ltii.GEBRA : DETERMINANTS.

MATRICES

• ~pie 2-24. For.the matri x A=[: ; ~]•

O -1 -1
•i
l&Ol&~ingular matrices P and Q such that PAQ is in the normal form.
Bent:e find the rank ofA . (Raipur, 1998)

• write A = IAI, i.e. [l1

8oL We 21 32] =[l0 1 0OJ A [1O 01
O 0OJ
I O -1 -1 0 0 1 0 0 1
, .
subjecting
all affect every elementary row (column) transformation of the product by
~ct or (post factor) ofA to the same.

Opei,ate C2 -Ci, C3-2 C1

:
[~
0
~
-1
. ~]=[~ ~ g]A[~ -! -~]
-1 0 0 1 0 0

~~R rR1 , [~
0 -1
~ ~]=[-! ~
~1 0 0 1
[l
, 0OJ AO
0
-11
0
-2]
0.
1

[! j OJ ['1
g = -~ ~ g]A[~ -! .:..1] .
-1
1
0 1 0. 0

~rat e R3 + R2, [~
0
~
0
g] .[-~
0 -1
~ . g]A[~0
·1 1
-1 -1]
1 .-1 .
0 1 . ...
. . .·.

12 0
·c1a ii of the normal form 0 0
0 0 1 -
• 1 •11 1 ... - • f

1 .. -1 and p(A) =2.

I
t
Bence 1 0 , Q= 0 • t

1 1 0 0 1. ..
.

) .
Pn>hh .'n1s 2 (>

....-..,.··■ne the rank of the following matrices (1-5):

1.
.-3
[l .-~ !] 1
-
2--
CAnclhra, 2000>
1
2.·[a1 : "ci.
3 4
:11J • (Madras,
.. .. ,,
2000 S>
,

1 2 3
(

0 5 6 7 8
3 2 4. 6 7 8_ 9 - !
2 4 (Osmania, 2003 S)
3. 3 11 12 1s·. •14
3 2 1
7. 5 16 . 17 18 19
6 8
5. Prove that the row equivalent matrices have the same rank. .. • : I , •••

8. Us[e G~u; 8~;]~ ~~tho d to find ~e ~vers e ofthe [r:n7 '3ma]trices:

2 (Mangalore, 1997)
(i) O 2 1 (Coimbatore, 1997) (ii) 2 1 1
5 2 -3 • • · . 1 2 1

7~ .If A;:[~ j !]
findA-1. Also find tw~ non-singular matrices P ~ Q : ~ ~t. PAQ. ~I,
•
~here
..
; •
I . o -1 1 ••
• • •
I is the unit matrix and verify that A-1 =- QP. ,,. - • • • • ...,_ r • • •· • • -- • • ••-~-
a HIGHER ENGtN
8. Reduce each of the following matrices to no
~ form and hence find their ran b :
(i)[ :
-8
~ :
-1 -3
:] (J.N.T.U., 2000) • • (ii)A =[: J -~ -:i
4 . 1 1 1 2 CU.P.r.~
9. Find no~-singular matrices P and Q such that
PAQ ia in the nonnal rorm for the matri
• [1 -1 -1]
(i)A = 1 1 1 CJ.N.T.U., 2002) (ii)A =[12 -22 . 31 -2]
3 _cea:
~
3 1 1 3 0 4 1

2 9 (1) VECTORS

Any quantity having _n-components is called

a
in a linear equation or the elements in a row or vector of order n. Therefore the
numbers .xi, .x2, ......, Xn written in a particular orde mn matrix will form a vedor Th
cc:,lu
r, constitute a vector x. • 111
(2) Lin ear dep end enc e. The vectors xi, X1,••····•
Xn are said to be line arly deJ>eJldt
there exist r numbers Ai, ~......, A.,, not all zero, such
that
A1X1 + ~X2 +.... + A,,Xr = O.
~
If no such numbers, other than zero, exist, the vect
ors are said to be linearly indel)en(
If A1 ~ 0, transposing A1x1 to the other side and divi
ding by - A1, we write (i) in the rr,
x 1 = µ2x2 + µ3x3 + ••·••· + µ,.xr
The n the vector x 1 is said to be a line ar com
~ination of the vectors x , x , ..., x,.
Example 2· 25. Are the vectors x = (1, 3, 4, 2), x . . 2 3
1 = (3, - 5, 2, 2)
dependent ? If so express one of these as a linear com2 and x 3 = (2, - 1, 3, 2) •
bination of the others.
Sol. The rela tion A.1x 1 + ~x2 + AaX3 = O. ·1
i.e., A1(l , 3, 4, 2) + ½(3 , - 5, 2, 2) ~ ~(2 , -:- 1,
3, 2) = 0
is equ ival ent to A.1 + 3½ + 2¼ = 0, 3A. - 5½
1 -1s = 0, i • i._ •
4A.1 + 2A.-i + 3A.3 = 0, 2A.1 + 2½ + 2Aa = 0
As thes e are satisfied by_ the values A. = 1,
vectors are line arly dependent. Also we hav e
1 A-2 = lfs = - 2 which are not zero, the •
the relation,
+ X2 - 2X3 = 0
X1 • : '•• . .
by means of~hi ch any of the given vectors can
be expressed as a linear combination ofthe o
' . . . ~

Obs. Applying elementary row operations to the

vectors s1, st, s8, we see that the matrices
St
and B= Si
s1 +s2 -h3
have the same rank. The rank ofB being 2, the rank '
ofA is also 2.' M o~ ~ , ~ are linearly inde
1

~~
and x3 can be e~presaed as a linear combination
ofx1 and 1'2 [ :::·:S:- <•1 +-~} "'½ reaulU
hold for column operations and for any matrix.
In general, we have the following results :
If a giuen mat ru ha, r liMarly independent vecto
linear combination, of the• r vector,, then ranl r, (row_• ~r columna) and ~ r,mai_ni?f "";, .
contain, r linu.rly indepenunt vector, and rem of the ~ , u II r, ConW'11'1, if a ma w~ of
ainin, uecton (if any) mn 6, ,xp m• a, 0 ,~
combinatio~ of the• vector,.
 8 -;

.45
~EAR ALGEBRA : DETERMINANTS. MATRICES

Pn,hl.-111 s 2 7

•re the following vecton linearly dependent. If 10, find the relation between them :
1. (3, 2, 7), (2, 4, 1), (1, - 2, 6).
2. (1, 1, 1, 3), (1, 2, 3, 4), (2, 3, 4, 9).
3. Xi= (1, 2, 4), :1t2 = (2, - 1, 3), z 3 = (0, 1, 2), z.. = (- 3, 7, 2). (U.P.T.U., 2003; Nagpur, 2001>

l 0. SOLUTION OF LINEAR SYSTEM OF EQUATIONS

I (1) Method of determ inants - Cramer 's* rule
t ~~~sider the equations a 1x + bJY + c1z•= d 1
02-X + b'})' + C2Z = d2
1 , . •...(i)

_ ._ OaX + b3Y + CsZ := d3

-.._ ·. •• ~ · ~l • b1 c1
If the determinant of coefficients be~= a2 b2. c2
•------ .-- -- --.:.........-~--~---·----··•·- a3 ____ b3·- c3 ---------·· - -------·· ------- ····•··- ·------·
.,

~ , ~

-.
m1 bi C1 ,

. . .
t b2 [Operate C1 + yC2 + zCg] " • • I •

xi!.=
•

1en ~ C2
ms bs C3

a1x + b].'Y + c1z b1 c1 dt b1 Ct

[by (i)]
- a,p + b21 + C'J.Z b2 c2 - d2 b2 C2
a~ + b3Y + CsZ bs C3 ds bs C3

d1 b1 Ct a1 b1 Ct
.
b2 provided ~ ¢ o.·_; ...(ii)
Thus x= "2 b2 c2 + °2 C2
ds bs C3 as bs C3

a1- d1 : C1 a1 bi Ct
...<iii)
b2 c2

i Similarly, y= a2 d2 c2 + a2 f ,

, ( . G3 . d3. .,C3 as bs C3
•• .. ,I

.. '
< •

a1 b1 d1 a1 b1 Ct
...(iv)
f.nd z= G2 b2 d2 + a2 b2 C2
aa bs ds as ba C3

Equatio ns (ii), (iii) and (iv) gi,ving the values of x, y, z •constitute the Cramer's rule, which .
equations (i) to a problem in evaluation of determinants.
educes the solution of the - linear•

(2) Matrix inversio n method

If A= :
a3
::
bs
:: ,x=[;Ja ndD =
c3 . z d3
:~i •

n the equations (i) are. equivalent to the ~atrix equationAX=D •••(v)

here A is the coefficient matrix. · ·.-· . •

• Gabriel CratMr (1704-1752), a Swiaa mathematician.

