Dr.R.Venkatesan Matrices

1.
54 ■ Engineering Mathematics
1.6 EIGEN VALUES AND EIGEN VECTORS

1.6.0 Introduction
In this section we study certain numbers associated with a square matrix, called eigen values and
certain vectors associated with them, called eigen vectors. The problem of determining eigen values
and eigen vectors of a square matrix is called an eigen value problem. The eigen value problem arise in
a wide range of physical and engineering applications such as mechanical system, electrical networks,
Markov processes, elastic deformations, etc.
Eigen value problems are used in diagonalisation of a matrix of a quadratic form.
1.6.1 Vector
The vector 2i + 3 j − 4 k can be regarded as the triplet (2, 3, −4).
Definition 1.28 An ordered n-tuple (x1, x2, …, xn) of numbers x1, x2, …, xn is called an n-dimensional
vector.
For example the triplet (2, 3, −4) is a 3-dimensional vector. (1, 0, −2, 3) is a 4-dimensional vector.
A row matrix is also called a row vector and a column matrix is called a column vector.
Definition 1.29 If X1 = (a1, a2, …, an), X2 = (b1, b2, …, bn) be two n-dimensional vectors, then their
sum and scalar multiplications are
X1 + X2 = (a1 + b1, a2 + b2, …, an + bn),

aX1 = (aa1, aa2, …, aan), which are n-dimensional vectors.
X1 = X2 if and only if a1 = b1, a2 = b2, …, an = bn
Definition 1.30 Linear Combination

If X1, X2, …, Xr are r vectors of n-dimension and if a1, a2, …, ar are numbers, then the vector
a1X1 + a2X2 + … + arXr is called a linear combination of the vectors X1, X2, …, Xr.
Definition 1.31 Linearly dependent and independent vectors
(a) The set of vectors X1, X2, …, Xr is said to be linearly dependent if there exist numbers
a1, a2, …, ar, not all zero, such that a1X1 + a2X2 + … + arXr = 0
(b) The set of vectors X1, X2, …, Xr is said to be linearly independent if any relation of the form
a1X1 + a2X2 + … + arXr = 0 ⇒ a1 = 0, a2 = 0, …, ar = 0
Note
(i) If X1, X2, …, Xr are linearly dependent, then some vector is a linear combination of others.
(ii) In a plane or 2-dimensional space, non-collinear vectors are linearly independent vectors whereas
collinear vectors are dependent vectors.
In 3-dimesional space, non-coplanar vectors are linearly independent vectors whereas coplanar
vectors are dependent vectors.
Example: i = (1, 0, 0 ), j = (0, 1, 0 ), k = (0 , 0, 1) are linearly independent vectors.
(iii) Any set of vectors containing zero vector 0 is a linearly dependent set.
M01_ENGINEERING_MATHEMATICS-I _CH01_Part A.indd 54 5/30/2016 4:36:18 PM

Matrices ■ 1.55
(iv) Rank of an m × n matrix A is equal to the maximum number of independent column vectors or
row vectors of A.
(v) A useful result to test linear independence:
Let X1, X2, …, Xn be n vectors of n-dimensional space. Let A be the matrix having these n-vectors as
columns (or rows). Then A is a square matrix of order n. If A ≠ 0 , then X1, X2, …, Xn are linearly
independent.
If A = 0 , then X1, X2, …, Xn are linearly dependent.
WORKED EXAMPLES
EXAMPLE 1
Show that the vectors (1, 2, 3), (3, 22, 1), (1, 26, 25) are linearly dependent.
Solution.
⎡1 3 1⎤
Let A = ⎢⎢ 2 −2 −6 ⎥⎥ with the vectors as columns.
⎢⎣ 3 1 −5⎥⎦
1 3 1
Then A = 2 −2 −6 = 1⋅ (10 + 6) − 3( −10 + 18) + 1⋅ ( 2 + 6) = 16 − 24 + 8 = 0
3 1 −5
∴ the vectors (1, 2, 3), (3, −2, 1) and (1, −6, −5) are linearly dependent.
EXAMPLE 2
Show that the vectors X1 = (1, 2, 23, 4), X2 = (3, 21, 2, 1), X3 = (1, 25, 8, 27) are linearly
dependent and find the relation between them.
Solution.
⎡ 1 2 −3 4 ⎤
Let A = ⎢⎢3 −1 2 1⎥⎥ with the vectors as rows.
⎢⎣ 1 −5 8 −7⎥⎦
We shall use elementary row operations.
⎡ 1 2 −3 4⎤
∴ A ∼ ⎢0 −7 11 −11⎥⎥ R 2 → R 2 − 3R1 = R 2′
⎢
⎣⎢0 −7 11 −11⎥⎦ R 3 → R 3 − R1 = R 3′
⎡ 1 2 −3 4⎤
⎢
∼ ⎢0 −7 11 −11⎥⎥
⎢⎣0 0 0 0 ⎥⎦ R 3 → R 3′ − R 2′ = R 3′′
Since the maximum number of non-zero rows is 2, which is less than the number of vectors, the given
vectors are linearly dependent.
M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 55 5/30/2016 5:03:24 PM

1.56 ■ Engineering Mathematics
The relation between them is obtained as below.

R 3′′ = 0 ⇒ R 3′ − R 2′ = 0
⇒ R 3 − R1 − ( R 2 − 3R1 ) = 0
⇒ R 3 − R 2 + 2R1 = 0
Since the rows are vectors, we get X3 − X2 + 2X1 = 0 which is the relation between the vectors.
Note The rows of the matrix are the given vectors. So, only row operations must be used to find the
relationship between the vectors.
1.6.2 Eigen Values and Eigen Vectors

Definition 1.32 Let A be a square matrix of order n. A number l is called an eigen value of A if there
exists a non-zero column matrix X such that AX = lX. Then X is called an eigen vector of A cor-
responding to l.
⎡ x1 ⎤
⎢x ⎥
If A = ⎡⎣aij ⎤⎦ and X = ⎢ 2 ⎥ , then AX = lX ⇒ (A − lI)X = 0.
n×n ⎢:⎥
⎢ ⎥
⎣x n ⎦
This will represent a system of linear homogeneous equations in x1, x2, …, xn.
Since X ≠ 0 at least one of the xi ≠ 0.
Hence, the homogeneous system has nontrivial solutions.
∴ the determinant of coefficients A − lI = 0.
This equation is called the characteristic equation of A.
The determinant A − lI , on expansion, will be a nth degree polynomial in l and is known as the
characteristic polynomial of A.
The roots of the characteristic equation are the eigen values of A.
Definition 1.33 Characteristic Equation and Characteristic Polynomial

If l is a characteristic root of a square matrix A, then A − lI = 0 is called the characteristic equation
of A.
The polynomial A − lI in l is called the characteristic polynomial of A.
Note
(1) The word ‘eigen’ is German, which means ‘characteristic’ or ‘proper’. So, an eigen value is also
known as characteristic root or proper value. Sometimes it is also known as latent root.
⎡a a12 ⎤ ⎡1 0 ⎤
(2) If A = ⎢ 11 ⎥ , I=⎢ ⎥ , then the characteristic equation of A is
⎣ a21 a22 ⎦ ⎣0 1⎦
A − lI = 0
a11 − l a12
⇒ =0
a21 a22 − l
⇒ (a11 − l )(a22 − l ) − a21a12 = 0

Matrices ■ 1.57
⇒ a11 ⋅ a22 − (a11 + a22 )l + l 2 − a21a12 = 0

⇒ l 2 − (a11 + a22 )l + (a11a22 − a21a12 ) = 0
⇒ l 2 − S 1l + S 2 = 0
where S1 = a11 + a22 = sum of the diagonal elements of A.
S2 = a11 ⋅ a22 − a21 ⋅ a12 = A
⎡ a11 a12 a13 ⎤ ⎡1 0 0 ⎤

(3) If A = ⎢a21 a22 a23 ⎥⎥ , I = ⎢⎢0 1 0 ⎥⎥ , then the characteristic equation of A is
⎢
⎢⎣a31 a32 a33 ⎥⎦ ⎢⎣0 0 1 ⎥⎦
A − lI = 0
a11 − l a12 a13
⇒ a21 a22 − l a23 = 0
a31 a32 a33 − l
Expanding this determinant we will get
l3 − S1l2 + S2l − S3= 0,
where S1 = sum of the diagonal elements of A

S2 = sum of the minors of elements of the main diagonal
S3 = A
We will use this formula in problems.
Definition 1.34 The set of all distinct eigen values of the square matrix A is called the spectrum of A.
The largest of the absolute values of the eigen values of A is called the spectral radius of A. The set
of all eigen vectors corresponding to an eigen value l of A, together with zero vector, forms a vector
space which is called the eigenspace of A corresponding to l.
1.6.3 Properties of Eigen Vectors

Theorem 1.3
(1) Eigen vector corresponding to an eigen value is not unique.
(2) Eigen vectors corresponding to different eigen values are linearly independent.
Proof
(1) Let l be an eigen value of a square matrix A of order n.
Let X be an eigen vector corresponding to l.
Then AX = lX
Multiply by a constant C
∴ C(AX) = C(lX) ⇒ A(CX) = l(CX)
Since C ≠ 0, X ≠ 0 we have CX ≠ 0
∴ CX is an eigen vector corresponding to l for any C ≠ 0. Hence, eigen vector is not unique
for the eigen value l.

(2) Let l1, l2 be two different eigen values of A.

Let X1, X2 be the corresponding eigen vectors.
∴ AX1 = l1X1 (1)

and AX2 = l2X2 (2)
We have to prove X1 and X2 are linearly independent.

Suppose a1X1 + a2X2 = 0 (3)
then A(a1X1 + a2X2) = 0
⇒ a1 ( AX1 ) + a 2 ( AX 2 ) = 0
⇒ a1 (l1X1 ) + a 2 (l 2 X 2 ) = 0
⇒ (a1l1 )X1 + (a 2 l 2 )X 2 = 0 (4)
Multiply (3) by l1, we get
l1 (a1X1 ) + l1 (a 2 X 2 ) = 0
⇒ (a1l1 )X1 + (a 2 l1 )X 2 = 0 (5)

(4) − (5) ⇒ a 2 ( l 2 − l1 ) X 2 = 0 (6)
Since l1 ≠ l 2 ⇒ l 2 − l1 ≠ 0 and X 2 ≠ 0
∴ ( l 2 − l1 ) X 2 ≠ 0
∴ (6) ⇒ a 2 = 0 ∴ (3) ⇒ a1X1 = 0 ⇒ a1 = 0, since X1 ≠ 0.

Thus, a1X1 + a 2 X 2 = 0 ⇒ a1 = 0 and a 2 = 0
∴ X1 and X2 are linearly independent.
Note
(1) If all the n eigen values l1, l2, …, ln of A are different, then the corresponding eigen vectors
X1, X2, …, Xn are linearly independent.
(2) A given eigen vector of A corresponds to only one eigen value of A.
(3) Eigen vectors corresponding to equal eigen values may be linearly independent or dependent.
WORKED EXAMPLES
EXAMPLE 1
⎡4 1 ⎤
Find the eigen values and eigen vectors of the matrix ⎢ ⎥.
⎣3 2 ⎦
Solution.
⎡4 1⎤
Let A = ⎢ ⎥.
⎣ 3 2⎦
The characteristic equation of A is A − lI = 0

Matrices ■ 1.59
4−l 1
⇒ =0 ⇒ l 2 − S1l + S2 = 0
3 2−l
where S1 = sum of the diagonal elements of A = 4 + 2 = 6
4 1
S2 = A = = 8−3= 5
3 2
∴ the characteristic equation is l2 − 6l + 5 = 0 ⇒ (l − 1) (l − 5) = 0 ⇒ l = 1, 5
which are the eigen values of A.
To find eigen vectors:

⎡ x1 ⎤
Let X = ⎢ ⎥ be an eigen vector of A corresponding to l.
⎣x 2 ⎦
( A − lI)X = 0 ⎡4 − l 1 ⎤ ⎡ x 1 ⎤ ⎡0 ⎤
Then ⇒ ⎢ 3 ⎢ x ⎥ = ⎢0 ⎥
⎣ 2 − l ⎥⎦ ⎣ 2⎦ ⎣ ⎦
( 4 − l )x 1 + x 2 = 0 ⎫
⇒ ⎬
3x 1 + ( 2 − l )x 2 = 0 ⎭ (I)
Case (i) If l = 1, then equations (I) become
3x1 + x2 = 0 and 3x1 + x2 = 0 ∴ x2 = −3x1
⎡ 1⎤
Choosing x1 = 1, we get x2 = −3. ∴ eigen vector is X1 = ⎢ ⎥
⎣ −3⎦
Case (ii) If l = 5, then equations (I) become
−x1 + x2 = 0 and 3x1 − 3x2 = 0 ∴ x1 = x2
⎡1⎤
Choosing x1 = 1, we get x2 = 1 ∴ eigen vector is X2 = ⎢ ⎥
⎣1⎦
⎡ 1⎤ ⎡1⎤
Thus, eigen values of A are 1, 5 and the corresponding eigen vectors are ⎢ ⎥ , ⎢ ⎥
⎣ −3⎦ ⎣1⎦
Note In case (i) we have only one equation 3x1 + x2 = 0 to solve for x1 and x2. So, we have infinite
number of solutions x1 = k, x2 = −3k, for any k ≠ 0. We have chosen the simplest solution.
⎡ k⎤ ⎡ 1⎤
Infact ⎢ ⎥ = k ⎢ ⎥ is an eigen vector for l = 1 for any k ≠ 0. So, for l = 1 there are many eigen
⎣ −3k ⎦ ⎣ −3⎦
vectors. This verifies property 1.
EXAMPLE 2
⎡1⎤ ⎡ 1⎤
Show that the real matrix ⎡ a b ⎤ has two eigen vectors
⎢i ⎥ and ⎢ −i ⎥ , where b ≠ 0.
⎢⎣2b a ⎥⎦ ⎣ ⎦ ⎣ ⎦
Solution.
⎡ a b⎤
Let A = ⎢ ⎥.
⎣ −b a⎦

a−l b
⇒ =0 ⇒ l 2 − S1l + S2 = 0
−b a−l
where S1 = sum of the diagonal elements of A = a + a = 2a

a b
S2 = A = = a2 + b 2
−b a
∴ the characteristic equation is
l2 − 2al + (a2 + b2) = 0 ⇒ l = 2a ± 4a2 − 4(a2 + b 2 )

2a ± −4b 2 2a ± 2ib
= = = a + ib or a − ib
2 2
which are the eigen values of A.

⎡x ⎤
Let X = ⎢ 1 ⎥ be an eigen vector of A corresponding to l.
⎣x 2 ⎦
⎡a − l b ⎤ ⎡ x 1 ⎤ ⎡0⎤
Then ( A − lI)X = 0 ⇒ ⎢ ⎥ ⎢ ⎥=⎢ ⎥
⎣ −b a − l ⎦ ⎣ x 2 ⎦ ⎣0 ⎦
(a − l )x 1 + bx 2 = 0⎫
⇒ ⎬
−bx 1 + (a − l )x 2 = 0⎭ (I)
Case (i) If l = a + ib, then the equations (I) become
(a − (a + ib ))x 1 + bx 2 = 0
and −bx 1 + (a − a − ib )x 2 = 0
⇒ −ibx 1 + bx 2 = 0 ⇒ x 2 = ix 1
and −bx 1 − ibx 2 = 0 ⇒ − x 1 = ix 2
⇒ i x 1 = ix 2
2
⇒ x 2 = ix 1
So, we have only one equation x 2 = ix 1

⎡1⎤
Choosing x1 = 1, we get x2 = i ∴ an eigen vector is X1 = ⎢ ⎥
⎣i ⎦
Case (ii) If l = a − ib, then the equations (I) become
(a − a + ib )x 1 + bx 2 = 0
and −bx 1 + (a − a + ib )x 2 = 0
⇒ ibx 1 + bx 2 = 0 ⇒ x 2 = −ix 1
−bx 1 + ibx 2 = 0 ⇒ x 2 = −ix 1
⎡ 1⎤
Choosing x1 = 1, we get x2 = −i ∴ an eigen vector is X 2 = ⎢ ⎥
⎣ −i ⎦

Matrices ■ 1.61
Thus, the eigen values of A are a + ib, a − ib and the corresponding eigen vectors are
⎡1⎤ ⎡1⎤
X1 = ⎢ ⎥ and X 2 = ⎢ ⎥
⎣i ⎦ ⎣ −i ⎦
EXAMPLE 3
⎡3 24 4 ⎤
Find the eigen values and eigen vectors of the matrix ⎢1 22 4 ⎥⎥ .
⎢
⎢⎣1 21 3 ⎥⎦
Solution.
⎡3 −4 4 ⎤
Let A = ⎢⎢1 −2 4 ⎥⎥
⎣⎢1 −1 3 ⎦⎥
3−l −4 4
⇒ 1 −2 − l 4 =0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
1 −1 3−l
where S1 = sum of main diagonal elements of A = 3 + (−2) + 3 = 4
S2 = sum of minors of diagonal elements of A
−2 4 3 4 3 −4
= + + = −6 + 4 + 9 − 4 + ( −6) + 4 = −2 + 5 + ( −2) = 1
−1 3 1 3 1 −2
and S3 = A = 3( −6 + 4) + 4(3 − 4) + 4( −1 + 2) = −6 − 4 + 4 = −6
∴ the characteristic equation is l3 − 4l2 + l + 6 = 0
We choose integer factors of constant term 6 for trial solution. We find l = −1 is a root. To find the
other roots we perform synthetic division
Other roots are given by
⇒ l 2 − 5l + 6 = 0 1 −4 1 6
−1
⇒ (l − 2)(l − 3) = 0 0 −1 5 −6
⇒ l = 2 or 3 1 −5 6 0
∴ the eigen values are l = −1, 2, 3 [different roots]

⎡ x1 ⎤
Let X = ⎢⎢ x 2 ⎥⎥ be an eigen vector corresponding to the eigen value l.
⎢⎣ x 3 ⎥⎦
⎡3 − l −4 4 ⎤ ⎡ x 1 ⎤ ⎡0 ⎤
Then ( A − lI)X = 0 ⇒ ⎢⎢ 1 −2 − l 4 ⎥⎥ ⎢ x ⎥ = ⎢0 ⎥
⎢ 2⎥ ⎢ ⎥
⎢⎣ 1 −1 3 − l ⎥⎦ ⎢⎣ x 3 ⎥⎦ ⎢⎣0 ⎥⎦

(3 − l ) x 1 − 4 x 2 + 4 x 3 = 0 ⎫
⇒ ⎪
x 1 − ( 2 + l )x 2 + 4 x 3 = 0⎬ (I)
x 1 − x 2 + (3 − l )x 3 = 0⎪⎭
Case (i) If l = −1, then the equations (I) become

4x1 − 4x2 + 4x3 = 0 ⇒ x1 − x2 + x3 = 0
x1 − x2 + 4x3 = 0 and x1 − x2 + 4x3 = 0
The different equations are x1 − x2 + x3 = 0 and x1 − x2 + 4x3 = 0
x1 x2 x3
By rule of cross multiplication, we get
−1 1 1 −1
x1 x x3
= 2 =
−4 + 1 1 − 4 −1 + 1 −1 4 1 −1
x1 x x x1 x2 x3
⇒ = 2 = 3 ⇒ = =
−3 −3 0 1 1 0
⎡1 ⎤
Choosing x1 = 1, x2 = 1, x3 = 0, we get an eigen vector X1 = ⎢1 ⎥
⎢⎣0 ⎥⎦

x1 − 4x2 + 4x3 = 0, x1 − 4x2 + 4x3 = 0 and x1 − x2 + x3 = 0
∴ the different equations are x1 − 4x2 + 4x3 = 0 and x1 − x2 + x3 = 0
By the rule of cross multiplication, we get
x1 x x3 x1 x2 x3
= 2 =
−4 + 4 4 − 1 −1 + 4 −4 4 1 −4
x1 x 2 x x1 x 2 x 3
⇒ = = 3 ⇒ = = −1 1 1 −1
0 3 3 0 1 1
⎡0 ⎤
Choosing x1 = 0, x2 = 1, x3 = 1, we get an eigen vector X 2 = ⎢1 ⎥
⎢⎣1 ⎥⎦
Case (iii) If l = 3, then equations (I) become

0x1 − 4x 2 + 4x 3 = 0 ⇒ 0x1 − x 2 + x 3 = 0
x 1 − 5x 2 + 4 x 3 = 0 and x 1 − x 2 + 0 x 3 = 0
The equations are different, but only two of them are independent. So, we can choose any two of them
to solve. From the first two equations, we get
x1 x x x1 x2 x3
= 2 = 3
−4 + 5 1 − 0 0 + 1 −1 1 0 −1
x1 x 2 x
⇒ = = 3 −5 4 1 −5
1 1 1

Matrices ■ 1.63
⎡1⎤
Choosing x1 = 1, x2 = 1, x3 = 1, we get an eigen vector X 3 = ⎢1⎥
⎢⎣1⎥⎦
Thus, the eigen values of A are −1, 2, 3 and corresponding eigen vectors are
⎡1 ⎤ ⎡0 ⎤ ⎡1⎤
X1 = ⎢1 ⎥ , X 2 = ⎢1 ⎥ , X 3 = ⎢1⎥
⎢⎣0 ⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣1⎥⎦
Note
(1) We are using the following integer root theorem for trial solution. “For the equation of the form
xn + an − 1 xn − 1 + an − 2 xn − 2 + … + a1x + a0 = 0 with integer coefficients ai, any rational root is an
integer and is a factor of the constant term a0”. So, it is enough we try factors of the constant term
for integer solutions. If there is no integer solution, then real roots should be irrational.
(2) In the above problem the eigen values −1, 2, 3 are different. So, by property (2) the eigen vectors
are linearly independent. We shall verify this:
⎡1 1 0⎤
Consider B = ⎢0 1 1 ⎥ with the eigen vectors as rows.
⎢ ⎥
⎢⎣1 1 1 ⎥⎦
Then B = 1⋅ 0 − 1( −1) + 0 = 1 ≠ 0
∴ X1, X2, X3 are linearly independent.
EXAMPLE 4
⎡2 2 1⎤
Find the eigen values and eigen vectors of ⎢ 1 3 1⎥ .
⎢⎣ 1 2 2 ⎥⎦
Solution.
⎡2 2 1⎤
Let A = ⎢1 3 1 ⎥
⎢⎣1 2 2⎥⎦
The characteristic equation of A is A − l I = 0
2−l 2 1
⇒ 1 3−l 1 =0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
1 2 2−l
where S1 = sum of the diagonal elements of A = 2 + 3 + 2 = 7
S2 = sum of minors of the diagonal elements of determinant A
3 1 2 1 2 2
= + + = 6 − 2 + 4 − 1 + 6 − 2 = 11
2 2 1 2 1 3
S3 = A = 2(6 − 2) − 2( 2 − 1) + 1( 2 − 3) = 8 − 2 − 1 = 5
∴ the characteristic equation is l3 − 7l2 + 11l − 5 = 0

Choose the integer factors of constant term −5 for trial.

