8

Matrices
8.1 Eigen }'aloes, Eigen-Vectors : Characteristic equation of a matrix
Let A= (a;J1 be a square matrix of order n . Suppose there is an n~
dimensional nonzero column vector X ,such that the action of A on X ( i.e.
the matrix product AX, gives a vector which is just a multiple of X , that is'
AX= AX .
where A is a scalar. In other words, the transformation represented by
the matrix A just multiplies the vector X by a scalar A. The vector X is then
called an eigen vector of the matrix A. A is calJed an Eigen value of A
corresponding to the eigenvector X . The problem of finding the eigenvectors
and· the eigenvalues of a matrix is called the eigenvalue problem.
Definition of Eigenvector. A nonzero vector X is called an eigenvector
of a matrix A if there is a number A such that AX = AX .
Here A is called an eigen value of A corresponding to the eigenvector X
and vice versa.
We have, AX = AX =Al X, I being a unit matrix.
or (A - A I) X = 0.
This is the matrix form of an eigenvalue problem.
Since X :I: 0, the matrix ( A - Al ) is singular, so that
IA -All=O ... (1)
Equation (I) is called the characteristic equation of A .The eigenvalues
are just the roots of the equation obtained by expanding the determinant in Eq.
(1). Then- roots A1, Ai, .. .An of the characteristic equation are not necessarily
aJJ different.

r~ ·;!]
Example 1. Determine the eigenvalues and eigenvectors ofthe matrix·

A=
0 0 5
Sol. The characteristic-equation of A 1s I A - Al I= O i.e.,
3-.A. I 4 .
0 2
_A = 0 i.e., (A - 2) (A - 3) (.A. - 5) = 0
6
0 0 _
5 A :. At = 2, ½ = 3, A.:3 = 5
These are the eigenvalues of A.
To determine eigen vectors let us consider the eigen values one by one.
(i) When l 1 = 2 the eigenvector X 1 is given by (A - 2l)X 1 = 0

130
 0

[H <, [ •

fhe rank of coefficient m tru bean 1

e
I linearly independent lut1
1-- 2 I .•
ri,ese are equ1 valent to x 1 + .x + 4
fr 0
3x -0
ri,e last two g ive x3 = 0 Then fo t one g1\ ,: + X

l
Takex 1 = I, then x2=- l andx3 =0.
Hence 1
C bem
X ~c, [~
(ii) When½= 3, the eigenvector X 2 1s g1\len by (A - 31) X2 =O

i.e., [g ~I i][~~]=[g]
These are equivalent to x 2 + 4x 3 = 0
-X2 + fu-3 = 0
2x3 = 0
giving x 3 = 0, x 2 =0 and x 1 is arbitrary, say x 1 = 1.
C2 being a sca lar.
Then

These are equivalent to -2x 1 + x2 + 4.x3 = 0

-3x2 + 6x3 = 0

g1vmg x 2 = 2x3 = ix 1, i.e., 2x 1 = 3x2 =6x3

Take x3 =1, so that x2 = 2 and x 1 =3
Hen~ X3 = C3 m, C3 being a scalar.

Eb
 Mechanics and Mathematica/ Mer/
IOd.1
Example 2. Find thl' eiRen l'{,/11es and normalised et,?en i•ectors of the
11/{l// ; ,

[~ ~ :1
Sol.
Let A=[~
A- 11.I =[~ 0
O
1
I -11. 0
= 0 I - 11.
[ 0 l
The characteristic equation of A 1s
I -11. 0 0
I A - 11.II = 0 l - 11. I =O
0 I l-11.
i.e., (l-11.)[(l-11.) 2 - l]=O
I.e., ( l - 11.) (11. 2 - 211.) = 0
I.e.,
11. ( I - 11.) (11. - 2) = 0
i.e.,
11. = 0, I, 2.
Thus the eigen 1·a/ues of the matrix A are 0, l, 2.
Eigenvalue equation is
(A -11.I) X = 0
... (I)
For').= 0, Eq. (I) reduces to

[g : : ] [::I]= [g]
Th1s 1s equivalent to the following equations

XI= 0
=Q X2 +X3
x 2 +x3 =O
Solving these equations, we get

xi =0. x 2 =- x 3 =k (say)
Xi
N
0
·- ru X1 = X2 ={xi, x 2• x3 } = { 0, k. -k }
lN X3
0
:--1
0 .
 ,ctS
Jf thee gen, ed rs be normahsed to unit), then I x1 I= 1

