UNIVERSITY EXAMINATIONS UNIVERSITEITSEKSAMENS

university
of south africa

MAT3701 May/June 2020

LINEAR ALGEBRA III

Duration : 3 Hours 100 Marks

EXAMINERS :
FIRST : PROF JD BOTHA
EXTERNAL : PROF LM PRETORIUS

This is an open book examination. The examination question paper remains the property of the University of
South Africa.

This examination question paper consists of 3 pages.

Answer all the questions and show all calculations. Please read the questions carefully.
Since this is an open book examination, if the solution to a similar question is given, no marks will be awarded.
The originality of submissions will be verified by the software Turnitin.

TO UPLOAD YOUR ANSWER FILE

1. Access the myExams site at https://myexams.unisa.ac.za/portal and login using your student number and myUnisa
password.
2. Click on the site MAT3701 Exam May 2020 (in the horizontal navigation bar at the top or click on the Sites
button at the top right-hand corner and select the site from the drop-down menu).
3. Once the site has loaded, click on eAssessment in the left navigation menu.
4. A page with the heading “Assignment List” will appear. Click on MAT3701 MayJune Exam 2020 under “Assign-
ment Title”.
5. A new page will open. Under Attachments, click Choose file next to Select a file from computer. Now browse
your device for your answer file and select it for uploading. Remember, the file should be in pdf format and the name of
the file should be in the format Studentnumber MAT3701.
7. Once you have attached your answer file, the name of the file as well as the file size and upload time stamp will be
displayed under Attachments.
6. Tick the “Honour Pledge” button if you agree.
7. Click the “Submit” button.

Open Rubric
 2 MAT3701
May/June 2020

QUESTION 1
 
0 1
Let A = and consider the following subspaces of M2×2 (C) defined by
1 0
W1 = {X ∈ M2×2 (C) : AX = XA} and W2 = {X ∈ M2×2 (C) : AX = X}.

(1.1) Find a basis for W1 . (8)

(1.2) Find a basis for W1 ∩ W2 . (8)

(1.3) Explain whether M2×2 (C) = W1 ⊕ W2 . (1)

[17]

QUESTION 2

Let T : C 3 → C 3be a linear 2

 operator such that T = T and dim(N (T )) = 2. Show there exists a basis β for C
3

1 0 0
such that [T ]β =  b 0 0 where b, c ∈ C.
c 0 0
[15]

QUESTION 3

Let      
0 0 0 0 x1 1
1 0 0 0 x2  0
A=
0
 , x =   and w =   .
1 1 0 x3  0
0 0 0 0 x4 0
Let T : R4 → R4 be the linear operator defined by T (x) = Ax and let W be the T –cyclic subspace of R4 generated
by w.

(3.1) Find the T –cyclic basis for W generated by w. (8)

(3.2) Find the characteristic polynomial of TW. (2)

(3.3) For each eigenvalue of TW , find a corresponding eigenvector expressed as a linear combination of the (8)
T -cyclic basis for W.
[18]

QUESTION 4

Consider the inner product space P2 (R) over R with h·, ·i defined by
hg, hi = g (a) h (a) + g(b)h (b) + g (c) h (c)
where a, b and c are distinct real numbers. Let β = {fa , fb , fc } be the set of Lagrange polynomials associated with
a, b and c respectively, and let P : P2 (R) → P2 (R) be the orthogonal projection on W = span{fa , √12 (fb + fc )}.

(4.1) Show that ha1 fa + b1 fb + c1 fc , gi = a1 g(a) + b1 g(b) + c1 g(c) for all a1 , b1 , c1 ∈ R and g ∈ P2 (R). (6)

(4.2) Show that {fa , √12 (fb + fc )} is orthonormal. (7)

(4.3) Find a formula for P (g) expressed as a linear combination of β. (7)

[20]

[TURN OVER]
 3 MAT3701
May/June 2020

QUESTION 5

Let T : V → V be a linear operator on a finite-dimensional inner product space V over C.