The integer factors of −5 are −5, 1, or −1, 5.
We find l = 1 is a root.
Other roots are given by l2 − 6l + 5 = 0 1 1 −7 11 −5
⇒ (l − 1) (l − 5) = 0 0 1 −6 5
⇒ l = 1, 5 1 −6 5 0
∴ the eigen values are l = 1, 1, 5 (Two equal eigen values)

⎡ x1 ⎤
Let X = ⎢⎢ x2 ⎥⎥ be an eigen vector of A corresponding to the eigen value l.
⎢⎣ x3 ⎥⎦
⎡2 − l 2 1 ⎤ ⎡ x 1 ⎤ ⎡0 ⎤
Then ( A − lI)X = 0 ⇒ ⎢ 1 3−l 1 ⎥ ⎢ x 2 ⎥ = ⎢0 ⎥
⎢⎣ 1 2 2 − l ⎥⎦ ⎢ x ⎥ ⎣⎢0 ⎥⎦
⎣ 3⎦
⇒ ( 2 − l ) x 1 + 2x 2 + x 3 = 0 ⎫
⎪
x 1 + (3 − l ) x 2 + x 3 = 0 ⎬ (I)
x 1 + 2x 2 + ( 2 − l )x 3 = 0 ⎪⎭
Case (i) If l = 5, then the equations (I) become
−3x1 + 2x2 + x3 = 0, x1 − 2x2 + x3 = 0 and x1 + 2x2 − 3x3 = 0
These 3 equations are different, but only 2 of them are independent.
So, we can choose any two of them to solve for x1, x2, x3.
From last two equations, by the rule of cross multiplication, we get
x1 x x x1 x2 x3
= 2 = 3
6 − 2 1+ 3 2+ 2 −2 1 1 −2
⇒ x1 x
= 2 =
x3
4 4 4 2 −3 1 2
⇒ x1 x2 x3
= =
1 1 1
⎡1⎤
Choosing x1 = 1, x2 = 1, x3 = 1, we get an eigen vector X1 = ⎢⎢1⎥⎥
⎣⎢1⎦⎥
Case (ii) If l = 1, then the equations (I) become
x1 + 2x2 + x3 = 0, x1 + 2x2 + x3 = 0 and x1 + 2x2 + x3 = 0
We have only one equation x1 + 2x2 + x3 = 0 to solve for x1, x2, x3. Assign arbitrary values for two var-
iables and solve for the third.
Choose x3 = 0, then x1 + 2x2 = 0 ⇒ x1 = −2x2
⎡ −2⎤
Choose x2 = 1, ∴ x1 = −2, we get an eigen vector X 2 = ⎢⎢ 1⎥⎥
⎢⎣ 0 ⎥⎦

Matrices ■ 1.65
We shall find one more solution from x1 + 2x2 + x3 = 0

Choose x2 = 0 then x1 + x3 = 0 ⇒ x3 = − x1
Choose x1 = 1 ∴ x3 = −1
⎡ 1⎤
∴ another eigen vector corresponding to l = 1 is X 3 = ⎢ 0 ⎥⎥
⎢
⎢⎣ −1⎥⎦
Thus, eigen values of A are 5, 1, 1 and the corresponding eigen vectors are
⎡1⎤ ⎡ −2⎤ ⎡ 1⎤
X1 = ⎢1⎥ , X 2 = ⎢ 1⎥ and X 3 = ⎢⎢ 0 ⎥⎥
⎢ ⎥ ⎢ ⎥
⎣⎢1⎦⎥ ⎣⎢ 0 ⎦⎥ ⎣⎢ −1⎦⎥
Note Though the eigen values are not different, we could find independent eigen vectors.
⎡ 1 1 1⎤
For, consider B = ⎢⎢ −2 1 0 ⎥⎥ with the vectors as rows
⎣⎢ 1 0 −1⎦⎥
Then B = 1( −1 − 0) − 1( 2 − 0) + 1(0 − 1) = −1 − 2 − 1 = −4 ≠ 0
∴ X1, X2, X3 are linearly independent.
EXAMPLE 5
⎡ 6 −6 5 ⎤
⎢ −13 10 ⎥⎥ .
Find the eigen values and eigen vectors of the matrix ⎢ 14
Solution. ⎢⎣ 7 −6 4 ⎥⎦
⎡ 6 −6 5⎤
Let A = ⎢⎢14 −13 10 ⎥⎥
⎢⎣ 7 −6 4 ⎥⎦
6−l −6 5
−13 − l 10 = 0 ⇒ l − S1l + S2 l − S3 = 0
3 2
⇒ 14
7 −6 4−l
where S1 = sum of the diagonal elements of A = 6 + (−13) + 4 = −3
S2 = sum of minors of elements of the diagonal of A
−13 10 6 5 6 −6
= + +
−6 4 7 4 14 −13
= ( −52 + 60) + ( 24 − 35) + ( −78 + 84) = 8 − 11 + 6 = 3
S3 = A = 6( −52 + 60) + 6(56 − 70) + 5( −84 + 91)
= 48 + 6( −14) + 5(7) = 48 − 84 + 35 = −1
∴ the characteristic equation is l3 + 3l2 + 3l + 1 = 0
⇒ (l + 1)3 = 0 ⇒ l = −1, −1, −1
Three equal eigen values.


⎡ x1 ⎤
Let X = ⎢⎢ x2 ⎥⎥ be an eigen vector corresponding to the eigen value l.
⎢⎣ x3 ⎥⎦
Then ( A − lI)X = 0
⎡6 − l −6 5 ⎤ ⎡ x 1 ⎤ ⎡0⎤
⎢ 14 −13 − l 10 ⎥⎥ ⎢ x ⎥ = ⎢0⎥
⇒ ⎢ ⎢ 2⎥ ⎢ ⎥
⎢⎣ 7 −6 4 − l ⎥⎦ ⎢⎣ x 3 ⎥⎦ ⎢⎣ 0 ⎥⎦
( 6 − l ) x 1 − 6 x 2 + 5x 3 = 0 ⎫
⎪
⇒ 14 x 1 − (13 + l )x 2 + 10 x 3 = 0 ⎬
7x 1 − 6 x 2 + ( 4 − l )x 3 = 0 ⎪⎭ (I)
If l = −1, then the equations (I) become
7 x1 − 6 x2 + 5 x3 = 0
14 x1 − 12 x2 + 10 x3 = 0 ⇒ 7 x1 − 6 x2 + 5 x3 = 0
7 x1 − 6 x2 + 5 x3 = 0
We have only one equation 7x1 − 6x2 + 5x3 = 0
Assign arbitrary values to two variables and find the third.
We shall find 3 vectors.
x x
Putting x1 = 0, we get −6x2 + 5x3 = 0 ⇒ 6x2 = 5x3 ⇒ 2 = 3
5 6
⎡0 ⎤
Choosing x2 = 5, x3 = 6, we get an eigen vector X1 = ⎢⎢ 5⎥⎥
⎣⎢6 ⎦⎥
x x
Putting x2 = 0, we get 7x1 + 5x3 = 0 ⇒ 7x1 = −5x3 ⇒ 1 = 3
5 −7
⎡ 5⎤
Choosing x1 = 5, x3 = −7, we get the second eigen vector X 2 = ⎢ 0 ⎥⎥
⎢
Again, putting x3 = 0, we get ⎢⎣ −7⎥⎦
x1 x2
7x1 − 6x2 = 0 ⇒ 7x1 = 6x2 ⇒ =
6 7
⎡6 ⎤
Choosing x1 = 6, x2 = 7, we get the third eigen vector X 3 = ⎢⎢ 7⎥⎥
⎢⎣0 ⎥⎦
Thus, eigen values of A are −1, −1, −1 and the corresponding eigen vectors are
⎡0 ⎤ ⎡ 5⎤ ⎡6 ⎤
⎢ ⎥ ⎢
X1 = ⎢ 5 ⎥ , X 2 = ⎢ 0 ⎥ ⎥ and X 3 = ⎢ 7⎥⎥ ,
⎢
⎢⎣6 ⎥⎦ ⎢⎣ −7⎥⎦ ⎢⎣0 ⎥⎦
which are different.

Matrices ■ 1.67
⎡0 5 6⎤
Note If B = ⎢ 5 0 7⎥⎥ with the eigen vectors as rows, then
⎢
⎢⎣6 −7 0 ⎥⎦
B = 0 − 5(0 − 42) + 6( −35 − 0) = 210 − 210 = 0
∴ the vectors X1, X2, X3 are linearly dependent. However, any two of them are linearly independent.
Geometrically, it means that all the vectors are coplanar, but any two of them are non-collinear.
In this example we have seen −1 is the only eigen value of 3 × 3 matrix and two linearly
independent eigen vectors.
1.6.4 Properties of Eigen Values

1. A square matrix A and its transpose AT have the same eigen values.
Proof Eigen values of A are the roots of its characteristic equation
A − lI = 0 (1)
We know ( A − l I)T = A T − (l I)T [{ ( A + B)T = A T + BT ]
= A T − l IT = A T − l I [{ I T = I]
∴ ( A − l I)T = A T − l I (2)
For any square matrix B, BT = B

∴ ( A − l I)T = A − l I (3)
From (2) and (3), A − lI = AT − lI .

This shows that the characteristic polynomial of A and AT are the same.
Hence, the characteristic equations of A and AT is (1).
∴ A and AT have the same eigen values. ■
2. Sum of the eigen values of a square matrix A is equal to the sum of the elements on its main
diagonal.
Proof Let A be a square matrix of order n.
Then the characteristic equation of A is A − lI = 0
⇒ l n − S1l n −1 + S2 l n − 2 − ... + ( −1) n Sn = 0 (1)
where S1 = sum of the diagonal elements of A

If l1, l2, …, ln are the roots of (1), then l1, l2, …, ln are the eigen values of A.
From theory of equations,
−coefficient of l n −1
sum of the roots of (1) is =
coefficient of l n
⇒ l1+ l2 + … + ln = −(−S1) = S1
∴ the sum of the eigen values = l1+ l2 + … + ln = S1
= sum of the diagonal elements of the matrix A. ■

Note Sum of the diagonal elements of A is called the trace of A.

∴ Sum of the eigen values = trace of A
3. Product of the eigen values of a square matrix A is equal to A .
Proof Let A be a square matrix of order n.

Then its characteristic equation is A − lI = 0
⇒ l n − S1l n −1 + S2 l n − 2 − ... + ( −1) n Sn = 0 (1)
where Sn = A .
If l1, l2, …, ln are the n roots of (1), then from theory of equations,
constant term
the product of roots = ( −1) n
coefficient of l n
⇒ l1 l 2 … l n = ( −1) n ( −1) n Sn = ( −1) 2 n Sn = Sn = A [{ (−1)2n = 1]
∴ the product of the eigen values = l1 l 2 … l n = Sn = A . ■
Note If at least one eigen value is 0, then A = 0 ∴ A is a singular matrix.

If all the eigen values are non-zero, then A ≠ 0
∴ A is a non-singular if all the eigen values are non-zero.
1 1 1
4. If l1, l2, …, ln are non-zero eigen values of square matrix of order n, then , , …, are
l1 l 2 ln
21
eigen values of A .
Proof Let l be any non-zero eigen value of A, then there exists a non-zero column matrix X such that
AX = lX. Since all the eigen values are non-zero, A is non-singular.
−1
∴ A exists.
−1 −1
∴ A (AX) = A (lX)
−1 −1
⇒ (A A)X = l(A X)
−1
⇒ IX = l(A X)
−1 1 −1 −1 1
⇒ X = l( A X ) ⇒ X = A X ⇒ A X = X. [{ l ≠ 0]
l l
1 −1
So, is an eigen value of A .
l
This is true for all the eigen values of A.
1 1 1 −1
∴ , , …, are the eigen values of A . ■
l1 l 2 ln
−1 1
Note that the eigen vector for A corresponding to is also X.
l

Matrices ■ 1.69
5. If l1, l2, ???, ln are the eigen values of A, then

(i) cl1, cl2, ???, cln are the eigen values of cA, where c ≠ 0
(ii) l1m , l 2m , …, l mn are the eigen values of Am, where m is a positive integer.
Proof
Let l be any eigen value of A, then there exists a non-zero column matrix X such that
AX = lX (1)
(i) Multiply by c ≠ 0 then c(AX) = c(lX) ⇒ (cA) X = (cl) X
∴ cl is an eigen value of cA.
This is true for all eigen values of A.
∴ cl1, cl2, …, cln are the eigen values of cA.
(ii) Now A2X = A(AX) = A(lX) = l(AX) = l(lX) = l2X [using (1)]
∴ AX=lX
2 2
⇒ l is an eigen value of A .
2 2
Similarly, A X = A(A X) = A(l2X) = l2(AX) = l2(lX) = l3X

3 2
A3X = l3X ⇒ l3 is an eigen value of A3.

Proceeding in this way, we have AmX = l mX for any positive integer m.
This is true for all eigen values.
∴ l1m , l 2m , …, l mn are the eigen values of Am. ■
6. If l1, l2, …, ln are the eigen values of A, then

(i) l1 2 K, l2 2 K, …, ln 2 K are the eigen values of A 2 KI.
(ii) a 0 l1 1 a1 l1 1 a 2 , a 0 l 2 1 a1 l 2 1 a 2 , …, a 0 l n 1 a1 l n 1 a 2 are the eigen values of

2 2 2
a 0 A 2 1 a1 A 1 a 2 I.
Proof
Let l be any eigen value of A. Then AX = lX (1)
where X ≠ 0 is a column matrix.
∴ AX − KX = lX − KX
⇒ (A − KI)X = (l − K)X
∴ l − K is an eigen value of A − KI.
∴ l1 − K, l2 − K, …, ln − K are the eigen values of A − KI.
(ii) We have AX = lX and A2X = l2X.
∴ a0 (A2X) = a0 (l2X) and a1 (AX) = a1 (lX)
∴ a0 (A2X) + a1 (AX) = a0 (l2X) + a1 (lX)
Adding a2X on both sides, we get
a0(A2X) + a1(AX) + a2X = a0(l2X) + a1(lX) + a2X
⇒ (a0A2 + a1A + a2I)X = (a0l2 + a1l + a2)X
This means a0l2 + a1l + a2 is an eigen value of a0A2 + a1A + a2I.
∴ a 0 l 12 + a1l 1+ a 2 , a 0 l 22 + a1l 2 + a 2 , …, a 0 l 2n + a1l n + a 2 are the eigen values of
a 0 A 2 + a1A + a 2 I. ■

Note
⎡1 0 0 ⎤
⎢ ⎥
1. The eigen values of the unit matrix ⎢0 1 0 ⎥ are 1, 1, 1 and the corresponding eigen vectors are
⎡1 ⎤ ⎡0 ⎤ ⎡0 ⎤ ⎢⎣0 0 1 ⎥⎦
⎢0 ⎥ , ⎢1 ⎥ , ⎢0 ⎥ ,
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ which are independent.
⎢⎣0 ⎥⎦ ⎢⎣0 ⎥⎦ ⎢⎣1 ⎥⎦
⎡l1 a12 a13 ⎤
⎢ ⎥
2. The eigen values of a triangular matrix ⎢ 0 l 2 a23 ⎥ are the main diagonal elements l1, l2, l3.
⎢⎣ 0 0 l 3 ⎥⎦
3. If l is an eigen value of A then AX = lX. We have seen
A2X = l2X, …, AmX = lmX.
Thus, the eigen values of A, A2, …, Am are l, l2, …, lm which are all different. But they all have
the same eigen vector X.
Similarly, l and a0l2 + a1l + a2 are eigen values of A and
a0A2 + a1A + a2I. But they have the same eigen vector X.
WORKED EXAMPLES
EXAMPLE 1
⎡ 1 2 −2 ⎤
⎢ 3 ⎥⎥ .
Find the sum and product of the eigen values of the matrix ⎢ 1 0
⎢⎣ −2 −1 −3 ⎥⎦
Solution.
⎡ 1 2 −2 ⎤
Let A = ⎢⎢ 1 0 3 ⎥⎥
⎢⎣ −2 −1 −3 ⎥⎦
Sum of the eigen values = Sum of the elements on the main diagonal
= 1 + 0 + (−3) = −2
1 2 −2
Product of the eigen values = A = 1 0 3
−2 −1 −3
= 1(0 + 3) − 2 (−3 + 6) − 2(−1 − 0) = 3 − 6 + 2 = −1
EXAMPLE 2
⎡ 3 10 5⎤
⎢ ⎥ −1
If 2 and 3 are eigen values of A 5 ⎢22 23 24 ⎥ , find the eigen values of A and A3.
⎢⎣ 3 5 7 ⎥⎦
Solution.
⎡ 3 10 5⎤
Given A = ⎢ −2 −3 − 4 ⎥
⎢ ⎥
⎢⎣ 3 5 7⎥⎦

Matrices ■ 1.71
Also given 2 and 3 are two eigen values of A.

