Reduction Of

Quadratic Form
To
Canonical Form

1
 Quadratic Form

Definition: The quadratic form in n variables 𝑥1 , 𝑥2. , … . 𝑥𝑛 is the general homogenous

function of second degree in the variables
i.e. Y=𝑓(𝑥1 , 𝑥2. , … . 𝑥𝑛 )= σ𝑛𝑖,𝑗=1 𝑎𝑖𝑗 𝑥𝑖 𝑥𝑗
In terms of matrix notation , the quadratic form is given by
𝑎11 𝑎12 … 𝑎1𝑛 𝑥1
𝑇
𝑎21 𝑎22 … 𝑎2𝑛 𝑥2
Y=𝑋 𝐴𝑋 = 𝑥1 𝑥2 … 𝑥𝑛 … … … … …
𝑎𝑛1 𝑎𝑛2 … 𝑎𝑛𝑛 𝑥𝑛

2
 In Matrix Notation

In two variables (𝑥,𝑦): 𝑎𝑥 2 +2ℎ𝑥𝑦 + 𝑏𝑦 2

𝑎 ℎ 𝑥
𝑄 𝑥, 𝑦, 𝑧. = 𝑋 𝑇 𝐴𝑋
= 𝑥 𝑦
ℎ 𝑏 𝑦
Here A is known as matrix of quadratic form

3
 In three variables

𝑄(𝑥, 𝑦, 𝑧)=𝑎𝑥 2 + 𝑏𝑦 2 + 𝑐𝑧 2 + 2ℎ𝑥𝑦 + 2𝑓𝑦𝑧 + 2𝑔𝑧𝑥

In matrix notation
𝑎 ℎ 𝑔 𝑥
𝑄 𝑥, 𝑦, 𝑧 = 𝑋 𝑇 𝐴𝑋= 𝑥 𝑦 𝑧 ℎ 𝑏 𝑓 𝑦
𝑔 𝑓 𝑐 𝑧

4
 To write matrix of quadratic form in three variables

𝑄 𝑥, 𝑦, 𝑧 = 𝑎𝑥 2 + 𝑏𝑦 2 + 𝑐𝑧 2 + 2ℎ𝑥𝑦 + 2𝑓𝑦𝑧 + 2𝑔𝑧𝑥

1 1
𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑥 2 ) 2
𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑥𝑦) 2
𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑥𝑧)
1 1
A= 2 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑥𝑦) 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑦 2 ) 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑦𝑧)
2
1 1
𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑥𝑧) 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑦𝑧) 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡(𝑧 2 )
2 2

5
 Ex. Obtain the matrix of the Quadratic form
Q= 𝑥 2 + 2 𝑦 2 − 7𝑧 2 − 4𝑥𝑦+8𝑥𝑧 + 5𝑦𝑧

Sol. Compare the given quadratic form by standard form

𝑄 𝑥, 𝑦, 𝑧 = 𝑎𝑥 2 + 𝑏𝑦 2 + 𝑐𝑧 2 + 2ℎ𝑥𝑦 + 2𝑓𝑦𝑧 + 2𝑔𝑧𝑥

5
Here 𝑎 = 1 , 𝑏 = 2 , 𝑐 = −7; ℎ = −2 ; 𝑔 = 4 ; 𝑓 = 2
Therefore coefficient matrix is
1 −2 4
𝑎 ℎ 𝑔 5
−2 2
𝐴= ℎ 𝑏 𝑓 = 2
𝑔 𝑓 𝑐 5
4 −7
2
Which is a symmetric matrix.

