Unit 7 Matrices-I

UNIT 7
MATRICES-I
Structure

7.1 Introduction 7.4 Orthonormal Vectors

Expected Learning Outcomes Orthonormal Bases
7.2 Linear Equations and Matrices Construction of Orthonormal Bases
Definition of a Matrix Signature of the Metric
Multiplication of Matrices 7.5 Complex Vector Spaces: Hilbert
7.3 Various Types of Matrices Space
Transpose of a Matrix and Symmetric Hermitian Inner Product
Matrix Pythagoras Theorem and Bessel’s
Complex Conjugate and Hermitian Adjoint Inequality
Inverse, Orthogonal and Unitary Matrices Schwarz and Triangle Inequalities
Functions of a Matrix Orthonormal Bases
7.6 Summary
7.7 Terminal Questions
7.8 Solutions and Answers

7.1 INTRODUCTION
In the previous unit, you have studied concepts related to vector spaces. In
this unit, you will learn basic concepts related to matrices most of which you
may already know from your UG physics. Still we have presented these
concepts in some detail as these are used extensively in physics courses.
In Sec. 7.2, we introduce matrices that arise when we are solving systems of
linear equations. We define matrices and then explain multiplication of
matrices. Then in Sec. 7.3, we define different types of matrices that you may
already be familiar with such as symmetric matrix, hermitian adjoint,
orthogonal and unitary matrices. You will learn about the transpose and
inverse of a matrix as well. Next, we deal with matrix algebra in Sec. 7.4 and
explain elementary operations on matrices. In Sec. 7.5, we discuss
determinants and their properties along with proofs of some important results.
Finally, in Sec. 7.6, we discuss linear operators and matrices. In the next unit,
we continue the discussion on matrices and solve eigenvalue problems for
different types of matrices with examples.
Expected Learning Outcomes
After studying this unit, you should be able to:
 define real and complex matrices, determine their sum, difference,
products with a number and another matrix; 33
Block 2 Vector Spaces, Matrices and Tensors
 obtain the complex conjugate, transpose and inverse of a matrix;
 define symmetric, hermitian, orthogonal and unitary matrices and function
of a matrix;
 solve problems on commutator algebra and partitioning of matrices;
 define a determinant and determine its value;
 use the properties of determinants to solve problems; and
 establish the connection between linear operators and matrices.

7.2 LINEAR EQUATIONS AND MATRICES

The rectangular array of real or complex number that we call a ‘matrix’ occurs
naturally when we deal with linear equations. Take the simple example of
You have studied linear equations for two unknown quantities x 1 and x 2 :
these preliminaries
a11 x1  a12 x 2  c1 (7.1a)
in your school and
UG courses in a21 x1  a22 x 2  c 2 (7.1b)
physics and
mathematics. where coefficients a’s and c’s are given numbers. We could have written these
equations also as:
ax1  bx 2  c

dx1  cx 2  f

But the convention to write the coefficients as a11, a12 ,... etc. has many
advantages as we shall see.

Now we ask the question: What can we say about the solutions of the
unknown variables x 1 and x2 ?

We try to eliminate x 2 from Eqs. (7.1a and b): multiply the first equation by
a22 and the second by a12 :

a11 a22 x1  a12 a22 x 2  a22 c1 (7.2a)

a12 a 21 x1  a12 a 22 x 2  a12 c 21 (7.2b)

and subtract Eq. (7.2b) from Eq. (7.2a), so that

 a11a22  a12 a21  x1  a22 c1  a12 c 2 (7.2c)
or, if  a11 a22  a12 a21   0, then

x1 
 a22 c1  a12 c2 
 a11 a22  a12 a21   b11 c1  b12 c2 (7.2d)

where
a22
b11 
 a11 a22  a12 a21 
 a12
b12 
34
and
 a11 a22  a12 a21  (7.2e)
Unit 7 Matrices-I
Similarly, we can calculate the other unknown x 2 as:
 a11 c 2  a21 c1 
x2   b21c1  b22 c 2
 a11 a22  a12 a21  (7.2f)
where
 a21
b21 
 a11 a22  a12 a21 
a11
and b22 
 a11 a22  a12 a21  (7.2g)

Eqs. (7.1a and b) and their solutions are written in the convenient matrix
notation as:
 a11 a12   x1   c1 
      , (7.3a)
 a21 a22   x 2  c 2 
 x1   b11 b12   c1 
     (7.3b)
 x 2  b22 b22  c 2 

Or even more briefly as:

AX  C, X  BC (7.4a)

Where A and B are matrices given by:

 a11 a12   b11 b12 
A  , B   (7.4b)
 a21 a22   b21 b22 
The matrix B with elements bij , i , j  1, 2 is called the inverse of the matrix A
and written as:
B  A 1 (7.4c)
and the quantity
A  a11 a22  a12 a21 (7.4d)

is called the determinant of the matrix A. The solutions show us that

a a a a
b11  22 , b12   12 , b21   21 and b22  11
A A A A
(7.5)
This was a brief revision of basic concepts that you already know from school
and UG courses. We will discuss all these concepts in detail now.
7.2.1 Definition of a Matrix
A set of n  m real or complex numbers t ij , i  1,..., n j  1,..., m when written
as follows, in a rectangular form as n rows and m columns:
t 11 t 12  t 1m 
 
t 21 t 22  t 2m 
T   (7.6)
   
 
t  t nm 
 n1 t n 2 35
Block 2 Vector Spaces, Matrices and Tensors
is called an n  m real matrix if t ij are real, and a complex matrix if t ij are
complex numbers. The two indices i and j are called the row index and the
column index, respectively. The special case when there is only one column,
(the n  1 matrix) is also called a column vector:
 x1 
 
x2 
X   (7.7a)

 
x 
 n

Similarly, a 1 m matrix with just one row, is called a row vector:

Y   y 1, y 2 ,..., y m  (7.7b)

The matrix having all elements equal to zero is called the zero matrix. Two
matrices A  [aij ] and B  [bij ] are said to be equal if and only if:

aij  bij for all i and j (7.8a)

If A  [aij ] and B  [bij ] are two matrices of the same order, say, n  m
matrices, then their sum (A  B) and difference (A  B) are the matrices C
 [cij ] and D  [dij ], respectively, such that:

cij  aij  bij (7.8b)

and dij  aij  bij (7.8c)

Matrix addition is commutative and associative:

AB BA (7.8d)

A  (B  C )  ( A  B )  C (7.8e)

We define the multiplication of a matrix A by a real or complex number c by

simply multiplying every element of it by that number:
c Aij  cAij (7.9)

Let us now revise matrix multiplication.