Let l be the 3rd eigen value.
We know, sum of the eigen values = sum of the diagonal elements.
⇒ 2 + 3 + l = 3 + (−3) + 7 ⇒ l = 2
So, eigen values of A are 2, 2, 3
−1 1 1 1
∴ the eigen values of A are , , and the eigen values of A3 are 23, 23, 33 ⇒ 8, 8, 27.
2 2 3
EXAMPLE 3
⎡4 ⎤ ⎡5 4 ⎤
If ⎢ ⎥ is an eigen vector of the matrix ⎢ ⎥ , then find the corresponding eigen value.
⎣1 ⎦ ⎣1 2⎦
Solution.
⎡5 4 ⎤ ⎡4⎤
Let A = ⎢ ⎥ and X = ⎢ ⎥ . If l is the eigen value corresponding to an eigen vector X, then
⎣1 2 ⎦ ⎣1 ⎦
(A − lI)X = 0
⎡5 − l 4 ⎤ ⎡ 4⎤ ⎡0⎤
⇒ ⎢ 1 = ⇒ (5 − l ) ⋅ 4 + 4 = 0 ⇒ l = 6
⎣ 2 − l ⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣0 ⎥⎦
EXAMPLE 4
⎡3 1 4⎤
If A 5 ⎢0 2 6 ⎥ , then find the eigen values of A2 2 2A + I.
⎢⎣0 0 5 ⎥⎦
Solution. ⎡3 1 4⎤
Given A = ⎢0 2 6 ⎥
⎢⎣0 0 5 ⎥⎦
Since A is a triangular matrix, the eigen values are the diagonal elements.
∴ 3, 2, 5 are the eigen values of A.
∴ the eigen values of A − 2A + I are 32 − 2 ⋅ 3 + 1, 22 − 2 ⋅ 2 + 1, 52 − 2 ⋅ 5 + 1
2
i.e., the eigen values of A2 − 2A + I are 4, 1, 16.
EXAMPLE 5
⎡ 6 22 2⎤
The product of two eigen values of the matrix A 5 ⎢22 3 21⎥ is 16. Find the third eigen
value. ⎢⎣ 2 21 3 ⎥⎦
Solution.
Let l1, l2, l3 be the eigen values of A.
Given l1 ⋅ l2 = 16
We know that l1 ⋅ l2 ⋅ l3 = A
6 −2 2
⇒ 16l 3 = −2 3 −1 = 6(9 − 1) + 2( −6 + 2) + 2( 2 − 6)
2 −1 3
⇒ 16l 3 = 48 − 8 − 8 = 32 ⇒ l 3 = 2

EXAMPLE 6
Find the eigen values of the matrix ⎡ 1 22 ⎤ . Hence, find the matrix whose eigen values are
1 ⎢⎣25 4 ⎥⎦
and 21.
6
Solution.
1 −2⎤
Let A = ⎡⎢ .
⎣ −5 4 ⎥⎦
1− l −2
⇒ = 0 ⇒ l 2 − S1l + S2 = 0
−5 4 − l
where S1 = 1 + 4 = 5 and S2 = A = 4 − 10 = −6
∴ the characteristic equation is l2 − 5l − 6 = 0 ⇒ (l − 6) (l + 1) = 0 ⇒ l = 6, −1
1 −1
Since 6, −1 are the eigen values of A, by property (4), , − 1 are the eigen values of A .
6
T
−1 1 1 ⎡ 4 5⎤ 1 ⎡ 4 2⎤
So, the required matrix is A = adj A = − ⎢ =− ⎢
A 6 ⎣ 2 1⎥⎦ 6 ⎣ 5 1 ⎥⎦
EXAMPLE 7
⎡ 3 21⎤
If a, b are the eigen values of ⎢ form the matrix whose eigen values are a3, b3.
⎣21 5 ⎥⎦
Solution.
⎡ 3 −1⎤
Let A = ⎢ ⎥.
⎣ −1 5⎦
Since a, b are the eigen values of A, by property 5(ii), a3, b 3 are the eigen values of A3.
⎡ 3 −1⎤ ⎡ 3 −1⎤ ⎡10 −8⎤

Now A2 = A ⋅ A = ⎢ ⎥ ⎢ ⎥=⎢ ⎥
⎣ −1 5⎦ ⎣ −1 5⎦ ⎣ −8 26 ⎦
⎡ 10 −8⎤ ⎡ 3 −1⎤ ⎡ 38 −50 ⎤
and A3 = A 2 A = ⎢ ⎥ ⎢ ⎥=⎢ ⎥
⎣ −8 26 ⎦ ⎣ −1 5⎦ ⎣ −50 138⎦
EXERCISE 1.5
Find eigen values and eigen vectors of the following matrices.
⎡ 4 −20 −10 ⎤ ⎡ 7 −2 0 ⎤
⎡ 2 0⎤ ⎢ ⎡ −2 −3⎤
4 ⎥⎥
2 2
1. ⎢ 2 1⎥ 2. ⎢ −2 10 3. ⎢ 2 −6 ⎥ 4. ⎢ −2 6 −2⎥
1 1 ⎢ ⎥
⎢⎣ −7 2 −3⎥⎦ ⎢⎣ 6 −30 −13⎥⎦ ⎢⎣ −1 −2 0 ⎥⎦ ⎢⎣ 0 −2 5⎥⎦
⎡1 1 3⎤ ⎡ 3 10 5⎤
⎡ 6 −2 2⎤ ⎢ ⎥ ⎡ 2 −1 1⎤ ⎢ ⎥
5. ⎢ −2 3 −1⎥ 6. ⎢1 5 1⎥ 7. ⎢ −1 2 −1⎥ 8. ⎢ −2 −3 −4 ⎥
⎢⎣ 2 −1 3⎥⎦ ⎢⎣3 1 1⎥⎦ ⎢⎣ 1 −1 2⎥⎦ ⎢⎣ 3 5 7⎥⎦

Matrices ■ 1.73
⎡ −5 −5 −9⎤ ⎡ 2 1 1⎤
9. ⎢⎢ 8 9 18⎥⎥ 10. ⎢⎢1 2 1⎥⎥
⎢⎣ −2 −3 −7⎥⎦ ⎢⎣0 0 1⎥⎦
ANSWERS TO EXERCISE 1.5

⎡ 2⎤ ⎡ 2⎤ ⎡ 1⎤ ⎡ 5 ⎤ ⎡ 2⎤ ⎡ 0 ⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
1. l = 1, 3, −4; eigen vectors ⎢⎢ −1⎥⎥ , ⎢⎢ 1⎥⎥ , ⎢⎢ −3⎥⎥ 2. l = 0, −1, 2; eigen vectors ⎢1 ⎥ , ⎢0 ⎥ , ⎢ 1⎥
⎣⎢ −4 ⎥⎦ ⎢⎣ −2⎥⎦ ⎢⎣ 13⎥⎦ ⎢⎣0 ⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣ −2⎥⎦
⎡ 2⎤ ⎡ 3⎤ ⎡ −1⎤ ⎡ 1 ⎤ ⎡ 2⎤ ⎡ 2⎤
3. l = −3, −3, 5; eigen vectors ⎢ −1⎥ , ⎢0 ⎥ , ⎢ −2⎥ 4. l = 3, 6, 9; eigen vectors ⎢ 2⎥ , ⎢ 1⎥ , ⎢ −2⎥
⎢⎣ 0 ⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣ 1⎥⎦ ⎢⎣ 2⎥⎦ ⎢⎣ −2⎥⎦ ⎢⎣ 1⎥⎦
⎡ 1⎤ ⎡1 ⎤ ⎡ 2⎤ ⎡ −1⎤ ⎡ 1⎤ ⎡1 ⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
5. l = 2, 2, 8; eigen vectors ⎢ 0 ⎥ , ⎢ 2⎥ , ⎢ −1⎥ 6. l = −2, 3, 6; eigen vectors ⎢⎢ 0 ⎥⎥ , ⎢⎢ −1⎥⎥ , ⎢⎢ 2⎥⎥
⎢⎣ −2⎥⎦ ⎢⎣0 ⎥⎦ ⎢⎣ 1⎥⎦ ⎢⎣ 1⎥⎦ ⎢⎣ 1⎥⎦ ⎢⎣1 ⎥⎦
⎡1 ⎤ ⎡0 ⎤ ⎡ 1⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
7. l = 1, 1, 4; eigen vectors ⎢1 ⎥ , ⎢1 ⎥ , ⎢ −1⎥
⎢⎣0 ⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣ 1⎥⎦
⎡ 1⎤ ⎡ 5⎤
8. l = 3, 2, 2; eigen vectors X1 = ⎢⎢ 1⎥⎥ and X 2 = X 3 = ⎢⎢ 2⎥⎥
⎢⎣ −2⎥⎦ ⎢⎣ −5⎥⎦
⎡ 3⎤
⎢ ⎥
9. l = −1, −1, −1; eigen vectors X1 = X2 = X3 ⎢ −6 ⎥
=
⎣⎢ 2⎦⎥
⎡ 1⎤ ⎡ 0⎤ ⎡1 ⎤
10. l = 1, 1, 3; eigen vectors X1 = ⎢ 0 ⎥ , X 2 = ⎢ 1⎥ , X 3 = ⎢1 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ −1⎥⎦ ⎢⎣ −1⎥⎦ ⎢⎣0 ⎥⎦
1.6.5 Cayley-Hamilton Theorem

Theorem 1.4 Every square matrix satisfies its characteristic equation
Proof
Let A = [ aij ]n × n be a square matrix of order n.
Then the Characteristic polynomial is
a11 − l a12 a13 … a1n
a21 a22 − l a23 … a2 n
A − lI =
:
an1 an 2 an3 … ann − l

The coefficient of ln is (−1)n from the product of (a11 − l) (a22 − l)... (ann − l).
∴ let A − lI = ( −1) n ⎡⎣l n + a1l n −1 + a2 l n − 2 + … + an ⎤⎦ (1)
Since the elements of A − lI are at most first degree in l, the elements of adj (A − lI) are ordinary
polynomials in l of degree at most (n − 1).
∴ adj (A − lI) can be written as a matrix polynomial in l of degree n − 1.
Let adj( A − lI ) = B 0 l n −1 + B1l n − 2 + B 2 l n − 3 + … B n n −1 (2)
where B0, B1,…, Bn − 1 are n × n matrices.

The elements of these matrices are function of aij.
We know that if A is a n × n matrix, then A (adj A ) = A I = (adj A) A
where I is n × n identity matrix.
∴ we have ( A − lI )adj( A − lI ) = A − lI = I
Substituting from (1) and (2), we have
( A − lI )[B 0 l n −1 + B1l n − 2 + B 2 l n − 3 + … + B n − 2 l + B n −1 ]
= ( −1) n (l n + a1l n −1 + a2 l n − 2 + …an −1l + an )
Equating the coefficients of lm, ln − 1, ln − 2… l and term independent of l, we get

−IB0 = (−1)n, AB0 − IB0 = (−1)n a1I,
AB1 − IB2 = (−1)n a2I, … , ABn − 2 − IBn − 1 = (−1) an − 1 I, ABn − 1= (−1)n anI
Pre-multiplying the above equation by An, An − 1, An − 2, …, A, I, we get

− An B0 = ( −1) n An
An B0 − An −1 B1 = ( −1) n a1 An −1
An −1 B1 − An − 2 B2 = ( −1) n a2 An − 2
:
A Bn − 2 − ABn −1 = ( −1) n an −1 A
2
ABn −1 = ( −1) n an I
Adding we get,
( −1) n [A n + a1A n −1 + a2 A n − 2 + ... an I ] = 0
⇒ A n + a1A n −1 + a2 A n − 2 + ... an I = 0
This means A satisfies the equation l n + a1l n −1 + a2 l n − 2 + … + an = 0 ,
which is the characteristic equation of A.
Hence, the theorem.
Properties: Cayley–Hamilton Theorem has the following two important properties:
1. To find the inverse of a non-singular matrix A
2. To find higher integral power of A

Matrices ■ 1.75
WORKED EXAMPLES
EXAMPLE 1
⎡1 2⎤
Verify that A 5 ⎢ 4
⎥ satisfies its characteristic equation and hence find A .
⎣ 2 21⎦
Solution.
⎡ 1 2⎤
Given A=⎢ ⎥
⎣ 2 −1⎦
The characteristic equation of A is A − lI = 0 ⇒ l 2 − S1l + S2 = 0
where S1 = 1 + ( −1) = 0 and S2 = A = −1 − 4 = −5
∴ the characteristic equation is l2 − 5 = 0 (1)

By Cayley-Hamilton theorem, A satisfies (1). That is A − 5I = 0 2
(2)
We shall now verify this by direct computation.
⎡1 2 ⎤ ⎡1 2 ⎤ ⎡ 5 0 ⎤
A2 = A ⋅ A = ⎢ ⎥⎢ ⎥=⎢ ⎥
⎣ 2 −1⎦ ⎣ 2 −1⎦ ⎣0 5⎦
⎡1 0 ⎤
∴ I=⎢ ⎥
⎣0 1 ⎦
⎡5 0⎤ ⎡1 0⎤ ⎡ 5 0⎤ ⎡ 5 0⎤ ⎡0 0⎤
A 2 − 5I = ⎢ ⎥ − 5⎢ ⎥=⎢ ⎥−⎢ ⎥=⎢ ⎥
⎣0 5⎦ ⎣ 0 1 ⎦ ⎣ 0 5⎦ ⎣ 0 5⎦ ⎣ 0 0 ⎦
2
⇒ A − 5I = 0.
Hence, A satisfies its characteristic equation.
2
To find A4: We have A = 5I [from (2)]
⎡ 5 0 ⎤ ⎡ 25 0 ⎤
∴ A4 = 5A2 = 5 ⎢ ⎥=⎢ ⎥.
⎣0 5⎦ ⎣ 0 25⎦
EXAMPLE 2
⎡1 4 ⎤
Verify Cayley-Hamilton theorem for the matrix A 5 ⎢ ⎥ and find its inverse. Also express
⎣2 3 ⎦
A5 2 4A4 2 7A3 1 11A2 2 A 2 10I as a linear polynomial in A.
Solution.
⎡1 4 ⎤
Given A=⎢ ⎥
⎣ 2 3⎦
The characteristic equation of A is A − lI = 0 ⇒ l 2 − S1l + S2 = 0
where S1 = 1 + 3 = 4, S2 = A = 3 − 8 = −5

∴ the characteristic equation is l2 − 4l − 5 = 0 (1)

By Cayley-Hamilton theorem, A satisfies (1).
∴ A2 − 4A − 5I = 0 (2)
We shall now verify this by direct computations.
⎡1 4 ⎤ ⎡1 4 ⎤ ⎡ 9 16 ⎤
A2 = A ⋅ A = ⎢ ⎥⎢ ⎥=⎢ ⎥
⎣ 2 3 ⎦ ⎣ 2 3⎦ ⎣8 17⎦
⎡9 16 ⎤ ⎡1 4 ⎤ ⎡1 0 ⎤
A 2 − 4A − 5I = ⎢ ⎥ −4⎢ ⎥ −5⎢ ⎥
⎣8 17⎦ ⎣ 2 3⎦ ⎣0 1 ⎦
⎡9 16 ⎤ ⎡ 4 16 ⎤ ⎡ 5 0 ⎤ ⎡9 − 4 − 5 16 − 16 − 0 ⎤ ⎡0 0 ⎤
=⎢ ⎥−⎢ ⎥−⎢ ⎥=⎢ ⎥=⎢ ⎥
⎣8 17⎦ ⎣ 8 12⎦ ⎣ 0 5⎦ ⎣ 8 − 8 − 0 17 − 12 − 5 ⎦ ⎣0 0 ⎦
⇒ A2 − 4A − 5I = 0.
Hence, the theorem is verified.
21
To find A : We have 5I = A2 − 4A
−1 −1 −1 −1
Multiply by A , we get 5A = A A2 − 4 A A
⎡1 4 ⎤ ⎡1 0⎤
5 A−1 = A − 4I = ⎢ ⎥ − 4 ⎢0 1⎥
⎣ 2 3 ⎦ ⎣ ⎦
⎡1 4 ⎤ ⎡ 4 0 ⎤ ⎡1 − 4 4 − 0 ⎤ ⎡ −3 4 ⎤
=⎢ ⎥−⎢ ⎥=⎢ ⎥=⎢ ⎥
⎣ 2 3 ⎦ ⎣ 0 4 ⎦ ⎣ 2 − 0 3 − 4 ⎦ ⎣ 2 −1⎦
1 ⎡ −3 4 ⎤
∴ A −1 =
5 ⎢⎣ 2 −1⎥⎦
Finally, to find A5 − 4A4 − 7A3 + 11A2 − A − 10I:
Consider the polynomial l5 − 4l4 − 7l3 + 11l2 − l − 10 (3)
Divide the polynomial (3) by l2 − 4l − 5.
Division is indicated below.
l 3 − 2l + 3
l 2 − 4l − 5 l 5 − 4l 4 − 7l 3 + 11l 2 − l − 10
l 5 − 4l 4 − 5l 3
− 2l 3 + 11l 2 − l
−2l 3 + 8l 2 + 10l
3l 2 − 11l − 10
3l 2 − 12l − 15
l+5
∴ We get the quotient l3 − 2l + 3 and remainder l + 5.

Matrices ■ 1.77
∴ l5 − 4l4 − 7l3 + 11l2 − l − 10 = (l2 − 4l − 5)(l3 − 2l + 3) + l + 5

Replace l by A, we get
A5 − 4A4 − 7A3 + 11A2 − A − 10I = (A2 − 4A − 5I) (A3 − 2A + 3I) + A + 5I
= 0 + A + 5I = A + 5I [using (2)]
which is a linear polynomial in A.
EXAMPLE 3
⎡ 2 21 1⎤
⎢
Find the characteristic equation of the matrix A given A 5 ⎢21 2 21⎥⎥ . Hence, find A21 and A4.
⎣⎢ 1 21 2 ⎥⎦
Solution.
⎡ 2 −1 1⎤
Given A = ⎢⎢ −1 2 −1⎥⎥
⎢⎣ 1 −1 2⎥⎦
2−l −1 1
⇒ −1 2 − l −1 = 0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
1 −1 2 − l
where S1 = sum of the diagonal elements of A = 2 + 2 + 2 = 6

S2 = sum of the minors of the diagonal elements of A
2 −1 2 1 2 −1
= + + = 4 −1+ 4 −1+ 4 −1 = 9
−1 2 1 2 −1 2
S3 = A = 2( 4 − 1) + ( −2 + 1) + (1 − 2) = 6 − 1 − 1 = 4

By Cayley-Hamilton theorem, A satisfies its characteristic equation
∴ A3 − 6A2 + 9A − 4I = 0 (1)
⇒ 4I = A3 − 6A2 + 9A
Multiply by A−1, 4IA−1 = A3A−1 − 6A2 ⋅ A−1 + 9A A−1
⇒ 4A−1 = A2 − 6A + 9I
⎡ 2 −1 1⎤ ⎡ 2 −1 1⎤
But A 2 = ⎢⎢ −1 2 −1⎥⎥ ⎢⎢ −1 2 −1⎥⎥
⎢⎣ 1 −1 2⎥⎦ ⎢⎣ 1 −1 2⎥⎦
⎡ 4 + 1 + 1 −2 − 2 − 1 2 + 1 + 2⎤ ⎡ 6 −5 5⎤
⎢
= ⎢ −2 − 2 − 1 1 + 4 + 1 −1 − 2 − 2⎥⎥ = ⎢⎢ −55 6 −5⎥⎥
⎣⎢ 2 + 1 + 2 −1 − 2 − 2 1 + 1 + 4 ⎥⎦ ⎢⎣ 5 −5 6 ⎥⎦

⎡ 6 −5 5⎤ ⎡ 2 −1 1⎤ ⎡1 0 0⎤
4 A = ⎢ −5 6 −5⎥ − 6 ⎢ −1 2 −1⎥ + 9 ⎢⎢ 0 1 0 ⎥⎥
⎢ ⎥ ⎢ ⎥
−1
∴
⎢⎣ 5 −5 6 ⎥⎦ ⎢⎣ 1 −1 2⎥⎦ ⎢⎣ 0 0 1 ⎥⎦
⎡6 − 12 + 9 −5 + 6 5 − 6 ⎤ ⎡ 3 1 −1⎤
⎢
= ⎢ −5 + 6 6 − 12 + 9 −5 + 6 ⎥⎥ = ⎢⎢ 1 3 1⎥⎥
⎢⎣ 5−6 −5 + 6 6 − 12 + 9⎥⎦ ⎢⎣ −1 1 3⎥⎦
⎡ 3 1 −1⎤
A = ⎢ 1 3 1⎥⎥
1⎢
−1
∴
4
⎢⎣ −1 1 3⎥⎦
(1) ⇒ A 3 = 6 A 2 − 9A + 4I
∴ A 4 = 6 A 3 − 9A 2 + 4 A [Multiplying by A]
= 6[6 A − 9A + 4I] − 9A + 4A
2 2
= 27A 2 − 50A + 24I

⎡ 6 −5 5⎤ ⎡ 2 −1 1⎤ ⎡1 0 0⎤
⎢ ⎥ ⎢ ⎥
= 27 ⎢ −5 6 −5⎥ − 50 ⎢ −1 2 −1⎥ + 24 ⎢⎢ 0 1 0 ⎥⎥
⎢⎣ 5 −5 6 ⎥⎦ ⎢⎣ 1 −1 2⎥⎦ ⎢⎣ 0 0 1 ⎥⎦
⎡162 − 100 + 24 −135 + 50 135 − 50 ⎤ ⎡ 86 −85 85⎤
= ⎢⎢ −135 + 50 162 − 100 + 24 −135 + 50 ⎥⎥ = ⎢⎢ −85 86 −85⎥⎥
⎢⎣ 135 − 50 −135 + 50 162 − 100 + 24 ⎥⎦ ⎢⎣ 85 −85 86 ⎥⎦
EXAMPLE 4
Use Cayley-Hamilton theorem to find the matrix
⎡2 1 1 ⎤
A 2 5A 1 7A 2 3A 1 8A 2 5A 1 8A 2 2A 1 I if the matrix A 5 ⎢0 1 0 ⎥⎥ .
8 7 6 5 4 3 2 ⎢
⎢⎣1 1 2 ⎥⎦
Solution.
⎡2 1 1⎤
Given A = ⎢⎢0 1 0 ⎥⎥
⎢⎣1 1 2⎥⎦
The characteristic equation is A − lI = 0
⇒ 2−l 1 1
0 1− l 0 = 0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
1 1 2−l
where S1 = 2 + 1 + 2 = 5

