You are on page 1of 7

spanda@maths.iitkgp.ac.

in 1

Solution of Tutorial Problems set-II


Note: All these problems can be solved using the results of Chapter-2.
n
P
[0.0.1] Exercise Find a necessary and sufficient condition for hx, yi = αi xi yi to be an inner product
i=1
on Rn .
n
αi xi yi is an inner product on Rn . Take ei . Then hei , ei i = αi > 0
P
Sol. We assume that hx, yi =
i=1
for i = 1, . . . , n.
n
P
Converse: Assume that αi > 0 for i = 1, . . . , n. To show hx, yi = αi xi yi is an inner product
i=1
on Rn .

n
αi x2i > 0 as αi > 0 for i = 1, . . . , n.
P
1(a). hx, xi =
i=1

n
αi x2i = 0 =⇒ xi = 0 for i = 1, . . . , n.
P
1(b). hx, xi =
i=1

2. It is trivial.

3. It is trivial.

4. It is trivial.

!
a11 a12
[0.0.2] Exercise Let A = be a 2 × 2 matrix with real entries. Let fA : R2 → R be a map
a21 a22
defined by fA (x, y) = y t Ax, where x, y ∈ R2 . Show that fA is an inner product on R2 if and only if
A = At , a11 > 0, a22 > 0 and det(A) > 0.

Sol. We first assume that fA is an inner product. Using definition of inner product,
! ! we have
a11 a12 1
fA (x, x) > 0 for all non-zero x ∈ R2 . Then fA (e1 , e1 ) = et1 Ae1 = (1, 0) > 0 =⇒
a21 a22 0
a11 > 0. Using e2 , we can show that a22 > 0.
Using definition of inner product we have fA (x, y) = fA (y, x) = fA (y, x) as this is a real inner
product space. Therefore we have

y t Ax = xt Ay

=⇒ (y t Ax)t = xt Ay

=⇒ xt At y = xt Ay
spanda@maths.iitkgp.ac.in 2

=⇒ xt (At − A)y = 0. This is true for all x, y ∈ R2 .


! !
0 a21 − a12 0
Take x = (1, 0)t and y = (0, 1)t . Then we have (1, 0) = 0. This
a12 − a21 0 1
implies that a21 − a12 = 0 =⇒ a12 = a21 . Hence A = At .

To prove det (A) > 0, we take x = (a22 , −a12 ).


! !
a11 a12 a22
Since fA (x, x) > 0, we have (a22 , −a12 ) >0
a12 a22 −a12
 
Then a22 a11 a22 − a212 > 0.
 
a11 a22 − a212 > 0 as a22 > 0.

Hence det(A) > 0.

We now prove the converse.


! !
a11 a12 x1
1(a). fA (x, x) = (x1 , x2 )
a21 a22 x2

= a11 x21 + 2a12 x1 x2 + a22 x22 (as a12 = a21 )

a12 2 a11 a22 −a212 2


= a11 (x1 + a11 x2 ) + a11 x2 (as a11 > 0)

>0

a12 2 a11 a22 −a212 2


1(b). fA (x, x) = 0. Using a11 (x1 + a11 x2 ) + a11 x2 , we have x1 = 0 and x2 = 0.

2. It is trivial.

3. It is trivia.

4. It is trivia.

[0.0.3] Exercise Let V be a finite-dimensional vector space and let B = {u1 , . . . , un } be a basis for V.
Let hx, yi be an inner product on V. If c1 , . . . , cn are any n scalars, show that there is exactly one
vector x in V such that hx, ui i = ci for i = 1, . . . , n.

Sol. This solution will be sent later.


spanda@maths.iitkgp.ac.in 3

[0.0.4] Exercise Let (V, h, i) be an inner product space. Show that hx, yi = 0 for all y ∈ V, then x = 0.

Sol. Given that that hx, yi = 0 for all y in V. To show that x = 0. Since hx, yi = 0 for all y ∈ V,
then hx, xi = 0 as x is an element in V. Using definition of inner product, hx, xi = 0 =⇒ x = 0.
n
xi yi is not an inner product on Cn .
P
[0.0.5] Exercise Show that hx, yi =
i=1

Sol. This is not an inner product on Cn . It does not satisfy homogeneity property, that is
hαx, yi 6= αhx, yi. For example, take x = (1, 0, 0, . . . , 0), y = (1, 0, 0, . . . , n) and α = i. Then
hαx, yi = −i and αhx, yi = i They are not equal.

[0.0.6] Exercise Let (V, h, i) be a finite dimensional inner product space. Prove that for v ∈ V − {0},
the set W = {w ∈ V : hw, vi = 0} is a subspace of V of dimension dim V − 1.

Sol. The definition of W says that W = {v}⊥ . Hence W is a subspace of V. To find the
dimension of W, we use the following fact. Let S be a subset of V, then S ⊥ = (LS(S))⊥ . Using
this fact {v}⊥ = LS({v})⊥ = W. Then V = W + LS({v}) (internal direct sum). We know that
dim (LS({v})) = 1. Hence dim W = dim V − 1.