{[o2 + k2 + - k -] = 1 or k = _l....
'\J ( 2)

Normalised eigen , ector

X
1=
{ 1
0, ✓(2)' ✓(°"2)j
-1 l
for ;. . = 1, Eq. ( 1) reduces to

[~ ~ 11(:}lgJ
This 11- equt' alent to the following equations
,;c., = 0
X3 =Q
;. Xz = l t. 0,0} in normalised form.
For :l = :. Eq. ( 1) reduces to

[-i Jl]l:~1 =[gl

~I

'-inch 1s equ1, alent to the following equations

-X1 =0
-1.2 +:t3 = O
X2 -X3 =0
Solving these equations. we get
(1 =0,.x_ = (3
\\ nhm the arbitrary scale f ctor. the e1ge11 iector corresponding to

=21s gnen b)
X3={:c, ,X2 1.3}= {0.k,kj
For X normahsed to unat)
3
l
l e ., k=~
-.J2

1 1} in normalised fonn.
X3 = 0. - ( ? ) ' ,
"- ~ -
1'1lm the normalised ergen ,ectors of the gn·e11 matrix A correspo11d-

llw ~•ien l:al~s O. 1. 2 art

l · - TiI . respectively
o. (2) {1. 0. 0} , { 0. ~·
1 ~l}
 134 Mec han ics and Mat hem atic al Me h
t Ods
8 _2 . Ca) ley- Ham ilto n The ore m
Stat eme nt. E,•ery r;quare mat dx sati sfie s its own cha ract eris tic eq« .
at10,1
OR
If IA - Al I = a0 + a I A+ a 2 A2 + . . . + a An = 0
11
be the cha ract eris tic equ atio n of a squ are
mat rix A, then
a 0 I+ a 1 A+ a 2 A 2 + ... + a A 11 = 0
11
Pro of. The cha ract eris tic poly nom ial is
I A - Al I= a0 + a A+ a ')/ + ···· + _a A0
1 2 11 .. . (1)
Eac h elem ent of cha ract eris tic mat rix (A -
Al) is an ord inar y ~olynornial
of deg ree n (at mos t). The refo re the co-f acto
r of eve ry elem ent of I A - A.I I is
an ordi nary poly nom ial of deg ree n - 1 · (at
mos t). Con seq uen tly eac h element
of

B = adj (A - A.I) ... (2)

is an ord inar y poly nom ial of deg ree (n - 1)
(at mos t).
B = adj (A - Al) = B0 + B A + B2A2 + · · · +
1 B 0 _ 1
... (3)An- 1
Her e B0 , B1 , B2 , ... B _ are all squ are mat
0 1 rice s of the sam e orde r n
who se elem ents are poly nom ials in the elem
ents of A.
Now , (A - Al) adj (A - Al) = IA - Al I I
Usi ng Eqs. (3) and (1), we get
(A - Al) [Bo + B 1 A + B2 A.2 + . .. + Bn - 1 A.n-1]
= (a 0 + a 1 A+ a 2 A.2 + ... + an A.°) I
... (4)
Com pari ng the coe ffic ient s of like pow ers
of A on both the side s, we get
AB0 =aaI
AB 1 - B0 = a 1 I
AB .i- B 1 = a 2I
································
······················· ·········
A Bn - I - Bn - 2 = an - 1 I
-B n-1 = an I
Pre mul tipl yin g thes e by I, A, A 2 , A 3 .... An
in ord er and add ing ,
aoJ + a1A +a 2A 2 + ··· +an An =O
Thi s is Cay ley- Ham ilto .
n theorem.
Corollary. To det erm ine A - 1 by using Cay ley-
Ham ilto n theorem.
Let A be a non -sin gul ar mat rix of ord er n so
tha t I A I"# 0.
 rff' e, 135
Acctmhng to Cayley-Ham1lton theoicm
ool+a,A+u2A2+ ... +a"A"==O .. (1)
J1ic char actenstic pol) nomial j~
I A - Al I== a 0 + a 1 A+ a 2A.2 + .. a A" ... (2)
11
flJr;... = 0, Eq. (2) gives I A I= a 0 • .-. ao -:t 0
Now dividing Eq. (I) by a 0 . ,,, e get

l=-(:1 A+ a2 A2+ ... + an An\) ..(3)

o ao ao
...(I J 1•
pre-multiplying Eq. (3) b) \ - we get
OllJ1a1
All rs
fllelll
A - I= -(~I+
ao
a2 A+ ... + an An - 1\
ao ao )
... (4)