6
 Ex. Obtain the matrix of the Quadratic form
Q= 2𝑥 2 + 3 𝑦 2 + 𝑧 2 − 3𝑥𝑦+2𝑥𝑧 + 4𝑦𝑧

3
𝑎 ℎ 𝑔 2 − 1
2
Ans. 𝐴= ℎ 𝑏 𝑓 = 3
𝑔 𝑓 𝑐 − 3 2
2
1 2 1

7
 EX. Write down the quadratic form corresponding to the
matrix
1 2 5
𝐴= 2 0 3
5 3 4 𝐴=
𝑥
𝑆𝑜𝑙. Quadratic form 𝑖𝑠 𝑋 𝑇 𝐴𝑋 𝑤ℎ𝑒𝑟𝑒 𝑋 = 𝑦
𝑧
1 2 5 𝑥
𝑋 𝑇 𝐴𝑋 = 𝑥 𝑦 𝑧 2 0 3 𝑦
5 3 4 𝑧

or 𝑄 𝑥, 𝑦, 𝑧 = 𝑥 2 + 0𝑦 2 + 4𝑧 2 + 4𝑥𝑦 + 6𝑦𝑧 + 10𝑧𝑥

is the required Quadratic form.

8
EX. Write down the quadratic form corresponding to the matrix
3 2 4
𝐴= 2 0 4
4 4 3

𝑥
Sol. Quadratic form = 𝑋 𝑇 𝐴𝑋 𝑤ℎ𝑒𝑟𝑒 X = 𝑦
𝑧
3 2 4 𝑥
𝑋 𝑇 𝐴𝑋 = 𝑥 𝑦 𝑧 2 0 4 𝑦
4 4 3 𝑧

or 𝑄 𝑥, 𝑦, 𝑧 = 3𝑥 2 + 0𝑦 2 + 3𝑧 2 + 4𝑥𝑦 + 8𝑦𝑧 + 8𝑧𝑥

is the required Quadratic form.

9
 CANONICAL FORM OR
SUM OF SQUARE FORM

The sum of square form of a real quadratic form 𝑋 𝑇 𝐴𝑋 is

𝒀𝑻 𝑫𝒀 = 𝝀𝟏 𝒚𝟐𝟏 + 𝝀𝟐 𝒚𝟐𝟐 +…… +𝝀𝒏 𝒚𝟐𝒏
is known as canonical form which is formed with the help of the
orthogonal transformation 𝑋 = 𝐵𝑌 where B is the modal matrix of
𝑦1
𝑦2
normalized eigen vectors , Y= … and D is a diagonal matrix whose
𝑦𝑛
diagonal elements are the eigen values of the matrix A.

10
 Working Method

1. First write the coefficient matrix of the given quadratic

form.
2. Find the eigen values and eigen vector of the coefficient
matrix A.
3. Now check whether eigen vectors are pairwise
orthogonal or not.
Orthogonal Vector; Two vectors 𝑋1 , 𝑋2 are said to be
orthogonal if their inner product is zero i.e. 𝑋1 𝑇 . 𝑋2 = 0

11
4. Form a modal matrix B of normalized eigen
vector
i.e. B = 𝑋1 𝑋2 𝑋3
𝑥1
𝑥1 2 + 𝑥2 2 + 𝑥3 2
𝑥1 𝑥2
If X1 = 𝑥2 then 𝑋1 =
𝑥3 𝑥1 2 + 𝑥2 2 + 𝑥3 2
𝑥3
𝑥1 2 + 𝑥2 2 + 𝑥3 2
Similarly for 𝑋2 and 𝑋3

Here B is matrix of transformation

12
 5. Here B matrix is an orthogonal matrix
∴ 𝐵𝐵𝑇 = 𝐼
∵ 𝐵𝑇 =𝐵 −1
Find 𝐵 −1 𝐴𝐵
𝐵−1 𝐴𝐵= 𝐷 (diagonal matrix whose diagonal elements are
eigen values of the matrix A)
The required canonical form is
𝜆1 0 0 𝑦1
𝑌 𝑇 (𝐵 −1 𝐴𝐵) 𝑌= 𝑌 𝑇 𝐷 𝑌= 𝑦1𝑋 𝑦2 𝑦3 0 𝜆2 0 𝑦2
0 0 𝜆3 𝑦3
i.e. 𝜆1 𝑦1 2 + 𝜆2 𝑦2 2 + 𝜆3 𝑦3 2 is the required canonical form and
𝑋=𝐵𝑌 is required orthogonal transformation .

