Assignment 10

Andrea Massaro
May 2025

Collaboration statement and last personal reflec-

tion
I declare that every answer reported in this assignment is my own.

I will keep this reflection short as the assignment is long enough. I am very
happy I am finally done with all these, and now I can enjoy my day and a half
before my final exam for microeconomics. Yay!

Post Scriptum: I really enjoyed Q6, it was fun messing around with the golden
ratio.

Problem 1
Question:

R x4 √
1. Let f : [1, ∞) → R, x → 1
e t
dt. Find f ′ for x ∈ [1, ∞).

R cos x
2. Let g : [ π8 , π2 ) → R be defined by g(x) = 1
sin x 1−t2
dt. Find g ′ for
x ∈ [ π8 , π2 ).

Part 1
R x4 √
t
Let f : [1, ∞) → R be defined as f (x) = 1
e dt.
√ R x4
x
Denote with h the function h(x) = e . Then f (x) = 1 h(t)dt. Denote with
H(x) a primitive of h. Observe that h is continuous on [1, ∞) and therefore
H is differentiable on [1, ∞) and H ′ (c) = h(c), for all c ∈ [1, ∞), as proved
in class. By the second fundamental theorem of calculus and the chain rule of
differentiation, it follows that
Z x4
f (x) = h(t)dt = H(x4 ) − H(1)
1

1
 d 4
f ′ (x) = H ′ (x4 ) · (x ) − H ′ (1)
dx
f ′ (x) = 4x3 · h(x4 )
√ 4 2
f ′ (x) = 4x3 e (x ) = 4x3 ex

Part 2
R cos x 1
Let g : [ π8 , π2 ) → R be defined by g(x) = sin x 1−t2 dt. Denote with h the func-
1
tion h(x) = 1−x2 . Denote with H a primitive of h. Observe that h is continuous
on [0, 1).

Then H is differentiable on [0, 1) and H ′ (c) = h(c), for all c ∈ [0, 1), as proved
in Spivak and in class.1
Now observe that, by the Second Fundamental Theorem of Calculus,
Z cos x
1
g(x) = dt = H(cos(x)) − H(sin(x))
sin x 1 − t2
d d
g ′ (x) = H ′ (cos(x)) ·
(cos(x)) − H ′ (sin(x)) · (sin(x))
dx dx
This will be true as we are differentiating with respect to x ∈ [ π8 , π2 ), and
√ √ √ √
therefore cos(x) ∈ [ 2+
2
2
, 1) ⊂ [0, 1) and sin(x) ∈ [ 2−2
2
, 1) ⊂ [0, 1)

g ′ (x) = h(cos(x))(− sin(x)) − h(sin(x))(cos(x))

1 1
=− (sin(x)) − (cos(x))
1 − (cos(x))2 1 − (sin(x))2
1 1
=− −
sin(x) cos(x)
cos(x) + sin(x)
=−
sin(x) cos(x)

Alternative approach for part 2

We know that Z b
1
dt = artanh(b) − artanh(a)
a 1 − t2
√ √ √ √
as long as |a| < 1 and |b| < 1. As 2+ 2
2
≤ cos(x) < 1 and 0 < sin(x) ≤ 2− 2
2
for x ∈ [π/8, π/2), we can use this rule to find a primitive of h. Hence,

g(x) = artanh(cos(x)) − artanh(sin(x))

1 Observe that the proof in Spivak shows that a continuous function on [a, b], when inte-

grated on [a, b], yields in a function F such that F ′ (c) = f (c) for all c ∈ (a, b). As we proved
in class that the statement is true for all c ∈ [a, b], then it follows that it will be true for all
c ∈ [a, b)

2
 d
g ′ (x) = (artanh(cos(x)) − artanh(sin(x)))
dx
1 1
g ′ (x) = (− sin(x)) − (cos(x))
1 − cos2 (x) 1 − sin2 (x)
From the previous calculations, it follows that
sin(x) + cos(x)
g ′ (x) = −
sin(x) cos(x)

Problem 2
Question:

Evaluate the following integrals:

1.
√
Z
x
e dx

2. Z
x
dx
1 + sin(x)

3. Z
log(a2 + x2 )dx

Part 1
Consider the integral
√
Z
x
e dx

Let u2 = x. Then 2udu = dx.

