THE CHINESE UNIVERSITY OF HONG KONG

Department of Mathematics
MATH2040A/B (First Term, 2020-21)
Linear Algebra II
Solution to Homework 10

Sec. 6.3
Q3(b). For each of the following inner product spaces V and linear operators T on V , evaluate T ∗
at the given vector in V .

V = C2 , T (z1 , z2 ) = (2z1 + iz2 , (1 − i)z1 ) , x = (3 − i, 1 + 2i)

Sol. Solution: Denote β = {(1, 0), (0, 1)} as the standard ordered basis for V under field F = C.
Then
   

2 i ∗ ∗ 2 1+i
[T ]β = [T (1, 0)]β [T (0, 1)]β = ⇒ [T ]β = ([T ]β ) =
1−i 0 −i 0

It follows that
    
∗ ∗ 2 1+i 3−i 5+i
[T (x)]β = [T ]β [x]β = =
−i 0 1 + 2i −1 − 3i

Therefore, T ∗ (x) = (5 + i, −1 − 3i)

Z 1
Q3(c). V = P1 (R) with hf, gi = f (t)g(t)dt, T (f ) = f 0 + 3f . f (t) = 4 − 2t.
−1

Sol. Let β = {1, t} be the standard basis of V . Write T ∗ (4 − 2t) = a + bt for some a, b ∈ R. Then
for any g(t) = c + dt ∈ V with c, d ∈ R, we have T (g(t)) = d + 3c + 3dt and

hd + 3c + 3dt, 4 − 2ti = hT (g(t), 4 − 2ti = hg(t), T ∗ (4 − 2t)i = hc + dt, a + bti .

Now hd + 3c + 3dt, 4 − 2ti = 2(4)(d + 3c) + (3d)(−2) 23 = 4d + 24c and hc + dt, a + bti =
2ac + 23 bd. Since c, d are arbitrary, the coefficients of them on both sides of the equation
must equal respectively. Therefore 24 = 2a and 23 b = 4. Hence a = 12 and b = 6. So
T ∗ (4 − 2t) = 12 + 6t.

Q6. Let T be a linear operator on an inner product space V . Let U1 = T + T ∗ and U2 = T T ∗ .

Prove that U1 = U1∗ and U2 = U2∗ .

Sol.
U1∗ = (T + T ∗ )∗ = T ∗ + (T ∗ )∗ = T ∗ + T = U1 .
U2∗ = (T T ∗ )∗ = (T ∗ )∗ T ∗ = T T ∗ = U2 .

Q8. Let V be a finite-dimensional inner product space, and let T be a linear operator on V . Prove
that if T is invertible, then T ∗ is invertible and (T ∗ )−1 = (T −1 )∗ .

1
 Sol. Suppose x ∈ N(T ∗ ). Then

0 = T −1 (x), T ∗ (x) = T T −1 (x), x = hx, xi .




*
Hence x = 0 and thus T ∗ is an injective linear operator on V . So T ∗ is invertible by finiteness
of dimension of V . Also we have

x, (T −1 )∗ (y) = T −1 (x), T ∗ (T ∗ )−1 (y) = T T −1 (x), (T ∗ )−1 (y) = x, (T ∗ )−1 (y)






for all x, y ∈ V . Therefore (T ∗ )−1 = (T −1 )∗ .

Q9. Prove that if V = W ⊕ W ⊥ and T is the projection on W along W ⊥ , then T = T ∗ . Hint:

Recall that N(T ) = W ⊥ . (For definitions, see the exercises of Sections 1.3 and 2.1.)

Sol. From the assumption V = W ⊕ W ⊥ , for all v, w ∈ V , there exist unique v1 , w1 ∈ W and
v2 , w2 ∈ W ⊥ such that v = v1 + v2 and w = w1 + w2 . We check that

hT (v), wi = hv1 , w1 + w2 i = hv1 , w1 i + hv1 , w2 i = hv1 , w1 i

and so
hv, T (w)i = hT (w), vi = hw1 , v1 i = hv1 , w1 i = hT (v), wi .
Therefore T ∗ exists and T = T ∗ .

Q13. Let T be a linear operator on a finite-dimensional inner product space V . Prove the following
results.

(a) N(T ∗ T ) = N(T ). Deduce that rank(T ∗ T ) = rank(T ).

(b) rank(T ) = rank(T ∗ ). Deduce from (a) that rank(T T ∗ ) = rank(T ).
(c) For any n × n matrix A. rank(A∗ A) = rank(AA∗ ) = rank(A).

