QUADRATIC FORMS

Definition: A Homogeneous polynomial of the second degree in any number

of variables is called a Quadratic Form.

Example: 𝑥 2 + 4𝑥𝑦 − 2𝑦 2 , 𝑥12 − 3𝑥1 𝑥2 + 2𝑥22 are quadratic forms.

Canonical Form of a Quadratic Form:

The Quadratic Forms Q=𝑋 𝑇 𝐴𝑋 in the variables 𝑥1 , 𝑥2 , 𝑥3 , 𝑥4 , … 𝑥𝑛 can be

reduced to a quadratic form 𝑑1 𝑦12 + 𝑑2 𝑦22 + 𝑑3 𝑦32 + 𝑑4 𝑦42 + ⋯ +𝑑𝑛 𝑦𝑛2 which is
called the Canonical Form (or) Sum of squares in the form of Q.

Nature of Quadratic Forms:

The quadratic form is said to be

(i) Positive definite if all the eigenvalues of A are positive numbers.

(ii) Negative definite if all the eigenvalues of A are negative

numbers.

(iii) Positive Semi-definite if all the eigenvalues of A are positive and

atleast one of the eigenvalue is zero.

(iv) Negative Semi-definite if all the eigenvalues of A are Negative

and atleast one of the eigenvalue is zero.

(v) Indefinite in all other cases.

Index of the Quadratic Form:

The number of Positive square terms in the canonical form is called the
index(p) of the Quadratic Form.

Signature of the Quadratic Form:

The Signature(S) of the Quadratic Form S= 2p-r

Reduction of Quadratic Form by means of Orthogonal reduction:

 The transformation X=NY where y is a column vector of new variables
𝑦1 , 𝑦2 , 𝑦3 , 𝑦4 , … 𝑦𝑛 reduces the given quadratic form Q into 𝑌 𝑇 𝐷𝑌, D is a Diagonal
Matrix. The resulting quadratic form takes the form 𝑌 𝑇 𝐷𝑌.

(i.e) 𝛼1 𝑦12 + 𝛼2 𝑦22 + 𝛼3 𝑦32 + ⋯ 𝛼𝑛 𝑦𝑛2 is called the Canonical Form.

Nature of Quadratic Forms through Principal Minors:

Let A=aijbe the matrix of the given quadratic form,

𝑎11 𝑎12
Let D1=|𝑎11 |D2=|𝑎 𝑎22 |
21

𝑎11 𝑎12 𝑎13

D3=|𝑎21 𝑎22 𝑎23 | and ……Dn = |𝐴|
𝑎31 𝑎32 𝑎33

As D1,D2, D3,…..Dnare the Principal minors of A.

Then the Quadratic Form is

(i) Positive definite: If D1,D2, D3,…..Dnare all positive if Dn>0.

(ii) Negative definite: If D1, D3, D5,….. are all negative and if D2, D4,
D6,….. are all positive.
(iii) Positive semi definite: If Dn≥0 and atleast one Di=0.
(iv) Negative semi definite: If (-1)nDn≥0.
(v) Indefinite in all other cases.

Example 1:

By an orthogonal transformation, reduce the Quadratic Form 3𝑥 2 − 2𝑦 2 − 𝑧 2 +

12𝑦𝑧 + 8𝑧𝑥 − 4𝑥𝑦 to a Canonical Form. Also find it’s nature.

Solution:

The given quadratic form is, 3𝑥 2 − 2𝑦 2 − 𝑧 2 + 12𝑦𝑧 + 8𝑧𝑥 − 4𝑥𝑦

3 −2 4
The matrix A of the Quadratic Form is, 𝐴 = [−2 −2 6]
4 6 −1

The Characteristic equation is λ3−𝑆1 λ2 + 𝑆2 λ − 𝑆3 = 0

Where S1 = 0; S2 = -63; S3= -162

Hence the Characteristic equation becomes, λ3 − 0λ2 − 63λ + 162 = 0

 λ3 − 63λ + 162 = 0

λ = 𝟑, 𝟔, −𝟗are the Eigen values.