Matrices ■ 1.79
1 0 2 1 2 1
S2 = + + = 2 + 4 −1+ 2 = 7
1 2 1 2 0 1
S3 = A = 2 ⋅ 2 − 1⋅ 0 + 1( −1) = 4 − 1 = 3

By Cayley-Hamilton theorem, we get A3 − 5A2 + 7A − 3I = 0 (1)
We have to find the matrix
A8 − 5A7 + 7A6 − 3A5 + 8A4 − 5A3 + 8A2 − 2A + I = f(A), say
We shall rewrite this matrix polynomial in terms of
A3 − 5A2 + 7A − 3I
∴ the polynomial
f(A) = A5(A3 − 5A2 + 7A − 3I) + 8A4 − 5A3 + 8A2 − 2A + I

= 8A4 − 5A3 + 8A2 − 2A + I [Using (1)]
= 8A(A3 − 5A2 + 7A − 3I) + 35A3 − 48A2 + 22A + I
= 35A3 − 48A2 + 22A + I [Using (1)]
= 35(A3 − 5A2 + 7A − 3I) + 127A2 − 223A + 106I
= 127A2 − 223A + 106I [Using (1)]
⎡2 1 1⎤ ⎡2 1 1⎤ ⎡5 4 4⎤
But A = ⎢⎢0 1 0 ⎥⎥ ⎢⎢0 1 0 ⎥⎥ = ⎢⎢ 0 1 0 ⎥⎥
2
⎣⎢1 1 2⎦⎥ ⎣⎢1 1 2⎦⎥ ⎣⎢ 4 4 5 ⎦⎥

⎡5 4 4⎤ ⎡2 1 1⎤ ⎡1 0 0⎤
∴ f ( A ) = 127 ⎢ 0 1 0 ⎥ − 223 ⎢0 1 0 ⎥ + 106 ⎢⎢0 1 0 ⎥⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ 4 4 5 ⎥⎦ ⎢⎣1 1 2⎥⎦ ⎢⎣0 0 1 ⎥⎦
⎡635 − 446 + 106 508 − 223 508 − 223 ⎤ ⎡ 295 285 285⎤
=⎢⎢ 0 127 − 223 + 106 0 ⎥ = ⎢ 0 10 0 ⎥⎥
⎥ ⎢
⎢⎣ 508 − 223 508 − 223 635 − 446 + 106 ⎥⎦ ⎢⎣ 285 285 295⎥⎦
Note: Otherwise divide l8 − 5l7 + 7l6 − 3l5+ 8l4− 5l3 + 8l2 − 2l + 1 by l3 − 5l2 + 7l − 3 and
proceed as in example 2.
EXAMPLE 5
⎡1 0 0 ⎤
If A 5 ⎢⎢1 0 1 ⎥⎥ , then show that An 5 An 2 2 1 A2 2 I for n $ 3. Hence, find A50.
⎢⎣0 1 0 ⎥⎦
Solution.
⎡1 0 0 ⎤
Given A = ⎢⎢1 0 1 ⎥⎥ .
⎢⎣0 1 0 ⎥⎦


⇒ l 3 − S1l 2 + S2 l − S3 = 0
0 1 1 0 1 0
where S1 = 1, S2 = + + = −1 + 0 + 0 = −1
1 0 0 0 1 0
and S3 = A = −1
∴ the characteristic equation is l − l − l + 1 = 0
3 2
(1)
By Cayley-Hamilton theorem, A statisfies (1)
∴ A3 − A2 − A + I = 0
⇒ A3 − A2 = A − I (2)
Multiplying (2) by A, A2, …, An − 3, we get the equations
A 4 − A3 = A 2 − A
A5 − A 4 = A3 − A 2
A6 − A5 = A 4 − A3
: :
A n − A n −1 = A n − 2 − A n − 3
Adding all these equations, we get
An − A2 = An−2 − I
⇒ A n = A n − 2 + ( A 2 − I) for all n ≥ 3 (i)
n−2 n−4
A =A +A −I
2
n−4 n −6
A =A + A2 − I
: : :
∴ A n = ( A n − 4 + A 2 − I) + ( A 2 − I)
⇒ A n = A n − 4 + 2( A 2 − I) (ii)
n −6
=A + A − I + 2( A − I)
2 2
n −6
⇒ A =A n
+ 3( A 2 − I) (iii)
n −8
⇒ A =A n
+ 4( A − I)
2
(iv)
A A
⎛ n − 2⎞ 2
A n = A n −( n − 2) + ⎜ ( A − I)
⎝ 2 ⎟⎠
If n is even, then
⎡ 2 4
⎢observe the coefficients of A − I in (i), (ii), (iii) … and index of A. We see 2 = 1 in (i), 2 = 2 in
2
⎣
6 8 n−2 ⎤
(ii), = 3 in (iii), = 4 in (iv) and so on in the last one ⎥
2 2 2 ⎦

Matrices ■ 1.81
⎛ n − 2⎞ 2 n − 2 n 2 ⎛ n − 2⎞
∴ An = A2 + ⎜ A − ⇒ An = A −⎜
⎝ 2 ⎟⎠ ⎝ 2 ⎟⎠
I I
2 2
Putting n = 50, we get A50 = 25A2 − 24I
⎡1 0 0⎤ ⎡1 0 0 ⎤ ⎡1 0 0 ⎤
But A = A ⋅ A = ⎢⎢1 0 1 ⎥⎥
2 ⎢1 0 1 ⎥ = ⎢1 1 0 ⎥
⎢ ⎥ ⎢ ⎥
⎢⎣0 1 0 ⎥⎦ ⎢⎣0 1 0 ⎥⎦ ⎢⎣1 0 1 ⎥⎦
⎡1 0 0 ⎤ ⎡1 0 0⎤ ⎡ 1 0 0⎤
∴ A 50 ⎢
= 25 ⎢1 1 0 ⎥ − 24 ⎢⎢0 1 0 ⎥⎥ = ⎢⎢ 25 1 0 ⎥⎥
⎥
⎢⎣1 0 1 ⎥⎦ ⎢⎣0 0 1 ⎥⎦ ⎢⎣ 25 0 1 ⎥⎦
EXAMPLE 6
⎡1 2⎤
If A 5 ⎢ n
⎥ , then find A in terms of A and I.
⎣ 2 21⎦
Solution.
⎡ 1 2⎤
Given A=⎢ ⎥
⎣ 2 −1⎦
The characteristic equation is l2 − 5 = 0 [see example 17]

By Cayley-Hamiltons theorem A2 − 5I = 0 (1)
To find An, consider the polynomial ln
Dividing ln by l2 − 5, we get ln = (l2 − 5) f(l) + al + b (2)
where f(l) is the quotient and al + b is the remainder.
We shall now find the values of a and b.
The eigen values of A are l = 5, − 5
( 5) a 5 + b = ( 5)
n n
Substitute l = 5 in (2) then = 0+a 5+b ⇒ (3)
Substitute l = − 5 in (2), then ( − 5 ) = 0 + a ( − 5 ) + b

n
−a 5 + b = ( − 5 )
n
⇒ (4)
2b = ( 5 ) + ( − 5 )
n n
(3) + (4) ⇒
( 5 ) + (− 5 )
n n
(1 + ( −1) n )
= ( 5)
n
∴ b=
2 2
(3) − (4) ⇒ 2 5a = ( 5 ) − ( − 5 )
n n

( 5 ) − (− 5 )
n n
(1 − ( −1) n )
= ( 5)
n
∴ a=
2 5 2
Replacing l by A in (2), we get
A n = ( A 2 − 5I)f( A ) + aA + bI = 0 + aA + bI
n ⎛ 1 − ( −1) n ⎞ n ⎛ 1 + ( −1) n ⎞
∴ An = ( 5) ⎜ ⎟ A + ( 5) ⎜ ⎟⎠ I.
⎝ 2 ⎠ ⎝ 2
EXERCISE 1.6
Verify Cayley-Hamilton theorem for the following matrices and hence find their inverses.
⎡ 1 3 7⎤ ⎡1 0 3⎤
1. ⎢ 4 2 3⎥ 2. ⎢ 2 1 −1⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ 1 2 1⎥⎦ ⎢⎣ 1 −1 1⎥⎦
⎡ −1 0 3⎤ ⎡ 7 2 −2⎤
⎢ 8 1 −7⎥ ⎢ ⎥
3. ⎢ ⎥ 4. ⎢ −6 −1 2⎥
⎢⎣ −3 0 8⎥⎦ ⎢⎣ 6 2 −1⎥⎦
⎡ 2 −1 2⎤
5. Verify that the matrix A = ⎢⎢ −1 2 −1⎥⎥ satisfies its characteristic equation and hence find A4.
⎢⎣ 1 −1 2⎥⎦
⎡ 7 3⎤
6. A = ⎢ n
⎥ , find A in terms of A and I using Cayley-Hamilton theorem and hence find A .
3
⎣2 6⎦
⎡ 1 2 3⎤
7. Find A using Cayley-Hamilton theorem for the matrix A = ⎢⎢ 2 −1 4 ⎥⎥ . Find A4 + A3 − 18A2 − 39A + 2I
4
⎢⎣ 3 1 −1⎥⎦
8. Find the eigen values and eigen vectors of the system of equations 10x1 + 2x2 + x3 = lx1, 2x1 + 10x2 + x3 =
lx2, 2x1 + x2 + 10x3 = lx3
[Hint: Equations can be rewritten as (10 − l)x1 + 2x2 + x3 = 0, 2x1 + (10 − l)x2 + x3 = 0,
2x1 + x2 + (10 − l)x3 = 0
⎡10 2 1 ⎤ ⎡ x1 ⎤
If A = ⎢⎢ 2 10 1 ⎥⎥ and X = ⎢⎢ x 2 ⎥⎥ then these equations in matrix form is (A − lI)X = 0 and so A − lI = 0
⎢⎣ 2 1 10 ⎥⎦ ⎢⎣ x 3 ⎥⎦
is the characteristic equation of A and l = 8, 9, 13. Eigen vectors are given by (I)]
A
9. If l is an eigen value of a non-singular matrix A, show that
is an eigen value of the matrix adj A.
l
⎡ A ⎤
⎢Hint:
( AX = lX ⇒ (adj A) (AX) = (adj A) (lX) A X = l(adj A )X ⇒ (adj A )X =
l ⎥⎦
X
⎣

Matrices ■ 1.83
⎡ 1 2⎤
10. Verify Cayley-Hamilton theorem for the matrix A = ⎢ ⎥ and find its inverse and also find
⎣ −1 3⎦
A6 − 4A5 + 8A4 − 12A3 + 14A2.
⎡ 13 −3 5⎤
⎢ ⎥
11. Verify Cayley-Hamilton theorem and find the inverse of A = ⎢ 0 4 0 ⎥
⎢⎣ −15 9 −7⎥⎦

⎡ −4 11 −5⎤ ⎡0 3 3⎤ ⎡ 8 0 −3⎤
1⎢
3 2 −7⎥⎥ ⎢ −43 1 17⎥
1 ⎢
1. −1 −6 25⎥⎥ 2. 3.
35 ⎢ 9⎢ ⎢ ⎥
⎢⎣ 6 1 −10 ⎥⎦ ⎢⎣ 3 −1 −1⎥⎦ ⎢⎣ 3 0 −1⎥⎦
⎡ −3 −2 2⎤ ⎡ 124 −123 162⎤

1⎢ ⎢ 96 −123⎥⎥
5 −2⎥⎥ 5. A = ⎢ −95
4
4. 6
3⎢ ⎢⎣ 95 −95 124 ⎥⎦
⎢⎣ −6 −2 5⎥⎦
9n − 4 n ⎡ 7 3⎤ 9 ⋅ 4 n − 4 ⋅ 9 n ⎡1 0 ⎤ ⎡ 463 399⎤
6. A n = ⎢2 6⎥ + ⎢0 1 ⎥ ; A = ⎢ 266 330 ⎥
3
5 ⎣ ⎦ 5 ⎣ ⎦ ⎣ ⎦
⎡ 248 101 218⎤ ⎡ 3⎤ ⎡ 1⎤ ⎡1⎤

⎢ ⎥ ⎢ ⎥ ⎢ 1⎥ , ⎢1⎥
7. A 4 = ⎢ 272 109 50 ⎥ 8. ⎢ −2⎥ , ⎢ ⎥ ⎢⎥
⎢⎣104 98 204 ⎥⎦ ⎢⎣ 2⎥⎦ ⎢⎣ −3⎥⎦ ⎢⎣1⎥⎦
⎡ 7 −6 5⎤
1 ⎡3 −2⎤ ⎡ 1 −8⎤ 1 ⎢
0 ⎥⎥
−1 −1
10. A = ⎢
5 ⎣1 −1⎥⎦ ⎢⎣ 4 −7⎥⎦
, 11. A = 0 4
64 ⎢
⎢⎣15 18 −13⎥⎦
1.7 SIMILARITY TRANSFORMATION AND ORTHOGONAL TRANSFORMATION

1.7.1 Similar Matrices
Definition 1.35 Let A and B be square matrices of order n. A is said to be similar to B if there exists a
non-singular matrix P of order n such that
A = P−1BP (1)
The transformation (1) which transforms B into A is called a similarity transformation. The matrix P
is called a similarity matrix.
Note We shall now see that if A is similar to B then B is similar to A.
A = P−1BP ⇒ PA P−1 = B (Premultiplying by P and postmultiplying by P−1)
⇒ (P−1)−1 A(P−1) = B
⇒ Q−1AQ = B (2)
where Q = P−1

The relation (2) means B is similar to A. Thus, if A is similar to B, then B is similar to A.

Hence, we simply say A and B are similar matrices.
An important property of similarity transformations is that they preserve eigen values, which is
proved in the next theorem.
Theorem 1.5 Similar matrices have the same eigen values.
Proof Let A and B be two similar matrices of order n.
Then B = P−1 AP, by definition.
∴ the characteristic polynomial of B is B − lI
Now B − lI = P −1AP − lI = P −1AP − lP −1IP
⇒ = P −1 ( A − lI)P
= P −1 A − lI P [{ AB = A B ]
= A − lI P −1 P = A − lI P −1P = A − lI I
⇒ B − lI = A − lI [{ I = 1]
∴ A and B have the same characteristic polynomial and hence have the same characteristic equation.
So, A and B have the same eigen values.
Note Similar matrices A and B have the same determinant value i.e., A = B .
For B = P −1AP ⇒ B = P −1AP = P −1 A P = A P −1P = A I = A
1.7.2 Diagonalisation of a Square Matrix

Definition 1.36 A square matrix A is said to be diagonalisable if there exists a non-singular matrix P
such that P−1 AP = D, where D is a diagonal matrix. The matrix P is called a modal matrix of A.
The next theorem provides us with a method of diagonalisation.
Theorem 1.6 If A is a square matrix of order n, having n linearly independent eigen vectors and M
is the matrix whose columns are the eigen vectors of A, then M−1 AM = D, where D is the diagonal
matrix whose diagonal elements are the eigen values of A.
Proof Let X1, X2, …, Xn be n linearly independent eigen vectors of A corresponding to the eigen
values l1, l2, …, ln of A.
∴ AXi = liXi, i = 1, 2, 3, …, n.
Let M = [X1 X2 … Xn] be the matrix formed with the eigen vectors as columns.
Then AM = [AX1 AX2 AX3 … AXn]
= [l1X1 l2X2 l3X3 … lnXn]
⎡ l1 0 0 0⎤
⎢ 0 l 0 …⎥⎥
⎢ 2
= [ X1 X 2 … X n ] ⎢ 0 0 l 3 …⎥
⎢ ⎥
⎢: : : : ⎥
⎢ 0 0 0 l n ⎥⎦
⎣

Matrices ■ 1.85
⎡ l1 0 0⎤
⎢ 0 l 0 ⎥⎥
⇒ AM = MD, where D = ⎢ 2
⎢: : :⎥
⎢ ⎥
⎣ 0 … ln ⎦
⇒ M −1AM = D
The matrix M which diagonalises A is called the modal matrix of A and the resulting diagonal matrix
D whose elements are eigen values of A is called the spectral matrix of A.
1.7.3 Computation of the Powers of a Square Matrix

Diagonalisation of a square matrix A is very useful to find powers of A, say Ar.
By the theorem 1.6, D = M−1AM
∴ D 2 = ( M −1AM ) ( M −1AM )
= M −1A( MM −1 )AM = M −1AIAM = M −1A 2 M
Similarly, D3 = D 2 D
= ( M −1A 2 M ) ( M −1AM )
= M −1A 2 ( MM −1 )AM = M −1A 2 IAM = M −1A 3 M
Proceeding in this way, we can find
D r = M −1A r M
⎡l1r 0 … 0⎤
⎢ ⎥
0 l r
… 0⎥
∴ A r = MD r M −1 , where D r = ⎢ 2
⎢… … … …⎥
⎢ ⎥
⎢⎣ 0 0 … l rn ⎥⎦
Note
(1) If the eigen values l1, l2, …, ln of A are different then the corresponding eigen vectors X1,
X2, …, Xn are linearly independent by theorem 1.2.1 (2). So, A can be diagonalised.
(2) Even if 2 or more eigen values are equal, if we can find independent eigen vectors corresponding
to them (see worked example 6), then A can be diagonalised.
Thus, independence of eigen vectors is the condition for diagonalisation.
Working rule to diagonalise a n 3 n matrix A by similarity transformation:
Step 1: Find the eigen values l1, l2, …, ln
Step 2: Find linearly independent eigen vectors X1, X2, …, Xn
Step 3: Form the modal matrix M = [X1 X2 … Xn]
Step 4: Find M−1 and AM.
⎡ l1 0 0 … 0⎤
⎢ 0 l 0 … 0 ⎥⎥
Step 5: Compute M−1AM = D = ⎢ 2
⎢ : : : : :⎥
⎢ ⎥
⎣ 0 0 … … ln ⎦

1.7.4 Orthogonal Matrix

Definition 1.37 A real square matrix A is said to be an orthogonal matrix if AAT = ATA = I, where I
is the unit matrix of the same order as A.
From this definition it is clear that AT = A−1.
So, an orthogonal matrix is also defined as below.
Definition 1.38 A real square matrix A is orthogonal if AT = A−1.
EXAMPLE 1.25
⎡ cos u sin u ⎤
Prove that A 5 ⎢ ⎥ is orthogonal.
⎣2sin u cos u⎦
Solution.
⎡ cos u sin u ⎤
Given A=⎢ ⎥
⎣ − sin u cos u⎦
⎡ cos u sin u ⎤
∴ A =⎢ ⎥ = cos u + sin u = 1
2 2
∴ A is non-singular.
T
−1 adj A ⎡ cos u sin u ⎤
Hence, A = =⎢ ⎥ =A
T
[{ A = 1]
A ⎣ − sin u cos u ⎦
∴ A is orthogonal.
1.7.5 Properties of Orthogonal Matrix

1. If A is orthogonal, then AT is orthogonal.
Proof Given A is orthogonal. ∴ AAT = ATA = I
Reversing the roles of A and AT, we see ATA = AAT = I ⇒ AT is orthogonal.
Note Since AT = A−1, it follows A−1 is orthogonal.
2. If A is an orthogonal matrix, then A 561.

Proof Given A is orthogonal. Then AAT = 1
⇒ AA T = 1 ⇒ A AT = 1
But we know AT = A
A A =1 ⇒ A = 1 ⇒ A = ±1
2
∴
1
3. If l is an eigen value of an orthogonal matrix A, then is also an eigen value of A.
l
Proof Given A is orthogonal and l is an eigen value of A.

Matrices ■ 1.87
Then AT = A−1. By property (4) of eigen values,

1
is an eigen value of A−1 and so an eigen value of AT.
l
By property (1), A and AT have same eigen values.
1 1
∴ is an eigen of A and hence l, are eigen values of orthogonal matrix A.
l l
4. If A and B are orthogonal matrices, then AB is orthogonal.
Proof Given A and B are orthogonal matrices.
∴ AT = A−1 and BT = B−1
Now (AB)T = BT AT = B−1 A−1 = (AB)−1
∴ AB is orthogonal.
5. Eigen values of an orthogonal matrix are of magnitude 1.

