Introduction to Analysis - BX27

Tutorial II

Intervals

Exercise 1. (i) Prove that

ß ™
1
x ∈ R : there exists n ⩾ 2 such that ⩽ x < 1 = (0, 1).
n

Observe that, if
ß ™
1
z ∈ x ∈ R : there exists n ⩾ 2 such that ⩽x<1 ,
n

then z ∈ (0, 1) and hence

ß ™
1
x ∈ R : there exists n ⩾ 2 such that ⩽ x < 1 ⊆ (0, 1).
n

Now, take z ∈ (0, 1). Thanks to the Archimedean property of R, we can find n ∈ N
such that n > z1 and n ⩾ 2. Then, n1 ⩽ z < 1 and hence
ß ™
1
z ∈ x ∈ R : there exists n ⩾ 2 such that ⩽x<1 .
n
So ß ™
1
(0, 1) ⊆ x ∈ R : there exists n ⩾ 2 such that ⩽x<1 .
n

(ii) Prove that ß ™

1
x ∈ R : for all n ⩾ 1 − < x < 1 = [0, 1).
n

Observe that, if z ∈ [0, 1), then

1
− < z < 1,
n
for all n ⩾ 1 and hence
ß ™
1
[0, 1) ⊆ x ∈ R : for all n ⩾ 1, − < x < 1 .
n

1
 Now, if z ⩾ 1, then
ß ™
1
z∈
/ x ∈ R : for all n ⩾ 1 − < x < 1 .
n

Finally, if z < 0, thanks to the Archimedean property of R, we can find n ∈ N such

that n > − z1 and n ⩾ 1. Then, z < − n1 and hence
ß ™
1
z∈/ x ∈ R : for all n ⩾ 1 − < x < 1 .
n

Therefore, we conclude that

ß ™
1
x ∈ R : for all n ⩾ 1, − < x < 1 = [0, 1).
n

(iii) Determine the subset of R defined by

∞ Å ã
[ 1 n
, .
n+2 n+2
n=2

Since Å ã
1 n
, ⊆ (0, 1),
n+2 n+2
for all n ⩾ 2, we get
∞ Å ã
[ 1 n
, ⊆ (0, 1).
n=2
n+2 n+2

Now, take x ∈ (0, 1). Since x ̸= 0 and x ̸= 1, we can use the Archimedean property
of R to choose n ∈ N, n ⩾ 2, such that n + 2 > x1 and n + 2 > 1−x
2
. With is choice

1 n
<x< ,
n+2 n+2
and hence
∞ Å ã
[ 1 n
(0, 1) ⊆ , ,
n=2
n+2 n+2
and we conclude that
∞ Å ã
[ 1 n
, = (0, 1).
n=2
n+2 n+2

(iv) Determine the subset of R defined by

\ Å 1 n + 2ã
− , .
n n+1
n⩾1

2
 Since
1 n+2
− <0 and 1< ,
n n+1
for all n ⩾ 0, we get the inclusion
Å ã
1 n+2
[0, 1] ⊆ − , ,
n n+1
for all n ⩾ 0 and we can already conclude that
\ Å 1 n + 2ã
[0, 1] ⊆ − , .
n n+1
n⩾1

Now, if x < 0, we use the Archimedean property of R to prove

Ä the existence
ä of n ∈ N
1 1 1 n+2
such that n > − x . Hence x < − n and this shows that x ∈
/ − n , n+1 . Therefore
Å ã
\ 1 n+2
x∈/ − , .
n n+1
n⩾1

Similarly, if x > 1, we use the Archimedean property of R to prove the existence

Ä of
ä
1 n+2
n ∈ N such that n + 1 > x−1 . Hence, x > n+1 / − n1 , n+2
and this shows that x ∈ n+1 .
Therefore
\ Å 1 n + 2ã
x∈
/ − , .
n n+1
n⩾1

We conclude that
\ Å 1 n + 2ã
− , = [0, 1].
n n+1
n⩾1

Exercise 2. Assume A and B are two non-empty intervals of R.

(i) Show that A ∩ B is an interval.

Assume we are given x, y ∈ A ∩ B with x ⩽ y. Let z ∈ R be such that x ⩽ z ⩽ y.

Since x, y ∈ A, x ⩽ z ⩽ y and A is an interval, then z ⊂ A. Similarly, since x, y ∈ B,
x ⩽ z ⩽ y and B is an interval, then z ⊂ B. Therefore, z ∈ A ∩ B, which proves that
A ∩ B is an interval.