[0.0.7] Exercise Decide which of the following functions define an inner product C2 . For x = (x1 , y1 ),
y = (y1 , y2 ).

1. hx, yi = x1 y2

2. hx, yi = x1 y1 + x2 y2

3. hx, yi = x1 y1 + x2 y2

4. hx, yi = 2x1 y1 + i(x2 y1 − x1 y2 ) + 2x2 y2

Sol.

1. Not an inner product. Take x = (1, 0). hx, xi = 0 but x is not equal to zero.

2. Yes, inner product.

3. Not an inner product. Conjugate symmetry does not satisfy.

4. Not an inner product. Conjugate symmetry does not satisfy. Take x = (1, i) and y = (i, 1).

[0.0.8] Exercise Let V = P3 (x) be a subspace of real polynomials of degree at most 3. Equip V with
the inner product
Z1
hf, gi = f (x)g(x)dx
0
.

1. Find the orthogonal complement of the subspace of scalar polynomials.


spanda@maths.iitkgp.ac.in 4

2. Apply the Gram Schmidt process to the basis {1, x, x2 , x3 }.

Sol. 1. To find the orthogonal complement of of the subspace of scalar polynomials (scalar
polynomial means zero degree polynomial).
Let W be the orthogonal complement of the subspace of scalar polynomials.
Let P (x) = a0 + a1 x + a2 x2 + a3 x3 be an arbitrary element in W. Then h1, P (x)i = 0 =⇒
R1
P (x)dx = 0
0

R1
(a0 + a1 x + a2 x2 + a3 x3 )dx = 0
0

a1 a2 a3
=⇒ a0 + 2 + 3 + 4 =0

=⇒ 12a0 + 6a1 + 4a2 + 3a3 = 0

−6a1 −4a2 −3a3


=⇒ a0 = 12 .

−6a1 −4a2 −3a3


P (x) = 12 + a1 x + a 2 x 2 + a3 x 3

= a1 (x − 1/2) + a2 (x2 − 1/3) + a3 (x3 − 1/4)

This says that P (x) is a linear combination of x − 1/2, x2 − 1/3 and x3 − 1/4.

Hence W = ls({x − 1/2, x2 − 1/3, x3 − 1/4}).

The set of scalar polynomials is equal the R and we know the dimension of R is 1.

We also know that P3 (x) = R ⊕ W. Hence dim W = 3.

Therefore {x − 1/2, x2 − 1/3, x3 − 1/4} is basis of W.

2. Consider u1 = 1, u2 = x, u3 = x2 , u4 = x3 .

v1 = u 1 = 1

hu2 ,v1 i
v2 = u 2 − hv1 ,v1 i v1 = x − 1/2.

hu3 ,v2 i hu3 ,v1 i


v3 = u 3 − hv2 ,v2 i v2 − hv1 ,v1 i v1 = x2 − x + 1/6.
spanda@maths.iitkgp.ac.in 5

hu4 ,v3 i hu4 ,v2 i hu4 ,v1 i


v4 = u 4 − hv3 ,v3 i v3 − hv2 ,v2 i v2 − hv1 ,v1 i v1 = x3 − 32 x2 + 53 x − 1
20 .

[0.0.9] Exercise Find an inner product on R2 such that he1 , e2 i = 2.

Sol. Exercise 0.0.2 helps you to solve Exercise 0.0.9. If you are able to find a symmetric matrix
et1 Ae2 = 2 then you are done and
A with each diagonal entry is positive and det(A) > 0 such "that #
3 2
your desire inner product will be hx, yi = y t Ax. Take A = . You can easily check that A is
2 3
symmetric, each diagonal entry of A is positive and det(A) > 0. Notice that et1 Ae2 = 2.
" #" #
h i 3 2 x
1
Hence your desire inner product is hx, yi = y t Ax = y1 y2 = 3x1 y1 + 2x2 y1 +
2 3 x2
2x1 y2 + 3x2 y2 .

[0.0.10] Exercise Let V be the space of all n × n over R with the inner product hA, Bi = trace(AB t ).
Find the orthogonal complement of the subspaces of diagonal matrices.

[0.0.11] Exercise Let (V, h, i) be an IPS. Let α, β ∈ V. Then show that α = β if and only if hα, γi =
hβ, γi for all γ ∈ V.

Sol. First we assume that hα, γi = hβ, γi for all γ ∈ V. Then hα − β, γi = 0 for all γ ∈ V. Using
Exercise 0.0.4, we have α − β = 0. Hence α = β.

Now we assume that alpha=beta, that is α − β = 0. Then hα − β, γi = 0 for all γ ∈ V. This


implies that hα, γi = hβ, γi for all γ ∈ V.

[0.0.12] Exercise Apply Gram Schmidt process to the vectors u1 = (1, 0, 1), u2 = (1, 0, −1) and u3 =
(0, 3, 4) to obtain an orthonormal basis for R3 with the standard inner product.

