Giovanni Leoni, Principles of Real Analysis II, Fall 2012, Carnegie Mellon University

Monday, August 27, 2012

1 Real Numbers
There are two ways to introduce the real numbers. The …rst is to give them
in an axiomatic way, the second is to construct them starting from the natural
numbers. We will use the …rst method.
The real numbers are a set R with two binary operations, addition

+:R R!R
(x; y) 7! x + y

and multiplication

:R R!R
(x; y) 7! x y

and a relation such that

(A) (R; +) is an commutative group, that is,

(A1 ) for every a; b 2 R, a + b = b + a,

(A2 ) for every a; b; c 2 R, (a + b) + c = a + (b + c),
(A3 ) there exists a unique element in R, called zero and denoted 0, such
that 0 + a = a + 0 = a for every a 2 R,
(A4 ) for every a 2 R there exists a unique element in R, called the opposite
of a and denoted a, such that ( a) + a = a + ( a) = 0,

(M )

(M1 ) for every a; b 2 R, a b = b a,

(M2 ) for every a; b; c 2 R, (a b) c = a (b c),
(M3 ) there exists a unique element in R, called one and denoted 1, such
that 1 6= 0 and 1 a = a 1 = a for every a 2 R with a 6= 0,
(M4 ) for every a 2 R with a 6= 0 there exists a unique element in R, called
the inverse of a and denoted a 1 , such that a 1 a = a a 1 = 1,

(O) is a total order relation, that is,

(O1 ) for every a; b 2 R either a b or b a,

(O2 ) for every a; b; c 2 R if a b and b c, then a c,
(O3 ) for every a; b 2 R if a b and b a, then a = b,
(O4 ) for every a 2 R we have a a,

1
(AM ) for every a; b; c 2 R, a (b + c) = (a b) + (a c),
(AO) for every a; b; c 2 R if a b, a + c b + c,
(M O) for every a; b 2 R if 0 a and 0 b, then 0 a b,
(S) (supremum property)

Remark 1 Properties (A), (M ), (O), (AM ), (AO), (M O), and (S) completely
characterize the real numbers in the sense that if (R0 ; ; ; 4) satis…es the same
properties, then there exists a bijection T : R ! R0 such that T is an isomor-
phism between the two …elds, that is,

T (a + b) = T (a) T (b) ; T (a b) = T (a) T (b)

for all a; b 2 R, and a b if and only if T (a) 4 T (b). Hence, for all practical
purposes, we cannot distinguish R from R0 .

If a b and a 6= b, we write a < b.

Exercise 2 Using only the axioms (A), (M ), (O), (AO), (AM ) and (M O) of
R, prove the following properties of R:

(i) if a b = 0 then either a = 0 or b = 0,

(ii) if a 0 then a 0,
(iii) if a b and c < 0 then ac bc,
(iv) for every a 2 R we have a2 0,
(v) 1 > 0.

De…nition 3 Let E R be a nonempty set.

(i) An element L 2 R is called an upper bound of E if x L for all x 2 E;

(ii) E is said to be bounded from above if it has at least an upper bound;
(iii) if E is bounded from above, the least of all its upper bounds, if it exists, is
called the supremum of E and is denoted sup E.
(iv) E has a maximum if there exists L 2 E such that x L for all x 2 E.
We write L = max E.

We are now ready to state the supremum property.

(S) (supremum property) every nonempty set E R bounded from above

has a supremum in R.

The supremum property says that in R the supremum of a nonempty set

bounded from above always exists in R. We will see that this is not the case for
the rationals numbers.

2
Remark 4 (i) Note that if a set has a maximum L, then L is also the supre-
mum of the set.
(ii) If E R is a set bounded from below, to prove that a number L 2 R is
the supremum of E, we need to show that L is an upper bound of E, that
is, that x L for every x 2 E, and that any number s < L cannot be an
upper bound of E, that is, that there exists x 2 E such that s < x.

Wednesday, August 29, 2012

De…nition 5 Let E R be a nonempty set.

(i) An element ` 2 R is called a lower bound of E if ` x for all x 2 E;

(ii) E is said to be bounded from below if it has at least an lower bound;

(iii) if E is bounded from above, the greatest of all its lower bounds, if it exists,
is called the in…mum of E and is denoted inf E;
(iv) E has a minimum if there exists ` 2 E such that ` x for all x 2 E. We
write ` = min E.

Remark 6 (i) Note that if a set has a minimum `, then ` is also the in…mum
of the set.
(ii) If E R is a set bounded from above, to prove that a number ` 2 R is
the in…mum of E, we need to show that ` is a lower bound of E, that is,
that ` x for every x 2 E, and that any number ` < s cannot be a lower
bound of E, that is, that there exists x 2 E such that x < s.

Example 7 Consider the set

n2 + 2n
E= y2R: y= ;n2N :
n2 + 2
2
Since nn2+2n
+2 > 0, the set E is bounded from below. To see if it is bounded from
above, let’s sketch the graph of the function

x2 + 2x
f (x) =
x2 + 2
for x 1. We have

x2 + 2x x2 1 + x2
lim = lim 2
2
x!1 x + 2 x!1 x 1 + x22
1 + x2 1+ 12
1+0
= lim 2 = 2 = = 1:
x!1 1 + 2
x 1 + 12 1+0

3
Moreover,
d x2 + 2x 2 x2 + 2x + 2
f 0 (x) = = 2 0
dx x2 + 2 (x2 + 2)
p p p
for all 3+1 x 3 + 1. Hence, f is increasing in 1; 3 + 1 and
p p
decreasing for x 3 + 1. Not that 2 3 + 1 3. This implies that
4 15 15
sup E = max E = max ff (2) ; f (3)g = max ; = ;
3 11 11
while n o
inf E = min f (1) ; lim f (n) = min f1; 1g = 1;
n!1
so actually inf E = min E = f (1) = 1.
Example 8 Consider the set
xy
E= t2R: t= ; x; y 2 R; x < y :
x2 + y 2
xy
The set E is bounded since 21 x2 +y 2
1
2 for all x; y 2 R, with x < y.
Moreover, by taking x = 1 and y = 1, we get t = 21 , so
1
inf E = min E = :
2
Let’s prove that
1
sup E =
:
2
We need to show that any s < 21 is not an upper bound for the set E. If s 1
2,
then we can take x = 1 and y = 0, so that s < t = 0. Thus assume that
1 1 1
2 < s < 2 . Take the sequence xn = 1 n < yn = 1. Then
1
xn yn 1 n
tn = = 2 > s;
x2n + yn2 1 n1 +1
which gives
!
2
1 1
1 >s 1 +1 or 2sn2 2sn + s < n2 n
n n
or 0 = (1 2s) n2 + (2s 1) n s;
that is, p
1 2s + 4s2 + 1
n>
2 (1 2s)
Note that 4s2 + 1 > 0 and 1 2s > 0 for 21 < s < 12 . Hence, for all n
su¢ ciently large, tn > s, which shows that s is not an upper bound of E. Thus,
1 xy 1
2 is the supremum of the set. Note that t = x2 +y 2 = 2 only if x = y, which is
not allowed, so the set does not have a maximum.
Friday, August 31, 2012

4
2 Natural Numbers
De…nition 9 A set E R is called an inductive set if it has the following
properties

(i) the number 1 belongs to E,

(ii) if a number x belongs to E, then x + 1 also belongs to E.

