MATH 125 Fall 2021

Questions from past final exams - Solutions

This is a list of questions which have appeared on final exams for MATH 125 in the past.
They were conceived by various instructors, not necessarily those you currently have. Some
of the questions on this list will appear on the final exam on December 18, although the
numbers in the matrices and vectors could be different. There could also be questions on
the final exam this fall which are not on the list below.

You probably don’t have time to practice with all these questions. We suggest that you
give priority to the topics with which you think you need more practice. For instance, if
you are comfortable working with determinants but less when it comes to working with
eigenvalues and eigenvectors, that you should focus more on the questions about this second
topic.

Contents

1 Questions about matrix inverses 2

2 Questions about subspaces 7

3 Questions about bases 13

4 Questions about linear transformations 24

5 Questions about determinants 29

6 Questions about eigenvalues, eigenvectors and diagonalization 33

7 Multiple Choice Questions 44

8 True or false questions 46

9 Other questions 48

1
1 Questions about matrix inverses

1. Let » fi
1 5 3
A“– 0 1 ´1fl .
´2 ´10 ´7
A is invertible. Find its inverse A´1 .

Solution:
» fi
1 5 3 1 0 0
–0 1 ´1 0 1 0fl
´2 ´10 ´7 0 0 1
R3 ` 2R1
» fi
1 5 3 1 0 0
–0 1 ´1 0 1 0fl
0 0 ´1 2 0 1
´R3
» fi
1 5 3 1 0 0
–0 1 ´1 0 1 0 fl
0 0 1 ´2 0 ´1
R2 ` R3 , R1 ´ 3R3
» fi
1 5 0 7 0 3
–0 1 0 ´2 1 ´1fl
0 0 1 ´2 0 ´1
R1 ´ 5R2
» fi
1 0 0 17 ´5 8
–0 1 0 ´2 1 ´1fl
0 0 1 ´2 0 ´1
» fi
17 ´5 8
The inverse of A is the matrix ´2
– 1 ´1fl.
´2 0 ´1

2. Let A, B be two n ˆ n matrices.

a) Suppose B is not invertible. Without reference to the determinant, show that also
AB is not invertible.

Solution: This is the same as showing that if AB is invertible, then B must be invertible.

2
 AB invertible ùñ there is a matrix C such that CpABq “ I ùñ pCAqB “ I ùñ B is
invertible with inverse equal to CA.

b) Suppose A is not invertible. Without reference to the determinant, show that also
AB is not invertible.

Solution: This is the same as showing that if AB is invertible, then A must be invertible.

AB invertible ùñ there is a matrix C such that pABqC “ I ùñ ApBCq “ I ùñ A is

invertible with inverse equal to BC.

3. This is the same question as no. 37. Let A be a n ˆ n matrix such that A2 ´ 3A “ 2In
(where In is the n ˆ n identity matrix).

(a) Show that A is invertible.

`1 ˘
Solution: A2 ´ 3A “ 2In ùñ ApA ´ 3Iq “ 2In ùñ A 2
A ´ 32 I “ In : this last equality
shows that A is invertible and its inverse is 21 A ´ 32 I.

(b) Can 3 be an eigenvalue of A? Justify your answer.

Solution: Suppose that 3 is an eigenvalue of A. Then there exists a non-zero vector v such
that Av “ 3v, so pA ´ 3Iqv “ 0.

Applying both sides of ApA ´ 3Iq “ 2In to v gives ApA ´ 3Iqv “ 2In v, hence 0 “ 2v
since pA ´ 3Iqv “ 0. However, 0 “ 2v is not possible since v is supposed to be a non-zero
eigenvector.

Since a contradiction has been obtained, it follows that the initial assumption that 3 is
an eigenvalue must be rejected. Consequently, 3 is not an eigenvalue of A.

» fi
1 ´1 2 “ ‰
4. Let A “ –´2 3 ´6fl and let B “ 2 ´1 3 .
0 2 ´3

(a) The matrix A is invertible. Find its inverse, A´1 .

3
Solution:
» fi
1 ´1 2 1 0 0
–´2 3 ´6 0 1 0fl
0 2 ´3 0 0 1
R2 ` 2R1
» fi
1 ´1 2 1 0 0
–0 1 ´2 2 1 0fl
0 2 ´3 0 0 1
R3 ´ 2R2
» fi
1 ´1 2 1 0 0
–0 1 ´2 2 1 0fl
0 0 1 ´4 ´2 1
R2 ` 2R3 , R1 ´ 2R3
» fi
1 ´1 0 9 4 ´2
–0 1 0 ´6 ´3 2 fl
0 0 1 ´4 ´2 1
R1 ` R2
» fi
1 0 0 3 1 0
–0 1 0 ´6 ´3 2fl
0 0 1 ´4 ´2 1
» fi
3 1 0
The inverse of A is the matrix –´6 ´3 2fl.
´4 ´2 1

(b) Find the unique matrix X for which XA “ B.

fi»
“ 3 1 0
‰ “ ‰
Solution: XA “ B ðñ X “ BA´1 ðñ X “ 2 ´1 3 –´6 ´3 2fl “ 0 ´1 1 .
´4 ´2 1
“ ‰
Therefore, X “ 0 ´1 1 .

» fi
k 0 2
5. Let A “ –1 1 1 ´ k fl, where k is an arbitrary real number.
0 1 ´2

(a) For which k is A an invertible matrix?

4
Solution: A is invertible when detpAq ‰ 0. Let’s compute detpAq using expansion along the
first row:
ˇ ˇ ˇ ˇ
ˇ1 1 ´ k ˇ ˇ1 1 ˇ 2
detpAq “ k ˇˇ ˇ´0`2ˇ
ˇ0 1ˇ “ kp´2 ´ p1 ´ kqq ` 2p1 ´ 0q “ kpk ´ 3q ` 2 “ k ´ 3k ` 2.
ˇ
1 ´2 ˇ
?
´p´3q˘ p´3q2 ´4¨2
The roots of the polynomial k 2 ´ 3k ` 2 are 2
, that is, its roots are 1 and 2.

In conclusion, for any k ‰ 1 and k ‰ 2, the matrix A is invertible.

(b) Compute the inverse of A when k “ 0.

Solution:
» fi
0 0 2 1 0 0
–1 1 1 0 1 0fl
0 1 ´2 0 0 1
R1 Ø R2 , R2 Ø R3
» fi
1 1 1 0 1 0
–0 1 ´2 0 0 1fl
0 0 2 1 0 0
1
R3
2
» fi
1 1 1 0 1 0
–0 1 ´2 0 0 1fl
1
0 0 1 2
0 0
R2 ` 2R3 , R1 ´ R3
» fi
1 1 0 ´ 12 1 0
–0 1 0 1 0 1fl
1
0 0 1 2
0 0
R1 ´ R2
» fi
1 0 0 ´ 23 1 ´1
–0 1 0 1 0 1 fl
1
0 0 1 2
0 0
» 3 fi
´2 1 ´1
The inverse of A when k “ 0 is the matrix 1 0
– 1 fl.
1
2
0 0

» fi
7 3 10
6. This is the same question as no.2 (a) Let A “ –13 6 18fl. Find the inverse of A. Show
3 2 3
your calculations.

5
Solution:
» fi
7 3 10 1 0 0
–13 6 18 0 1 0fl
3 2 3 0 0 1
R1 ´ 2R3
» fi
1 ´1 4 1 0 ´2
–13 6 18 0 1 0 fl
3 2 3 0 0 1
R2 ´ 13R1 , R3 ´ 3R1
» fi
1 ´1 4 1 0 ´2
–0 19 ´34 ´13 1 26 fl
0 5 ´9 ´3 0 7
R2 ´ 4R3
» fi
1 ´1 4 1 0 ´2
–0 ´1 2 ´1 1 ´2fl
0 5 ´9 ´3 0 7
R3 ` 5R2
» fi
1 ´1 4 1 0 ´2
–0 ´1 2 ´1 1 ´2fl
0 0 1 ´8 5 ´3
R2 ´ 2R3 , R1 ´ 4R3
» fi
1 ´1 0 33 ´20 10
–0 ´1 0 15 ´9 4 fl
0 0 1 ´8 5 ´3
R1 ´ R2 , ´R2
» fi
1 0 0 18 ´11 6
–0 1 0 ´15 9 ´4fl
0 0 1 ´8 5 ´3

The inverse of A is thus » fi

18 ´11 6
–´15 9 ´4fl
´8 5 ´3

» fi
1
(b) The columns of A form a basis B of R3 . Let v be the vector – 2 fl. Determine the
´1
coordinate vector rvsB of v with respect to the basis B.

6
Solution: rvsB is the unique solution of Ax “ v, so
» fi » fi » fi
18 ´11 6 1 ´10
´1
rvsB “ A v “ ´15 9
– ´4fl – 2 fl “ – 7 fl .
´8 5 ´3 ´1 5

2 Questions about subspaces

7. Let v1 , v2 , v3 be some fixed but unknown vectors in Rn .

a) State the definition of a subspace of Rn .

Solution: A subspace of Rn is a subset S that contains 0 and for which the following two
properties are satisfied:

- If u P S and v P S, then u ` v P S.

- If u P S and c P R, then cu P S.

b) Let S be the subset of R3 consisting of all vectors

» fi
x
x “ y fl such that xv1 ` yv2 ` zv3 “ 0.
–
z

(So x is an element of S if and only if xv1 ` yv2 ` zv3 “ 0.)

Verify that S is a subspace of R3 .

Solution: 0 P S since 0v1 ` 0v2 ` 0v3 “ 0.

» fi » fi
x1 x2
Suppose that u P S and v P S. Write these vectors as u “ y1 and u “ y2 fl, so that
– fl –
z1 z2

x1 v1 ` y1 v2 ` z1 v3 “ 0 and x2 v1 ` y2 v2 ` z2 v3 “ 0.

Adding the last two equations shows that

px1 ` x2 qv1 ` py1 ` y2 qv2 ` pz1 ` z2 qv3 “ 0.

7
 » fi
x1 ` x2
Since u ` v “ – y1 ` y2 fl, this shows that u ` v is in S.
z1 ` z2
» fi
cx1
Now let c P R. Then x1 v1 `y1 v2 `z1 v3 “ 0 ùñ cx1 v1 `cy1 v2 `cz1 v3 “ 0 ùñ cy1 fl P S,
–
cz1
that is, cu P S.

c) Suppose the dimension of spanpv1 , v2 , v3 q is equal to 1. With S defined as in b), what

is the dimension of S? Justify your answer.

Solution: If spanpv1 , v2 , v3 q as dimension 1, then all these vectors are multiples of each
other. Suppose that v1 ‰ 0. Then we can write v2 “ c2 v1 and v3 “ c3 v1 , so the condition
xv1 ` yv2 ` zv3 “ 0 becomes px ` yc2 ` zc3 qv1 “ 0, hence x ` yc»2 `fizc3 “ 0 because v1 ‰ 0.
x
This means that S is the subspace consisting of all the vectors y fl satisfying the equation
–
z
x ` yc2 ` zc3 “ 0: therefore, S is a plane in R3 and has dimension 2.

8. (a) State what it means for a subset S of Rn to be a subspace of Rn .

