fancy
example is f (A) = k ak Ak , where ak are scalar coefficients. Another
P.

common series is defined by

∞
X Ak
.

eA =
Dr

k=0
k!

There are several techniques for lifting a real function to a square matrix
©

function. If the real function f (x) has the Taylor series expansion

x2
f (x) = f (0) + f ′ (0)x + f ′′ (0) + ...
2!

then a matrix function f (A) can be defined by substituting x by a

matrix A i.e.

A2
f (A) = f (0)I + f ′ (0)A + f ′′ (0) + ...
2!
where the powers become matrix powers, the additions become matrix
sums and the multiplications become scaling operations.
1
A−n = A−1 A−1 ...A−1 (n times) is defined if A is a nonsingular matrix i.e., |A| =
̸ 0
6 CHAPTER 1. MATRICES

6. Transpose of a matrix: For any matrix A, the transpose matrix AT

is obtained by interchanging corresponding rows and columns of A, i.e.
if A = (aij ), AT = (aji ). For example,
 
  1 3
1 1 −2
A= ⇒ AT =  1 0 
3 0 1
−2 1

Properties:

ˆ (AT )T = A
ˆ (A + B)T = AT + B T
ˆ (AB)T = B T AT

al
7. Complex Conjugate: For any matrix A, the complex conjugate ma-

d
trix A∗ is formed by taking the complex conjugate of each element of
an
A, i.e. if A = (aij ), A∗ = (a∗ij ) for all i and j. For example,
M
   
1 2+i ∗ 1 2−i
A= ⇒A =
3−i i 3 + i −i
P.

Properties:

ˆ (A∗ )∗ = A
.

ˆ (A + B)∗ = A∗ + B ∗
Dr

ˆ (AB)∗ = A∗ B ∗
©

8. Hermitian Conjugate: For an arbitrary matrix A, the Hermitian

conjugate matrix A† is obtained by taking the complex conjugate of
the matrix, and then the transpose of the complex conjugate matrix
i.e. if A = (aij ), A† = (a∗ji ) for all i and j. For example,
 
2 + 3i 4 − 3i
A =
4i 3
 T  
† 2 − 3i 4 + 3i 2 − 3i −4i
⇒A = =
−4i 3 4 + 3i 3

Properties:

ˆ (A† )† = A.
ˆ (A + B)† = A† + B † .
1.2. MATRIX ALGEBRA 7

ˆ (AB)† = B † A† .

9. Trace of a matrix: The trace of a square P matrix is defined as the

sum of its diagonal elements i.e. Tr(A) = i aii . For example,
 
2 i
A= ⇒ Tr(A) = 2 + 3 = 5
0 3

Properties:
ˆ Tr(A) = Tr(AT ).
ˆ Tr(A + B) = Tr(A) + Tr(B).
ˆ Tr(AB)=Tr(BA).

al
10. Determinant of a matrix: The determinant of a square matrix A is

d
defined as the determinant having same array as that of the matrix and
is generally denoted as |A|or det(A). For example, the determinant
an
2 3 2 3
of the matrix A = is |A| = = 5. If the determinant
1 4 1 4
M
of a matrix is zero i.e. |A| = 0, A is called a singular matrix.
Properties:
P.

ˆ |AB| = |BA| = |A||B|

ˆ |AT | = |A|
.

11. Cofactor matrix:

Dr

The cofactor matrix is defined as Ac = Aij . For example,

©

   
a11 a12 a13 A11 A12 A13
A=  a21 a22 a23  ⇒ Ac =  A21 A22 A23 
a31 a32 a33 A31 A32 A33
where
a22 a23 a21 a23 a21 a22
A11 = (−1)1+1 , A12 = (−1)1+2 , A13 = (−1)1+3
a32 a33 a31 a33 a31 a32
a12 a13 a11 a13 a11 a12
A21 = (−1)2+1 , A22 = (−1)2+2 , A23 = (−1)2+3
a32 a33 a31 a33 a31 a32
a12 a13 a11 a13 a11 a12
A31 = (−1)3+1 , A32 = (−1)3+2 , A33 = (−1)3+3
a22 a23 a21 a23 a21 a22
   
2 −1 c 3 0
The cofactor matrix of A = is A =
0 3 1 2
8 CHAPTER 1. MATRICES

12. Adjoint of a matrix:

The adjoint of a matrix is defined as the transpose of its cofactor
 matrix

cT 2 −1
i.e. adj(A) = A . For example, consider the matrix A = .
  0 3
3 0
The cofactor matrix Ac = .
1 2
 T  
3 0 3 1
Hence, the adjoint matrix adj(A) = = .
1 2 0 2
13. Inverse of a matrix:
For a given nonsingular matrix A, there exists another matrix B such
that AB = BA = I. Matrix B is called the inverse matrix of A

al
(B = A−1 ). For example, consider the following matrices:

d
   
2 −5 3 5
A=
−1 3
& B= an 1 2
M
Note that,
    
2 −5 3 5 1 0
AB = = =I
P.

−1 3 1 2 0 1

Similarly,
.