Example 10 The sets [0; 1) = fx 2 R : 0 xg, [1; 1) = fx 2 R : 1 xg,

and R are all inductive sets.

De…nition 11 The set of the natural numbers N is de…ned as the intersection

of all inductive sets of R.

Note that N is nonempty, since 1 belongs to every inductive set, and so also
to N. We also de…ne
N0 = N [ f0g :

Proposition 12 The set N is an inductive set.

Proof. We already know that 1 belongs to N. If x belongs to N, then it

belongs to every inductive set E but then, since E is an inductive set, it follows
that x + 1 belongs E. Hence, x + 1 belongs to every inductive set, and so by
de…nition of N, we have that x + 1 also belongs to N.
The next result is very important.

Theorem 13 (Principle of mathematical induction) Let fpn g, n 2 N, be

a family of propositions such that

(i) p1 is true,
(ii) if pn is true for some n 2 N, then pn+1 is also true.

Then pn is true for every n 2 N.

Proof. Let E := fn 2 N such that pn is trueg. Note that E N. It follows

by (i) and (ii) that E is an inductive set, and so E contains N (since N is the
intersection of all inductive sets). Hence, E = N.
If x 2 R and n 2 N, we de…ne

xn := x x:
n tim es

If x 6= 0, we de…ne x0 := 1. We do not de…ne 00 .

The following example and exercises will be used later on.

5
Example 14 Let x 1. Let’s prove that
n
(1 + x) 1 + nx (1)
1
for every n 2 N. For n = 1, we have (1 + x) 1 + 1x, which is true. As-
sume that for some n 2 N, the inequality (1) holds. We want to prove that
n+1
(1 + x) 1 + (n + 1) x. To see this, observe that
n+1 n
(1 + x) = (1 + x) (1 + x) (1 + x) (1 + nx)
2
= 1 + (n + 1) x + nx 1 + (n + 1) x + 0 = 1 + (n + 1) x;

where in the …rst inequality we have used the fact that 1 + x 0. Hence, by the
principle of mathematical induction, the inequality (1) holds for every n 2 N.

Remark 15 For x < 1 the inequality (1) is false in general, take x = 3 and
n 2 N. Then
n n n ??
(1 3) = ( 2) = ( 1) 2n 1 3n:
n ?? ??
For n odd, ( 1) = 1, and so 2n 1 3n, or, equivalently, 2n 3n 1,which
is false for all n odd large. Take n = 4, you get 24 = 16 12 1, which is false.

Example 16 Prove that

n (n + 1)
1+ +n= (2)
2
for every n 2 N

Exercise 17 Let x 6= 1. Prove that

xn+1 1
1+x + xn =
x 1
for all n 2 N.

In what follows 0! := 1, 1! := 1 and n! := 1 2 n for all n 2 N. The

number n! is called the factorial of n. For n 2 N and k 2 N0 , we de…ne

n n!
:= :
k k! (n k)!

Theorem 18 (Binomial theorem) Let x; y 2 R n f0g and let n 2 N. Then

n
X
n n k n k
(x + y) = x y : (3)
k
k=0

6
 Proof. Step 1: Let’s prove that

n n n+1
+ = :
k 1 k k
We have
n n n! n!
+ = +
k 1 k (k 1)! (n + 1 k)! k! (n k)!
k! (n k)!n! + (k 1)! (n + 1 k)!n!
=
k! (n k)! (k 1)! (n + 1 k)!
k + (n + 1 k)
= (n! (k 1)! (n k)!)
k! (n k)! (k 1)! (n + 1 k)!
n! (n + 1) (n + 1)! n+1
= = = :
k! (n + 1 k)! k! (n + 1 k)! k

Step 2: Let’s prove (3). The proof is by induction on n. For n = 1 we have

1
X 1 k 1 k 1 0 1 1 1 0 1
x y = x y + x y = (x + y) ;
k 0 1
k=0

since 11 = 10 = 1.
Assume that the formula (3) holds for n and let’s prove it for n + 1, we have
n+1 n n n
(x + y) = (x + y) (x + y) = x (x + y) + y (x + y) :

By the induction hypothesis for n, the right-hand side of the previous identity
equals to
n
X n
X n
X n
X
n k n k n k n k n k+1 n k n k n+1 k
x x y +y x y = x y + x y
k k k k
k=0 k=0 k=0 k=0
n+1
X n
X
n n k n+1
= xl y n+1 l
+ x y k
l 1 k
l=1 k=0
n
X n
X
n 0 n+1 0 n n k n+1 n n+1 0
= x y + xl y n+1 l
+ x y k
+ x y
0 l 1 k n
l=1 k=1
n
X
n + 1 0 n+1 0 n + 1 k n+1 k n + 1 n+1 0
= x y + x y + x y
0 k n+1
k=1

n n+1
where we have set k + 1 = l and used Step 1 and the facts that 0 = 0 =1
and nn = n+1
n+1 = 1.

Remark 19 If in Theorem 13 we replace property (i) with

(i)0 if pn0 is true for some n0 2 N,

7
 then we can conclude that pn is true for all n 2 N with n n0 . To see this,
it is enough to de…ne

E := fn 2 N such that pn+n0 1 is trueg ;

which is still an inductive set.

Exercise 20 Prove that

nn > 2n n!
for all n > 6. Hint: Use the binomial theorem.

3 The Rationals Numbers and the Supremum

Property
In the previous section we have de…ned the natural numbers. Note that (N; +; ; )
does not satisfy properties (A3 ), (A4 ), and (M4 ). In particular, we cannot sub-
tract two numbers a; b 2 N unless, a b + 1. For this reason, we de…ne the set
of integers Z as follows

Z := f n : n 2 Ng [ f0g :

Now (Z; +; ; ) satis…es properties (A3 ), (A4 ), but not (M4 ). To resolve this
issue, we introduce the set of rational numbers Q de…ned by

p
Q := : p; q 2 Z; q 6= 0 ;
q

where pq := p q 1 . Then (Q; +; ; ) satis…es properties (A), (M ), (O), (AM ),

(AO), (M O). So, what’s wrong?

Theorem 21 There does not exist a rational number r such that r2 = 2.

Proof. The proof is by contradiction. Assume that there exists r 2 Q such

that r2 = 2. Write r = pq , where p; q 2 Z and q 6= 0. Then
2
p
= 2:
q
By dividing by common factors, without loss of generality, we may suppose that
p and q have no common integral factors other that 1. Since p2 = 2q 2 , it follows
that p2 is an even number. We claim that p is even. Indeed, if p = 2k + 1 for
some k 2 Z, then
2
p2 = (2k + 1) = 2 2k 2 + k + 1;
which is an odd number, which is a contradiction. Hence, p = 2k, for some
k 2 Z. But then
4k 2 = p2 = 2q 2 ;

8
which implies that 2k 2 = q 2 . Hence, q 2 is even, and so reasoning as before, we
conclude that q must be even. This contradicts the fact that p and q have no
common integral factors (they are both divisible by 2). p
Thus in the set of rational numbers the square root r is not de…ned, in
general.
Monday, September 3, 2012
Memorial day, no classes
Wednesday, September 5, 2012

Theorem 22 The rational numbers do not satisfy the supremum property.