Answer: A subset S of Rn is a subspace if 0 P S and if the following two conditions on S

are satisfied:

- If u P S and v P S, then u ` v P S.

- If u P S and c P R, then cu P S.

(b) Suppose that T : R2 Ñ R2 is a linear transformation, and let S be the set of all
vectors in the range of T , i.e.,
S “ tx P R2 | x “ T pvq for some v P R2 u.
Show that S is a subspace of R2 .

(Note: It may help to recall that T p0q “ 0 for any linear transformation T : R2 Ñ R2 .)

Solution: Since T p0q “ 0, the zero vector is in S.

Let u, v P S. Then there exists x1 P R2 and x2 P R2 such that T px1 q “ u and T px2 q “ v.

Since T is a linear transformation, T px1 ` x2 q “ T px1 q ` T px2 q “ u ` v, which shows

that u ` v is in S.

8
 Now let c P R. Since T is a linear transformation, T pcx1 q “ cT px1 q “ cu, so cu is in S
also.

S is thus closed under vector addition and scalar multiplication, so it is a subspace of

R2 .

9. (a) State the definition of the dimension of a subspace S of Rn .

Answer: The dimension of a subspace S is the number of vectors in a basis of S.

(b) Give an example of a subspace of R5 which has dimension 2. (You don’t need to
justify your answer.)

» fi
1
—0ffi
— ffi
—0ffi and
Solution: spanpu, vq where u and v are the linearly independent vectors u “ — ffi
–0fl
» fi 0
0
—1ffi
— ffi 5
v“— —0ffi. This subspace of R has dimension 2.
ffi
–0fl
0

(c) Find the dimension of the subspace S of R3 spanned by the vectors

» fi » fi » fi
1 3 ´1
v1 “ ´2 , v2 “ ´8
– fl – fl and v3 “ 6 fl ,
–
2 6 1

i.e., S “ spanpv1 , v2 , v3 q. Justify your answer.

» fi
1 3 ´1
Solution: Let A “ rv1 v2 v3 s “ –´2 ´8 6 fl. Then S “ colpAq, so let’s determine the
2 6 1
dimension of colpAq by transforming A into row echelon form:
» fi » fi
1 3 ´1 1 3 ´1
–´2 ´8 6 fl R2 ` 2R1 , R3 ´ 2R1 –0 ´2 4 fl .
2 6 1 0 0 3

The last matrix is in row echelon form and has 3 pivots, so the rank of A is 3 and this is
also the dimension of colpAq. In conclusion, the dimension of S is 3.

9
10. Let A be a 6 ˆ 5 matrix such that the row space of A has dimension 4.

(a) What is the nullity of A? Justify your answer.

Solution: Since the row space of A has dimension 4, a row echelon form REFpAq of A must
have 4 non-zero rows. Consequently, the rank of A is 4.

By the Rank-Nullity Theorem, rankpAq ` nullitypAq “ 5, so 4 ` nullitypAq “ 5 and thus

nullitypAq “ 1.

(b) Suppose that B and C are 6 ˆ 5 matrices such that B ´ C “ A. Show that there
exists a non-zero vector v P R5 such that Bv “ Cv.

Solution: Since nullitypAq ě 1, there exists one non-zero vector v in R5 such that Av “ 0.
Since A “ B ´ C, it follows that pB ´ Cqv “ 0, hence Bv ´ Cv “ 0 and Bv “ Cv.

11. Let W Ă R3 be the span of the two vectors

» fi » fi
1 1
v “ 2 , w “ 0 fl .
– fl –
0 ´1

» fi
1
(a) Show that b “ –4fl P W by writing it explicitly as a linear combination of v and w.
1

» fi
1 1
Solution: Let A “ rv ws “ –2 0 fl. Let’s solve Ax “ b to show that b P W . The
0» ´1 fi
1 1 1
augmented matrix of Ax “ b is –2 0 4fl, so let’s put it in reduced row echelon form:
0 ´1 1
» fi » fi » fi
1 1 1 1 1 1 1 1 1
–2 0 4fl R2 ´ 2R1 –0 ´2 2 R2 ´ 2R3 , ´R3 0 0
fl – 0 fl
0 ´1 1 0 ´1 1 0 1 ´1
» fi » fi
1 1 1 1 0 2
R2 Ø R3 0– 1 ´1 R1 ´ R2 0 1
fl – ´1fl
0 0 0 0 0 0
„ 
2
It follows from this last augmented matrix that the only solution of Ax “ b is .
´1
Therefore, b “ 2v ´ w.

10
 (b) Find a vector b P R3 which is not contained in W and justify your answer. (Note:
if you write down a vector b which satisfies the conditions but without appropriate work
shown, you will get 0 marks.)

Solution: We just need to change the vector b so that » no

fi row of zeros is produced when
1
performing the row operations above. For instance –4fl is not in W . The augmented
» fi » fi 0
1 1 1 1
matrix of Ax “ 4 is 2 0
– fl – 4fl, so let’s put it in row echelon form:
0 0 ´1 0
» fi » fi
1 1 1 1 1 1
–2 0 4fl R2 ´ 2R1 –0 ´2 2fl
0 ´1 0 0 ´1 0
» fi » fi
1 1 1 1 1 1
R2 ´ 2R3 , ´R3 –0 0 2fl R2 Ø R3 –0 1 0fl
0 1 0 0 0 2
» fi
1
The last row of this matrix is of the form r0 0 | 2s, so the equation Ax “ 4fl does not have
–
» fi 0
1
a solution and it follows that 4fl is not in W .
–
0

12. (a) Suppose S consists of all points in R2 which are on the x-axis or y-axis (or both). Is
S a subspace of R2 ? Justify your answer.

„  „ 
2 1 0
Solution: No, S is not a subspace of R . For instance, is on the x-axis and is on
„  „  „  „  „0  1
1 0 1 0 1
the y-axis, but ` is not in S since ` “ , which is neither on the x-axis
0 1 0 1 1
nor the y-axis.

(b) Let A be any 2 ˆ 2 matrix, and A2 “ A ¨ A, the square of A. Let

S “ tv P R2 | A2 v “ 3vu Ă R2 .

Is S a subspace? Justify your answer.

11
Solution: Yes, S is a subspace of R2 . First, 0 P S since A2 0 “ 0 “ 30.

If u P S and v P S, then A2 u “ 3u and A2 v “ 3v, so

A2 pu ` vq “ A2 u ` A2 v “ 3u ` 3v “ 3pu ` vq,

hence A2 pu ` vq “ 3pu ` vq, which means that u ` v P S.

If u P S and c P R, then A2 pcuq “ cA2 u “ 3cu, so cu P S.

All this shows that S is a subspace of R2 .

» fi
w
— x ffi
13. Let S be the subset of R4 consisting of all the vectors —
– y fl such that w “ 2y ` z and
ffi

z
3x ´ y “ 2z. Using the definition of a subspace, prove that S is a subspace of R4 .

Solution: S contains 0 since the equations w “ 2y ` z and 3x ´ y “ 2z are satisfied when

w “ 0, x “ 0, y “ 0 and z “ 0.
» fi » fi
w1 w2
— x1 ffi — x2 ffi
S is closed under addition: suppose that —
– y1 fl and – y2 fl are in S. Then
ffi — ffi

z1 z2

w1 “ 2y1 ` z1 and w2 “ 2y2 ` z2 .

Adding these two equations shows that

w1 ` w2 “ 2y1 ` z1 ` 2y2 ` z2 “ 2py1 ` y2 q ` z1 ` z2 . (1)

Moreover, since those two vectors are in S, it is also the case that 3x1 ´ y1 “ 2z1 an
3x2 ´ y2 “ 2z2 . Adding these two equations shows that

3x1 ´ y1 ` 3x2 ´ y2 “ 2z1 ` 2z2 , hence 3px1 ` x2 q ´ py1 ` y2 q “ 2pz1 ` z2 q. (2)

» fi » fi » fi
w1 ` w2 w1 w2
— x1 ` x2 ffi — x1 ffi — x2 ffi
It follows from (1) and (2) that the vector —– y1 ` y2 fl is in S, that is, – y1 fl ` – y2 fl is in
ffi — ffi — ffi

z1 ` z2 z1 z2
S.
» fi
w
— x ffi
S is also closed under scalar multiplication: suppose that — – y fl is in S and c P R. Then
ffi

12
 » fi
cw
— cx ffi
w “ 2y ` z and 3x ´ y “ 2z, so cw “ 2cy ` cz and 3cx ´ cy “ 2cz also. Therefore, —
– cy fl is
ffi

» fi cz
w
— x ffi
in S, that is, c —
– y fl is in S.
ffi

3 Questions about bases

14. The matrix

» fi » fi
0 2 4 ´1 ´1 0 1 2 0 ´1
A “ –0 ´2 ´4 3 ´1fl has reduced row echelon form –0 0 0 1 ´1fl .
0 2 4 1 ´3 0 0 0 0 0

a) State the definition of a basis of a subspace of Rn .

Solution: A basis of a subspace S of Rn is an ordered set of vectors tv1 , . . . , vk u such that:

- S “ spanpv1 , . . . , vk q

and

- The vectors v1 , . . . , vk are linearly independent.

b) Find a basis for ColpAq.

Solution: The pivot columns of the reduced row echelon form of A are its 2nd and 4th columns,
so a basis for colpAq is provided by columns 2 and 4 of A, that is, a basis of colpAq is:
$» fi » fi,
& 2 ´1 .
–´2fl , – 3 fl .
2 1
% -

c) Find a basis for NulpAq.

13
Solution: The reduced row echelon form shows that x P nullpAq ðñ
"
x2 ` 2x3 ´ x5 “ 0
x4 ´ x5 “ 0
x1 , x3 and x5 are free variables, so set x1 “ r, x3 “ s and x5 “ t. Then x2 “ ´2s ` t, x4 “ t
and » fi » fi » fi » fi
r 1 0 0
—´2s ` tffi —0ffi —´2ffi —1ffi
— ffi — ffi — ffi — ffi
x“— — s ffi “ r —0ffi ` s — 1 ffi ` t —0ffi .
ffi — ffi — ffi — ffi
– t fl –0fl – 0 fl –1fl
t 0 0 1
Therefore, a basis for nullpAq is
$» fi » fi » fi,
’
’ 1 0 0 / /
&—0ffi —´2ffi —1ffi
’
’— ffi — ffi — /
/
ffi.
—0ffi , — 1 ffi , —0ffi .
— ffi — ffi — ffi/
’
’
’–0fl – 0 fl –1fl/ /
’ /
0 0 1
% -

d) Find a basis for RowpAq.

Solution: A basis for rowpAq, the row space of A, is provided by the non-zero rows of
RREFpAq, so a basis is
“ ‰ “ ‰
t 0 1 2 0 ´1 , 0 0 0 1 ´1 u.

e) Is r0, 1, 2, ´1, 0s an element of RowpAq? Justify your answer.

“ ‰ “ ‰ “ ‰ “ ‰
Solution: 0 1 2 ´1 0 “ 0 1 2 0 ´1 ´ 0 0 0 1 ´1 , so 0 1 2 ´1 0 is a
vector in rowpAq since the other two vectors belong to rowpAq (because RREFpAq is obtained
from A by applying row operations, hence its rows are linear combinations of the rows of
A).