(5.1) Define what is meant by the adjoint operator T ∗ of T . (2)

(5.2) Define what is meant by a normal operator T . (2)

(5.3) If T is normal, show that kT (v)k = kT ∗ (v)k for all v ∈ V . (5)

(5.4) Suppose V = C 3 and T : C 3 → C 3 is defined by T (z1 , z2 , z3 ) = (z1 + iz2 − iz3 , −iz1 , iz1 + iz3 ). Find (9)
a formula for T ∗ (z1 , z2 , z3 ).
[18]

QUESTION 6

It is given that A ∈ M3×3 (C) is a normal matrix with eigenvalues 1 and −1 and corresponding eigenspaces
 
1 1
E1 = span (2, 3, 6) , (6, 2, −3)
7 7

and  
1
E−1 = span (3, −6, 2) .
7

(6.1) Find the spectral decomposition of A. (11)

(6.2) Find A. (1)

[12]

TOTAL MARKS: [100]

c
UNISA 2020
 MAT3701/000/1/2020

Tutorial letter 000/1/2020

LINEAR ALGEBRA III

MAT3701

Semester 1

Department of Mathematical Sciences

Exam Paper and Memo

MAT3701 May/June 2020

BARCODE

university
Define tomorrow. of south africa
Question 1: 17 Marks
 
0 1
Let A = and consider the following subspaces of M2×2 (C) defined by
1 0
W1 = {X ∈ M2×2 (C) : AX = XA} and W2 = {X ∈ M2×2 (C) : AX = X}.

(1.1) Find a basis for W1 . (8)

(1.2) Find a basis for W1 ∩ W2 . (8)

(1.3) Explain whether M2×2 (C) = W1 ⊕ W2 . (1)

Solution
 
a b
Let X = ∈ M2×2 (C) .
c d
(1.1)
X ∈ W1 ⇔ AX = XA
   
c d b a
⇔ =
a b d c
⇔ a = d and b = c
     
a b 1 0 0 1
⇔X= =a +b .
b a 0 1 1 0
   
1 0 0 1
A basis for W1 is therefore , . (8)
0 1 1 0
(1.2)
X ∈ W1 ∩ W2 ⇔ AX = XA and AX = X
     
a b b a a b
⇔X= from (1.1), and =
b a a b b a
   
a a 1 1
⇔X= =a .
a a 1 1
 
1 1
A basis for W1 ∩ W2 is therefore . (8)
1 1
(1.3) No, since W1 ∩ W2 6= {0} . (1)
[17]

Question 2: 15 Marks

Let T : C 3 → C 3 be a linear operator

 such 2
 that T = T and dim(N (T )) = 2. Show there exists a
1 0 0
3
basis β for C such that [T ]β = b 0 0 where b, c ∈ C.

c 0 0

2
 MAT3701/000/1/2020

Solution

Since dim(N (T )) = 2 there exists a basis {v2 , v3 } for N (T ) consisting of two vectors. Extend it to a
basis β = {v1 , v2 , v3 } for C 3 . Then  
a 0 0
[T ]β =  b 0 0
c 0 0
for some a, b, c ∈ C. Note that [T 2 ]β = [T ]2β = [T ]β since it is given that T 2 = T . Comparing the
(1, 1)-entries on both sides of [T ]2β = [T ]β yields a2 = a, hence a = 0 or a = 1. But a = 0 yields
[T ]2β = [T ]β = 0, hence T = 0 which is contrary to the assumption that dim(N (T )) = 2. Hence [T ]β
is of the required form. [15]

Question 3: 18 Marks

Let      
0 0 0 0 x1 1
1 0 0 0 x2  and w = 0 .
    
A=
0 , x =
1 1 0 x3  0
0 0 0 0 x4 0
Let T : R4 → R4 be the linear operator defined by T (x) = Ax and let W be the T –cyclic subspace
of R4 generated by w.