7.2.2 Multiplication of Matrices

Matrix multiplication also appears naturally in linear equations. Suppose we
have three variables and two equations:
z1  a11 x1  a12 x 2  a13 x 3 (7.10a)

z2  a21 x1  a22 x2  a23 x3 (7.10b)

And, if there are, say, four variables y 1 to y 4 in terms of which x’s can be
written as:
x1  b11y 1  ...  b14 y 4 (7.11a)

x2  b21 y1  ...  b24 y 4 (7.11b)

x 3  b31 y 1  ...  b34 y 4 (7.11c)

36
Unit 7 Matrices-I
then
z1  a11  b11y 1  ...  b14 y 4   a12  b21y 1  ...  b24 y 4  

...  a13 b31 y1  ...  b34 y 4 

  a11b11  a12 b21  a13 b31  y 1  ...

  a11 b14  a12 b24  a13 b34  y 4

 c11 y 1  c12 y 2  c13 y 3  c14 y 4

and similarly,
z2  c21 y1  c22 y 2  c23 y 3  c24 y 4

We can define c ij for i  1, 2 and j  1, ...,4, in a succinct notation as follows:

3
c ij   aik bkj (7.12)
k 1

The above equation shows us the matrix multiplication rule. Here it is a

product of two matrices: one of them 2  3 matrix of a’s and the other a 3  4
matrix of b’s giving as a result a 2 4 matrix of c’s. Let us write it as:

 b11 b12 b13 b14 

c11 c12 c14  a11 a12 a13  
b24 
c13
   b21 b22 b23
c21 c22 c23 c24  a21 a22 a23  
b31 b32 b33 b34 
(7.13)
The multiplication rule can be summarized as follows:
1. Matrix multiplication is defined only if the number of columns of the first
matrix is equal to the number of rows of the second matrix.
2. To calculate the (i, j ) element of the product matrix, we take the i-th row of
the first matrix and j-th column of the second matrix and add the products
of the corresponding elements.
We can imagine the calculation for two matrices given above as follows. For
example, to calculate (24) element of the product, the 2-nd row of the first
matrix is taken, rotated by a right angle and brought next to the 4-th column of
the second matrix, and multiplied element by element and added to give the
(24) element:
 a 21   b14 
   
a21, a 22 , a 23   a 22   b24 
   
a 23  b34 

 a21b14  a22 b24  a23 b34  c 24

Actually, the equations for variables z1, z 2 , x1, x 2 , x 3 and y 1,..., y 4 given above
in this section are matrix equations too: first, as the product of 2  3 and 3 1
to give 2  1 matrix of z' s :
 x1 
 z1  a11 a12 a13   
     x 2  (7.14a)
z 2  a 21 a 22 a 23   
 x 3 
37
Block 2 Vector Spaces, Matrices and Tensors
and then as a product of 3  4 of b’s to a matrix 4  1 to give a 3 1 matrix of
the x' s :
 y1 
 x1   b11 b12 b13 b14   
    y 2 
 x 2   b21 b22 b23 b24    (7.14b)
    y 3 
 x 3  b31 b32 b33 b34   
y 
 4
If we substitute the matrix for x' s from Eq. (7.14b) in Eq. (7.14a), then we get:
 y1 
 b11 b12 b13 b14   
 z1  a11 a12 a13    y 2 
   b21 b22 b23 b24    (7.15a)
z 2  a 21 a 22 a 23    y 3 
b31 b32 b33 b34   
y 
 4
 y1 
 
c11 c12 c13 c14   y 2 
    (7.15b)
c 21 c 22 c 23 c 24   y 3 
 
y 
 4
Now that we have defined matrices and explained elementary operations on
them, we will discuss various types of matrices in the next section.

7.3 VARIOUS TYPES OF MATRICES

A matrix is called a square matrix, if the number of rows in it is the same as
In the discussion in
the number of columns. A matrix of n rows and n columns is called a square
Sec. 7.3.1 and 7.3.4,
we largely consider matrix of size n.
square matrices. For a square matrix A of size n, the elements A11, A22 ,..., Ann are called the
diagonal elements of the matrix.
If all elements of a matrix are equal to zero, it is called the zero matrix.
If all elements away from the diagonal of a square matrix are equal to zero, it
is called a diagonal matrix. (Some, though not all, elements of the diagonal
can possibly be zero too.) Such a matrix looks like
d 11 0  0 
 
 0 d 22  0 
D  (7.16)
    
 
 0  d nm 
 0
The identity matrix or the unit matrix is the diagonal matrix in which every
diagonal element is equal to 1. We denote the identity matrix by the symbol 1:
1 0  0
 
0 1  0
1   (7.17a)
   
 
0 0  1
38  
Unit 7 Matrices-I
The components of the unit matrix can be written as:
1   iJ , i , j  1, 2, ..., n (7.17b)
where the symbol  ij called the Kronecker Delta is defined as:
 ij  0, i  j , and 11   22  ...   nn  1 (7.17c)

Sometimes it is useful to also denote the size of the identity matrix. Then we
write it as 1n. The identity matrix when multiplied to any square matrix of the
same size does not change that matrix:
A1  A, 1A  A (7.17d)
With these preliminary definitions, we discuss the transpose of a matrix and
the symmetric matrix in the next section.
7.3.1 The Transpose of a Matrix and Symmetric Matrix
We have already defined real and complex matrices, as well as the column
and row vector matrices.
The transpose AT of an n  m (real or complex) matrix A is an m  n matrix in
which the rows of A become columns of AT :
AT  ij  A ji (7.18)

Thus if, for example,

a11 a12 a13 
A  (7.19a)
a21 a22 a23 
then
 a11 a 21 
 
AT  a12 a 22  (7.19b)
 
a13 a 23 

The interesting fact about transpose of a matrix is that if we take the transpose
of the product of two matrices, then the result is the product of the transpose
of the two matrices in reverse order:
C  AB, C T  B T AT (7.20a)
This is so because, for example, the (i, j)-element of C is
C ij   Aik Bkj (7.20b)
k

therefore,
CT  ji  C ij   Aik Bkj   Bkj Aik   BT  jk AT  ki
k k k
(7.20c)
A symmetric matrix is the matrix, which is equal to its transpose:
AT  A (symmetric) (7.21a)
On the other hand, if the transpose is equal to its negative, then it is called
anti-symmetric:
AT   A, (anti-symmetric) (7.21b)
REMEMBER: A symmetric or anti symmetric matrix is a square matrix. 39
Block 2 Vector Spaces, Matrices and Tensors
7.3.2 Complex Conjugate and Hermitian Adjoint

As the name indicates, the complex conjugate A of a matrix is one in which

the elements are the complex conjugate of the elements of A:

A  ij  Aij  (7.22)

The Hermitian adjoint of a matrix is the transpose of the complex conjugate

of the matrix, which is the same as the complex conjugate of the transpose of
the matrix. It is denoted by A† :

 A † ij  A     A ji   AT  

T 
(7.23)
  ij   ij

A matrix A is called Hermitian if it is equal to its Hermitian adjoint:

A  A† (7.24)

Note that a Hermitian matrix is also a square matrix.

7.3.3 Inverse, Orthogonal and Unitary Matrices

Given a square matrix A, another square matrix of the same size, denoted by
A 1 is called the inverse of the matrix A if

AA 1  A 1A  1 (7.25)

It is not necessary that for any square matrix A, the inverse matrix exists.

Just as the inverse of a real or complex number exists if and only if it is not
zero, similarly, the inverse of a matrix exists if and only if its determinant A is
not zero. You are familiar with determinants from school and UG courses.
However, we will explain the concept in Sec. 7.5.