15. Consider matrices

» fi » fi
3 12 ´27 9 ´9 1 0 ´1 3 0
—1 3 ´7 3 ´2ffi —0 1 ´2 0 0ffi
— ffi — ffi
A“— — 2 8 ´18 6 ´6ffi and B “ —0
ffi — 0 0 0 1ffi
ffi .
–´1 ´2 5 ´3 1 fl –0 0 0 0 0fl
1 4 ´9 3 ´1 0 0 0 0 0

14
 Given that B is the reduced row echelon form of A:

(a) Find a basis of ColpAq, the column space of A.

Solution: The pivot columns of B are columns 1,2 and 5. It follows that a basis for colpAq
consists of the columns 1,2 and 5 of A, that is, a basis of colpAq is
$» fi » fi » fi,
’
’ 3 12 ´9 / /
&— 1 ffi — 3 ffi —´2ffi
’
’ — ffi — ffi — /
/
ffi.
— 2 ffi , — 8 ffi , —´6ffi .
— ffi — ffi — ffi/
’
’
’ –´1fl –´2fl – 1 fl/ /
’ /
1 4
% -
´1

(b) Find a basis of ColpAT q, the column space of AT .

Solution: colpAT q is equal to the transpose of the row space of A. A basis for the row space
of A is given by the three non-zero rows of B, so a basis for colpAT q is given by the transpose
of those rows: $» fi » fi » fi,
’ 1
’ 0 0 / /
&— 0 ffi — 1 ffi —0ffi
’
’ — ffi — ffi — /
/
ffi.
—´1ffi , —´2ffi , —0ffi .
— ffi — ffi — ffi/
’
’
’ – 3 fl – 0 fl –0fl/ /
’ /
0 0 1
% -

(c) Find a basis of NulpAq, the null space of A.

Solution: x P NulpAq ðñ Bx “ 0
$
& x1 ´ x3 ` 3x4 “ 0
ðñ x2 ´ 2x3 “ 0
x5 “ 0
%

x3 and x4 are free variables, so set x3 “ s and x4 “ t. Then x1 “ s ´ 3t, x2 “ 2s and

» fi » fi » fi
s ´ 3t 1 ´3
— 2s ffi —2ffi — 0 ffi
— ffi — ffi — ffi
x“— — s ffi “ s —1ffi ` t — 0 ffi .
ffi — ffi — ffi
– t fl –0fl – 1 fl
0 0 0

15
A basis for nullpAq is thus given by
$» fi » fi,
’
’ 1 ´3 / /
&—2ffi — 0 ffi
’
’— ffi — /
/
ffi.
—1ffi , — 0 ffi .
— ffi — ffi/
’
’
’–0fl – 1 fl/ /
’ /
0 0
% -

(d) What are the rank and the nullity of A?

Solution: The rank of A is 3 and the nullity of A is 2.

16. Let A be the matrix » fi

9 ´15 18 ´5 ´26
—6 ´18 12 8 3 ffi
–3 ´3 6 ´3 ´13fl .
— ffi

3 0 6 16 ´9
A row echelon form of the matrix A is:
» fi
3 ´9 6 1 0
—0 ´3 0 1 6ffi
REFpAq “ —
–0 0 0 2 1fl .
ffi

0 0 0 0 0

(a) Find a basis for the column space of A. Give only your answer, no justification needed.

Solution: The row echelon form above shows that the pivot columns of A are its first, second
and fourth, so a basis for the column space of A is
» fi » fi » fi
9 ´15 ´5
—6ffi —´18ffi — 8 ffi
— ffi , —
–3fl – ´3 fl , –´3fl .
ffi — ffi

3 0 16

(b) Find a basis for the null space of A. Show how you got your answer.

16
Solution: Let’s put A in reduced row echelon form, starting with the row echelon form above.
» fi
3 ´9 6 1 0
—0 ´3 0 1 6ffi
— ffi
–0 0 0 2 1fl
0 0 0 0 0
1
R3 , R2 ´ R3 , R1 ´ R3
2» fi
3 ´9 6 0 ´ 21
—0 ´3 0 0 11 ffi
— 2 ffi
–0 0 0 1 1 fl
2
0 0 0 0 0
R1 ´ 3R2
» fi
3 0 6 0 ´17
—0 ´3 0 0 11 ffi
— 2 ffi
–0 0 0 1 1 fl
2
0 0 0 0 0

1 1
R1 , ´ R2
» 3 3 fi
1 0 2 0 ´ 17
3
—0
— 1 0 0 ´ 11 ffi
6 ffi
–0 1 fl
0 0 1 2
0 0 0 0 0
If x is in the null space of A, then
17
x1 ` 2x3 ´ 3 5
x “ 0
11
x2 ´ 6 5
x “ 0
1
x4 ` x
2 5
“ 0
x3 and x5 are free variables, so set s “ x3 and t “ x5 .
» fi » fi » fi » 17 fi
x1 ´2x3 ` 173
x 5 ´2 3
—x2 ffi — 11 — 11 ffi
— ffi — 6 5
x ffi
ffi
— 0 ffi
— ffi — 6 ffi
—x3 ffi “ — x3
x“— ffi — ffi “ s — 1 ffi ` t — 0 ffi .
1
ffi — ffi — 1 ffi
–x4 fl – ´ 2 x5 fl – 0 fl –´ fl
2
x5 x5 0 1
A basis for nullpAq is given by $» fi » fi,
17
’
’ ´2 3 /
/
11
— 0 ffi
’
’ — ffi/
&— ffi
6
— ffi.
/
— 1 ffi , — 0 ffi .
— ffi — 1 ffi/
’
’– 0 fl –´ fl/
’
’ 2 / /
0 1
% -

17
(c) What is the rank of A? Give only your answer, no justification needed.

Answer: 3

(d) What is the nullity of A? Give only your answer, no justification needed.

Answer: 2

the same as question 23. Let `„ be the line through the origin in R2 with normal
17. This is „
1 ´1
vector n “ and direction vector d “ .
1 1

Note that B “ tn, du is a basis for R2 .

a) Let S be a subspace of Rn with basis C “ tv1 , v2 , v3 u. State the definition of the

coordinate vector rxsC of a vector x in S with respect to the basis C.

» fi
c1
Solution: The coordinate vector rxsC is the 3 ˆ 1 vector c2 fl whose entries are such that
–
c3
x “ c1 v 1 ` c2 v 2 ` c3 v 3 .

b) With respect to the basis B given above, compute re1 sB and re2 sB where
„  „ 
1 0
e1 “ and e2 “ .
0 1

„  „ 
0 2
Solution: Observe that n ` d “ “ 2e2 and n ´ d “ “ 2e1 . Therefore, e2 “ 21 n ` 12 d
2 0
and e1 “ 21 n ´ 12 d, so the coordinate vectors re1 sB and re2 sB are given by:
„ 1  „1
re1 sB “ 2 , re2 sB “ 21 .
´ 21 2

c) Let T : R2 Ñ R2 be the reflection across the line `, so T pnq “ ´n and T pdq “ d. Find
the standard matrix rT s of T , that is, the matrix rT s such that for x in R2 , T pxq “ rT sx.

18
Solution: The columns of the matrix rT s are T pe1 q and T pe2 q.
„ 
1 1 1 1 1 1 0
e1 “ n ´ d ùñ T pe1 q “ T pnq ´ T pdq “ ´ n ´ d “ ´e2 “
2 2 2 2 2 2 ´1

and „ 
1 1 1 1 1 1 ´1
e2 “ n ` d ùñ T pe2 q “ T pnq ` T pdq “ ´ n ` d “ ´e1 “ .
2 2 2 2 2 2 0

„ 
0 ´1
In conclusion, rT s “ .
´1 0

18. This is the same question as no. 24. The set

"„  „ *
3 6
B“ , 1
0 2

is an ordered basis of R2 .
„ 
0
(a) Find re2 sB , the coordinate vector of e2 “ with respect to the ordered basis B.
1

Solution:
„ The
 coordinate vector of e2 is the unique solution of the equation Ax “ e2 where
3 6
A“ . Let’s put the augmented matrix rA | e2 s in reduced row echelon form:
0 21
„  „  „ 
3 6 0 1 1 2 0 1 0 ´4
rA | e2 s “ R , 2R2 R1 ´ 2R2
0 12 1 3 1 0 1 2 0 1 2
„ 
´4
The unique solution is thus and this is the coordinate vector re2 sB .
2

(b) Find the standard matrix of the unique linear transformation T : R2 Ñ R3 for which
» fi » fi
ˆ„ ˙ 3 ˆ„ ˙ 1
3 6
T “ ´6
– fl and T 1 “ 1 fl .
–
0 2
´3 ´2

Solution: The columns of the standard matrix are T pe1 q and T pe2 q, so let’s determine these
vectors.
» fi » fi
ˆ„ ˙ ˆ „ ˙ ˆ„ ˙ 3 1
1 1 3 1 3 1 – fl – fl
T pe1 q “ T “T “ T “ ´6 “ ´2 .
0 3 0 3 0 3
´3 ´1

19
 „  „ 
3 6
By part (a), e2 “ ´4 ` 2 1 , so, since T is a linear transformation,
0 2
» fi » fi » fi
ˆ„ ˙ ˆ„ ˙ 3 1 ´10
3 6
T pe2 q “ ´4T ` 2T 1 “ ´4 –´6fl ` 2 – 1 fl “ – 26 fl .
0 2 ´3 ´2 8

» fi
1 ´10
In conclusion, the standard matrix of T is –´2 26 fl.
´1 8

19. Determine all values of k such that the vectors

» fi » fi » fi
k k 0
– 0 fl , –k fl , –k fl
k2 2 k

form a basis of R3 .
»
fi
k k 0
Solution: Those 3 vectors form a basis of R3 exactly when the matrix – 0 k k fl is invert-
k2 2 k
ible, which happens when its determinant is ‰ 0.

Expanding the determinant along the first row, we obtain:

ˇ ˇ
ˇk k 0ˇˇ ˇ ˇ ˇ ˇ
ˇ ˇk k ˇ ˇ 0 kˇ
ˇ0 k k ˇˇ “ k ˇˇ ˇ´kˇ 2 ˇ
ˇ 2
ˇk 2 kˇ ˇk k ˇ
2 k ˇ
“ kpk 2 ´ 2kq ´ kp0 ´ k 3 q “ k 4 ` k 3 ´ 2k 2 “ k 2 pk 2 ` k ´ 2q.
?
2 2 ´1˘ 12 ´4¨p´2q
The determinant is ‰ 0 if k ‰ 0 and k ` k ´ 2 ‰ 0. The roots of k ` k ´ 2 are 2
,
that is, 1 and ´2.

In conclusion, those vectors form a basis of R3 if k ‰ 0, k ‰ 1 and k ‰ ´2.

20. Let A be the matrix » fi

2 ´4 0 2 1
–´1 2 1 2 3fl .
1 ´2 1 4 4

(a) Put A in row echelon form.

