MAT3701

May/June 2012
LINEAR ALGEBRA

Duration : 2 Hours 100 Marks

EXAMINERS :
FIRST : PROF JD BOTHA
EXTERNAL : PROF LM PRETORIUS
This examination question paper remains the property of the University of South Africa
and may not be removed from the examination venue.

This paper consists of 3 pages

Answer ALL the questions.

QUESTION 1
Let V be the vector space C 2 with scalar multiplication over the real number R. Let

U = {(z1 , z2 ) ∈ V : z 1 = z1 and z 2 = iz2 } .

(a) Show that U is a subspace of V. (6)

(b) Find a basis for U. (6)

(c) Given that

W = {(z1 , z2 ) ∈ V : z 1 = z2 }

is a subspace of V , determine whether V = U ⊕ W. (10)

[22]

QUESTION 2
Let T : M2×2 (C) → M2×2 (C) be the linear operator defined by
   
a b 2a b+c
T = .
c d b+c 2d

(a) Show that T satisfies the test for diagonalizability. (14)

(b) Is T invertible? State your reasons(s). (2)

[16]

16

Open Rubric
 MAT3701/102

QUESTION 3

Let
1 1
 
2 2
1
1
A= 2
0 0 .
1
0 2
0
(a) Show that A is a regular transition matrix. (4)
(b) Find lim Am . (8)
m→∞

(c) Describe the Gerschgorin discs in which the eigenvalues of A lie. (4)
[16]

QUESTION 4

n o
Let P denote the orthogonal projection of C 3 on W = span (1, 0, 0) , √1 (0, 1, i) .
2

(a) Find the formula for P (z1 , z2 , z3 ). (7)

(b) Find the eigenvalues of P and a basis for each of the corresponding eigenspaces. (6)
(c) Find a basis τ for C 3 and a diagonal matrix D such that [P ]τ = D. (2)
[15]

QUESTION 5

It is given that A ∈ M3×3 (C) is a normal matrix with eigenvalues −1 and 1 and corresponding
eigenspaces  
1
E−1 = span (0, 1, 0) , √ (i, 0, 1)
2
and  
1
E1 = span √ (1, 0, i) , respectively.
2
Find the spectral decomposition of A. [15]

QUESTION 6
Let
 
1 3
A= .
3 1

17
 (a) Find kAk , kA−1 k and cond(A) . (9)

(b) Suppose x and x e are vectors such that Ax = b, kbk = 1 and kb − Ae xk ≤ 0.001. Use (a)
to determine upper bounds for ke x − A−1 bk (the absolute error) and ke
x − A−1 bk / kA−1 bk (the
relative error). (7)

[16]

TOTAL: [100]

18
 MAT3701/102

Memorandum: May/June 2012 Exam

QUESTION 1

(a) Let (z1 z2 ) , (z3 , z4 ) ∈ U and a ∈ R.

S1 : (0, 0) ∈ U since 0 = 0 and 0 = i0.

S2 : (z1 , z2 ) + (z3 , z4 ) = (z1 + z3 , z2 + z4 ) ∈ U since
z1 + z3 = z1 + z3 = z1 + z3 and z2 + z4 = z2 + z4 = iz2 + iz4 = i (z2 + z4 ) .
S3 : a (z1 , z2 ) = (az1 , az2 ) ∈ U since az 1 = az1 (a real) = az1 and az 2 = az2 = a (iz2 ) = i (az2 ) .

(6)

(b)
(a + ib, c + id) ∈ U ⇔ a − ib = a + ib and c − id = ic − d
⇔ a ∈ R arbitrary, b = 0, and d = −c
∴U = {(a, c − ic) : a, c ∈ R}
= {a (1, 0) + c (0, 1 − i) : a, c ∈ R}
= span {(1, 0) , (0, 1 − i)}

A basis for U is therefore α = {(1, 0) , (0, 1 − i)} . (6)

(c) (a + ib, c + id) ∈ W ⇔ a − ib = c + id ⇔ a = c and b = −d

∴W = {(a + ib, a − ib) : a, b ∈ R}

= {a (1, 1) + b (i, −i) : a, b ∈ R}

A basis for W is therefore β = {(1, 1) , (i, −i)}. The set α ∪ β is linearly independent, since

x1 (1, 0) + x2 (0, 1 − i) + x3 (1, 1) + x4 (i, −i) = (0, 0)


x1 + x3 + ix4 = 0
⇔
(1 − i) x2 + x3 − ix4 = 0


 x1 +x3 =0
x4 =0

⇔

 x2 +x3 =0
x2 x4 =0

⇔ x1 = x2 = x 3 = x4 = 0

Thus, α ∪ β is a basis for V, and therefore V = U ⊕ W. (10)