(3.1) Find the T –cyclic basis for W generated by w. (8)

(3.2) Find the characteristic polynomial of TW. (2)

(3.3) For each eigenvalue of TW , find a corresponding eigenvector expressed as a linear (8)
combination of the T -cyclic basis for W.

Solution

(3.1)        
1 0 0 0
0
, T (w) = Aw =  , T (w) = AT (w) =  , T (w) = AT 2 (w) = 0.
1 2 0 3  
w=
0 0 1 1
0 0 0 0
The T –cyclic basis for W is {w, T (w), T 2 (w)} since it is linearly independent and
T 3 (w) = T 2 (w). (8)

(3.2) Since T 3 (w) = T 2 (w) from (3.1), the characteristic polynomial of TW is

cTW (t) = (−1)3 (t3 − t2 ) = −t2 (t − 1).

(2)

3
(3.3) From (3.1)
T 2 (T − I4 )(w) = 0, so T (T − I4 )(w) = T 2 (w) − T (w) is an eigenvector of TW
associated with 0. Similarly
(T − I4 )T 2 (w) = 0, so T 2 (w) is an eigenvector of TW associated with 1.
(8)[18]

Question 4: 20 Marks

Consider the inner product space P2 (R) over R with h·, ·i defined by
hg, hi = g (a) h (a) + g(b)h (b) + g (c) h (c)
where a, b and c are distinct real numbers. Let β = {fa , fb , fc } be the set of Lagrange polynomials
associated with a, b and c respectively, and let P : P2 (R) → P2 (R) be the orthogonal projection on
W = span{fa , √12 (fb + fc )}.

(4.1) Show that ha1 fa + b1 fb + c1 fc , gi = a1 g(a) + b1 g(b) + c1 g(c) for all a1 , b1 , c1 ∈ R and (6)
g ∈ P2 (R).

(4.2) Show that {fa , √12 (fb + fc )} is orthonormal. (7)

(4.3) Find a formula for P (g) expressed as a linear combination of β. (7)

Solution

(4.1)
ha1 fa + b1 fb + c1 fc , gi
= (a1 fa + b1 fb + c1 fc )(a)g(a) + (a1 fa + b1 fb + c1 fc )(b)g(b) + (a1 fa + b1 fb + c1 fc )(c)g(c)
= (a1 fa (a) + b1 fb (a) + c1 fc (a))g(a) + (a1 fa (b) + b1 fb (b) + c1 fc (b))g(b)
+ (a1 fa (c) + b1 fb (c) + c1 fc (c))g(c)
= a1 g(a) + b1 g(b) + c1 g(c).
(6)
(4.2)
hfa , fa i = 1 · fa (a) from (4.1)
= 1.
and
1 1
hfa , √ (fb + fc )i = h √ (fb + fc ), fa i
2 2
1 1
= √ fa (b) + √ fa (c) from (4.1)
2 2
=0

4
 MAT3701/000/1/2020

and
1 1
h √ (fb + fc ), √ (fb + fc )i
2 2
1
= hfb + fc , fb + fc i
2
1
= (1 · (fb + fc )(b) + 1 · (fb + fc )(c)) from (4.1)
2
=1

Thus {fa , √12 (fb + fc )} is orthonormal. (7)

(4.3)
1 1
P (g) = hg, fa ifa + hg, √ (fb + fc )i √ (fb + fc )
2 2
1
= hg, fa ifa + hg, fb + fc i(fb + fc )
2
1
= hfa , gifa + hfb + fc , gi(fb + fc )
2
1
= (1 · g(a))fa + (1 · g(b) + 1 · g(c))(fb + fc ) from (4.1)
2
1
= g(a)fa + (g(b) + g(c))(fb + fc ).
2
(7)[(20)]

Question 5: 18 Marks

Let T : V → V be a linear operator on a finite-dimensional inner product space V over C.