As an example, we take the simple matrix:

1 1
A  
1 1

a b
Assuming that A 1   
c d 

exists and is such that

1 1 a b 1 0
AA 1         1
1 1 c d  0 1

we get a contradiction that 1 0! You should verify this result by multiplying

the two matrices before proceeding further. Therefore, no such inverse matrix
exists for this matrix A.

An orthogonal matrix R is a real square matrix such that

RRT  RT R  1 (7.26)
40
Unit 7 Matrices-I
Therefore, the inverse of an orthogonal matrix is always equal to its
transpose matrix.

A unitary matrix U is a complex square matrix such that

UU †  U †U  1 (7.27)

Therefore, the inverse of a unitary matrix is equal to its Hermitian adjoint.

As real numbers are a subset of complex numbers, a real Hermitian matrix is
a symmetric matrix and a real unitary matrix is orthogonal.
We will discuss these matrices in detail in the next unit.
7.3.4 Functions of a Matrix
Given a square matrix A of size n, i.e., an n  n matrix, we can multiply it to
itself any number of times, as well as multiply it by constants. Therefore, we
can construct a polynomial of a matrix, which is itself a matrix of the same
size. Thus, for example,

M  a0 1  a1 A  ...  ar Ar (7.28a)

is a polynomial. Here 1 is the unit matrix of size n and

Ar  A A ... A , r factors (7.28b)

We can also define convergent series in powers of the matrix A. The most
important and useful of these convergent series is the exponential:

exp A  1  A 
1 2 1
A  ...  Ar  ... (7.28c)
2! r!
The discussion so far was essentially a revision of what you have studied in
your UG courses. We will now discuss matrix algebra.

7.4 MATRIX ALGEBRA

Let L (n) be the collection or set of square matrices of the same fixed size n,
with members like A, B, C … etc. Then the sum and multiplication by a
number of these matrices is again a matrix of the same type. Therefore, L (n )
forms a vector space.
Not just that. We can multiply any two matrices of L (n ) to get another matrix
in L (n ). This additional structure makes L (n ) an algebra.

This algebra is associative, because for any matrices A, B, C  L (n ) :

A(BC )  ( AB )C (7.29)

But the algebra is not commutative because, in general,

AB  BA (7.30)
The difference of the two products
AB  BA (7.31)
is called the commutator and is written as
A, B   AB  BA (7.32) 41
Block 2 Vector Spaces, Matrices and Tensors
You may like to pause now and attempt an SAQ.

SAQ 1
a) What is the dimension of the real and complex vector spaces L (n ) when
the matrices are all (i) real matrices and (ii) complex matrices? Construct a
basis for these two cases.
b) Show by giving an example of 2  2 Hermitian matrices A and B that their
multiplication is not Hermitian.

7.4.1 Elementary Operations

Given a square matrix

 b11 b12 b13 

 
B  b21 b22 b23 
 
b31 b32 b33 

we can interchange the first and second row and obtain another matrix:

b21 b22 b23 

~  
B   b11 b12 b13 
 
b31 b32 b33 

This operation of interchanging the rows is called an elementary operation. It

is interesting that this interchange can be done by multiplying B on the left
(also called pre-multiplication) by the matrix:

0 1 0
 
I12   1 0 0
 
 0 0 1 

that is

b21 b22 b23  0 1 0  b11 b12 b13 

     
 b11 b12 b13    1 0 0 b21 b22 b23 
     
b31 b32 
b33   0 0 1  b31 b32 b33 

Similarly, an interchange of columns 1 and 2 can be effected by multiplying on

the right (also called post-multiplication) by the same matrix:

 b11 b12 b13  0 1 0  b12 b11 b13 

     
b21 b22 b23  1 0 0  b22 b21 b23 
     
b31 b32 b33   0 0 1  b32 b31 b33 

Note that the matrix I12 is the unit matrix with its first and second rows (or
equivalently the first and second columns) interchanged.
42
Unit 7 Matrices-I
You can try the following SAQ to consolidate this concept.

SAQ 2

Construct a matrix which will interchange the i-th and j-th rows when
pre-multiplied to a square matrix of size n.

Another elementary operation on rows is adding to a given row, a linear

combination of other rows. For example, if  is a number,

 b11 b12 b13  b11  b21 b12  b22 b13  b23 

   
b21 b22 b23    b21 b22 b23 
   
b31 b32 b33   b31 b32 b33 

Here the number  times the elements of the second row are added to the first
row. You can see that this operation can also be effected by pre-multiplication
by a matrix:

 1  0  b11 b12 b13  b11  b21 b12  b22 b13  b23 

     
0 1 0 b21 b22 b23    b21 b22 b23 
     
0 0 1 b31 b32 b33   b31 b32 b33 

Try the following SAQ for practice.

SAQ 3

a) What will be the effect of pre-multiplying a 3  3 matrix by the following

matrix?
1   
 
0 1 
 
0 0 1

b) How will the operation of adding a column with a linear combination

of other columns to it be effected?

7.4.2 Partitioning of Matrices

We can build larger matrices out of smaller matrices of appropriate sizes. For
example, from two matrices of sizes 2  2 we can create 2  4, 4  2 or 4  4
matrices. From

 a1 a12  b11 b12 

A , B ,
a21 a22  b21 b22 

we can create,

a11 a12 b11 b12 

A B    ,
a21 a22 b21 b22 
43
Block 2 Vector Spaces, Matrices and Tensors
a11 a12 
 
 A  a 21 a 22 
or,   ,
B   b b12 
 11 
b 
 21 b22 

a11 a12 b11 b12 

 
 A B  a 21 a 22 b21 b22 
or,   ,
B A  b11 b12 a11 a12 
 
b a 21 a22 
 21 b22

where lines inside this matrix are drawn just to show the partitioning.

Conversely, a matrix can be partitioned into several parts by grouping rows

and columns in many ways, where the smaller matrices (called sub-matrices)
can be treated as elements of the larger matrix.

For example, we can partition an n  m matrix A into four sub-matrices by

grouping rows into first n 1 and next n 2  n  n1 as well as columns into first
m1 and next m2  m  m1,

 a11  a1 m1 a1( m11)  a1 m 

 
       
 
 an 1  an1 m1 an1( m11)  an1 m 
A 
1

a  a( n11 ) m1 a( n11)(m11)  a( n11)m 

 ( n11)1 
 
       
 
 an1  anm1 a n( m11)  anm 

B C 
 
D E 

Here B is an n1  m1 matrix, C is an n1  m2 , D is an n 2  m1 and E is

n 2  m 2 matrix. Suppose F is an m  r matrix, which is similarly partitioned
into m1  r1, m1  r2 , m2  r1, m2  r2 sub-matrices

G H 
F 
 J K 

Then the product of the matrices A and F is:

m m1 m
 AF ij   Aik Fkj   Aik Fkj   Aik Fkj
k 1 k 1 k  m1  1

i  1,..., n1, n1  1,.., n ,

j  1,..., r1, r1  1,...r

44
Unit 7 Matrices-I
The product matrix AF thus gets partitioned as four n1  r1 , n1  r2 , n2  r1

and n2  r2 matrices:

B C  G H  BG  CJ BH  CK 
AF      
D E   J K  DG  EJ DH  EK 

This product rule is just as if we had two 2  2 matrices with sub-matrices as

elements. The partitioning has to match, of course, because columns of the
first matrix should be partitioned in the same way as the rows of the second to
allow multiplication.