Proof Let A be an orthogonal matrix and let l be an eigen value of A.
Then AX = lX, where X ≠ 0 (1)
Taking complex conjugate, we get A X = l X
But A is real matrix ∴ A=A
Hence, AX = l X
Taking transpose, ( AX )T = (l X )T
T T T −1 T
⇒ X AT = l X ⇒ X A =lX [{ A T = A −1 ] (2)
Multiplying (2) and (1) we get
T −1 T
( X A )( AX ) = (l X )(lX )
T −1 T
⇒ X ( A A )X = ll X X
T T T
X X= l X X [{ X X ≠ 0 as X ≠ 0]
2
⇒
1= l ⇒ l =1
2
⇒

Hence, eigen values of A are of absolute value 1. ■
1.7.6 Symmetric Matrix

Definition 1.39 Real Square Matrix
The matrix A = [aij]n × n is said to be symmetric if AT = A.
⎡ 1 −1 3⎤
⎡ 1 −1⎤ ⎢ ⎥
Example: ⎢
− ⎥ ⎢ −1 0 4 ⎥ are symmetric matrices.
,
⎣ 1 0 ⎦ ⎢
⎣ 3 4 2⎥⎦
Note that the elements equidistant from the main diagonal are the same.

1.7.7 Properties of Symmetric Matrices

1. Eigen values of a symmetric matrix are real.
Proof Let A be a symmetric matrix of order n and l be an eigen value of A.
Then there exists X ≠ 0 such that AX = lX (1)
Taking complex conjugate,
AX=lX
⇒ AX = l X [{ A is real A = A ] (2)
T T
Taking transpose, ( AX )T = (l X )T ⇒ ( X )T A T = l ( X )T ⇒ X A=lX [{ A T = A ]
Post multiplying by X,
T
( X A)X = (l X)T X
⇒
T T
X (AX ) = l (X X ) T
T T
[{ X X ≠ 0 as X ≠ 0]
⇒ X (lX ) = l ( X X )
T T
⇒ l( X X ) = l( X X )
⇒ l=l ∴ l iss real
This is true for all eigen values.
∴ eigen values of a symmetric matrix are real. ■
2. Eigen vectors corresponding to different eigen values of a symmetric matrix are orthogonal
vectors.
Proof Let A be a symmetric matrix of order n. ∴ AT = A.

Let l1, l2 be two different eigen values of A. Then l1, l2 are real, by property (1).
∴ there exist X1 ≠ 0, X2 ≠ 0 such that
AX1 = l1X1 (1)
and AX2 = l2X2 (2)

T
Premultiplying (1) by X , we get
2 X ( AX1 ) = X l1X1
T
2
T
2
⇒ X T2 AX1 = l1 ( X T2 X1 ) (3)
Premultiplying (2) by X1T , we get X1T (AX 2 ) = X1T l 2 X 2

⇒ X1T AX 2 = l 2 ( X1T X 2 ) (4)
Taking transpose of (3), we get ( X AX1 ) = (l1X X1 )
T
2
T T
2
T
⇒ X1T A T X 2 = l1X1T X 2
⇒ X1T AX 2 = l1 (X1T X 2 ) (5)
From (4) and (5) we get, l1 ( X X 2 ) = l 2 ( X X 2 )
T
1
T
1

Matrices ■ 1.89
⇒ (l1 − l 2 )( X1T X 2 ) = 0
Since l1 ≠ l 2 , l1 − l 2 ≠ 0, ∴ X1T X 2 = 0
⇒ X1 and X2 are orthogonal. ■
Remark: If X1 = (a1, b1, c1) and X2 = (a2, b2, c2) be two 3-dimensional vectors, they are orthogonal if
their dot product is 0 ⇒ a1a2 + b1b2 + c1c2 = 0
⎡ a1 ⎤ ⎡ a2 ⎤
If we treat them as column matrices, X1 = ⎢ b1 ⎥ X 2 = ⎢⎢ b2 ⎥⎥ , then the matrix product
⎢ ⎥
⎣⎢ c1 ⎦⎥ ⎣⎢ c2 ⎦⎥
⎡ a2 ⎤
X X 2 = [a1 b1 c1 ] ⎢⎢ b2 ⎥⎥ = a1a2 + b1b 2 + c1c 2
T
1
⎢⎣ c2 ⎥⎦
So, X1 and X2 are orthogonal if X1T X 2 = 0 or X T2 X1 = 0 .
Thus, we can treat column matrices as vectors and verify dot product = 0.
X1
2. The unit vector in X1 is and it is called a normalised vector.
a 1 b 21 c 2
2
Note For any square matrix eigen vectors corresponding to different eigen values are linearly
independent, but for a symmetric matrix, they are orthogonal, pairwise.
1.7.8 Diagonalisation by Orthogonal Transformation or Orthogonal Reduction

Definition 1.40 A Square Matrix
A is said to be orthogonally diagonalisable if there exists an orthogonal matrix N such that
N−1 AN = D ⇒ NT AN = D [ { NT = N−1]
This transformation which transforms A into D is called an orthogonal transformation.
The next theorem gives a method of orthogonal reduction.
Theorem 1.7 Let A be a symmetric matrix of order n. Let X1, X2, …, Xn be eigen vectors of A which are
pairwise orthogonal. Let N be the matrix formed with the normalised eigen vectors of A as columns.
Then N is an orthogonal matrix such that N−1 AN = D ⇒ NT AN = D.
N is called normalised modal matrix of A or normal modal matrix of A.
Working rule for orthogonal reduction of a n 3 n symmetric matrix.

Step 1: Find the eigen values l1, l2, …, ln
Step 2: Find the eigen vectors X1, X2, …, Xn which are pairwise orthogonal.
Step 3: Form the normalised modal matrix N with the normalised eigen vectors as columns.
Step 4: Find NT and AN.
⎡ l1 0 0⎤
⎢ 0 l 0 ⎥⎥
Step 5: Compute N T AN = D = ⎢ 2
⎢ : … …⎥
⎢ ⎥
⎣ 0 … ln ⎦

WORKED EXAMPLES
EXAMPLE 1
⎡3 1 1⎤
Diagonalise the matrix A = ⎢1 3 21⎥⎥ by means of an orthogonal transformation.
⎢
⎢⎣1 21 3 ⎥⎦
⎡3 1 1⎤
⎢
Solution. Given A = ⎢1 3 −1⎥⎥
⎣⎢1 −1 3 ⎦⎥
A is a symmetric matrix, since the elements equidistant from the main diagonal are the same.
The characteristic equation of A is A − l I = 0
3−l 1 1
⇒ 1 3 − l −1 = 0 ⇒ l3 – S1l2 + S2l – S3 = 0
1 −1 3−l
where S1 = 3 + 3 + 3 = 9
3 −1 3 1 3 1
S2 = + + = (9 − 1) + (9 − 1) + (9 − 1) = 24
−1 3 1 3 1 3
3 1 1
S 3 = 1 3 −1 = 3(9 − 1) − 1(3 + 1) + ( −1 − 3) = 24 − 4 − 4 = 16
1 −1 3
∴ The characteristic equation is l3 − 9l2 + 24l − 16 = 0
By trial we find l = 1 is a root.
Other roots are given by l2 – 8l + 16 = 0
⇒ (l − 4)2 = 0 1 −9 24 −16
1
⇒ l = 4, 4 0 1 −8 16
∴ the eigen values are l = 1, 4, 4. 1 − 8 16 0
⎡x1 ⎤
⎢ ⎥
Let X = ⎢ x 2 ⎥ be an eigen vector corresponding to l.
⎢⎣ x 3 ⎥⎦
⎡3 − l 1 1 ⎤ ⎡ x 1 ⎤ ⎡0 ⎤
Then (A − lI)X = 0 ⇒ ⎢
⎢ 1 3−l −1 ⎥⎥ ⎢⎢ x 2 ⎥⎥ = ⎢⎢0 ⎥⎥
⎣⎢ 1 −1 3 − l ⎦⎥ ⎣⎢ x 3 ⎦⎥ ⎣⎢0 ⎥⎦
(3 − l ) x 1 + x 2 + x 3 = 0 ⎫
⇒ ⎪
x 1 + (3 − l ) x 2 − x 3 = 0 ⎬ (I)
x 1 − x 2 + (3 − l )x 3 = 0 ⎪⎭

Matrices ■ 1.91
Case (i) If l = 1 then equations (I) become

2x 1 + x 2 + x 3 = 0, x 1 + 2x 2 − x 3 = 0 and x 1 − x 2 + 2x 3 = 0
Choosing the first two equations, we have

x1 x2 x3
x1 x x
= 2 = 3 1 1 2 1
−1 − 2 1 + 2 4 − 1
x1 x 2 x x1 x 2 x3 2 −1 1 2
⇒ = = 3 ⇒ = =
−3 3 3 −1 1 1
⎡ −1⎤
⎢
Choosing x1 = −1, x2 = 1, x3 = 1, we get an eigen vector X1 = 1⎥
⎢ ⎥
⎢⎣ 1⎥⎦
Case (ii) If l = 4, then the equations (I) become
−x1 + x 2 + x 3 = 0 ⇒ x1 − x 2 − x 3 = 0
x 1 − x 2 − x 3 = 0 and x 1 − x 2 − x 3 = 0
We get only one equation x1 – x2 – x3 = 0 (1)
To solve for x1, x2, x3, we can assign arbitrary values for two of the variables and we shall find 2
orthogonal vectors.
⎡1 ⎤
Put x3 = 0, x2 = 1, then x1 = 1, we get an eigen vector X 2 = ⎢1 ⎥⎥ .
⎢
⎢⎣0 ⎥⎦
⎡a⎤
Let X 3 = ⎢⎢ b ⎥⎥ be ⊥ to X2. Then a + b = 0 ⇒ b = −a and X3 should satisfy (1) ∴ a – b – c = 0
⎢⎣ c ⎥⎦
⎡ 1⎤
⎢ ⎥
Choosing a = 1, we get b = −1 and c = 2, ∴ X 3 = ⎢ −1⎥
⎣⎢ 2⎦⎥
Thus, the eigen values are l = 1, 4, 4, and the corresponding eigen vectors are
⎡ −1⎤ ⎡ 1⎤ ⎡ 1⎤
X1 = ⎢ 1⎥ , X 2 = ⎢ 1⎥ , X 3 = ⎢ −1⎥⎥
⎢ ⎥ ⎢ ⎥ ⎢
⎢⎣ 1⎥⎦ ⎢⎣0 ⎥⎦ ⎢⎣ 2⎥⎦
⎡ 1 ⎤ ⎡ 1 ⎤
⎢− ⎥ ⎡ 1 ⎤ ⎢ ⎥
⎢ 3⎥
⎢ ⎥ ⎢ 6⎥
The normalised eigen vectors are ⎢ 1 ⎥ ⎢ 2⎥ ⎢ −1 ⎥
⎢ ⎥, ⎢ 1 ⎥ ⎢ ⎥
⎢ 3⎥ ⎢ 6⎥
⎢ ⎥,
⎢ 1 ⎥ ⎢ 2⎥ ⎢ 2 ⎥
⎢ ⎥ ⎢⎣ 0 ⎥⎦ ⎢ ⎥
⎣ 3⎦ ⎣ 6⎦

⎡ −1 1 1 ⎤ ⎡ −1 1 1 ⎤
⎢ ⎥ ⎢ ⎥
⎢ 3 2 6⎥
⎢ 3 3 3⎥
⎢ 1 1 −1 ⎥ ⎢ 1 1 ⎥
So, the normalised modal matrix N = ⎢ ⎥ ∴ NT = ⎢ 0⎥
⎢ 3 2 6⎥ ⎢ 2 2 ⎥
⎢ 1 2 ⎥ ⎢ 1 −1 2 ⎥
⎢ 0 ⎥ ⎢ ⎥
⎣ 3 6⎦ ⎣ 6 6 6⎦
⎡ 1 1 1 ⎤ ⎡ −1 4 4 ⎤
⎢− ⎥ ⎢ ⎥
⎢ 3 2 6⎥ ⎢ 3 2 6⎥
⎡3 1 1⎤
⎢ −1 ⎥ ⎢ 1 −4 ⎥
AN = ⎢1 3 −1 ⎥⎥
⎢ 1 1 4
⎢ ⎥=⎢ ⎥
⎢ 3 2 6⎥ ⎢ 3 2 6⎥
⎢⎣1 −1 3⎥⎦
⎢ 1 2 ⎥ ⎢ 1 8 ⎥
⎢ 0 ⎥ ⎢ 0 ⎥
⎣ 3 6⎦ ⎣ 3 6⎦
⎡ 1 1 1 ⎤ ⎡ 1 4 4 ⎤
⎢− ⎥ ⎢− ⎥
⎢ 3 3 3⎥ ⎢ 3 2 6⎥
∴ ⎢ 1 1 ⎥ ⎢ 1 4 −4 ⎥
N AN = ⎢
T
0⎥ ⎢ ⎥
⎢ 2 2 ⎥ ⎢ 3 2 6⎥
⎢ 1 1 2 ⎥ ⎢ 1 8 ⎥
⎢ − ⎥ ⎢ 0 ⎥
⎣ 6 6 6⎦ ⎣ 3 6⎦
⎡ 1 0 0⎤
= ⎢⎢0 4 0 ⎥⎥ , which is a diagonal matrix.
⎢⎣0 0 4 ⎥⎦
EXAMPLE 2
The eigen vectors of a 3 3 3 real symmetric matrix A corresponding to the eigen values 2, 3, 6
are [1, 0, 21]T, [1, 1, 1]T, [21, 2, 21]T respectively, find the matrix A.
Solution.
Given A is symmetric and the eigen values are different.
So, the eigen vectors are orthogonal pairwise.
The normalised eigen vectors are
⎡ 1 ⎤ ⎡ 1 ⎤
⎡ 1 ⎤ ⎢ ⎥ ⎢− ⎥
⎢ ⎥ ⎢ 3⎥ ⎢ 6⎥
⎢ 2⎥
⎢ 1 ⎥ ⎢ 2 ⎥
⎢ 0⎥ , ⎢ ⎥, ⎢ ⎥
⎢ ⎥ ⎢ 3⎥ ⎢ 6⎥
⎢− 1 ⎥
⎢ 1 ⎥ ⎢ −1 ⎥
⎢⎣ 2 ⎥⎦ ⎢ ⎥ ⎢ ⎥
⎣ 3⎦ ⎣ 6⎦

Matrices ■ 1.93
⎡ 1 1 1 ⎤
⎢ − ⎥
⎢ 2 3 6⎥
⎢ 1 2 ⎥
So, the normalised modal matrix is N = ⎢ 0 ⎥
⎢ 3 6⎥
⎢ 1 1 1 ⎥
⎢− − ⎥
⎣ 2 3 6⎦
Then by orthogonal reduction theorem 1.7, page 1.89
⎡2 0 0⎤
N AN = D = ⎢0 3 0 ⎥⎥ , since 2, 3, 6 are the eigen values.
T ⎢
⎢⎣0 0 6 ⎥⎦
But NT = N−1 ∴ N−1AN = D ⇒ A = ND N−1 = NDNT

⎡ 1 1 1 ⎤ ⎡2 0 0⎤ ⎡ 1 1 ⎤
⎢ − ⎥ ⎢ ⎥ ⎢ 0 − ⎥
⎢ 2 3 6⎥ ⎢ ⎥ ⎢ 2 2⎥
⎢ 1 2 ⎥ ⎢0 3 0 ⎥ ⎢ 1 1 1 ⎥
⇒ A=⎢ 0 ⎢ ⎥ ⎢
⎥ ⎥
⎢ 3 6⎥ ⎢ ⎥ ⎢ 3 3 3⎥
⎢ −1 1 1 ⎥ ⎢⎣0 0 6 ⎥⎦ ⎢ 1 2 1 ⎥
⎢ − ⎥ ⎢− − ⎥
⎣ 2 3 6⎦ ⎣ 6 6 6⎦
⎡ 2 3 − 6⎤ ⎡ 1 1 ⎤
⎢ ⎥ ⎢ 0 − ⎥
⎢ ⎥ ⎢ 2 2⎥
=⎢ 0 3 2 6⎥
⎥ ⎢ 1 1 1 ⎥
⎢ ⎢ ⎥
⎢ ⎥ ⎢ 3 3 3⎥
⎢ ⎥ ⎢ 1 2 1 ⎥
⎢⎣ − 2 3 − 6 ⎥⎦ ⎢− − ⎥
⎣ 6 6 6⎦
⎡ 3 −1 1⎤
= ⎢⎢ −1 5 −1⎥⎥ , which is the required matrix.
⎢⎣ 1 −1 3⎥⎦
EXAMPLE 3
⎡ 2 0 1⎤
Diagonalise the matrix A = ⎢⎢0 3 0 ⎥⎥ . Hence, find A3.
⎢⎣ 1 0 2 ⎥⎦
Solution.
⎡ 2 0 1⎤
Given A = ⎢0 3 0 ⎥ which is a symmetric matrix.
⎢ ⎥
⎢⎣ 1 0 2⎥⎦
So, we shall diagonalise by orthogonal transformation.

2−l 0 1
⇒ 0 3−l 0 = 0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
1 0 2−l
where S1 = 2 + 3 + 2 = 7
3 0 2 1 2 0
S2 = + + = 6 + 3 + 6 = 15
0 2 1 2 0 3
S3 = A = 2(6) − 0 + 1( −3) = 12 − 3 = 9
∴ the characteristic equation is l 3 − 7l 2 + 15l − 9 = 0 1 −7 15 −9

1
By trial l = 1 is a root. 0 1 −6 9
Other roots are given by l 2 − 6l + 9 = 0
1 −6 9 0
⇒ (l − 3) 2 = 0 ⇒ l = 3, 3
∴ the eigen values are l = 1, 3, 3
⎡ x1 ⎤
Let X = ⎢⎢ x 2 ⎥⎥ be an eigen vector corresponding to eigen value l.
⎢⎣ x 3 ⎥⎦
⎡2 − l 0 1 ⎤ ⎡ x 1 ⎤ ⎡0 ⎤
⎢
( A − lI)X = 0 ⇒ ⎢ 0 3−l 0 ⎥⎥ ⎢⎢ x 2 ⎥⎥ = ⎢⎢0 ⎥⎥
Then
⎢⎣ 1 0 2 − l ⎥⎦ ⎢⎣ x 3 ⎥⎦ ⎣⎢0 ⎥⎦
( 2 − l )x 1 + 0x 2 + x 3 = 0 ⎫
⇒ ⎪
0 x 1 + (3 − l ) x 2 + 0 x 3 = 0 ⎬ (I)
x 1 + 0 x 2 + ( 2 − l )x 3 = 0 ⎪⎭

x1 + x3 = 0 ⇒ x3 = −x1, 2x2 = 0 ⇒ x2 = 0 and x1 + x3 = 0 ⇒ x3 = −x1
⎡ 1⎤
Choose x1 = 1. ∴ x3 = −1 ∴ an eigen vector is X1 = ⎢ 0 ⎥⎥
⎢
⎢⎣ −1 ⎥⎦
− x 1 + x 3 = 0, 0 x 2 = 0 and x 1 − x 3 = 0 ⇒ x 1 = x 3 (2)
and x2 can take any value
⎡1 ⎤
Choosing x1 = 1, we get x3 = 1 and choose x2 = 0 ∴ an eigen vector is X 2 = ⎢⎢0 ⎥⎥
⎢⎣1 ⎥⎦

