Professional Documents
Culture Documents
54 ■ Engineering Mathematics
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
(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.
(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.
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.
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
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.
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.
l1 (a1X1 ) + l1 (a 2 X 2 ) = 0
∴ ( l 2 − l1 ) X 2 ≠ 0
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
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.
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⎦
The characteristic equation of A is A − lI = 0
a−l b
⇒ =0 ⇒ l 2 − S1l + S2 = 0
−b a−l
⎡ 1⎤
Choosing x1 = 1, we get x2 = −i ∴ an eigen vector is X 2 = ⎢ ⎥
⎣ −i ⎦
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 ⎦⎥
The characteristic equation of A is A − lI = 0
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]
(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⎪⎭
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 ⎥⎦
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
⎡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⎥⎦
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
⎡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 ⎥⎦
⎣⎢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 ⎥⎦
The characteristic equation of A is A − lI = 0
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.
⎡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.
A − lI = 0 (1)
= A T − l IT = A T − l I [{ I T = I]
∴ ( A − l I)T = A T − l I (2)
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]
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.
This is true for all eigen values of A.
∴ 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.
This is true for all eigen values of A.
∴ 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⎥⎦
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
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 ⎥⎦
The characteristic equation of A is A − lI = 0
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.
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⎥⎦
⎡ −5 −5 −9⎤ ⎡ 2 1 1⎤
9. ⎢⎢ 8 9 18⎥⎥ 10. ⎢⎢1 2 1⎥⎥
⎢⎣ −2 −3 −7⎥⎦ ⎢⎣0 0 1⎥⎦
⎡ 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 ⎥⎦
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)
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
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
⎡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⎦
where S1 = 1 + 3 = 4, S2 = A = 3 − 8 = −5
∴ A2 − 4A − 5I = 0 (2)
⎡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
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
⎡ 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
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⎥⎦
⇒ 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
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
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 ⎦
⎛ 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⎦
( 5) a 5 + b = ( 5)
n n
Substitute l = 5 in (2) then = 0+a 5+b ⇒ (3)
−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
⎣
⎡ 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⎥⎦
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 ⎣ ⎦ ⎣ ⎦
⎡ 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⎥⎦
⇒ 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
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 ⎥⎦
⎣
⎡ 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.
⎢ : : : : :⎥
⎢ ⎥
⎣ 0 0 … … ln ⎦
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
⎣ − sin u cos u⎦
∴ 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.
⇒ 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.
⇒ X T2 AX1 = l1 ( X T2 X1 ) (3)
⇒ 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
⇒ (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.
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.
⎢ : … …⎥
⎢ ⎥
⎣ 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
To find eigen vectors:
⎡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 ⎪⎭
⎡ −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⎦
⎡ 1 1 1 ⎤
⎢ − ⎥
⎢ 2 3 6⎥
⎢ 1 2 ⎥
So, the normalised modal matrix is N = ⎢ 0 ⎥
⎢ 3 6⎥
⎢ 1 1 1 ⎥
⎢− − ⎥
⎣ 2 3 6⎦
⎡2 0 0⎤
N AN = D = ⎢0 3 0 ⎥⎥ , since 2, 3, 6 are the eigen values.
T ⎢
⎢⎣0 0 6 ⎥⎦
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
( 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 ⎪⎭
⎡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 ⎥⎦
⎢ …⎥ ⎢… … … …⎥
⎢ ⎥ ⎢ ⎥
⎣x n ⎦ ⎣an1 an2 … ann ⎦
form
n n
Q = ∑ ∑ aij x ij can be written as Q = XT AX.
j =1 i =1
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
⎡ l1 0 0 … 0⎤
⎢ 0 l 0 … 0 ⎥⎥
D= ⎢ 2
⎢ … … … … …⎥
⎢ ⎥
⎣ 0 0 0 … ln ⎦
⎡ 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 ⎦
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.
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 ⎦
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.
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.
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)
∴
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.
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⎥⎦
S3 = A = 6(9 − 1) + 2( −6 + 2) + 2( 2 − 6) = 48 − 8 − 8 = 32
⎡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 ⎦
⎡ 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 ⎥⎦
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⎦
⎡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 ⎭
Case (i) If l = 2, then equations (I) become
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 ⎦
y12 y22
∴ 2 y12 + 32 y22 = 128 ⇒ + = 1, which represents an ellipse.
64 4
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
⇒ 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 ⎥⎦
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
∴ x = 1, y = 2, z = 2
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⎥⎦
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
⎡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
⎡ 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⎦
⎡ 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 ⎥⎦
(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 .
⎡ 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⎥⎦
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
⎣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.
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.
⎡1 5⎤ ⎡ −2 5⎤
2. If A − 2B = ⎢ ⎥ , 2A − 3B = ⎢ 0 7⎥ , then B = ________.
⎣ 3 7 ⎦ ⎣ ⎦
⎡ 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 ⎦
12. The nature of the quadratic form 2xy + 2yz + 2zx is ________.
⎡ cos u sin u ⎤
14. The matrix A = ⎢ ⎥ is ________. 15. If A is an orthogonal matrix, then A−1 is ________.
⎣ − sin u cos u⎦
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
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⎦
⎡ 3 −1 3 ⎤
⎢ ⎥
13. ⎢ −1 3 −3⎥ 14. orthogonal 15. A −1 = A T
⎢⎣ −3 −3 −5⎥⎦