Example 1. Fi11d the characteristic equation of the matrix

... (2

... (3 3 1 1
A=[~ -~ !Ij
and verify the Cayley-Hamz ton theorem for it. Hence or othenvise
der n 1
Jind A -

Sol. A - A I = [1 -~ !]-
3 1 1
A. [~ ~ gl
0 0 1j 3 1
1
= [ ; A. _ ; _ A \
1-AJ
~
... (4 :. I A- :U I= 2
1-/\.
- 1- A 4
2 31
= -A.3 + A.2 + 18/\.+ 30
3 1 1 -/\.
get
Hence the characteristic equation is
3 2
- /\. + /\. + 18/\. + 30 = 0
Now, in order to verify Cayley-Hamilton theorem, we have to show
that - A3 + A2 + 18A + 30 I = 0

Here I = [~ ~ ~] , A =\ 1-~ !l
oa 1 l3 1 1J
A2 = [; - ~ ;J [~ -i !lj=I~~ ~ !l l
3 1 I 3 l 14 j
1 I 8 6

3
A =A2A=(~~
s
! :1li -~
1
!l =I~~~;
l3 J l62 62J
6 14
~!l 1 1 24
-A3 + A2 + 18A + 301
 ,1e 'i.Ci.T.

f ,, ,,
-
r>2 ~ '"1'
- 4
~1 2 "'
24 f2
- ':I
- !j
/.
6
4
~

(J , = J

5
L

'!nfiec
- .., ,
To Juul A - ~, [ - A - A - - ~ A =
I - -
I = - A -A- - 'It' A
3J
Premulupl., ing

..,:::
-H
j
30 30 30
")
JO 8
A - J=
30 30 30
5 5 5
30 30 30
Example 2. Illustrate the Ca~leJ-Hamilton :heorerr. _:or ;he 1r.c.:~..z
1 0 (J
A= 2 - 1 0
0 0 !
J
j - ). ") 0
I A - i.I I = 2 - l - i. 0
0 0 l - i.
= - 5 ..- 5i. - ).2 - i. 3
Hence the characteristic equation is .
- 5 ..- Si. - i.: - )..3 = 0
We have to show that - 51 + 5A ~ A 2 -A 3 = O

Now. - 51 + 5A + A2 - A3
-5
~~ g]+ (~~
-10
= 0 -~ g)'
0 -5
+(
ll~ 5 0 0 0 0 I
0O -5
J+ I-10
0
5
( 0 \ 0
 /t,/ c !ta u,( und Marli e111ar, u,/ vT
fhcurc rn JJ. / In , , Ju ,/ a rltu rmul
1
111011 t , r
/11 I
I ,1,,, dwg rmal
111 11tr 111

Proof. LetA = diag. la , 1 u _2

Then (A - I.I) == diag. [ a, J - i. a 22 I.
'Jhc churn<.:ten~t1<.: eyu8.ll<;n I A - i.l 1 :::. r1 gi ve

(11
11 - i..J (a 22 - i.) • (r.1 ,,,, -iJ=( J

a 11 , a22 · ann are the diagon al eleme nts of A . The refore the ti•
values A of a diagon al matrix are the eleme nts in the diagon al. / ?,

8.4 DiagonaJization of matrices.

Let A1, Ai· .. 'J...n be n dhtin ct eigen value s of a matrix A
X 1, X2, ..... Xn be then corres pondin g eigenv ectors . an,
Let X; be the col umn vector given by

X;=[1xm~J ...(;
Consi der a matrix P whose colum n vectors are n eigenv ectors such tha.t