To Find Eigen Vectors:

Case (i): Ifλ= 𝟑

If λ = 3 then (A- λ𝐼)𝑋 = 0

3 −2 4 1 𝑥1
[(−2 −2 6 ) − 3 [1]] (𝑥2 ) = 0
4 6 −1 1 𝑥3

0 −2 4 𝑥1
(−2 −5 6 ) ( 𝑥2 ) = 0
4 6 −4 𝑥3

Hence the equation from the above matrix are,

0𝑥1− 2𝑥2 + 4𝑥3= 0

−2𝑥1 − 5𝑥2 + 6𝑥3= 0
4𝑥1 +6𝑥2 − 4𝑥3= 0

By the multiplication rule from equations 1 and 2,

𝑥1 𝑥2 𝑥3
= =
−12 + 20 −8 + 0 0 − 4

x1 x2 x3
= = = k(say as a constant)
8 −8 −4

x1 x2 x3
(i. e) = = =k
2 −2 −1

Hence, 𝑥1 = 2𝑘; 𝑥2 = −2𝑘; 𝑥3 = −𝑘

 𝒙𝟏 2k
As, X = (𝒙𝟐 ) (i.e) X = (−2k)
𝒙𝟑 −k

2
If, k=1, then X = (−2)
−1
2
So, X1 = (−2)is an Eigen vector.
−1

Case (ii): If λ = 𝟔

If λ = 6 then (A- λ𝐼)𝑋 = 0 becomes as

3 −2 4 1 𝑥1
[(−2 −2 6 ) − 6 [1]] (𝑥2 ) = 0
4 6 −1 1 𝑥3

−3 −2 4 𝑥1
(−2 −8 6 ) ( 𝑥2 ) = 0
4 6 −7 𝑥3

Hence the equation from the above matrix are,

-3𝑥1− 2𝑥2 + 4𝑥3= 0

−2𝑥1 − 8𝑥2 + 6𝑥3= 0
4𝑥1 +6𝑥2 − 7𝑥3= 0
By the multiplication rule from equations 1 and 2,

𝑥1 𝑥2 𝑥3
= =
−12 + 32 −8 + 18 24 − 4

x1 x2 x3
= = = k(say as a constant)
20 10 20

x1 x2 x3
(i. e) = = =k
2 1 2

Hence, 𝑥1 = 2𝑘; 𝑥2 = 𝑘; 𝑥3 = 2𝑘
 𝒙𝟏 2k
As, X = ( 𝟐 ) (i.e) X = ( k )
𝒙
𝒙𝟑 2k

2
If, k=1, then X = (1)
2
2
So, X2 = (1)is an Eigen vector.
2

Case (iii): If λ = −𝟗

If λ = −9 then (A- λ𝐼)𝑋 = 0 becomes as

3 −2 4 1 𝑥1
[(−2 −2 6 ) + 9 [1]] (𝑥2 ) = 0
4 6 −1 1 𝑥3

12 −2 4 𝑥1
(−2 7 6) ( 𝑥 2 ) = 0
4 6 8 𝑥3

Hence the equation from the above matrix are,

12𝑥1− 2𝑥2 + 4𝑥3= 0

−2𝑥1 + 7𝑥2 + 6𝑥3= 0
4𝑥1 +6𝑥2 + 8𝑥3= 0

By the multiplication rule from equations 1 and 2,

𝑥1 𝑥2 𝑥3
= =
−12 − 28 −8 − 72 84 − 4

x1 x2 x3
= = = k(say as a constant)
−40 −80 80

x1 x2 x3
(i. e) = = =k
−1 −2 2
 Hence, 𝑥1 = −𝑘; 𝑥2 = −2𝑘; 𝑥3 = 2𝑘

𝒙𝟏 −k
As, X = (𝒙𝟐 ) (i.e) X = (−2k)
𝒙𝟑 2k

−1
If, k=1, then X = (−2)
2
−1
So, X3 = (−2)is an Eigen vector.
2
2 2 −1
Hence X1 = (−2), X2 = (1) , X3 = (−2)are theEigen vectors.
−1 2 2

To check for Pairwise orthogonality:

𝑋1𝑇 𝑋2 = 𝑋2𝑇 𝑋3 = 𝑋3𝑇 𝑋1 = 0

𝐻ence X1, X2 and X3 are pairwise orthogonal.