13
Ex. Reduce the quadratic form
Q(𝑥, 𝑦, 𝑧)=𝑥12 + 3𝑥22 + 3𝑥32 − 2𝑥2 𝑥3 into canonical
form using orthogonal transformation.
Sol. Write the coefficient matrix of given quadratic
form
1 0 0
𝐴 = 0 3 −1
0 −1 3

The eigen values of A are 1,2,4

14
Eigen Vectors for 𝜆 = 1
𝐴 − 𝜆1 𝐼 𝑋1 = 𝟎 𝑜𝑟 𝐴 − 𝐼 𝑋1 = 𝟎

0 0 0 𝑥1 0
or 0 2 −1 𝑥2 = 0
0 −1 2 𝑥3 0

0 0 0 𝑥1 0
or 0 2 −1 𝑥2 = 0 Applying 𝑅3 → 𝑅3 + 1 𝑅2
3 2
0 0 𝑥3 0
2

2𝑥2 − 𝑥3 = 0. . . . . . . (1)
𝑥3 = 0. . . . . . . . . . . . . (2)
Here no. of equations (m) are 2 and no. of
variables (n) are 3
n−m = 3−2 = 1 variable is independent.
Let 𝑥1 = k
𝑥3 = 0; 𝑥2 = 0
𝑥1 = 𝑘
15
 𝑥1 𝑘
𝑋1 = 𝑥2 = 0
𝑥3 0
If 𝑘 = 1 𝑡ℎ𝑒𝑛 𝑋1
1
= 0 is the eigen vector
0
corresponding to 𝜆 = 1

16
Eigen vector for 𝜆 = 2
𝐴 − 𝜆2 𝐼 𝑋2 = 𝟎 or 𝐴 − 2𝐼 𝑋2 = 𝟎
−1 0 0 𝑥1 0
0 1 −1 𝑥2 = 0
0 −1 1 𝑥3 0
Applying 𝑅3 → 𝑅3 + 𝑅2
−1 0 0 𝑥1 0
0 1 −1 𝑥2 = 0
0 0 0 𝑥3 0
𝑥1 = 0. . . . . . . (1)
𝑥2 − 𝑥3 = 0. . . . . . . . . . . . . (2)
Here no. of equations (m) are 2 and no. of
variables (n) are 3
n−m = 3−2 = 1 variable is independent.
Let 𝑥3 = k
𝑥2 = 𝑥3 ,
𝑥3 = 𝑘 = 𝑥2

17
 𝑥1 0
𝑋2 = 𝑥2 = 𝑘
𝑥3 𝑘
0
If 𝑘 = 1 𝑡ℎ𝑒𝑛 𝑋2 = 1
1
is the eigen vector
corresponding to 𝜆 = 2

18
Eigen Vector for 𝜆 = 4
𝐴 − 𝜆3 𝐼 𝑋3 = 𝟎 or 𝐴 − 4𝐼 𝑋3 = 𝟎
−3 0 0 𝑥1 0
or 0 −1 −1 𝑥2 = 0
0 −1 −1 𝑥3 0
𝐴𝑝𝑝𝑙𝑦𝑖𝑛𝑔 R 3 → R 3 − R 2
−3 0 0 𝑥1 0
0 −1 −1 𝑥2 = 0
0 0 0 𝑥3 0
𝑥1 = 0; 𝑥2 + 𝑥3 = 0

Here no. of equations are (m) 2 and

no. of variables (n) are 3
n−m = 3−2 = 1 variable is independent.
∴ Let 𝑥3 = k
𝑥2 = −𝑥3
𝑥2 = −𝑘; 𝑥1 = 0

19
 𝑥1 0
𝑋3 = 𝑥2 = −𝑘
𝑥3 𝑘
0
𝐼𝑓 𝑘 = 1 𝑡ℎ𝑒𝑛 𝑋3 = −1
1
is the eigen vector
corresponding to 𝜆 = 4