Matrices ■ 1.95
⎡a⎤
We shall now choose X 3 = ⎢⎢b ⎥⎥ orthogonal to X2
⎣⎢c ⎦⎥
∴ dot product = 0 ⇒ a + c = 0 and X3 should satisfy equations (2)
∴ a – c = 0 and 0b = 0
⎡0 ⎤
⎢
Solving, we get a = c = 0 and choose b = 1 ∴ X 3 = 1 ⎥
⎢ ⎥
⎢⎣0 ⎥⎦
⎡ 1⎤ ⎡1 ⎤ ⎡0 ⎤
⎢ 0⎥ , ⎢0⎥ , ⎢1⎥
Thus, the eigen values are 1, 3, 3 and the corresponding eigen vectors are ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ −1 ⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣0 ⎥⎦
Clearly they are pairwise orthogonal vectors.
⎡ 1 ⎤ ⎡ 1 ⎤
⎢ ⎥ ⎢ ⎥ ⎡0 ⎤
⎢ 2⎥ ⎢ 2⎥
The normalised eigen vectors are ⎢ 0⎥ , ⎢1⎥
⎢ 0 ⎥, ⎢ ⎥
⎢ ⎥ ⎢ ⎥ ⎢⎣0 ⎥⎦
⎢− 1 ⎥ ⎢ 1 ⎥
⎢⎣ 2 ⎥⎦ ⎢⎣ 2 ⎥⎦
⎡ 1 1 ⎤
⎡ 1 1 ⎤ ⎢ 0 − ⎥
⎢ 0⎥ 2 2⎥
2 2 ⎢
⎢ ⎥ ⎢ 1 1 ⎥
∴ the normalised modal matrix N = ⎢ 0 0 1⎥ ∴ N =⎢
T
0 ⎥
⎢ ⎥ ⎢ 2 2⎥
⎢− 1 1
0⎥ ⎢⎣ 0 1 0 ⎥⎦
⎢⎣ 2 2 ⎥⎦
⎡ 1 1 ⎤ ⎡ 1 3 ⎤
⎢ 0⎥ ⎢ 0⎥
⎡2 0 1⎤ 2 2 2 2
⎢ ⎥ ⎢ ⎥
AN = ⎢⎢0 3 0 ⎥⎥ ⎢ 0 0 1⎥ = ⎢ 0 0 3⎥
⎢⎣1 0 2⎥⎦ ⎢ ⎥ ⎢ ⎥
⎢− 1 1 1 3
0⎥ ⎢ − 0⎥
⎢⎣ 2 2 ⎥⎦ ⎢⎣ 2 2 ⎦⎥
⎡ 1 1 ⎤ ⎡ 1 3 ⎤
⎢ 0 − ⎥ ⎢ 0⎥
2 2⎥ 2 2 ⎡1 0 0 ⎤
⎢ ⎢ ⎥ ⎢
∴ N T AN = ⎢ 1 1 ⎥ ⎢ 0 0 3⎥ = ⎢0 3 0 ⎥⎥ = D
⎢ 0 ⎥ ⎢ ⎥ ⎢
0 0 3⎥⎦
0⎥ ⎣
⎢ 2 2⎥ ⎢− 1 3
⎢⎣ 0 1 0 ⎥⎦ ⎢⎣ 2 2 ⎥⎦
∴ N−1AN = D ⇒ A = ND N−1
∴ A3 = ND3N−1 = ND3NT NT = N−1]
∴
[

⎡ 1 1 ⎤ ⎡ 1 0 0⎤ ⎡ 1 −1 ⎤
⎢ 0⎥ ⎢ ⎥ ⎢ 0 ⎥
⎢ 2 2 ⎥ ⎢ ⎥ ⎢ 2 2⎥
⇒ A3 = ⎢ 0 0 1⎥ ⎢ 0 27 0⎥ ⎢ 1 1 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ 0 ⎥
⎢− 1 1 ⎢ ⎥ 2 2⎥
0⎥ ⎢
⎢⎣ 2 2 ⎥⎦ ⎢⎣ 0 0 27⎦⎥ ⎢⎣ 0 1 0 ⎥⎦
⎡ 1 1 ⎤ ⎡ 1 −1 ⎤ ⎡14 0 13⎤
⎢ 0⎥ ⎢ 0 ⎥ ⎢ ⎥
⎢ 2 2 ⎥ ⎢ 2 2⎥ ⎢ ⎥
=⎢ 0 0 1⎥ ⎢ 27 27 ⎥ = ⎢ 0 27 0 ⎥
⎢ ⎥ ⎢ 0 ⎥ ⎢ ⎥
⎢− 1 1 ⎢ 2 2⎥ ⎢ ⎥
0⎥
⎢⎣ 2 2 ⎦⎥ ⎢⎣ 0 27 0 ⎥⎦ ⎣⎢13 0 14 ⎥⎦
1.8 REAL QUADRATIC FORM. REDUCTION TO CANONICAL FORM

Definition 1.41 A homogeneous polynomial of second degree in any number of variables is called a
quadratic form.
For example
(i) x 2 + 4 xy + 4 y 2
(ii) ax 2 + by 2 + cz 2 + 2hxy + 2gyz + 2fzx
(iii) x 12 + x 22 + 3x 32 + x 42
are quadratic forms in 2, 3 and 4 variables respectively.
n n
Definition 1.42 The general quadratic form in n variables x1, x2, …, xn is ∑∑a x x
j =1 i =1
ij i j , where aij are
real numbers such that aij = aji for all i, j = 1, 2, 3, …, n.

n n
Usually the quadratic form is denoted by Q and Q = ∑ ∑ aij xi x j .
j =1 i =1
1. Matrix form of Q
⎡ x1 ⎤ ⎡ a11 a12 … a1n ⎤

⎢x ⎥ ⎢a a22 … a2n ⎥⎥
If X = ⎢ ⎥ , A = ⎢ , where aij = aji, then A is a symmetric matrix and the quadratic
2 21
⎢ …⎥ ⎢… … … …⎥
⎢ ⎥ ⎢ ⎥
⎣x n ⎦ ⎣an1 an2 … ann ⎦
form
n n
Q = ∑ ∑ aij x ij can be written as Q = XT AX.
j =1 i =1
A is called the matrix of the quadratic form.

⎡1 2⎤ ⎡ x ⎤
For example the quadratic form x2 + 4xy + 4y2 can be written in the matrix form [x y ] ⎢ ⎥ ⎢ ⎥.
⎣2 4⎦ ⎣ y ⎦
⎡x ⎤ ⎡1 2⎤
Here X = ⎢ ⎥ and A = ⎢ ⎥.
⎣y ⎦ ⎣2 4⎦

Matrices ■ 1.97
Note Since the quadratic form is Q = XT AX, it is obvious that the characteristics or properties of Q
depend on the characteristics of A.
2. Canonical form of Q
Definition 1.43 A quadratic form Q which contains only the square terms of the variables is said to
be in canonical form.
For example x2 + y2, x2 – y2, x2 + y2 – 4z2 and x 1 + x 2 + 2x 3 + x 4 are in canonical forms because they
2 2 2 2
contain only square terms of the variables.

3. Reduction of Q to canonical form by orthogonal transformation
Let Q = XT AX be a quadratic form in n variables x1, x2, …, xn and A = [aij] be the symmetric matrix
of order n of the quadratic form. We will reduce A to diagonal form by an orthogonal transformation
X = NY, where N is the normalised modal matrix of A. Then NT AN = D, where D is the diagonal
matrix containing the eigen values of A.
If l1, l2, …, ln are the eigen values of A, then
⎡ l1 0 0 … 0⎤
⎢ 0 l 0 … 0 ⎥⎥
D= ⎢ 2
⎢ … … … … …⎥
⎢ ⎥
⎣ 0 0 0 … ln ⎦
∴ Q = X T AX = (NY)T A(NY) = Y T (N T AN)Y = Y T DY
⎡ y1 ⎤ ⎡ l1 0 … 0⎤ ⎡ y 1 ⎤
⎢y ⎥ ⎢ 0 l … 0 ⎥⎥ ⎢⎢ y 2 ⎥⎥
Y = ⎢ 2 ⎥ , then Q = [ y 1 y 2 … y n ] ⎢
2
If
⎢…⎥ ⎢… … … …⎥ ⎢ … ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎣y n ⎦ ⎣ 0 0 … ln ⎦ ⎣y n ⎦
⎡ y1 ⎤
⎢y ⎥
= [l1 y 1 l 2 y 2 … l n y n ] ⎢ 2 ⎥ = l1 y 12 + l 2 y 22 + … + l n y n2
⎢…⎥
⎢ ⎥
⎣y n ⎦
This is the required quadratic form.
Note In the canonical form the coefficients are the eigen values of A. Since A is a symmetric matrix,
the eigen values of A are all real. So, the eigen values may be positive, negative or zero. Hence, the
terms of the canonical form may be positive, negative or zero. By using the canonical form or the
eigen values, we can characterise the quadratic form.
Definition 1.44 If A is the matrix of the quadratic form Q in the variables x1, x2, …, xn, then the rank
of Q is equal to the rank of A.
If rank of A < n, where n is the number of variables or order of A, then A = 0 and Q is called a
singular form.

4. Index, signature and rank of quadratic form

Definition 1.45 Let Q = XT AX be a quadratic form in n variables x1, x2, …, xn.
X = [x1, x2, …, xn]T and A is the matrix of the quadratic form.
(i) Index of the quadratic form is the number of positive eigen values of A.
(ii) Signature of the quadratic form is the difference between the number of positive and
negative eigen values of A.
(iii) Rank of the quadratic form is the number of positive and negative eigen values of A.
Usually index is denoted by p, signature by s and rank by r.
5. Definite and indefinite quadratic forms
Definition 1.46 Let Q = XT AX be a quadratic form in n variables x1, x2, …, xw.
i.e., X = [x1 … xn]T and A is the matrix of the quadratic form.
(i) Q is said to be positive definite if all the n eigen values of A are positive.
i.e., if r = n and p = n
e.g., y 12 + y 22 + … + y n2 is positive definite.
(ii) Q is said to be negative definite if all the n eigen values of A are negative.
i.e., if r = n, p = 0
e.g., − y 12 − y 22 − … − y n2 is negative definite.
(iii) Q is said to be positive semidefinite if all the n eigen values of A are ≥ 0 with at least one
eigen value = 0.
i.e., if r < n and p = r
e.g., y 12 + y 22 + … + y r2 , where r < n, is positive semi-definite.
(iv) Q is said to be negative semidefinite if all the n eigen values of A are ≤ 0 with at least one
value = 0.
i.e., if r < n and p = 0
e.g., − y 12 − y 22 − … − y r2 , where r < n, is negative semi definite.
(v) Q is said to be indefinite if A has positive and negative eigen values.
e.g., y 12 + y 22 − y 32 − y 42 + … + y n2 is indefinite.
6. We can also find the nature of a quadratic form without finding the eigen values of A or without
reducing to canonical form but by using the principal minors of A as below.
Definition 1.47 Let Q = XT AX be a quadratic form in n variables x1, x2, …, xn and let the matrix of
the quadratic form be
⎡ a11 a12 … a1n ⎤
⎢a a … a2n ⎥
A = ⎢ 21 22 ⎥
⎢… … … …⎥
⎢ ⎥
⎣an1 an2 … ann ⎦
a11 a12 a13

a11 a12
Let D1 = a11 = a11, D2 = D
, 3 = a21 a22 a23 and so on.
a21 a22
a31 a32 a33
Finally Dn = A .

Matrices ■ 1.99
The determinants D1, D2, D3, …, Dn are called the principal minors of A.
The quadratic form Q is said to be
(i) positive definite if Di > 0 for all i = 1, 2, …, n
(ii) negative definite if (−1)i Di > 0 for all i = 1, 2, …, n
i.e., D1, D3, D5, … are negative and D2, D4, D6 … are positive.
(iii) positive semi-definite if Di ≥ 0 for all i = 1, 2, 3, …, n and at least one Di = 0.
(iv) negative semi-definite if (−1)i Di ≥ 0 for all i = 1, 2, 3, …, n and at least one Di = 0.
(v) indefinite in all other cases.
7. Law of intertia of a quadratic form

In the reduction of a quadratic form to canonical form the number of positive and negative terms
are independent of the choice of the transformation.
In other words, the signature of a real quadratic form is invariant under a real non-singular
transformation. This property is called the law of inertia of the quadratic form.
WORKED EXAMPLES
EXAMPLE 1
Write down the matrix of the quadratic form 2 x 12 2 2 x 22 1 4 x 32 1 2 x 1 x 2 2 6 x 1 x 3 1 6 x 2 x 3 .
Solution.
The Q.F is 2x 12 − 2x 22 + 4 x 32 + 2x 1x 2 − 6 x 1x 3 + 6 x 2 x 3
It has 3 variables x1, x2, x3.
So, the matrix of the quadratic form is a 3 × 3 symmetric matrix.
⎡ a11 a12 a13 ⎤
A = ⎢⎢a21 a22 a23 ⎥⎥
⎢⎣a31 a32 a33 ⎥⎦
Here
1 1
a11 = coefficient of x 12 = 2 a12 = a21 = (coefficient of x1x2) = ( 2) = 1
2 2
1 1
a22 = coefficient of x 22 = −2 a13 = a31 = (coefficient of x1x3) = ( −6) = −3
2 2
1 1
a33 = coefficient of x 32 = 4 a23 = a32 = (coefficient of x2x3) = (6) = 3
2 2
⎡ 2 1 −3⎤
⎢
A = ⎢ 1 −2 3⎥⎥
∴
⎢⎣ −3 3 4 ⎥⎦

EXAMPLE 2
⎡2 4 5⎤
⎢ ⎥
Write down the quadratic form corresponding to the matrix ⎢ 4 3 1 ⎥ .
⎢⎣ 5 1 1 ⎥⎦
Solution.
⎡ 2 4 5⎤
⎢ ⎥
Let A = ⎢ 4 3 1⎥ , which is a 3 × 3 symmetric matrix.
⎣⎢ 5 1 1⎦⎥
So, the quadratic form has 3 variables x1, x2, x3.
⎡ x1 ⎤
Let X = ⎢⎢ x 2 ⎥⎥ , then the quadratic form is
⎢⎣ x 3 ⎥⎦
⎡ 2 4 5⎤ ⎡ x1 ⎤
Q = X AX = [ x1 x2 x3 ] ⎢⎢ 4 3 1⎥⎥
T ⎢x ⎥
⎢ 2⎥
⎢⎣ 5 1 1⎥⎦ ⎢⎣ x3 ⎥⎦
⎡ x1 ⎤
= [2 x1 + 4 x2 + 5 x3 4 x1 + 3 x2 + x3 5 x1 + x2 + x3 ] ⎢ x2 ⎥
⎢x ⎥
⎣ 3⎦
= ( 2 x1 + 4 x2 + 5 x3 ) x1 + ( 4 x1 + 3 x2 + x3 ) x2 + (5 x1 + x2 + x3 ) x3
= 2 x12 + 4 x1 x2 + 5 x3 x1 + 4 x1 x2 + 3 x22 + x3 x2 + 5 x1 x3 + x2 x3 + x32
= 2 x12 + 3 x22 + x32 + 8 x1 x2 + 10 x1 x3 + 2 x2 x3
⎡ a11 a12 a13 ⎤ ⎡ 2 4 5⎤
Aliter Given A = ⎢⎢a21 a22 a23 ⎥⎥ = ⎢⎢ 4 3 1⎥⎥
⎢⎣a31 a32 a33 ⎥⎦ ⎢⎣ 5 1 1⎦⎥
Then the quadratic form in 3 variables x1, x2, x3 is
Q = X T AX = a11x 12 + a22 x 22 + a33 x 32 + 2(a12 )x 1x 2 + 2(a13 )x 1x 3 + 2(a23 )x 2 x 3
= 2x 12 + 3x 22 + x 32 + 2( 4)x 1x 2 + 2(5)x 1x 3 + 2(1)x 2 x 3
= 2x 12 + 3x 22 + x 32 + 8x 1x 2 + 10x 1x 3 + 2x 2 x 3
EXAMPLE 3
Discuss the nature of the following quadratic forms.
(i) 6x2 + 3y2 + 3z2 – 4xy 2 2yz + 4zx
(ii) 6 x 12 1 3 x 22 1 14 x 32 1 4 x 2 x 3 1 18 x 1 x 3 1 4 x 1 x 2
(iii) xy + yz + zx
(iv) 10x2 + 2y2 + 5z2 + 6yz 2 10zx 2 4xy.
Solution.
(i) The Q.F is 6x2 + 3y2 + 3z2 – 4xy − 2yz + 4zx, having 3 variables x, y, z.

Matrices ■ 1.101
The matrix of the quadratic form is

⎡ 6 −2 2⎤
A = ⎢⎢ −2 3 −1⎥⎥ .
⎢⎣ 2 −1 3⎥⎦
The principal minors are
6 −2
D1 = | 6 | = 6 > 0; D 2 = = 18 − 4 = 14 > 0
−2 3
6 −2 2
D3 = A = −2 3 −1 = 6(9 − 1) + 2( −6 + 2) + 2( 2 − 6) = 48 − 8 − 8 = 32 > 0
2 −1 3
Since D1, D2, D3 are positive, the quadratic form is positive definite.
(ii) The quadratic form is 6 x 12 + 3x 22 + 14 x 32 + 4 x 2 x 3 + 18x 1x 3 + 4 x 1x 2 , having 3 variables x1, x2, x3.
The matrix of the Q.F is
⎡6 2 9 ⎤
A = ⎢⎢ 2 3 2 ⎥⎥ .
⎢⎣ 9 2 14 ⎥⎦
6 2
The principal minors are D1 = 6 > 0; D2 = = 18 − 4 = 14 > 0
2 3
6 2 9
and D3 = A = 2 3 2
9 2 14
= 6(42 − 4) – 2(28 − 18) + 9(4 − 27) = 6(38) – 20 + 9(−23) = 228 – 20 – 207 = 1 > 0
Since D1, D2, D3 are positive, the quadratic form is positive definite.
(iii) The quadratic form is xy + yz + zx in 3 variables x, y, z
⎡ 1 1⎤
⎢0 2 2⎥
⎢ ⎥
The matrix of the quadratic form is A = ⎢
1 1⎥
0
⎢2 2⎥
⎢ ⎥
⎢1 1 0⎥
⎢⎣ 2 2 ⎥⎦
1
0
2 1
The principal minors are D1 = 0, D2 = = − <0
1 4
0
2
1 1
0
2 2 0 1 1
1 1 1 1 1 1
and D3 = 0 = 1 0 1 = [ 0 − 1(0 − 1) + (1 − 0) ] = × 2 = > 0
2 2 8 8 8 4
1 1 0
1 1
0
2 2

Since D1 = 0, D2 < 0, D3 > 0, the quadratic form is indefinite.

(iv) The Q.F 10 x 2 + 2 y 2 + 5z 2 + 6 yz − 10 zx − 4 xy is in three variables x, y, z.
⎡ 10 −2 −5⎤
⎢
The matrix of the quadratic form A = ⎢ −2 2 3⎥⎥
⎢⎣ −5 3 5⎥⎦
10 −2
The principal minors are D1 = 10 > 0, D 2 = = 20 − 4 = 16 > 0
−2 2
10 −2 −5
D3 = −2 2 3 = 10(10 − 9) + 2 (−10 + 15) – 5(−6 + 10) = 10 +10 – 20 = 0
−5 3 5
Since D1 > 0, D2 > 0 and D3 = 0, the quadratic form is positive semi-definite.
EXAMPLE 4
Determine l so that l(x2 1 y2 1 z2) 1 2xy 2 2yz 1 2zx is positive definite.
Solution. ⎡l 1 1⎤
The matrix of the quadratic form is A = ⎢⎢ 1 l −1⎥⎥
⎢⎣ 1 −1 l ⎥⎦
l 1
The principal minors are D1 = l, D 2 = = l 2 − 1;
1 l
D3 = A = l (l 2 − 1) − (l + 1) + ( −1 − l )
= (l + 1)(l(l + 1) − 1 − 1)
l + 1)(l 2 − l − 2 = (l + 1) 2 (l − 2)
= (l
Given the quadratic form is positive definite.
∴ D1 > 0, D2 > 0 and D3 > 0
⇒ l > 0, l2 – 1 > 0 ⇒ (l + 1)(l − 1) > 0 ⇒ l>1 ( l > 0)
∴
and (l + 1) (l − 2) > 0 ⇒ l − 2 > 0

2
⇒ l>2 [∴ (l + 1)2 > 0]
∴ the common values of l are l > 2
EXAMPLE 5
Show that the quadratic form ax 12 2 2bx 1 x 2 1 cx 22 is positive definite if a . 0 and ac 2 b2 . 0.
Solution.
⎡ a −b ⎤
The matrix of the quadratic form is A = ⎢
⎣ −b c ⎥⎦
a −b
The principal minors are D1 = a, D 2 = A = = ac − b 2 .
−b c
Given a > 0 and ac – b2 > 0. ∴ D1 > 0 and D2 > 0.
Hence, the Q.F is positive definite.