 HIGHER ENGINEERING lAb.
Multiplying both sides of 1 ~
(v) by the • roca1
recip matr ix A - • w e get 11
A-1 AX =A -1D ...
or IX =A -1D
A1 A2 A3 I·: I
d1
~
or i.e.,
l1 ; :
z
Bl
C1 • C2
B2 B3 X d2
C3 d3

where A 1, B 1 etc• are th e cofactors of a1

ai , b1 etc. in th e determ b1 C1
inant-A= a
2 b2 C2 (A
Hence equating th e va b3 C3 a3
side of (vi), we get th e de ~ues of x, ~, z to th e corresponding elements in th
sired solutions. e product on~
. _
• Obs. When A is a sin
gu la r matrix, i.e. A= 0,
of equations an d the nu th e a~ ve ~t ho ds fail.
mb er of un kn ow ns are These also fail when Utt,
wi th such equations as unequal. M at nc es can,
wi ll be seen in § 2·11(2 ho we ver, be usefully applied
).
1
Example 2-26. Solve the
equations 3x + y + 2z = 3,
by (i) determinants (ii
) matrices. 2x - 3y - z = - 3, x + 2y +
z =4
(i) Sol. by determin
ants

Here 11 = 1~ -; -ii=
1 2 1
3( - 3 + ~) - 2( 1- 4) + (-
=
1 + 6) 8 _[Expanding by
c1
1
... x= .!
11 .I _;
4
-i2 -i1i [ExpandbyC11

i
= (3 (- 3 + 2) + 3( 1- 4)
+ ~( -1 + 6)1 = 1
Similarly, y = !.
A 1
1~ -! -~1 4 1
=2 and z =!
6
1~ -i
l
Hence x = 1, y = 2, z = - 1. -! l= -1
No te . Th e us e of Cr
such an d ot he r cases, th amer's ru le involves a lo t of lab ou r wh en th e
e nu me number of equations exceed
(ii) Sol. by Matrices: rical methods given in § 22·4 to 22·6 ar e preferab le.
si
-

•
Here A=
• . 1. 2
1: j -~1 1~ :: : \
1
=
a 3 b3 c ~say).
3
Then A1 = - l,A 2 = 3,
A3 = 5; B1 = - 3,-B = 1,
Also A= a 1A1 + a~ 2 + ay \3 2 B3 = 7; C = 7, C =~ 5, C3
= 8. 1 2 =-ll.

... x1
_ y = ~
A1
B1
A2 Aa d1
B2 B3 x d 2
lz C1 C2 Ca d3
= ! =~ ~ ~ x -~t !r-:=;:~~tr ~1
1 - s
ce X = 1, y = 2, Z = - 1.
- 11 "J l21 + 1s - «J l-1J
~ A1
'.INEAR
I
ALGEBRA:
t
DETERMINANTS. MATRICES

Example 2·27. Solve the equations x 1 -x2 +x3 +x4 =2 ; xz +x2-xa +x4 = -4
:,1 + xa + x3 - X4 = 4 ; x 1 + x 2 + x 3 + x 4 = 0, by finding the inverse by elementary row operations.
Sol. Given system can be written as AX= B, where
1 -1 1 1 X1 2
1 1 1 •1 X2 -4
A= : X= , B= 4
1 1 -1 -1 %3
1 1 1 1 X4
0

To find A- 1, we write
1 -1 1 1 : 1 0 0 0
1 1 -1 1 : 0 1 0 0 [R2-R1
[A:/]= 0 0 1 0 [Ra +R1
1 1 1 -1 :
1 1 1 1 : 0 0 0 1 [R4 +R1

1 -1 1 1: 1 0 0 0 1
l2R2
0 2 -2 0 :-1 1 0 0
- 2 0 2 0-: 1 0 1 0 [21Ra
2022:1001 1
[2R4

1 - 1 1 1 : 1 0 0 0 [R1 + R2
0 1 - 1 0 : -1/2 1/2 0 0
- 1 0 1 0 : 1/2 - 0 1/2 0
1 0 1 1 : 1/2 0 0 1/2 [R4 - Ra

1 0 0 1: 1/2 1/2 0 0
0 1 -1 0 : -1/2 1/2 0 0
- 1 0 •1 0 : 1/2 0 1/2 0
0 0 0 1 : 0 0 -l/2 1/2

1 0 0 0 : 1/2 1/2 + 1/2 -1/2

1 1 0 0 : 0 1/2 1/2 0
- 1 0 1 0 : 1/2 0 1/2 0
0 0 0 1 : 0 0 -1/2 1/2

1 0 0 0: 1/2 1/2 1/2 -1/2

0 1 0 0 : -1/2 0 0 1/2
= 0 0 1 0 : 0 -1/2 ·o 1/2 ..
0 0 0 1 : 0 0 -1/2 1/2

1/2 1/2 1/2 -1/2

-1/2 0 0 1/2
Thus, 0 -1/2 0 1/2
0 0 -1/2 1/2
1/2 -1/2
1/2 1/2 2 1
-1/2 0 0 1/2 -4 -1
Hence, X =A- 1 B = 0 -1/2 0 1/2 4 - 2
0 0 -1/2 1/2 0 -2
. Xt = 1, %2 = - 1, %3 = 2, %4 = - 2.
i.e.
 HIGHER ENGINEERING MATHE¼

Solve the following equation• with the help of determinantl (1 to 4)

:
I. .x + y + z = 4 ; x - y + z • 0 ; 2x + .Y + z • 5.
<O,mania, 2t
' • % + 3y + 6z : 2 ; 3x - Y ♦ 4z a 9 ; % - 4y ♦
2z C 7•
S. .X + ,Y + Z s 6-6 ; X - y + Z = 2·2 ; % + 2y + 3z • 15·2.
4 . .x2z3 /y = e8 ; y 2z/:r = e4 ; x3y/z4 = 1.
5. 2vw - wu + uv = 3uvw ; 3vw + 2wu + 4uu = 19uvw ; &,u, + 7wu - uv
= l 7uvw.
Solve the following system of equations by matrix method (6 to 8) :
6. 2.x + Sy + 3z = 1 ; - .x + 2y + z = 2 ; x + y + z = 0.
(J.N. T. U, 20
7 . .x + y + z = 3 ; .x + 2y + 3z = 4; .x + 4y + 9z = 6.
<Bhopal, 20
8. 3x - y + z = 6 ; 4% - y + 2z = 7 ; 2x - y + z = 4.
<Andhra, 1a
9 . .x - 3y - 8z + 10 = 0 ; 3x + y - 4 = 0 ; 2x + 5y + 6i = 13.
(Madras, l~
10. :r2yz = :ry2z3 = .x3y2z = e. .

11. By finding A-1, solve the linear ~uati on AX= B, where A = [ !~

<A.M.I.E., 1997)
12.. In a given electrical network, the equations for the curren ts i1,
i2, • ◄-
is are
. 3i1 +it+ ia = 8 ; 2i173 i2 - 2ia = - 5 ; 7i1 + 2i2 - Sia= O.
Calculate i 1 and i3 by Cramer'■ rule.
U. Usina the loop mnen t method on a circuit, the following equations are
obtain ed :
7i1- ~c 12,-4 i1 + l2i2- 6ii·•0 ,-8ii+ 14ia~O.' '
By matri& met.hod. aolve for i , i and .i . •·, : ' ; : :: 1 n•
1 2 3
14. Solve the foJ.lowint . t i ~ by caicu 1atiq the·i n~·b ; elem en~
row operations :
2z1 +2.z t+'-a
, -ax.• 2 ;:U1 •~-2 za+s ,•8 ;s1 +st-: Ua-4
. ... -. " .x,•- l; 2x1
.
+~+ 5x3+.r4= .,..