Sol. (a) It is clear that N(T ) ⊂ N(T ∗ T ). Let x ∈ N(T ∗ T ). Then hT (x), T (x)i = hx, T ∗ T (x)i =
* *
hx, 0 i = 0. Hence T (x) = 0 and x ∈ N(T ). It follows that

rank(T ∗ T ) = n − nullity(T ∗ T ) = n − nullity(T ) = rank(T )

where n = dim(V ).
(b) By Q12(b), R(T ∗ ) = N(T )⊥ . Since V = N(T ) ⊕ N(T )⊥ by Sec 6.2 Q13(d), we have
n = nullity(T ) + dim(N(T )⊥ ) and

rank(T ∗ ) = dim(N(T )⊥ ) = n − nullity(T ) = rank(T ).

(c) Note that L∗A = LA∗ . Hence by applying part (a) and (b) with T = LA , we have
rank(A∗ A) = rank(LA∗ LA ) = rank(L∗A LA ) = rank(LA ) = rank(A). Similarly, rank(AA∗ ) =
rank(A).

Q14. Let V be an inner product space, and let y, z ∈ V . Define T : V → V by T (x) = hx, yi z
for all x ∈ V . First prove that T is linear. Then show that T ∗ exists, and find an explicit
expression for it.

2
Sol. For all x, w ∈ V , we have
D E
hT (x), wi = hhx, yi z, wi = hx, yi hz, wi = x, hz, wiy = hx, hw, zi yi .

Note that w 7→ hw, zi y is a linear operator on V since

hw1 + cw2 , zi y = (hw1 , zi + c hw2 , zi)y = hw1 , zi y + c hw2 , zi y

for all w1 , w2 ∈ V and scalar c. Therefore this gives the adjoint of T .

Sec. 6.4
2. For each linear operator T on an inner product space V, determine whether T is normal,
self-adjoint, or neither. If possible, produce an orthonormal basis of eigenvectors of T for V
and list the corresponding eigenvalues.

(c). V = C2 and T is defined by T (a, b) = (2a + ib, a + 2b)

Sol. Take β = {(1, 0), (0, 1)} as the ordered basis for V . Then
   
2 i ∗ ∗ 2 1
[T ]β = ⇒ [T ]β = ([T ]β ) =
1 2 −i 2

Therefore, we have
    
∗ ∗ 2 1 2 i 5 2 + 2i
[T T ]β = [T ]β [T ]β = =
−i 2 1 2 2 − 2i 5

and also
    
∗ ∗ 2 i 2 1 5 2 + 2i
[T T ]β = [T ]β [T ]β = = = [T ∗ T ]β
1 2 −i 2 2 − 2i 5

As the matrix representation map is an isomorphism, we have T ∗ T = T T ∗ , i.e. T is normal.

Also, as [T ]β 6= [T ∗ ]β , T 6= T ∗ and hence T is not self-adjoint operator. We then solve for the
 
2−λ i
eigenvalue of T Consider fT (t) = det ([T ]β − λI2 ) = det = (λ − 2)2 − i = 0.
1 2−λ
√ √ √
Solving λ − 2 = i = ± 22 (1 + i). Then we have the eigenvalue given by λ1 = 2 + 22 (1 + i)
√
2
and λ2 = 2 − 2 (1 + i)

√
2
For λ1 = 2 + 2 (1 + i), consider
√ ! ( √ ! )
2 2
2− 2 (1 + i) i
Eλ1 = N (T − λ1 I2 ) = N √
2
= t 1
2 :t∈C
1 2 − 2 (1 + i) 2 (1 − i)
 √ 
Obviously we have 22 , 21 (1 − i) = 1

3
 √
2
For λ2 = 2 − 2 (1 + i). consider
√ ! ( ! )
2+ 2
2 (1 + i) i − 12 (1
√
+ i)
Eλ2 = N (T − λ2 I2 ) = N √
2
= t 2 :t∈C
1 2 + 2 (1 + i) 2
 √ 
Obviously we have − 21 (1 + i), 22 = 1 Therefore, we can take the orthonormal basis of

eigenvectors of T for V can be taken as
( √ ! !)
2 − 12 (1
√
+ i)
2 ,
1 2
2 (1 − i) 2