20
Solution: » fi » fi
2 ´4 0 2 1 1 ´2 1 4 4
–´1 2 1 2 3fl R1 Ø R3 ´1 2 1 2
– 3fl
1 ´2 1 4 4 2 ´4 0 2 1
» fi » fi
1 ´2 1 4 4 1 ´2 1 4 4
R2 ` R1 –
0 0 2 6 7 fl R3 ` R2 0 0
– 2 6 7fl
R3 ´ 2R1
0 0 ´2 ´6 ´7 0 0 0 0 0

(b) Compute a basis of the row space of A.

“ ‰ “ ‰
Solution: basis for the row space of A is t 1 ´2 1 4 4 , 0 0 2 6 7 u.

(c) Compute a basis for the column space of A.

Solution: The pivot columns in the row echelon form above are columns 1 and 3. Therefore,
columns
$» fi 1 »andfi,3 of A form a basis for the column space of A, that is, a basis for colpAq is
& 2 0 .
–´1fl , –1fl .
1 1
% -

(d) Compute a basis for the null space of A.

Solution: To obtain a basis for the null space of A, let’s transform A into reduced row echelon
form:
» fi » fi » fi
1 ´2 1 4 4 1 ´2 1 4 4 1 ´2 0 1 21
–0 0 2 6 7fl 1 R2 –0 0 1 3 7 fl R1 ´ R2 –0 0 1 3 7 fl .
2 2
2
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

This last matrix shows that if x P nullpAq, then

" 1
x1 ´ 2x2 ` x4 ` x
2 5
“ 0
7
x3 ` 3x4 ` x
2 5
“ 0

x2 , x4 and x5 are free variables, so set x2 “ r, x4 “ s and x5 “ t, so x1 “ 2r ´ s ´ 12 t and

x3 “ ´3s ´ 72 t. It follows that
» fi » fi » fi » 1 fi
2r ´ s ´ 12 t 2 ´1 ´2
—
— r ffi
ffi
—1ffi
— ffi
— 0 ffi
— ffi
— 0 ffi
— ffi
7 ffi —0ffi ` s —´3ffi ` t —´ 7 ffi .
x“— — ´3s ´ t
2 ffi
“ r — ffi — ffi — 2 ffi
– s fl –0fl – 1 fl – 0 fl
t 0 0 1

21
A basis for nullpAq is $» fi » fi » 1 fi,
’
’ 2 ´1 ´2 / /
&—1ffi — 0 ffi — 0 ffi/
’
’— /
ffi — ffi — 7 ffi.
—0ffi , —´3ffi , —´ ffi .
— ffi — ffi — 2 ffi/
’
’
’–0fl – 1 fl – 0 fl//
’ /
0 0 1
% -

» fi
1 2 ´1
21. Let A “ –0 4 4 fl.
2 7 1

(a) Compute RankpAq, the rank of the matrix A.

Solution: Let’s put A in row echelon form:

» fi » fi » fi » fi
1 2 ´1 1 2 ´1 1 2 ´1 1 2 ´1
1 1
A“ 0 – 4 4 fl R3 ´ 2R1 –0 4 4 fl R2 , R3 –0 1 1 fl R3 ´ R2 –0 1 1 fl .
4 3
2 7 1 0 3 3 0 1 1 0 0 0

The last matrix is in row echelon form and has two non-zero rows. Therefore, the rank of A
is 2.

(b) Determine a basis for ColpAq, the column space of A.

Solution: The pivot columns of the row echelon form above are the first and second columns.
Therefore,
$ » fi » fi a,
basis for ColpAq consists of the first two columns of A, that is, a basis is
& 1 2 .
–0fl , –4fl .
2 7
% -

(c) Determine a basis for NulpAq, the null space of A.

Solution: Let’s put the last matrix above in reduced row echelon form:
» fi » fi
1 2 ´1 1 0 ´3
–0 1 1 fl R1 ´ 2R2 –0 1 1 fl .
0 0 0 0 0 0

This last matrix shows that if x P nullpAq, that is, if Ax “ 0, then

"
x1 ´ 3x3 “ 0
x2 ` x3 “ 0.

22
 » fi » fi
3t 3
x3 is a free variable. Set x3 “ t, so x1 “ 3t, x2 “ ´t and x “ ´t “ t ´1fl. A basis for
– fl –
$» fi, t 1
& 3 .
nullpAq is –´1fl .
1
% -

» fi
7 3 10
22. This is the same question as no.6 (a) Let A “ –13 6 18fl. Find the inverse of A.
3 2 3
Show your calculations.

Solution:
» fi
7 3 10 1 0 0
–13 6 18 0 1 0fl
3 2 3 0 0 1
R1 ´ 2R3
» fi
1 ´1 4 1 0 ´2
–13 6 18 0 1 0 fl
3 2 3 0 0 1
R2 ´ 13R1 , R3 ´ 3R1
» fi
1 ´1 4 1 0 ´2
–0 19 ´34 ´13 1 26 fl
0 5 ´9 ´3 0 7
R2 ´ 4R3
» fi
1 ´1 4 1 0 ´2
–0 ´1 2 ´1 1 ´2fl
0 5 ´9 ´3 0 7
R3 ` 5R2
» fi
1 ´1 4 1 0 ´2
–0 ´1 2 ´1 1 ´2fl
0 0 1 ´8 5 ´3
R2 ´ 2R3 , R1 ´ 4R3
» fi
1 ´1 0 33 ´20 10
–0 ´1 0 15 ´9 4 fl
0 0 1 ´8 5 ´3

23
 R1 ´ R2 , ´R2
» fi
1 0 0 18 ´11 6
–0 1 0 ´15 9 ´4fl
0 0 1 ´8 5 ´3

The inverse of A is thus » fi

18 ´11 6
–´15 9 ´4fl
´8 5 ´3

»
fi
1
(b) The columns of A form a basis B of R3 . Let v be the vector – 2 fl. Determine the
´1
coordinate vector rvsB of v with respect to the basis B.

Solution: rvsB is the unique solution of Ax “ v, so

» fi » fi » fi
18 ´11 6 1 ´10
rvsB “ A´1 v “ –´15 9 ´4fl – 2 fl “ – 7 fl .
´8 5 ´3 ´1 5

4 Questions about linear transformations

2
23. (This is„the
 same as question 17.) Let„` bethe line through the origin in R with normal
1 ´1
vector n “ and direction vector d “ .
1 1

Note that B “ tn, du is a basis for R2 .

a) Let S be a subspace of Rn with basis C “ tv1 , v2 , v3 u. State the definition of the

coordinate vector rxsC of a vector x in S with respect to the basis C.

» fi
c1
Solution: The coordinate vector rxsC is the 3 ˆ 1 vector c2 fl whose entries are such that
–
c3
x “ c1 v 1 ` c2 v 2 ` c3 v 3 .

24
 b) With respect to the basis B given above, compute re1 sB and re2 sB where
„  „ 
1 0
e1 “ and e2 “ .
0 1

„  „ 
0 2
Solution: Observe that n ` d “ “ 2e2 and n ´ d “ “ 2e1 . Therefore, e2 “ 21 n ` 12 d
2 0
and e1 “ 21 n ´ 12 d, so the coordinate vectors re1 sB and re2 sB are given by:
„ 1  „1
re1 sB “ 2 , re2 sB “ 21 .
´ 21 2

c) Let T : R2 Ñ R2 be the reflection across the line `, so T pnq “ ´n and T pdq “ d. Find
the standard matrix rT s of T , that is, the matrix rT s such that for x in R2 , T pxq “ rT sx.

Solution: The columns of the matrix rT s are T pe1 q and T pe2 q.

„ 
1 1 1 1 1 1 0
e1 “ n ´ d ùñ T pe1 q “ T pnq ´ T pdq “ ´ n ´ d “ ´e2 “
2 2 2 2 2 2 ´1
and „ 
1 1 1 1 1 1 ´1
e2 “ n ` d ùñ T pe2 q “ T pnq ` T pdq “ ´ n ` d “ ´e1 “ .
2 2 2 2 2 2 0
„ 
0 ´1
In conclusion, rT s “ .
´1 0

24. This is the same question as no. 18 The set

"„  „ *
3 6
B“ , 1
0 2

is an ordered basis of R2 .
„ 
0
(a) Find re2 sB , the coordinate vector of e2 “ with respect to the ordered basis B.
1

Solution:
„ The
 coordinate vector of e2 is the unique solution of the equation Ax “ e2 where
3 6
A“ . Let’s put the augmented matrix rA | e2 s in reduced row echelon form:
0 21
„  „  „ 
3 6 0 1 1 2 0 1 0 ´4
rA | e2 s “ R , 2R2 R1 ´ 2R2
0 12 1 3 1 0 1 2 0 1 2
„ 
´4
The unique solution is thus and this is the coordinate vector re2 sB .
2

25
 (b) Find the standard matrix of the unique linear transformation T : R2 Ñ R3 for which
» fi » fi
ˆ„ ˙ 3 ˆ„ ˙ 1
3 6
T “ ´6
– fl and T 1 “ 1 fl .
–
0 2
´3 ´2

Solution: The columns of the standard matrix are T pe1 q and T pe2 q, so let’s determine these
vectors.
» fi » fi
ˆ„ ˙ ˆ „ ˙ ˆ„ ˙ 3 1
1 1 3 1 3 1
T pe1 q “ T “T “ T “ –´6fl “ –´2fl .
0 3 0 3 0 3
´3 ´1

„  „ 
3 6
By part (a), e2 “ ´4 ` 2 1 , so, since T is a linear transformation,
0 2
» fi » fi » fi
ˆ„ ˙ ˆ„ ˙ 3 1 ´10
3 6
T pe2 q “ ´4T ` 2T 1 “ ´4 –´6fl ` 2 – 1 fl “ – 26 fl .
0 2 ´3 ´2 8

» fi
1 ´10
In conclusion, the standard matrix of T is –´2 26 fl.
´1 8

25. This question has two parts.

2 2
(a) Find„ the
 standard
„  matrix
„  for„ the linear transformation T : R Ñ R that sends
2 4 0 2
the vector to and to .
0 8 1 4

ˆ„ ˙ „ 
2 4
Solution: T “ , so
0 8
ˆ„ ˙ ˆ „ ˙ ˆ„ ˙ „  „ 
1 1 2 1 2 1 4 2
T “T “ T “ “ .
0 2 0 2 0 2 8 4

ˆ„ ˙ „  „ 
0 2 2 2
Since it is already known that T “ , the standard matrix of T is .
1 4 4 4

(b) Is T invertible? If so, find the standard matrix of its inverse.

26
Solution:
„ No,T is not invertible because its standard matrix is not invertible: the determi-
2 2
nant of is 2 ¨ 4 ´ 2 ¨ 4, that is, the determinant is 0, so that matrix is not invertible.
4 4

» fi
ˆ„ ˙ 2x1 ´ x2
x1
26. Let T : R2 Ñ R3 , T “ – x1 ` x2 fl.
x2
´x2

(a) Compute the standard matrix of T , i.e. the matrix A such that T pvq “ Av for all v
in R2 .
» fi
2 ´1
Solution: The standard matrix of T is –1 1 fl. To obtain the first column, set x1 “ 1 and
0 ´1
x2 “ 0; to obtain the second column, set x1 “ 0 and x2 “ 1.

