Inner Product Space

Definition:
Real Inner Product: Let V be a vector space. A real inner product on V is a
mapping f :V ×V → R that assigns to each ordered pair vectors (α , β ) of V a real
number f ( α , β ) , generally denoted by α ⋅ β or by ( α , β ) , satisfying the following
properties-

1. ( α , β )= ( β , α ) , ∀ α , β ∈ V (Symmetry)
2. c ( α , β )=( cα , β ) =( α , cβ ) , ∀ α , β ∈V , c ∈ R (Homogeneity)
3. ( α + β , γ )= ( α , γ )+ ( β , γ ) , ∀ α , β , γ ∈ V (Linearity)
4. ( α , α ) >0 , ∀ α ≠ θ ∈V . (Positivity)
A real vector space V together with real inner product defined on it, is said to be
a Euclidean Space.
Examples:

1. In the real vector space Rn , let α =(a1 , a2 , … , an ) and β=(b1 , b2 , … ,b n) be two

vectors and let us define
( α , β )=a1 b1 +a2 b2 +⋯+a n b n.
This function satisfies all the conditions of real inner product. This inner
product is called standard inner product and often is called the dot
product, denoted by α ⋅ β . The vector space equipped this inner product
becomes a Euclidean space.
2. In R2 , let us define ( α , β )=2a 1 b 1+ a1 b2 +a2 b1 +a 2 b 2 . Then this functions
satisfies all the 4 conditions of inner product. This function becomes a real
inner product in R2 . Therefore the vector space R2 becomes a Euclidean
space under this inner product.
This example shows that the real vector space can be made a Euclidean
space in many ways.
3. In the real vector space Pn, let us define for polynomials f , g ∈ Pn ,
1
( f , g ) =∫ f ( t ) g ( t ) dt .
0
This function satisfies all the conditions of real inner product. Therefore
the vector space Pn becomes a Euclidean space under this inner product.
4. In the real vector space Pn, let us define for polynomials f , g ∈ Pn ,
1
( f , g ) =∫ f ( t ) g ( t ) dt .
−1
This function satisfies all the conditions of real inner product. Therefore
the vector space Pn becomes a Euclidean space under this inner product.

Norm: Let α be a vector in a Euclidean space V with the inner product the norm
of α , denoted by ‖α‖, is defined by ‖α‖=√ (α , α ).

Theorem: Let α be a vector in a Euclidean space V and ‖α‖, be its norm. Then

(i) ‖cα‖=|c|‖α‖ , c being a real number;

(ii) ‖α‖> 0 unless α =θ and ‖θ‖=0.

√
Proof: (i) ‖cα‖=√ (cα , cα)= c 2 (α , α )=|c| √( α , α )=|c|‖α ‖.
 Inner Product Space

For α ≠ θ implies ( α , α ) >0 and therefore ‖α‖> 0.

If α =θ , then ( α , α )= (θ ,θ )=0 , and therefore ‖α‖=0.

Schwarz’s inequality: For any two vector α , β in a Euclidean space V,

|( α , β )|≤‖α‖‖ β‖ ,
The equality holds when α , β are linearly dependent.

Triangle Inequality: For any two vector α , β in a Euclidean space V, then

‖α + β‖≤‖α ‖+‖ β‖.

Proof: By Schwarz’s inequality, for any two vector α , β in a Euclidean space V,

|( α , β )|≤‖α‖‖ β‖
⇒−‖α‖‖β‖≤ ( α , β ) ≤‖α ‖‖ β‖

⇒ ( α , β ) ≤‖α ‖‖β‖ … … .(i)

2 2 2
Now, ‖α + β‖ = ( α + β , α + β ) =( α , β )+ ( β , α ) + ( α , β ) + ( β , β )=‖α ‖ +2 ( α , β )+‖β‖ .

Form (i),

2 ( α , β ) ≤ 2‖α‖‖ β‖
2 2 2 2
⇒‖α ‖ +2 ( α , β )+‖β‖ ≤‖α ‖ +2‖α ‖‖β‖+‖β‖
2 2
⇒‖α + β‖ ≤ (‖α‖+‖β‖)

⇒‖α + β‖≤ (‖α‖+‖β‖)

This completes the proof.
Unit vector: Let α be a vector in a Euclidean space V is called a unit vector if
β
‖α‖=1. Along any non-zero vector β , the unit vector is .
‖β‖
Orthogonal set of vectors: A set of vectors { α 1 , α 2 , … , α n } in an Euclidean inner
product space V is called an orthogonal set of vectors if ( α i , α j )=0 , if i≠ j.