19
 QUESTION 2

(a) Let
        
1 0 0 1 0 0 0 0
E= E1 = ; E2 = ; E3 = ; E4 =
0 0 0 0 1 0 0 1
 
2 0
T (E1 ) = = 2E1
0 0
 
0 1
T (E2 ) = = E2 + E3
1 0
 
0 1
T (E3 ) = = E2 + E3
1 0
 
0 0
T (E4 ) = = 2E4
0 2
 
2 0 0 0
 0 1 1 0 
∴ [T ]E = 
 0

1 1 0 
0 0 0 2

λ−2 0 0 0
0 λ − 1 −1 0
= (λ − 2)2 (λ − 1)2 − 1
 
det (λI4 − [T ]E ) =
0 −1 λ−1 0
0 0 0 λ−2

= (λ − 2)2 (λ − 2) (λ)

The eigenvalues of T are 0 and 2 (multiplicity 3).

 
0 0 0 0
 0 1 −1 0 
2I4 − [T ]E = 
 0 −1 1

0 
0 0 0 0

which is of rank 1. Since

4− rank (2I4 − [T ]E ) = 3, the multiplicity of λ = 2, it follows that T satisfies the test for
diagonalizability

(14)

(b) No, since 0 is an eigenvalue of T . (2)

[16]

20
 MAT3701/102

QUESTION 3

(a) A is a transition matrix since all entries are nonnegative and each column adds up to 1.

1 3 1 5 1 1
   
2 4 2 8 2 2
A2 =  1
4
1
4
1
2
 ; A3 =  1
4
3
8
1
4

1 1 1 1
4
0 0 8 8 4

(4)
3
Since all entries in A are positive, A is also regular.

(b)
− 21 1
     
2
1 −1 1 2 −1 0 4
1
A−I =  2
−1 0  → 0
 − 21 1  → 0
 1 −2 
1
0 2
−1 0 1 −2 0 0 0
   
4 4 4 4
1
∴  2  ∈ E1 (A) ∴ lim Am =  2 2 2 .
m→∞ 7
1 1 1 1

(8)

(c)
1 3

C1 = z∈C : z− 2
≤ 2
(1st row)

1


C3 = C2 = z ∈ C : |z| ≤ 2
2nd and 3rd rows

(4)
[16]

QUESTION 4

(a)
D E
P (z1 , z2 , z3 ) = h(z1 , z2 , z3 ) , (1, 0, 0)i (1, 0, 0) + (z1 , z2 , z3 ) √12 (0, 1, i) √12 (0, 1, i) ,
= z1 (1, 0, 0) + 21 (z2 − iz3 ) (0, 1,
i)
= z1 , 12 (z2 − iz3 ) , 21 (iz2 + z3 )

(7)
n o
(b) λ = 1, E1 = W, with basis (1, 0, 0) , √1 (0, 1, i)
2

λ = 0, E0 = W ⊥ = span{(0, i, 1)} , with basis {(0, i, 1)} (6)

 
1 0 0
(c) τ = {(1, 0, 0) , (0, 1, i) , (0, i, 1)} , D =  0 1 0 . (2)
0 0 0

[15]

21
 QUESTION 5
Let
0 √i2 √1
 
2
U = 1 0 0 .
0 √12 √i
2

Then
A = −1 · P1 + 1 · P2
where
0 √i 1
0 2i
     
1 0 0 2
 
∗ 0 1 0 2
P1 =U 0 1 0 U = 1 0 = 0 1 0 
− √i2 √1
   
0
0 0 0 0 √1 2 − 2i 0 12
2

 1  1
0 − 2i
   
0 0 0 √
2 h i 2
P2 =U 0 0 ∗
0 U =  0  √1 0 − √i2 = 0 0 0 
2
0 0 1 √i i
0 12
2 2

[15]

QUESTION 6
 
∗ 2 10 6
(a) A A = A =
6 10

∗ 10 − λ 6 = (10 − λ)2 − 62
Eigenvalues of A A :
6 10 − λ = (λ − 4) (λ − 16)
∴ λ = 4, 16.
√
∴ kAk = 16 = 4, kA−1 k = √1
4
= 1
2

cond (A) = ||A|| ||A−1 || = 2.