X11 X12 X1n

P=
X21 X22 X2n
=[Xu ]
.. '2,
Xn1 Xn2 xnn
Suppo se that Dis a diagon al matrix such that
A-1 0 0 0
0 Ai O 0
D= O O ½ 0 =diag [ A- 1, ½, ... An ] ... (3)
0 0 O An
A1X11 JviX12 ...... ')..nX1n
Then PD= A1X21 ½X22 ...... A,rX2n =[A•X .. ]
'] lJ ... (4)
A1Xn1 ½Xn2 ....... AnXnn
(no summation over j )
= (AX1, AX2, ... AXn) (expressing matrix as vectors)
= A (X 1, X2 , ... X,,)
=AP.. ...(5)
 143
ftCC-1
{J / 1, Wl!
Jf l' be a non srngular matrix, then premultiplying (')) by p get
D=P - 1
AP ... ( ())
1
fhUS, p~ernultiplyi_ng A by p - and postmultiplying by P, we g<..,;t the
, nal rnatnX wh~se ?iagonal clements are the eigenvalues. This proces., i~
J1J!/1 he diagonaltzat w11 oft he matrix A.
r~11ed t -[3 4]
txarnPle. Let A - 4 -3
The characteristic equation 1s
IA-All= -A
3
4
4
-3-A - \-o
- 9 + A.2 - 16 = 0 or A = ± 5
i.e.,
"-1 = Ai= 5.
::=r~rr.::
- 5 and
i.e.,
Toe corresponding e•:~: mined •

Let P=[-; i]
Then p-l=-H-1 =i]
AP= -H-1 =i] [! _j} [~; :i]
1
Hence, P-

=
1
-H-~ =i] [-1~ ~]
=-![2t-2~]=[-~ ~]
D=[Ab ~]
Practical Method of Diagonalization. To reduce a given square ma
Ato diagonal form, we first write the characteristic equation for the m
a·d e\'aluate the characteristic roots At, ½ ... "-n· Then the required diag
ill~"' D of A will be
A.1 0 0 0
o Iv;_ 0 0
D= 0 0 ½ O

0 0 0 /1.n
Example 1. Diagonalise the following matrix
eas 0 - sin e 01
l
A = sin 8 cos 8 0
0 0 1
 12
Matrices II
~~.1. Special Types of ~at~ces
Here we define cert.am important matrices.
1. Square Matrix. A matrix having the same number of :-c :vs as the
number of columns is called a square matrix. For example f/ / n matnx
al I a12 ........... a,n

ra21 a22 ........... a2r.

ar.2 ar.ri

is a square matrix of order n. The elements a 1, a 22 ••••••a/lll are caUed its

diagonal elements. The sum of the diagona l elements of a square matrix
is called the trace of that matrix.
Tr A= I,au
,-
2. Diagonal Matrix. If all the elements of a square matrix are zero
except those m the leading diagonal, then the matrix is said to be a dragonal
matrix. For example
a 11 0 0 .. ········ 0
0 a22 0 ........... 0
0 0 a33 .......... 0

0 0 0

is a diagonal matrix of order n.

Thus a n-square matrix A= [a,)n x n is diagonal matrix if and only if
a = f O for i *j
1
1 i "# Q
\.
for l =]

a= a,1 8iJ
It can also be written as
A =cliag. [a 11 , a22.. .......a,)
3. Scalar Matri.L A diagonal matrix in v.-hich all the diagonal elements
are equal is called a scalar matrix.. Thus
183
 184 A1echamcs and Mathemat1ca/ p1,,.....
".YSlcs

'}.. 0 0 ........... 0
0 1.. 0 ........... 0
0 0 J.. .......... 0
1s a n x n scalar rnati
rx.

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

Thus a n-square matrix A = [a,)n x n is a scalar matrix if and only if

01
1-
-{i..
0
for
for
i=j
i -:t- j
4. Identity or Unit Matrix. A scalar matrix in which each diagonal
element is unity. is called an identity or unit matrix. Thus an-square matrix
[a,) 1s a unit matrix if and only if
I for I= J
01
1= {0 for

5. Row l\-1atrix and Column Matrix. A matrix having one row only
i.e. . a matnx of order 1 x 11 is called a row matrix or a row vector and is
expressed as [X] 1 xn = [a 11 a 12 a 13........... a,n]
A matrix having one column only i.e .. a matrix of order m x 1 is called
a coumn matrix or a column vector. It is expressed as
a,1
a:n
a3,
{X}mx I=

6. Null Matrix. The m x n matrix whose elements arc all O 1::i called
the null matrix of type m x n.
7. Upper-Triangular Matrix. A square matrix, whose every element
a,1 for i > j is zero, is caJJed an upper triangular matrix. Thus
all a, 3 ........... a,n
0 a 13···· ........ a2n
0 a 33 ............. a 3n 1s an upper triangular matrix .