(5.1) Define what is meant by the adjoint operator T ∗ of T . (2)

(5.2) Define what is meant by a normal operator T . (2)

(5.3) If T is normal, show that kT (v)k = kT ∗ (v)k for all v ∈ V . (5)

(5.4) Suppose V = C 3 and T : C 3 → C 3 is defined by (9)

T (z1 , z2 , z3 ) = (z1 + iz2 − iz3 , −iz1 . iz1 + iz3 ) ,

Find a formula for T ∗ (z1 , z2 , z3 ).

5
 Solution

(5.1) The adjoint operator of T is the linear operator T ∗ : V → V such that

hT (v), wi = hv, T ∗ (w)i for all v, w ∈ V.

(2)

(5.2) T is a normal operator if T T ∗ = T ∗ T . (2)

(5.3) For all v ∈ V

kT (v)k2 = hT (v), T (v)i

= hv, T ∗ T (v)i
= hv, T T ∗ (v)i since T is normal
= hT ∗ (v), T ∗ (v)i
= kT ∗ (v)k2 .

Thus kT (v)k = kT ∗ (v)k. (5)

(5.4) Let β be the standard basis for C 3 , then

   
1 i −i 1 i −i
[T ]β =  −i 0 0  , hence [T ∗ ]β = [T ]∗β =  −i 0 0 
i 0 i i 0 −i

since β is an orthonormal basis for C 3 . It follows that

    
1 i −i z1 z1 + iz2 − iz3
[T ∗ (z1 , z2 , z3 )]β = [T ∗ ]β [(z1 , z2 , z3 )]β =  −i 0 0   z2  =  −iz1 
i 0 −i z3 iz1 − iz3

therefore
T ∗ (z1 , z2 , z3 ) = (z1 + iz2 − iz3 , −iz1 , iz1 − iz3 ) .
(9)[18]

Question 6: 12 Marks

It is given that A ∈ M3×3 (C) is a normal matrix with eigenvalues 1 and −1 and corresponding
eigenspaces  
1 1
E1 = span (2, 3, 6) , (6, 2, −3)
7 7
and  
1
E−1 = span (3, −6, 2) .
7

(6.1) Find the spectral decomposition of A. (11)

6
 MAT3701/000/1/2020

(6.2) Find A. (1)

Solution

(6.1) ∗ −6
2 6
 2 6
2 6
  40 18

7 7 7 7 7 7 49 49 49
 2 3 6
 
     
3 2
 3 2
   7 7 7  
P1 =    = 3 2   =  18 13 12 
7 7  7 7  7 7  6 2 −3
 49 49 49 
      
7 7 7
6 −3 6 −3 6 −3 −6 12 45
7 7 7 7 7 7 49 49 49

and ∗
3 3 3 9 −18 6
     
7 7 7 49 49 49
      
 −6   −6
 =  −6 −6
   3 2
  −18 36 −12 

P−1 =
7

 7  7

 7 7 7
=
 49 49 49 
.
      
2 2 2 6 −12 4
7 7 7 49 49 49

The spectral decomposition is

A = 1 · P1 + (−1) · P−1
 40 18 −6   9 −18 6

49 49 49 49 49 49
   
 18 13 12 
  −18 36 −12 

=1·
 49 49 49 
+ (−1) · 
 49 49 49 
.
   
−6 12 45 6 −12 4
49 49 49 49 49 49

(11)

(6.2)
31 36 −12
 
49 49 49
 
−23
 36 24

A=
 49 49 49
.

 
−12 24 41
49 49 49

(1)[12]

TOTAL[100]

7
 UNIVERSITY EXAMINATIONS UNIVERSITEITSEKSAMENS

university
of south africa

MAT3701 October/November 2020

LINEAR ALGEBRA III

Duration : 2 Hours 100 Marks

EXAMINERS :
FIRST : PROF JD BOTHA
EXTERNAL : PROF LM PRETORIUS

This is an open book examination. The examination question paper remains the property of the University of
South Africa.

This examination question paper consists of 3 pages.

The exam is IRIS invigilated. This requires that IRIS be activated at the start of the exam and kept on for the full duration
of the exam. The originality of submissions will also be verified with Turnitin.