(ii) Is it always true that A ∪ B is an interval?

This is not always true. For example, [0, 1] ∪ [2, 3] is the union of two intervals but it
is not an interval.

Exercise 3. Let A ⊆ R, A ̸= ∅ be given. We define

B := {|x| : x ∈ A}.

Give counterexamples for the statements that are false and a proof for the statements
that are true.

3
(i) If B is an interval, then A is an interval.
This is false. Consider for example A := [−2, −1] ∪ [1, 2], for which B = [1, 2].

(ii) If A is an interval, then B is an interval.

This is true. Take any x, y ∈ B with x < y. Then there exists α, β ∈ A such that
x = |α| and y = |β|. We distinguish cases according to the sign of α and β.
(i) If both α ⩾ 0 and β ⩾ 0. Then 0 ⩽ α ⩽ β and [α, β] ⊆ A. In this case we
conclude that [a, b] ⊆ B.
(ii) If α ⩽ 0 and β ⩾ 0. Since α2 ⩽ β 2 , we conclude that 0 ⩽ −α ⩽ β and −α ∈ A
since −α ∈ [α, β] which is included in A. Therefore, [a, b] ⊆ B also in this case.
(iii) If β ⩽ 0 and α ⩾ 0. Since α2 ⩽ β 2 , we conclude that β ⩽ −α ⩽ 0 and
−α ∈ [β, α] which is included in A. Therefore, [a, b] ⊆ B also in this case.
(iv) Finally, if α ⩽ 0 and β ⩽ 0, then β ⩽ α ⩽ 0 and [β, α] ⊆ A. In this case we also
conclude that [a, b] ⊆ B.

(iii) If A has an upper bound, then B has an upper bound.

This is false. Consider for example A = (−∞, 0] which has 0 as an upper bound and
for which B = [0, +∞) has no upper bound.

(iv) If B has an upper bound, then A has an upper bound.

This is true. Indeed, if M is an upper bound for B, then |x| ⩽ M for all x ∈ A.
Hence M is an upper bound for A.

Density

Exercise 4. Prove that the set of irrational numbers is dense in R.

√
We know that 2 is not a rational number. Take a < b two real numbers. Since Q
is dense in R, there exists r ∈ Q (observe that one can always assume that r ̸= 0)
such that
a b
√ <r< √ .
2 2
Hence √
a < r 2 < b,
√
and q 2 is an irrational number.

Exercise 5. Given x ∈ R and n ∈ N, prove that there exists k ∈ Z such that

|2n x − k| < 1.
ß ™
k
Deduce from this that the set : k ∈ Z, n ∈ N is dense in R.
2n

4
 Given x ∈ R, we define k := ⌊2n x⌋. By definition

k ⩽ 2n x < k + 1,

and hence |2n x − k| < 1. It is important to observe that k depends on n (and on

x).

Now, given x, y ∈ R with x < y, we set

x+y
z := ,
2
and choose n ∈ N such that
1 y−x
< .
2n 2
To prove that such a choice is possible, just invoque the Archimedean Property of R
together with the fact that 2n ⩾ n for all n ⩾ 1. Observe that, in particular, this
choice implies that
1 1
x<z− n and z + n < y.
2 2
According to the first part, we know that there exists k ∈ N such that

|2n z − k| < 1.

So
1 k 1
z− < n < z + n,
2n 2 2
and we conclude that
k
x< < y,
2n
ß ™
k
which proves that : k∈Z n ∈ N is dense in R.
2n

Exercise 6. Prove that the set {r3 : r ∈ Q} is dense in R.

Consider 0 ⩽ x < y. Then x1/3 < y 1/3 and, since Q is dense in R, there exists
r ∈ Q such that x1/3 < r < y 1/3 and hence x < r3 < y. Similarly, if x < y ⩽ 0, then
according to what we have already proven there exists r ∈ Q such that |y|1/3 < r <
|x|1/3 and hence x < −r3 < y. The situation where x < 0 < y can be covered since
we now know that there exists r ∈ Q such that 0 < r3 < y and hence x < r3 < y.