.. .......... ............ .............. ......

0 0 0............... ann

8. Lower-Triangular Matrix. A square matrix whose every element

a,1 for i < j is zero, is called the lower triangular matrix. Thus
 I I

(J I/
(J
I I r,
')

a, a l a

9. fr 1m.spo e of o Mntnx A 11111 , ix ,f rd , n, II ,n r

fl 111 Ill ,ow and c. l1lfr111 (ti I(/// /. n) II frt

I 1 ind , dcnul cl I y II y11Jl1ol \' ,A r

rA 1A
fJ I I a, a,n
11 a I
t1 I
a,
If A- th n A AT

am, llm2 .a a,
p10 perties
1
/1) (/\ / =A.
I •1 ,, T 1 /
(11) (M) rv1 , fl, ueIng w,) \Cldw 1re£1I or comp[ r)
7 7
(111) (A I /1) A 11P,1\ ,111d/J bci111: ,on{omiahlefo, uddu1m1
I I I
(n) (All) IJ A ,A :uul IJ ht111g 11m/01111<1hl1 J-n 11111lr1rJl1 a11011

10. Complcx-Conjugufr Mutrix. 'I h1: rnm11fn nm111gI11e of ,n 11 h111 11 )

111ut11x A 1s formed by 1ak111g thl lompln lo111ug11k ol e II h clc111e111 I h n1 1.
we have
A. = u~ (Im all i and Jl
If A· = A, then A is a real m,1tr1 ~.

Propertic!:i
0
(1) (A / = A.
0

(11) (A + B)° = A• -+ B A and IJ brmg , to11Jo111wl,/,,.

0 0
(lit) (M = "A.." A A. being wn ,omple., rtu111hcr.
) ,

(h·) (AJJ)° = A• JJ°, A nnd B !J,,ing um/01111al>I,· /or 11111lt1plicat1m1.

1

ll. Hcrmil ian Con,iu~atc (Th(• conju~nfr transpose of a matrix)

'fhc llrrmitian ronJugat,• of an 111h1t1a1y mat11x A is ohtamcd by t,1k111g the
complex conJugatc of the matrix and then thl' t,an~posl' of till' co111plex
conJugatc matrix.
• I r •
At= (A ) = (A )
 186
Propcrfie.c; ,
. (A t)t == A. t+ Bt ,A an dB b em
. g con fom iab le for addu1on '
11
(1) -A
(ii) (A+ B)t - ple x number. . .
• t A being anY· .com iformable for mult1p hcation .
.. ') (M )t == A AA •t A and B bem g con
(111
(iv) (AB )t == .B~nt isy mm etr ic Ma tric es
t .
12 Sym me tnc and wil
. 'd to be a symmetric ma tnx .

:x:~:: A[iss;, ~]
•

e .
c d
. imetric (skew) ma tnx .

:x:::A[, ~ ~s sait tJ an ant ,sy "

VI

-b -c 0
.
13. Hermitian matrix
· · ma tnx·
If A t = A, A is said to be a He mu tla n 1

Example. (
0
1
-~ J
ical obs erv abl es are rep res ent ed by Her-
In quantum mecham•cs, aII Phys
mitian operators (matrices).
14. Skew-Hermitian Ma trix
ermitian matrix.
If A t = -A , A is said to be a skew-H
1
Ex am ple . [-IO+; ; ;]
es
15. Sin gul ar and No n-s ing ula r Ma tric
A square matrix A is said to be singul
ar if I A I = 0

A square matrix A is said to be non

-singular if I A I =I- 0.
16. Un itar y Ma trix
A square finite matrix A is said to
be unitary if
At A =I
Thi s implies AA t =/

where A t is conjugate transpose of A and I is a un it ma trix .