    
Dr

3 5 2 −5 1 0
BA = = =I
1 2 −1 3 0 1
©

 
3 5
∴B= = A−1 , the inverse matrix of A.
1 2
The inverse matrix can be found by using the relation

adj(A)
A−1 = (1.2)
|A|

 
2 −1
Problem 1: Find out the inverse matrix of A = .
0 3
 
3 1
Solution: Adjoint matrix of A i.e. adj(A) = . The determi-
0 2
3 1
nant of the matrix i.e. |A| = =6
0 2
1.3. SPECIAL SQUARE MATRICES 9
 
−1 1 3 1
∴A = 6
.
0 2

Properties:

ˆ (A−1 )−1 = A.
ˆ (A + B)−1 = A−1 + B −1 .
ˆ (AB)−1 = B −1 A−1 .

14. Derivative of a matrix: The derivative of a matrix with respect to

a variable say, x is equal to the derivative of each element with respect
x separately. For example,

al
   
d x x2 1 1 2x 0
=
dx ex 0 2x3 ex 0 6x2

d
an
15. Integral of a matrix: The integral of a matrix with respect to a
variable say, x is equal to the integral of each element with respect x
M
separately. For example,
Z    x2   
x 3x2 1 2
x3 x c1 c2 c3
dx = +
P.

4
ex 0 2x3 ex c x2 c4 0 c5

1.3 Special Square Matrices

.
Dr

1. Singular and non-singular matrices: If the determinant of a matrix

is zero, it is called singular i.e. for a singular matrix A, |A| = 0.
©

Consider the following matrix for example:

 
1 −1
A=
1 −1

As seen, |A| = −1 + 1 = 0. Therefore, it is a singular matrix.

For a non-singular matrix, |A| =
̸ 0.

2. Symmetric and skew-symmetric matrices: A matrix is said to be

symmetric if the transpose matrix is equal to the matrix itself i.e. for
a symmetric matrix A, AT = A. For example, consider the following
matrix:
   
0 1 T 0 1
A= ⇒A = =A
1 0 1 0
10 CHAPTER 1. MATRICES

If AT = −A, the matrix A is called anti-symmetric or skew-symmetric.

For an example, consider the matrix below:
   
0 −1 T 0 1
A= ⇒A = = −A
1 0 −1 0

Problem 2: Diagonal elements of a skew-symmetric matrix are zero.

Solution: For a skew-symmetric matrix A, AT = −A. In terms of the
ij-th element, aij = −aji .
Now, for the diagonal elements i = j.
Therefore, aii = −aii or, aii = 0 for all i.

al
Problem 3: Any square matrix can be uniquely written as the sum of
a symmetric matrix and a skew-symmetric matrix.

d
Solution: Let A is a square matrix.
1
an 1
A = (A + AT ) + (A − AT ) = P + Q
2 2
M
1 1 T 1
Now, P = (A + A ) = {A + (A ) } = (AT + A) = P
T T T T T
2 2 2
T 1 T T 1 T T T 1 T
P.

Q = (A − A ) = {A − (A ) } = (A − A) = −Q
2 2 2
i.e. P is a symmetric matrix and Q is a skew-symmetric matrix. So
any square matrix can be written as the sum of a symmetric matrix
.
Dr

and a skew-symmetric matrix. To prove the representation unique, we

assume A = R + S where R is a symmetric matrix and S is a skew-
symmetric matrix i.e. RT = R and S T = −S.
©

AT = R T + S T = R − S
1 1
⇒ R = (A + AT ), S = (A − AT )
2 2

3. Hermitian and skew-Hermitian matrices: A matrix is said to be

Hermitian if the Hermitian conjugate matrix is equal to the matrix
itself i.e. for a Hermitian matrix H, H † = H. For example, consider
the following matrix:
 
0 i
H =
−i 0
 T  
† 0 −i 0 i
⇒H = = =H
i 0 −i 0
1.3. SPECIAL SQUARE MATRICES 11

If H † = −H, the matrix H is called anti-Hermitian or skew-Hermitian.

For an example, consider the matrix below:
 
0 i
H =
i 0
 T  
† 0 −i 0 −i
⇒H = = = −H
−i 0 −i 0

Note that,

ˆ Every square matrix can be uniquely represented as the sum of a

Hermitian and a skew-Hermitian matrix.
ˆ The diagonal elements of a skew-Hermitian matrix are either zero

al
or purely imaginary.

d
an
Problem 4: For an arbitrary matrix A, show that A+A† and i(A−A† )
are both Hermitian.
M
Solution: A matrix H is Hermitian if H † = H. Now,

†
P.

A + A† = A† + (A† )†
= A† + A
.

Therefore, A + A† is Hermitian. Similarly,

Dr

†
i A − A† = −i A† + (A† )†
  
©

= −i A† − A

= i(A − A† )

Therefore, i(A − A† ) is also Hermitian.

Note that, (A − A† ) and i(A + A† ) both are skew-Hermitian matrices.

4. Orthogonal matrix: For a unitary matrix O, OOT = OT O = I, the

identity matrix. Consider the following example.
   
0 1 T 0 1
O = ⇒O =
1 0 1 0
    
T 0 1 0 1 1 0
∴ OO = = = I = OT O
1 0 1 0 0 1
12 CHAPTER 1. MATRICES

So, O is a unitary matrix.