To find Normalised Modal Matrix (N)

𝒙𝟏
As we know if X = 𝒙𝟐 ) then the norm of X is,
(
𝒙𝟑

‖𝑥 ‖ = √𝑥12 + 𝑥22 + 𝑥32

2
So for, X1 = (−2), then
−1
‖X1‖ = √𝑥12 + 𝑥22 + 𝑥32

‖X1‖ = √4 + 4 + 1 = 3

So normalized matrix for X1 becomes as

 𝟐
𝟑
−𝟐
N1 = 𝟑
−𝟏
(𝟑)
𝟐 𝟏
𝟑 𝟑
𝟏 𝟐
SimilarlyN2 = ; N3 =
𝟑 𝟑
𝟐 −𝟐
( 𝟑) (𝟑)

As
‖X2‖ = √4 + 1 + 4 = 3

‖X3‖ = √4 + 4 + 1 = 3

𝟐𝟐𝟏
𝟑𝟑𝟑
−𝟐 𝟏 𝟐
Hence N= is the Normalised modal matrix which is orthogonal as
𝟑𝟑𝟑
−𝟏 𝟐−𝟐
(𝟑𝟑𝟑)
NN =N N=I
T T

𝟐𝟐𝟏 𝟐𝟐𝟏
𝟑𝟑𝟑 3 −2 4 𝟑𝟑𝟑 3 0 0
−𝟐𝟏 𝟐 −𝟐𝟏 𝟐
So, N AN =
T (−2 −2 6 ) = (0 6 0 ) = D Hence
𝟑𝟑𝟑 𝟑𝟑𝟑
−𝟏𝟐−𝟐 4 6 −1 −𝟏𝟐−𝟐 0 0 −9
(𝟑𝟑𝟑) (𝟑𝟑𝟑)
[ ]
NTAN=D(3,6,-9,), where D is a diagonal matrix with 3,6,-9 as the principal
diagonal elements.

Therefore the Orthogonal transformation X=NY will reduce the Quadratic

Form to the Canonical form as 3𝑦12 + 6𝑦22 − 9𝑦32 .

The nature of the Quadratic form as,

Rank = 3 (Three non-zero rows = r)

Index= 2 (Number of Positive terms=p)

Signature = 1 (2p-r)

Indefinite in nature( Both positive and Negative Eigen values)

Example 2:

By an orthogonal transformation, reduce the Quadratic Form 6𝑥 2 + 3𝑦 2 +

3𝑧 2 − 4𝑥𝑦 − 2𝑦𝑧 + 4𝑥𝑧 to a Canonical Form. Also find it’s nature.

Solution:

The given quadratic form is, 6𝑥 2 + 3𝑦 2 + 3𝑧 2 − 4𝑥𝑦 − 2𝑦𝑧 + 4𝑥𝑧

6 −2 2
The matrix A of the Quadratic Form is, 𝐴 = [−2 3 −1]
2 −1 3

The Characteristic equation is λ3 − 𝑆1 λ2 + 𝑆2 λ − 𝑆3 = 0

Where S1 =12; S2 = 36; S3= 32

Hence the Characteristic equation becomes, λ3 − 12λ2 + 36λ − 32 = 0

λ = 𝟐, 𝟐, 𝟖are the Eigen values.

To Find Eigen Vectors:

Case (i): If λ = 𝟖

If λ = 8 then (A- λ𝐼)𝑋 = 0

6 −2 2 1 𝑥1
[(−2 3 −1) − 8 [1]] (𝑥2 ) = 0
2 −1 3 1 𝑥3

−2 −2 2 𝑥1
(−2 −5 −1) (𝑥2 ) = 0
2 −1 −5 𝑥3

Hence the equation from the above matrix are,

-2𝑥1− 2𝑥2 + 2𝑥3= 0

−2𝑥1 − 5𝑥2 − 𝑥3= 0
2𝑥1 −𝑥2 − 5𝑥3= 0

By the multiplication rule from equations 1 and 2,

 𝑥1 𝑥2 𝑥3
= =
2 + 10 −4 − 2 10 − 4

x1 x2 x3
= = = k(say as a constant)
12 −6 6

x1 x2 x3
(i. e) = = =k
2 −1 1

Hence, 𝑥1 = 2𝑘; 𝑥2 = −𝑘; 𝑥3 = 𝑘

𝒙𝟏 2k
As, X = (𝒙𝟐 ) (i.e) X = (−k)
𝒙𝟑 k

2
If, k=1, then X = (−1)
1
2
So, X1 = (−1)is an Eigen vector.
1

Case (ii): If λ = 𝟐

If λ = 2 then (A- λ𝐼)𝑋 = 0

6 −2 2 1 𝑥1
𝑥
[(−2 3 −1) − (2) [1]] ( 2 ) = 0
2 −1 3 1 𝑥3
4 −2 2 𝑥1
[(−2 1 −1)] (𝑥2 ) = 0
2 −1 1 𝑥3