Z √
Z √
x u2
e dx = e 2udu
Z
=2 ueu du

As we solved this integral in class, we have that

√ √
Z
2 ueu du = 2(ueu − eu ) = 2e x ( x − 1)

Therefore
√ √ √
Z
x x
e dx = 2e ( x − 1) + C C∈R

3
Part 2
Firstly, let us solve the integral
Z
1
dx
1 + sin(x)
We have that
1 − sin(x)
Z Z
1 1
· 1dx = dx
1 + sin(x) 1 + sin(x) 1 − sin(x)
1 − sin(x)
Z
= dx
1 − sin2 (x)
Z
1 sin(x) 1
= 2
− · dx
cos (x) cos(x) cos(x)
Z
= sec2 (x) − tan(x) sec(x)dx
Z Z
= sec2 (x)dx − tan(x) sec(x)dx

= tan(x) − sec(x) + C

Now consider the integral Z

x
dx
1 + sin(x)
1
Let u = x and dv = 1+sin(x) , then du = 1 and v = tan(x) − sec(x), as proved
previously. Therefore,
Z Z
x
dx = x(tan(x) − sec(x)) − 1 · (tan(x) − sec(x))dx
1 + sin(x)
  Z 
sin(x) 1 sin(x) sec(x) + tan(x)
=x − − − sec(x) · dx
cos(x) cos(x) cos(x) sec(x) + tan
sec2 (x) + sec(x) tan(x)
  Z 
sin(x) − 1 − sin(x)
Z
=x − − dx − dx
cos(x) cos(x) sec(x) + tan(x)
 
sin(x) − 1
=x − (− ln | cos(x)| − ln | sec(x) + tan(x)|)
cos(x)
 
sin(x) − 1
=x + ln | cos(x)(sec(x) + tan(x)|
cos(x)
 
sin(x) − 1
=x + ln |1 + sin(x)| + C C∈R
cos(x)

Part 3: Real analysis method

Consider the integral Z
log(x2 + a2 )dx

4
 −2x
Let u = log(x2 + a2 ) and dv = 1, then du = x2 +a2 and v = x Therefore,

2x2
Z Z
2 2 2 2
log(a + x )dx = x log(a + x ) − dx
x2 + a2
Z 2
x + a2 a2
= x log(a2 + x2 ) − 2 − dx
x2 + a2 x2 + a2
Z Z 1 
2 2 a
= x log(x + a ) − 2 dx − a dx
1 + ( x )2
  x a
= x log(x2 + a2 ) − 2 x − a arctan
 xa
2 2
= x log(x + a ) − 2x + 2a arctan +C C∈R
a

5
Problem 3
1. Find the Taylor Polynomial of degree 2 at a = 1 for the function
Z x2
2
f (x) = 5 + et dt
1

2. Airy functions are the solutions to the differential equation

f ′′ (x) = xf (x)

Suppose f is an Airy function with f (0) = 0 and f ′ (0) = 1. Explain

why f is at least four times differentiable and find the degree 4 Taylor
polynomial about x = 0 for f .

Part 1
We know that, for an at least n times differentiable function f : A → B, its
Taylor polynomial P with degree n at x = a ∈ A is
n
X f (i) (a)(x − a)i
Pn,a =
i=0
i!

where f 0 (a) represents the function f at x = a. Consider the function

Z x2
2
f (x) = 5 + et dt
1

2
Let h(x) = ex . Observe that h is continuous in R. Therefore, by the First
fundamental Theorem of Calculus, its primitive H is differentiable in A ⊆ R
and H ′ (c) = h(c) for all c ∈ A. As f is the sum of differentiable functions, then
f is differentiable. Therefore, by the second fundamental theorem of calculus
and the addition and chain rules of differentiation,

f (x) = 5 + H(x2 ) − H(1)

d d d
f ′ (x) = (5) + [H(x2 )] − [H(1)]
dx dx dx
d 2
= 0 + H ′ (x2 ) · (x ) − 0
dx
2 2
= e(x ) · 2x
4
= 2xex