20
Continued.

Eigen vector corresponding to matrix A

1 0 0
𝑋1 = 0 , 𝑋2 = 1 , 𝑋3 = −1
0 1 1
Now to check whether 𝑋1 , 𝑋2 , 𝑋3 are pairwise orthogonal.
Pairwise Orthogonal –Two vectors 𝑋1 , 𝑋2 are said to be orthogonal
if their inner product i.e. 𝑋1𝑇 . 𝑋2 =0

T
1 0  0 
X 1 . X 2 = 0 1 = 1 0 01 = 0
T
   
0 1 1
Therefore X1 , X 2 are pairwise orthogonal.
Similarly we can check for others vectors also.
Here X1 , X 2 , X 3 are pairwise orthogonal
21
Now we form normalized modal matrix
1 0 0 1 0 0
1 1 1 1
0 − 0
𝐵= 2 2 , 𝐵 −1 = 2 2
1 1 1 1
0 0 −
2 2 2 2

1 0 0 1 0 0
1 1 1 0 0 1 1
0 0 −
𝐵 −1 𝐴𝐵 = 2 2 0 3 −1 2 2
1 1 0 −1 3 1 1
0 − 0
2 2 2 2

22
 1 0 0
𝐵 −1 𝐴𝐵 = 0 2 0
0 0 4

i.e. 𝐵 −1 𝐴𝐵=D

where D is a diagonal matrix whose diagonal

elements are the eigen values of the matrix of
quadratic form.

23
 The canonical Form

𝑌 𝑇 (𝐵−1 𝐴𝐵)𝑌 = 𝑌 𝑇 𝐷𝑌
1 0 0 𝑦1
= 𝑦1 𝑦2 𝑦3 0 2 0 𝑦2
0 0 4 𝑦3
= 𝑦1 2 + 2𝑦2 2 + 4𝑦3 2

𝑦1 2 + 2𝑦2 2 + 4𝑦3 2
is the required canonical form.

24
Orthogonal
Transformation
𝑋 = 𝐵𝑌
1 0 0
𝑥1 1 1 𝑦1
0 −
𝑥2 = 2 2 𝑦2
𝑥3 1 1 𝑦3
0 𝑋 = 𝐵𝑌
𝑥1
𝑥2 =
𝑥3
2 2
𝑥1 = 𝑦1 ,
1 1
𝑥2 = 𝑦2 − 𝑦3 ,
2 2
1 1
𝑥3 = 𝑦2 + 𝑦3
2 2
is the required
orthogonal transformation.

25
Reduce 6𝑥 2 + 3𝑦 2 + 3𝑧 2 − 4𝑥𝑦 − 2𝑦𝑧 + 4𝑥𝑧 into
canonical form by orthogonal transformation.
Sol. Write the coefficient matrix of given quadratic
form
6 −2 2
𝐴 = −2 3 −1
2 −1 3

The eigen values of A are 2,2,8

−1 1
The eigen vectors corresponding to 2, 2, 8 are 0 , 2
2 0
2
and −1 respectively.
1
26
 −1 1 2
𝑋1 = 0 , 𝑋2 = 2 , 𝑋3 = −1
2 0 1

Here 𝑋1 , 𝑋2 , are not pairwise orthogonal

Let 𝑋1 is a vector which is orthogonal to 𝑋2 and 𝑋3
𝑎 1
𝑇
Let 𝑋1 = 𝑏 than 𝑋1 . 𝑋2 = 0 or 𝑎 𝑏 𝑐 2 =0
𝑐 0
or a+2b=0……(1)
2
similarly 𝑋1𝑇 . 𝑋3 = 0 or 𝑎 𝑏 𝑐 −1 =0 or 2a - b + c=0…..(2)
1
Assuming b =𝑘 , 𝑎𝑛𝑑 𝑠𝑜𝑙𝑣𝑖𝑛𝑔 𝑤𝑒 𝑔𝑒𝑡
−2𝑘 −2
𝑎 = −2𝑘 𝑎𝑛𝑑 𝑐 = 5𝑘 or 𝑋1 = 𝑘 or if 𝑘 =1 than 𝑋1 = 1
5𝑘 5
Now 𝑋1 , 𝑋2 , and 𝑋3 are pairwise orthogonal .
Now rest of the process is same as in previous example.