Proof. We need a nonempty set E Q bounded from above but for which
there exists no supremum in Q. De…ne

E := x 2 Q : 0 < x and x2 < 2 :

Then E is nonempty, since 1 2 E. Moreover, E is bounded from above, since 2

is an upper bound.
Assume by contradiction that there exists L 2 Q such that L = sup E. It
cannot be L 0, since 1 2 E and 1 > 0. Hence, L > 0. Let’s prove that it
cannot be L2 < 2. Choose n 2 N so large that n > 22L+1
L2 (we will see later on
that this can be done). Then
2
1 1 2L 1 2L 2L + 1
L+ = L2 + + < L2 + + = L2 + < 2;
n n2 n n n n

by the choice of n. Hence, L + n1 belongs to E, which contradicts the fact that

L is an upper bound of E. Similarly, taking L n1 , for n large, we can show
2
that L n1 > 2 and L n1 > 0. Let’s prove that L n1 is an upper bound of
1 2
E. If x 2 E, then x > 0 and x2 < 2 < L n . Hence,

1 1
0< L +x L x :
n n

Since L n1 > 0 and x > 0, we have that L n1 + x > 0. It follows from

the previous inequality that 0 < L n1 x , that is x < L n1 . Hence,
1
L n is an upper bound of E, which contradicts the fact that L is the least
upper bound of E. Hence, it cannot be L2 > 2, Thus, L2 = 2, which is again a
contradiction by Theorem 21.
The set R n Q is called the set of irrational numbers.

Theorem 23 The set of irrational numbers is nonempty.

Proof. Take
E := x 2 R : 0 < x and x2 < 2 :
Exactly as in the previous proof, we have that E is nonempty and bounded
from above. Hence, by the supremum property there exists L 2 R such that

9
L = sup E. It follows as in the previous proof that L2 = 2, and so L belongs to
R n Q. p
The number L is denoted 2 and called square root of 2. Similarly, for every
n 2 N with n even and every x 2 R with x 0, we can show that there exists a
unique y 2 R with y 0 such that xn = y. On the other hand, for every n 2 N
with n odd and every x 2 R, we can show that there exists a unique y 2 R such
that xn = y. p
The number y is denoted n x and called n-th root of x.

Exercise 24 (The n-th root of a) Given a > 0 and n 2 N, we want to de…ne

the n-th root of a.

(i) Prove that the set

E := fx 2 R : x 0g [ fx 2 R : x > 0 and xn < ag

is nonempty and bounded from above.

(ii) Let L = sup E. Prove that Ln = a. Hint: Use the binomial theorem.
(iii) Prove that for every x; y 2 R, x 6= 0, y 6= 0, and n 2 N,

xn y n = (x y) xn 1
+ xn 2
y+ + xy n 2
+ yn 1
:

(iii) Prove that L is the only positive solution of the equation xn = a.

In Theorem 22 we have used the fact that there exist arbitrarily large natural
numbers. This follows from the Archimedean Property.

Proposition 25 (Archimedean Property) If a; b 2 R with a > 0, then there

exists n 2 N such that na > b.

Proof. If b 0, then n = 1 will do. Thus, assume that b > 0. Assume by

contradiction that na b for all n 2 N and de…ne the set

E = fna : n 2 Ng :

Then the set E is nonempty and has an upper bound, b. By the supremum
property, there exists L = sup E. Hence, for every m 2 N, we have that
(m + 1) a L, or, equivalently, ma L a for all m 2 N. But this shows that
L a is an upper bound of E, which contradicts the fact that L is the least
upper bound.

Exercise 26 Prove that every nonempty subset of the integers bounded from
below has a minimum.

The next result is left as an exercise.

Theorem 27 (The integer part) Given a real number x 2 R, there exists an

integer k 2 Z such that k x < k + 1.

10
De…nition 28 Given a real number x 2 R, the integer k given by the previous
corollary is called the integer part of x and is denoted bxc. The number x bxc
is called the fractional part of x and is denoted fr x (or fxg). Note that 0
fr x < 1.

Friday, September 7, 2012

Corollary 29 (Density of the rationals) If a; b 2 R with a < b, then there

exists r 2 Q such that a < r < b.

Proof. We want to …nd r 2 Q such that a < r < b. Any r 2 Q can be

written as r = pq , where p 2 Z and q 2 N. So we want to …nd p 2 Z and q 2 N
such that
p
a < < b;
q
or, equivalently,
qa < p < qb:
Suppose we have chosen q. Then to …nd p, we apply the previous corollary to
obtain an integer p 2 Z such that

p 1 qa < p: (4)

We would like p < qb. By (4),

?
p 1 + qa < qb;

provided b 1 a < q. Hence, we …rst use the Archimedean property (applied with
1 and b 1 a in place of a and b) to …nd q 2 N such that b 1 a < q and then …nd p
as above, namely, p = bqac + 1.

Corollary 30 (Density of the irrationals) If a; b 2 R with a < b, then there

exists x 2 R n Q such that a < x < b.
p p
Proof. Since a < b, we have that p2a < 2b. p By the density of the
rationals, there exists r 2 Q such that 2a < r < 2b. Without loss of
generality, we may assume that r 6= 0 (why?). Hence, a < pr2 < b. Since pr2 is
irrational (why?), the result is proved.

4 Topological Properties of the Real Line

Given a number x 2 R, the absolute value of x is the number

+x if x 0;
jxj :=
x if x < 0:

The absolute value satis…es the following properties, which are left as as exercise.

11
Theorem 31 Let x; y; z 2 R. Then the following properties hold.

(i) jxj 0 for all x 2 R, with jxj = 0 if and only if x = 0,

(ii) j xj = jxj for all x 2 R,
(iii) if y 0 and x 2 R, then jxj y if and only if y x y,
(iv) jxj x jxj for all x 2 R,
(iii) jxyj = jxj jyj for all x; y 2 R,
(iv) jx + yj jxj + jyj for all x; y 2 R.

Given r > 0 and x0 2 R, the ball of center x0 and radius r is the set

B (x0 ; r) := fx 2 R : jx0 xj < rg :

A subset U R is open if for every x 2 U there exists r > 0 such that

B (x; r) U .

Example 32 Some simple examples of sets that are open and of some that are
not.

(i) The set (a; 1) = fx 2 R : x > ag is open. Indeed, if x > a, take r :=

x a > 0. Then B (x; r) (a; 1). Similarly, the set ( 1; a) is open.
(ii) The set (a; b) = fx 2 R : a < x < bg is open. Indeed, given a < x < b,
take r := min fb x; x ag > 0. Then B (x; r) (a; b).
(iii) The set (a; b] = fx 2 R : a < x bg is not open, since b belongs to the set
but there is no ball B (b; r) contained in (a; b].

Example 33 Consider the set

1
E =Rn : n2N :
n
Let’s prove that E is not open. The point x = 0 belongs to E, but for every
r > 0, by the Archimedean principle we can …nd n 2 N such that n > 1r , and
so 0 < n1 < r, which shows that n1 2 ( r; r). Since n1 does not belong to E, the
ball ( r; r) is not contained in E for any r > 0. Hence, E is not open.

