MAM2011F, Linear Algebra 26–28 March 2025

Tutorial 6
Determinants, eigenvalues, eigenvectors, and vector spaces

 
1 0 0
1. Let A = 1 3 2.
1 2 3
(a) Find all the eigenvalues of A.

Solution: To find the eigenvalues of A, you need to solve the equation

det(λI − A) = 0.

λ−1 0 0
0 = det(λI − A) = −1 λ − 3 −2
−1 −2 λ − 3
λ − 3 −2
= (λ − 1)
−2 λ − 3
(expanded along the first row, which is most convenient
in this case)
= (λ − 1)[(λ − 3)(λ − 3) − 4]
= (λ − 1)[λ2 − 6λ + 5]
= (λ − 1)2 (λ − 5).

So the eigenvalues of A are 1 and 5.

(b) For each eigenvalue you found in (a), find a corresponding eigenvector.
[Hint: For each eigenvalue, you have to solve a system of linear equations to find
a corresponding eigenvector.]

Solution: To find an eigenvector x of A corresponding to the eigenvalue λ,

you need to find a non-zero solution x of the system of equations

(λI3 − A)x = 0.

(If your system has only the zero solution, there is something wrong with
your calculation of λ.)
 
4 0 0
ˆ For λ = 5, λI3 − A = −1 2 −2.
−1 −2 2

1
 The solutions of the system

(5I3 − A)x = 0
 
0
are all of the form µ 1, with µ ∈ R. Any vector of this form with
1
µ ̸= 0 is an eigenvector of A corresponding to the eigenvalue 5.
 
0 0 0
ˆ For λ = 1, λI3 − A = −1 −2 −2.
−1 −2 −2
The solutions of the system

(1I3 − A)x = 0
   
−2 −2
are all of the form α 1
  + β 0  with α, β ∈ R. Any non-zero

0  1  
−2 −2
vector of this form, such as  1  or  0 , will be an eigenvector
0 1
of A corresponding to the eigenvalue 1.

 
7 4 −4
2. Let A =  4 −8 −1.
−4 −1 −8
(a) Write down the determinant you have to evaluate to find the eigenvalues of A
(there will be λ’s in the determinant).

Solution:
λ − 7 −4 4
det(λI − A) = −4 λ + 8 1 .
4 1 λ+8

(b) Perform a row operation on this determinant to obtain a zero in the first column.

Solution: Add row 2 to row 3; this row operation does not change the value

2
 of the determinant:
λ − 7 −4 4
det(λI − A) = −4 λ + 8 1
4 1 λ+8
λ − 7 −4 4
= −4 λ + 8 1
0 λ+9 λ+9

(c) Then make use of the properties of determinants to find the value of the deter-
minant, in terms of λ.

Solution: Expand the last determinant in (b) above along the last row and
use the fact that λ + 9 is a common factor in this row:

−7 −4 4 λ − 7 −4 4
−4 λ + 8 1 = (λ + 9) −4 λ + 8 1
0 λ+9 λ+9 0 1 1
= (λ + 9)(λ2 − 81)
= (λ − 9)(λ + 9)2 .

(d) Now find all the eigenvalues of A.

Solution: The eigenvalues of A are the solutions to the equation

0 = det(λI − A) = (λ − 9)(λ + 9)2

and therefore are 9, −9, −9.

(e) For each eigenvalue you have found, find a corresponding eigenvector.

Solution: To find an eigenvector x of A corresponding to the eigenvalue λ,

we need to find a non-zero solution x of the system of equations
(λI3 − A)x = 0.
(If your system has only the zero vector as solution, there is something wrong
with your calculation of λ.)
ˆ For λ = 9,
 
2 −4 4
λI3 − A = −4 17 1  .
4 1 17

3
 We can reduce this matrix to the matrix
 
1 −2 2
0 1 1 .
0 0 0

The solutions of the system

(9I3 − A)x = 0
 
4
are therefore all of the form s  1 , with s ∈ R. Any vector of this
−1
form with s ̸= 0 is an eigenvector of A corresponding to the eigen-
value 9.

ˆ For λ = −9,
 
−16 −4 4
λI3 − A =  −4 −1 1  .
4 1 −1

We can reduce this matrix to the matrix

 
4 1 −1
0 0 0  .
0 0 0

The solutions of the system

(−9I3 − A)x = 0
   
−1 1
are therefore all of the form s 4
  + t 0, with s, t ∈ R. Any

0 4
vector x ̸= 0 of this form is an eigenvector of A corresponding to the
eigenvalue −9.

3. Let A and P be n × n matrices, with P invertible.

(a) Show that A and P −1 AP have the same eigenvalues.
[Hint: I = P −1 IP .]

4
 Solution:

det(λI − P −1 AP ) = det(λP −1 IP − P −1 AP )
= det(P −1 (λI − A)P )
= det(P −1 ) · det(λI − A) · det(P )
= det(P −1 ) · det(P ) · det(λI − A)
= det(P −1 P ) · det(λI − A)
= det(I) · det(λI − A)
= det(λI − A)

Make sure that you can justify each step. Note that in the above we use the
fact that real number multiplication is commutative; whereas matrix multi-
plication of course is not.
Since the eigenvalues of A and P −1 AP are respectively the solutions of
det(λI − A) = 0 and det(λI − P −1 AP ), it follows that the eigenvalues are
the same.

