2021 Fall MAS109 Introduction to Linear Algebra Week 5 Solutions

Section 4.3
11.    
2 3 4 1 3 4
   
Solution. Let A = 
1 −2 −1 , A1 = 2
 −2  , Use Cramer’s rule,
−1
3 1 1 4 1 1
det(A1 )
Then, x = det(A) .

2 3 4
−2 −1 3 4 3 4
det(A) = 1 −2 −1 = 2 − +3 = 2 · (−1) − (−1) + 3 · 5 = 14
1 1 1 1 −2 −1
3 1 1
1 3 4
−2 −1 3 4 3 4
det(A1 ) = 2 −2 −1 = −2 +4 = (−1) − 2 · (−1) + 4 · 5 = 21
1 1 1 1 −2 −1
4 1 1
so x = 23 .

13.
cos θ sin θ 0
cos θ sin θ
Solution. det(A) = − sin θ cos θ 0 = = cos2 θ + sin2 θ = 1 6= 0.
− sin θ cos θ
0 0 1
So A is invertible for all values of θ.
cos θ 0 − sin θ 0 − sin θ cos θ
C11 = = cos θ, C12 = − = sin θ, C13 = =0
0 1 0 1 0 0
sin θ 0 cos θ 0 cos θ sin θ
C21 = − = − sin θ, C22 = = cos θ, C23 = − =0
0 1 0 1 0 0
sin θ 0 cos θ 0 cos θ sin θ
C31 = = 0, C32 = − = 0, C33 = = cos2 θ + sin2 θ = 1
cos θ 0 − sin θ 0 − sin θ cos θ
   
cos θ sin θ 0 cos θ − sin θ 0
  T
 
− sin θ cos θ 0. Then, adj(A) = C =  sin θ
So C =    cos θ 0 .
0 0 1 0 0 1
 
cos θ − sin θ 0
−1 1
 
By Theorem 4.3.3, A = det(A) adj(A) =  sin θ cos θ 0 
0 0 1

36. !
2 −1 3 −1 3 2
Solution. (a) u × v = ,, − = (−4, 9, 6)
2 −3 −3 0 2 0
!
2 −3 0 −3 0 2
v×w = , − , = (32, −6, −4)
6 7 2 7 2 6
!
9 6 −4 6 −4 9
(u × v) × (v × w) = , − , = (0, 176, −264)
−6 −4 32 −4 32 −6
Actually,it is parallel
 to v. This is no coincidence.
w
 
Let A =  u  . Use formula a × (b × c) = b(a · c) − c(a · b). (Check it out)

v

1 Typeset by LATEX
2021 Fall MAS109 Introduction to Linear Algebra Week 5 Solutions

Then, (u × v) × (v × w) = (w · (u × v))v = det(A)v

!
2 −1 3 −1 3 2
(b) u × w = , − , = (20, −23, 14)
6 7 2 7 2 6
Then u × (v − 2w) = u × v − 2u × w = (−44, 55, −22)

(c) (u × v) − 2w = (−4, 9, 6) − (4, 12, 14) = (−8, −3, −8)

50. !
v2 v3 v1 v3 v1 v2
Solution. (a) v × w = , − ,
w2 w3 w1 w3 w1 w2
u1 u2 u3
v2 v3 v1 v3 v1 v2
Then u · (v × w) = u1 − u2 + u3 = v1 v2 v3
w2 w3 w1 w3 w1 w2
w1 w2 w3

(b) |u · (v × w)| = kukkv × wk| cos θ| where θ is angle between u and v × w.

Geometric interpretation of kv × wk is area of the parallelogram P that has v and w as adja-
cent sides. And v × w is orthogonal to this parallelogram P . So kuk| cos θ| can be interpreted
as height (parallelogram P base) of parallelepiped that generated by u, v, and w.
Therefore geometric interpretation of |u · (v × w)| is volume of parallelepiped that generated
by u, v, and w.

D8.
Solution. (a) True.
Let xij be the ith row,jth column entry of A adj(A).
By Theorem 4.3.1, if i 6= j, then xij = 0.
So A adj(A) is a diagonal matrix.

(b) False.
If determinant is zero, then denominator of formula is zero. So Cramer’s rule cannot be used
for this system of linear equation.

(c) True.
Since 1
det(A) A adj(A) = I, so adj(A)−1 = 1
det(A) A.

(d) False. " # " #

1 1 0 −1
Let A = . Then adj(A) = . But adj(A) has no row of zeros.
0 0 0 1

(e) True.
u1 u2 u3
u · (v × w) = v1 v2 v3 (∵ Exercise 4.3.50.(a))
w1 w2 w3
w1 w2 w3
= − v1 v2 v3 (∵ Theorem 4.2.2.(b))
u1 u2 u3

2 Typeset by LATEX
2021 Fall MAS109 Introduction to Linear Algebra Week 5 Solutions

w1 w2 w3
= u1 u2 u3 (∵ Theorem 4.2.2.(b))
v1 v2 v3
= w · (u × v) (∵ Exercise 4.3.50.(a))
= (u × v) · w (∵ Theorem 1.2.6.(a))

Section 4.4
13.
Solution. Observe that each matrix is triangular, so that one can easily observe that the diagonal
components are the eigenvalues. We give a specific calculation for students’ convenience.
" #
−1 6 λ+1 −6
(a) Let A = . Then p(λ) = det(λI − A) = = (λ + 1)(λ − 5).
0 5 0 λ−5
Eigenvalues of A: −1, 5.
 