Conaidel- the -,.tem of m linea r equations

G11S1 + Gt~2 + ••• + Gl,sX,a = 11
• : •Gt1S1 + ~ 2 +.~. +~,a =Ii
r • . : . :
......... ......... ......... • '..
......... ......... ......
4 m1S1 +Gm ~+·... +G~ n
., =•;..
. cootaioing then unknowm s1; zl, •..., Sn~ To d-.n ine ~~• · lhe. ~tb
pa11e11 a eoluti on) or not, we
ul (i) are consistent
. the pltric ea..~
OODli
.. . der.the rao)y of _; :. : .. •
...
4 12 ... , Gt,a :, au G12 '.) .•. 1 G1n k1
422 ·•· Clt,a and- i -K= f21 ~~2
•••
4 ml
• ••
4 m2
•••
•••
•••
am,a
•
.'.. ~
••• ... -~· . a2n k2
...
. - .
A is the co-ef(u:knt matri s and Kia called the augmented. mat, u of the \
Gml . Gm2 ;;. amn km
equations (i).
(S) Boue be'• di-- --. The v•tem of ~uati ons (i) is
consistent if and only if the coe
matr u A and tlN IJll61'Mnl«I. motrls Kare of the aa.me rank-otherwise the system is
. I incon s. {

l
Proof- We conak;ler the following two .poeaib
, •
le. cues : ..
• •
,
. •
: •
I. BanJi ofA • rani ofK • r (r $ the rmaller of the n~b en m and n). The equat
by •uit.able n,w opera tlou, be reduced to ions Ci)
. . . , ._
• , ' ' ~
~AR ALGEBRA : DETERMINANTS. MATRICES 49

b11x1 + b1~2 + ...... + b1nXn =l1

0.x1 + b2~2 + .. :.. , + b2nXn = l2 ' ... (ii)
.................................................
0.x1 + 0,x2 + ...... + b,,,x11 =l,

d the remaining m - r equations being all of the form O.x 1 + 0.x2 + ...... + O.xn = 0.
The equations (ii) will have a solution, though n - r of the unknowns may be chosen
)itrarily. The solution will be unique only when r =n. Hence the equations (i) are consistent.
,~ II: Ran~ o( A (i.e. r) < ran~ of K In particular, let the rank of K be r + 1. In this case, the
~uabons (i) will reduce, by suitable row operations, to •
~ b11x1 + b12X2 + ...... + b1nXn =l1,
\ 0.xl + b2~2 + ...... + b2nXn = l2,
....................................................
0.x1 + 0.x2 + ...... + brnXn =l,,
0.x1 + 0.x2 + ...... + O.xn =l,. +i,
d the remaining m - (r + 1) equations are of the form 0.x 1 + 0.x2 + ...... + O.xn =0.
a Clearly the {r + l)th equation cannot be satisfied by any set of values for the unknowns.
?nee the equations (i) are inconsistent. •
[Procedure to test the consistency of a system of equations in n unknowns :
Find the ranks of the coefficient matrix A and the augmented matrix K, by reducing A to the
angular form by elementary row operations. Let the rank of A be r and that of K be r' •
(i) If r ~ r', .the equations are inconsistent, i.e. there is no solution.
<ii) If r =r' =n, the equation•s are consistent and there is a unique solution.
(iii) If r = r' < n, the equations are consistent and there are infinite number ofsolutions. Giving
•.bitrary values ton - r of the unknowns, we may express .the other r unknowns in terms of '
·ese.] ' •
E~ample 2-2~. Test for consistency and solve ,••
I +
5x + 3y + 7z ~ 4, 3x + 26y + 2z = 9, 7x 2y + lOz =5. • (V. T. U., •2004 ; Andhrd, 1998)

Sol. We have
[! 2! .1!][~]=[i]
9
Operate 3R1, 5R2, ·[15
15
7
130
2 !g][i] =[!!]
r~
9
Operate R 2 -Ri, 121
2 -mm=r~n
7 1 305 21
Operate R 1, 5R3, ii R2,
3 [ 35
1
11
10 :!][i] =[:I]
1
Operate Ra - R1 + R2, R1,
1 1
! _-l][i]=W
~ ~50_ _ _ _ _ _ _ _ _ _~..=.:...: .:::..~~
_,, The ranks of coefficient matrix and augment~d matrix for_ the l~st set of equation,.
2. Hence the equations are consistent. Also the given system 1s equivalent to

5x + 3y + 7z = 4, lly - z = 3, ... 3
y = 11 +
z
ii and X =
7 18
ii-u•
where z is a parameter.

Hence x = _J_ y = ~ and z = 0, is a particular solution.

11' 11

Obs. In the above solution, the coefficient matrix is reduced to an upper triangular
row-transformations.

Example 2-29. Investigate the values of "A. andµ so that the equations
h+~+~= ~h+~-~ =~h+~+ b=~
have (i) no solution, (ii) a unique solution and (iii) an infinite number of solutions.
• (Ranchi, 2

Sol. We have [~
2 3
~
-~] [;] = [:]
"A. z µ
•
··
The system admits of unique solution if, and only if, the coefficient matrix is of rank
requires that
2 3 5
7 3 -2 = 15(5 - A.) ~ 0 • I
2 3 A.
Thus for a unique solution A.~ 5 and µ may have any value. If A.= 5, the system will hav
solution for those values ofµ for which the matrices
5
A= 2
7 33 5] and K = [2
-2 7 33 -2
[2 3 5 2 3 5
are not of the same rank. ButA is of rank 2 andKis not of rank 2 unlessµ= 9. Thus ifA.=5
µ it 9, the system will have no solution.
If A= 5 and µ = 9, the system will have an infinite number of solutions.
<3) System of linear homogene ous equations. Consider the
equations
a11x1 + a12X2 + •···· + a1n%n =0
a21x1 + a22X2 + ...... + a2ftXn =0
....................................................

Find t~e rank r of the coefficient matrix A bY. r~ucil)g it to the triangular form by elemen
row operations. . , .
l. If r = n,· the equations (iii) have on/,:y a trivial zero solution
X1 = X2 = ...... = %n = 0
If r < n, the equations (iii) have (n - r) linearly independent solutions.
The number of linearly independen t solutions is (p - r) means, if arbitrary values
igned to (n - r) of the variables, the values of the remaining variables can be uniquely fou
us the equations (iii) will have an infinite number of solutions.
 --
51
LlNEAR ALGEBRA : DETERMINANTS. MATRICES
than the num ber of variables), the
II. When m < n (i.e. the num ber of equations is less
solution is alwa ys other than x 1 = x 2 = ...... = Xn = 0.
The num ber of solutions is infinite.
ber of variables), the necessary and
III. When m = n (i.e. the num ber of equa tion s= the num
= ...... = Xn = 0, is that the dete rmin ant of the
sufficient condition for solutions other than x 1 = x 2
coefficient matr ix is zero. In this case the equations
are said to be consistent and such a sof ution
d the elim inan t of the equations.
is called non-trivial solution. The determinant is calle
Exa mpl e 2·30. Solve the equations
+ 12z =0 (Osmania, 1999)
' (i) x + 2y + 3z =0, 3x + 4y + 4z =0, 7x + lOy
(ii) 4x + 2y + z + 3w =0, 6x + 3y + 4z + 7w =
0, 2x + y + w =0.
Sol. (i) Ran k of the coefficient matrix

I
2
4
10
!]-[~ -~ -~]
12 7 10 12
[Operating R3 - 3R1

-[g -~ -i]
"
(Operating Ra - 7R1 - 2R2

l is 3 whic h= the number of variables (i.e. r = n)

z = 0.
:. The equations have only a trivial solution : x = y =
(ii) Ran k of the coefficient matrix
2 1 2 1