Problem 5: Show that the determinant of an orthogonal matrix is ±1.

Solution: For an orthogonal matrix O, OOT = OT O = I, the identity
matrix. Now,

|OOT | = |I| = 1
⇒ |O||OT | = 1
⇒ |O|2 = 1 (∵ |OT | = |O|)
⇒ |O| = ±1.

Problem 6: Show that the inverse of an orthogonal matrix is equal to

al
its transpose i.e. O−1 = OT .

d
Solution: For an orthogonal matrix O, OOT = OT O = I. Since
|O| = ±1, the inverse matrix O−1 exists. Now,
an
O−1 OOT = O−1 I
M
⇒ O−1 = OT (∵ O−1 O = I)
P.

5. Unitary matrix: For a unitary matrix U , U U † = U † U = I, the

identity matrix. Consider the following example.
.
Dr

   
0 i † 0 −i
U = ⇒U =
i 0 −i 0
©

    
† 0 i 0 −i 1 0
∴ UU = = = I = U †U
i 0 −i 0 0 1

Hence, U is a unitary matrix.

Problem 7: Show that the inverse of a unitary matrix is equal to its

Hermitian conjugate i.e. U −1 = U † .
Solution: For a unitary matrix U , U U † = U † U = I. If U −1 is the
inverse matrix of U , U −1 U = I. Now,

U −1 U U † = U −1 I
⇒ U −1 = U †
1.4. EIGENVALUE PROBLEMS 13

6. Self-adjoint matrix: If the transpose of the cofactor matrix i.e. the

adjoint of any arbitrary matrix is equal to the matrix itself, it is called
a self-adjoint matrix i.e. for a self-adjoint matrix adj(A) = A. For
example,
   
−1 0 c −1 0
A= ⇒A =
0 −1 0 −1
 T  
−1 0 −1 0
adj(A) = = =A
0 −1 0 −1

1.4 Eigenvalue Problems

Every square matrix A satisfies a relation

al
AX = λX (1.3)

d
where λ is a scalar (real or complex) and X is a column matrix. Eq. (1.3) is
an
called the eigenvalue equation of matrix A with eigenvalue λ and eigenvector
X. If A is a square matrix of order n, X is a column matrix of order n × 1.
M
From eq. (1.3), (A − λI)X = 0. In terms of the elements of the matrices
A and X,
  
a11 − λ
P.

a12 ... ... a1n x1

 a21
 a22 − λ ... ... a2n    x2 
 

 ... ... ... ... ...   ...  = 0
  (1.4)
.

 ... ... ... ... ...   ... 

Dr

an1 an2 ... ... ann − λ xn

⇒ (a11 − λ)x1 + a12 x2 + ...... + a1n xn = 0

©

a21 x1 + (a22 − λ)x2 + ...... + a2n xn = 0

..................................................... = 0
..................................................... = 0
an1 x1 + an2 x2 + ...... + (ann − λ)xn = 0
Thus we have a set of n number of linear homogeneous equations. Non-
trivial solution exists if the determinant of the coefficients vanishes, i.e.
a11 − λ a12 ... ... a1n
a21 a22 − λ ... ... a2n
D(λ) = ... ... ... ... ... =0
... ... ... ... ...
an1 an2 ... ... ann − λ
⇒ |A − λI| = 0 (1.5)
14 CHAPTER 1. MATRICES

D(λ) is a polynomial of degree n. It is called the characteristics polyno-

mial of the given matrix A.

D(λ) = |A − λI| = 0 (eq. 1.5) is the characteristic equation of the matrix

A. The equation has n roots i.e. n number of possible values of λ - say
λ1 , λ2 , ...., λn (some of them may be equal). Thus we conclude that a matrix
of order n has n number of eigenvalues.
The polynomial D(λ) of degree n can be expressed as

D(λ) = |A − λI| = c0 + c1 λ + c2 λ2 + ..... + cn−1 λn−1 + cn λn (1.6)

which implies that c0 = |A|.

As λ1 , λ2 , ...., λn are the roots of the characteristic equation (eq. 1.5),

al
D(λ) = (λ1 − λ)(λ2 − λ).....(λn − λ) (1.7)

d
By speculation of eq. 1.6 and eq. 1.7 an
λ1 λ2 ....λn = c0 = |A| (1.8)
M
Thus the product of the eigenvalues of a matrix is equal to its determinant.
Similary, by inspection of eq. 1.6 and eq. 1.7 (equating the coefficients of
P.

n−1
λ ) we find

cn−1 = (−1)n−1 (a11 + a22 + .... + ann ) = (−1)n−1 (λ1 + λ2 + .... + λn )

.
Dr

⇒ λ1 + λ2 + .... + λn = a11 + a22 + .... + ann = Tr(A) (1.9)

Thus the sum of the eigenvalues is equal to the trace of the matrix.
©

 
2 −1
Problem 8: Find the trace and determinant of the matrix A =
3 −2
and hence determine its eigenvalues.
Solution: The trace of the matrix is the sum of its diagonal elements i.e.
Tr(A) = 2 − 2 = 0.
The determinant of the matrix is
2 −1
|A| = = −4 + 3 = −1
3 −2

If λ1 and λ2 are the eigenvalues of the matrix, by eq. 1.8 and eq. 1.9 we have

λ1 + λ2 = Tr(A) = 0
λ1 λ2 = |A| = −1
1.4. EIGENVALUE PROBLEMS 15

Solving these equations, we find the eigenvalues as λ1 = −1, λ2 = 1.