Complex Numbers

5
Exercise 7. Write the following complex numbers in the cartesian form (i.e. a + ib,
with a, b ∈ R):
Ç √ å3
−i π4 (1 + i)9 1 3 10
3e , , − +i and .
(1 − i)7 2 2 1 − 3i

The first one is straightforward

√ √
π π 3 2 3 2
3 cos(− ) + 3i sin(− ) = +i .
4 4 2 2
For the second one, we uses the polar form of a complex number:
Ç √ å3
1 3 Ä 2π ä3
− +i = ei 3 = e2πi = 1.
2 2

For the last one, we multiply by the complex conjugate

10 10 (1 + 3i) 10 (1 + 3i)
= = = 1 + 3i.
1 − 3i (1 − 3i)(1 + 3i) 10

Exercise 8. Write the following complex numbers in polar form i.e. as reiθ with r ⩾ 0
and θ ∈ [0, 2π[:
√ 1 − i tan φ
1 + i 3, and eiφ + eiψ ,
1 + i tan φ
where φ, ψ ∈ R (for the second number, we assume that φ ̸= π/2 + kπ for all k ∈ Z).
The first one is √ å
√
Ç
1 3 π
1+i 3=2 +i = 2 ei 3 .
2 2
The second one is
1 − i tan φ cos φ − i sin φ
= = e−2iφ .
1 + i tan φ cos φ + i sin φ
The last one is
φ−ψ
Å ã
i i i i
Ä ä
iφ iψ 2 (φ+ψ) 2 (φ−ψ) 2 (ψ−φ)
e +e =e e +e = 2 cos e 2 (φ+ψ) .
2

Exercise 9. (i) Prove that

Å ã
7x x
sin(4x) − sin(3x) = 2 cos sin .
2 2

6
 We use the “mean angle” trick:
Ä 7x x x

ä
sin(4x) − sin(3x) = ℑ e4xi − e3xi = ℑ e 2 i e 2 i − e− 2 i ,

7x x



and then sin(4x) − sin(3x) = 2 cos 2 sin 2 .

(ii) Solve the following equation for x ∈ R:

sin x − sin(3x) + sin(4x) = 0.

We use (i) together with the formula sin x = 2 sin x2 cos x2 to show that the equation



can be rewritten as x Å Å ã  x ã
7x
sin cos + cos = 0.
2 2 2
Therefore, either sin x2 = 0 (i.e. x ∈ 2πZ) or cos 7x = − cos x2 (i.e. x = π3 + 2π




2 3 Z
or x = π4 + π2 Z). Finally, the set of solutions is given by
nπ π o ßπ 2π
™
{kπ : k ∈ Z} ∪ +k : k ∈Z ∪ +k : k∈Z .
4 2 3 3

Exercise 10. Express (cos x)3 in terms of cos(3x) and cos x.

Using (x + y)3 = x3 + 3x2 y + 3xy 3 + y 3 , which is valid for any x, y ∈ C, we compute

Ç å3
3 eiθ + e−iθ e3iθ + e−3iθ eiθ + e−iθ 1 3
(cos x) = = +3 = cos(3x) + cos x.
2 8 8 4 4

Exercise 11. For each of the following equations, find the solutions z ∈ C:

1
|z| = = |z − 1|, z = i (z − 1) and |z + 3i| = 3 |z| .
z

(i) The first equality shows that necessarily |z| = 1. Using this information, the
second equality reads |z − 1| = 1. Writing z := x + iy, where x, y ∈ R, we get
the equations x2 + √y 2 = 1 and (x − 1)2 + y 2 = 1. The solutions of this system are
x =
¶ √ 1/2 and y
√ ©
= ± 3/2. So the set of complex numbers satisfying the equations is
1+i 3 1−i 3
2 , 2 .
(ii) We write z := x + iy, where x, y ∈ R. Since the real and the imaginary part are
unique, the equation z̄ = i(z − 1) is equivalent to the system
ß
−y = x − 1
x = −y

and thus, there are no solutions.

7
 (iii) Writing again z := x+iy, where x, y ∈ R, we see that the equation is equivalent
to
8x2 + 8y 2 − 6y = 9.

2
Let us remark that 8x2 + 8y 2 − 6y = 8x2 + 8 y − 83 − 98 . Therefore, the set of pairs
(x, y) ∈ R2 verifying the above equation is the circle of center (0, 38 ) and of radius
9
8 . We conclude that the set of complex numbers verifying |z + 3i| = 3 |z| is given by
{ 83 i + 89 eiθ : θ ∈ R}.

Exercise 12. Solve in C

z 2 − (4 + i)z + 5 + 5i = 0.
We compute the discriminant

∆ := (4 + i)2 − 4 (5 + 5i) = −5 − 12i.