Notation: We often encounter matrices in which there are zero sub-matrices.

It is common practice to leave these blocks blank,

For example, a matrix such as

 a11  a1 m1 0  0 
 
      
 
a n 1  a n1 m1 0  0 
A 
1

 0  0 a( n11)(m11)  a( n11)m 
 
 
      
 
 0  0 a n( m11)  a nm 

is written as:

 a11  a1 m1 
 
   
 
a n 1  a n m 
A 
1 1 1

 a( n11)1  a( n11)m 
 
 
    
 
 an( m11)  anm 

A number of square matrices, when arranged along the diagonal, define a

larger square matrix, called a block diagonal matrix. As an example

 
 A 
 
 
 
 
 B 
 
 
 
 C 
 
 

is a block diagonal matrix with A, B, C as square matrices of different sizes, in

general. Let us now learn about determinants. 45
Block 2 Vector Spaces, Matrices and Tensors
7.5 DETERMINANTS
Before we discuss determinants, it is good to recall some facts about
permutations, which is a revision of what you have learnt in earlier courses.

7.5.1 Permutations
You know that n distinct objects n  2 can be arranged in a line in many
ways. We can denote these objects by any symbols, but it is convenient to
denote them by numbers, like 1,2 ,…. , n. You know that there are factorial n
n!  n n  1 ... 3.2.1

ways of arranging them, because in the first (leftmost) place we can put any of
the n objects, and for each choice we can put any of the remaining n  1
objects in the second place. This will make n n  1 different ways to fill the
first two places. Next the object for the third place can be chosen in n  2 
possible ways for every choice of the first two places. And so on, till we have
arranged all the n objects in a row.

These different ways of arranging objects in a line are called different

permutations of n objects. We can think of each of the n permutations as a
one-to-one mapping P of the set Sn  1, 2, 3,..., n onto itself:

P : Sn  Sn , 1, 2, 3 ,..., n   P (1), P (2), P (3),..., P (n ) (7.33)

A permutation can be arranged by starting from the original order 1, 2,..., n  by
a number of interchanges. Note that in Eq. (7.33), each element of set S is
replaced by the corresponding P (i ), i  1, 2, ..., n. For example, the
permutation (3, 2, 1) is described by the functions P(1)  3, P (2)  2, P(3)  1.

A permutation is called even or odd according as the number of interchanges

required is even or odd.
For example if n  4, the permutation [3, 1, 2, 4] can be obtained from
1, 2, 3, 4 in two interchanges, first by interchanging 3 and 1 to get
3, 2, 1, 4 and then by interchanging 1 and 2 to get 3, 1, 2, 4 :
1, 2, 3, 4  3, 2, 1, 4  3, 1, 2, 4

A simple way to find how many interchanges are requires is to count the
number of ‘crosses’ in the lines joining two copies of the set with elements
1,2,3,4 as shown in the diagram below:

1 1

2 2

3 3
4 4 Two crosses.

There are two crosses. Therefore, this is an even permutation. One can arrive
at the same permutation in a number of ways, but the number of interchanges,
46
if it is even, remains even, or if it is odd, remains odd. For example,
Unit 7 Matrices-I
1, 2, 3, 4  1, 4, 3, 2 one interchange 2  4
or 1, 2, 3, 4  1, 4, 3, 2 three interchanges 4  3, 4  2, 3  2

The second set of three interchanges is shown in the diagram below.

1 1

2 2

3 3
4 4 Three crosses.
For any n, half the permutations are even and the remaining half odd. This is
so because for every permutation there is another which differs from it by a
single interchange.
Now, work through the following Example. But first a word about the notation
we use.
Notation: The  sign of a permutation P is denoted by ( 1)P , which is  for
even P and  for odd P.
Example 7.1
Write down all the 4!  24 permutations of four objects and classify them as
even and odd.
We omit commas and brackets for simplicity of writing. The sign is put before
the permutation
1234, 1342, 1423, 1324, 1243, 1432
 2134,  2341,  2413, 2314, 2143, 2431
 3214,  3142,  3421, 3124, 3241, 3412
 4231,  4312,  4123,  4321, 4213, 4132 Each element s of S in
Since a permutation P is a one-one onto mapping of the set { 1,2,…..n}, the Eq. (6.33) is replaced
consecutive application of two such permutations P1 and P2 is also a by the corresponding
P(s). So, for example,
permutation: for a set [1, 2, 3], the
P1  P2  (i )  P1 P2 (i ) (7.34) permutation
(3, 2, 1) is described
The identity mapping I which does not change the order 1, 2,..., n is such that by the function
I  P  P I  P (7.35) P (1)  3, P (2)  2,
P (3)  1.
Moreover, the inverse mapping exists which simply reverses the order
P P1
1, 2,..., n   P (1), P (2),..., P (n ) 
 1, 2,..., n 
You can see that since P 1 retraces each step backwards, the even or odd
nature of P and P 1 is the same.

With this preliminary information on permutations, we discuss determinants.

7.5.2 Definition of Determinant and its Properties
You have already encountered the determinant of a 2  2 matrix in Eq. (7.4d)
of Sec. 7.2:
A  a11 a22  a12 a21 47
Block 2 Vector Spaces, Matrices and Tensors
A determinant of a square matrix of size 3,
a11 a12 a13 
 
A  a 21 a 22 a 23 
 
a31 a32 a33 
is defined as:
A   a11 a22 a33  a11 a23 a32  a12 a23 a31

 a12 a21 a33  a13 a21 a32  a13 a22 a31

The general definition of the determinant for an n  n matrix

a11 a12  a1 n 
 
a 21 a 22  a 2n 
A 
    
 
a  a nm 
 n1 a n 2
is
A    1P a1 P (1) a2 P (2) ...an P (n ) (7.36)
P

Note from the definition that the determinant has the following features:
1. The determinant is a homogeneous polynomial of degree n in the
elements with each element appearing only as first power.
2. There are n terms in the polynomial. Each term is a product of n matrix
elements with the row indices fixed and column indices corresponding to a
permutation of 1, 2,…,n. The terms have coefficients  1 depending on the
even or odd nature of P.
A convenient way to evaluate a determinant is by the rule called Laplace
expansion. You should know this rule but we give it here for ready reference.
It goes like this:
Laplace expansion by the first row:
1. Take the first element a11 of the first row. Multiply it by the determinant of
the (n  1)  (n  1) matrix which will be left after deleting the first row and
the first column on which a11 falls.
2. Take the second element a12 of the first row. Multiply it similarly by the
determinant of the matrix left after deleting the first row and the second
column on which a12 falls. Multiply further by  1 and add to the term with
factor a11.
3. Take the third element a13 of the first row. Multiply it similarly by the
determinant of the matrix after deleting the first row and the third column
on which a13 falls. Multiply by  1 and add to the previous terms.
4. Continue in this way, till the end of the row, with alternating  1 and  1
factors.
5. In this process the n  n determinant has been reduced to n terms
containing (n  1)  (n  1) determinants, which, in turn, can be reduced to
(n  2)  (n  2) determinants, and so on until there are no determinants
48 left to evaluate.
Unit 7 Matrices-I
Thus, for example,
a11 a12 a13

A  a21 a22 a23

a31 a32 a33

a22 a23 a21 a23 a21 a22

 a11  a12  a13
a32 a33 a31 a33 a31 a32

 a11 a22 a33  a23 a32   a12 a21 a33  a23 a31 

 a13 a 21 a32  a 22 a31 

It is not necessary to expand only by the first row. We can choose any fixed
row or a fixed column and continue as above with the following rule in mind:
The sign of the term whose coefficient is aij is determined by  1 . Thus,
i j

as in the above example the sign is  1   1 for a11 , and  1   1

1 1 1 2
for
a12 , and so on.