6
Observe that f ′ is the product of differentiable functions. Then f ′ is differen-
tiable and
 
d x4
f ′′ (x) = 2 xe
dx
 4 4

= 2 ex + xex · 4x3
4
= 2ex (1 + 4x4 )

Therefore, the Taylor polynomial P of degree 2 at x = 1 of f is

f ′′ (1)(x − 1)2
P2,1 = f (1) + f ′ (1)(x − 1) +
2!
Z 1
t2 10e(x − 1)2
= (5 + e dt) + 2e(x − 1) +
1 2
2
= 5 + 0 + 2e(x − 1) + 5e(x − 1)
= e(x − 1)(5x − 3) + 5

Part 2
Consider the Airy function f such that f (0) = 0 and f ′ (0) = 1. Observe that,
as f is an Airy function, then

f ′′ (x) = xf (x)

which implies that f is twice differentiable. Furthermore, it must be true that

f ′′ (0) = 0 · f (0) = 0

To prove that f (3) exists, we have to show that

f ′′ (x + h) − f ′′ (x)
lim exists
h→0 h
Observe that
f ′′ (x + h) − f ′′ (x) (x + h)f (x + h) − xf (x)
lim = lim
h→0 h h→0 h
xf (x + h) + hf (x + h) − xf (x)
= lim
h→0 h
x(f (x + h) − f (x)) + hf (x + h)
= lim
h→0 h
f (x + h) − f (x) hf (x + h)
= lim x +
h→0 h h
f (x + h) − f (x) h
= x lim + lim · lim f (x + h)
h→0 h h→0 h h→0

7
This must be true because we are taking the limit as h → 0 of continuous
functions and their respective limits exist2 . Moreover, observing that the first
limit is exactly equal to the definition of f ′ , that h ̸= 0 and that f is continuous,
then it follows that
f ′′ (x + h) − f ′′ (x)
lim = xf ′ (x) + f (x)
h→0 h

which is a defined quantity. Therefore f (3) exists and

f (3) (x) = xf ′ (x) + f (x)

Observe that
f (3) (0) = 0 · f ′ (0) + f (0) = 0 + 0 = 0
Similarly, we can claim that f (4) exists if and only if

f (3) (x + h) − f (3) (x)

lim exists
h→0 h
Observe that
f (3) (x + h) − f (3) (x) ((x + h)f ′ (x + h) + f (x + h)) − (xf ′ (x) + f (x))
lim = lim
h→0 h h→0 h
xf ′ (x + h) + hf ′ (x + h) + f (x + h) − xf ′ (x) − f (x)
= lim
h→0 h
f ′ (x + h) − f ′ (x) f (x + h) − f (x) h ′
= lim x + + f (x + h)
h→0 h h h
f ′ (x + h) − f ′ (x) f (x + h) − f (x) h
= x lim + lim + lim · lim f ′ (x + h)
h→0 h h→0 h h→0 h h→0

Similarly as explained before, this must be true as we are taking the limit of
continuous functions and their respective limits exist. Observing that x is a
constant with respect to h, that the first and the second limit are equal to the
definition of f ′′ and f ′ respectively, and that h ̸= 0, then

f (3) (x + h) − f (3) (x)

lim = xf ′′ (x) + 2f ′ (x) = x2 f (x) + 2f ′ (x)
h→0 h
As the limit expressed above exists and it is finite, then

f (4) (x) = x2 f (x) + 2f ′ (x)

This proves that f is at least 4 times differentiable. Observe that

f (4) (0) = 02 · f (0) + 2f ′ (0) = 0 + 2 = 2

2 We can pull x out of the first limit because it is a constant with respect to h

8
Its Taylor polynomial P of degree 4 at x = 0 is equal to

f ′′ (0)(x)2 f (3) (0)(x)3 f (4) (0)(x)4

P4,0 = f (0) + f ′ (0)(x) + + +
2! 3! 4!
2x4
=0+x+0+0+
24
x3
 
=x 1+
12

9
Problem 4
Question:

Let A be an n × n matrix. Determine whether the following statements are

true or false:
1. The transpose of A has the same eigenvalues of A;

2. The transpose of A has the same eigenvectors of A;

3. If A is diagonalisable then AT is diagonalisable.

Part 1
The answer is true. We know that λ is an eigenvalue for an n × n matrix A if
and only if it is the root of the characteristic polynomial pA (λ) = det(Iλ − A).
We want to show that pA (λ) = pAT (λ). Recall the identity proved in class
det(A) = det(AT ). Let λ be an eigenvalue of A. Then
pA (λ) = det(Iλ − A)
= det((Iλ − A)T )
= det(I T λT − AT )
= det(Iλ − AT )
= pAT (λ)

Therefore λ is also an eigenvalue for AT .