(b) If λ is an eigenvalue of A and x a corresponding eigenvector of A, is it also an

eigenvector of P −1 AP ? If not, find an eigenvector of P −1 AP corresponding to λ.

Solution:
The answer is No.
Claim 1. Let x is an eigenvector of A corresponding to λ. Then P −1 x is
an eigenvector of P −1 AP corresponding to λ.
Proof. We need to verify that

P −1 AP (P −1 x) = λ(P −1 x).

Indeed,

P −1 AP (P −1 x) = P −1 A(P P −1 )x
= P −1 Ax
= P −1 λx (since x is eigenvector of A corresponding to λ)
= λP −1 x.

5
4. This question asks you to check the axioms for a vector space in a specific example.
You will need to have your notes open at the definition of a vector space (page 91).
Let R2 = {(x1 , x2 ) : x1 , x2 ∈ R}. We have seen that, with the usual definitions for
addition and scalar multiplication of vectors, R2 is a vector space. In this question
we use the usual scalar multiplication, but define a different addition to the usual
one on R2 (we’ll denote it by ⊕ to emphasize the difference). Your task will be to
test whether R2 is a vector space with these two operations.

ˆ Addition on R2 is defined as follows: For x = (x1 , x2 ) and y = (y1 , y2 ) in R2 ,

x ⊕ y = (x1 + y1 + 1, x2 + y2 − 2).

ˆ Scalar multiplication on R2 is defined as follows: For x = (x1 , x2 ) in R2 and

λ ∈ R,

λx = (λx1 , λx2 ).
(a) Check that x ⊕ y ∈ R2 and λx ∈ R2 . (This is close to obvious; simply say why
it is obvious.)

Solution: It is clear that if x = (x1 , x2 ), y = (y1 , y2 ) ∈ R2 , then x ⊕ y =

(x1 + y1 + 1, x2 + y2 − 2) ∈ R2 , and if λ ∈ R, then λ(x1 , x2 ) = (λx1 , λx2 ) ∈ R2 .

We now want to see whether R2 is a vector space over R when addition and
scalar multiplication is defined as above. This will only be the case if every one
of the eight axioms listed in Definition 2.1.4 for a vector space hold for R2 with
the above addition and scalar multiplication.
(b) Do the axioms 1 to 4 hold? If you claim an axiom holds, you must show that
it holds (write out your check in full). If you claim that it does not hold, you
must give an example to show that it does not hold.
[Hint: The zero vector and negatives may not be quite what you expect. Get
your work for axioms 3 and 4 checked by a tutor.]

Solution: We check the first four axioms.

(1)

x ⊕ y = (x1 + y1 + 1, x2 + y2 − 2) = (y1 + x1 + 1, y2 + x2 − 2) = y ⊕ x.

(2)

(x ⊕ y) ⊕ z = (x1 + y1 + 1, x2 + y2 − 2) ⊕ (z1 , z2 )
= ((x1 + y1 + 1) + z1 + 1, (x2 + y2 − 2) + z2 − 2)
= ((x1 + (y1 + z1 + 1) + 1, x2 + (y2 + z2 − 2) − 2)
= x ⊕ (y ⊕ z).

6
 (3) For axiom 3, we need a vector, let’s call it z = (z1 , z2 ), such that for
every x = (x1 , x2 ) ∈ R2 , (x1 , x2 ) = x = x ⊕ z = (x1 + z1 + 1, x2 + z2 − 2).
This means that we must have z1 = −1, z2 = 2. Therefore axiom 3 will be
satisfied if we take as zero vector for R2 the vector z = (−1, 2).
(4) For axiom 4 we need, for every x = (x1 , x2 ) ∈ R2 , a negative, let’s say
y = (y1 , y2 ) ∈ R2 such that x ⊕ y = z, where z is the zero vector found
above. Check that y = (−x1 − 2, −x2 + 4) satisfies this equation. Clearly y
is in R2 .
This shows that the axiom 1 to 4 are satisfied.

(c) Do the axioms 5 to 8 all hold? Again, give proofs or counter-examples to support
your claims.

Solution:
(5) For λ ∈ R, x = (x1 , x2 ), y = (y1 , y2 ) ∈ R2 we have λ(x ⊕ y) = (λx1 +
λy1 + λ, λx2 + λy2 − 2λ) while λx ⊕ λy = (λx1 + λy1 + 1, λx2 + λy2 − 2) and
these are in general not equal (take λ = 2 for example), so the axiom 5 does
not hold.
(6) For λ, µ ∈ R, x ∈ R2 we have (λ + µ)x = ((λ + µ)x1 , (λ + µ)x2 ) while
λx ⊕ µx = (λx1 + µx1 + 1, λx2 + µx2 − 2), and these two are in general not
equal. (Take, for example λ = µ = 1). So axiom 6 does not hold.
(7) For λ, µ and x as above, check that (λµ)x = λ(µx), so axiom 7 holds.
(8) For x ∈ R2 1x = x, so axiom 8 holds.

(d) Is R2 a vector space with the addition and scalar multiplication as defined above?

Solution: R2 is not a vector space with the addition ⊕ and usual scalar
multiplication, since two of the axioms fail. (You can already say this as
soon as you find one axiom that fails.)