With this brief revision of evaluating a determinant, we write the properties of

determinants.
Properties of Determinants
From the definition of the determinant given in Eq. (7.36), the following
properties can be deduced. We give the proofs later on in the unit.
1. The determinant of a matrix A and its transpose are the same AT  A .

2. If in a matrix any two rows are interchanged, the determinant of the

changed matrix is the original determinant with a change of sign.
Similarly, if any two columns are interchanged, then also the determinant
changes sign
3. If a row or column is multiplied by a number, then the determinant gets
multiplied by that number.
4. If we add a row (or column) to another row (or column), then the
determinant does not change. More generally if to a row (or column) a
linear combination of other rows (or columns) is added, then the
determinant does not change.
5. The determinant of a diagonal matrix is just the product of its diagonal
elements. The determinant of a block diagonal matrix is the product of the
determinants of the sub-matrices on the diagonal.
6. The determinant of the product of two matrices is equal to the product of
their determinants:
AB  A B (7.36)

Most of these properties are consequences of the definition. Only the proof of
AB  A B is non-trivial.

We now show the reason behind these properties. 49

Block 2 Vector Spaces, Matrices and Tensors
To show that AT  A , let the elements of AT be written aij  a ji

A    1P a1 P(1) a2P(2) ...an P(n)

P

   1P aP (1)1 aP (2) 2 ...aP (n)n

P

We can re-arrange the factors so that the first, second etc. row indices of AT
appear in their natural order. Then the column indices 1, 2,…,n will get
arranged in reverse permutation:
A    1P a1 P 1(1) a2 P 1(2) ...anP
 1(n )
P
1
 1 1P a1 P 1(1) a2 P 1(2) ...anP
 1(n )
P
 AT

The last step follows because the even or odd nature of P and P 1 is the
same.
If the two rows of the matrix are interchanged, then in the expression for the
determinant only one interchange is needed to bring it into the original order.
That is responsible for the extra minus sign. The same applies to the column
interchange because by transpose the process is equivalent to the row
exchange. The matrix with two identical rows or two identical columns has
zero determinant because A   A .

Also, adding a row to another row, the expression for the determinant
becomes equal to sum of two expressions. One of these expressions is the
original determinant, and the other is the determinant for the matrix with two
identical rows, which is zero.
We now come to block matrices. Let a block diagonal r  s   r  s  matrix C
with elements Cij be given with two blocks as:

 
 A 
 
 
C 
 
 B 
 

Here A is an r  r matrix with elements aij and B an s  s matrix with elements

bij . The determinant is:

C    1P c1P(1) ...c rP(r )c(r 1) P(r 1) ...c(r s) P(r s)
P

But among all permutations P if a row index i is in the first r, that is 1  i  r

and its column index P (i ) in the last s, that is r  1  P (i )  r  s  then
ciP(i )  0. Therefore, the only permutations that contribute to the determinant
are those in which P permutes 1, …, r among themselves say by P1 and
independently, r  1... r  s  permute among themselves by permutation P2 .
50 The number of interchanges in permutation P will be the sum of interchanges
Unit 7 Matrices-I

in the permutations P1 and P2 . The sign  1P will then be the product of the
two signs. So,

C    1P  1P a1P (1) ... arP (r ) b1P (1) ... bsP (s )
1 2
1 1 2 2
P1 P2

 A B

You should now solve an SAQ.

SAQ 4
Show by similar arguments that the determinants of 2n  2n matrices D1 and
D 2 of the form

   
 A   A C 
   
   
D1   , D2   
   
 C B   B 
   

where A, B and C are n  n matrices, are given by

D1  A B,
D2  A B

We will now give the proof of some of these properties.

Proof of AB  A B

Let A and B be two n  n matrices, and 1 the unit matrix of size n. The proof
depends on the identity:

1 A A AB

1 1 B 1 B

(7.37)

The matrix pre-multiplying on the left side of the equation simply replaces rows
of the matrix with linear combinations of other rows, so that the determinant of
the matrix with A and B on the diagonal does not change. Therefore,

A AB

1 B 1 B

51
Block 2 Vector Spaces, Matrices and Tensors
This means that
A B   1n  1 A B

  12n AB  AB
Proof of the Laplace Expansion
We have explained the Laplace expansion by the first row as an example
earlier in this section.
It is sufficient to prove it for the first row because any other row can be
interchanged and brought as the first row and a change of sign. Also as the
determinant remains unchanged by taking transpose of the matrix, the
Laplace expansion can be done by any column as well.
Let
A    1P a1P(1) a2 P(2) ...anP(n)
P

The permutations P of 1, ……, n can be divide into the following subsets:

1. all those n  1! permutations P1 for which P (1)  1. These are essentially
the permutations of 2,3,….. n. For such permutations  1   1 1 ,
P P

2. all those n  1! permutations P2 for which P (1)  2. These are essentially
the permutations of 1, 3,… n, after the interchange of 1 and 2 so that
 1P   1P2
3. and so on for P3 ,P4 etc.

Therefore, A    1P a1P (1) a2P (2)... anP (n )

P

 a11   1P1a2P1( 2). .. anP1( n )

P1

 a12   1P2 a1 P2 (1) ... anP ( n )  ...

1
P2
giving the Laplace expansion formula.
Notation: Cofactors
Given a matrix A, the determinant of the matrix which remains after removing
the i-th row and j-th column, (that is, the row and column corresponding to the
element aij ) when multiplied with a factor ( 1) i  j is called the cofactor of aij .
We denote it by:
a11  a1( j 1) a1( j 1)  a1n

     

a( i 1)1  a( i 1)( j 1) a( i 1) ( j 1)  a( i 1)n

a~ij   1 i  j
a( i 1)1  a( i 1)( j 1) a( i 1) ( j 1)  a( i 1) n

     

a n1  a n( j 1) a n( j 1)  a nn

52 (7.38)
Unit 7 Matrices-I
The Laplace expansion by the i-th row can be written as:
n
A   aij a~ij , (7.39a)
j 1

and expansion by j-th column as

n
A   aij a~ij (7.39b)
i 1

You should now solve an SAQ.

SAQ 5

Show that if the elements of a fixed row are multiplied by corresponding

cofactors of a different row and added, the result will be zero:
n
 aij a~kj  0 i  k, (7.40a)
j 1

Similarly, if the elements of a column are multiplied by cofactors of a different

column and added, the result will be zero:
n
 a ij a~ik  0 j k (7.40b)
j 1

Hint: The expressions in the first of these cases will look like the determinant
of the matrix in which the k-th row is identical to the i-th row. But the
determinant of a matrix with two identical rows is zero. Similar argument
holds for the second case.