Answer all the questions and show all calculations. Since this is an open book examination, if the solution to a simi-
lar question is given, no marks will be awarded.

Please stop writing at 11h00. You then have one hour in which to scan your answer sheets (please number the pages
clearly and write your student number and module code on the front page) and submit it as a pdf file under the name
studentnumber MAT3701 on this, the myExams platform. No submissions will be accepted after 12h00.

TO UPLOAD YOUR ANSWER FILE

1. When you have finished writing and created a pdf answer file, go to Submission on the Exam site and under At-
tachments, click the Choose file button next to Select a file from computer. Then browse your computer for your
answer file and select it for uploading. Remember to submit your answer file as a pdf file under the name studentnum-
ber MAT3701.
2. Once you have attached your answer file, the name of the file as well as the file size and upload time stamp will be
displayed under Attachments.
3. Tick the “Honour Pledge” button if you agree.
4. Click the “Submit” button.
5. After you have finished uploading your answer file, click on the “Submit” button on the IRIS pop-up screen. IRIS will
then upload your session recording files. Remember not to close the window until IRIS is finished.

Open Rubric
 2 MAT3701
October/November 2020

QUESTION 1

Consider the vector space V = C 2 with scalar multiplication over the real numbers R and let W and U be the
subspaces of V defined by
W = {(z1 , z2 ) ∈ V : z2 = z1 + z̄1 } and U = {(z1 , z2 ) ∈ V : z2 = z1 − z̄1 }.

(1.1) Find a basis for W . (4)

(1.2) Find a basis for U . (4)

(1.3) Find a basis for W ∩ U . (4)

(1.4) Explain whether V = W ⊕ U . (4)

(1.5) Explain whether W is a subspace over C. (2)

[18]

QUESTION 2

Let fa , fb , fc ∈ P2 (R) denote the Lagrange polynomials associated with the distinct real numbers a, b, c respectively.
Let V = span{fa , fb } and W = span{fa + fc }, and let g ∈ P2 (R).

(2.1) Show that B = {fa , fb , fa + fc } is linearly independent. (6)

(2.2) Show that P2 (R) = V ⊕ W . (2)

(2.3) Find a formula for T (g) expressed as a linear combination of {fa , fb , fc }, where T : P2 (R) → P2 (R) (8)
denotes the projection on V along W .
[16]

QUESTION 3

Let T : R4 → R4 be the linear operator defined by T (a, b, c, d) = (a, b + c, c, d).

(3.1) Show that the T -cyclic subspace of R4 generated by an arbitrary fixed (a, b, c, d) ∈ R4 is of dimension (8)
at most 2.

(3.2) Find the values of a, b, c, d such that the T -cyclic subspace in (3.1) is of dimension 1. (8)

(3.3) Find the values of a, b, c, d such that the T -cyclic subspace in (3.1) is of dimension 0. (2)
[18]

QUESTION 4

Consider the inner product space C 2 over C with inner product h·, ·i defined by
h(x1 , x2 ), (y1 , y2 )i = x1 y 1 + (1 + i)x1 y 2 + (1 − i)x2 y 1 + 3x2 y 2 .

(4.1) Show that h(x1 , x2 ), (x1 , x2 )i > 0 for all (x1 , x2 ) 6= (0, 0). (8)

(4.2) Apply the Gram-Schmidt orthogonalization process to {(1, 0), (0, 1)} to conctruct an orthonormal basis (4)
for C 2 with respect to h·, ·i.
[12]

[TURN OVER]
 3 MAT3701
October/November 2020

QUESTION 5

Let V be an inner product space over R with dimension n ≥ 2, and let u ∈ V be a fixed unit vector. Let T : V → V
be the linear operator defined by T (v) = −v + 2hv, uiu for all v ∈ V .