How to determine the eigenvalues and the normalized eigenvectors of a

matrix? Let us understand with the following examples.
 
2 −1
Example 1: A =
3 −2
The eigenvalue equation of the matrix is AX = λX or, (A − λI)X =
0 where λ is the eigenvalue and X is the corresponding eigenvector. The
characteristic equation is |A − λI| = 0 i.e.

2−λ −1
= 0
3 −2 − λ

al
⇒ (2 − λ)(2 + λ) + 3 = 0

d
⇒ λ2 = 1
or, λ an
= ±1

Thus the eigenvalues are λ1 = −1 and λ2 = 1.

M
Let, X1 is the eigenvector of A which corresponds to the eigenvalue λ1 =
−1. From the eigenvalue equation AX1 = λ1 X1 , we have
P.

(A − λ1 I)X1 = 0
  
3 −1 x1
⇒ = 0
.

3 −1 x2
Dr

⇒ 3x1 − x2 = 0
or, x2 = 3x1
©

If x1 = a, x2 = 3a where a is an arbitrary number (̸= 0). The eigenvector

 of

a
the given matrix corresponding to the eigenvalue λ1 = −1 is X1 = .
  3a
1
In normalized form, X1n = √110 .
3
Similarly, let us consider X2 as the eigenvector of A corresponding to the
eigenvalue λ2 = 1. From the eigenvalue equation AX2 = λ2 X2 , we have

(A − λ2 I)X2 = 0
  
1 −1 x1
⇒ = 0
3 −3 x2
⇒ x1 − x2 = 0
or, x1 = x2
16 CHAPTER 1. MATRICES

If x1 = b, x2 = b where b is another arbitrary number (̸= 0). Therefore, the

eigenvector
 of the given matrix corresponding  tothe eigenvalue λ2 = 1 is
b 1
X2 = . In normalized form, X2n = √12 .
b   1
0 −1
Example 2: A =
1 0
The eigenvalue equation of the matrix is AX = λX or, (A − λI)X =
0 where λ is the eigenvalue and X is the corresponding eigenvector. The
characteristic equation is |A − λI| = 0 i.e.
−λ 0
= 0
1 −λ
⇒ λ2 + 1 = 0

al
or, λ = ±i

d
Thus the eigenvalues are λ1 = −i and λ2 = i.
an
Let, X1 is the eigenvector of A which corresponds to the eigenvalue λ1 =
−i. From the eigenvalue equation AX1 = λ1 X1 , we have
M
(A − λ1 I)X1 = 0
  
i −1 x1
⇒ = 0
1 i x2
P.

⇒ ix1 − x2 = 0
or, x2 = ix1
.

If x1 = a, x2 = ia where a is an arbitrary number (̸= 0). The eigenvector of

Dr

 
a
the given matrix corresponding to the eigenvalue λ1 = −1 is X1 = .
ia
©

 
1 1
In normalized form, X1n = 2 √ .
i
Similarly, let us consider X2 as the eigenvector of A corresponding to the
eigenvalue λ2 = i. From the eigenvalue equation AX2 = λ2 X2 , we have
(A − λ2 I)X2 = 0
  
−i −1 x1
⇒ = 0
1 −i x2
⇒ x1 − ix2 = 0
or, x1 = ix2
If x2 = b, x1 = ib where b is another arbitrary number (̸= 0). Therefore,
the eigenvector
  of the given matrix corresponding
  to the eigenvalue λ2 = i is
ib i
X2 = . In normalized form, X2n = √12 .
b 1
1.4. EIGENVALUE PROBLEMS 17
 
1 0 0
Example 3: A =  0 0 2 
0 2 0
The eigenvalue equation of the matrix is AX = λX or, (A − λI)X =
0 where λ is the eigenvalue and X is the corresponding eigenvector. The
characteristic equation is |A − λI| = 0 i.e.

1−λ 0 0
0 −λ 2 = 0
0 2 −λ
⇒ (1 − λ)(λ2 − 4) = 0
or, λ = 1, ±2

al
Thus the eigenvalues are λ1 = −2, λ2 = 1 and λ3 = 2.

d
Let, X1 is the eigenvector of A which corresponds to the eigenvalue λ1 =

(A − λ1 I)X1 = 0
an
−2. From the eigenvalue equation AX1 = λ1 X1 , we have
M
  
3 0 0 x1
⇒  0 2 2   x2  = 0
P.