I A ti I A I
We know that I Ati = I A*1 and I At A I =
I A I = I / I = I.
Hence if At A = I, we have I At A I = I A ti
tary matrix is always of unit
This shows that the determinant of uni
non-singular.
mo dul us an d hence a unitary matrix is
 Analysis 187
vector~
11"'2 il..Jfl
t:ia,nple. [ -il...ff 2°j
- ll'✓

orthogonal Matrix
11, fl . . A . .
A square imte matn.x is sa,d to be orthogonal if
AT A =I
'[his implies A AT= I
Ar is the transpose of A and / is unit matrix ·
where
We know that I A Tl = I A I and I AT A I = I A Tl I A I
2
Hence if AT A= I, we have I A 1 = I i.e., I A I=+ 1
This shows that the determinant of an orthogonal matrix can only have
values + 1 or - 1.
cos 0 -sin 0]
. 0
Example. [ sm cos 0

JS. Adjoint of a Matrix

The adjoint of matrix A is defined as the transpose of the matrix formed

b) the cofactors of the elements of I A I.
all a12........ aln

'½1 a22 ........ a2n

Let A=

Let us denote the cofactors of elements by the corresponding capital

letters (e.g., the cofactor of a 11 is A 11 etc.). Then,
a
A 11 = (- 1) I + I 22
G32

A
13
= (- 1) 1 + J a21 a22
a31 a32

A 23 = (- 1)2+3 au a12
a31 a32
 188 Mechanics and Mathematica/ Ph
1
~Sic~
a a
•A =(- 1)3+2 II 13
32 a 21 G23

The matrix of cofactors of det A is

A11 Al2"'"" "AIII
A21 A22 ........ Ai,,
[A I)..]= ···························· ······
··································

Now taking the transpose of the matrix of cofactors of det A,

All A21""""Anl
A12 A 22 ........ A112
adj A=

12.2. Inverse of a Matrix

Let A be a square matrix of order n. If there exists a n-square matrix
B such that
AB= BA = I,, where /11 is n-square unit matrix,

then B is called the inverse of A and A is called the invertible matrix.

1
The inverse of matrix A is denoted by A- •
Theorem. The necessary and sufficient condition for a square matrix
A to possess an inverse is that it is non-singular; i.e., I A I:;; 0.
Proof. To prove that the condition is necessary, let us assume tha1
A is a given square matrix and let B be its inverse. Then by definition.
AB=B A =I ...(ll

Taking determinant of either side, we get

I AB I= I i.e., I A II BI= I/ I
t.e., I A II BI= 1
This implies I A .II B I :;; 0
Consequently I A I:;; 0 i.e., the matrix A is non-singular.
Thus the necessary condition for a square matrix A to poss~ss an inverst
is that I A I :;t 0.
,. t4rtO(l'SlS
\89

o"e that the condition is sufficient, let us assume that A is any

fo pr rnatrix i.e. , I A I* 0. Then we have simply to prove that the
A e,dsts. Let us. d e iime a matnx
£!1.llaf
!'l",1fl~
. B such that
of• .
,-!r.t B = adJ A
IAI

'f}lell•
_ A (adj
AB - IA I
A)= A (IadjA IA) =~
IA I
AB=!
t• -(~dj AJA= (adj A)· A= lA\l
.A.Jso BA - IA I IA I IA I
BA =I
(· AB=BA = I
ThUS
. hows that B is the inverse of A.
htCh
11 C
s .
The inverse of A 1s defined as
or.
B =A-1 = adJ A
IA I
Example 1. Find the invelrseof_r;e m;]tri.x
1
-1 1 2
3 2 -1

1
Solution. I A I = -1
3
-~ l\
2 -1
=-25

A 11 =-5, A 12 =5, A 13 =-5

A =5, A 22 =- 10, A 23 =- 5
21
A31 = - S, A32 = - S,A33 = O

Therefore, the matrix of cofactors of I A I is

-5 5 -SJ
[A, ]=
1
l 5
-5

Taking the transpose of the above matrix, we get

-10
-5
-5
0

l
-5 5 -5~
adj A= 5 -10 -5
-5 -5 0
-VS
d.
A-1=~- _1
IA I -- 25
l-5S
5
-10 2/5
vs
1/51
1/S
0
-5 -5
- ~A
t~ hcorcin, ff
Mechanics and Mathematical physic.,
1
d Bare rwo 11-square, non-singular matrices • hen
i,l;;.,i

yec101

1
011 (ABf I == B- 1 A-1
T I 1
Proof. since A and B arc non-singular matrices, I A I CFO and B
I AB I =" A II B \ CF O "fl

... '!1iis impli,•s that AB is non-singular matrix and hence invertible

Let A- and B- 1 be the inverse of matrices A and B respectiv~l;
1
AA-I == A-I A := / .