[i3 4
1 0 !]-[i .o 5/2
0 -1/2 -1/2
5/23] [Operating R2- : R1, Ra- ~ R1

❖]
n
[Operating Ra + ~ R2
2 1

-H 0
0
5/2
0
is 2 whic h< the num ber of variable (i.e. r < n)
system is equivalent to
:. Number of independent solutions= 4- 2 = 2. Given
5: 4x + 2y + z + 3w =0, z + w =0.
:. We have z =-w and y = - 2x- w
x and w being the pa;r:ameters.
which give an infinite number of non-trivial solutions,
tions
Example 2·31. Find the values of Afor which the equa
IL

(l - l)x + (3l + l)y + 2'A.z = 0

(A - 1) x + (4A - 2)y + (A + 3')z = 0
•.1 2x + (3l + l)y + 3(l- 1) z = 0
the smallest ofthese values. What happens
are consistent, and find the ratios ofx : y : z when Ahas (Delhi, 2002)
when Ahas the greater of these values.
1L Sol. The given equations will be consistent, if
A-1 3l+ 1 2A
A+ 3 = 0 (Operate R2 - R i
A - 1 4l - 2
2 3A + 1 3(A - 1)
l-1 3A+ 1 2A
0 l-3 3-A =0 [Operate C3 + C2
or if,
J!
2 3A+ 1 3(l- 1)
52 HIGHER ENGINEERING MATHEM

or if,
A-1
0
2
3A. + 1
A-3
3A+ 1
5A.+ 1
0
6A.-2
=0 . ..
:

or if, (A_ 3) IA.

1 SA.+ 1 = 0 or if, 2(A. - 3) [(A- 1)(3A. - 1) - (5A. + 1)) = 0
2 2(3A. + 1)

or if, 6A.(A - 3)2 = 0 or if, A.= 0 or 3.

ta> \Vhen A= 0, the equati ons become - x + Y =0
-x-2 y +3z =0
2x+y -3z= 0
Solving (ii) and (iii), we get
6
~ 3 = 6 :_ 3 = _ t
+ . Hence x = Y =z.
4
<b) \Vhen A= 3, equati ons becomes identical.

Prohl<.~ms 2 9
1. Invest igate for consist ency of the following equatio ns and if possib
le find the solutions
4x - 2y + 6z = 8, x + y - 3z = - 1, 15x - 3y + 9z = 21.
(Osmania, 19
2. For what values of k the equatio ns x + y + z = 1, 2x + y + 4z = k, 4x + y + lOz 2
= k have a solution
solve them completely in each case.
(V. T.U
3. Investi gate for what values of Aand µ the simulta neous equatio ns
X + y + Z = 6, X + 2y + 3z = 10, X + 2y + AZ = µ,
have Ci) no solution, (ii) a unique solution, (iii) an infinite numbe r of solutions.
(P. T. U., 2002 ; Samba lp
4. Test for consistency and solve
(i) 2.x - 3y + 7z = 5, 3x + y - 3z = 13, 2x + 19y- 47z = 32.
(Delhi, 2
(ii) x + 2y + z = 3, 2x + 3y + 2z = 5, 3x - 5y + 5z = 2, 3x + 9y - z !
= 4. (Madras, 2
(iii) 2x + 6y + 11 = 0, 6x + 20y - 6z + 3 = 0, 6y - 1& + 1 = 0.
5. Find the values of a and b for which the equations
x +ay+z =3,.x+ 2y + 2z = b, x+ 5y + 3z =9
are consistent. When will these equations have a unique solution ? (Mi
•
6. Show that if A. ;t - 5, the system of equati
3x - y + 4z ~ 3, % + 2y - 3z = - 2,
have a unique solution. If A.= - 5, show
each case.
7. Show that the equation.a
3%+4y + fiz=a, 4%+ 5y + & = b,
do not have a solution unless a + c = 21,: •
8. Prove that the equati ons~+ 3y + 2z = 12,
e = 74; and in that cue the equations are If, •
arbitra ry quantity. ,
9 .. Find the value, or A. for which the e
2% + 5y + (6 - A)= 0 are consistent and fmd the Vi.,..,. 9'1
of)...
10.. Show that there are three real values of or w c
:::::: :+az •O, cx+o :+~b rI=irr 9 • UII
 S
TERMINANTS. MATRICE
UNEAQ ALGEBRA : DE 2 o
2
l+ a + 6 +c = .
2
th at
=b n- 1, an =c l-m ,p ro ve - XJ + 2x i = ~t; J ca n
JJ .If bl =a m -n ,c m - 2.r 2 + %3 = A.r1, 2r 1 - 3x 2 + 2x3 = Ar2,
th e sy ste m of eq ua tio
ns 2.r1 case.
12. Sh ow th at ne ra l so lu tio n in ea ch
3. Ob ta in th e ge
ss a no n- tri vi al so lu tio n only if A= 1, A= - eq ua tio ns m ay po ss es s no n- tri
vi al solution:
posse fo llowi ng se t of
es of A for v.·hich th e
18. De ter m in e th e va lu + 4.r2 + Ar3 = o.
+ %2 - A.r3 = 0, 4.r1 - 2.r2 - 3%3 = 0, 2) J1
3% 1 l solution.
pe rm iss ib le va lu e of A, de ter m in e th e ge ne ra
Fo r ea ch
e sy ste m of eq ua tio ns 5z + 8w = 0 ; 5% - 7y +
2z - W = 0.
14. Solve co mp let ely th W : 0 ; 4.r +
2z + 3w = 0 ; + Z - + 13z + 3w = 0
-

y - 14z - 9w = 0; .r + 3y
y

% -
(i) X +
2 y

- 3w = 0; 2r + (J .N .T .U ., 20 02 SJ
-

+ 2z
y

= 0; 2.r + 3y
(ii ) 3:x + 4y -z - 6w

VERSE
D OF FINDING THE IN
2 12. PARTITION METHO in ve rs e of a m at rix An of or de r n is
e, if th e
d of fin di ng th e in ve rs ng (n + l) th row an d
According to th is m et ho n ea si ly be ob ta in ed by ad di
e of th e m at rix An + 1 ca
known, th en th e in ve rs
(n + l) th co lu m n
to An .

Le t
.. ..
.. s
w
...
ve ct or s (b ei ng tra ns po se s of column vector
e ro
n ve ct or s an d A 3', X 3' ar
where A2 , X2 ar e co lu m as su m
1
e th at A 1 is known .
be rs . W e al so
di na ry nu m
A3 , X3 ) an d a, x ar e or
•
..
Th en A A - =I n+ 1 i.e.
1

...(i)

gi ve s A1 X1 +A ~a '= ln ...(ii )

A1 X2 +A2-1" = 0 ...(ii i)
A 3'X1 + aX3' = 0 ...(iv )
A3 'X 2 +a x= 1
1 )- 1
1 ~ an d us in g th is , (iv ) gives x = (ex -A3' A 1 A 2
Fr om (ii ), X2 = -A i A
e given.
H en ce x an d th en X 2 ar
1
(In -A 2 X3 ')
Also from li) , X1 = A i 1
1 (ex -A ' A 1 A 2)- 1 = -A 3' A i x
s Xa ' = -A ' A i 3 1
an d us in g th is , (ii i) gi ve
3
1 d.
X i is de te rm in ed an d he nc e A - is compute
Th en
need to determ ine
aluation ofA - we only
1
ala tor me th od : Fo r ev
th e 'Esc
O la fl us is a1ao knIown· as
-

r iD ve ne matricea A j an d (a
1
-A3' Aj ' A 2,- .
1 1

.. .. .. .. .. .. U6UI/I th e pa
rtiti on method, fin d th e inve
rse
.
oi;
. [3 -J}-
1
3
5
 aw
--~

HIGHER ENGINEERING MATHE~J

1 1 1
Sol. Let A= _4 3 -11 (A
... ·a = A~'
3 5

so that
1 -[1
A -1 - 4
1]-l
3
=-[-4 3

Let A-1 =[~~

:: X21
... so that AA- 1 =I.
Xa' : X

a-A 3' Ai 1 A2= 3 + (3 5] [-! -n[ -n= - lO

X = (a- A 3'A-1 lA2)-1 =- 10

1

Also

Then -1]( 1) 1
1 - 10 = - 10 [- 11 21

Fina lly - 1] 1 [ - 34
1 - 10 - 11] [ - 11] 11 21
[-

-1~)-[-i:: ~·2]
1·4
Henc e A- 1 = -1·5
[ 1·1

Proh ll'm s 2 10
Find the inve rse of each of the following matrices usin g
the parti tion method.
1. (!
1
!3 :]4 (Nagpur; 1997) 2.(~ •
1 3
~ 1)
5

s.[! ! i :J-l
1 1 1 1
~[! -~ :i !]
1 -2 2 1
s.[! i -: ~;s] ·
2 - 3 ;.. 1 • (Na,pur, 200