(b)
x − A−1 b|| = ||A−1 (Ae
||e x − b)|| ≤ ||A−1 || ||Ae
x − b||
≤ 12 × 0.001 = 0.0005

x − A−1 b||
||e ||e
x − x||
−1
= ≤ cond(A) b − eb / kbk
||A b|| kxk
= cond(A) kb − Ae xk / kbk
= 2 × 0.001
1
= 0.002

(7)
[16]

TOTAL [100]

22
 MAT3701/102

MAT3701
October/November 2012
LINEAR ALGEBRA

Duration : 2 Hours 100 Marks

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

Closed book examination

This examination question paper remains the property of the University of South Africa
and may not be removed from the examination venue.

This paper consists of 3 pages

Answer ALL the questions.

QUESTION 1

Let fa , fb , fc denote the Lagrange polynomials associated with the distinct real numbers a, b, c
respectively. Let T : P2 (R) → P2 (R) be defined by T (g) = g (a) fa .

(a) Write down the formula that expresses g ∈ P2 (R) as a linear combination of fa , fb and fc . (3)

(b) Write down fa (a) , fa (b) and fa (c) . (2)

(c) Show that T is a linear operator. (6)

(d) Explain whether or not T is a projection. (5)

(e) Find [T ]β , where β = {fa , fb , fc } . (4)

[20]

QUESTION 2

Let T : R3 → R3 be the linear transformation defined by

T (a, b, c) = (a + b, b + c, c + a) ,

and let W be the T -cyclic subspace of R3 generated by v = (1, −1, 0) .

(a) Find the T -cyclic basis β of W. (6)

(b) Find the characteristic polynomial of TW (the restriction of T to W ). (3)

23
 (c) Explain whether the subspace

U = (a, b, c) ∈ R3 a + b + c = 0


is T -invariant. (6)

[15]

QUESTION 3

Let T : P2 (R) → P2 (R) be the linear operator defined by

T a + bx + cx2 = (2a + b) + (a + 2b) x + (a + b + c) x2 .

(a) Show that T satisfies the test for diagonalizability. (12)

(b) Show that T satisfies the polynomial p (x) = (x − 1) (x − 3) . (5)

[17]

QUESTION 4

Let T : V → V be a linear operator on a finite-dimensional inner product space V over C and suppose
W is a T − invariant subspace of V.

(a) Prove that W ⊥ is T ∗ invariant. (6)

(b) Suppose T is self-adjoint, and β 1 and β 2 are basis forW and W ⊥, respectively. Let β = β 1 ∪ β 2 .
B1 0
Show that [T ]β is block diagonal, that is, [T ]β = where B1 and B2 are square
0 B2
matrices (not necessarily of the same order). (6)

[12]

QUESTION 5

∗
Let M2×2 (R) be the inner product space with inner product defined  hA, Bi =tr (B A) and
 by let
1 1 1 1 1 1
P : M2×2 (R) → M2×2 (R) be the orthogonal projection on W = span , .
2 −1 −1 2 1 1
 
a b
(a) Find a formula for P . (7)
c d

24
 MAT3701/102

(b) Find a basis for the subspace along which P projects. (5)

[12]

QUESTION 6

(a) Let V be a finite-dimensional inner product space over C, and let T : V → V be a linear
operator.

(i) State necessary and sufficient conditions for T to be an orthogonal projection. (2)
(ii) Show that if T is an orthogonal projection then I − T is also an othogonal projection. (5)
 
1 i 0
(b) Show that A =  i 1 0  is normal. (6)
0 0 1+i

(c) Given that the eigenvalues of A in (b) are 1 + i and 1 − i, find the spectral decomposition of A.
(11)

[24]

TOTAL: [100]

25
Memorandum: October/November 2012 Exam
QUESTION 1
(a) g = g (a) fa + g (b) fb + g (c) fc (3)
(b) fa (a) = 1, fa (b) = fa (c) = 0 (2)
(c) Let g, h ∈ P2 (R) and k ∈ R.
T (g + h) = (g + h) (a) fa
= (g (a) + h (a)) fa
= g (a) fa + h (a) fa
= T (g) + T (h)

and
T (kg) = (kg) (a) fa
= k g (a) fa
= k T (g)
(6)
(d)
T 2 (g) = T (g (a) fa )
= g (a) T (fa )
= g (a) fa (a) fa
= g (a) fa since fa (a) = 1
= T (g)
Hence T is a projection since T 2 = T. (5)
(e)
T (fa ) = fa (a) fa = fa
T (fb ) = fb (a) fa = 0 since fb (a) = 0
T (fc ) = fc (a)f
a = 0 
1 0 0
Thus, [T ]β =  0 0 0  .
0 0 0
(4)
[20]