» fi
´4
2
(b) Determine all v in R such that T pvq “ 7 fl.
–
´6

» fi » fi
„  ´4 ´4
x1
Solution: If v “ and T pvq “ 7 , then Av “ 7 fl. The augmented matrix of this
– fl –
x2
» fi ´6 ´6
2 ´1 ´4
system is –1 1 7 fl. Let’s put it in reduced row echelon form:
0 ´1 ´6
» fi » fi » fi
2 ´1 ´4 1 1 7 1 1 7
–1 1 7 fl R1 Ø R2 –2 ´1 ´4fl R2 ´ 2R1 –0 ´3 ´18fl
0 ´1 ´6 0 ´1 ´6 0 ´1 ´6
» fi » fi
1 1 7 1 0 1
1 R3 ` R2 –
´ R2 –0 1 6 fl 0 1 6fl
3 R1 ´ R2
0 ´1 ´6 0 0 0
» fi
´4 „ 
1
The unique solution of Av “ 7 is thus
– fl .
6
´6

(c) Determine a vector which is not in the image of T . Justify your answer!

27
Solution: A vector b which
» fiis not in the image of T is a vector for which
» there isfino x»such
fi
x 2x1 ´ x2 x
that T pxq “ b. If b “ –y fl, for instance, then T pxq “ b implies that – x1 ` x2 fl “ –y fl,
0 ´x2 0
so x2 “ 0. In order to have a solution, we then need to have 2x1 “ x and x1 “ y, so if we
set x “ 1 and y “ 1, then there
» fi is no common solution to 2x1 “ 1 and x1 “ 1, so there is no
1
vector x such that T pxq “ –1fl.
0

27. Let T : R2 Ñ R2 be the linear transformation given by the reflection with respect to the
line y “ ´2x. Determine the standard matrix rT s of T .

Solution: The„ equation of that line can be written as 2x ` y “ 0, so a normal vector n to

2 ? ?
that line is . Note that k n k“ 22 ` 12 “ 5.
1

The formula for the reflection T is provided by:

ˆ ˙
n¨v
T pvq “ v ´ 2 n.
k n k2
„ 
1
To determine the standard matrix of T , we have to compute T pe1 q and T pe2 q where e1 “
„  0
0
and e2 “ .
1
„  „ 
1 2
ˆ ˙ „  ¨ „  „  „  „ 3
e1 ¨ n 1 0 1 2 1 2¨2 2 ´5
T pe1 q “ e1 ´ 2 n“ ´2 “ ´ “
knk 2 0 5 1 0 5 1 ´ 45

and
„  „ 
0 2
ˆ ˙ „  ¨ „  „  „  „ 4
e2 ¨ n 0 1 1 2 0 2¨1 2 ´
T pe2 q “ e2 ´ 2 n“ ´2 “ ´ “ 35 .
knk 2 1 5 1 1 5 1 5

The columns of the stardard matrix rT s of T are T pe1 q and T pe2 q. Therefore,
„ 3 
´ 5 ´ 45
rT s “ .
´ 54 35

28
28. Fix a vector w in Rn . Consider the function T : Rn Ñ Rn given by
T pvq “ pv ¨ wqw
for all v P Rn . Using the second definition of a linear transformation in the notes, prove that
T is a linear transformation.

Solution: Let u, v P R2 . Then

` ˘
T pu ` vq “ pu ` vq ¨ w w
“ pu ¨ w ` v ¨ wqw (by distributivity)
“ pu ¨ wqw ` pv ¨ wqw (by distributivity)
“ T puq ` T pvq. (3)

Let c P R, v P R2 . Then
` ˘
T pcvq “ pcvq ¨ w w
“ cpv ¨ wqw
“ cT pvq, (4)
where the second equality follows from a property of the dot product.

(3) and (4) show that T is a linear transformation (according to Definition 2).

5 Questions about determinants

29. Compute the determinant of the matrix

» fi
0 1 0 0
—2 ´1 1 2ffi
A“— –0 7
ffi .
1 0fl
1 9 ´10 2
Is A invertible? Justify your answer.

2 1 2
Solution: Expanding along the first row gives detpAq “ ´ 0 1 0 , and the latter deter-
1 ´10 2
minant can be expanded along the second row:
2 2
detpAq “ ´ “ ´p4 ´ 2q “ ´2.
1 2
Since detpAq ‰ 0, the matrix A is invertible.

29
30. Evaluate the determinant of the matrix
» fi
0 ´2 1 0
—´3 4 ´6 1 ffi
A“— –0
ffi .
1 3 ´5fl
1 ´1 2 0

(You may use whichever method(s) you prefer).

Solution: There are two zeros in the first row, so let’s just expand the determinant along the
first row:
» fi
0 ´2 1 0 » fi » fi
—´3 4 ´6 1 ffi ´3 ´6 1 ´3 4 1
det — ffi “ ´ p´2q det – 0 3 ´5fl ` det – 0 1 ´5fl
–0 1 3 ´5fl
1 2 0 1 ´1 0
1 ´1 2 0

ˆ „  „ ˙
3 ´5 ´6 1
“ 2 ¨ ´3 det ` det
2 0 3 ´5
ˆ „  „ ˙
1 ´5 4 1
` p´3q det ` det
´1 0 1 ´5
using expansion along the first column;
` ˘ ` ˘
“ 2 ¨ p´3q ¨ 10 ` p30 ´ 3q ` p´3q ¨ p0 ´ 5q ´ 20 ´ 1q
“ 2 ¨ p´3q ´ 6
“ ´ 12

Another way to proceed would have been to put A in row echelon form:
» fi » fi » fi
0 ´2 1 0 1 ´1 2 0 1 ´1 2 0
—´3 4 ´6 1 ffi
ffi R1 Ø R4 —´3 4 ´6 1 ffi R2 ` 3R1 —0 1 0 1 ffi
— ffi — ffi
—
–0 1 3 ´5fl –0 1 3 ´5fl –0 1 3 ´5fl
1 ´1 2 0 0 ´2 1 0 0 ´2 1 0
» fi » fi » fi
1 ´1 2 0 1 ´1 2 0 1 ´1 2 0
R3 ´ R2 — —0 1 0 1 ffi R3 Ø R4 —0 1 0 1 ffi R4 ´ 3R3 —0 1 0 1 ffi .
ffi — ffi — ffi
R4 ` 2R2 –0 0 3 ´6fl –0 0 1 2 fl –0 0 1 2 fl
0 0 1 2 0 0 3 ´6 0 0 0 ´12
The only row operations above that change the determinant are R1 Ø R4 and R3 Ø R4 and
their effect is just to change the sign, so applying both leaves the determinant unchanged.
The last matrix is a triangular matrix, so its determinant is the product of the entries on
its diagonal, that is, its determinant is ´12. It follows that the determinant of A is also
´12.

30
31. Assume that the determinant of
» fi
a b c
–d e f fl “ 4.
g h i

Compute the determinant of

» fi
´1 a b c
— 0 a ` 2g b ` 2h c ` 2i ffi
– 0 3d ` 2g 3e ` 2h 3f ` 2ifl .
— ffi

0 g h i

Solution: ˇ Expanding along the firstˇ column shows that the determinant »to be computed
fi
ˇ a ` 2g b ` 2h c ` 2i ˇ a b c
ˇ ˇ
equals ´ ˇˇ3d ` 2g 3e ` 2h 3f ` 2iˇˇ . Let’s see how to obtain the matrix –d e f fl from
ˇ g h i ˇ fi g h i
»
a ` 2g b ` 2h c ` 2i
the matrix 3d ` 2g 3e ` 2h 3f ` 2ifl:
–
g h i
» fi » fi » fi
a ` 2g b ` 2h c ` 2i a b c a b c
–3d ` 2g 3e ` 2h 3f ` 2ifl R1 ´ 2R3 –3d 3e 3f fl 1 R3 –d e f fl .
R2 ´ 2R3 3
g h i g h i g h i

The row operations R1 ´ 2R3 and R2 ´ 2R3 do not change the determinant and 31 R3 divides
it by 3, so ˇ ˇ ˇ ˇ
ˇ a ` 2g b ` 2h c ` 2i ˇ ˇa b c ˇ
1ˇ ˇ ˇ ˇ ˇ
3d ` 2g 3e ` 2h 3f ` 2i ˇ “ ˇd e f ˇ “ 4
3 ˇˇ ˇ ˇ ˇ
g h i ˇ ˇg h i ˇ
ˇ ˇ
ˇ a ` 2g b ` 2h c ` 2i ˇ
ˇ ˇ
and thus ˇˇ3d ` 2g 3e ` 2h 3f ` 2iˇˇ “ 12. The determinant we had to compute equals ´12
ˇ g h i ˇ
(because of the ´1 in the upper-left corner).

» fi
a b c
32. Suppose that the determinant of the matrix A “ –d e f fl is equal to 6. Compute the
g h i
determinants of the following matrices:
» fi
a b c
(a) – ´2d ´2e ´2f fl.
a`g b`h c`i

31
Solution: » fi » fi
a b c a b c
– ´2d ´2e ´2f fl ´ 1 R2 , R3 ´ R1 –d e f fl .
2
a`g b`h c`i g h i
The row operation R3 ´ R1 does not change the determinant, so
ˇ ˇ
ˇ a b c ˇˇ
1 ˇ
detpAq “ ´ ˇˇ ´2d ´2e ´2f ˇˇ
2ˇ
a ` g b ` h c ` iˇ
and it follows that this last determinant equals ´12 since detpAq “ 6.
» fi
a ´ 3d b ´ 3e c ´ 3f
(b) – g ´ a h ´ b i ´ c fl.
d e f

Solution: » fi » fi
a ´ 3d b ´ 3e c ´ 3f a b c
– g ´ a h ´ b i ´ c fl R1 ` 3R3 –g ´ a h ´ b i ´ cfl
d e f d e f
» fi » fi
a b c a b c
R2 ` R1 –g h i fl R2 Ø R3 –d e f fl .
d e f g h i
The row operations R1 ` 3R3 and R2 ` R1 do not change the determinant and R2 Ø R3 only
changes the sign, so ˇ ˇ
ˇa ´ 3d b ´ 3e c ´ 3f ˇ
ˇ ˇ
ˇ g ´ a h ´ b i ´ c ˇ “ ´ detpAq “ ´6.
ˇ ˇ
ˇ d e f ˇ

» fi
a`b d`e g`h
(c) – b e h fl.
c f i

Solution: » fi » fi
a`b d`e g`h a d g
– b e h fl R1 ´ R2 – b e hfl “ AT .
c f i c f i
The row operation R1 ´ R2 does not change the determinant and detpAT q “ detpAq “ 6, so
ˇ ˇ
ˇa ` b d ` e g ` hˇ
ˇ ˇ
ˇ b e h ˇˇ “ 6.
ˇ
ˇ c f i ˇ

32
6 Questions about eigenvalues, eigenvectors and diag-
onalization

33. Let » fi
0 2 2
A “ –2 0 2fl .
2 2 0

a) Find the characteristic polynomial of A.

Solution: Using expansion along the first row, we obtain:

´λ 2 2
´λ 2 2 2 2 ´λ
detpA ´ λIq “ 2 ´λ 2 “ ´λ ´2 `2
2 ´λ 2 ´λ 2 2
2 2 ´λ
“ ´ λpλ2 ´ 4q ´ 2p´2λ ´ 4q ` 2p4 ´ p´2λqq “ ´λ3 ` 4λ ` 4λ ` 8 ` 8 ` 4λ
“ ´ λ3 ` 12λ ` 16

b) For each eigenvalue of A find a basis for the corresponding eigenspace. (Hint: ´2, 4
are eigenvalues.)