2 13. (1) LINEAR TRANSFORMATIONS

Let (.x, y) be the co-ordinates of a point P referred to set of

its co-ordinates (.x', y') referred to OX, OY', obtained by rotat rectan
angle 8 are given by ing the
•
x' = x cos 8 + y sin 8, }
:l = - x sin 8 + y cos 8
~EAR ALGEBRA : DETERMINANTS. MATRICES 55
A more general transform ation than (i) is
x' =a1x + b11} ,..<ii'
y' =a~ +bzY

hich in matrix notation is ~ } [ =~ :~] ~]

Such transformations as (i) and (ii), are called linear transformations in two dimensions.
Similarly the relations of the type x' = l ix + m JY + n 1z
y' = l2X + m'JY + n2-Z ...(iii'
z' = laX + m3Y + naZ
ive a·linear transformation from (x, y, z) to (x', y', z') in three dimensional problems.
Y1 a1 b1 c1 ... k1 xi
In general, the relation Y = AX where Y = Y2 , A = a 2 b2 c2 •• • k2 , X = x 2 •••Civ)
Yn an bn Cn ••• kn Xn
ive linear transformation from n variables x1, x2, ..., Xn to the variables y 1, Y2, ... , Yn i.e. the
·ansformation of the vector X to the vector Y.
This transformation is called linear because the linear relations A(X + X2) = AX1 + AX2 and
1
(bX) = bAX, hold for this transformation.
If the transformation matrix A is singular, the transformation also is said to be singular
therwise non-singular. For a non-singular transformation Y = AX, we can also write the inverse
·ansformation X = A-1 Y. A non-singular transformation is also called a regular transformation.
Cor. If a transformation from (x1 , x2 , x3 ) to (y , y , y ) is given by Y = AX and another transform
1 2 3 ation
: (.y 1
,y ,y to (z ,z ,_z is given by Z =BY, then the transformation from (x ,x ,.x to (z ,z ,z is given
2
3
)

1 2 3
)

)
)

1 3
2

y
1 2 3

Z =BY= B(AX) = (BA) X.

(2) Orthogonal transformation. The linear transformation {iv), i.e. Y =AX, is said to be
rthogon al if it transform.~
2 2
Y1 +y2 + ... +Yn 2·into
xi2 +x22 + ... +Xn.
2
'he matrix of an orthogonal transformation is called an orthogo nal matrix.
X1

We have X'X = [x1x2·····.xn1 x x 2 = x 12 + x22 + ... + Xn2

Xn
nd similarly Y'Y = y 12 + y 22 + ... + Yn 2.
:. If Y = AX is an orthogonal transformation, then
X'X = Y'Y = (AX)' (AX) = X'A'AX which is possible only ifA'A = I.
ButA-1 A =1, therefore, A' =A-1 for an orthogonal transformation.
Hence a square matrix A is said to be orthogonal if AA' = A'A = I.
I
---1.--- --- --- --- --- --- --- --- --- -
l If A is orthogonal, A' and A- 1 are also orthogonal.
'Obs.
l ~ I \ is orthogonal,A' =A-1.
(Assam, 1999)
1
:. (A;);;;; (A-1)' = (A'f1 i.e. B' = B-1 where B = A'
~ B (i6,A') is orthogonal. AsA' =A-1,A-1 is also orthogonal.
Obt. I, If A ii orthogonal, then I A I = ± 1.
smce AA'= A'A = 1 :. 1A 1 1A' 1 = 11 1
But I A' I • I A I, :. I A I I A I = 11
I A 12 = 1 i.e. I A I = ± 1.
::,S 6~- ---- ---- ---- ---- ---- ---- H_I G_H E_R
.::. _E_ N_G _IN _E_ ER. ....; IN. .:.:G ~M AT HE
Exa mp le 2-33. Sho w that the transformat
~
ion
y 1 = 2x 1 + x2 + X3 , Y2 = x I + x2 + 2x3 , Y3
=XI - 2x3
,s regular. \\'rite down the inverse transformat
ion. (Kerala, 199!
Sol . The given tran sfor mat ion may be wri
tten as
Y= AX

l
1
whe re X = x11
X2 , Y ,= lYl]
X3
Y2 , A• = (21
Y3
2 11
l
1
0 J1
Now lA l= 1 1
2 =-1
1 -20
Thu s the mat rix A is non -sin gula r and hen
ce the tran sfor mat ion is regu lar..
:. The inve rse tran sfor mat ion is given by
X= A-1 Y

whe re A-
1
=[-; j =i1
Thu sx 1 =2y 1 -2y 2-y 3; x2= -4y 1+6 y2+
3y3 ; x3= Y1- Y2- Y3
is the inve rse tran sfor mat ion.
Exa mp le 2-34. Prove that the following mat .
rix is orth ogo nal:
-21 3 113 2131
213 213 113
[ 1/3 -2/ 3 213 <.A.M.I.E., 1

Sol . We hav e AA' = r-;

l j; ~j; i~;1 r-i~
x! ;~; 1/3 1
11a -2/ 3 213 l 21a 113
-2/
2/3
3 .

l
4/9 + 1/9 + 4/9 - 4/9 + 2/9 + 2/9
= - 4/9 + 2/9 + 2/9 - 2/9 - 2/9 + 4/91
4/9 + 4/9 + 1/9 2/9 - 4/9 + 2/9 =I.
- 2/9 - 2/9 + 4/9 2/9 - 4/9 + 2/9 1/9 + 4/9 + 4/9
Hen ce the mat rix is orth ogo nal.

Pro hlP n1s 2 11

1. Represent each of the transformations
xi = 311 + 212, 11 = z1 + 2z2 and X2 = -
Y1 + 4y2, Y2 = 3z1
by the use of matrices and find the composite.
transformation which expresses xi, x2 in te
z1, z2.
2. If ~ = x cos a ~ y sin a, 11 = x sin a + y cos a, writ (Ktrala,
e the mat rix A of tran sformation and prove
A-1 =A'. Hence write the inverse transformation.
3. A transformation from the variables x
1, .x2, x3 to y 1, y2, y3 is give n by Y = AX, and
transformation fromy1,Y2,Y3 toz1,z2,z3 is give
n by Z=B Y, where
A= 20 11 -201 , B = (11 21 1) 3. Obtain the transfor•mation from.x ,.x ,.x3 toz1
[ -1 2 1 1 3 5 1 2 ,z2,'
. . ~
. •
~INEAR ALGEBRA: DETERMINANTS, MATRICES 57
• ·
4. Find the inverse transfo rmatio n of Y1 = x1 + 2x2 + 5x3 ,· Y2 = 2x 1 + 4x2 + l lx3 ; Y3 = - X2 + 2%3.
5. Verify that the following matrix is orthog onal:

8]
l
l/3 2/3 2/3] cos 8 0
(i) 2/3 1/3 -2/3 (P. T. U., 2003) (ii) 0 1
si~ (A.M.J.E., 1997WJ
2/3 -2/3 1/3 [ sin 0 0 cos 0

6. ls the matrix 4
l
-3
2 -3
3
1
l]
orthogonal ?

7. Prove th) ~ ~ _!-~]is orthogonal when l = 2/7, m = 3/7, n = 6/7.

l-m n -l 0
8. If A and B are orthogo nal matrice s, prove that AB is also orthog onal.