QUESTION 2
(a)
v = (1, −1, 0)
T (1, −1, 0) = (0, −1, 1)
T 2 (1, −1, 0) = T (0, −1, 1) = (−1, 0, 1)
= −v + T (v) · · · (i)
Hence β = {(1, −1, 0) , (0, −1, 1)}
(6)

26
 MAT3701/102

(b) cTW (t) = (−1)2 (t2 − t + 1) , from (i) (3)

(c) Suppose (a, b, c) ∈ U, hence a + b + c = 0. Then T (a, b, c) = (a + b, b + c, c + a) , and

(a + b) + (b + c) + (c + a) = 2 (a + b + c) = 0 so that T (a, b, c) ∈ U.
Therefore U is T − invariant. (6)
[15]

QUESTION 3

(a) Let β = {1, x, x2 } . Now

T (1) = 2 + x + x2
T (x) = 1 + 2x + x2
T (x2 ) = x2

Therefore  
2 1 0
[T ]β = 1
 2 0 
1 1 1

  λ − 2 −1 0
det λI − [T ]β = −1 λ−2 0
−1 −1 λ−1

= (λ − 1) (λ − 2)2 − 1
 

= (λ − 1)2 (λ − 3)

and we have that the characteristic polynomial of T splits over R.

 
−1 −1 0
E1 : −1
 −1 0 
−1 −1 0
 
Since rank I − [T ]β = 1, it follows that dim (E1 ) = 2 = multiplicity of (λ = 1) .
Therefore T satisfies the test for diagonalizability.

   
   1 1 0 −1 1 0
(b) [T ]β − I [T ]β − 3I = 1
 1 0   1 −1 0 =0 (2)
1 1 0 1 1 −2
and therefore T satisfies p (x) . (5)

[17]

27
 QUESTION 4
(a) Let w ∈ W ⊥ . For all w ∈ W,
hT ∗ (w) , wi = hw, T (w)i
= 0, since T (w) ∈ W

Thus T ∗ (w) ∈ W ⊥ , and therefore W ⊥ is T ∗ − invariant. (6)

(b) For each w ∈ β 1 it follows that T (w) will be a linear combination of the vectors in β 1 since W
is T − invariant, and similarly for the vectors in W ⊥ since W ⊥ is also T − invariant. Therefore
[T ]β is of the given form. (6)
[12]

QUESTION 5
(a) Since the given spanning set for W is orthonormal,
       
a b a b 1 1 1 1 1 1
P = , +
c d c d 2 −1 −1 2 −1 −1
     
a b 1 1 1 1 1 1
,
c d 2 1 1 2 1 1
   
1 1 1 1 1 1
= (a + b − c − d) + (a + b + c + d)
4 −1 −1 4 1 1
 
1 2a + 2b 2a + 2b
=
4 2c + 2d 2c + 2d
 
1 a+b a+b
=
2 c+d c+d

(b) The subspace along which P projects is ker (P ) .

Now  
a b
P = 0 ⇔ a = −b and c = −d
c d
  
a −a
∴ ker (P ) = : a, c ∈ R
c −c
   
1 −1 0 0
= span ,
0 0 1 −1
   
1 −1 0 0
A basis is therefore , (5)
0 0 1 −1
[12]

28
 MAT3701/102

QUESTION 6

(a) (i) T = T ∗ and T 2 = T.

(ii) (I − T )∗ = I ∗ − T ∗ = I − T since T ∗ = T and I ∗ = I and (I − T )2 = I − 2T + T 2 = I − T
since T 2 = T (5)

(b)    
1 i 0 1 −i 0
AA∗ = i 1 0   −i 1 0 
0 0 1+i 0 0 1−i
 
2 0 0
= 0 2 0 
0 0 2

= A∗ A

Thus A is normal. (6)

(c)

I = P1+i + P1−i
A = (1 + i) P1+i + (1 − i) P1−i
∴ A − (1 + i) I = ((1 − i) − (1 + i)) P1−i = −2iP
 1−i   
−i i 0 1 −1 0
1 1 1
∴ P1−i = − (A − (1 + i) I) = −  i −i 0  = 2
−1 1 0 
2i 2i
0 0 0 0 0 0
and  
1 1 0
1
P1+i = I − P1−i =  1 1 0 
2
0 0 2
Spectral decomposition of A :  1 1  1
− 12
 
2 2
0 2
0
1
A = (1 + i)  2 2 0  + (1 − i)  − 2 12
1 1
0 .
0 0 1 0 0 0

(11)
[24]

TOTAL [100]

29