0 2 2 x3
⇒ 3x1 = 0 & x2 + x3 = 0
or, x1 = 0 & x2 = −x3
.
Dr

If x3 = a, x2 = −a where a is an arbitrary number (̸= 0). The eigenvector

 of

0
©

the given matrix corresponding to the eigenvalue λ1 = −2 is X1 =  −a .

  a
0
In normalized form, X1n = √12  −1 .
1
Similarly, let us consider X2 as the eigenvector of A corresponding to the
eigenvalue λ2 = 1. From the eigenvalue equation AX2 = λ2 X2 , we have

(A − λ2 I)X2 = 0
  
0 0 0 x1
⇒  0 −1 2   x2  = 0
0 2 −1 x3
⇒ −x2 + 2x3 = 0 & 2x2 − x3 = 0
or, x2 = x3 = 0
18 CHAPTER 1. MATRICES

Let x1 = b, an arbitrary number (̸= 0). The eigenvector

  of the given matrix
b
corresponding to the eigenvalue λ2 = 1 is X2 =  0 . In normalized form,
  0
1
X2n =  0 .
0
If X3 is the eigenvector of A corresponding to the eigenvalue λ3 = 2, from
the eigenvalue equation AX3 = λ3 X3 we have

(A − λ3 I)X3 = 0
  
−1 0

al
0 x1
⇒  0 −2 2   x2  = 0

d
0 2 −2 x3
an
⇒ x1 = 0 & x2 = x3
M
Let x2 = x3 = c, where c is an arbitrary number (̸= 0). The eigenvector
  of
0
the given matrix corresponding to the eigenvalue λ3 = 2 is X3 = c . In
P.


  c
0
1 
normalized form, X3n = 2√ 1 .
.
Dr

1
©

1.4.1 Corollaries

1. Eigenvalues of a diagonal matrix are equal to its diagonal elements.

Proof: Consider a diagonal matrix of order n:

 
a11 0 ... ... 0

 0 a22 ... ... 0 


 ... ... ... ... ... 

 ... ... ... ... ... 
0 0 ... ... ann
1.4. EIGENVALUE PROBLEMS 19

The characteristic equation of the matrix is

|D − λI| = 0
a11 − λ 0 ... ... 0
0 a22 − λ ... ... 0
⇒ ... ... ... ... ... = 0
... ... ... ... ...
0 0 ... ... ann − λ
⇒ (a11 − λ)(a22 − λ)...(ann − λ) = 0

i.e. λ = a11 , a22 , ..., ann , the diagonal elements of the matrix.

2. At least one eigenvalue of a singular matrix is zero.

al
Proof: Consider a singular matrix A i.e. |A| = 0.

d
If λ1 , λ2 , λ3 , ... are the eigenvalues of the matrix A, the product of the
an
eigenvalues must be equal to the determinant of A (eq. 1.8) i.e.

λ1 .λ2 .λ3 ... = |A| = 0

M
Therefore, at least one of the eigenvalues must be zero.
P.

3. If λ is the eigenvalue of a non-singular matrix A, the eigenvalue of A−1

is 1/λ corresponding to a given eigenvector.
Proof: Let λ and λ′ are respectively the eigenvalues of a non-singular
.
Dr

matrix A and its inverse matrix A−1 corresponding to the same eigen-
vector X. The eigenvalue equations are
©

AX = λX
A X = λ′ X
−1

Now multiplying the first equation by A−1 from left

A−1 AX = λA−1 X
⇒ X = λλ′ X
or, (1 − λλ′ )X = 0

Since X is the eigenvector (X ̸= 0), λ′ = 1/λ i.e. the eigenvalues of

the inverse matrix are the reciprocal of the eigenvalues of the original
matrix.

4. Eigenvalues of a unitary matrix are of unit magnitude.

20 CHAPTER 1. MATRICES

Proof: Consider a unitary matrix U having an eigenvalue λ correspond-

ing to an eigenvector X. The eigenvalue equation is
U X = λX (1.10)
Taking the Hermitian conjugate of eq. 1.10
(U X)† = (λX)†
⇒ X † U † = λ∗ X † (1.11)
Multiplying eq. (1.11) by eq. (1.10) from left
X †U †U X = λ∗ λX † X
⇒ X †X = λ∗ λX † X (∵ U is unitary, U † U = I)

al
or, (1 − |λ|2 )X † X = 0
⇒ 1 − |λ|2 0 (∵ X † X ̸= 0)

d
=
⇒ |λ|2 = 1 an
5. The eigenvalues of a Hermitian matrix are real and the eigenvectors
M
corresponding to different eigenvalues are orthogonal.
Proof: Let us consider a Hermitian matrix H having an eigenvalue λ
P.

corresponding to an eigenvector X. The eigenvalue equation is

HX = λX (1.12)
.

Multiplying eq.(1.12) by X † from left

Dr

X † HX = λX † X (1.13)
©

Taking the hermitian conjugation of eq. (1.12),

(HX)† = (λX)†
⇒ X † H † = λ∗ X †
or, X † H = λ∗ X † (∵ H is Hermitian, H † = H) (1.14)
Multiplying eq.(1.14) by X from right
X † HX = λ∗ X † X (1.15)
Comparing eq.(1.13) and eq.(1.15),
λX † X = λ∗ X † X
⇒ (λ − λ∗ )X † X = 0
⇒ λ∗ = λ (∵ X † X ̸= 0)
1.4. EIGENVALUE PROBLEMS 21

Thus the eigenvalues of a Hermitian matrix are real.