7.5.3 Inverse of a Matrix

Dividing Eq. (7.39a) by A (if it is non-zero), we get
n a~ij
1   aij A
, for fixed i (7.41a)
j 1
n a~kj
and 0  aij A
, for i k (7.41b)
j 1

This suggests that we define a matrix of cofactors:

A  a~ 
~
ij (7.42)

Then the above equations for all i and k can be written as:
n a~kj
 ik   a ij
j 1 A


1
A
A A~T  ik
Since  ik are elements of the unit matrix, the inverse of the matrix can be
defined as

A 1 ik  1 ~
a ki (7.43)
53
A
Block 2 Vector Spaces, Matrices and Tensors
so that A A 1  1 (7.44)

It is important to remember that the ik-element of the inverse matrix is

associated with ki element of the cofactor matrix. The interchange of the row
~
and column appears because of the transpose of A in the formula above. You
may like to solve an SAQ.

SAQ 6
Prove that A1 A  1. [Hint: Use Eq. (7.39b)].

7.6 LINEAR OPERATORS AND MATRICES

You have learnt that a linear operator T mapping an n-dimensional vector
space U into another vector space V of dimension m defines a rectangular
array or set of numbers t ij if we choose bases in the two spaces,

Let e1, e 2,..., e n be a basis in U and similarly f1, f2,..., fm in V.

Since Tei  V , i  1,. . ., n, each of these can be expanded with respect to the
basis in V:
Te1  t11f1  t12f2  ...  t1mfm

Te 2  t21 f1  t22 f2  ...  t2m fm

.. .  .. .

Ten  tn1 f1  tn 2f2  ...  tnm fm

then the elements t ij of the matrix are real numbers and it is a real matrix. If
U and V are complex spaces then the elements t ij of the matrix are complex
numbers and it is a complex matrix.
As T is a linear operator, it is sufficient to define it on the basis vectors
because on any other vector u  U its action is then known:

n  n n m
Tu  T   x i ei    x iT ei    x i tij f j
 
 i 1  i 1 i 1 i 1

Therefore, if the components of the vector v  Tu  V with respect to the

basis f j , j  1, . . ., m are y j :

m n m
Tu  v   y j fj    x i tij f j
j 1 i 1 i 1

Then comparing the coefficients of the basis vectors f j on the two sides
above, we obtain
n
yj   x i t ij
i 1

You can see that this is a matrix equation where the row vectors
Y  [ y 1,..., y m ]
54
Unit 7 Matrices-I
are obtained by multiplying the row vector
X  [ x 1,..., x n ]
on the right by the n  m matrix T with elements t ij :

n
Y  XT , yj   x i t ij j  1, 2, ... , m (7.45)
i 1

We can also write this equation in terms of column vectors. Taking the
transpose of this equation, we get:
YT  TT XT (7.46)

or

 y1   t11 t 21t n1 
  x1 
     
 y2   t12 t 22  rn 2   x2 
      (7.47)
     
     
y  t t 2m  t mn  x 
 m  1m  n

7.6.1 Two Linear Operators

Suppose there are three vector spaces U, V and W of dimensions n, m and r,
respectively, and two linear operators T : U  V and S : V  W. Let
g1, g2, ... , gr be a basis in W.

Then, the operator ST : U  W which is the composite of two linear operators,

allows us to calculate the vector STei for i  1, 2, ... , n directly as

m 
 
m
ST ei  STei   S  tij f j    tij Sf j ,
 
i  1, 2, ..., n
 j 1  j 1

Let Sf j  W be expanded in the basis of space W

m
Sf j   s jkgk , j  1, 2, ..., m
k 1

Then
m r
ST  ei    tij s jkgk , i  1, 2, ..., n
j 1 k 1

If we had expanded STei directly in the basis gk , the corresponding matrix

will be (using pik as coefficients):
r m r
ST  ei   pikgk    tij s jkgk ,
k 1 j 1 k 1

This shows that the matrix corresponding to the product ST is related to the
matrices T and S as:
m
p ik   t ij s jk
j 1 55
Block 2 Vector Spaces, Matrices and Tensors
This shows that the matrix [ST] corresponding to the linear operator
ST : U  W where T : U  V and S : V  W is given by the product of
matrices in reverse order:
ST  T S ,
or,
 p11 p12  p1r  t11 t 12  t1m   s11 s12  s1r 
     
 p 21 p 22  p 2r  t 21 t 22  t 2m   s 21 s 22  s 2r 
    
             
     
p p n 2  p nr  t n1 t n 2  t nm  s  s mr 
 n1  m1 s m 2

To bring the same order of S and T, we can take the transpose and write:

ST T  S T T  T
We now end this unit and summarise its contents.

7.7 SUMMARY
In this unit, we have covered the following concepts:
 Elementary concepts related to matrices including their definition,
elementary algebraic operations on matrices such as their sum,
difference, multiplication by a number and multiplication of matrices
(Sec. 7.2).
 Various types of matrices such as real, complex, symmetric, Hermitian
adjoint, unitary and orthogonal matrices, evaluation of the transpose,
complex conjugate and inverse of a matrix, and function of matrices
(Sec. 7.3).
 Elementary operation in matrix algebra and partitioning of matrices
(Sec. 7.4).
 Permutations, definition of determinants, evaluation and properties of
determinants (Sec. 7.5).
 Linear operators and matrices and two linear operators (Sec. 7.6).

7.8 TERMINAL QUESTIONS

1. Show that the product of two diagonal matrices is diagonal.
2. Separate the real and imaginary parts of a Hermitian matrix H by writing it
as sum of a real matrix A and i (the imaginary unit) times another real
matrix B:
H  A  iB
What is the nature of matrices A and B
3. Let c  x  iy be a complex number with x and y its real and imaginary
parts. Define
 x y
C 
 y x 
56
Unit 7 Matrices-I
Construct a similar matrix C from c   x   iy . Show that

CC  CC

has the same form as C or C.

4. Apart from the identity matrix itself, find three real 2  2 matrices such that
their square is equal to the identity matrix.
5. Show that the cube of the following matrix is equal to the identity matrix:
 1  1
 
 1 0 

6. A rotation about origin by an angle of the orthogonal axes in a plane

leads to the transformation of coordinates (of the same point) with respect
to the two sets of axes by
 x   cos  sin   x 
    
y   sin  cos  y 

Show that any transformation matrix is an orthogonal matrix. Show that

any 2  2 orthogonal matrix has this form.
7. Show that the determinant of an orthogonal matrix is either 1 or 1.
8. Show that the set of all 2  2 Hermitian matrices form a four dimensional
real vector space. Find a suitable basis for this vector space.
9. The commutator of two square matrices has been defined as
[ A, B]  AB  BA.