3 0 0 λ−3 0 0
 
(b) Let B = 
−2 7 0
 . Then p(λ) = det(λI − B) = 2 λ−7 0
4 8 1 −4 −8 λ−1
= (λ − 3)(λ − 7)(λ − 1).
Eigenvalues of B: 1, 3, 7.
 
− 13 0 0 0 λ+ 1
3 0 0 0

 0 − 3 0 01  1
. Then p(λ) = det(λI − C) = 0 λ+ 3 0 0
(c) Let C =   0
 0 1 0  0 0 λ−1 0
1 1
0 0 0 2 0 0 0 λ− 2
= (λ + 13 )2 (λ − 12 )(λ − 1).
Eigenvalues of C: 1, 12 , − 13 .

18.
λ−1 −3 −7 −11
1
0 λ− −3 −8
Solution. p(λ) = det(λI − A) = 2
= λ(λ − 1)(λ − 21 )(λ − 2)
0 0 λ −4
0 0 0 λ−2
Eigenvalues of A: 1, 12 , 0, 2.
        
0 −3 −7 −11 x 0 x 1
 1    
0 2 −3 −8   y  0
 
y 
 
  = t 0.
 
Eigenvectors of A: Case λ = 1,    =   ⇒
0 0
 1 −4 
  z  0
    z 
 
0
 
0 0 0 −1 w 0 w 0
        
1
− −3 −7 −11 x 0 x 6
 2        
 0 0 −3 −8   y  0
    
 y  = t −1.
   
Case λ = 12 ,  1
= ⇒
 0
 0 2 −4   z  0
    
z 
 
0
 

0 0 0 − 32 w 0 w 0

3 Typeset by LATEX
2021 Fall MAS109 Introduction to Linear Algebra Week 5 Solutions

        
−1 −3 −7 −11 x 0 x 11
 1    
 0 − 2 −3 −8   y  0
   
   =   ⇒  y  = t −6.
   
Case λ = 0, 
0
 0 0 −4   z  0
    
z 
  1
 
0 0 0 −2 w 0 w 0
        
1 −3 −7 −11 x 0 x 53
0 32 −3 −8   y  0
        
y  28/3
Case λ = 2,    =   ⇒  z  = t  2 .
         
0 0 2 −4 
  z  0    
0 0 0 0 w 0 w 1
1
By Theorem 4.4.6, eigenvalues of A9 : 1, 512 , 0, 512
       
1 6 11 53
       
0 −1 −6 28/3
And corresponding eigenvectors: t   , t   , t   , t 
       
0  0   1   2 


0 0 0 1
26.
λ−3 0 0
Solution. p(λ) = det(λI − A) = 0 λ−x −2 = (λ − 3)(λ2 − 2xλ + x2 − 4)
0 −2 λ−x
= (λ − 3)(λ − x − 2)(λ − x + 2)
Hence A has eigenvalues 3, x − 2, x + 2. Here x − 2 6= x + 2 for any x, so the repeated eigenvalue
must be 3. Therefore, it must be the case that x + 2 = 3 or x − 2 = 3. Then x = 1 or x = 5. These
are only value of x that make A have at least one repeated eigenvalue.

32.
Solution. (Ax) · x = (λx) · x = λ(x · x) = λkxk2
Since x is nonzero vector, kxk =
6 0.
(Ax)·x
So λ = kxk2

33.
Solution. (a) Inductionhon n.
i
If n = 1, then C = −c0 . So p(λ) = λ + c0 . (Statement is true if n = 1)
Suppose statement is true for n = k − 1, consider n = k case.
p(λ) = det(λI − C)
λ 0 0 ··· 0 c0
−1 λ 0 ··· 0 c1
= 0 −1 λ ··· 0 c2
.. .. .. .. ..
. . . . .
0 0 0 ··· −1 λ + ck−1
2
0 λ 0 ··· 0 c0 + c1 λ
−1 λ 0 ··· 0 c1
= 0 −1 λ ··· 0 c2 (∵ Theorem 4.2.2.(c))
.. .. .. .. ..
. . . . .
0 0 0 ··· −1 λ + ck−1

4 Typeset by LATEX
2021 Fall MAS109 Introduction to Linear Algebra Week 5 Solutions

λ2 0 ··· 0 c0 + c1 λ
−1 λ ··· 0 c2
= .. .. .. .. (Expand by cofactors along the first column).
. . . .
0 0 ··· −1 λ + ck−1
c0
λ 0 ··· 0 λ + c1
−1 λ ··· 0 c2
=λ .. .. .. .. (∵ Theorem 4.2.2.(a))
. . . .
0 0 · · · −1 λ + ck−1
= λ(( cλ0 + c1 ) + c2 λ + · · · + ck−1 λk−2 + λk−1 ) (∵ Induction hypothesis)
= c0 + c1 λ + c2 λ2 + · · · + ck−1 λk−1 + λk .
Therefore statement is true for any positive integer n.

