Determinant-Characteristic Roots ] t 173

Solation. Let a eF be eigen value of T. Then 3 v(£0) ¢V

such that

(1) T(V)=Av
Let Ti-ST S-
then T S=ST
So (T1 S) (v)=(ST) (V)
or Ti (S. (v))==S(T(M))
=S(Av) [By (1).
Hence Ti (S(v))=A S(V).
Putting S(v)=x1 we get
T (x)=Ax.
Since v0, So x=S(v) #0 since S in Non Singular.
Thus 3 x=S(v)*0, such that
Ti (*)=A x
Hence is an eigen value of Ti
That, is an eigen value of STS°l and x=S (v) is an eigen
vector of STS-1

V26). Show that the non-zero characteristic vectors of a

linear transformation T corresponding to distinct characteristic
roots are linearly independent.
Solution. Let vi, Vas., Vn be non-zero characteristic vectors
of the linear map T on V corresponding to distinct characteristic
roots A Ass " An-
We will prove {'1, Vas . , "} to be linearly independent by
induction.
Ifn1, then {v,} is linearly independent since v,z£0.

Hence the Theorem is true for n=1.

Assume the theorem to be true for (n-1) non-zero charac
teristic vetors v, Vay .., Yn-1 with distinct eigon values As dg .
n-1. et

(1) a ' "2 V» l.. -1 Va-1t" V=0where a,eF

Tben T (o V, t"a Va t . 4-1 V-1t4, V»)=T(0)=-0
 Determna11-Charabteristie Roois
74 1
+a-1 T(Vn-1)+a
T(vn)=0.
Or a T(v,)+a T(v)+... Hence
T(v)=A4 vi for i=1, 2,,
n.
But
"ita2 Aa V a t + - 1 An-1 Va-1t+ An Vs=0
(2) A we get
and substracting from (2)
Multiplying (1) by An
V , + . + a A n - 1 - A » ) Va-1=0
A-An) V1+aa A»-A,)
(3) a induction hypothesis,
Va .Va-1} is L. I. by
Since V
so (3) gives
Aa-An)=0,..g-1
An-1-A»)=0.
A - A ) = 0 , aa
n-1.
Since AAn for i=1, 2, ..,

..=a,-1=0
Hence
have a, Vn =0
Therefore from (1) we

or a=0 for v,z£0.

a1 Vt+ag Vat...+a,V,=0
Thus
1=ag*. =",=0
is linearly independent.
Therefore {Vi, Va, -. Va}
a vector
over a field F of dimension n.
(2/7). Let V be space
V, associated with the
Let and T" be two linear maps on

with respeet to bases {*1, *2, ., *) and

same matrix A=[ay]
) respectively. Prove that Tland'T" have the same
*a,.,
cigen values.
Solution: Let St V>V be defined by

(1) S(x)=* forj=1, 2, ..., n.

It can be easily proved that S is a bijection and hence non

singular and invertible.

T'=STS-1
Now we prove that
Now for j=1, 2, ..n, we get
(STS-") (x'))=ST (S(*'))
Since by (1), S-1 (x') = xs]
ST (x)
S(T(*))
Hence (STS-1) (x'i)=S (a x (Since Matrix of Trelative
to basis Nj} is [as})

S (a) (since S is linear)

i
Determinant-Characteristic Roots j 173

(from (1))

-T' (*') {Since matrix of T' relative

to basis fx'} is [al}
Thus (STS-) (*i)=T' (*') for j=1, 2, . n.

So (STS-1) ()=T' (v) v eV. (Prove)

Hence STS-1=T'
Hence by Ex. (2-5), T and T'=STS-1 have the same eigen
values.
We can thus define the eigen value or characteristic root of
an n-square matrix [a,,] as follows:

Definition (22). Let A=[4,] be an n-square matrix over a

field F.
aEF is called an eigen value of the matrix [a] if A is an
n-dimensional
eigen valuc of any linear transformation T on an

vector space V, asscciated with

the matrix A.