Matrices ■ 1.103
EXAMPLE 6
Reduce 6 x 2 1 3 y 2 1 3 z 2 2 4 xy 2 2 yz 1 4 xz into a canonical form by an orthogonal
reduction and find the rank, signature, index and the nature of the quadratic form.
Solution.
Given quadratic form is 6 x 2 + 3y 2 + 3z 2 − 4 xy − 2 yz + 4 xz
⎡ 6 −2 2⎤
The matrix of the Q.F is A = ⎢⎢ −2 3 −1⎥⎥
⎢⎣ 2 −1 3⎥⎦

6 − l −2 2
⇒ −2 3 − l −1 = 0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
2 −1 3 − l
where S1 = 6 + 3 + 3 = 12
3 −1 6 2 6 −2
S2 = + + = 8 + 14 + 14 = 36
−1 3 2 3 −2 3
S3 = A = 6(9 − 1) + 2( −6 + 2) + 2( 2 − 6) = 48 − 8 − 8 = 32
∴ the characteristic equation is l 3 − 12l 2 + 36l − 32 = 0

1 −12 36 −32
By trial, l = 2 is a root. 2
Other roots are given by l2 − 10l + 16 = 0 0 2 −20 32
⇒ (l − 2)(l − 8) = 0 ⇒ l = 2, 8 1 −10 16 0
∴ the eigen values are l = 2, 2, 8

⎡ x1 ⎤
Let X = ⎢ x ⎥ be an eigen vector corresponding to eigen value l.
⎢ 2⎥
⎢⎣ x 3 ⎥⎦
⎡6 − l −2 2 ⎤ ⎡ x 1 ⎤ ⎡0 ⎤
⎢
Then ( A − lI)X = 0 ⇒ ⎢ −2 3 − l −1 ⎥⎥ ⎢ x ⎥ = ⎢0 ⎥
⎢ 2⎥ ⎢ ⎥
⎢⎣ 2 −1 3 − l ⎥⎦ ⎢⎣ x 3 ⎥⎦ ⎢⎣0 ⎥⎦
(6 − l ) x 1 − 2 x 2 + 2 x 3 = 0 ⎫
⇒ ⎪
−2x 1 + (3 − l )x 2 − x 3 = 0 ⎬ (I)
2x 1 − x 2 + (3 − l )x 3 = 0 ⎪⎭
−2 x1 − 2 x2 + 2 x3 = 0 ⇒ x1 + x2 − x3 = 0

−2 x1 − 5 x2 − x3 = 0 ⇒ 2 x1 + 5 x2 + x3 = 0
2 x1 − x2 − 5 x3 = 0
From the first two equations we get
x1 x2 x x1 x2 x3
= = 3
1 + 5 −2 − 1 5 − 2 1 −1 1 1
x1 x 2 x
⇒ = = 3 5 1 2 5
6 −3 3
x1 x 2 x3
⇒ = =
2 −1 1
⎡ 2⎤
Choosing x1 = 2, x2 = −1, x3 = 1, we get an eigen vector X1 = ⎢⎢ −1⎥⎥
Case (ii) If l = 2, then equations (I) become ⎢⎣ 1⎥⎦
4 x 1 − 2x 2 + 2x 3 = 0 ⇒ 2x 1 − x 2 + x 3 = 0
−2x 1 + x 2 − x 3 = 0 ⇒ 2x 1 − x 2 + x 3 = 0
2x 1 − x 2 + x 3 = 0
So, we get only one equation 2x1 − x2 + x3 = 0 (1)
Choosing x3 = 0, we get 2x1 – 2x2 = 0 ⇒ x2 = 2x1
⎡1 ⎤
⎢
Choosing x1 = 1, we get x2 = 2 and x3 = 0 ∴ an eigen vector is X 2 = ⎢ 2⎥⎥
⎡a⎤ ⎢⎣0 ⎥⎦
⎢ ⎥
We shall find another eigen vector X 3 = ⎢b ⎥ orthogonal to X2
⎢⎣c ⎥⎦
∴ a + 2b = 0 ⇒ a = −2b
Also X3 satisfies (1)
∴ 2a – b + c = 0 ⇒ −4b – b + c = 0 ⇒ c = 5b
⎡ −2⎤
Choosing b = 1, we get c = 5 and a = −2 and eigen vector X 3 = ⎢ 1 ⎥
⎢ ⎥
⎣⎢ 5⎦⎥
Thus, the eigen values are 8, 2, 2 and the corresponding eigen vectors are
⎡ 2⎤ ⎡1⎤ ⎡ −2⎤
X1 = ⎢ −1⎥ , X 2 = ⎢ 2⎥ , X 3 = ⎢⎢ 1⎥⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ 1⎥⎦ ⎢⎣ 0 ⎥⎦ ⎢⎣ 5⎥⎦
Clearly X3 is orthogonal to X1 and X2. Also X1, X2 are orthogonal.
⎡ 2 ⎤ ⎡ 1 ⎤ ⎡ −2 ⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ 6⎥ ⎢ 5⎥ ⎢ 30 ⎥
⎢ −1 ⎥ ⎢ 2 ⎥ ⎢ 1 ⎥
The normalised eigen vectors are ⎢ ⎥ , ⎢ ⎥, ⎢ ⎥
⎢ 6⎥ ⎢ 5⎥ ⎢ 30 ⎥
⎢ 1 ⎥ ⎢ ⎥ ⎢ 5 ⎥
⎢ ⎥ ⎢ 0 ⎥ ⎢ ⎥
⎣ 6⎦ ⎣ ⎦ ⎣ 30 ⎦

Matrices ■ 1.105
⎡ 2 1 −2 ⎤
⎢ ⎥
⎢ 6 5 30 ⎥
⎢ −1 2 1 ⎥
The normalised modal matrix N = ⎢ ⎥
⎢ 6 5 30 ⎥
⎢ 1 5 ⎥
⎢ 0 ⎥
⎣ 6 30 ⎦
⎡8 0 0 ⎤
Then N AN = D = ⎢⎢0 2 0 ⎥⎥
T
⎢⎣0 0 2⎥⎦
⎡ y1 ⎤
The orthogonal transformation X = NY, where Y = ⎢⎢ y 2 ⎥⎥ reduces the given quadratic form to
⎢⎣ y 3 ⎥⎦
⎡8 0 0 ⎤ ⎡ y 1 ⎤
⎢ ⎥⎢ ⎥
YT DY = [y1 y2 y3] ⎢0 2 0 ⎥ ⎢ y 2 ⎥
⎢⎣0 0 2⎥⎦ ⎢⎣ y 3 ⎥⎦
= 8y 12 + 2y 22 + 2 y 32 , which is the canonical form.

∴ rank of the Q.F = 3, index = 3, signature = 3
The Q.F is positive definite, since all the eigen values are positive.
EXAMPLE 7
Find out the type of conic represented by 17x2 2 30xy 1 17y2 5 128 after reducing the quadratic
form 17x2 2 30xy 1 17y2 to canonical form by an orthogonal transformation.
Solution.
Given quadratic form is 17x2 – 30xy + 17y2
⎡ 17 −15⎤
The matrix of the Q.F is A=⎢ ⎥
⎣ −15 17⎦

17 − l −15
⇒ =0
−15 17 − l
⇒ (17 − l ) 2 − 152 = 0 ⇒ (17 − l ) 2 = 152 ⇒ 17 − l = ±15 ⇒ l = 2 or 32.
⎡x ⎤
If X = ⎢ 1 ⎥ be an eigen vector corresponding to eigen value l.
⎣x 2 ⎦

⎡17 − l −15 ⎤ ⎡ x1 ⎤ ⎡0 ⎤
Then ( A − lI)X = 0 ⇒ ⎢ ⎥ ⎢ ⎥=⎢ ⎥
⎣ −15 17 − l ⎦ ⎣ x2 ⎦ ⎣0 ⎦
⇒ (17 − l )x 1 − 15x 2 = 0 ⎫ (I)
⎬
−15x 1 + (17 − l )x 2 = 0 ⎭
15x 1 − 15x 2 = 0, − 15x 1 + 15x 2 = 0 ⇒ x 1 = x 2
⎡1⎤
Choosing x1 = 1, we get x2 = 1 ∴ an eigen vector is X1 = ⎢ ⎥
⎣1⎦
Case (ii) If l = 32, then equation (I) become
−15 x1 − 15 x2 = 0 and − 15 x1 − 15 x2 = 0 ⇒ x2 = − x1
⎡ 1⎤
Choose x1 = 1, we get x2 = −1 ∴ an eigen vector is X 2 = ⎢ ⎥
⎣ −1⎦
⎡ 1 ⎤ ⎡ 1 ⎤
⎢ ⎥ ⎢ ⎥
∴ the normalised eigen vectors are ⎢ 2⎥ ⎢ 2⎥
,
⎢ 1 ⎥ ⎢ 1 ⎥
⎢ ⎥ ⎢− ⎥
⎣ 2⎦ ⎣ 2⎦
The normalised modal matrix
⎡ 1 1 ⎤
⎢ ⎥
2 2⎥
N= ⎢
⎢ 1 1 ⎥
⎢ − ⎥
⎣ 2 2⎦
⎡ 1 1 ⎤ ⎡ 1 1 ⎤
⎢ ⎥ ⎡ 17 −15⎤ ⎢ ⎥
2 2⎥ 2 2 ⎥ = ⎡2 0⎤ = D
⇒ N T AN = ⎢ ⎢ ⎥ ⎢ ⎢ ⎥
⎢ 1 1 ⎥ ⎢ ⎥ ⎢ 1 −1 ⎥ ⎣0 32⎦
⎢ − ⎥ ⎢⎣ −15 17 ⎥⎦ ⎢ ⎥
⎣ 2 2⎦ ⎣ 2 2⎦
⎡y ⎤
The transformation X = NY, where Y = ⎢ 1 ⎥ , reduces the given quadratic form to
⎣ y2 ⎦
⎡ 2 0 ⎤ ⎡ y1 ⎤
Y T DY = [ y1 y2 ] ⎢ ⎥ ⎢ ⎥ = 2 y1 + 32 y2 , which is the canonical form.
2 2
⎣0 32⎦ ⎣ y2 ⎦
But the given quadratic from = 128
y12 y22
∴ 2 y12 + 32 y22 = 128 ⇒ + = 1, which represents an ellipse.
64 4

Matrices ■ 1.107
EXAMPLE 8
Reduce the quadratic form 8x2 1 7y2 1 3z2 2 12xy 1 4xz 2 8yz to the canonical form by an
orthogonal transformation. Find one set of values of x, y, z (not all zero) which will make the
quadratic form zero.
Solution.
Given quadratic form is 8x2 + 7y2 + 3z2 – 12xy + 4xz – 8yz
⎡ 8 −6 2⎤
⎢
The matrix of the quadratic form is A = ⎢ −6 7 −4 ⎥⎥
⎢⎣ 2 −4 3⎥⎦
8−l −6 2
⇒ −6 7 − l −4 = 0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
2 −4 3 − l
where S1 = 8 + 7 + 3 = 18
7 −4 8 2 8 −6
S2 = + +
−4 3 2 3 −6 7
= ( 21 − 16) + ( 24 − 4) + (56 − 36) = 5 + 20 + 20 = 45
S3 = A = 8( 21 − 16) + 6( −18 + 8) + 2( 24 − 14) = 40 − 60 + 20 = 0
∴ the characteristic equation is l 3 − 18l 2 + 45l = 0
⇒ l (l 2 − 18l + 45) = 0 ⇒ l(l − 3)(l − 15) = 0 ⇒ l = 0, 3,15.

⎡ x1 ⎤
If X = ⎢⎢ x 2 ⎥⎥ be an eigen vector of the eigen value l of A, then
⎢⎣ x 3 ⎥⎦
⎡8 − l −6 2 ⎤ ⎡ x1 ⎤
( A − lI)X = 0 ⇒ ⎢⎢ −6 7 − l −4 ⎥⎥ ⎢⎢ x 2 ⎥⎥ = 0
⎢⎣ 2 −4 3 − l ⎥⎦ ⎢⎣ x 3 ⎥⎦
(8 − l )x 1 − 6 x 2 + 2x 3 = 0 ⎫
⇒ ⎪
−6 x 1 + (7 − l )x 2 − 4 x 3 = 0 ⎬ (I)
2x 1 − 4 x 2 + (3 − l )x 3 = 0 ⎪⎭


8 x1 − 6 x2 + 2 x3 = 0 ⇒ 4 x1 − 3 x2 + x3 = 0
−6 x1 + 7 x2 − 4 x3 = 0 and 2 x1 − 4 x2 + 3 x3 = 0
From the first and third equations, we get
x1 x2 x3 x1 x2 x3
= =
−9 + 4 2 − 12 −16 + 6 −3 1 4 −3
⇒ x1 x
= 2 = 3
x
−5 −10 −10 −4 3 2 −4
⇒ x1 x 2 x3
= =
1 2 2
⎡1 ⎤
Choosing x1 = 1, x2 = 2, x3 = 2, we get an eigen vector X1 = ⎢ 2⎥
⎢ ⎥
⎢⎣ 2⎥⎦
⎡ 2⎤ ⎡ 2⎤
Similarly, we can find for l = 3, X 2 = ⎢ 1⎥ and for l = 15, X 3 = ⎢⎢ −2⎥⎥
⎢ ⎥
⎢⎣ −2⎥⎦ ⎢⎣ 1⎥⎦
⎡1 2 2⎤
⎢3 3⎥
⎢
3
⎥ ⎡1 2 2⎤
∴ the normalised modal matrix N = ⎢2 1 −2 ⎥ = 1 ⎢ 2 1 −2⎥⎥
⎢3 3 3⎥ 3⎢
⎢ ⎥ ⎢⎣ 2 −2 1⎥⎦
⎢2 −2 1⎥
⎢⎣ 3 3 3 ⎥⎦
⎡0 0 0 ⎤
T
= = ⎢0 3 0 ⎥
∴ N AN D ⎢ ⎥
⎢⎣0 0 15⎥⎦
⎡ y1 ⎤
⎢ ⎥
The transformation X = NY, where Y = ⎢ y 2 ⎥ , reduces the quadratic form to the canonical form
⎢⎣ y 3 ⎥⎦
⎡0 0 0 ⎤ ⎡ y1 ⎤
Y DY = [ y 1 y 2 y 3 ] ⎢⎢0 3 0 ⎥⎥
T ⎢y ⎥
⎢ 2 ⎥ = 0 y 1 + 3y 2 + 15y 2
2 2 2
⎢⎣0 0 15⎥⎦ ⎢⎣ y 3 ⎥⎦
⎡x ⎤ ⎡1 2 2⎤ ⎡ y1 ⎤
The transformation is ⎢⎢ y ⎥⎥ = ⎢⎢ 2 1 −2⎥⎥ ⎢y ⎥
1
3 ⎢ 2⎥
⎢⎣ z ⎥⎦ ⎢⎣ 2 −2 1⎥⎦ ⎢⎣ y 3 ⎥⎦

Matrices ■ 1.109
1 1 1
∴ x = ( y1 + 2 y2 + 2 y3 ), y = ( 2 y1 + y2 − 2 y3 ) and z = ( 2 y1 − 2 y2 + y3 )
3 3 3
Quadratic form = 0
⇒ 0 y 12 + 3y 22 + 15y 32 = 0 ⇒ 3y 22 + 15y 32 = 0 ⇒ y 2 = 0 and y 3 = 0
and y1 can take any value, we shall choose y1 = 3
∴ x = 1, y = 2, z = 2
Hence, this set of values will make the quadratic form = 0
EXAMPLE 9
Reduce the quadratic form x 12 1 2 x 22 1 x 32 2 2 x 1 x 2 1 2 x 2 x 3 to the canonical form through an
orthogonal transformation and hence show that it is positive semi-definite. Also give a non-zero
set of values (x1, x2, x3) which makes the quadratic form zero.
Solution.
Given the quadratic form is x 12 + 2x 22 + x 32 − 2x 1x 2 + 2x 2 x 3
⎡ 1 −1 0 ⎤
The matrix of the quadratic form is A = ⎢⎢ −1 2 1⎥⎥
⎢⎣ 0 1 1⎥⎦

1− l −1 0
⇒ −1 2 − l 1 = 0 ⇒ l 3 − S1l 2 + S2 l − S3 = 0
0 1 1− l
where S1 = 1 + 2 + 1 = 4
2 1 1 0 1 −1
S2 = + + = 1+1+1 = 3
1 1 0 1 −1 2
S3 = A = 2 − 1 + 1( −1) = 0
∴ the characteristic equation is l 3 − 4l 2 + 3l = 0

⇒ l(l 2 − 4l + 3) = 0 ⇒ l(l − 1)(l − 3) = 0 ⇒ l = 0, 1, 3

⎡ x1 ⎤
If X = ⎢⎢ x 2 ⎥⎥ be an eigen vector corresponding to an eigen value l of A, then
⎢⎣ x 3 ⎥⎦

⎡1 − l −1 0 ⎤ ⎡ x 1 ⎤ ⎡0 ⎤
⎢ −1 2 − l 1 ⎥⎥ ⎢ x ⎥ = ⎢0 ⎥
(A − lI)X = 0 ⇒ ⎢ ⎢ 2⎥ ⎢ ⎥
⎢⎣ 0 1 1 − l ⎥⎦ ⎢⎣ x 3 ⎥⎦ ⎢⎣0 ⎥⎦
(1 − l ) x1 − x2 + 0 x3 = 0
⇒ − x1 + ( 2 − l ) x2 + x3 = 0 (I)
0 x1 + x2 + (1 − l ) x3 = 0

x1 − x2 = 0 ⇒ x1 = x2 , − x1 + 2 x2 + x3 = 0 and x2 + x3 = 0 ⇒ x3 = − x2
⎡ 1⎤
Take x2 = 1, then x1 = 1, x3 = −1 ∴ an eigen vector is X1 = ⎢⎢ 1⎥⎥
⎢⎣ −1⎥⎦
Case (ii) If l = 1, the equations (I) become
0x1 − x 2 = 0 ⇒ x 2 = 0
−x1 + x 2 + x 3 = 0 ⇒ x1 = x 3
and
⎡ 1⎤
Take x3 = 1, then x1 = 1 ∴ an eigen vector is X 2 = ⎢⎢0 ⎥⎥
⎢⎣ 1⎥⎦
Case (iii) If l = 3, then equations (I) become
−2 x1 − x2 = 0 ⇒ x2 = −2 x1 , − x1 − x2 + x3 = 0 and x2 − 2 x3 = 0 ⇒ x2 = 2 x3
⎡ −1⎤
⎢
Take x2 = 2, then x1 = −1, x3 = 1 ∴ an eigen vector is X 3 = ⎢ 2⎥⎥
⎢⎣ 1⎥⎦
Thus, the eigen values are l = 0, 1, 3 and the eigen vectors are
⎡ 1⎤ ⎡ 1⎤ ⎡ −1⎤
⎢ ⎥ ⎢
X1 = ⎢ 1⎥ , X 2 = ⎢0 ⎥ , X 3 = ⎢⎢ 2⎥⎥
⎥
⎢⎣ −1⎥⎦ ⎢⎣ 1⎥⎦ ⎢⎣ 1⎥⎦
⎡ 1 ⎤ ⎡ 1 ⎤
⎢ ⎥ ⎡ 1 ⎤ ⎢− ⎥
⎢ 3⎥ ⎢ ⎥ ⎢ 6⎥
⎢ 2⎥
⎢ 1 ⎥ ⎢ 2 ⎥
∴ the normalized eigen vectors are ⎢ ⎥ , ⎢ 0 ⎥, ⎢ ⎥
⎢ 3⎥ ⎢ ⎥ ⎢ 6⎥
⎢ 1 ⎥
⎢ −1 ⎥ ⎢ 1 ⎥
⎢ ⎥ ⎢⎣ 2 ⎦⎥ ⎢ ⎥
⎣ 3⎦ ⎣ 6⎦