Example 34 Consider the set

1
U = R n f0g [ : n2N :
n

Let’s prove that U is open. If x < 0, take r = x > 0, then B (x; r) = ( 2x; 0)
1
U . If x > 1, take r = x 1, then B (x; r) = (1; 2x 1) U . If n+1 < x < n1 ,
n o
take r = min n1 x; x n+1 1 1
= n+1 , then B (x; r) U . Hence, U is open.

12
 Monday, September 10, 2012

The main properties of open sets are given in the next proposition.

Proposition 35 The following properties hold:

(i) ; and R are open.

(ii) If Ui R, i = 1; : : : ; n, is a …nite family of open sets of R, then U1 \ \Un
is open.
S
(iii) If fUi gi2I is an arbitrary collection of open sets of R, then i2I Ui is
open.

Proof. To prove (ii), let x 2 U1 \ \ UM . Then x 2 Ui for every

i = 1; : : : ; n, and since Ui is open, there exists ri > 0 such that B (x; ri ) Ui .
Take r := min fr1 ; : : : ; rn g > 0. Then

B (x; r) U1 \ \ Un ;

which shows that U1 \ \ Un is

S open.
To prove (iii), let x 2 U := i2I Ui . Then there is i 2 I such that x 2 Ui
and since Ui is open, there exists r > 0 such that B (x; r) Ui U . This shows
that U is open.
Properties (i)–(iii) are used to de…ne topological spaces.

De…nition 36 Let X be a nonempty set and let be a family of sets of X. The

pair (X; ) is called a topological space if the following hold.

(i) ;; X 2 .
(ii) If Ui 2 for i = 1; : : : ; M , then U1 \ : : : \ UM 2 .
S
(iii) If fUi gi2I is an arbitrary collection of elements of , then i2I Ui 2 .

Remark 37 The intersection of in…nitely many open sets is not open in gen-
1 1
eral. Take Un := n ; n for n 2 N. Then

1
\ 1 1
; = f0g ;
n=1
n n

but f0g is not open. Indeed, for every r > 0, the ball ( r; r) is not contained in
f0g.

Given a set E R, a point x 2 E is called an interior point of E if there

exists r > 0 such that B (x; r) E. The interior E of a set E R is the
union of all its interior points.
The proof of following proposition is left as an exercise.

13
Proposition 38 Let E R. Then
(i) E is an open subset of E,
(ii) E is given by the union of all open subsets contained in E; that is, E is
the largest (in the sense of union) open set contained in E,
(iii) E is open if and only if E = E ,
(iv) (E ) = E .
Example 39 Consider the set E = [0; 1). Then 0 is not an in interior point
of E, so E (0; 1). On the other hand, since (0; 1) is open and contained
in E, by part (ii) of the previous proposition, E (0; 1), which shows that
E = (0; 1).
Exercise 40 Some properties of the interior.
(i) Prove that if E; F are subsets of R, then
E \ F = (E \ F ) ;
E [F (E [ F ) :

(ii) Show that in general E [ F 6= (E [ F ) .

(iii) Let fEi gi2I be an arbitrary collection of sets of R. What is the relation,
T T S
if any, between i2I (Ui ) and i2I Ui ? And between i2I (Ui ) and
S
i2I Ui ?

A subset C R is closed if its complement R n C is open.

The main properties of closed sets are given in the next proposition.
Proposition 41 The following properties hold:
(i) ; and R are closed.
(ii) If Ci R, i = 1; : : : ; n, is a …nite family of closed sets of R, then C1 [
[ Cn is closed.
T
(iii) If fCi gi2I is an arbitrary collection of closed sets of R, then i2I Ci is
closed.
Wednesday, September 12, 2012
The proof follows from Proposition 35 and De Morgan’s laws. If fEi gi2I is
an arbitrary collection of subsets of a set R, then De Morgan’s laws are
!
[ \
Rn Ei = (R n Ei ) ;
i2I i2I
!
\ [
Rn Ei = (R n Ei ) :
i2I i2I

14
Next we present a very useful proposition for identifying open an closed sets.
We will prove it later on when we discuss about continuous functions.
Proposition 42 Let D R and let f : D ! R be continuous.
1
(i) If D is open, then f (U ) is open for every open set U R.
1
(ii) If D is closed, then f (C) is closed for every closed set C R.
Example 43 Consider the set
1
E= x 2 R : sin x > ; cos x < 1 :
2
To see if this is open or closed, let’s rewrite E as follows
1
E= x 2 R : sin x > \ fx 2 R : cos x < 1g :
2
The function f (x) = sin x is continuous and de…ned in R, which is open. Note
that
1 1
x 2 R : sin x > =f 1 ;1
2 2
and since 21 ; 1 is open, by the previous proposition f 1 1
2; 1 is open.
Similarly, setting g (x) = cos x, we have that the set
1
fx 2 R : cos x < 1g = g (( 1; 1))
is open. Thus E is open, since intersection of two open sets.
Remark 44 Note that the majority of sets are neither open nor closed. The
set E = (0; 1] is neither open nor closed.
Given a set E R, the closure of E, denoted E, is the intersection of all
closed sets that contain E; in other words, the closure of E is the smallest (with
respect to inclusion) closed set that contains E. It follows by Proposition 41
that E is closed.
The proof of following proposition is left as an exercise.
Proposition 45 Let C R. Then C is closed if and only if C = C.
Proposition 46 Let E R, and let x 2 R. Then x 2 E if and only if
B (x; r) \ E 6= ; for every r > 0.
Proof. Let x 2 E and assume by contradiction that there exists r > 0 such
that B (x; r) \ E = ;. Since B (x; r) is open and B (x; r) \ E = ;, it follows
that R n B (x; r) is closed and contains E. By the de…nition of E we have that
E R n B (x; r), which contradicts the fact that x 2 E.
Conversely, let x 2 R and assume that B (x; r) \ E 6= ; for every r > 0. We
claim that x 2 E. Indeed, if not, then x 2 R n E, which is open. Thus, there
exists B (x; r) R n E, which contradicts the fact that B (x; r) \ E 6= ;.
The previous proposition leads us to the de…nition of accumulation points.

15
De…nition 47 Given a set E R, a point x 2 R is an accumulation point, or
cluster point of E if for every r > 0 the ball B (x; r) contains at least one point
of E di¤ erent from x.

Note that x does not necessarily belong to the set E.

Example 48 Consider the set

E := f2g :

Then 2 is not an accumulation points of E. Indeed, for every r > 0, the ball
B (2; r) intersects E only at the point 2, and so in B (2; r) there are no point of
E di¤ erent from 2. If x 6= 2, then x is not an accumulation point of E, since
taking r = j2 xj > 0, we have that the ball B (x; j2 xj) does not intersect E.