2 14. (1) CHARACTERISTIC EQUATION

If A is any square matrix of order n, we can form the ~at~x A - M, where
I is the nth order
.1nit matrix . The determ inant of this matrix equate d to zero, i.e.
au -A a12 a1n
IA-A ll= a21 a22 - A a2n
=0
anl an2 a nn -'A

is called the characteristic equati on of A. On expan ding the determ

~quation takes the form inant, the charac teristi c

1~ (-ltA.n +k1A n-l +k2A.n -2 + ... +kn =0,

Nhere k's are expres sible in t~rms of the eleme nts aij. The roots of this equati
on are called the
:haracteristic roots or latent roots or eigen-values of the matrix A.
(2) Eigen vecto rs

x1 au a12 a1n
If X = x2 and A = a21 a22
a2n then the linear transf ormat ion Y =AX ...(i)

:arri~s t~e. colum n vec~or X into the colum n vector Y by means of the
squar e-mat rix A. In
practice, 1t 1s often requir ed to find such vector s which transf orm into thems
multiple of themse lves. elves or to a scalar
Let X be such a vector which transf orms into 'AX by means of the transf ormat
ion (i).
Then 'AX =AX or AX-A IX = O or [A- ).JJX = O
~
This matrix equati on repres ents n homog eneous linear equati ons ...(ii)
JJ
(au - A)x1 + a1ix2 + ... + a1nX11 = 0
a21X1 + (a22 - A)x2 + _
,
oil
... + a2nXn = 0
........................................................
...(iii)

which will have a non-tr ivial solutio n only if the ffi

I A- Al I = 0. coe icient matri x is singu lar, i.e. if
 tr l q'►ci---

HIGHER ENGINEERING

This is called the characteristic equation of the transformation and is s

characteristic equation of the matrix A. It has n roots and corresponding to ea
equation (ii) [or (iii)] will have a non-zero solution
X = [x1, x2, ...... , xnl', which is known as the eigen vector or latent vec

Obs. 1. Corresponding to n distinct eigen values, we get n independent eigen vectors, But
or more eigen values are equal, it may or may not be possible to get linearly independent ei
corresponding to the repeated roots. •-
Obs. 2. If~; is a solution for a eigen value )..i then it follows from (ii) that cX; is also a solu
c is an arbitrary constant. Thus the eigen vector corresponding to a eigen value is not unique
any one of the vectors cX;.

Example 2-35. Find the eigen values and eigen vectors of the matrix [ ~
Sol. The characteristic equation is [A - Al] = 0
5-11. 4 2
2 _ "- = 0 or A - 711. + 6 = 0
l.e.
1
or (11. - 6) (11. - 1) = 0 A= 6, 1.
Thus the eigenvalues are 6 and 1.
If x, y be the components of an eigenvector corresponding to the eigenvalue A, the
IA - All X = [ · /
5
2 ~ i] [;] = 0
Corresponding to A= 6, we have [- ~ -!] [;] =0
which gives only one independent equation -x + 4y =.0
: = f giving the eigenvector (4, 1).
I

Corresponding to :\. = 1, we have [ f f] [;] =0

which gives only one independent equation x + y = 0.

~=~giving the eigenvector (1, - 1).

Example 2-36. Find the eigen v.alues and eigen vectors of the matrix
. • 3 1
5
1
[1
. .
1-A 1 3]
Sol. The characteris • 1 5 .;.1 1 i.e. 'J...3 - 7A2 +
8 1 1-A
Since A~ -2 satisfies i -w te tliij iua
=
(A + 2) (A2 - 9A t 18) 0 or (A+ 2) (A- 3) (A - 6) = 0.
Thus the eigen values of A are A= - 2, 3, 6.
If x, y, z be the components of an eigen vector corresponding to the eipn value A, we

LA -AlJX =[
1

3
5-
1 1-A z
1~
=.0 t • ~] [;]
Putting A=-2, we have k+y+Sz =0,· s+7y +••0, 'k+y +Is •O. •
 59
a.I? ALGEBRA : DETERMINANTS. MA
TRICES
fir ,t two
ng the same, we have from the
The first and thi rd equations bei
.....!_ =l =~ or ...!.... =l =1
z
-20 0 20 -1 0
, 0, 1). Als o eve ry non -ze ro mu JtipJe of thi s vector is an eigen
Hence the eigen vector is (-1
tor corresponding to A= -2. -zero
tor s cor res pon din g to A= 3 and A= 6 are the arb itra ry non
: Similarly the eigen vec
(1, 2, 1) which are obtained from
(i).
ltiples of the vectors (1, -1, l) and
(-1, 0, 1), (1, -1, 1), (1, 2, 1).
ee eigenvectors may be taken as
I

ij Hence the thr

. PROPERTIES OF EIGEN VALUES

.
are ma tri x A an d its tra nsp ose A' have the sam e eigen values
'i I. An y squ
We hav e (A - Al)' =A ' -A l' =A ' -A l
I(A - U) ' I = IA' - Al I f•. • I B' I = I B I
IA -A ll= IA' -A ll
- )JI = 0
IA - U I = 0 is and only if IA'
valve of A'.
1 ~ i;-a n eig en val ue
ofA if and only if it is an eigen .
tria ngu lar ma trix are jus t the diagonal elements of the ma trix
II. The eigen values of a
a11 a12
aln
0 a22 a2n n.
be a tria ng ula r ma trix of ord er
Le t A=

0 0 ann
A).....(ann - A).
Th en IA - )JI = (au - A) (a22 -
au , a22, ... , ann·
:. Roots of IA - Ml = 0 are A.=
au , a22 , ... ann·
the diagonal ele me nts of A i.e.
Hence the eig env alu es ofA are
ments of the ma trix .
gonal ma trix are jus t the diagonal ele
Cor. The eigen. ualues of a dia
ty.
otent ma trix are either zero or uni
III. The eigen values ofan idemp sts
trix so tha t A 2
= A. If A be an eigen val ue ofA, the n the re exi
Le t A be an idempotent ma ... ( 1)
in-zero vec tor X suc h tha t AX = ').X

i.e. A 2X = :l(AX)
2
[·.·A =A &A X= 'AX
i.e. AX =A (U )
...(2)
AX=}..2X
2 or (A2 - A) X = o
From ( 1) and (2) , we get A X = ').X
2
}.. - A= 0 whence A= 0
or 1.
,wel
Hence the result.
pri nci pa l dia gon al.
ofthe eigen val ues ofa ma trix is the sum ofthe elements ofthe
IV. The sum ensio
for a ma trix of order 3, but the method will be capable of easy ext n
{~ propertyordwil l be proved
atnces of any er.]
 ~-- ~~-~..,,
- I am • a + • -w:&,>+¼'K

f
I
60

Consider the square matrix

HIGHER ENGINEERING

au a12
a13]
A= a21 a22 a23
[
a31 a32 a33

au -A a12 a13

I
!
I
so that IA- M I= a21
a31
a22 -
a32
A a23
a33-A
(On expanding)

= - A.3 + A2(a 11 + a22 + a33)- .....•

If Ai, "-2, l 3 be the eigenv alues ofA, then I A - M I = (-1)3 (A- A1)(A- "'2)(A- A3)
= - A3 + A2(A1 + ¼ + A3) - ..... .

Equati ng the right hand sides of(ii) and (iii) and comparing coefficients of A2, we
A1 + ¼ + A3 =a 11 + a 22 + a 33. Hence the result.
V. The produc t of the eigen values of a matrix A is equal to its determinant.
Putting A = 0 in (iii), we get the result.
VI. If A is an eigen value of a matrix A, then 1IA is the (;igen value ofA-1.
If X be the eigen vector corresponding to A, then AX= 'J..X
Prcmul tiplyin g both sides by A- 1, we getA-1AX =A- 1'J..X
i.e. IX= AA- 1x or X =A(A-1X) i.e. A- 1 X =(1/A) X
This being of the same form as (i), shows that 1/A is an eigen value of the inverse
A-1.
VII. If A is an eigen value of an orthogonal matrix, then 1/A. is also its eigen va/ue.
We know that if).. is an eigenv alue ofa matrix A, then 1/A- is an eigenv alue ofA-1.
VJ. Since A is an orthogonal matrix, A-1 is same as its transpo se A'.
. .
:. 1/A. is an eigen value ofA'.
But.th e matrice s A andA' have the same eigenv alues, since the determ inants1
and I A' - 'JJ I are the same.
Hence 1/A- is also an eigen value ofA.
VIII. If Ai,~' ......, An are the eigen values of a matrix A, then Am has the eigen
l. 1m, AIJ,m, ...... , Anm (m being a positive integer).
Le_tA; be the eigenv alue ofA and Xi the corresponding eigenv ector. Then
AX;=A;X;.
We have 2
A X;. =A(AX;,) =A(A-;X;.) =~(AX ,)= 'A,fl;X;.) = ),2 X;
Simila rly,A3X; = 'A./Xt In genera l,Amxi = 'A;mX;. which is of the same form as (i!-
Hence 'A;m is. an eigenv alue ofAm.
The corresponding eigen vector is the same X;.
 @ffi I

't
'EAR ALGEBRA : DETERMINANTS. MATRICES
61

16. CAVLEY-HAMILTON THEOREM*

Euery square matrix satisfies its own characteristic equation ; i.e. if the characteristi c
uation for the nth order square matrix A is
I A-iJ. I =(-ItAn+k1 An-l+ ...... +kn=O ... (i)

en (-ItAn + k1An - l + ...... +kn= 0.