Now consider two distinct eigenvalues λ1 and λ2 of the Hermitian
matrix H corresponding to the eigenvectors X1 and X2 respectively.
λ∗1 = λ1 , λ∗2 = λ2 and λ1 ̸= λ2 . The eigenvalue equations are
HX1 = λ1 X1 (1.16)
HX2 = λ2 X2 (1.17)

Taking the hermitian conjugation of eq. (1.17),

(HX2 )† = (λX2 )†
⇒ X2† H = λ2 X2† (∵ H † = H & λ∗2 = λ2 ) (1.18)
Multiplying eq.(1.18) by X1 from right

al
X2† HX1 = λ2 X2† X1 (1.19)

d
Multiplying eq.(1.16) by X2† from left
an
X2† HX1 = λ1 X2† X1 (1.20)
M
Comparing eq. 1.19 and eq. 1.20,
λ1 X2† X1 = λ2 X2† X1
⇒ (λ1 − λ2 )X2† X1 = 0
P.

⇒ X2† X1 = 0 (∵ λ2 ̸= λ1 )
Thus X1 and X2 are orthogonal.
.
Dr

6. If two matrices commute, they will have simultaneous eigenfunction.

Proof: Let two matrices A and B commute i.e. AB = BA.
©

If X is an eigenvector of A and λ is the associated eigenvalue, AX =

λX.
Multiplying by B from left,
BAX = λBX
or, ABX = λBX (∵ AB = BA)
⇒ A(BX) = λ(BX)
Thus BX is another eigenfunction of A for the same eigenvalue λ. BX
is therefore, a scalar multiple of X i.e.
BX = µX (1.21)
This is the eigenvalue equation of matrix B with eigenvalue µ and
associated eigenfunction X. Thus X is the simultaneous eigenfunction
for the matrices A and B.
22 CHAPTER 1. MATRICES

1.5 Cayley-Hamilton Theorem

Cayley-Hamilton theorem states that every square matrix satisfies its own
characteristic equation. Let us consider a matrix A of order n. If λ is the
eigenvalue of A, the characteristic equation is D(λ) = |A − λI| = 0. We
rewrite eq. 1.6 as

n
X
2 n
D(λ) = c0 + c1 λ + c2 λ + ..... + cn λ = ci λ i (1.22)
i=0

is a polynomial of order n. The Cayley-Hamilton theorem states that sub-

stituting the matrix A for λ in this polynomial (eq. 1.22) results in the null

al
matrix i.e.
n

d
X
D(A) = ci Ai = 0 (1.23)
i=0 an
The theorem can be verifiedwith the 
following example.
M
2 −1
Consider a matrix A = . The characteristic equation of the
3 −2
P.

matrix is
.

2−λ −1
Dr

D(λ) = =0
3 −2 − λ
⇒ D(λ) = λ2 − 1 = 0 (1.24)
©

By Cayley-Hamilton theorem, the characteristic eq. 1.24 will be satisfied by

the matrix A i.e. D(A) = A2 − I = 0 or, A2 = I. Now,
    
2 2 −1 2 −1 1 0
A = = = I (q.e.d.)
3 −2 3 −2 0 1

Cayley-Hamilton theorem is often used to determine the inverse of a ma-

trix. Given the characteristic equation (eq. 1.22) of a matrix A, Cayley-
Hamilton theorem implies

n
X
D(A) = ci Ai = c0 I + c1 A + c2 A2 + ..... + cn An = 0 (1.25)
i=0
1.5. CAYLEY-HAMILTON THEOREM 23

Multiplying eq. 1.25 by A−1 ,

c0 IA−1 + c1 AA−1 + c2 A2 A−1 + ..... + cn An A−1 = 0

or, c0 A−1 + c1 I + c2 A + ..... + cn An−1 = 0
or, c0 A−1 = −(c1 I + c2 A + ..... + cn An−1 )
n
1 1 X
⇒ A−1 n−1
= − (c1 I + c2 A + ..... + cn A ) = − ci Ai−1
c0 c0 i=1


2 0 1
Problem 9: Determine the inverse of the matrix A =  1 1 2  by
0 1 1

al
using the Cayley-Hamilton theorem.
Solution: The characteristic equation of the matrix is

d
2−λ
1
0
1−λ
1
2
an = 0
0 1 1−λ
M
⇒ (2 − λ){(1 − λ)2 − 2} + 1 = 0
⇒ λ3 − 4λ2 + 3λ + 1 = 0 (1.26)
P.

By Cayley-Hamilton theorem, eq. 1.26 will be satisfied by the matrix A itself

i.e.
.

A3 − 4A2 + 3A + I = 0 (1.27)
Dr

Multiplying eq. 1.27 by A−1 ,

©

A3 A−1 − 4A2 A−1 + 3AA−1 + IA−1 = 0

⇒ A−1 = −(A2 − 4A + 3I)
    
2 0 1 2 0 1 4 1 3
Now, A2 =  1 1 2   1 1 2  =  3 3 5 
0 1 1 0 1 1 1 2 3
     
4 1 3 2 0 1 1 0 0
∴ A−1 = −A2 + 4A − 3I = −  3 3 5  + 4  1 1 2  − 3  0 1 0 
1 2 3 0 1 1 0 0 1
 
1 −1 1
=  1 2 3 
−1 2 2
24 CHAPTER 1. MATRICES

1.6 Diagonalization of Matrices

A diagonal matrix corresponding to a square matrix is a matrix of same order
having its diagonal elements as the eigenvalues of the original matrix and all
other elements are zero. For example, consider
 
a11 a12 ... ... a1n
 a21 a22 ... ... a2n 
 
A=  ... ... ... ... ... 