Example 49 Consider the set

1 1
E := [ 1+ :
n n2N n n2N

We want to prove that 0 and 1 are accumulation points of E. Note that 0 2 = E,

while 1 2 E (so accumulation points may or may not be in the set E). For
r > 0, by taking a natural number n > 1r , we have that 0 < n1 < r, and so
1 1
n 2 B (0; r) \ E (of course n 6= 0). This shows that 0 is an accumulation points
of E.
Similarly, for r > 0, by taking a natural number n > 1r , we have that 0 <
1 1 1
n < r, and so 1 < 1 + n < 1 + r, which shows that 1 + n 2 B (1; r) \ E (of
1
course 1 + n 6= 1). This shows that 1 is an accumulation points of E. Next we
show that there are no other accumulation points of E.
Indeed, if x < 0, take r = x > 0, then B (x; r) = ( 2x; 0) does not intersect
E. If x > 2, take r = x 1, then B (x; n r) = (1; 2x 1) o does not intersect E.
1 1 1 1
If n+1 < x < n, take r = min n x; x n+1 , then B (x; r) does not
n o
1 1 1 1 1
intersect E. If x = n, with n > 1, take r = min n ;
n+1 n 1 n . Then
1
B (x; r) intersects E only in n. Hence, U is nopen. o
1 1
If 1 + n+1 < x < 1+ n, take r = min 1 + n1 x; x 1
1 + n+1 , then
n o
B (x; r) does not intersect E. If x = 1 + n1 , take r = min n1 1 1
n+1 ; n 1
1
n .
Then B (x; r) intersects E only in 1 + n1 .

The set of all accumulation points of E is denoted acc E.

Remark 50 Note take if x 2 R is an accumulation point of E, then by taking

r = n1 , n 2 N, there exists a sequence fxn g E with xn 6= x for all n 2 N such
that jxn xj < n1 ! 0. Thus fxn g converges to x. Conversely, if there exists a
sequence fxn g E with xn 6= x for all n 2 N such that jxn xj ! 0, then x is
an accumulation point of E.

16
 Friday, September 14, 2012
It turns out that the closure of a set is given by the set and all its accumu-
lations points.

Proposition 51 Let E R Then

E = E [ acc E:

In particular, a set C R is closed if and only if C contains all its accumulation

points.

Proof. Let x 2 E and assume by contradiction that x 2 = E [ acc E. Since

x2 = acc E, then there exists a ball B (x; r) that contains no other point of E
other than x, but since x 2
= E, it follows that B (x; r) R n E. This contradicts
Proposition 46.
Conversely, let x 2 E [ acc E. If x 2 E, then since E E, there is nothing
to prove. If x 2 acc E, then the result follows from Proposition 46.

Exercise 52 (i) Prove that if E1 ; : : : ; En are subsets of R, then

E1 \ \ En E1 \ \ En ;
E1 [ [ En = E1 [ [ En :

(ii) Show that in general E1 \ \ En 6= E1 \ \ En .

(iii) Let fEi gi2I be an arbitrary collection of sets of R. What is the relation,
T T S
if any, between between i2I Ui and i2I Ui ? And between i2I Ui and
S
i2I Ui ?

Theorem 53 (Bolzano–Weierstrass) Every bounded set E R with in…-

nitely many elements has at least one accumulation point.

Proof. Since E is bounded, let b 2 R be its supremum and a 2 R be its

in…mum. Then E [a; b]. Divide [a; b] into two closed intervals of equal length,
a; a+b2 and a+b 2 ; b . Since E has in…nitely many elements, at least one of
these two closed intervals contains in…nitely many elements of E. Let’s call this
closed interval I1 = [a1 ; b1 ] (if both interval do the job, just pick one). Then
[a1 ; b1 ] [a; b], b1 a1 = b 2 a and I1 contains in…nitely many elements of E.
Divide [a1 ; b1 ] into two closed intervals of equal length. Since E has in…nitely
many elements, at least one of these two closed intervals contains in…nitely many
elements of E. Let’s call this closed interval I2 = [a2 ; b2 ] (if both interval do
the job, just pick one). Then [a2 ; b2 ] [a1 ; b1 ], b2 a2 = b22a and I2 contains
in…nitely many elements of E. By induction, we construct a sequence of intervals
[an ; bn ], n 2 N, with

[an+1 ; bn+1 ] [an ; bn ] [a1 ; b1 ] [a; b] ;

17
 b a
and for every n 2 N, bn an = 2n and [an ; bn ] contains in…nitely many elements
of E. Note that
a a1 an an+1 ; (5)
b b1 bn bn+1 : (6)
Let
A := fa1 ; : : : ; an ; : : :g ;
B := fb1 ; : : : ; bn ; : : :g :
We claim that sup A inf B. To see this, let n; m 2 N and …nd k m; n. Then
by (5) and (6),
an ak bk bm :
Hence, an bm for all n; m 2 N. Taking the supremum over all n 2 N, we get
that
sup A = sup an bm
n2N
for all m 2 N. Taking the in…mum over all m 2 N, we get that sup A inf B.
Next we claim that sup A = inf B. Since sup A an for all n 2 N and
inf B bn for all n 2 N, we have that
b a
0 t := inf B sup A bn an =
2n
for all n 2 N. If t > 0, by (1), we have that
b a
1+n 2n
t
for all n 2 N, which contradicts the Archimedean property. This shows that
x0 := sup A = inf B:
Finally, we prove that x0 is an accumulation point of E. Fix r > 0 and consider
the ball B (x0 ; r). Since x0 r < x0 = sup A, we have that x0 r is not an
upper bound of A, and so there exists an such that
x0 r < an :
On the other hand, since x0 = inf B < x0 + r, we have that x0 + r is not a lower
bound of B, and so there exists bm such that
bm < x0 + r:
Let k m; n. Then by (5) and (6),
x0 r < an ak bk bm < x0 + r:
This shows that [ak ; bk ] B (x0 ; r). Since [ak ; bk ] contains in…nitely many
elements of E, the same holds for B (x0 ; r) and so x0 is an accumulation point
of E.
Monday, September 17, 2012

18
De…nition 54 Given a set E R, a point x 2 R is a boundary point of E if
for every r > 0 the ball B (x; r) contains at least one point of E and one point
of R n E. The set of boundary points of E is denoted @E.

The following theorem is left as an exercise.

Theorem 55 Let E R. Then

(i) E = E [ @E,
(ii) E is closed if and only if it contains all its boundary points,
(iii) @E = @ (R n E),
(iv) @E = (R n E) \ E.

5 Sequences
De…nition 56 A sequence of real numbers is a function

f :N!R
n 7! f (n)

from the natural numbers to the real numbers.

If f (n) = xn for n = 1; 2; : : :, usually we denote the sequence f by the

symbol fxn g or fxn gn2N or x1 ; x2 ; : : :.

De…nition 57 We say that a sequence fxn g of real numbers converges to a

number ` 2 R if for every " > 0 there exists an integer N = N (") such that

jxn `j "

for all n N . In this case we say that ` is the limit of fxn g and we write

lim xn = ` or xn ! `:
n!1

Let’s see some examples

n2 +sin n
Example 58 Consider the limit limn!1 n2 +2n 1 . We have

n2 + sin n n2 1 + sin
n2
n
1 + sin
n2
n
1+0
lim = lim = lim 1 = 1+0 = 1:
n!1 n2 + 2n 1 n!1 n2 1 + n2 1 n!1 1 + 2 0
n2 n n2

Let’s prove it using the previous de…nition. Fix " > 0, we want to …nd N =
N (") 2 N such that
n2 + sin n
1 "
n2 + 2n 1

19
for all n N . We have

n2 + sin n n2 + sin n n2 2n + 1
1 =
n2 + 2n 1 n2 + 2n 1
2n 1 sin n 4n 4
= 2 2
= "
n + 2n 1 n +0 n

for all n 4" , where we have used the fact that 2n 2 0. It is enough to take
N = 4" + 1.