(5.1) Show that T is self-adjoint. (5)

(5.2) Show that T 2 = I. (5)

(5.3) Given that λ = −1 and λ = 1 are eigenvalues of T , find the associated eigenspaces. (10)
[20]

QUESTION 6

Let  
1 0 −1
A= 0 i 0 .
1 0 1
Given that the eigenvalues of A are i, 1 + i, and 1 − i, find the spectral decomposition of A. Write each orthogonal
projection matrix as a single matrix.
[16]

TOTAL MARKS: [100]

c
UNISA 2020
 MAT3701/000/2/2020

Tutorial letter 000/2/2020

LINEAR ALGEBRA III

MAT3701

Semester 2

Department of Mathematical Sciences

Exam Paper and Memo

MAT3701 October/November 2020

BARCODE

university
Define tomorrow. of south africa
Question 1: 18 Marks

Consider the vector space V = C 2 with scalar multiplication over the real numbers R and let W
and U be the subspaces of V defined by

W = {(z1 , z2 ) ∈ V : z2 = z1 + z̄1 } and U = {(z1 , z2 ) ∈ V : z2 = z1 − z̄1 }.

(1.1) Find a basis for W . (4)

(1.2) Find a basis for U . (4)

(1.3) Find a basis for W ∩ U . (4)

(1.4) Explain whether V = W ⊕ U . (4)

(1.5) Explain whether W is a subspace over C. (2)

Solution

(1.1)
W = {(a + ib, 2a) : a, b ∈ R} = {a(1, 2) + b(i, 0) : a, b ∈ R}
hence W = span{(1, 2), (i, 0)}, and therefore {(1, 2), (i, 0)} is a basis for W . (4)

(1.2)
U = {(a + ib, 2ib) : a, b ∈ R} = {a(1, 0) + b(i, 2i) : a, b ∈ R}
hence U = span{(1, 0), (i, 2i)}, and therefore {(1, 0), (i, 2i)} is a basis for U . (4)

(1.3)
(z1 , z2 ) ∈ W ∩ U ⇒ z2 = z1 + z̄1 = z1 − z̄1 ⇒ 2z̄1 = 0 ⇒ z1 = 0 ⇒ z2 = 0.
Thus W ∩ U = {(0, 0)}, hence φ is a basis for W ∩ U . (4)

(1.4) Since dim(W ) = dim(U ) = 2 and dim(W ∩ U ) = 0, it follows that

dim(W + U ) = dim(W ) + dim(U ) − dim(W ∩ U ) = 4.

Hence V = W +U since W +U ⊆ V and dim(V ) = 4. Since we also have that W ∩U = {(0, 0)}
from (1.3), it follows that V = W ⊕ U . (4)

(1.5) No. For example, (1, 2) ∈ W but i(1, 2) = (i, 2i) ∈

/ W since (i, 2i) ∈ U . (2)[18]

Question 2: 16 Marks

Let fa , fb , fc ∈ P2 (R) denote the Lagrange polynomials associated with the distinct real numbers
a, b, c respectively. Let V = span{fa , fb } and W = span{fa + fc }, and let g ∈ P2 (R).

(2.1) Show that B = {fa , fb , fa + fc } is linearly independent. (6)

2
 MAT3701/000/2/2020

(2.2) Show that P2 (R) = V ⊕ W . (2)

(2.3) Find a formula for T (g) expressed as a linear combination of {fa , fb , fc }, where (8)
T : P2 (R) → P2 (R) denotes the projection on V along W .

Solution

(2.1) Suppose
a1 fa (x) + b1 fb (x) + c1 (fa + fc )(x) = 0.
Then  
x = a ⇒ a1 · 1 + b1 · 0 + c1 · (1 + 0) = 0
 a1
 + c1 =0
x = b ⇒ a1 · 0 + b1 · 1 + c1 · (0 + 0) = 0 ⇒ b1 =0
 
x = c ⇒ a1 · 0 + b1 · 0 + c1 · (0 + 1) = 0 c1 =0
 

Thus a1 = b1 = c1 = 0, hence B = {fa , fb , fa + fc } is linearly independent. (6)

(2.2) Since B is linearly independent and dim(P2 (R)) = 3, it follows that B is a basis for P2 (R).
Hence
P2 (R) = span(B) = span{fa , fb } ⊕ span{fa + fc } = V ⊕ W.
(2)

(2.3) According to the Lagrange interpolation formula

g = g(a)fa + g(b)fb + g(c)fc = (g(a) − g(c))fa + g(b)fb + g(c)(fa + fc )

hence
T (g) = (g(a) − g(c))fa + g(b)fb .
(8)[16]

Question 3: 18 Marks

Let T : R4 → R4 be the linear operator defined by T (a, b, c, d) = (a, b + c, c, d).