Let the adjoint of the matrix A - ')J be P. Clearly the elements ofP will be polynomials of the
- l)th degree in A, for the cofactors of the elements in I A - iJ. I will be such polynomials.
:. P can be split up into a number of matric~s, containing terms with the same powers of
such that
... (ii)

1ere P 1, P2, ...... , Pn are all the square matrices of order n whose elements are functions of the
?ments of A.
Since the product of a matrix by its adjoint= determinant of the matrix x unit matrix.
~
[A - iJJP = I A - iJ. IXI
:. by (i) and (ii), [A-U][P1A n-l +P2An- 2 + ...... +Pn-1A+Pn l

= [(-ltAn + k1An - I+ ..... + _kn- 1A + kn][.

Equating the coefficients of various powers of A, we get

-P1 = (-ItI
AP1 -P2=ki/,
AP2 - Pa = k2I,

AP11 - 1- Pn = kn - if,
APn =kn[.
1
n Now pre-multiplying the equations by A", An- , ...... , A, I respectively and adding, we get
(-ItAn+ k1An- l + ...... +kn -1A + knl = 0, ... (iii)
the terms on the left cancel in pairs. This proves the theorem.
Cor. Another method of finding the inverse.
Multiplying (iii) by A-1, we get
(-ltAn-I+k1A n-2+ ..... +kn-1l+k,t4- I=O .

1ence A- 1 =-J II
2
((-l)'1An-I+k1 An- + ...... +kn-1n,

This result gives the inverse ofA in terms of n - 1 powers ofA and is considered as a practical method
~ the computation of the inverse of the large matrices. As a by-product of the computation, the
aracteristic equation and the determinant of the matrix are also obtained.

Example 2-37. Verify Cayley-Hamilton theorem for the matrix A=[~ ;J and find its

erse. Also express A

5
- 4A 4 - 1A3 + IIA 2 -A - IOI as a linear polynomal in A.

~ See fo~tnote on .P• 17. William Rowan Hamilton (1805-1865) an Irish mathematicia n who is known
his work m dynamics.
v62~__ __ __ __ __ __ __ __
- __ __ __H_IG_H_E_R_E_N...;..G__
IN:.:.:EE::.:.:R~ING M
Sol. The characteristic equation of A is ~
1 1= O
l - A or i 2 - 4A - 5 = 0
12 3- i
By Cay ley- Ham ilto n theo rem , A mu st sati
sfy its cha rac teri stic equ atio n (i), that
2 80

=[~ :m !)-4[~
A -4A -51 =0
2
Now A - 4A - 51
!)- 5[~

=[: !~)-[: !~)-[~ ~)=[~

Thi s ver ifie s the theo rem .
Mu ltip lyin g (ii) by A- 1, we get A- 41 -
SA- 1 = 0
or A- ,=~ (A -41 )=~ [~ !)-4[~ ~]=H-~ -~)
Now div idin g the polynomial i 5 - 4">..4 -
7">..3 + 11">..2 - "A.- 101 by the polynomial
we obt ain ;.2_

"A.5 - 4">..4 - 7">..3 - "A.-101 = ("A.2 - 4">..

- 5)(">..3 - 2">.. + 3) +"A.+ 5
=A .+5
Hen ce A 5 - 4A4 - 7A3 + 11A2 -A -10 1 =A
+ 5 whi ch is a line ar poly nom ial in A.
Exa mp le 2·38. Fin d the characteristic equ
ation of the mat rix A of Ex. 2-20, and he
its inverse.
1- A. 1 3
Sol . The cha ract eris tic equ atio n is 1 3-A . -3 = 0 i.e., "A.3 - 20).. + 8 =0.
-2 -4 -4- A.
By Cay ley- Ham ilto n the ore ~, A 3

- 20A +B l=~ ' whe nce A- = ~I - !A 1

2

= 5 [10 01 00) - S
1 [-10
4 -22
8 -1261 = [ 3 3/2]
2
0
-5/ 4 -1/ 41 -3/4
0 1 2 2 22 -1/ 4 -1/ 4 -1/4
Pro hh• n1s 2 12

L Find the 1uan and product of the eip n val 2 3

uu of - 2 1
• ( 1 0
2. Find the product of the eigen valuu of 7
-6 - 21 22)
( 6 2 -1
a. PiDd the eiie n valuea and eigen vecton of the matrices :
(a) i1 2] <Anclhra, 199 • 8)
[ 4 Cb) [1 4] (Bhopat•
3 2
4. Piad the characteristic roota and the cha
ract eristic vec ton of the matrices :
(4) 8
-6 -6
7 2]
-4 T. U., 2004 ·; Osma.nia., 2003 S; p,~
[ 2 -4 3
( V .

. [2
(b) ~
 63
tlNEAR ALGEBRA : DETERMINANTS. MATRICES

-2 2 S>
(c) 2 1 -31
-6 (Bhopal, 2002 S; Madras, 2000S,· V. T. U., 2000
[ -1 -2 0

-2
3 <Delhi, 2002 ,· Madras, 2002 ; P. T.U., 1999>
(d) H
-1 -1]
ix
5. If~ be an eigen value of a non-singular matr
ix A, show that / A //}.. is an eigen value of the matr )
(U.P. T. U., 2001
adJ A.

6. Two eigen valu es of the matr ix A~ ~ I]

= [ are = 1 each. Find the eigen values ofA- .
1

1 2 2
of a matr ix A, then A has the laten t roots
2
7. ·show that if >..1, >..2, ... , A,i are the laten t roots
;_f, ~' ... ~- eigen
the eigen vectors corresponding to two unequal 1996
8. For a symmetrical squa re matrix, show that (Kanpur, )
values are orthogonal.
rse of
9. Using Cayley-Hamilton theorem, find the inve

(i)[~ ~] ~Delhi,_1997), _ (ii)[~ J -~] .

(Osmania, 2000 S)

(iii) r~ ~ _:
/ (Bhopal, 2002 S>
-4 -4
1
H
2

10. Find the characteristic equation of the matr

ixA =[t ~ Show that the equation is satisfied by

:~ (Madra.o;, 2003)
A and hence obtain the inverse of the given matr
ix.
ix A an[d ;md ~ts i~~e]rse.
11. Ve_rify[ ~ayl ~;H a~]l ton theorem for the matr
(ii) -6 -1 2 (Coimbatore, 2001 )
Ci) -1 -1 (J.N.T.U., 2002)
2
6 2 -1
:j 1 2 -1
8
= [~ -~}
: 12. Using Cayley-Hamilton theorem,_flnd A , ifA (Anna, 2003 )

13. IfA=[ f ! f ], eval uate A-1, A-

2
andA - 3.
-1 -4 -3
~
1 .
11
14. Find the characteristic equation of the matrix
A =[ 1 glj and, hence,. find the matrix represented
1
(U.P.T.U., 2003 )
lb by AB - 5A1 + 7A6 -3A 5+A • - 5A3 + BA2 -2A +/.

17. (1) REDUCTION TO DIAGONAL FORM

foun d
pend ent eigen vectors, then a matr ix P can be
°' If a square matr ix A of order n has n linearly inde
1
tch that P- AP is a diagonal matrix. n
resu lt will be proved for a squa re matr ix of orde r 3 but the method will be capable of easy exte nsio
(This
, matrices of any order.J
,1; Let A be a squa re mat rix of orde
r 3. Let A1, A2, A.3 be its eigen valu es and

~ X1 = [;:] . X2 = [;:] and X3 = [;:]be the corresponding eigen vectors.