 ... ... ... ... ... 
an1 an2 ... ... ann

and the eigenvalues of A are λ1 , λ2 , ... λn . Therefore, the diagonal matrix of

A is

al
 
λ1 0 ... ... 0

d
 0 λ2 ... ... 0 
 
D=  ... ... ... ... ... 
an 
 ... ... ... ... ... 
0 0 ... ... λn
M
If A has n number of linearly independent eigenvectors, a matrix S can be
found such that S −1 AS = D, the diagonal matrix. The matrix S is called
the diagonalizing matrix.
P.

Let, X1 , X2 , ... Xn are the linearly independent eigenvectors of A. Thus

the diagonalizing matrix
.
Dr

 
x11 x12 ... ... x1n
 x21 x22 ... ... x2n 
 
S = (X1 X2 ...Xn ) =  ... ... ... ... ... 
©

 
 ... ... ... ... ... 
xn1 xn2 ... ... xnn

where we denote the eigenvectors Xi by the column matrices having elements

x1i , x2i , ..., xni .
Note that the diagonalizing matrix S is not unique as we could arrange
the eigenvectors X1 , X2 , ... Xn in any order to construct it. The following
steps may be followed to diagonalize a matrix:

ˆ Find the eigenvalues of original matrix.

ˆ Find out corresponding eigenvectors. The eigenvectors must be lin-

early independent. Otherwise, the matrix will not be diagonalizable
(Linearly independent eigenvectors may exit for a matrix having de-
generate eigenvalues).
1.6. DIAGONALIZATION OF MATRICES 25

ˆ Construct the diagonalizing matrix S with its column elements as the

linearly independent eigenvectors.
ˆ Determine the inverse matrix S −1 .

ˆ The matrix D = S −1 AS is the diagonal matrix with λ1 , λ2 , ... λn as its

successive diagonal elements, where λi is the eigenvalue corresponding
to the eigenvector Xi .

Problem 10: Diagonalize the matrix

 
2 −1
A=
3 −2

al
Solution: Note that the eigenvalues
  of A are λ1 =−1, λ2 = 1. Corre-

d
1 1
sponding eigenvectors are X1 = and X2 = respectively. The
3 an 1
eigenvectors are linearly independent2 .  
1 1
M
Thus the diagonalizing matrix S =
 3 1
1 −1
The inverse matrix S −1 = − 12
−3 1
P.

Therefore, the diagonal matrix

     
−1 1 1 −1 2 −1 1 1 −1 0
D = S AS = − =
.

2 −3 1 3 −2 3 1 0 1
Dr
©

1.6.1 Corollaries
1. Diagonalizing matrix of a real symmetric matrix is orthogonal.
Proof: Let us consider a symmetric matrix A i.e. AT = A. If λi are
the eigenvalues of A and S is the diagonalizing matrix,
S −1 AS = D = diag(λ1 , λ2 , ...λn )
⇒ (S −1 AS)T = DT
⇒ S T AT (S −1 )T = D
⇒ S T A(S −1 )T = S −1 AS
⇒ ST = S −1
⇒ ST S = I
2
The task is left for the readers
26 CHAPTER 1. MATRICES

i.e. S is an orthogonal matrix.

2. Diagonalizing matrix of a Hermitian matrix is unitary.

Proof: Let us consider a Hermitian matrix H i.e. H † = H. If D is the
diagonal matrix and S is the corresponding diagonalizing matrix,

S −1 HS = D
⇒ (S −1 HS)† = D†
⇒ S † H † (S −1 )† = D
⇒ S † H(S −1 )† = S −1 HS
⇒ S† = S −1
⇒ S †S = I

al
i.e. S is unitary.

d
1.7 Similarity Transformation
an
M
Consider a square matrix A of order n and a non-singular matrix S such
that S −1 AS = B, another square matrix of same order as A. Matrix B
P.

is similar to A and the transformation from A to B through the relation

S −1 AS = B is called similarity transformation. Diagonalization is a special
type of similarity transformation.
.
Dr

Problem 11: Eigenvalues of a matrix remain invariant under similarity

transformation
Solution: Consider a similarity transformation S −1 AS = B. If λ is the
©

eigenvalue of B, the characteristic equation is |B − λI| = 0 i.e.

|S −1 AS − λI| = 0
⇒ |S −1 AS − S −1 λIS| = 0
⇒ |S −1 (A − λI)S| = 0
⇒ |S −1 ||A − λI||S| = 0
⇒ |S −1 S||A − λI| = 0
⇒ |A − λI| = 0

which is the characteristic equation of the original matrix A with same eigen-
value λ. Thus the eigenvalues remain invariant under similarity transforma-
tion.
1.8. UNITARY TRANSFORMATION 27

1.8 Unitary Transformation

The similarity transformation may be done by a unitary matrix U . The
transformation U −1 AU = B is called unitary transformation. Since for a
unitary matrix U , U −1 = U † ; the unitary transformation may be defined as
B = U † AU .