Part 2
False. Let A be  
−6 12
A=
−6 11
 T
Observe that v = 3 2is an eigenvector with eigenvalue λ = 2.
      
−6 12 3 6 3
Av = = =2 = 2v
−6 11 2 4 2
 
T −6 −6
Then A = . Then v is not an eigenvector for AT .
12 11
    
T T −6 −6 3 −30
A v=A = = ̸= λv
12 11 2 58
for all λ ∈ F . This proves that the statement is not true in general.

10
Part 3
True. Suppose A is diagonalisable. Then for some diagonal matrix D and
invertible matrix P we have
A = P DP −1
Taking the transpose of both sides, it follows that

AT = (P DP −1 )T
= (P −1 )T DT P T
= (P T )−1 DP T as DT = D for diagonal matrices and (A−1 )T = (AT )−1
= QDQ−1

for a matrix Q = (P T )−1 . Therefore AT is similar to some diagonal matrix D

(an specifically, the same diagonal matrix of A). Hence AT is also diagonalisable.

11
Problem 5
Question:

Either construct a 2 × 2 matrix A with eigenvalues 2 and 3 and correspond-

ing eigenvectors [3 2]T and [4 3]T , or explain why such an A cannot exist.

Solution:
     
−6 12 3 4
The matrix A = has as eigenvectors v1 = and v2 = , with
−6 11 2 3
respective eigenvalues λ1 = 2 and λ2 = 3.
In part 2 of Problem 4 we showed that Av1 = λ1 v1 . I will report the same
calculations for the sake of marking.
      
−6 12 3 6 3
Av1 = = =2 = λ 1 v1
−6 11 2 4 2
Observe that
      
−6 12 4 12 4
Av2 = = =3 = λ 2 v2
−6 11 3 9 3
We can find the matrix A by setting up the following equality
        
a b 3 6 3a + 2b 6
= =⇒ =
c d 2 4 3c + 2d 4
and         
a b 4 12 4a + 3b 12
= =⇒ =
c d 3 9 4c + 3d 9
The equalities imply
3a + 2b = 6 (1)
4a + 3b = 12 (2)
and
3c + 2d = 4 (3)
4c + 3d = 9 (4)
Solving for a and b first, subtract (1) from (2),
a + b = 6 =⇒ a = 6 − b
Substituting this value into (1), we get
3(6 − b) + 2b = 6 =⇒ b = 12
And therefore a = 6 − 12 = −6. Following a similar procedure for c and d,
subtract (3) from (4) to get
c + d = 5 =⇒ c = 5 − d

12
Substitute this value into (3),

3(5 − d) + 2d = 4 =⇒ d = 11

and so c = 5 − 11 = −6.

13
Problem 6
Define a variation of the Fibonacci numbers by a1 = 1, a2 = 3, and an =
an−1 + an−2 , for all n ≥ 3
   
a a
1. Find a 2 × 2 matrix with A n+1 = n+2 for all n ≥ 3.
an an+1

2. Diagonalise A

3. Use this to find a formula to an

Part 1
 
1 1
The matrix A = fits our purpose. We can find it by letting
1 0
    
a b an+1 a
= n+2
c d an an+1
    
a · an+1 + b · an an+1 an+1 + an
=
c · an+1 + d · an an an+1
By inspection, we find that this is true for a = b = c = 1 and d = 0.