Solution: E´2 “ nullpA ` 2Iq, so let’s put A ` 2I in reduced row echelon form:
» fi » fi » fi
2 2 2 2 2 2 1 1 1
1
A ` 2I “ –2 2 2fl R2 ´ R1 , R3 ´ R1 –0 0 0fl R1 –0 0 0fl .
2
2 2 2 0 0 0 0 0 0

Therefore x P nullpA ` 2Iq ðñ x1 ` x2 ` x3 “ 0. x2 and x3 are free variables, so set x2 “ s

and x3 “ t. Then x1 “ ´s ´ t and
» fi » fi » fi
´s ´ t ´1 ´1
x“ – s fl “ s 1 ` t 0 fl .
– fl –
t 0 1

A basis for nullpA ` 2Iq, and thus for the eigenspace E´2 , is
$» fi » fi,
& ´1 ´1 .
– 1 fl , – 0 fl .
0 1
% -

33
 As for the other eigenspace, E4 , it equals nullpA ´ 4Iq, so let’s put A ´ 4I in reduced row
echelon form:
» fi » fi » fi
´4 2 2 2 ´4 2 2 ´4 2
A ´ 4I “ – 2 ´4 2 fl R1 Ø R2 –´4 2 2 fl R2 ` 2R1 , R3 ´ R1 –0 ´6 6 fl
2 2 ´4 2 2 ´4 0 6 ´6
» fi » fi » fi
2 ´4 2 1 ´2 1 1 0 ´1
1 1
R3 ` R2 –0 ´6 6fl R1 , ´ R2 –0 1 ´1fl R1 ` 2R2 –0 1 ´1fl .
2 6
0 0 0 0 0 0 0 0 0
Therefore x P nullpA ´ 4Iq ðñ
"
x1 ´ x3 “ 0
x2 ´ x3 “ 0
x3 is a free variables, so set x3 “ t. Then x1 “ t, x2 “ t and
» fi » fi
t 1
x “ –tfl “ t –1fl .
t 1
» fi
1
A basis for nullpA ´ 4Iq, and thus for the eigenspace E4 , consists of the vector –1fl.
1

» fi
´1 ´12 ´5 2
—´3 8 5 ´2ffi
34. Let A be the matrix —– 6 ´24 ´14 4 fl. The characteristic polynomial of this
ffi

´3 ´3 0 ´1
2 2
matrix is pλ ´ 1qpλ ` 8λ ` 16q.

(a) ´1 and 1 are eigenvalues of A. What are all the eigenvalues of A?

Solution: The other eigenvalues of A are the roots of λ2 ` 8λ ` 16.

λ2 ` 8λ ` 16 “ pλ ` 4q2 , so ´4 is the only other eigenvalue.

»fi
1
—´1ffi
(b) The eigenspace E´1 is equal to the span of —
– 2 fl and the eigenspace E1 is equal to the
ffi

» fi ´1
1
—´1ffi
span of —– 2 fl. Find a basis for all the eigenspaces of A other than these two. Show your
ffi

0
calculations.

34
Solution:
» fi » fi
´1 ` 4 ´12 ´5 2 3 ´12 ´5 2
— ´3 8`4 5 ffi “ —´3 12
´2 ffi — 5 ´2ffi
A ´ p´4qI “ A ` 4I “ —
– 6
ffi .
´24 ´14 ` 4 4 fl – 6 ´24 ´10 4 fl
´3 ´3 0 ´1 ` 4 ´3 ´3 0 3
Let’s put this matrix in reduced row echelon form.
» fi
3 ´12 ´5 2
—´3 12 5 ´2ffi
— ffi
– 6 ´24 ´10 4 fl
´3 ´3 0 3
R2 ` R1 , R3 ´ 2R1 , R4 ` R1
» fi
3 ´12 ´5 2
—0 0 0 0ffi
— ffi
–0 0 0 0fl
0 ´15 ´5 5
1
R4 , R2 Ø R4
»5 fi
3 ´12 ´5 2
—0 ´3 ´1 1ffi
— ffi
–0 0 0 0fl
0 0 0 0
R1 ´ 4R2
» fi
3 0 ´1 ´2
—0 ´3 ´1 1 ffi
— ffi
–0 0 0 0 fl
0 0 0 0

1 1
R1 , ´ R2
» 3 3 fi
1 0 ´ 13 ´ 23
—0 1 1 ´ 1 ffi
— 3 3 ffi
–0 0 0 0 fl
0 0 0 0
This shows that x is in the eigenspace E´4 exactly when
1 2
x1 ´ x
3 3
´ x
3 4
“ 0
1 1
x2 ` x
3 3
´ x
3 4
“ 0
x3 and x4 are free variables, so set x3 “ s and x4 “ t. Then x P E´4 exactly when
» fi » 1 fi » 1 fi » 2 fi
x1 3
s ` 23 t 3 3
—x2 ffi —´ 1 s ` 1 tffi —´ 1 ffi — 1 ffi
x“— ffi — 3 3 ffi “ s — 3 ffi ` t — 3 ffi .
–x3 fl “ – s fl – 1 fl – 0 fl
x4 t 0 1

35
A basis for E´4 is thus provided by
$» 1 fi » 2 fi,
&— 31 ffi — 31 ffi/
’
’ /
.
—´ 3 ffi , — 3 ffi
’– 1 fl – 0 fl/
’ /
0 1
% -

or, after multiplying the two vectors by 3, by

$» fi » fi,
’
’ 1 2 /
&— ffi — ffi/
— ffi , —1ffi .
.
´1
’ – 3 fl –0fl/
’ /
0 3
% -

(c) Diagonalize A by finding a diagonal matrix D and an invertible matrix P such that
A “ P DP ´1 . (You don’t have to compute P ´1 .) Give only your answer, no justification
needed.

Solution: » fi » fi
´4 0 0 0 1 2 1 1
— 0 ´4 0 0ffi —´1 1 ´1 ´1ffi
D“—
–0
ffi and P “ — ffi .
0 ´1 0fl –3 0 2 2 fl
0 0 0 1 0 3 ´1 0

35. Suppose A is a 3 ˆ 3 matrix with the following properties: It has eigenvalues 1 and 0.
The eigenspace for 0 is the line spanpnq, where
» fi
1
n “ ´1fl .
–
1

The eigenspace for 1 is the plane P through the origin with normal vector n. That is P is
the plane with normal equation n ¨ x “ 0.

a) State the definition of the eigenspace associated to an eigenvalue of a matrix B.

Solution: The eigenspace Eλ associated to the eigenvalue λ is the subspace spanned by all
the eigenvectors of B of eigenvalue λ.

b) For the two eigenvalues 0, 1, state their geometric multiplicities. Justify your answer.

36
Solution: The geometric multiplicity of an eigenvalue is the dimension of its eigenspace.
The eigenspace for the eigenvalue 0 is a line, so it has dimension one and thus the geometric
multiplicity of 0 is one. The eigenspace for the eigenvalue 1 is a two-dimensional plane, so
the geometric multiplicity of 1 is two.

c) A is similar to » fi
1 0 0
D “ –0 0 0fl .
0 0 1
Find an invertible matrix P such that P ´1 AP “ D.

Solution: The columns of P must» befilinearly independent eigenvectors of A. An eigenvector

1
for E0 is already given, namely ´1fl. Eigenvectors in E1 are vectors orthogonal to n. Two
–
» fi » fi 1
1 0
such vectors are 1 and 1fl and these two are linearly independent. It follows that these
– fl –
0 1
can be taken to be columns 1 and 3 of P .
» fi
1 1 0
Therefore, P can be the matrix –1 ´1 1fl.
0 1 1

d) Find the determinant of A. Justify your answer.

Solution: Since A “ P DP ´1 , detpAq “ detpP DP ´1 q “ detpP q detpDq detpP q´1 “ detpDq “

0.

e) True or false: An “ A for all n ą 0. Justify your answer.

» n fi
1 0 0
Solution: True. An “ P Dn P ´1 “ P DP ´1 “ A because Dn “ – 0 0 0 fl “ D.
0 0 1n

» fi
3 ´3 2
36. Let A “ –2 1 0 fl.
1 2 ´4

(a) Compute the characteristic polynomial ppλq of A. (You do NOT have to factor it out.)

Note that the determinant of A is equal to pp0q, the constant term of ppλq.

37
Solution:

ppλq “ detpA ´ λIq

ˇ ˇ
ˇ3 ´ λ ´3 2 ˇ
ˇ ˇ
“ ˇˇ 2 1´λ 0 ˇˇ
ˇ 1 2 ´4 ´ λˇ
ˇ ˇ ˇ ˇ
ˇ´3 2 ˇ ˇ3 ´ λ 2 ˇ
“ ´ 2 ˇˇ ˇ ` p1 ´ λq ˇ ˇ
2 ´4 ´ λˇ ˇ 1 ´4 ´ λˇ
(using expansion along the second row)
“ ´ 2p´3 ¨ p´4 ´ λq ´ 2 ¨ 2q ` p1 ´ λqpp3 ´ λqp´4 ´ λq ´ 2q
“ ´ 2p12 ` 3λ ´ 4q ` p1 ´ λqpλ2 ` λ ´ 12 ´ 2q
“ ´ 16 ´ 6λ ` λ2 ` λ ´ 14 ´ λ3 ´ λ2 ` 14λ
“ ´ λ3 ` 9λ ´ 30

(b) Do the columns of A span R3 ? Answer yes or no and justify briefly your answer.

Solution: pp0q “ ´30, so detpAq ‰ 0. Therefore, the columns of A span R3 .

37. This is the same question as no. 3. Let A be a n ˆ n matrix such that A2 ´ 3A “ 2In
(where In is the n ˆ n identity matrix).

(a) Show that A is invertible.

`1 ˘
Solution: A2 ´ 3A “ 2In ùñ ApA ´ 3Iq “ 2In ùñ A 2
A ´ 32 I “ In : this last equality
shows that A is invertible and its inverse is 21 A ´ 32 I.

(b) Can 3 be an eigenvalue of A? Justify your answer.

Solution: Suppose that 3 is an eigenvalue of A. Then there exists a non-zero vector v such
that Av “ 3v, so pA ´ 3Iqv “ 0.

Applying both sides of ApA ´ 3Iq “ 2In to v gives ApA ´ 3Iqv “ 2In v, hence 0 “ 2v
since pA ´ 3Iqv “ 0. However, 0 “ 2v is not possible since v is supposed to be a non-zero
eigenvector.

Since a contradiction has been obtained, it follows that the initial assumption that 3 is
an eigenvalue must be rejected. Consequently, 3 is not an eigenvalue of A.

38
 » fi
2 3 ´6
38. Let A “ –0 1 2 fl.
0 ´1 4

(a) Determine the characteristic polynomial of A.