If we take V=F" and consider the linear map T; FF"

given by.
T(X)=[a] where X = 1 F
(1)

We have proved before that matrix of T relative to the usual

basis of F is [a]
Hence if AeF be an eigen value of 7, then A is also an eigen

value of the matrix [aj] associated with T (By Defin. (2:2)

Now AEF is an eigen value of T, if 3 X*0
such that T(X)=AX
or
2
 Deter nina t-Characteristic Ro»ts ] 177
Since
#0, so 3 x 0 where i=l or 2 or 3.... or n,

such that A x=A; *;

or A=s [ Since xz40]
Hence A, Ap...An (the diagonal elements) are the eigen values.
Sccond part obviously follows since a matrix A and the
diagonal matrix similar to A are matrices of the same linear
transformation T on V relative to different bases and so they
have the same eigen values.
(29If A and B be n-square matrices over F, then show
that 4B and BA have the same cigen values.
Solution. If AB has an eigen value 0, then AB is Singular
by Ex (2.4) (ii). So either A or B is singular which implies that
BA is singular.
Therefore BA also must have an eigen value 0.
Now let AF£0 be an eigen value of AB, ihen 3a column
matrix X£0, such that

(1) (AB) X=AX

or AY=AX where BX=Y ( a columu matrix )
Now (BA) Y=B (AY)
=B (AX)==A(BX) =AY
Now Y40.
For Y=0 BX=0
(AB) X=0
AX=0 By (1)1
X = 0 since a <0.

Thus we get (BA) Y=AY where Y0

Hence is an cigen valuc of BA also.
Similarly it can be proved that any non-zero cigen value of

BA is also an eigen valuc of AB.

Thus A and BA have thc same cigen values.

12
 Determinant-Characteristic Roots
178 1
Preblem-Set
T on V with
(2.1) If v is an eigen vector of a lirear map
¬F is also an
eigen value à, then prove that av where (40)
a

vector v
eigen vector with eigen value A. Also prove any eigen
can not correspond to two diferent eigen values A and Ap.

(2.2) Give an example of a square matrix of order 3 with

entries in Z whose eigen values are 2, 1, 3

Hint. Apply Ex (2.8) required Matrix= 2 0 0

0 1 0
0 0 3
IfA is eigen value of invertible linear trans-
(23)/ an an

formation 7, then prove that a-1 is an eigen value of T-1

Hint: Tis Non-singular. Hence A#0 so - l exists.
Now 3 a vector v-£0 such that
Tv)=Av
Hence T- (T())=T ' (Ar)
This implies v=AT-(v).
or T (v)==a-l.
Theorem (2.1). Let A =[ayl of order nx n. aEF is a charac
teristic root of A iff
4-A l=0
Proef: If A is a characteristic root of A, then 3 a non-zero
column matrix X, such that
AY=AX [By defin. (2.2) (a)]
or AX=AI,X
(1) or (4- 1) X=0
which is a system of n-linear homogeneous equation in n'unknowns
in matrix form.
The system (1) has a Non-zero solution X iff
A-A 1, =0 [By Cor. (6:3) Ch 3]
Hence is a characteristic root of A iff
TA-A , I=0
Deter aina 1t-Charate istic R»o1s] ] 179

Deinitipn (2.3) Characteristic polynomial of a matrix.

1f 4-[a,,] be an n-square matrix, then determinant
which is a polynomial in
4-A
of degrec n, is called the Characteristic
poyno mial of matrix A and is denoted by $(a).
Thus the Characteristic Polynomial of A is givcn by
6(A)|4-aI,=|d1-A 1
d21 d2gA. 2n
****************

1 2 . .nn-a
The cquation d(A)=0 is called the Characteristic equation
of A.
Note. (2.1). The characteristic roots of the matrix A arc
the zeros of its characteristic
polynomial $(a) which is of degree
n. Thus an n-square matrix A may have at most n distinct charac-
teristic roots.

Examplc-Set
(2.10) Show that the eigen values of the matrix.
01 0 01
00 1 0 are +l and i
0 00 1
0 0 0-
Solution. The characteristic polynomial of the natrix is
given by
By R,R
FAR]
0 0

A-1(A 1) (A1) (A-0 (at)

lenc 1he eigen values aIe

C a h of the oliowg mauices lind their egen
aiurs and a dis ieit c bpaues
i)A ()
 Determinant-Characterlstie_Roots
180
polynomial of A is
Solution (0) The characteristic
#A)-14-A
By CC
+CCC
-SA
-6 4-A -2-A 4-Al +C3

-2+A0! -(2+AP!0 3[RR,-R,]

0 1 4-A

=(2+A (4-3)
Hence eigen values of A are -2,-2, 4.
Va when a=-2.
We find the eigen space
vector of A with eigen value -2.
Let X=[x] be the eigen

(A-A n) X=0 Equ (1) Theor. (2,1)]1

Then

Putting A=-2 and X=| we get

E-E! -1 11
The coefficient matrix r!-!
0 0J
The last matrix is in echclon form with only one non-zero
row. Hence Rank of coeff. Matrix =p=1.
Hence the system will have n-P=3-l=2 L, I, solutions.