Hence the equation from the above matrix are,

4𝑥1− 2𝑥2 + 2𝑥3= 0
−2𝑥1 + 𝑥2 − 𝑥3= 0
2𝑥1− 𝑥2 + 𝑥3= 0

As all the above three equations are same as

 2𝑥1− 𝑥2 + 𝑥3= 0

An independent solutions can be obtained as follows,

Putting 𝑥2 = 1 𝑎𝑛𝑑 𝑥3 = 0 𝑤𝑒 𝑔𝑒𝑡 𝑥1 = 1

1
So let 𝑥2 = [2] arbitrarily,
0

Let us choose arbitrarily 𝑥3 such that it satisfies orthogonality

𝑎
condition. Let 𝑥3 = [𝑏]
𝑐
So 𝑥3 𝑥2 = 𝑥3 𝑥1= 0
𝑇 𝑇

𝑥3 𝑥2𝑇 = 0 𝑇
𝑥3 𝑥1= 0
𝑎 0 𝑎 0
(𝑏 ) (1 2 0 =(0)
)𝑇 (𝑏 ) (2 −1 1)𝑇 =(0)
𝑐 0 𝑐 0
2a-b+c=0
a+2b=0

Hence the two equations are,

a+2b+0c=0;
2a- b+ c=0

By the multiplication rule,

𝑥1 𝑥2 𝑥3
= =
2 + 0 0 − 1 −1 − 4

x1 x2 x3
= = = k(say as a constant)
2 −1 −5

Hence, 𝑥1 = 2𝑘; 𝑥2 = −𝑘; 𝑥3 = −5𝑘

𝒙𝟏 2k
As, X = ( 𝟐 ) (i.e) X = ( −k )
𝒙
𝒙𝟑 −5k
 2
If, k=1, then X = (−1)
−5
2
So, X3 = (−1)is an Eigen vector
−5
2 1 2
Hence X1 = (−1), X2 = (2) , X3 = (−1)are theEigen vectors satisfying
1 0 −5
orthogonal property.

To check for Pairwise orthogonality:

𝑋1𝑇 𝑋2 = 𝑋2𝑇 𝑋3 = 𝑋3𝑇 𝑋1 = 0

𝐻ence X1, X2 and X3 are pairwise orthogonal.

To find Normalised Modal Matrix (N)

𝒙𝟏
As we know if X = (𝒙𝟐 ) then the norm of X is,
𝒙𝟑

‖𝑥 ‖ = √𝑥12 + 𝑥22 + 𝑥32

2
So for, X1 = (−1), then
1
‖X1‖ = √𝑥12 + 𝑥22 + 𝑥32

‖X1‖ = √4 + 1 + 1 = √6

So normalized matrix for X1 becomes as

𝟐
√𝟔
−𝟏
N1 = √𝟔
𝟏
(√𝟔 )
 𝟐
𝟏
√𝟑𝟎
√𝟑 −𝟏
Similarly N2 = ( 𝟐 ); N3 = √𝟑𝟎
√𝟑
−𝟓
𝟎 (√𝟑𝟎)

As
‖X2‖ = √4 + 1 + 4 = √𝟑

‖X3‖ = √4 + 4 + 1 = √𝟑𝟎
𝟐 𝟐
𝟏
√𝟔 √𝟑𝟎
−𝟏 √𝟑 −𝟏
Hence N= 𝟐 is the Normalised modal matrix which is orthogonal as
√𝟔 √𝟑𝟎
𝟏 √𝟑 −𝟓
(√𝟔 𝟎 √𝟑𝟎)
NNT=NTN=I

8 0 0
So, NTAN = = (0 2 0) = D
0 0 2

Hence NTAN=D(8,2,2), where D is a diagonal matrix with 8,2,2 as the

principal diagonal elements.

Therefore the Orthogonal transformation X=NY will reduce the Quadratic

Form to the Canonical form as 8𝑦12 + 2𝑦22 + 2𝑦32 .

The nature of the Quadratic form as,

Rank = 3 (Three non-zero rows)

Index= 3 (Number of Positive terms)

Signature = 3

Positive definite in nature( Only positive Eigen values)