(b) c0 = 2, c1 = −3, c2 = 1, c3 = −5.

 
0 0 0 −2
 
1 0 0 3 
Companion matrix: 0 1 0 −1

 
0 0 1 5
.
D8.
Solution. (a)"False.#
1 0
Let A = . Then eigenvalues of A are 1, 2.
0 2
But the row reduced echelon form of A is I, and I have only one eigenvalue 1.

(b) True.
Let x be a eigenvector of A with eigenvalue λ1 and y be a eigenvector of A with eigenvalue
λ2 . (λ1 6= λ2 )
Suppose x + y is eigenvector of A with eigenvalue λ.
Then λ(x + y) = A(x + y) = Ax + Ay = λ1 x + λ2 y.
⇒ (λ − λ1 )x = (λ2 − λ)y
Since λ1 6= λ2 , one of them should be distinct to λ.
Without loss of generality, let λ1 6= λ.
λ2 −λ
Then x = λ−λ1 y
2 −λ λ2 −λ λ2 −λ
⇒ λ1 x = Ax = A( λλ−λ 1
y) = λ−λ1 (Ay) = λ2 λ−λ 1
y = λ2 x
⇒ (λ1 − λ2 )x = 0 ⇒ x = 0.
It is a contradiction to x is a nonzero vector.

(c) True.
Let p(λ) = λ3 + a1 λ2 + a2 λ + a3 be the characteristic polynomial for matrix A.
Since lim p(λ) = ∞ and lim p(λ) = −∞.
λ→∞ λ→−∞
There exist some a < b such that p(a) < 0 < p(b).
By the intermediate value theorem, there exists some c ∈ (a, b) such that p(c) = 0. (polynomial
is continuous)
So 3 × 3 matrix A has at least one real eigenvalue.

5 Typeset by LATEX
2021 Fall MAS109 Introduction to Linear Algebra Week 5 Solutions

(Another solution) If one is not familiar with the analytic tool, then one may have an algebraic
approach. One may observe that if p(λ) = 0, then p(λ) = 0. Namely, if p(λ) = λ3 + a1 λ2 +
3 2
a2 λ + a3 = 0, then p(λ) = λ + a1 λ + a2 λ + a3 = λ3 + a1 λ2 + a2 λ + a3 = 0 = 0.1 Therefore,
p(λ) must have an even number of non-real roots. Since a cubic polynomial always have 3
roots(counting with multiplicity), at least one root must be real.

(d) True.
Since p(λ) = det(λI − A), p(0) = det(−A) = (−1)n det(A).
And p(0) = 1 6= 0. So A is invertible.
.
P2.
Solution. (a) Let pA (λ) be the characteristic polynomial for matrix A.
Then pA (λ) = det(λI − A) = det((λI − A)T ) (∵ Theorem 4.2.1)
= det((λI)T − AT ) = det(λI − AT ) = pAT (λ).
So A and AT have the same chararcteristic polynomial.

λ−2 0
(b) p(λ) = det(λI − A) = = (λ − 2)(λ − 3).
−2 λ−3
So A have eigenvalues 2, 3. " #" # " # " # " #
0 0 x 0 x 1
Eigenvectors of A: Case λ = 2, = ⇒ =t
−2 −1 y 0 y −2
" #" # " # " # " #
1 0 x 0 x 0
Case λ = 3, = ⇒ =t
−2 0 y 0 y 1
" #" # " # " # " #
0 −2 x 0 x 1
Eigenvectors of AT : Case λ = 2, = ⇒ =t
−1 y 0 0 y 0
" #" # " # " #
" #
1 −2 x x 0 2
Case λ = 3, = ⇒ =t
0 0 y 0 y 1
So A and AT need not have same eigenspaces.
.
P3.
Solution. Since x is eigenvector of A with eigenvalue λ, Ax = λx.
Since A is invertible, x = A−1 Ax = A−1 (λx) = λA−1 x.
If λ = 0, then det(A) = 0 (By Theorem 4.4.12.(a)).
It is a contradiction to A is invertible.
So λ 6= 0 then A−1 x = λ1 x.
Therefore 1
λ is an eigenvalue of A−1 , and x is a corresponding eigenvector.
P7.
Solution. Since Ax = λx and Bx = x, then ABx = Ax = λx and BAx = B(λx) = λBx = λx
So λ is an eigenvalue of both AB and BA and that x is a corresponding eigenvector.

1 Here we used the fact that the conjugate of a real number is itself. Also the addition/multiplication of conjugates
are equal to the conjugate of addition/multiplication.

6 Typeset by LATEX