The system in reduced form becomes

x-y+2=0.
Choosing z, y to be arbitrary variables and setting
we get the following two L. I. solution
y
Determinant-Churacteristic Roots} 181
Thus the basis of the eigen space of A=-2 is
(0, 1, 1), (1,0,-1)}
Similarly if (x\ be the eigen vector with eigen value A=4,

then we get

0
-
The coefficient Matrix is equivalent to

(Prove)
0 0 0
Hence Rank of coef. Matrix=2, so the system has 3-2=11
independent non-zero solution.
The system in reduced form is
X-z/2=0
J-z/2=0
Choosing z to be the arbitrary variable of taking z=2
we get x=l, y=l, z=2.
Hence the basis of eigen space of A=4 is
(1, 1, 2)}
Thus the matrix A has three. L. I. eigen

(0, 1, 1). (1,0,-1), (1, 1, 2)

(i) Solution is left to the readers.
Hint. The eigen values of B are: -2, -2,4.
The eigen space of A=-2 is generatei by (1, 1, 0)
The eigen space of a =4 is generated by (0, 1, 1)

Thus B has two L. I. eigen vectors

(1, 1,0) and (0, 1, 1)

(2.12). Prove that similar Matriees have the same charac

tetistic polynomial and bence the same eigen values
 Determinant-Characteristic Roo1s
182 1
similar Matrices. Then 3 a non-
Solution : Let A and B be

Singular matrix S such that

defin (4.3) Ch. 4]
B=SAS-1 [By
Polynomial of B is
Now the characteristic

B-Al=|SAS-1-S(AI) S-||
= |S(A-AI) S-|
= |S| |4-A/| |S-|.
=|4-u| |SS-|.
=|4-A| |/1.
=|4-l|.
= Characteristic Polynomial of A.
characteristic polynomial and
Thus B and A have the
same

same eigen values.

so they havethe
V>V be an Endomorphism of a finite
(2:1. Let f :
dimensional vector space V over a field F.
values of f with
Let As An eF be the distinict eigen
A2,....
vectors *1, X2,...Xn ¬V. Let W be the subs-
corresponding eigen
that
pace of V generated
by {x1, *2,.,*n Prove
Hence f: W->W is an Endomorphism,
(a) Sx) eW Vx¬W.
(6) dim W >n.
Solution : Let x e W, then
x=at+ag*at..+ *n where a ¬F [Since W is generated

by {x]
Hence S)=a f*)+a, S*)+...+a, S*»)

the eigen values of f with corresponding eigen

[Since {A;} are
vectors{x}].
Thus flx) is a linear combination of elements of {X1, X2..Xn}
Hence x ) EW VxeW,
is f: W>W, which is therefore
Hence restriction of f to W
also an Endomorphism.
Determinant-Characteristic Roots 185

or

(5) or A A X= a [Since A=4]

From (3) and (5) we get
Aa=a
Hence A=A since aF0
Therefore characteristic root a is real

Problem-Set
Prove that the characteristic roots of a triangular
(2.4)
Matrix are the elements in the principal diagonal.
in which all
Hint. A triangular Matrix is a square Matrix
the elements either below or above the principal diagonal
are zeros,

a13 an IS a triangular Matrix

A= 11 d12
0 ag2 agiseoo *..aan
0 a33 .3n

0 0.
elergénts below thbe principal diagonal
are zeros.
All
A* have the
(2.3/ Prove that the matrix A and its transpose
same eigen values

Hint. (4-al)=4-Al.
Now 4-Al|=|(4-al)l.=14-l|
Let: R->R* be a linear map and let
(2.6).
A= [41 a1] be the matrix off with respect to a basis of R?.

be the eigen values of f. Define

Let A and Aa
Trace of f, noted T, (S, by 7, (f)=4ut42a
Prove that (a) T, (S)=A1tA2» bence T, (S ) is independeat
of the particular bases,
(b) Det A=1 A2
Hint. By Ex (2.14). we have AtA-[rac: of A=4ut4a

and d=4
 186 1 [ Determinant-Characteristic Roos

If B is any other matrix of f relative to another basis of

R then B and A are similar matrices, so their characteristic
roots A1» Aa are the same
Let B=[b11 bie
ba1 bes
Hence T (f) with/respect to the new
basis==bju+bg
=tAs
So T,(f) is ipdependent of the basis chosen.
Theorem (2.2). Cayley-Hamilton Theorem
Every Squar matrix is a root of its characteristic polynomial.
Proof. Let A=[ai] be an n-Square matrix over a ficld F.
Let its characteristic polynomial be