Sol. v1 = u1 = (1, 0, 1).

v2 = u 2

hu3 ,v2 i hu3 ,iv1


v3 = u 3 − hv2 ,v2 i v2 − hv1 ,iv1 v1 .

−4
= (0, 3, 4) − 2 (1, 0, −1) − 24 (1, 0, 1).

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

= (0, 3, 4) + (0, 0, −4)

= (0, 3, 0).
spanda@maths.iitkgp.ac.in 6

v1 = (1, 0, 1), v2 = (1, 0, −1) and v3 = (0, 3, 0) are orthogonal.


 
1 −1 2
 
[0.0.13] Exercise Consider the inner product hx, yi = y t AX on R3 where A =   1 1 0 . Find an

−1 0 3
orthonormal basis B of S := {(x1 , x2 , x3 ) : x1 + x2 + x3 = 0} and then extend it to an orthonormal
basis of R3 .

Sol. We first find a basis of S. Let (x1 , x2 , x3 ) be an arbitrary element in S. Then x1 + x2 + x3 = 0.


This implies (x1 , x2 , x3 ) = (−x2 − x3 , x2 , x3 ) = x2 (−1, 1, 0) + x3 (−1, 0, 1).

Notice that S = ls({(−1, 1, 0), (−1, 0, 1)}). It is easy to prove that {(−1, 1, 0), (−1, 0, 1)}
is linearly independent. Hence {(−1, 1, 0), (−1, 0, 1)}. Applying Gram Schmidt process on
{(−1, 1, 0), (−1, 0, 1)}. Let u1 = (−1, 1, 0) and u2 = (−1, 0, 1).

v1 = u1 = (−1, 1, 0).

hu2 ,v1 i
v2 = u 2 − hv1 ,v1 i v1 .

We have to calculate hu2 , v1 i and hv1 , v1 i.


  
h i 1 −1 −1 2
t
 
hu2 , v1 i = v1 Au2 = −1 1 0  1 1 0   0 
  
=2
−1 0 3 1
  
h i  2 1 −1 −1
hv1 , v1 i = −1 1 0   1 1 0   1  = 1.
 
−1 0 3 0

Then v2 = u2 − 2v1 = (−1, 0, 1) − 2(−1, 1, 0) = (1, −2, 1)


 
1 1
{√ v1 , √ v2 } = {(−1, 1, 0), (−1, 2, −1)} hv1 , v1 i = 1, hv2 , v2 i = −1 is an orthonormal
hv1 ,v1 i hv2 ,v2 i
basis of S.

You can notice that (1, 1, 1) is orthogonal to (−1, 1, 0) and (−1, 2, −1). Therefore
{(−1, 1, 0), (−1, 2, −1), ( √13 , √13 , √13 )} is an orthonormal set in R3 . They are linearly independent
and dimension of R3 is 3. Then {(−1, 1, 0), (−1, 2, −1), ( √13 , √13 , √13 )} is an orthonormal basis on R3 .
p
[0.0.14] Exercise Let (V, h, i) be an IPS. Let ||u|| = hu, ui for all u ∈ V be the norm induced by h, i.
Then prove that ||u + v||2 + ||u − v||2 = 2||u||2 + 2||v||2 .
p
Sol. Note: Before going to solve this problem, I would like to introduce something. ||x|| = hx, xi.
spanda@maths.iitkgp.ac.in 7

We have seen that this a norm on V. This is called a norm induced by the inner product h, i.
This problem says that any norm which is induced by an inner product that norm must satisfy
this condition ||u + v||2 + ||u − v||2 = 2||u||2 + 2||v||2 .

||u + v||2 = hu + v, u + vi = hu, ui + hu, vi + hv, ui + hv, vi.

||u − v||2 = hu − v, u − vi = hu, ui − hu, vi − hv, ui + hv, vi.

After adding them, we have ||u + v||2 + ||u − v||2 = 2||u||2 + 2||v||2 . This is clled Parallelogram
Identity.

Note: The Parallelogram Identity is not true in general for any arbitrary norm.

[0.0.15] Exercise Let (V, h, i) be a finite dimensional IPS. Let B = {u1 , u2 , . . . , un } be a basis of V. Then
prove that hu, vi = ȳ t Ax for all u, v ∈ V where x = (x1 , . . . , xn )t ,y = (y1 , . . . , yn )t are coordinates of
u and v with respect to basis B and aij = hui , uj i.

Sol. Given that (V, h, i) is a finite dimensional IPS and B = {u1 , u2 , . . . , un } is a basis of V. Let
u, v ∈ V. Then x = x1 u1 + · · · + xn un and y = y1 u1 + · · · + yn un . Here x = (x1 , . . . , xn )t and
y = (y1 , . . . , yn )t are the coordinates of u and v with respect to basis B.

hu, vi = hx1 u1 + · · · + xn un , y1 u1 + · · · + yn un i

n
P
= xi yj hui , uj i.
i,j=1

= y t Ax where aij = hui , uj i and x = (x1 , . . . , xn )t and y = (y1 , . . . , yn )t .

You might also like