We look for a, b ∈ R such that (a + ib)2 = −5 − 12i. Expanding the square, we get
the system ( 2
a − b2 = −5

2ab = −12,
whose solutions are given by (a, b) = (−2, 3) and (a, b) = (2, −3). Going back to the
equation we want to solve, we get the solutions z1 := 3 − i and z2 := 1 + 2i.

Exercise 13. Assume n ∈ N, n ⩾ 2. Find the set of complex numbers z ∈ C such that

(z − 1)n = (z + 1)n .

Observe that z solves the equation if and only if z−1 th

z+1 is a n -root of unity. Therefore,
the set of solutions is given by
1+ω
z= ,
1−ω
where ω ∈ C is a nth -root of unity, ω ̸= 1. Writing ω = eiθ , with θ = 2πk n , for
k = 1, . . . , n − 1 (observe that we exclude k = 0 since ω has to be different from 1),
we get
cos θ2


1 + eiθ
z= = i .
1 − eiiθ sin θ2

Therefore, the set of solutions is given by

cos kπ
®
´
n
i
: k = 1, . . . , n − 1 .
sin kπ
n

Exercise 14. (Difficult). In this exercise, we prove that cos(2π/7) is not a rational
number.

8
(i) Prove that
cos(2α) = 2 (cos α)2 − 1.
and
cos(4α) = 8 (cos α)4 − 8 (cos α)2 + 1.

We use
cos(2α) = (cos α)2 − (sin α)2 ,
and
(cos α)2 + (sin α)2 = 1,
to conclude that
cos(2α) = 2 (cos α)2 − 1.
Then,

cos(4α) = 2 (cos(2α))2 − 1 = 2 (2 (cos α)2 − 1) − 1 = 8 (cos α)4 − 8 (cos α)2 + 1.

(ii) Prove that

cos(3α) = 4 (cos α)3 − 3 cos α.

We compute

cos(3α) = cos α cos(2α) − sin α sin(2α)

= cos α((cos α)2 − (sin α)2 ) − 2 (sin α)2 cos α

= (cos α)3 − 3 cos α (sin α)2

= (cos α)3 − 3 cos α (1 − (cos α)2 )

= 4 (cos α)3 − 3 cos α.

(iii) Show that cos(2π/7) is a root of the polynomial

P (x) := 8x3 + 4x2 − 4x − 1.

Hint: Observe that cos(8π/7) = cos(6π/7).

We have 6π/7 = π − π/7 and 8π/7 = π + π/7 hence cos(8π/7) = cos(6π/7). We use
the previous questions with α = 2π/7, to conclude that

8 (cos α)4 − 8 (cos α)2 + 1 = 4 (cos α)3 − 3 cos α.

(iv) Prove that cos(2π/7) is not a rational number.

We argue by contradiction and assume that cos(2π/7) is a rational number. We write

cos(2π/7) = p/q where p, q ∈ N \ {0} where p and q are coprime. Thanks to the
previous question, we conclude that

8 p3 + 4 p2 q 2 − 4 p q 3 − q 4 = 0.

9
 This implies that q 4 is even and hence q is even. Writing q = 2 q̄, we

8 p3 + 16 p2 q̄ 2 − 32 p q̄ 3 − 16 q̄ 4 = 0,

which implies that

p3 + 2 p2 q̄ 2 − 4 p q̄ 3 − 2 q̄ 4 = 0.
and we conclude as above that p is even, which is a contradiction with the fact that p
and q are coprime.

Exercise 15. (Difficult) Let ω denote the principal n-th root of unity i.e. ω := e2iπ/n .
Compute
n−1
X
(1 + ω k )n .
k=0

We use the Binomial Formula to compute

n−1 n−1 n Ç å
! n Ç å X
n
!
X
k n
X X n k ℓ
X n ℓ k
(1 + ω ) = (ω ) = (ω ) .
ℓ ℓ
k=0 k=0 ℓ=0 ℓ=0 k=0

The key observation is

n−1
X 1 − ω nℓ
ω ℓk =
1 − ωℓ
k=0
ℓ
if ω ̸= 1. Therefore,
n
X
(ω ℓ )k = 0,
k=0

when ℓ = 1, . . . , n − 1. We conclude that

Ç å n ! Ç å n !
X
k n n X
0 k n X
n k
n − 1(1 + ω ) = (ω ) + (ω ) = 2n.
0 n
k=0 k=0 k=0

10