BB- 1 == B- 1 B == I
Now consider a matrix C given by 1
C == B- 1 A-
1 1
1
TI1cn, C (AB)== (JT 1 A- 1) (AB)== B- (A- A) B == B- JB
1
. == (B-1 B) == I (Since A- A == J and JB == B)
1 1 ...(la)
i.e.· (B- A- ) (AB)== J
Similarly it can be shown that

Equations (I) imply that

1
(AB) (B- 1 A- ) = I
is the inverse
-I 8 -1 tA-I . . of AB .
' z.e.,
...(lb1 I
B) == B- A-t ... (2) \
 = -.z:i\T 1'•)11
I

[Since trace of a matrix is the sum of its diagonal terms]

1 1
= L L P~y A.I'k pkI. =LL
I ./, k
L(Pk;P;;
I j k .
)A jk
.

= ~L(P p-l )k1 A·1k = "L(P p-J A)kk = "L Akk = Trace A
.! k k · k

i.e., Trace B = Trace A

12.13. Unitary Transformation
Consider a linear transfonnation
Y=AX ... {I)
such that X and Y are column vectors of order n x I and A is a square
transformation matrix of order n x n. If the matrix A of transformation is
unitary, the linear transformation is said to be a unitary transformation.
From definition of unitary matrix
 ~
I ?09
At A AAt I tl)
VY (AX)t ( U) \t It A\ \ 1,\ l 1 ,1)
ti

or L>'v1 L,x: \
I I
(3b)
I 1

:::, 1/Je norm of, , tor,

lll\'cir 1a111 um1er 11111/a,y /ramjorma/11111
1~

As nn '"' ei ,c \.nsc, ums1dc1 a linc.u t1 ,msfomrnt1on matrix A which

a, e the nonn ol 1;, Cl) , cch'f c)t· a vector . space um:hnngcd, thnt ,~
I l I IX I l,r I AX I x
\t X (A Y)t f\ \ t tt I\
L)b, 1l)l1~h At \ l. ~Ihm mg tlMt I is unit,\!)
I hu~, :-\ l1tH ar lransfor111011011 acr111g 0 11 a ,,l <tor 'l"'"e n w111w·y if
nl) 1 II /cm,•' llrt! norm of l'\'l 'J' ,·ector 111 the Vt'< 101 ,pace um hanged
2. t 4. Orthogonal frnnsformution
c~,n,1lkr th1.· folllm ing trnnsfonnation frnm the ,-variables to the x-
' nablc",
\ I ½- a , I ~
\ I <111 1 2 ll I n \ 11'

I a., --1 t (I
\2 a2, :\"
I
\ ' n'

ii,. I\ 1 a,. \' ➔ t (Ill/ \ 11'

AX.
1
'
,here , () , J , ; ,. )' and , (\' 1. , 2 ... .. ,,,)' ,md A ta) 1s the matrix
tr n fomuiuon In 1.:ase b) th(' II ,mslrn m,111on the sum ol square::,

v·, X'X - ,,.' .

\

n It 1 \.ulled nn ortlwglin,tl t1M1!:ifotm,1t1on. The m,1trix J\ l)I such a

mgul,ir
onnatIon I non-
Ob, ,ou I~ the ond111011 of trt1m(o, 1111111011 I \ to bl 011hogonal
ti t th tr m ft rmatIm1 111rItrI\ I '"11,Jie, thl.' u\m.i1t1on
\ \' A'\
Thcnn m. I he prtldlll:I of l\\ll ,Hthngnn,d t1,rns1~,,mutwns ti •
ortho 1onul tran 1orn1at1on
Proof. I ct , AX and \J HV \w t,H) tHthognn.11 t1,msllmnations

\'A I, HH'
1hen,
(AHHI\B)' AHif A' AIA'
t\1\' I