(3.1) Show that the T -cyclic subspace of R4 generated by an arbitrary fixed (a, b, c, d) ∈ R4 (8)
is of dimension at most 2.

(3.2) Find the values of a, b, c, d such that the T -cyclic subspace in (3.1) is of dimension 1. (8)

(3.3) Find the values of a, b, c, d such that the T -cyclic subspace in (3.1) is of dimension 0. (2)

3
 Solution

(3.1) The first three generators of the T -cyclic subspace W of R4 generated by (a, b, c, d) are

(a, b, c, d), T (a, b, c, d) = (a, b + c, c, d), T 2 (a, b, c, d) = T (a, b + c, c, d) = (a, b + 2c, c, d).

Since

(a, b + 2c, c, d) = 2(a, b + c, c, d) − (a, b, c, d), i.e. T 2 (a, b, c, d) = 2T (a, b, c, d) − (a, b, c, d)

it follows that dim(W ) ≤ 2. (8)

(3.2)
W = span{(a, b, c, d), (a, b + c, c, d)} = span{(a, b, c, d), (0, c, 0, 0)}
hence
dim(W ) = 1 ⇔ c = 0 and (a, b, d) 6= (0, 0, 0).
(8)

(3.3)
dim(W ) = 0 ⇔ (a, b, c, d) = (0, 0, 0, 0).
(2)[18]

Question 4: 12 Marks

Consider the inner product space C 2 over C with inner product h·, ·i defined by

h(x1 , x2 ), (y1 , y2 )i = x1 y 1 + (1 + i)x1 y 2 + (1 − i)x2 y 1 + 3x2 y 2 .

(4.1) Show that h(x1 , x2 ), (x1 , x2 )i > 0 for all (x1 , x2 ) 6= (0, 0). (8)

(4.2) Apply the Gram-Schmidt orthogonalization process to {(1, 0), (0, 1)} to conctruct an (4)
orthonormal basis for C 2 with respect to h·, ·i.

4
 MAT3701/000/2/2020

Solution

(4.1)

h(x1 , x2 ), (x1 , x2 )i = x1 x̄1 + (1 + i)x1 x̄2 + (1 − i)x2 x̄1 + 3x2 x̄2

= (x1 + (1 − i)x2 ) (x̄1 + (1 + i)x̄2 ) + x2 x̄2
= (x1 + (1 − i)x2 ) (x1 + (1 − i)x2 ) + x2 x̄2
= |x1 + (1 − i)x2 |2 + |x2 |2 ≥ 0

and

h(x1 , x2 ), (x1 , x2 )i = 0 ⇔ |x1 + (1 − i)x2 |2 + |x2 |2 = 0

⇔ |x1 + (1 − i)x2 |2 = |x2 |2 = 0
⇔ |x1 + (1 − i)x2 | = |x2 | = 0
⇔ x1 + (1 − i)x2 = x2 = 0
⇔ x1 = x2 = 0

(8)

(4.2)

v1 = (1, 0)
h(0, 1), (1, 0)i
v2 = (0, 1) − (1, 0)
h(1, 0), (1, 0)i
= (0, 1) − (1 − i)(1, 0)
= (−1 + i, 1)

The required basis is { kv11 k v1 , kv12 k v2 } = {(1, 0), (−1 + i, 1)} since h(1, 0), (1, 0)i = 1 and

h(−1+i, 1), (−1+i, 1)i = (−1+i)(−1−i)+(1+i)(−1+i)+(1−i)(−1−i)+3 = 2−2−2+3 = 1.