Wednesday, September 19, 2012

Next we discuss some important limits.

Example 59 Let x > 0. We want to calculate

p
lim n x:
n!1
p
Consider …rst the case x > 1. Then n x > 1 and so we may write
p
n
x = 1 + an ;

where an > 0. By inequality (1),

n
x = (1 + an ) 1 + nan ;

which implies that

1 x
0 < an :
n
This implies that an ! 0 as n ! 1. Indeed, given " > 0, we have that
x 1 x 1 x 1
n p " for all n n " . It is enough to take N = " + 1. In
turn, x = 1 + an ! 1 + 0, which shows that
n

p
lim n x = 1
n!1
p
n
for x > 1. If x = 1, then 1 = 1 ! 1 as n !p1.
Consider next the case 0 < x < 1. Then n x < 1 and so we may write
p
n
1
x= ;
1 + an
1
where an > 0. Hence, x = (1+an )n . By inequality (1),
n
(1 + an ) 1 + nan ;

which implies that

1 1
x= n ;
(1 + an ) 1 + nan

20
and so
1
1 x
0 < an :
n
1
1
This implies that an ! 0 as n ! 1. Indeed, given " > 0, we have that n
x
"
1 x 1 p 1 1
for all n " . In turn, x = 1+an ! 1+0 , which shows that
n

p
n
lim x = 1:
n!1

De…nition 60 We say that a sequence fxn g of real numbers diverges to plus

in…nity if for any real number M > 0 there exists an integer N = N (M ) such
that for all n N we have
xn > M: (7)
In this case we write

lim xn = 1 or xn ! 1.
n!1

Similarly we say that a sequence fxn g of real numbers diverges to minus in…nity
if for any real number M > 0 there exists an integer N = N (M ) such that for
all n N we have
xn < M: (8)
In this case we write

lim xn = 1 or xn ! 1.
n!1

If fxn g does not converge and does not diverge to plus in…nity or to minus
in…nity, we say that it oscillates.

De…nition 61 Given a sequence fxn gn2N of real numbers and a strictly in-
creasing sequence of natural numbers fnk gk2N (that is n1 < n2 < : : :) the se-
quence fxnk gk2N is called a subsequence of fxn g.

Theorem 62 A sequence fxn g of real numbers converges to ` if and only if

every subsequence fxnk g converges to `.

We will prove this theorem later on when we study the liminf and limsup of
a sequence.

Example 63 This theorem is very useful to show that a sequence diverges. It

is enough to …nd two subsequences which converge to di¤ erent numbers. As an
n
example, the sequence f( 1) g is divergent for both 1, 1,. . . and 1, 1, . . .
are subsequences and converge to di¤ erent limits.

Exercise 64 Let 0 < < 1 be a rational number. Show that the sequence
fsin (n )g oscillates.

21
Example 65 If x 2 R then
8
>
> 1 if x > 1;
<
1 if x = 1;
lim xn =
n!1 > 0
> if 1 < x < 1;
:
oscillates if x 1:

If x > 1, then we can write x = 1 + a, where a > 0. By inequality (1),

n
xn = (1 + a) 1 + na:

Hence, for any M > 0, by taking n Ma 1 , we have that xn M , which implies

that limn!1 xn = 1.
If x = 1, then xn = 1n = 1 ! 1 as n ! 1. If 0 < x < 1, then we can write
1
x = 1+a , where a > 0, and so again by inequality (1),

1 1
0 xn = n :
(1 + a) 1 + na
1 1
Hence, given " > 0, by taking n a " 1 , we have that 0 xn ", which
n
implies that limn!1 x = 0. The case 1 < x < 0 is similar and is left as an
exercise.
If x = 1, we have already seen that the limit does not exist. If x < 1,
then for n = 2k, we have that
2k 2k
xn = (x) = ( x) !1

as k ! 1, since x > 1. If instead we take n = 2k + 1, then

2k+1 2k+1
xn = (x) = ( x) ! 1

as k ! 1, since x > 1. Thus, the limit limn!1 xn does not exists.

Exercise 66 Prove that p

n
lim n = 1:
n!1

Exercise 67 Prove that

8
n < 0 if jxj > 1;
lim n = 1 if 0 < x 1; :
n!1 x :
does not exist if 1 x < 0:

Friday, September 21, 2012

De…nition 68 We say that a sequence fxn g of real numbers is

(i) bounded from above if there exists M 2 R such that xn M for all n 2 N;
(ii) bounded from below if there exists m 2 R such that xn m for all n 2 N;

22
(iii) bounded if it is bounded from above and from below.

Theorem 69 Let fxn g R be a sequence.

(i) If there exists limn!1 xn = ` 2 R, then fxn g is bounded.

(ii) If the limit of fxn g exists, it is unique.

Proof. (i) Let " = 1, then there exists an integer N = N (1) such that

jxn `j 1

for all n N . Thus, jxn j = jxn `j jxn `j + j`j 1 + j`j. Hence, it su¢ ces
to take
M = max fjx1 j ; : : : ; jxN j ; 1 + j`jg :
(ii) Let limn!1 xn = ` and limn!1 xn = L, with L 6= `. If one of the two is a
real number, say ` 2 R, then the sequence is bounded by part (i), and so also L
must be a real number. Let 0 < " < 21 jL `j. By the de…nition of limit, there
exist an integer N1 = N1 (") such that

jxn `j "

for all n N1 and an integer N2 = N2 (") such that

jxn Lj "

for all n N2 . Then for n max fN1 ; N2 g, we have that

jL `j = jL xn `j jxn `j + jxn Lj " + " < jL `j ;

which is a contradiction. It remains the case in which one limit is +1 and the
other 1, which again gives a contradiction (take M = 1 in (7) and (8)).

Theorem 70 Let fxn g and fyn g be two sequences of real numbers such that
there exist
lim xn = `1 2 R and lim yn = `2 2 R:
n!1 n!1

Then

(i) there exists lim (xn + yn ) = `1 + `2 ,

n!1

(ii) there exists lim xn yn = `1 `2 ,

n!1

xn `1
(iii) if yn 6= 0 for all n 2 N and `2 6= 0 then there exists lim = .
n!1 yn `2

23
 Proof. (i) and (ii) are left as an exercise. (iii) Write

xn `1 xn `2 yn `1 xn `2 `1 `2 yn `1
= =
yn `2 yn `2 yn `2
1
= j`2 (xn `1 ) + `1 (yn `2 )j
jyn j j`2 j
1 1 j`1 j
jxn `1 j + jyn `2 j :
jyn j jyn j j`2 j
1
Thus we need to bound jyn j from above, or, equivalently, we need jyn j to stay
away from zero. Since `2 6= 0, taking " = j`22 j > 0, there exist an integer
N1 = N1 (") such that
j`2 j
jyn `2 j
2
for all n N1 . Hence,

j`2 j j`2 j
jyn j = jyn `2 j j`2 j jyn `2 j j`2 j =
2 2
for all n N1 . It follows that

xn `1 1 1 j`1 j
jxn `1 j + jyn `2 j
yn `2 jyn j jyn j j`2 j
2 2 j`1 j
jxn `1 j + jyn `2 j :
j`2 j j`2 j j`2 j

Given " > 0 there exist an integer N2 = N2 (") such that

" j`2 j
jxn `1 j
4
for all n N2 and an integer N3 = N3 (") such that
2
" j`2 j
jyn `2 j
4 (1 + j`1 j)

for all n N3 . Then for n max fN1 ; N2 ; N3 g, we have that

xn `1 2 2 j`1 j
jxn `1 j + jyn `2 j
yn `2 j`2 j j`2 j j`2 j
" " j`1 j " "
+ + 1 = ":
2 2 1 + j`1 j 2 2

This completes the proof.