27
 INDEX ,SIGNATURE,RANK
Let 𝑄 = 𝑋 𝑇 𝐴𝑋
be a quadratic form in n variables 𝑥1 , 𝑥2 , . . . . . . . . . . . 𝑥𝑛 .

INDEX ∶ The no. of positive terms (p) in the canonical form.

Signature: The difference between positive and negative terms in the canonical form.

Rank: A matrix is said to be of rank r when

(i) it has at least one non-zero minor of order r,
and (ii) every minor of order higher than r vanishes.
Briefly, the rank of a matrix is the largest order of any non-vanishing minor of
the matrix.

28
Determine the rank of the following matrices:

𝟏 𝟐 𝟑
(i)A= 𝟏 𝟒 𝟐
𝟐 𝟔 𝟓

Sol. (i)Here A is a 3 x3 matrix

∴ 𝜌(𝐴) ≤ 3
After operating 𝑅2 → 𝑅2 - 𝑅1 , 𝑅3 → 𝑅3 −2𝑅1
𝟏 𝟐 𝟑
A~ 𝟎 𝟐 −𝟏
𝟎 𝟐 −𝟏
Operating 𝑅3 → 𝑅3 - 𝑅2

𝟏 𝟐 𝟑
A~ 𝟎 𝟐 −𝟏
𝟎 𝟎 𝟎
The only third order minor is zero, but the second order minor
1 2
= 2 ≠0
0 2
∴𝜌 𝐴 =2

29
 𝟏 𝟐 𝟑 𝟒
(ii) B=
−𝟐 𝟎 𝟓 𝟕

Sol. (ii)Here B is a 2 x 4 matrix.

∴ 𝜌(𝐴) ≤ 2, the smaller of 2 and 4.
1 2
The second order minor =4≠ 0
−2 0
∴𝜌 𝐴 =2

30
 NATURE OF QUADRATIC FORM
( using rank, signature)
 Let 𝑋 𝑇 𝐴𝑋 be a real quadratic form in n
variables 𝑥1, 𝑥2, …. 𝑥𝑛 with rank r and signature s.
 Then we say that the quadratic form is
(i) positive definite if r = n, s = n
(ii) negative definite if r = n, s = 0
(iii) positive semidefinite if r<n and s = r
(iv) negative semidefinite if r<n, s = 0
(v) indefinite in all other cases.

31
NATURE OF QUADRATIC FORM
(using eigen values)

 POSITIVE DEFINITE: If all the eigen values of

the coefficient matrix are POSITIVE.
 POSITIVE SEMI DEFINITE: If all the eigen
values of the coefficient matrix are POSITIVE and
at least one is zero.
 NEGATIVE DEFINITE: If all the eigen values
of the coefficient matrix are NEGATIVE.
 NEGATIVE SEMI DEFINITE: If all the eigen
values of the coefficient matrix are NEGATIVE
and
at least one is zero.
➢ INDEFINITE: If some of the eigen values are
POSITIVE and some are NEGATIVE.
32
 EX. Determine the nature ,index and signature of the quadratic form
𝑄(𝑥) = 2𝑥1 𝑥2 + 2𝑥1 𝑥3 + 2𝑥3 𝑥2
Solution. Here
0 1 1
𝐴= 1 0 1
1 1 0
Now the characteristic equation for A is
𝜆3 − 3𝜆 − 2 = 0
or 𝜆 = 2, −1, −1
Some of the eigen values are positive and
some are negative.
Hence 𝐐(𝐱) is indefinite.
Here the canonical form is 2𝑥12 − 𝑥22 − 𝑥32 .
Index =1 No. of positive term in canonical form
Signature=-1
(difference of positive and negative term in canonical form )

33