(d). V = P2 (R) and T is defined by T (f ) = f 0 , where

Z 1
hf, gi = f (t)g(t)dt
0

Sol. Solution: Take α = 1, x, x2 as the orthogonal basis for V and


√ hence we can√ apply
2
G-S pro-


cess to obtain the ordered orthonormal basis for V, β = 1, 2 3(t − 1/2), 6 5 t − t + 1/6
Check by definition we can obtain
 √   
0 2 3 √0 0
√ 0 0
[T ]β =  0 0 2 15  ⇒ [T ∗ ]β = ([Tβ ])∗ =  2 3 √0 0  6= [T ]β
0 0 0 0 2 15 0

Therefore, T is not an adjoint linear operator. Also, we have

 √    
0 2 3 √0 √0 0 0 12 0 0
[T T ∗ ]β = [T ]β [T ∗ ]β =  0 0 2 15   2 3 √0 0  =  0 60 0 
0 0 0 0 2 15 0 0 0 0

and
  √   
0
√ 0 0 0 2 3 √0 0 0 0
[T ∗ T ]β = [T ∗ ]β [T ]β =  2 3 √0 0  0 0 6 [T ∗ T ]β
2 15  =  0 12 0  =
0 2 15 0 0 0 0 0 0 60

Therefore, T ∗ T 6= T T ∗ and hence T is not normal operator. So, there exist no orthonormal
basis of eigenvectors of T for V .

6. Q: Let V be a complex inner product space, and let T be a linear operator on V. Define
1 1
T1 = (T + T ∗ ) and T2 = (T − T ∗ )
2 2i
(a) Prove that T1 and T2 are self-adjoint and that T = T1 + iT2 .
(b) Suppose also that T = U1 + iU2 , where U1 and U2 are self-adjoint. Prove that
U1 = T1 and U2 = T2
(c) Prove that T is normal if and only if T1 T2 = T2 T1 .

4
Sol: (a) We have
 ∗  ∗
1 1 1 1
T1∗ = ∗
(T + T ) = (T ∗ + (T ∗ )∗ ) = (T ∗ + T ) = (T + T ∗ ) = T1
2 2 2 2

T1 is self-adjoint. Also, we have

 ∗  ∗
1 i i
∗
T2 = (T − T ) = − (T − T ) = (T − T ∗ )∗
∗ ∗
2i 2 2
i i 2 1
= (T ∗ − T ) = (T ∗ − T ) = (T − T ∗ ) = T2
2 2i 2i
so T2 is also self-adjoint. It is clear that
 
1 1 1 1
T1 + iT2 = (T + T ∗ ) + i (T − T ∗ ) = (T + T ∗ ) + (T − T ∗ ) = T
2 2i 2 2

(b) From assumption, we have T = T1 + iT2 = U1 + iU2 and hence

(T1 − U1 ) + i (T2 − U2 ) = 0 (1)

As T1 , T2 , U1 , U2 are self-adjoint, takeing adjoint operator on both sides

(T1 − U1 ) − i (T2 − U2 ) = (T1∗ − U1∗ ) − i (T2∗ − U2∗ ) = ((T1 − U1 ) + i (T2 − U2 ))∗ = 0 (2)

Adding (1) and (2) to use

2 (T1 − U1 ) = 0 ⇒ T1 = U1

Consider (1) − (2) : 2i (T2 − U2 ) = 0 yields T2 = U2 . The proof is completed.

(c) (⇒) Suppose T is normal, then

T12 + iT1 T2 − iT2 T2 + T22 = (T1 − iT2 ) (T1 + iT2 ) = (T1 + iT2 )∗ (T1 + iT2 )
= T ∗ T = T T ∗ = (T1 + iT2 ) (T1 + iT2 )∗ = (T1 + iT2 ) (T1 − iT2 )
= T12 − iT1 T2 + iT2 T1 + T22

By swapping the terms in the equality above yields 2iT1 T2 = 2iT2 T1 and hence T1 T2 =
T2 T1 .
(⇐) Suppose T1 T2 = T2 T1 , we have

T ∗ T = (T1 + iT2 )∗ (T1 + iT2 ) = (T1 − iT2 ) (T1 + iT2 ) = T12 + T22 + iT1 T2 − iT2 T1
= T12 + T22 + iT2 T1 − iT1 T2 = (T1 + iT2 ) (T1 − iT2 ) = (T1 + iT2 ) (T1 + iT2 )∗ = T T ∗

Hence T is normal operator.

7. Q: Let T be a linear operator on an inner product space V , and let W be a T -invariant

subspace of V . Prove the following results.
(a) If T is self-adjoint, then TW is self-adjoint.
(b) W ⊥ is T ∗ -invariant.
(c) If W is both T - and T ∗ -invariant, then (TW )∗ = (T ∗ )W .