Problem 12: A Hermitian matrix remains Hermitian under unitary

transformation.
Solution: Let A is a Hermitian matrix i.e. A† = A.
The unitary transformation matrix B = U −1 AU = U † AU where U is a
unitary matrix (U U † = I & U −1 = U † ).

al
B = U † AU

d
⇒ B † = (U † AU )†
= U † A† (U † )†
an
= U † AU (∵ A† = A)
M
= B
P.

Thus B is Hermitian if A is Hermitian.

Problem 13: The norm of a matrix 3 remains unchanged under the
unitary transformation.
.

Solution: Consider a matrix A and its unitary transformation matrix

Dr

B = U −1 AU , where U is a unitary matrix i.e. U U † = U † U = I or U −1 = U † .

Now,
©

B † = (U −1 AU )† = (U † AU )† = U † A† U

Multiplying the above equation by B = U † AU from right

B † B = U † A† U U † AU = U † A† AU
⇒ |B † B| = |U † ||A† A||U | = |U † U ||A† A| = |A† A|

Thus the norm of the matrix remains invariant under the unitary transfor-
mation.

3
Norm of a matrix A is defined as |A† A|. Note that, if |X † X| = 1 for a column matrix
X it is said to be normalized.
28 CHAPTER 1. MATRICES

1.9 Evaluating Power of a Matrix

Consider diagonalization of a matrix A by the matrix S: S −1 AS = D or,
A = SDS −1 . For a function f (A) of matrix A, we have

f (A) = Sf (D)S −1 , (1.28)

where f (D) is similar function of D. Thus from eq. 1.28, for any power An
of matrix A
An = SDn S −1 (1.29)

 
2 −1
Problem 14: A = . Find A50 .

al
3 −2
 
−1 0

d
Solution: Refer to Problem 10. The diagonal matrix D =
0 1
and the diagonalising matrix S =

1 1
3 1

an
. The inverse matrix S −1 =
M
 
1 1 −1
−2
−3 1
P.

∴ By eq. 1.29,
  50  
50 50 −1 1 1 −1 −1 0 1 1
A = SD S = −
.

2 −3 1 0 1 3 1
Dr

   
1 1 −1 1 0 1 1
= −
2 −3 1 0 1 3 1
©

 
1 0
=
0 1

1.10 Solutions of Linear Coupled First Order

Ordinary Differential Equations
Consider the following pair of linear coupled first order differential equations:

y1′ (t) = a11 y1 (t) + a12 y2 (t)

y2′ (t) = a21 y1 (t) + a22 y2 (t)
1.10. SOLUTIONS OF LINEAR COUPLED FIRST ORDER ORDINARY DIFFERENTIAL EQUAT

The equations, in matrix form, can be represented as

 ′    
y1 a11 a12 y1
= i.e. Y ′ = AY
y2′ a21 a22 y2
 
c1
Let the boundary conditions are y1 (0) = c1 and y2 (0) = c2 i.e. Y (0) =
c2
Steps:
ˆ Determine the eigenvalues λ1 , λ2 , ... of the matrix A.

ˆ Find out the corresponding eigenvectors X1 , X2 , ...

X
ˆ The solutions of the coupled equations can be written as Y (t) = ai eλi t Xi ,

al
i
where ai are arbitrary constants.

d
ˆ Applying the boundary conditions ai can be determined and exact so-
lution is obtained.
an
Let us consider the following set of equations:
M
y1′ = 2y1 + 3y2
y2′ = 4y1 + y2
P.

The given initial conditions are y1 (0) = 2, y2 (0) = 1. The equations, in

matrix form, can be represented as
.
Dr

 ′    
y1 2 3 y1
= i.e. Y ′ = AY
y2′ 4 1 y2
©

 
2 3
where A = . The eigenvalues of the matrix A are λ1 = −2 and
4 1    
3 1
λ2 = 5. Corresponding eigenvectors are X1 = and X2 =
−4 1
respectively.
Thus, the general solutions are
X
Y (t) = ai eλi t Xi
i
     
y1 (t) −2t 3 5t 1
⇒ = a1 e + a2 e
y2 (t) −4 1
or, y1 (t) = 3a1 e−2t + a2 e5t
and y2 (t) = −4a1 e−2t + a2 e5t
30 CHAPTER 1. MATRICES

Applying the initial conditions,

y1 (0) = 3a1 + a2 = 2
y2 (0) = −4a1 + a2 = 1

Solving these equations, we have a1 = 1 and a2 = −1. Hence, the exact

solutions are

y1 (t) = 3e−2t − e5t

y2 (t) = −4e−2t − e5t

d al
an
M
. P.
Dr
©
Bibliography

[1] T. L. Chow, Mathematical Methods for Physicists: A Concise Introduc-

tion Cambridge University Press (2000)

[2] Charlie Harper, Introduction to Mathematical Physics Prentice Hall India

al
(1978)

d
[3] B. S. Rajput, Mathematical Physics Pragati Prakashan (2014)
an
M
. P.
Dr
©

31