Matrices ■ 1.111
⎡ 1 1 1 ⎤
⎢ − ⎥
⎢ 3 2 6⎥
⎢ 1 2 ⎥
∴ the normalised modal matrix is N = ⎢ 0 ⎥
⎢ 3 6⎥
⎢ −1 1 1 ⎥
⎢ ⎥
⎣ 3 2 6⎦
⎡0 0 0 ⎤
∴ the diagonal matrix is N AN = D = ⎢⎢0 1 0 ⎥⎥
T
⎢⎣0 0 3⎥⎦
The orthogonal transformation is X = NY
⎡ y1⎤
where Y = ⎢⎢ y 2 ⎥⎥ and the canonical form is Y T DY = y 22 + 3y 32
⎢⎣ y 3 ⎥⎦
∴ the quadratic form is positive semi-definite.
⎡ 1 1 1 ⎤
⎢ − ⎥
3 2 6⎥
⎡ x1 ⎤ ⎢ ⎡ y1⎤
⎢ ⎥ ⎢ 1 2 ⎥⎢ ⎥
The transformation X = NY ⇒ ⎢ x 2 ⎥ = ⎢ 0 ⎥ y2
3 6⎥⎢ ⎥
⎢⎣ x 3 ⎥⎦ ⎢ ⎢y ⎥
⎢ −1 1 1 ⎥ ⎣ 3⎦
⎢ ⎥
⎣ 3 2 6⎦
1 1 1 1 2 1 1 1
∴ x1 = y1 + y2 − y3 , x2 = y1 + y3 and x3 = − y1 + y2 + y3
3 2 6 3 6 3 2 6
These equation make the quadratic form = 0
⇒ y 22 + 3y 32 = 0 ⇒ y 2 = 0 and y 3 = 0
[Since sum of squares of real numbers = 0 ⇒ each number = 0] and y1 can take any real value.
1 1 1
∴ x1 = y1, x 2 = y1, x3 = − y1
3 3 3
Choosing y 1 = 3 , one set of values of x1, x2, x3 is x1 = 1, x2 = 1, x3 = −1.
EXERCISE 1.7
Diagonlaise the following matrices by orthogonal transformation.
⎡ 6 −2 2⎤ ⎡ 10 −2 −5⎤ ⎡ 8 −6 2⎤
1. A = ⎢⎢ −2 3 −1⎥⎥ 2. A = ⎢⎢ −2 2 3⎥⎥ ⎢
3. A = ⎢ −6 7 −4 ⎥⎥
⎢⎣ 2 −1 3⎥⎦ ⎢⎣ −5 3 5⎥⎦ ⎢⎣ 2 −4 3⎥⎦
⎡ 2 −1 1⎤ ⎡ 7 −2 0 ⎤ ⎡ 2 1 −1⎤
4. A = ⎢⎢ −1 2 −1⎥⎥ 5. A = ⎢⎢ −2 6 −2⎥⎥ 6. A = ⎢ 1 1 −2⎥⎥
⎢
⎢⎣ 1 −1 2⎥⎦ ⎢⎣ 0 −2 5⎥⎦ ⎢⎣ −1 −2 1⎥⎦

7. Find the symmetric matrix A whose eigen values and eigen vectors are given
⎡ 1⎤ ⎡1⎤
(i) eigen values are 0, 2, and eigen vectors ⎢ ⎥ , ⎢1⎥ .
⎣ −1⎦ ⎣⎦
⎡ 1⎤ ⎡0 ⎤ ⎡ 1⎤
(ii) eigen values are 1, 2, 3 and eigen vectors ⎢⎢ −1⎥⎥ , ⎢0 ⎥ ,
⎢ ⎥
⎢ 1⎥ .
⎢ ⎥
⎢⎣ 0 ⎥⎦ ⎢⎣ 1⎥⎦ ⎢⎣0 ⎥⎦
8. Reduce the quadratic form 8x 12 + 7x 22 + 3x 32 − 12x 1x 2 − 8x 2 x 3 + 4 x 3x 1 to the canonical form through an

orthogonal transformation and hence show that it is positive semi-definite.
9. Reduce the quadratic form 2x 12 + 6 x 22 + 2x 32 + 8x 1x 3 to canonical form by orthogonal reduction.
10. Reduce the quadratic form 3x 12 + 5x 22 + 3x 32 − 2x 2 x 3 + 2x 3x 1 − 2x 1x 2 to the canonical form by orthogonal

reduction.
11. Find the nature, index, signature and rank of the following Q.F, without reducing to canonical form.
(i) 3x 12 + 5x 22 + 3x 32 − 2x 2 x 3 + 2x 3x 1 − 2x 1x 2 . (ii) 10 x 2 + 2 y 2 + 5z 2 + 6 yz − 10 zx − 4 xy .
(iii) 3x 2 − 2 y 2 − z 2 − 4 xy + 8xz + 12 yz .
12. Determine the nature of the following quadratic form f(x1, x2, x3) = x 12 + 2x 22 .
13. Find the nature of the quadratic form 2x2 + 2xy + 3y2.
14. Find the index, signature and rank of the Q.F in 3 variables x 12 + 2x 22 − 3x 32 .
15. Reduce the quadratic form x 12 + 5x 22 + x 32 + 2x 1x 2 + 2x 2 x 3 + 6 x 3x 1 to canonical form through an orthogonal

transformation.
16. Reduce the quadratic form 2x 12 + 6 x 22 + 2x 32 + 8x 1x 3 to canonical form.
17. Reduce the given quadratic form Q to its canonical form using orthogonal transformation
Q = x2 + 3y2 + 3z2 − 2yz.
1. l = 2, 2, 8; eigen vectors [0 1 1]T, [1 1 −1]T, [2 –1 1]T
2. l = 0, 3, 14; eigen vectors [1 −5 4]T, [1 1 1]T, [−3 1 2]T
3. l = 0, 3, 15; eigen vectors [1 2 2]T, [2 1 −2]T, [2 −2 1]T
⎡ 22 −21 21⎤
⎢ ⎥
4. l = 1, 1, 4; eigen vectors [1 1 0] , [−1 1 2] , [1 −1 1] , A = ⎢ −21 22 −21⎥
T T T 3
⎢⎣ 21 −21 22⎥⎦
5. l = 3, 6, 9; eigen vectors [1 2 2]T, [2 1 −2]T, [2 −2 1]T

⎡ −1 0 0 ⎤ ⎡2 1 0⎤
⎡1 1⎤
6. D = ⎢⎢ 0 1 0 ⎥⎥ 7. (i) A = ⎢ ⎥ (ii) ⎢ 1 2 0 ⎥
⎢ ⎥
⎢⎣ 0 0 4 ⎥⎦ ⎣1 1⎦ ⎢⎣0 0 2⎥⎦

Matrices ■ 1.113
8. 3y 22 + 15y 32 9. −2 y 12 + 6 y 22 + 6 y 32 10. 2 y 12 + 3y 22 + 6 y 32
11. (i) Eigen values are 2, 3, 6; positive definite, index = 3, signature = 3, rank = 3.
(ii) Eigen values are 0, 3, 14; positive semi-definite, index = 2, signature = 2, rank = 2.
(iii) Eigen values are 3, 6, −9; indefinite, index = 2, signature = 1, rank = 3.
12. It is a positive semi-definite form 13. The quadratic form is positive definite.
14. Eigen value = 1, 2, −3, Index = 2, Signature = 1, Rank = 3
15. −2 y 12 + 3y 22 + 6 y 32 16. −2 y 12 + 6 y 22 + 6 y 32 17. y 12 + 2 y 22 + 4 y 32
SHORT ANSWER QUESTIONS

⎡5 4 ⎤
1. If A 5 ⎢ ⎥ , find the eigen values of A 2 2A 1 I.
2
⎣1 2⎦
2. Prove that A 5 ⎡
1 2⎤
and 23A21 have the same eigen values.
⎢ 2 1⎥
⎣ ⎦
⎡ 2 2 1⎤
3. Two eigen values of the matrix A 5 ⎢⎢ 1 3 1⎥⎥ are equal to 1 each, find the eigen values of A21.
⎣⎢ 1 2 2 ⎦⎥
⎡ 2 0 1⎤
⎢ ⎥
4. Find the sum and product of the eigen value of the matrix A 5 ⎢0 2 0 ⎥ .
⎢⎣ 1 0 2 ⎥⎦
⎡ 4 6 6⎤
5. Two eigen values of A 5 ⎢⎢ 1 3 2 ⎥⎥ are equal and they are double the third. Find the eigen values
⎢⎣21 25 22 ⎥⎦
of A2.
6. If l1, l2, ..., ln are the eigen values of an n 3 n matrix A, then show that l13 , l 32 ,…, l 3n are the eigen
values of A3.
⎡ 1 1 3⎤ ⎡21⎤
7. The matrix A 5 ⎢⎢ 1 5 1⎥⎥ has an eigen vector ⎢ 0 ⎥ , find the corresponding eigen value of A.
⎢ ⎥
⎣⎢3 1 1⎦⎥ ⎢⎣ 1⎥⎦
⎡ 1 4⎤
8. Using Cayley-Hamilton theorem find the inverse of ⎢ ⎥
⎣2 3⎦
9. If 2, 3 are eigen values of a square matrix A of order 2, express A2 in terms of A and I.
10. If A is an orthogonal matrix, show that A21 is also orthogonal.
11. For a given matrix A of order 3, A 5 32 and two of its eigen values are 8 and 2. Find the sum of the
eigen values
⎡ cos u sin u 0⎤
12. Check whether the matrix B = ⎢⎢2sin u cos u 0⎥⎥ is orthogonal? Justify.
⎢⎣ 0 0 1⎥⎦
⎡1 2 ⎤
13. Use Cayley-Hamilton theorem to find A4 2 4A3 2 5A2 1 A 1 2I when A 5 ⎢ ⎥.
⎣4 3 ⎦

14. True or false: If A and B are two invertible matrices then AB and BA have the same eigen values.
⎡2 2 1 ⎤
15. Find the eigen vector corresponding to the eigen value 1 of the matrix A 5 ⎢⎢1 3 1 ⎥⎥ .
⎢⎣1 2 2 ⎥⎦
⎡ 1 2 22 ⎤
⎢ 3 ⎥⎥ .
16. Find the sum and product of the eigen values of the matrix A 5 ⎢ 1 0
⎢⎣22 21 23 ⎥⎦
17. A is a singular matrix of order three, 2 and 3 are the eigen values. Find the third eigen value.
18. Find the nature of the quadratic form 2x2 + 2xy + 3y2.
19. If the quadratic form ax2 + 2bxy + cy2 is positive definite (or negative definite) then prove that the
quadratic equation ax2 + 2bx + c = 0 has imaginary roots.
20. Find the index and signature of the quadratic form x 12 1 2 x 22 2 5 x 32 .
OBJECTIVE TYPE QUESTIONS

A. Fill up the blanks
⎡1 9⎤ ⎡ x y 2⎤
1. If ⎢ ⎥=⎢ ⎥ , y < 0, then x − y + z + w = ________.
⎣2 0⎦ ⎣z w⎦
⎡1 5⎤ ⎡ −2 5⎤
2. If A − 2B = ⎢ ⎥ , 2A − 3B = ⎢ 0 7⎥ , then B = ________.
⎣ 3 7 ⎦ ⎣ ⎦
⎡ cos 2 a cos a sin a ⎤ p

3. If M (a) = ⎢ ⎥ , if a and b differ by , then M(a) ⋅ M(b) = ________.
⎣cos a sin a sin 2 a ⎦ 2
⎡ 1 0 2⎤
4. A = ⎢⎢0 2 1 ⎥⎥ satisfies the equation x3 − 6x2 + 7x + 2 = 0, then A−1 = ________.
⎢⎣ 2 0 3⎥⎦
⎡1 3 4 3⎤
⎡5 4 ⎤
5. The rank of the matrix ⎢ 3 9 12 9 ⎥ is ________. 6. Eigen values of ⎢ ⎥ are ________.
⎢ ⎥ ⎣1 2 ⎦
⎢⎣ −1 −3 −4 −3⎥⎦
⎡ −2 −1⎤
7. An Eigen vector corresponding to the Eigen value −1 of ⎢ ⎥ is ________.
⎣5 4⎦
⎡ 2 −2 2 ⎤
⎢ ⎥
8. If 2 is Eigen value of ⎢1 1 1 ⎥ , then the other Eigen values are ________.
⎢⎣1 3 −1⎥⎦
⎡ 1 7 5⎤
9. The sum of the square of the Eigen values of the matrix ⎢⎢0 2 9⎥⎥ is ________.
⎢⎣0 0 5⎥⎦
⎡a 4 ⎤
10. The Eigen values of the matrix ⎢ ⎥ are −2 and 3, then the values of a and b are ________.
⎣1 b ⎦

Matrices ■ 1.115
11. The nature of the quadratic form 2x 12 − x 22 is ________.
12. The nature of the quadratic form 2xy + 2yz + 2zx is ________.
13. The matrix of the quadratic form 3x 12 + 3x 22 − 5x 32 − 2x 1x 2 − 6 x 1x 3 − 6 x 2 x 3 is ________.
⎡ cos u sin u ⎤
14. The matrix A = ⎢ ⎥ is ________. 15. If A is an orthogonal matrix, then A−1 is ________.
B. Choose the correct answer

⎡2 ⎤
⎢x −1 2 ⎥
⎢ ⎥
1. If the matrix ⎢ 1 x 2x 2 ⎥ is singular, then the value of x is
⎢ 1 ⎥
⎢1 2 ⎥
⎣ x ⎦
(a) ±2 (b) ±1 (c) ±3 (d) None of these
⎡1 v2 v⎤
⎢ ⎥
2. If v is a complex cube root of unity, then the matrix ⎢v2 v 1 ⎥ is
⎢v 1 2⎥
v ⎦
⎣
(a) Singular (b) non-singular (c) symmetric (d) skew-symmetric
3. If A2 − A + I = 0, then the inverse of A is

(a) A − I (b) I − A (c) A + I (d) A
⎡ 5 5a a ⎤
4. Let A = ⎢⎢0 a 5a ⎥⎥ . If A 2 = 25, then a is equal to
⎢⎣0 0 5 ⎥⎦
1
(a) 25 (b) 1 (c) (d) 5
5
5. If a, b, c are in A.P, then the system of equations 3x + 4y + 5z = a, 4x + 5y + 6z = b, 5x + 6y + 7z = c is
(a) consistent (b) consistent with unique solution
(c) consistent with infinite number solutions (d) consistent with finite number solutions
6. If A is non-singular, then the equation AB = 0 implies

(a) B is non-singular (b) B is singular (c) B = 0 (d) None of these
7. The system of equations x + 2y − z = 6, 3x − y − 2z = 3, 4x + 3y + z = 9 is

(a) consistent with unique solution (b) consistent with infinite number of solutions
(c) inconsistent (d) None of these
⎡ 3 −1 1 ⎤
8. Two Eigen values of the matrix A = ⎢⎢ −1 5 −1⎥⎥ are 3 and 6. Then the Eigen values of A−1 are
⎢⎣ 1 −1 3 ⎥⎦
1 1 1 1 1 1 1 1 1 1
(a) 1, , (b) −1, , (c) − , , (d) , ,
3 6 3 6 2 3 6 2 3 6

9. If the Eigen values of a matrix A of order 3 are 2,3, and 4, then the Eigen values of adj A are
1 1 1 1 1 1
(a) 12, 8, 6 (b) , , (c) , , (d) None of these
2 3 4 12 8 6
10. If the sum of two Eigen values of 3 × 3 matrix A is equal to its trace, then the value of A is equal to
(a) 1 (b) −1 (c) 0 (d) None of these
⎡1 0 0 ⎤
11. The product of two Eigen values of the matrix A = ⎢0 3 −1⎥ is 2, then the third Eigen values is
⎢ ⎥
⎢⎣0 −1 3 ⎥⎦
(a) 2 (b) 4 (c) 5 (d) 6
12. If A and B are invertible matrices of the same order, such that AB = BA then A and B are
(a) similar (b) dissimilar
(c) have different Eigen values (d) None of these
13. The index and signature of the quadrature for x 12 + 2x 22 − 5x 32 are

(a) 1, 2 (b)1, −2 (c) 2, 1 (d) 2, −2
14. The matrix of the quadratic form 3x2 + 2y2 − 4xy is
⎡ 3 −2⎤ ⎡ 1 2⎤ ⎡ −3 2 ⎤ ⎡ −1 −2⎤
(a) ⎢ −2 2 ⎥ (b) ⎢ ⎥ (c) ⎢ 2 −2⎥ (d) ⎢ ⎥
⎣ ⎦ ⎣ −2 1 ⎦ ⎣ ⎦ ⎣ 2 −1⎦
15. The nature of the quadratic form 2xy + 2yz + 2zx is
(a) indefinite (b) definite (c) positive definite (d) negative definite
ANSWERS
A. Fill up the blanks ⎡ −3 0 2⎤
⎡ −4 −5⎤ ⎢ 1 1 ⎥⎥
1. 6 2. ⎢ ⎥ 3. 0 4. A −1
=⎢1
⎣ −6 −7⎦ ⎢ 2 2⎥
⎢2 0 −1⎥⎦
⎣
⎡1⎤
5. rank = 1 6. 1, 6 7. ⎢ ⎥ 8. −2, 2 9. 30
⎣ −1⎦
10. a = 2, b = −1 or a = 1, b = 2 11. indefinite 12. indefinite
⎡ 3 −1 3 ⎤
⎢ ⎥
13. ⎢ −1 3 −3⎥ 14. orthogonal 15. A −1 = A T
⎢⎣ −3 −3 −5⎥⎦
B. Choose the correct answer

1. (b) 2. (a) or (e) 3. (b) 4. (c) 5. (c) 6. (c) 7. (a) 8. (d) 9. (a)
10. (c) 11. (b) 12. (a) 13. (c) 14. (a) 15. (a)

Dr.R.Venkatesan Matrices

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dr.R.Venkatesan Matrices

Uploaded by

Copyright:

Available Formats

1.

1.6 EIGEN VALUES AND EIGEN VECTORS

X1 + X2 = (a1 + b1, a2 + b2, …, an + bn),

Definition 1.30 Linear Combination

Definition 1.31 Linearly dependent and independent vectors

M01_ENGINEERING_MATHEMATICS-I _CH01_Part A.indd 54 5/30/2016 4:36:18 PM

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 55 5/30/2016 5:03:24 PM

The relation between them is obtained as below.

1.6.2 Eigen Values and Eigen Vectors

Definition 1.33 Characteristic Equation and Characteristic Polynomial

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 56 5/30/2016 5:03:25 PM

⇒ a11 ⋅ a22 − (a11 + a22 )l + l 2 − a21a12 = 0

⎡ a11 a12 a13 ⎤ ⎡1 0 0 ⎤

Expanding this determinant we will get

l3 − S1l2 + S2l − S3= 0,

where S1 = sum of the diagonal elements of A

1.6.3 Properties of Eigen Vectors

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 57 5/30/2016 5:03:26 PM

(2) Let l1, l2 be two different eigen values of A.

∴ AX1 = l1X1 (1)

We have to prove X1 and X2 are linearly independent.

Multiply (3) by l1, we get

⇒ (a1l1 )X1 + (a 2 l1 )X 2 = 0 (5)

∴ (6) ⇒ a 2 = 0 ∴ (3) ⇒ a1X1 = 0 ⇒ a1 = 0, since X1 ≠ 0.

∴ X1 and X2 are linearly independent.

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 58 5/30/2016 5:03:29 PM

To ﬁnd eigen vectors:

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 59 5/30/2016 5:03:32 PM

where S1 = sum of the diagonal elements of A = a + a = 2a

l2 − 2al + (a2 + b2) = 0 ⇒ l = 2a ± 4a2 − 4(a2 + b 2 )

To ﬁnd eigen vectors:

So, we have only one equation x 2 = ix 1

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 60 5/30/2016 5:03:36 PM

To ﬁnd eigen vectors:

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 61 5/30/2016 5:03:38 PM

Case (i) If l = −1, then the equations (I) become

Case (ii) If l = 2, then equations (I) become

Case (iii) If l = 3, then equations (I) become

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 62 5/30/2016 5:03:40 PM

The characteristic equation of A is A − l I = 0

∴ the characteristic equation is l3 − 7l2 + 11l − 5 = 0

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 63 5/30/2016 5:03:42 PM

Choose the integer factors of constant term −5 for trial.

To ﬁnd eigen vectors:

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 64 5/30/2016 5:03:43 PM

We shall ﬁnd one more solution from x1 + 2x2 + x3 = 0

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 65 5/30/2016 5:03:45 PM

To ﬁnd eigen vectors:

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 66 5/30/2016 5:03:48 PM

1.6.4 Properties of Eigen Values

We know ( A − l I)T = A T − (l I)T [{ ( A + B)T = A T + BT ]

For any square matrix B, BT = B

From (2) and (3), A − lI = AT − lI .

⇒ l n − S1l n −1 + S2 l n − 2 − ... + ( −1) n Sn = 0 (1)

where S1 = sum of the diagonal elements of A

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 67 5/30/2016 5:03:50 PM

Note Sum of the diagonal elements of A is called the trace of A.

3. Product of the eigen values of a square matrix A is equal to A .

Proof Let A be a square matrix of order n.

⇒ l n − S1l n −1 + S2 l n − 2 − ... + ( −1) n Sn = 0 (1)

∴ the product of the eigen values = l1 l 2 … l n = Sn = A . ■

Note If at least one eigen value is 0, then A = 0 ∴ A is a singular matrix.

M01_ENGINEERING_MATHEMATICS-I _CH01_Part B.indd 68 5/30/2016 5:03:53 PM

5. If l1, l2, ???, ln are the eigen values of A, then