5
 (d) If W is both T - and T ∗ -invariant and T is normal, then TW is normal.
Sol: (a) ∀u, v ∈ W , since T is self-adjoint,

hTW (u), vi = hT (u), vi = hu, T (v)i = hu, TW (v)i,

whence TW is self-adjoint.
(b) Fix w0 ∈ W ⊥ and w ∈ W . As W is T -invariant, T (w) ∈ W . Then

hw, T ∗ (w0 )i = hT (w), w0 i = 0.

Therefore, T ∗ (w) ∈ W ⊥ . W ⊥ is T ∗ -invariant.

(c) Fix w ∈ W . We claim that (TW )∗ (w) = (T ∗ )W (w). If suffices to show that ∀w0 ∈ W ,
hw0 , (TW )∗ (w)i = hw0 , (T ∗ )W (w)i. Indeed, ∀w0 ∈ W ,

hw0 , (TW )∗ (w)i = hTW (w0 ), wi = hT (w0 ), wi = hw0 , T ∗ (w)i = hw0 , (T ∗ )W (w)i.

Therefore, (TW )∗ = (T ∗ )W .
(d) We have TW (TW )∗ = TW (T ∗ )W = (T T ∗ )W = (T ∗ T )W = (T ∗ )W TW = (TW )∗ TW .
Therefore, TW is normal.

9. Q: Let T be a normal operator on a finite-dimensional inner product space V . Prove that

N(T ) = N(T ∗ ) and R(T ) = R(T ∗ ).
Sol: Fix v ∈ N(T ). If v = ~0, then clearly v ∈ N(T ∗ ). If v 6= ~0, then v is an eigenvector of
T corresponding to eigenvalue 0 and by Theorem 6.15, v is also an eigenvector of T ∗
corresponding to eigenvalue 0 = 0, implying that v ∈ N(T ∗ ). We have N(T ) ⊂ N(T ∗ ).
Note that T ∗ is also normal. Applying the above argument on T ∗ yields N(T ∗ ) ⊂
N((T ∗ )∗ ) = N(T ). Hence, N(T ) = N(T ∗ ).
By Exercise 12 in Sec. 6.3, R(T ∗ ) = N(T )⊥ = N(T ∗ )⊥ = R((T ∗ )∗ ) = R(T ).

11. Q: Assume that T is a linear operator on a complex (not necessarily finite-dimensional)

inner product space V with an adjoint T ∗ . Prove the following results.
(a) If T is self-adjoint, then hT (x), xi is real for all x ∈ V .
(b) If T satisfies hT (x), xi = 0 for all x ∈ V, then T = T0 . Hint: Replace x by x + y and
then by x + iy and expand the resulting inner products.
(c) If hT (x), xi is real for all x ∈ V, then T = T ∗

Sol: (a) As T is self-adjoint, i.e.

hT (x), xi = hx, T ∗ (x)i = hT ∗ (x), xi = hT (x), xi

Therefore, we have
1 1
hT (x), xi = (hT (x), xi + hT (x), xi) = · 2 Re(hT (x), xi) = Re(hT (x), xi) ∈ R
2 2
The proof is completed.

6
(b) Pick x, y ∈ V, we have hT (x), xi = 0 and hT (y), yi = 0. Also, as x + y ∈ V, it follows that

0v = hT (x + y), x + yi = hT (x) + T (y), x + yi = hT (x), yi + hT (y), xi (3)

Similarly, as x + iy ∈ V, we have

0 = hT (x+iy), x+iyi = hT (x)+iT (y), x+iyi = īhT (x), yi+ihT (y), xi = −ihT (x), yi+ihT (y), xi
(4)
And hence (5): 0 = hT (x), yi − hT (y), xi. Summing (3) and (5) yields 2hT (x), yi = 0 and
so hT (x), yi = 0. As this statement holds for all x, y ∈ V, we have T = T0 .
(c) Suppose hT (x), xi ∈ R for all x ∈ V
(?)
hT (x), xi = hx, T ∗ (x)i = hT ∗ (x), xi = hT ∗ (x), xi ⇒ h(T − T ∗ ) (x), xi = 0

where (?) holds because taking conjugation on real number does not change the value.
As h(T − T ∗ ) (x), xi = 0 for all x ∈ V, it follows by (b) that T − T ∗ = T0 . Therefore, we
have T = T ∗