The following theorem is left as an exercise.

Theorem 71 Let fxn g and fyn g be two sequences of real numbers.

24
 (i) If there exists lim xn = 1 and fyn g is bounded from below, then there
n!1
exists lim (xn + yn ) = 1.
n!1

(ii) If there exists lim xn = 1 and fyn g is bounded from above, then there
n!1
exists lim (xn + yn ) = 1.
n!1

(iii) If there exist lim xn = 1 and limn!1 yn = ` 2 [ 1; 1] with ` 6= 0,

n!1
then there exists lim (xn yn ) = (sgn `) 1.
n!1

1
(iv) If there exist lim xn = 1, then there exists lim = 0.
n!1 n!1 xn

(v) If xn 6= 0 for all n 2 N su¢ ciently large and there exist lim xn = 0, then
n!1
.
8
1 < +1 if xn > 0 for all n su¢ ciently large,
lim = 1 if xn < 0 for all n su¢ ciently large,
n!1 xn :
does not exist otherwise.

Remark 72 (Important) In any of the following cases, nothing can be said

without further investigation.

(i) If there exist lim xn = 1 and limn!1 yn = 1, then the limit lim (xn + yn )
n!1 n!1
needs further investigation.

(ii) If there exist lim xn = 1 (or 1) and limn!1 yn = 0, then the limit
n!1
lim (xn yn ) needs further investigation.
n!1

(iii) If there exist lim xn = 1 (or 1) and limn!1 yn = 1, then the limit
n!1
xn
lim (yn ) needs further investigation.
n!1

(iv) If yn 6= 0 for all n 2 N and there exist lim xn = 1 (or 1) and

n!1
xn
limn!1 yn = 1 (or 1), then the limit lim needs further investi-
n!1 yn
gation.
(v) If yn 6= 0 for all n 2 N and there exist lim xn = 0 and limn!1 yn = 0,
n!1
xn
then the limit lim needs further investigation.
n!1 yn

Example 73 An important limit is

n
1
lim 1+ = e:
n!1 n

We will prove it later (see Example ??).

25
 Monday, September 24, 2012
The next theorem gives the precise speed at which n! goes to in…nity.

Theorem 74 (Stirling’s Formula) The following holds

n!
lim p = 1:
n!1 nn e n 2 n
Second proof. Consider the function f (x) = log x, for x > 0. Then
f 0 (x) = x1 and f 00 (x) = x12 < 0, so the function f is convex, and the slope
of f is positive and decreasing. Hence, the area of f in the interval [n; n + 1]
is greater than the area of the trapezoid of vertices, A = (n; 0), D = (n + 1; 0),
B = (n; log n), C = (n + 1; log (n + 1)), that is,
Z n+1
1
[log n + log (n + 1)] log x dx: (9)
2 n

Consider the line tangent to the curve at the point B. Its equation is
1
y log n = (x n) :
n
This line intersects the vertical line x = n+1 at the point G = n + 1; log n + n1 .
Since the slope of f is increasing, he area of f in the interval [n; n + 1] is less
than the area of the trapezoid ABGD, that is,
Z n+1
1 1
log x dx log n + log n + : (10)
n 2 n
Similarly, consider the line tangent to the curve at the point C. Its equation is
1
y log (n + 1) = (x n 1) :
n+1
1
This line intersects the vertical line x = n at the point E = n; log (n + 1) n+1 .
Since the slope of f is increasing, the area of f in the interval [n; n + 1] is less
than the area of the trapezoid AECD, that is,
Z n+1
1 1
log x dx log (n + 1) + log (n + 1) : (11)
n 2 n + 1
Summing the inequalities and dividing by 2 gives
Z n+1
1 1 1
log x dx 2 log n + 2 log (n + 1) + :
n 4 n n+1

Hence, also by (9),

Z n+1
1 1 1 1
[log n + log (n + 1)] log x dx log n + log (n + 1) + :
2 n 2 2n 2 (n + 1)

26
Summing between 1 and m 1 gives
m
X1 1
1
log m! log m = [log n + log (n + 1)]
2 n=1
2
Z m
m
log x dx = [x (log x 1)]1 = m log m m+1
1
m
X1 1 1 1
log n + log (n + 1) +
n=1
2 2n 2 (n + 1)
1 1 1
= log m! log m + :
2 4 4m
Hence,
Z m 1
1
0 cm := log x dx log m! log m
1 2
1 1 1
= m+ log m log m! m+1 :
2 4 4m

Since cm is given by the di¤erence between the area of f between x = 1 and

x = m and the trapezoidal approximation to this area, we have that the sequence
fcm g is increasing. Hence, by a theorem that we will prove later on, there exists

1
lim cm = c 2 0; :
m!1 4

Raising everything to the power e we have that

p
mm me m
= e(m+ 2 ) log m
1
cm log m! m
e e= e ! ec :
m!
That is,
m!
lim p = e1 c
=: `: (12)
m!1 mm e m m
To …nd the exact value of c we use your homework. Let
Z =2
n
In := (sin x) dx
0

Since in 0; 2 ,
n+1 n
(sin x) (sin x) ;
we have
I2k+1 < I2k < I2k 1:

Hence,
I2k I2k 1 2k + 1
1 = ;
I2k+1 I2k+1 2k

27
where we have used the formula nIn = (n 1) In 2. It follows by the squeeze
theorem that
I2k
lim = 1:
k!1 I2k+1

On the other hand, by your homework again,

k
Y 2i 1
I2k = ;
2 i=1
2i
k
Y 2i
I2k+1 = ;
i=1
2i + 1

and so
Qk k
I2k 2i 1 Y (2i 1) (2i + 1)
= Qki=1
2 2i
= 2
I2k+1 2i
i=1 2i+1
2 i=1 (2i)
2
[(2k)!] (2k + 1)
= 4 ! 1:
2 (2k k!)

Using (12), we …nally have

2k 2k
p 2
2
[(2k)!] (2k + 1) ` (2k) e 2k (2k + 1)
1 = lim 4 4
= lim p 4
k!1 2 (2k ) (k!) k!1 2 4
(2k ) `k k e k k
2k (2k + 1) 4
= lim = ;
k!1 2 `2 k 2 2 `2
p
and so ` = 2 .

Remark 75 Stirling’s formula shows that n! goes to in…nity slower than nn but
faster that bn . Indeed,
p p 1
n! n! e n 2 n n! 2 n2
= p = p ! 1 0;
nn nn e n 2 n nn e n 2 n en
1
where we have used the fact that nen2 ! 0 (since e > 1). On the other hand, if
b > 1,
p p n
bn bn n n e n 2 n nn e n 2 n be 1
= p = 1p ! 1 0;
n! n! nn e n 2 n n! n n2 2
be n
where we have used the fact that n ! 0 (why?).