Orthonormal set of vectors: A set of vectors { α 1 , α 2 , … , α n } in an Euclidean

inner product space V is called an orthonormal set of vectors if ( α i , α j )=0 , if i≠ j
and ( α i , α i )=1.

Pythagoras Theorem: For two orthogonal vectors α , β in an Euclidean space V,

2 2 2
‖α + β‖ =‖α ‖ +‖ β‖ .
Proof: Let α and β be two orthogonal vectors. Then ( α , β )=0.

Now,
2 2 2
L.H.S¿‖α + β‖ =( α + β , α + β ) =( α , β ) + ( α , β ) + ( β ,α ) + ( β , β )=‖α ‖ +‖β‖ =¿RHS.
 Inner Product Space
This completes the proof.

Parallelogram Law: For two vectors α , β in an Euclidean space V,

‖α + β‖ +‖α −β‖ =2 (‖α ‖ +‖β‖ ) .

2 2 2 2

Theorem: A set of orthogonal vectors in an Euclidean space is a linearly

independent set of vectors.

Proof. Let { α 1 , α 2 , … , α n } be a set of orthogonal vectors in an Euclidean space V.

Then

( α i , α j )=0 , for i≠ j.
Let us consider the scalars c 1 , c2 , … , c n such that c 1 α 1 +c 2 α 2 +⋯+ cn α n=θ . Then

( α i , c1 α 1+ c2 α 2+⋯+ c n α n )= ( α i , θ )
⇒ ( α i , c 1 α 1 ) + ( α i , c 2 α 2 ) +⋯+ ( α i , c i α i ) +⋯+ ( α i , c n α n )=0 (Using linearity axiom)

⇒ c1 ( α i , α 1 ) +c 2 ( α i , α 2 ) +⋯+ ci ( α i , α i ) +⋯+c n ( α i ,α n )=0 (Using homogeneity axiom)

⇒ ci ( α i , α i )=0
2
⇒ ci‖α i‖ =0

⇒ ci=0 ,i=1 , 2 , … , n .

Hence c 1 α 1 +c 2 α 2 +⋯+ cn α n=θ ⇒ c 1=c2=⋯=c n=0.

Therefore, the given set of vectors is linearly independent.

Example: In R3 , let α =( a1 ,a 2 , a 3 ) and β=( b1 ,b 2 , b3 ) , then find weather ( α , β ) is an

inner product where ( α , β )=a1 b1 +(a2+ b3 )(b2 +b 3).
2
Solution: Here ( α , β )=a1 b1 + ( a2 +b 3 ) ( b2 +b 3 )=a 1 b1 + a2 b 2+ a2 b3 +b 3 b 2+ b3 .

( β , α )=b1 a1 + ( b2 + a3 ) ( a2 +a 3 )=b 1 a 1+ b2 a2 +b2 a3 +a 3 a2+ a23 .

Hence ( α , β ) ≠ ( β , α ) . The symmetry axiom is not satisfied for the given function.
So the given function is not an inner product.

Example: Verify whether the following is an inner product Type equation here .
2
1. If α =( a1 ,a 2 ) , β=( b1 , b2 ) ∈ R , then ( α , β )=2a 1 b 1+ 6 a2 b2 .
2
2. If α =( a1 ,a 2 ) , β=( b1 , b2 ) ∈ R , then ( α , β )=a1 b1−2 a1 b2−2a 2 b 1+5 a2 b 2 .
2
3. If α =( a1 ,a 2 ) , β=( b1 , b2 ) ∈ R , then ( α , β )=a21 +2 a1 b1 +4 b22 +2 b2 .
2
4. If α =( a1 ,a 2 ) , β=( b1 , b2 ) ∈ R , then ( α , β )=b21−a21+ a2 b2 +3 a1 +5.

Solution:

1. It is seen that ( α , β )=2a 1 b 1+ 6 a2 b2, ( β , α )=2b 1 a 1+ 6 b2 a2. So, ( α , β )= ( β , α ) .

Hence the symmetricity axiom is satisfied.
 Inner Product Space

Let c ∈ R . Then c ( α , β )=c ( 2 a1 b1 +6 a 2 b2 ) =2 ca1 b1 +6 ca 2 b 2.