(1) #)=l|4-x Inl=(-1)* {x"+an.1 *-it..+a1 *+a}

Let the adjoint of A-x In be B.
Now the elements of B are the co-factors of the elements
of l4-x 1,| and hence are polynomials in x of degree n-1 or
less.
Hence the matrix B can be expressed as the sum of n matrices
of order n as follows
(2).B-B,-1 +B,- -4...+ B, xt+B
where B, Bi. Bg»..Bq-1 are n-Square matrices independent
of x.
By the property of Adjoint we have
(4-x ,) B=|4-x I, n [By Theor (2:1)]
or (A-x In) (Br-1 a*-14...+B *+B,)=
(-1" {&"+a- -1+... +a,} [From (1) and (2)]
or

(3)-1, B-1 x+(4B,-1-l, Ba-2) *1+(AB,--I, Bg-s)1+

+(AB1-1, B,) xtAB,=(-1"{*"+a-1 *-14+a}
Equating the coefficients of corresponding powers of x of
the two sids of (3) we get

(-1)" -B,-1
(-1" a- h=AB-1- Ba-
Determinant-Characteristic Roots] 187

(-1" ag-l-4 Ba--Ba-3

*********
(-1 a I,=AB1-B,
(-1" a I,=ABo
A", An-,aA, lp
Multiplying the above matrix equations by
respectively we get
(-1)" A"=-A" Bn-1
a,n-1 A"-T=A"
B-1-A-I B-2
(-1)"

a1 A=A' B,-AB,
-1
(-1" agI=AB,
we have
Adding the above equations
4-1+-+a1 A+4, In=0.
(-1" {4"+an-1
dA)=0.
or

the characteristic polynomial p{«) of A.

Thus A is a root of be
A can
of Inverse a non-singular matrix
Note. (2.2). as illustrated by the
Hamilton Theorem
computed by Cayley-
following example
Example-set
Theorem find the inverse of
Cayley-Hamilton
(2.16), /By
2 3
the matrix A=
2 4 3
Solution:
Characteristic polynomial of Ais given by
2
3 --9x+17x+15)(Prove)
)=| 5-
4 3-X
2
Theorem A is a zero of 4(x), so
Cayley-Hamilton
By A3-9A-+174+15 1=0

A3-94-+17.4-15
Or
4-1 we have
Miltiplying by
A-94-+17 Ia=--15 A
 188 Determinant-Char acieristic Roots

Hence A (4-9A+1715)
Now A-94+17 9 -18
1=[(3 24 45
27
0
8 36 15 -36-27
+17 0
0 17
0 0 17
15 -15
-10
Hence A- 15
0-3
6-15 -2/5
-11 0 5
0 2/3 0 -1/3
Definition (2.4). Characteristic Polynomial of a liaear map.
It T be a linear transformation on a vector space V. Then
the Characteristic Polynomial of T is defined to be the Charac-
teristic Polynomial of a6y matrix of T, since such matrices being
similar, ha ve the samg characte ristic polynomial by Ex. (2*12).
Definition (2Polynomial in a linear transformation T oa V.
Let T be a linear transformation on a vector space V over
field F. Let
Sx)=a, "+a-1 x"=...+a1 *t49
be a polynomial in x over F.
We define S(T)=4, T"+a,-, T"-t..+a1 T+4, l.
Now (T) is linear transformation
also/a V (Prove),
on
f(T)
is called a polynumit in T over the field F.
Theorem (2 3 Let 1T be a lnea transformation on V and a
be a charac ter isti: root of T. fS()¬F Ix| then f(A) is a Chara-
cteristic root f JCT)
Proof Since a is a Characteristic root of 7, so

340,such that 7()-Av.