Part 2
√
Denote with φ the golden ratio. That is, φ = 1+2 5 . Observe the following
relations that we will be using later in the problem
√
5−1 2 1
φ−1= = √ =
2 1+ 5 φ
√
1 2 5+2−2 √
φ + = φ + (φ − 1) = 2φ − 1 = = 5
φ 2
√ !2 √
2 1+ 5 3+ 5
φ = = =φ+1
2 2
√ √
φ−3 −5 + 5 1− 5 1
√ = √ = =−
5 2 5 2 φ
√ √
2+φ 5+ 5 1+ 5
√ = √ = =φ
5 2 5 2

14
  
1 1
Now let A = . It follows that its characteristic polynomial p(λ) is
1 0

p(λ) = det(Iλ − A)
 
λ − 1 −1
= det
−1 λ
= λ2 − λ − 1

Letting p(λ) = 0, then

√
2 1± 5
λ − λ − 1 =⇒ λ =
2
Therefore λ1 = φ and λ2 = − φ1 . To find the invertible matrix P so that
A = P DP −1 , we need to find Eλi = null(Iλi − A) for each λi . Let us find the
rref(Iλi − A) to find a basis for the eigenspaces of Eλi . Firstly, consider the
eigenspace for λ1
 
φ − 1 −1
Iλ1 − A =
−1 φ
 1 
−1
= φ
−1 φ
 
1 −φ
∼
−1 φ
 
1 −φ
∼ Adding R1 to R2
0 0
  
φ
Therefore Eλ1 = Span v1 = . Now consider the eigenspace for λ2
1
 1
− φ − 1 −1

Iλ2 − A =
−1 − φ1
 
−φ −1
=
−1 − φ1
1 φ1
 
1
∼ multiply R1 by − and then add R1 to R2
0 0 φ
  1 
−φ
Therefore Eλ2 = Span v2 = . Denote with B = {v1 , v2 } the eigen-
1
basis of A. Recall
 that D is the diagonal matrix with (D)ii = λi and that
P = PB→E = v1 . . . vn , with vi ∈ B and E the standard basis. It follows
φ − φ1
     
λ 0 φ 0
that D = 1 = and that P = [v v ] =
0 λ2 0 − φ1 1 2
1 1

15
To find P −1 , consider the det(P )
 
1
det(P ) = φ − −
φ
1
=φ+
φ
√
= 5 as proved previously
   
a b −1 1 d −b
Recall that if P = , then P = det(P )
c d −c a
 1
1
Therefore P −1 = √15 φ . Hence, it follows that
−1 φ

1 φ − φ1 φ 1
  
0 1
A = P DP −1 = √ 1
φ
5 1 1 0 − φ −1 φ

Part 3
Observe that    
an an+1
A =
an−1 an
Therefore,
        
an−1 a an a
A2 = A A n−1 =A = A n+1
an−2 an−2 an−1 an
 
a =3
Using an iterative process, we apply A n − 1 times to 2 to obtain the
  a1 = 1
a
vector n+1
an    
n−1 a2 an+1
A =
a1 an
Due to diagonalisation, we can now find a formula for the term an without using
recursion:    
3 a
P Dn−1 P −1 = n+1
1 an

16
Recall that, for any diagonal matrix D, (Dn )ii = dnii and (Dn )ij = 0 for all
i ̸= j. Moreover, as we proved in class, An = P Dn P −1 . It follows that

φ − φ1 φn−1 1   
     
3 0 1 1 3
P Dn−1 P −1 = √ φ
1 1 1 0 (−1)n−1 ( φ1 )n−1 5 −1 φ 1
φ − φ1 φn−1 3 + φ1
    
0 1
= √
1 1 0 (−1)n−1 ( φ1 )n−1 5 −3 + φ
 1   n−1
  
φ −φ φ 0 1 2+φ
= √
1 1 0 (−1)n−1 ( φ1 )n−1 5 −3 + φ
 1   n−1  
φ −φ φ 0 φ
= n−1 1 n−1 1
1 1 0 (−1) (φ) −φ
φ − φ1 φn
   
=
1 1 (−1)n ( φ1 )n
+ (−1)n+1 ( φ1 )n+1
 n+1 
φ
=
φn + (−1)n ( φ1 )n
 
a
= n+1
an

All the properties of φ used above where proved in part 2 of this problem. It
follows that
 n √ !n √ !n
n n 1 1+ 5 n 5−1
an = φ + (−1) = + (−1)
φ 2 2

17