Again, cα=c ( a 1 , a2 ) =( c a1 , c a 2 ) . So, ( cα , β )=2ca 1 b 1+ 6 ca2 b2

Hence, c ( α , β )=( cα , β ) .

Therefore, the homogeneity axiom is satisfied.

Definition: Let α and β be two vectors in a Euclidean space V. Then the angle
−1 (α , β )
between α and β is defined as cos .
‖α‖‖ β ‖
Let α −cβ be orthogonal to β . Then

( α , β)
( α −cβ , β )=0⇒ ( α , β ) −( cβ , β )=0⇒ ( α , β ) −c ( β , β )=0⇒ c= .
( β , β)
Here c is scalar component of α along β and cβ is called projection of α upon
β.
Example: Using Gram-schmidt process convert the basis { ( 1 , 0 ,1 ) , ( 1 ,1 , 1 ) ,(1 ,3 , 4) }
of R3 into orthogonal basis. Also, covert it into orthonormal basis.

Solution: Let α 1=( 1 , 0 ,1 ) , α 2=( 1 ,1 , 1 ) ,α 3=( 1 , 3 , 4 ) . Here the set of vectors { α 1 , α 2 , α 3 }

are linearly independent. Let β 1=α 1 and consider β 2=α 2−c β1 so that β 2 is
orthogonal to β 1 . Then

( β 1 , α 2 ) ( α 1 , α 2 ) 1 ×1+ 0× 1+ 1× 1
( β 1 , β 2 )=0⇒ ( β 1 , α 2−c β 1 )=0 ⇒ ¿⇒ ¿⇒ c= = = 2 2 2
=1.
( β 1 , β1 ) ( α 1 , α 1 ) 1 +0 +1

Hence β 2=α 2−c β1 =( 1, 1 , 1 )−1 × ( 1 , 0 ,1 )= ( 0 ,1 , 0 ) . Hence the set { β 1 , β 2 } is an

orthogonal set of vectors.

Now, we consider another vector β 3=α 3−d 1 β 1−d 2 β 2 so that it is orthogonal to the
vectors β 1 , β 2 . Then

( β 1 , β 3 )=0⇒ ( β 1 , α 3 −d 1 β 1−d 2 β 2) =0⇒ ¿⇒ ¿⇒ ¿

( β 1 , α 3 ) ( α 1 , α 3 ) 1×1+ 0× 3+1 ×4 5
⇒ d1 = = = 2 2 2
=
( β 1 , β1 ) ( α 1 ,α 1 ) 1 + 0 +1 2

and

( β 2 , α 3 ) 0 ×1+1 ×3+ 0× 4
( β 2 , β 3 ) =0⇒ ( β 2 , α 3 −d 1 β 1−d 2 β 2) =0⇒ ¿⇒ ¿⇒ ¿ ⇒ d2 = = 2 2 2
=3.
( β 2 , β2 ) 0 +1 +0
Hence,

5
β 3=α 3−d 1 β 1−d 2 β 2=( 1 , 3 , 4 )− ( 1 , 0 ,1 )−3 ( 0 ,1 , 0 )
2

(5 5
)
β 3=( 1 , 3 , 4 ) − ,0 , −( 0 ,3 , 0 ) =
2 2
−3
2 (,0 ,
3
2 )
 Inner Product Space

{
Therefore { β 1 , β 2 , β 3 } = ( 1 , 0 , 1 ) , ( 0 , 1 , 0 ) , ( −32 , 0 , 32 )} is the required orthogonal set of
vectors.

Consider,

( )
β1 (1 , 0 , 1 )
1 1 β2 ( 0 ,1 , 0 )
γ 1= = = , 0 , γ = = 2 2 2 =( 0 , 1, 0 )
‖β 1‖ √ 1 + 0 + 1 2 2
2 2 2 2
‖β 2‖ √ 0 +1 + 0
−3
(
,0,
3
) (
−3
,0 ,
3
)
( )
β 2 2 2 2 −1 1
γ 3= 3 = = = ,0 , .

√(
‖β 3‖ −3 2 2 3 2 3 √2 √2 √ 2
2
+0 + ) 2 () 2

Then { γ 1 , γ 2 , γ 3 }=
{( 1
2
1
) (
,0 , , ( 0 ,1 , 0 ) ,
2
−1
,0,
√2 √2
1
)}
is the required orthonormal basis.