Now T)-T(T0)-7(A)=AT(V)=av
Determinant-Characteristic Roots ]
189

Continuing thus we get

k.
(1) T (v)=akv for all positive integers
If S(«)=a, x"+a,-1 -14... +a1 x+ao
¬F[x]
T+a0 .
Then S(T)=4,T"+4-1 T"-1t...+a4
T"-14...+a1 T+40 ) (v)
So S(T) (V)=(4,Tn+a,-1
T(V)+a, (")
T"(v)+a.-1 TH- (v)4...+a1
=a,
v-+an-1 - l y+..+a,
Av+ag
=an A A+a,)v
-1+...+a,
=(a, A"+an-1
=f(a) v.
of S(T) and v is an
characteristic root
Therefore f(a) is a
value fa).
vector of S(T) with eigen
eigen matrix.
an n-square
Definition (2 6). Folynomial in
matrix over a field F.
Let A be an n-square
1 . . . + a 1 x+ag ¬F[x].
Let S(x)=Cn X"+41-1
An-1..+a1 A+4, ,
Then fA)=a, A"+a1-1
F.
polynomial in A
over
is called a
that:
as in Theor. (2:3).
We prove exactly
f(A) will be
can an
A, then
value of the matrix
If a be an cigen
eigen value of f(A). over F of dimension
V be a vector space
Theorem (2 Le There exists a
non-zero
n on-zero polyno-
polyno-
linear map on V.
n. Ler T be a

such that fT)=0.

eF[x]
mial fr) Dim Hom (V, V)=n*=m.
Froof (V, V) and
elements of Hom
T2,...Tm are (m + 1)
Now , 7,
hence are linearly dependent. that
all zero. such
. . ¬ F , not
Therefore 3 a0
taT-+a,7*+...+am Tm=0
+"ag...+an X ¬F[x]
if Su)=°% zeros.
Thus coefis, a; are not
since all the
f(x)#0
then f(r), such that
non zero
polynomial
Thus there cxistsa
S(T)=«,4+«,T+-ag7-...+anTi=0.
 190 Determbnant-Characteristic Roots
Defieition (27) Minimal Polynomial.
Let T be linear transformation on V. A non-zero monic
polynomial m(*) is said to be the minin al polynomial of T if
)m(T)==0
(i) If f(T)==0 for some polynomial ft), then m(x) divides
Sa), that is, m(x) is the polynomial of lowest degree such that
m(T)=0.
It can be proved that such a polynomial exists and is unique.
Similarly m() is said to be the minimal polynomial of a
square matrix A, if m(r) is the polynomial of lowest degree such
that mA)=0.
Theorem (25). 1 a«F is an eigen value of a linear trans-
Jormation T on á vector space V, then a is a zero of the minimal
polynomial m(x) of T.
Proof Since a is an eigen value of T, so m(a) is an eigen
value of m(T) [By Theor. (2*3)]
Hence 3 V#0, such that
m(T) (V)=m(a) v
or 0=m(a) [Since m(T)=0]
Hence m()=0 since v40.
Thus a is a zero of the minimal polynomial m(x).
Note (2:3). If A be a matrix of the linear transformation T
and o(x) be the Characteristc polynomial of A, then $(:) is also
the Characteristic polynomial of T. (By Defin. (2-4))
Now MA)=0 [By C. Hamilton Theorem]
Hence (T)-0
So the minimal polynomial m(x)
of T must be a factor of
d(x) and therefore every zero of m(x) is also a zero of p(x).
Thus every zero of m(r) is
characteristic root of T. This
a
result together with the previous Theorem. (2:5) shows that the zeros
of the minimal polynomial of T are precisely all its eigen values.
Determinant-Characteristic Roots ] 191

Example Set

(217). Find the minimal polynomial m(A) of tho matrix

A2 1 0 0
020o
00 2 0
000 5
Solotion: Characteristic polynomial of A is
0 0
) 2-A
2-A 0
0
2-
0 5-A
=(2-A) (5-A)
=(A-2 (A-5)
have the same zeros, hence
By note (2:3) m(a) and o(a)
m(a) must also divide
(A-2), (A-5) must be factors of m() and
be one of the following
dA), so m(1) must
mA)=(a-2) 0-5), ma(A)=(A-2(A-5), m,(a)=(A-2A-5)
that m(A)£0 but
We can verify
m(A)=0 and m,(4)=#{A)=0
Hence mz(A)==(A-2)"(A-5) is the required minimal polyno-

mial of A.

(2.18). Let M=1

Prove that Characteristic
A,, A, are square matrices.
W lhere

of the Characte ristic polynomials

polynomial of M is the product
of A, and A
B
Hint M-»l|-A,-A
0 A-AI

| 4 - 1 |4,-A
Ilemitian Matrix.
Dehuitin (2 6).
i q u a r c m a t I over tie field C'ol com
et 1 tc n

icxnunbhe