 HIGHER ENGINEERING

Den otin g the squ are mat rix (X X:iX 1=

1 3 r;: ;: ;:] by P, we hav e
lz 1 z2 z3
AP = AlX 1X2 X3} = [AX , AX2 , AX3] = ()..1X1,
1 ½X°2, A.~3 ]

J~l~: ~: ~::tr;: ;: ;:]x \~

1
~. ~1= PD ,wh ereD isth edia go
A.1z1 jl222 A.3Z~ lz1 z2 z3 lO O A.3
:. p-lA P = r 1 PD =D, which proves the theorem.

Obs . 1. The matr ix P whic h diagonalises A is calle

d the mod al mat rix ofA and the resulting
mat rix D is know n as the spec tral mat rix of
A.
2. The diag onal matr ix has the eigen va.lues
of A as its diagonal elements.
3. The mat rix P, whic h diagonalise A, cons
itutes the eigen vectors of A.

(2) Sim ilar ity of mat rice s. A square mat rix A/1. of
A of orde r n if
order n is called sim ilar to a square

A = P- 1AP for some non-singular· n x n matrix P.

Thi s tran sfor mat ion of a mat rix A by a non -sin /1.

tran sfo rma tion . gula r mat rix P to A is called a sim

Obs . If the matr ix A is simi lar to the matr ix A,

then A has the sam e eige nval ues as A.
If x is an eigen vector of A, then y = P- 1x is an
eigen vector of Acorresponding to the same e •
(3) Pow ers of a mat rix. Diagon
alisation of a mat rix is quite useful for obtainin
a mat rix. gp
Let A be the squ are mat rix. The n a non -sin gula
r mat rix P can be found such that
D =P-1 AP
D2 = (P-l AP) (P-l AP) =P- 1A2 P

Sim ilar ly
To obta in An, premultiply (i) by, P and post-multipl
y by P-1.
1
The n PD" P- =~ ;;;;z;·;·~~lfil~ • •es An.
4bus,

Wor king pro cedu re :

1. Find the eige nval ues of the square matrixA.
2. Find the corresponding eigen vectors and write
the modal matrix P.
3. Find the diagonal matrix D from D =:: Y 1 DP
4. Obta in A" from A" == PDP- 1.
Exa mp le 2-39. Fin d a mat rix P whi ch tranaforms
the mat ru: A of Es. 2·Jp ~ aaGfflt!
Hen ce calcuJ.ate A 4•
 65
LINEAR ALGEBRA : DETERMINANTS, MATRICES

-2, 3, 6 and the eigenvectors are C-1, 0,

Sol. The eige nva lues of A (found in Ex. 2·35) are
1 >, (1, -1, 1), (1, 2, 1). Wri ting thes e eigen
vectors as the thre e columns the requ ired
'
transformation mat rix (modal matrix) is

_P =n -! c1
i]
-1 1 11 a1 b1
To find P-1, I P I = 0 -1 2 = a2 b2 c2 (say)
1 1 1 a3 b3 c3

A1 =-3, B1 =2, C1 = l,A2 = 0, B 2 =-2, C2 =2,A3

=3, B 3 =2, C3 =1
Also I P I =a1A1 + b1B1 + c1C1 =6

... rt= I ; I [;:

C1
!:
C2
!:]
Ca
lr-3
=6 ~
-2
0

t
Thu s D= P- AP =n
1
g ~]
4

h D
4
=[(~)O ; ~]=[106 8ol g]
o 64 0 1296

=~ [~~ -I ~][1 g 1r-: _g2 21~1j

6
4 4 1 8~-
- He~ceA =PD P- 1296 1
1 1 1 0 0
ll

j ~f!:~
485
_f ~Jf ;~
_235]
-2~ 2~1 1051 485
In =r-~ 235 485 251
1 1 1 216 512 216

ONICAL FORM
18. REDUCTION OF QUADRATIC FORM TO CAN
A hom ogen eous expr essi on of the second degr ee in any num ber of variables is called a
f
uadratic form.

• For instance, ifA=[; ~ !],x=m and X' [% = y z], then

2 2 ...(i) -·
X'AX= ax + by + cz + 2/yz + 2.gzx + 2hxy
2

hich is a quadratic form.

and
Let A1, ½, Aa be the eige nva lues of the matrix A

X1 =[;:J,X2=[;:}Xa=[m
(i.e. each elem ent is divided by squ are
i its corresponding eigen vectors in the normalized form eigen vector).
ot of sum of the squa res of all the three elemen ts in the
0] .
i~
[Al • 0 [xl X2 %3]
1 Ys
The n by§ 2-17(1), P- AP= 0 ½ 0 whe reP = Y1 Y2

. 0 0 As Z1 Z2 Z3
66
HIGHER ENGINEERING
Hence the quadratic form (i) is reduce MATHE~
d to a can on ica l for m (or sum of
Pri nci pal axe • form.) squ
art s ti
•
A.1%2 + A?Y2 + A.a%2
and P is the ma trix of tra nsf orm ati
on which is an orthogonal ma trix .
Note. Conl(nlent (or orthogonal) tran
called congruent matrices and the abo sfo rm atio n. ~e ~agonal matrix
ve method of reduction is called con D and the mat.
transformation gru ent (or ort ~~
Index. The ~umber of a positive terms
in canonical form of a quadratic form
• aa,
the form.
,
is known as the
1
J
Sig nat ure of the quadratic form is .
the difference of positive term s and neg
furm. ative terms in ·ts
1 can
·
Ex am pl~ 2-40. Reduce the quadratic
form 3~ + 5y2 + 3z2 - 2yz + 2zx - 2xy
form. Also specify the ma trix of transf to the ca
ormation. (Madras, 200 2 ; Cochin~0
2
_ So l. The ma trix of the given quadratic
form is A
1 - -1 a
Jl-~ -! -!]
'
3-A . -1 1
Its cha rac teri stic equation is I A - Al
I = 0 i.e. -1 5- A -1
· 1 -1 3-A .
which gives A= 2, 3, 6 as its eigen
values. Hence the given qua dra tic
canonical form form reduces to
~

i.e. 2x2 + 3y2 + 6z2.

To fin d the ma trix of transformation
Fro m ( A - Al ] X = 0, we obtain the equ
ations
(3 - A) +z =0 ; -
+ (5 - A.)y - z = 0 ; + (3 - A.)z = 0.
x - y

Now corresponding to A= 2, we get x

x - y

-y +z = 0, -x + 3y -z = 0, and x -y
whence + z = 0,
x _ ,l •_ _!__'
1 - 0 - -1
:. The eigenvector is (1, 0, -1) and its
normalised form i~ (1/✓2, 0, -· 11✓ 2)."
Similarly corresponding to A= 3, th~
t 1/✓3, 1/~ 3. 1/✓3). eigen vector is (1, 1, 1) and its nor
malised fo
Finally, corresponding to. A= 6, the
(1/✓6, - 21✓6, 11✓6).
eigen vector is (1, - 2, i) and its nor
. malised fonn

•
Hence the ma trix of transforma (. 11✓ 2 1/✓ 3 11✓ 61
tion is P = O 11✓ 3 - 21✓ 6
· - 11✓2 11✓3 11✓ 6
Not.e. lDCIG of the quadratic form =
. .3. Its signature is also 3.
. .
:, ~1 r d-d URE OF A QU AD RA TIC FORM
A real. quadratic form 'JCAX in n variab
les is said to be
(i) po alt ive def ini te if all the eig
en values ofA > 0.
(ii) ae ,at lve def ini te if alt the eig ,
en values ofA·< 0. •'· . •
~) po sit ive eemldeftnlte ifall the eig • · .
en values ofA ;i: 0 an d at leaBt one e~e
(iu) ne ptl ve aemldetinit,e if all n :,:
the eigen values ofA ~ 0 and at l~ast o
(u) lndeftnlte if som e of the e;,, ~n;_..
en ~alue, ofA are + ve ancl oth ers ·- ve.