Solution: Using expansion along the first column, we obtain:

2´λ 3 ´6
1´λ 2
detpA ´ λIq “ 0 1´λ 2 “ p2 ´ λq
´1 4 ´ λ
0 ´1 4 ´ λ
“ p2 ´ λqpp1 ´ λqp4 ´ λq ` 2q “ ´pλ ´ 2qpλ2 ´ 5λ ` 6q
“ ´ pλ ´ 2qpλ ´ 2qpλ ´ 3q.

(b) The eigenvalues of A are 2 and 3. For each eigenvalue, find a basis of the corresponding
eigenspace of A.

Solution: E2 “ nullpA ´ 2Iq, so let’s put A ´ 2I is reduced row echelon form:

» fi » fi » fi
0 3 ´6 0 0 0 0 1 ´2
A ´ 2I “ –0 ´1 2 fl R1 ` 3R2 , R3 ´ R2 –0 ´1 2fl ´ R2 , R1 Ø R2 –0 0 0 fl .
0 ´1 2 0 0 0 0 0 0

This last matrix shows that ~x P nullpA ´ 2Iq ðñ x2 ´ 2x3 “ 0. x1 and x3 are free variables,
so let’s set x1 “ s and x3 “ t. Then x2 “ 2t and
» fi » fi » fi
s 1 0
~x “ 2t “ s 0 ` t 2fl .
– fl – fl –
t 0 1

A basis for E2 is provided by $» fi » fi,

& 1 0 .
–0fl , –2fl .
0 1
% -

Let’s turn to E3 , which equals nullpA ´ 3Iq. Let’s put A ´ 3I is reduced row echelon
form: » fi » fi
´1 3 ´6 1 ´3 6
A ´ 3I “ 0 ´2 2
– fl ´ R1 , R2 ´ 2R3 0 ´2
– 2fl
0 ´1 1 0 0 0
» fi » fi
1 ´3 6 1 0 3
1
´ R2 –0 1 ´1fl R1 ` 3R2 –0 1 ´1fl .
2
0 0 0 0 0 0

39
This last matrix shows that ~x P nullpA ´ 3Iq ðñ
"
x1 ` 3x3 “ 0
x2 ´ x3 “ 0

x3 is a free variable, so let’s set x3 “ t. Then x1 “ ´3t, x2 “ t and

» fi » fi
´3t ´3
~x “ – t fl “ t – 1 fl .
t 1
» fi
´3
A basis for E3 is provided by the single vector – 1 fl.
1

(c) The matrix A is diagonalizable. Explain why.

Solution: A is diagonalizable because the dimension of E2 is equal to the algebraic multiplic-

ity of 2, that is, it is equal to the multiplicity of 2 as a root of the characteristic polynomial
detpA ´ λIq. (This multiplicity is 2 since pλ ´ 2q2 divides detpA ´ λIq.) For the other
eigenvalue, it is also true that dim E3 is also equal to the algebraic multiplicity of 3 (which
is one).

(d) Find an invertible 3ˆ3 matrix P and a 3ˆ3 diagonal matrix D such that P ´1 AP “ D.
(You shouldn’t need to compute P ´1 for this!)

Solution: The columns of P are the eigenvectors found in (b) and the diagonal entries of D
are their corresponding eigenvalues, so
» fi » fi
1 0 ´3 2 0 0
P “ 0 2 1
– fl and D “ 0– 2 0fl .
0 1 1 0 0 3

(e) Is A125 diagonalizable? Justify your answer.

Solution: A125 is fi
» also diagonalizable since A125 “ P D125 P ´1 and D125 is the diagonal matrix
125
2 0 0
– 0 2125 0 fl.
0 0 3125

40
 » fi
1 1 1
39. This question has 4 parts, all concerned with the matrix A “ –0 ´1 0 fl.
0 0 ´1

(a) Compute the characteristic polynomial of A.

Solution: A ´ λI is a diagonal matrix, so its determinant is the product of the entries on its
diagonal:
ˇ ˇ
ˇ1 ´ λ 1 1 ˇ
ˇ ˇ
detpA ´ λIq “ ˇ 0
ˇ ´1 ´ λ 0 ˇˇ “ p1 ´ λqp´1 ´ λqp´1 ´ λq “ ´pλ ´ 1qpλ ` 1q2 .
ˇ 0 0 ´1 ´ λˇ

(b) Find the eigenvalues of A and list their algebraic multiplicities.

Solution: The eigenvalues of A are the roots of its characteristic polynomial, which are 1
and ´1. The algebraic multiplicity of 1 is one and the algebraic multiplicity of ´1 is two
because λ ` 1 appears with exponent 2 in detpA ´ λIq.

(c) Show that A is diagonalizable by finding explicitly a matrix P which diagonalizes it.

Solution: First, bases for the eigenspaces are needed.

Let’s start with E1 , which equals nullpA ´ Iq. Let’s put A ´ I in reduced row echelon
form:
» fi » fi » fi
0 1 1 0 1 1 0 0 0
1 1
A ´ I “ 0 ´2
– 0 fl ´ R2 , ´ R3 –0 1 0fl R1 ´ R2 , R1 ´ R3 –0 1 0fl
2 2
0 0 ´2 0 0 1 0 0 1
» fi
0 1 0
R1 Ø R2 , R2 Ø R3 –0 0 1fl .
0 0 0
It follows
» fithat if x P E1 , then x2 “ 0, x3 “ 0 and x1 is a free variable. Therefore, a basis for
1
E1 is –0fl.
0

Now let’s find a basis for E´1 , which equals nullpA ` Iq. Let’s put A ` I in reduced row
echelon form: » fi » 1 1 fi
2 1 1 1 2 2
1
A`I “ 0 0 0
– fl R1 0 0 0 fl .
–
2
0 0 0 0 0 0

41
x2 and x3 are free variables, so let’s set x2 “ s and x3 “ t. It follows that if x P E´1 , then
x1 ` 12 x2 ` 12 x3 “ 0, hence x1 “ ´ 12 s ´ 21 t. Therefore,
» 1 fi » 1 fi » 1 fi
´ 2 s ´ 21 t ´2 ´2
x“ – s fl “ s 1 ` t 0 fl
– fl –
t 0 1
$» 1 fi » 1 fi,
& ´2 ´2 .
and a basis for E1 is – 1 , 0 fl .
fl –
0 1
% -

» fi
1 ´ 21 ´ 21
To diagonalize A, we can thus use the matrix P given by P “ –0 1 0 fl and the
» fi 0 0 1
1 0 0
diagonal matrix D given by D “ –0 ´1 0 fl.
0 0 ´1

(d) Use the previous parts to compute A2015 .

Solution: A “ P DP ´1 with P and D as in part (c). It follows that

A2015 “ P D2015 P ´1
» 2015 fi
1 0 0
“P– 0 p´1q2015 0 fl P ´1
0 0 p´1q2015
“ P DP ´1 “ A.

» fi
4 0 1
40. Let A “ –1 0 2fl.
0 0 4

(a) Determine the eigenvalues of A.

Solution: The characteristic polynomial of A is detpA ´ λIq. Let’s expand this determinant
along the third row:
ˇ ˇ
ˇ4 ´ λ 0 1 ˇˇ ˇ ˇ
ˇ ˇ4 ´ λ 0 ˇ
detpA ´ λIq “ ˇˇ 1 ´λ 2 ˇˇ “ p4 ´ λq ˇˇ ˇ
ˇ 0 1 ´λˇ
0 4 ´ λˇ
“ p4 ´ λqp4 ´ λqp´λq “ ´λpλ ´ 4q2 .
The eigenvalues of A are the roots of its characteristic polynomial, which are 0 and 4.

42
 (b) For each eigenvalue λ, compute a basis for the associated eigenspace Eλ .

Solution: E0 “ nullpAq, so let’s put A in reduced row echelon form to obtain a basis of E0 :
» fi » fi » fi » fi
4 0 1 0 0 ´7 0 0 0 1 0 0
–1 0 2fl R1 ´ 4R2 , 1 R3 –1 0 2 fl
R1 ` 7R3 –
1 0 0fl
R1 Ø R2 –
0 0 1fl .
4 R2 ´ 2R3 R2 Ø R3
0 0 4 0 0 1 0 0 1 0 0 0

This last matrix shows

» fi that
» iffi Ax “ 0, then x1 “ 0 and
$»x3fi“,0. x2 is a free variable, so if
0 0 & 0 .
x2 “ t, then x “ t “ t 1 . A basis for E0 is thus –1fl .
– fl – fl
0 0 0
% -

Let’s now turn to the eigenspace E4 , which equals nullpA ´ 4Iq. Let’s put A ´ 4I in
reduced row echelon form:
» fi » fi » fi
0 0 1 1 ´4 2 1 ´4 0
A ´ 4I “ 1 ´4 2 R1 Ø R2 0 0 1 R1 ´ 2R2 0 0
– fl – fl – 1fl .
0 0 0 0 0 0 0 0 0

If x P E4 , then it follows from the last »

matrix
fi that
» x
fi1 ´ 4x2 “ 0 and x3 “ 0.$» x2 fi
is,a free
4t 4 & 4 .
variable. Set x2 “ t, so x1 “ 4t and x “ – t fl “ t –1fl. A basis for E4 is thus –1fl .
0 0 0
% -

(c) Is A diagonalizable? If no, then state why not; if yes, then give an invertible matrix
P and a diagonal matrix D such that P ´1 AP “ D.

Solution: A is not diagonalizable because the dimension of E4 is 1, so the geometric mul-

tiplicity of the eigenvalue 4 is 1, but its algebraic multiplicity is 2 since the characteristic
polynomial of A has pλ ´ 4q2 as a factor.

„ 
3 1
41. Let A “ . Using diagonalization, compute the matrix A50 , the 50th power of A.
2 4

Solution: Let’s start by computing the characteristic polynomial of A:

ˇ ˇ
ˇ3 ´ λ 1 ˇ
detpA ´ λIq “ ˇˇ ˇ “ p3 ´ λqp4 ´ λq ´ 2 “ λ2 ´ 7λ ` 10.
2 4 ´ λˇ
? ?
´p´7q˘ p´7q2 ´4¨10 7˘ 9
The roots of this characteristic polynomial are 2
, that is, 2
, so the eigen-
values of A are 2 and 5.

43
 To diagonalize A, we need one non-zero eigenvector for each eigenvalue.
„  „ 
1 1 1 1
A ´ 2I “ R2 ´ 2R1 . Therefore, an eigenvector for the eigenvalue 2 must
2 2 0„ 0 
1
satisfy x1 ` x2 “ 0: one choice is .
´1
„  „ 
´2 1 ´2 1
A ´ 5I “ R2 ` R1 . Therefore, an eigenvector for the eigenvalue 5
2 ´1 0 0„ 
1
must satisfy ´2x1 ` x2 “ 0: one choice is .
2
„  „ 
1 1 2 0
A can be diagonalized using the matrix P and D given by P “ and D “ :
´1 2 0 5
A “ P DP ´1 .

P ´1 will be needed below, so let’s compute it:

„  „  „  „ 2

1 1 1 0 1 1 1 0 1 1 1 1 0 1 0 3
´ 31
R2 ` R1 R2 1 1 R1 ´ R2 1 1
´1 2 0 1 0 3 1 1 3 0 1 3 3
0 1 3 3

It follows that
„ „  „2 1
 „ 50  „2 1

50 50 ´1 1 1 250 0 3
´ 3 “
2 550
3
´ 3
A “ PD P “
´1 2 0 550 13 13 ´250 2 ¨ 550 13 13
„ 251 `550 550 ´250

“ 2¨5503´251 3
250 `2¨550
3 3

42. (OPTIONAL) Let A be a nonzero square matrix (so A ‰ 0) such that A2 “ 0. Show
that A is not diagonalizable.