Prove the following identities for matrices A, B, C, D:

A, BC   A, B C  BA, C 
A B, C   AB, C   A, C  B
A, B,C   B, C, A  C, A, B    0, Jacobi identity

10. Define the anti-commutator as

A, B  AB  BA
Prove the following identity:
[ AB, CD]  A {B,C } D  { A,C } BD  C { A, D}B  CA {B, D}

Note: This identity is of great relevance in quantum field theory. If each of

the quantities A, B anti-commute with each of C, D, then their
products commute.
11. Let
1 a b  1 d e 
   
A  0 1 c  , B  0 1 f 
   
0 0 1 0 0 1

Calculate AB and show that it is also a matrix of the same form. This form
is called upper triangular matrix in which all elements below the diagonal 57
Block 2 Vector Spaces, Matrices and Tensors
are zero. What should d, e, f be so that AB  1? Will it make BA  1 too
Find the condition that AB  BA.
12. The first Pauli matrix is
0 1
1   
1 0
Calculate
2 2
U1  exp i1  1  i1     ...
2! 1
For real , show that U1 is unitary and has determinant 1.
13. Calculate exp i 2  and exp i 3  for the other Pauli matrices.
14. Determinants of 2n  2n matrices D 3 and D 4 of the form

   
 A   A 
   
   
D3   , D 4   
   
 B C   B C 
   

where A, B and C are n  n matrices, are

D3  ( 1) n A B , D4  ( 1) n A B

Hint: Here the permutations that contribute are those in which 1, …. n are
mapped into n  1, …, 2n (and therefore n  1, …, 2n are mapped
into 1, …, n. We can start such permutations with first interchanging
1 with n  1, 2 with n  2 and so on, which gives a factor of (1) n
followed by independent permutations of the two sets of indices
among themselves.
15. Velocity addition in special relativity: The Lorentz transformation
between two inertial frames, one of which X  moves with relative velocity
v with respect to X is given by the matrix equation:
X   Lv  X
with
t  t
X   , X   
 x   x 
and
1  1  v / c2
Lv    
1  v 2 / c 2  v 1 

where c is the constant equal to the velocity of light. Similarly, there is

another frame X  that moves with relative velocity w in the same direction
with respect to X . W hat is the velocity of X  with respect to X?
Show that if X   L(w ) X   L(w ) L(v ) X then

v w
L(w ) L(v )  L(u ), u
58 1  vw / c 2
Unit 7 Matrices-I
Hint: Use the identity
1  a 2 1  b 2   1  a 2  b 2  a 2 b 2  a  b 2  1  ab 2
7.9 SOLUTIONS AND ANSWERS
Self-Assessment Questions
1. a) The vector space of all n  n real matrices is of dimension n 2 because
any real matrix
t11 t12  t1m 
 
t 21 t 22  t 2m 
T  
   
 
t  t nm 
 n1 t n 2

can be written as a linear combination of n  n matrices e(jk),

(j, k = 1, ..., n) with 1 in the j-th row and k-th column and zero
elsewhere.
t11 t12  t1m  1 0  0 0 1  0
     
t 21 t 22  t 2m  0 0  0 0 0  0
T    t11    t12    ...
           
     
t  0   
 n1 t n 2  t nm   0  0 0 0  0

 T11e(11)  T12 e(12)  ...

Therefore, for the n 2 real n  n matrices e(ij) form a basis.

For n  n complex matrices, there are two vector spaces to consider.
As a complex vector space, it is of dimension n 2 . And the above
formula with e(jk) as the basis holds good but the coefficients are
complex numbers.
The set of complex matrices can also be considered as a real vector
space of dimension 2n 2 and the basis can be chosen as the 2n 2
matrices e(jk) and ie(jk) so that if the elements of complex matrix T are
t jk  r jk  is jk where r and s are the real and imaginary parts of
complex number t jk , then

T   r jk e jk  s jk (ie( jk ))
j ,k
b) We choose Hermitian Pauli matrices, say A  1 and B   2 :
0 1 0  i 
A , B ,
1 0  i 0 
Then their product
i 0 
AB   ,
0  i 
is not hermitian because
 AB †   AB . 59
Block 2 Vector Spaces, Matrices and Tensors
2. i) Take the unit matrix of size n. ii) put ii and jj elements on the diagonal
equal to zero. iii) place ij and ji elements equal to 1. This is the required
matrix:
1 2  i  j  n

1 1

2 1

 1

i 0 1

 

 

 

j 1 0

 

n 1

3. a) On multiplying the matrices, we get:

1     b11 b12 b13 
0 1   b b23 
   21 b22
0 0 1   b31 b32 b33 

b11  b21  b31 b12  b22  b32 b13  b23  b33 

  b21  b31 b22  b32 b23  b33 

 b31 b32 b33 

b) Consider the matrix similar to the one in part (a).

 1 0 0
 
 1 0 
 
  1

Multiplication of the following matrices

 b11 b12 b13   1 0 0
   
b21 b22 b23   1 0 
   
b31 b32 b33    1

gives
 b11  b12  b13 b12  b13 b13 
b  b  b b22  b23 b23 
 21 22 23
 b31  b32  b33 b32  b33 b33 

This is how the first column can add a linear combination of second
and third column, and the second column with a multiple of the third
60
Unit 7 Matrices-I
column. Suppose we want to change the third column by adding a
multiple of first column, then we choose
 b11 b12 b13   1 0    b11 b12 b13  b11 
 b23  0 1 0    b b23  b21 
b21 b22    21 b22
 b31 b32 b33  0 0 1 b31 b32 b33  b31 

Similarly for others. The rule is that for linear combination of other
columns to be added to some column you post-multiply a unit matrix
with additional entries added at appropriate places.
4. We show the result for D1 whose elements are d ij with i and j running
from 1 to 2n. We are given that
d ij  0, if i  n and j  n
d ij  c ij , if i  n and j  n
d ij  bij , if i  n and j  n.

Now, D1    1P d1P(1) d 2P(2) d nP(n)d(n1) P(n1) d 2nP(2n)

P

Here P denotes the 2n! permutations. From the above it is clear that when
P maps the set {1, 2, ..., n} into any permutation which contains any index
from n + 1, ..., 2n, those terms will be zero. Therefore,
D1    1P d1P(1) d 2P(2) d nP(n)d(n1) P(n1) d 2nP(2n)
P

where P  permutes {1, 2, ..., n} into themselves. Therefore, {n + 1, ..., 2n}

also permute among themselves. And the permutations P  breaks into two
independent permutations P1 and P2 such that

D1    1P d1P(1) d nP(n )

1
1 1
P1

   1P d (n1) P (n1) d 2nP (2n )

2
2 2
P2

 A B
The case for D2 is similar.
5. The ‘Hint’ given is actually the solution.
6. If we expand a determinant by j-th column
n
 aij a~ij  A for any fixed j
i 1

where a~ij is the cofactor of the element aij . Now if we take a matrix in
which j-th and k-th columns are identical then
n
 aij aik  determinant of matrix with two columns identical = 0 for j  k.
i 1

Therefore,
n n
 aik aij   aij aik  A  jk 61
i 1 i 1
Block 2 Vector Spaces, Matrices and Tensors
~T
~ is the same as the ki-element of matrix A
We realize that aik made from
the cofactors of the transpose matrix AT . If A  0, then dividing by A
we get:

 
1 n ~T
A ki aij   kj
A i 1

The right-hand side is the unit matrix. Therefore,

 
1 1 n ~T
 A ki  A ki ,
A i 1

which was to be proved.