(4)[12]

Question 5: 20 Marks

Let V be an inner product space over R with dimension n ≥ 2, and let u ∈ V be a fixed unit vector.
Let T : V → V be the linear operator defined by T (v) = −v + 2hv, uiu for all v ∈ V .

(5.1) Show that T is self-adjoint. (5)

(5.2) Show that T 2 = I. (5)

(5.3) Given that λ = −1 and λ = 1 are eigenvalues of T , find the associated eigenspaces. (10)

5
 Solution

(5.1) For all v1 , v2 ∈ V

hT (v1 ), v2 i = h−v1 + 2hv1 , uiu, v2 i

= h−v1 , v2 i + h2hv1 , uiu, v2 i
= hv1 , −v2 i + 2hv1 , uihu, v2 i
= hv1 , −v2 i + hv1 , 2uhu, v2 ii
= hv1 , −v2 i + hv1 , 2hv2 , uiui
= hv1 , −v2 + 2hv2 , uiui
= hv1 , T (v2 )i

thus T ∗ = T . (5)

(5.2) For all v ∈ V

T 2 (v) = T (−v + 2hv, uiu)

= T (−v) + T (2hv, uiu)
= −T (v) + 2hv, uiT (u)
= −(−v + 2hv, uiu) + 2hv, ui(−u + 2hu, uiu)
= v − 2hv, uiu + 2hv, ui(−u + 2u)
= v − 2hv, uiu + 2hv, uiu
=v

thus T 2 = I. (5)

(5.3) For λ = −1,

T (v) = −v ⇔ −v + 2hv, uiu = −v

⇔ 2hv, uiu = 0
⇔ hv, ui = 0
⇔ v ∈ span{u}⊥ .

Since dim(V ) ≥ 2, span{u}⊥ 6= {0}, hence E−1 (T ) = span{u}⊥ .

For λ = 1,

T (v) = v ⇒ −v + 2hv, uiu = v

⇒ 2hv, uiu = 2v
⇒ hv, uiu = v
⇒ v ∈ span{u}

thus E1 (T ) ⊆ span{u}. Since also T (u) = −u + 2hu, uiu = −u + 2u = u, it follows that

u ∈ E1 (T ) and therefore span{u} ⊆ E1 (T ). Thus E1 (T ) = span{u}. (10)[20]

6
 MAT3701/000/2/2020

Question 6: 16 Marks

Let  
1 0 −1
A =  0 i 0 .
1 0 1
Given that the eigenvalues of A are i, 1 + i, and 1 − i, find the spectral decomposition of A. Write
each orthogonal projection matrix as a single matrix.

Solution

To calculate the corresponding eigenspaces, we solve the following homogeneous systems:

     
1 − i 0 −1 1−i 0 −1  0 
Ei (A) :  0 0 0  →  0 0 0  −(1 − i)R3 + R1 , thus Ei (A) = span 1
1 0 1−i 0 0 −1 + 2i 0
 
      
−i 0 −1 −i 0 −1  1 
1 
E1+i (A) : 0 −1 0 → 0 −1 0 R3 − iR1 , thus E1+i (A) = span
    √ 0
 2
1 0 −i 0 0 0 −i

      
i 0 −1 i 0 −1  1 
1  
E1−i (A) : 0 −1 + 2i 0 → 0 −1 + 2i 0 R3 + iR1 , thus E1−i (A) = span
    √ 0
 2
1 0 i 0 0 0 i


The spectral decomposition of A is

A = iPi + (1 + i)P1+i + (1 − i)P1−i

where
   
0   0 0 0
Pi = 1 0 1 0 = 0 1 0
0 0 0 0
   
1  1 0 i
P1+i = 12  0  1 0 i = 21  0 0

0
−i −i 0 1
   
1   1 1 0 −i
1 
P1−i = 2 0 1 0 −i = 2 0 0 0 .
i i 0 1

[16]

TOTAL[100]