Solution: Suppose that A is diagonalizable, so A “ P DP ´1 for some invertible matrix P

and some diagonal matrix D. Then A2 “ 0 ùñ P D2 P ´1 “ 0 ùñ D2 “ 0. Since D is a
diagonal matrix, D2 “ 0 can only happen when D “ 0, and then A must be also the zero
matrix. However, this is not possible since it is stated in the question that A is non-zero.
Therefore, the assumption that A is diagonalizable must be rejected and it follows that A is
not diagonalizable.

7 Multiple Choice Questions

43. Let A and B be matrices with determinant 1 and ´1 respectively. Which of the following
is always true?

44
 (a) detpA ` Bq “ 0

(b) detpA´1 q “ ´1

(c) detpA2 B 2 q “ detpAq

(d) detpAB ´1 q “ 1

Solution: (c) is always true because detpA2 B 2 q “ detpAq2 detpBq2 “ 1.

44. Which of the following statements is true?

(a) The columns of a 2 ˆ 3 matrix could form a basis of R2 .

(b) There exists a square matrix A whose rows are linearly independent, but whose
columns are linearly dependent.

(c) There can exist a 4 ˆ 12 matrix with 3 linearly independent columns.

(d) There exists a 3 ˆ 3 matrix with nullity 2 and rank 2.

Answer: (c)

45. Let A be a 3 ˆ 3 matrix with distinct eigenvalues. Which of the following could be a
possible list of the geometric multiplicities of the eigenvalues of A?

(a) 1, 2

(b) 1, 1, 1

(c) 3

(d) Not enough information to determine.

Answer: (b)

46. Let A be a 3 ˆ 3 matrix with nullity one. Which of the following statements is always
true?

(a) 0 is an eigenvalue of A.

(b) 1 is an eigenvalue of A.

45
 (c) The rows of A are linearly independent.

(d) The system Ax “ 0 has a unique solution x P R3 .

Answer: (a)

47. Let A be an arbitrary 2 ˆ 2 matrix, and B an invertible 2 ˆ 2-matrix. Which of the

following is NOT always true?

(a) The row space of BA is equal to the row space of A.

(b) The row space of AB is contained in the row space of B.

(c) The column space of AB is equal to the column space of A.

(d) The column space of BA is contained in the column space of A.

Answer: (d)

8 True or false questions

48. Mark each of the following statements as “true” or “false”, and in each case justify why
your answer is correct, i.e. explain briefly why the statement is true or give a counterexample
to show that it is false:

(a) If A is a real matrix with three rows and two columns, then multiplication by A
defines a linear transformation from R3 to R2 .

Answer: False. If A is a 3 ˆ 2 matrix and v P R2 , then Av is in R3 , so multiplication by A

defines a linear transformation from R2 to R3 .

(b) If zero is an eigenvalue of a matrix A, then A is not invertible.

Answer: True. The eigenspace for 0 is the null space of A, so if 0 is an eigenvalue of A, then
there must be a non-zero vector v such that Av “ 0. This can only happen when A is not
invertible.

(c) If a linear transformation from Rm to Rn is onto, then m ě n.

46
Answer: True. If T : Rm ÝÑ Rn is onto and given by an n ˆ m matrix A, so T pvq “ Av,
then the rank of A is n. The rank of any matrix is always less than or equal to its number
of columns, so n ď m.

(d) If A is a matrix with three rows and five columns such that the reduced row echelon
form of A has exactly three pivots, then the null space NulpAq is three-dimensional.

Answer: False. The Rank-Nullity Theorem implies that rankpAq ` nullitypAq “ 5. The rank
of A is 3 because A has 3 pivots, so its nullity is 2. The nullity is the dimension of nullpAq,
so this null space has dimension 2.

(e) If S “ tv1 , v2 , v3 , v4 u is a subset of R3 then S is a linearly dependent set.

Answer: True. The dimension of R3 is 3, so any set of more than 3 vectors is a linearly
dependent set.

49. Answer only T (true) or F (false). You don’t have to justify your answer.

49.1 Suppose that A is a 4 ˆ 4 matrix with eigenvalues 3 and ´5 and these have geometric
multiplicities 1 and 2. Then A is diagonalizable.

Answer: False

49.2 If the square matrix A has nullity 0, then A is invertible.

Answer: True

49.3 n linearly independent vectors in Rn must always span Rn .

Answer: True

49.4 Let A and B be n ˆ n matrices. Suppose that λ1 is an eigenvalue of A and λ2 is an

eigenvalue of B. Then λ1 λ2 is an eigenvalue of AB.

Answer: False

49.5 If the matrix A has more columns than rows, then its columns are linearly dependent.

47
Answer: True

49.6 The matrix of a rotation in R3 always has an eigenvector.

Answer: True

49.7 If S1 and S2 are distinct subspaces of Rn , then the dimension of the subspace S1 ` S2
is equal to dimpS1 q ` dimpS2 q.

Answer: False

49.8 If λ is an eigenvalue of the matrix A of algebraic multiplicity equal to 1, then its

geometric multiplicity is also equal to 1.

Answer: True

9 Other questions

50. Let u and v be the two vectors in R3 given by

» fi » fi
3 1
u“ 1 – fl and v “ ´4fl .
–
´2 5
Use the cross-product to obtain the equation in general form of the plane that is parallel to
u and v and which passes through the point p3, 1, ´1q.

Solution:
ˇ ˇ
ˇe 1 3 1 ˇˇ
ˇ
n “ ˇˇe2 1 ´4ˇˇ
ˇe3 ´2 5 ˇ
ˇ ˇ ˇ ˇ ˇ ˇ
ˇ 1 ´4ˇ ˇ 3 1ˇ ˇ3 1 ˇ
“ e1 ˇˇ ˇ ´ e2 ˇ
ˇ ˇ ` e3 ˇ
ˇ ˇ
´2 5 ˇ ´2 5ˇ 1 ´4ˇ
“ p5 ´ 8qe1 ´ p15 ´ p´2qqe2 ` p´12 ´ 1qe3
“ ´ 3e1 ´ 17e2 ´ 13e3
» fi
´3
“ –´17fl
´13

48
 »fi
3
Now let p “ – 1 fl. Then n ¨ p “ ´3 ¨ 3 ´ 17 ¨ 1 ´ 13 ¨ p´1q “ ´9 ´ 17 ` 13 “ ´13. The
´1
equation of P in general form is thus 3x ` 17y ` 13z “ 13.

51. Only a short answer (one or two lines) is required below.

51.1 Let A be an n ˆ n matrix with detpAq ‰ 0. Give one property of A that can be deduced
from the fact that its determinant is non-zero.

Solution: A is invertible.

51.2 If A is a 4 ˆ 6 matrix of nullity 3, what is the rank of A? What is the name of the
theorem that you need to use?

Solution: By the Rank-Nullity Theorem, rankpAq ` 3 “ 6, so rankpAq “ 3.

51.3 Let v1 , . . . , vk be k vectors in a subspace S of Rn . What are the two properties that
those vectors must possess to form a basis of S?

Solution: They must be linearly independent and they must span S.

„  „ 
5 4 3 5
51.4 Give one reason why the matrices A and B given by A “ and B “ are
4 5 3 7
not similar.

Solution: detpAq “ 25 ´ 16 “ 9 and detpBq “ 21 ´ 15 “ 6. Since detpAq ‰ detpBq, the

matrices A and B are not similar.

51.5 If A is a square matrix and the?complex number 3 ` 5i is an eigenvalue of A, what is

another eigenvalue of A? (Here, i “ ´1.)

Solution: 3 ´ 5i

51.6 If v1 , . . . , vk are eigenvectors of the matrix A with eigenvalues λ1 , . . . , λk , give a condition

on these eigenvalues which implies that those vectors are linearly independent.

Solution: Those eigenvectors are linearly independent when the eigenvalues λ1 , . . . , λk are
all pairwise distinct: λi ‰ λj when i ‰ j.

49
51.7 Let v1 , . . . , vk be vectors in Rn . Give the definition of spanpv1 , . . . , vk q, the span of
those vectors.

Solution: spanpv1 , . . . , vk q is the set of all linear combinations of v1 , . . . , vk :

spanpv1 , . . . , vk q “ tc1 v1 ` ¨ ¨ ¨ ` ck vk | c1 , . . . , ck P Ru.

51.8 Let A be a 3 ˆ 3 matrix with eigenvectors v1 , v2 and v3 of eigenvalues 3, 1, ´4, respec-

tively. These form a basis B of R3 . Let v be the vector in R3 given by v “ 3v1 ´ 2v2 ´ 3v3 .
Let u “ A4 v. Determine the coordinate vector rusB of the vector u in the basis B.

Solution:

u “ 3A4 v1 ´ 2A4 v2 ´ 3A4 v3 “ 3 ¨ 34 v1 ´ 2v2 ´ 3p´4q4 v3 “ 35 v1 ´ 2v2 ´ 3 ¨ 44 v3

so » fi
35
rusB “ – ´2 fl .
´3 ¨ 44

"„  *
x 2
51.9 Let S “ P R | x ď y . Give one reason or one counterexample why S is not a
y
subspace of R2 .

„  „  „  „ 
0 0 0 0
Solution: P S but ´ R S since ´ “ and 0 ą ´1.
1 1 1 ´1

ˆ„ ˙ „ 
2 2 x x ` 3y
51.10 Let T : R Ñ R be the function given by T “ 2 . Give one reason or
y x ` y2
one counterexample that shows that T is not a linear transformation.

ˆ„ ˙ „  ˆ„ ˙ „  ˆ„ ˙ ˆ„ ˙
1 1 ´1 ´1 ´1 1
Solution: T “ and T “ , so T ‰ ´T .
0 1 0 1 0 0

52. Find all values of the real numbers h and k such that the system

x1 ´ x2 “ k
2x1 ` hx2 “ ´2

50
of linear equations has

(a) No solutions.

(b) A unique solution.

(c) Infinitely many solutions.

„ 
1 ´1 k
Solution: The augmented matrix of that system is . Let’s put it in row
2 h ´2
echelon form: „  „ 
1 ´1 k 1 ´1 k
R2 ´ 2R1
2 h ´2 0 h`2 ´2 ´ 2k

(a) The system has no solutions if the second row is of the form r0 0 ˚s where ˚ is a
non-zero number. This happens when h ` 2 “ 0 and ´2 ´ 2k ‰ 0, that is, when h “ ´2 and
k ‰ ´1.

(b) The system has a unique solution if the second row is of the form r0 a bs where a is a
non-zero number and b is any number. This happens when h ` 2 ‰ 0, that is, when h ‰ ´2,
and for any k P R.

(c) There are infinitely many solutions if the second row is of the form r0 0 0s. This
happens when h ` 2 “ 0 and ´2 ´ 2k “ 0, that is, when h “ ´2 and and k “ ´1.

51