Terminal Questions
1. Let A and B be two diagonal matrices. Then their matrix elements are of
the form:
( A)ij  ai ij , (B )ij  bi ij

Their product has elements

( AB)ij   ai ik ak kj  ai bi ij
k

2. If the real and imaginary parts of a Hermitian matrix are written as separate
matrices:
Hij  (hij )  aij  ibij
then as
hij*  aij  ibij  h ji  a ji  ib ji

we must have
aij  a ji , bij  b ji

Therefore, the matrix A  aij  is real symmetric and B  bij  is real anti-
symmetric.
 x y  x  y 
3. C , C   .
 y x   y  x 
therefore,
 xx   yy  xy   yx   xx   yy  xy   yx 
CC     and C C   
 ( xy   yx ) xx   yy    ( xy   yx ) xx   yy  
which are of the same form as C or C  .
4. There are infinitely many. These three are simple ones:
0 1 1 1  1 0 
 ,  ,  
1 0 0  1 0  1
 1  1  1  1  1  1  0 1  1  1
5.         
 1 0   1 0   1 0   1  1  1 0 
 1 0
 
0 1
62
Unit 7 Matrices-I
 cos  sin   cos   sin  
6. R    RT   
  sin  cos    sin  cos  

Therefore,

1 0
RT R   
 0 1

Let a 2  2 matrix A be

a b 
A 
c d 

then

a c 
AT   
b d 

and

 a c  a b   a  c ab  cd   1 0 
2 2
T
A A      
b d  c d  ab  cd b2  d 2  0 1

Therefore, a2  c 2  1, b2  d 2  1, ab  cd  0

We can think of these equations as if they are about two-dimensional unit

vectors x  (a, c ) and y  (b, d ) which can be written

x . x  1, y . y  1, x . y0

(/2 – ) x  (a, c )  (cos ,  sin )

y  (b, d )  (sin , cos )

x

This shows that any orthogonal matrix has the same form as R.

7. If AT A  1 then because  AT  A  ,

2
AT A  AT A  A 1

Therefore, A  1

8. Let A and B be two Hermitian matrices. Then their sum is also a Hermitian
matrix

 A  B †    A  B  *  i  A*  B*  A  B  ( A  B)
 ij  j ji ji ij ij ij

Similarly, if a is a real number then

 aA†ij  (aA)*ji  a ( A)*ji  aAij  (aA)ij

63
Block 2 Vector Spaces, Matrices and Tensors
Therefore, Hermitian matrices of (any dimension) form a real vector space.
A 2  2 matrix A

a b 
A 
c d 

with complex elements will be Hermitian if

a * c *  a b 
A†      A.
b * d * c d 

or, a  a*, d  d *, b  c *, c  b *.

This means that a and d are real and b and c are complex conjugate to
each other for any Hermitian matrix.

Let b = e + if, then we can write A as

 a e  if 
A 
e  if d 

1 0 0 0   0 1 0 i
a  d  e  f 
0 0   0 1  1 0  i 0 

Thus any Hermitian matrix A can be written as a real linear combination of

four Hermitian matrices:

 1 0 0 0 0 1 0 i
 ,  ,  ,  
0 0 0 1 1 0  i 0

which can be taken as a basis in the real vector space of 2  2 Hermitian

matrices.

The standard choice for a basis is the unit matrix and the three Pauli
matrices  i , i  1, 2, 3 :

 1 0 0 1 0  i  1 0 
1 , 1   ,  2   ,  3   
0 1 1 0  i 0  0  1

9. [ A, BC ]  ABC  BCA  ( AB  BA) C  BAC  B (CA  AC )  BAC

 [ A, B ] C  B [ A, C ]

[ AB, C ]  ABC  CAB  A(BC  CB )  ACB  (CA  AC ) B  ACB

 A [B, C ]C  [ A, C ] B

For Jacobi identity add the following equations to get zero:

 A, [B,C ]  A (BC  CB )  (BC  CB) A  ABC  ACB  BCA  CBA

B, [C, A]  B (CA  AC )  (CA  AC ) B  BCA  BAC  CAB  ACB
C, [ A, B]  C ( AB  BA)  ( AB  BA)C  CAB  CBA  ABC  BAC
64
Unit 7 Matrices-I
10. [ AB,CD]  ABCD  CDAB

 A (BC  CB ) D  ACBD  C (DA  AD ) B  CADB

 A {B,C } D  C { A, D} B  AC BD  CADB

 A {B,C } D  C { A, D} B  ( AC  CA) BD  CABD

 CA (DB  BD )  CABD

 A {B,C } D  C { A, D} B  ( AC  CA) BD  CA (DB  BD )

11. Given
1 a b 1 d e
A  0 1 c  , B  0 1 f 
0 0 1 0 0 1

we have:
1 a b 1 d e  1 a  d e  af  b 
     
AB  0 1 c  0 1 f   0 1 f c 
     
0 0 1 0 0 1 0 0 1 

For AB  1, d  a, f  c, e  b  ac. This makes B  A 1 and so BA  1

as well. But in general
0 0 af  cd 
 A, B   0 0 0 
0 0 0 

Therefore af  cd is the condition to ensure that  A, B   0

12. Using the fact that 12  1 we calculate powers

 cos  i sin 
exp i  1   cos  1  i sin  1   
i sin  cos  

0 i 
13. 2   ,  2 2  1,
i 0 
Therefore, all even powers of  2 are 1 and all odd powers are  2 itself.
Thus, using i 2  1, i 3  i , i 4  1 etc.,

i 2 i 3
exp (i  2 )  1  i  2    2  ...
2! 3!
 2   3 
 1   ...  1  i     ...   2
 2!   3! 
 cos  1  i sin   2

Similarly, all even powers of  3 are 1 and all odd powers are  3 itself so
that
exp (i  3 )  cos  1  i sin  3
65
Block 2 Vector Spaces, Matrices and Tensors
14. This question is similar to SAQ 4, and the only extra trick required is given
in the hint with the question.

1  1 w / c 2 
15. L(w )   
1  w 2 / c 2  w 1 

1 1 v / c 2 
L(v )   
1  v 2 / c 2  v 1 

Therefore,
L(w ) L(v )

1 1  1 w / c 2   1 v / c 2 
   
1 w 2 / c2 1  v 2 / c 2  w 1   v 1 

1 1 1  vw / c 2 (w  v ) / c 2 
  
1 w 2 / c2 1  v 2 / c 2  (w  v ) 1  vw / c 2 

1  wv c 2  1 (w  v ) / (c 2  wv )
  
1  w 2 / c 2 1  v 2 / c 2  (w  v ) (1  wv c ) 
2
1

Let us define the velocity

wv
V
1  wv / c 2

then

1  wv c 2  1 V / c 2 
L(w ) L(v )   
1  w 2 / c 2 1  v 2 / c 2  V 1 

The factor in front is simplified by using

(1  a2 )(1  b2 )  1  a2  b2  a2b2  (a  b )2  (1  ab )2

so that if a  w and b  v then

1 w 2 / c2 1 v 2 / c2    (w  v )2 / c2  (1 wv / c2 )2

 (1  wv / c 2 )2 1  V 2 / c 2 
 

Therefore,

1  1 V / c 2 
L(w ) L(v )   
1  V 2 / c 2  V 1 

This proves the relativistic velocity addition formula.

66