Wednesday, September 26, 2012

28
Example 76 Let’s calculate
n
n+1
lim :
n!1 n+2
We have
2 3 n+1
n
!n
n
n+1 1 6 1 7 1
= n+2 =4 n+1 5 !e ;
n+2 1
n+1 1+ n+1

n n 1
where we have used the previous limit and the fact that n+1 = n(1+ n
= 1 !
1
) 1+ n
1.
Theorem 77 (Squeeze) Let fxn g, fyn g, and fzn g be three sequences of real
numbers such that
xn yn zn
for all n 2 N.
(i) If there exist lim xn = lim zn = ` 2 R, then there exists lim yn = `.
n!1 n!1 n!1

(ii) If there exists lim xn = 1, then there exists lim yn = 1.

n!1 n!1

(iii) If there exists lim zn = 1, then there exists lim yn = 1.

n!1 n!1

Proof. We prove (i) and leave (ii) and (iii) as an exercise. Given " > 0,
there exist an integer N1 = N1 (") such that
` " xn `+"
for all n N1 and an integer N2 = N2 (") such that
` " zn `+"
for all n N2 . Then for n max fN1 ; N2 g, we have that
` " xn yn zn ` + ";
which implies that jyn `j ".
Corollary 78 Let fxn g and fyn g be two sequences of real numbers such that
there exists lim xn = 0 and fyn g is bounded. Then there exists lim (xn yn ) =
n!1 n!1
0.
Proof. Since lim xn = 0, we have that lim jxn j = 0. Since fyn g is
n!1 n!1
bounded, there exists M > 0 such that jyn j M for all n 2 N. Hence,
M jxn j xn yn M jxn j :
We are in a position to apply the squeeze theorem, since M jxn j ! 0 and
M jxn j ! 0.

29
Example 79 Consider
sin4 n
lim :
n!1 n3

sin4 n
Since 0 sin4 n 1 and 1
n3 ! 0, by the previous corollary, limn!1 n3 = 0.
n p
Example 80 Let’s prove that lim n = 0 if x > 1: Write x = 1 + y where
n!1 x
y > 0: Then by Bernoulli’s inequality
p n n 2
x = (1 + y) (1 + ny)

and so
2n 2
xn (1 + y) (1 + ny) n2 y 2 :
In turn
n n 1
0 = ! 0:
xn n2 y 2 ny 2
Example 81 Let x > 1 and let a > 0. Let’s prove that
na
lim = 0:
n!1 xn

If a = m 2 N, then
0 1m
na n n n
= @h in A = h in h in
xn 1
(x) m
1
(x) m
1
(x) m

1
and since b = (x) m > 1, we have that bnn ! 0 by the previous example. The
result now follow from Theorem 70(ii). If 0 < a < 1, then

na n1
0
xn xn
and the result follows from the squeeze theorem and the previous example. Fi-
nally. if a > 1, then bac a bac + 1, and so

nbac na nbac+1
xn xn xn
and the result follows from the squeeze theorem.

Exercise 82 Let fan g be a sequence of real numbers such that there exists
lim an = 1 and let x > 1.
n!1

(i) Prove that

an
lim = 0:
n!1 xan

30
 (ii) Prove that if b > 0, then
b
(an )
lim = 0:
n!1 xan

(iii) Use parts (i) and (ii) to prove that if a > 0 and b > 0, then
b
(log n)
lim = 0:
n!1 na

Example 83 Let’s calculate

n3 n2 + log2 n + 2n 3
lim :
n!1 4n log n + n7
We have
n3 n2 log2 n 3
n3 n2 + log2 n + 2n 3 2n 2n 2n + 2n +1 2n
=
4n log n + n7 4n 1 log n
+ n7
4n 4n

n3 n2 log2 n 3
2n 2n + 2n +1 2n (0 0 + 0 + 1 0) 1
= ! = =0
2n 1 log n
+ n7 1 (1 0 + 0) 1
4n 4n

as n ! 1, where we have used the facts that, by the previous exercise,

log2 n log2 n n log n log n n

= ! 0; = ! 0:
2n n 2n 4n n 4n

6 Monotone Sequences
De…nition 84 We say that a sequence fxn g of real numbers is

(i) increasing if xn xn+1 for all n 2 N;

(ii) decreasing if xn xn+1 for all n 2 N;
(iii) strictly increasing if xn < xn+1 for all n 2 N;
(iv) strictly decreasing if xn > xn+1 for all n 2 N.

We say that a sequence fxn g of real numbers is a monotone sequence if it

satis…es one of the four properties above.

Theorem 85 Let fxn g be a monotone sequence of real numbers and let

E = fxn : n 2 Ng :

31
 (i) If fxn g is increasing, then there exists limn!1 xn = sup E.
(ii) If fxn g is decreasing, then there exists limn!1 xn = inf E.

Proof. We prove (i). Let L := sup E 2 ( 1; 1]. Fix t < L. Since t is not
an upper bound of E, there exists N = N (t) 2 N such that t < xN L. Since
fxn g is increasing, for all n N , we have that

t < xN xn L:

We now distinguish two cases. If L 2 R then given " > 0, we may take t = L ",
to obtain that
L " xn L L + "
for all n N , which implies that there exists limn!1 xn = L. On the other
hand, if L = 1, then we can take t to be any large number, and so xn t for
all n N , which implies that there exists limn!1 xn = 1.
Friday, September 28, 2012
The previous theorem used the supremum property (S). Actually, we can
show that the opposite is also true.

Theorem 86 Assume that every increasing sequence fxn g of real numbers bounded
from above admits a limit in R. Then the supremum property holds, that is,
every nonempty set E R bounded from above has a supremum.

Proof. Exercise.
Using Theorem 85, we can de…ne the number e. Consider the sequence
n
X 1
sn := ;
k!
k=0

1
where we recall that 0! := 1 and n! := 1 2 n. Note that sn+1 = sn + (n+1)! >
sn , and so the sequence fsn g is strictly increasing. We leave it as an exercise to
prove that fsn g is bounded. Since fsn g is bounded and increasing there exists

lim sn 2 (0; 1) :
n!1

We call this limit e.

Next we study the behavior of sequences de…ned recursively.

Example 87 Consider the sequence de…ned recursively as follows

a1 = a 0;
p
an+1 = 4 an ;

where n 2 N. If a = 0, then by induction, we have that an = 0 for all n 2 N.

Similarly, if a = 1, then by induction, we have that an = 1 for all n 2 N. If
0 < a < 1, then we claim that 0 < an < 1 for all n 2 N. This can be proved by

32
induction. Indeed, it is true for n = 1. Assume that 0 < an < 1 for some n 2 N
p
and let’s prove that 0 < an+1 < 1. We have that 0 < an+1 = 4 an < 1. Hence,
0 < an < 1 for all n 2 N. In turn, in this case we have that the sequence is
p
increasing for all n 2 N. Indeed, an+1 = 4 an an for 0 < an < 1. It follows
from Theorem 85 that there exists

lim an = ` 2 [a; 1] :
n!1

p p p
Hence, ` an+1 = 4 an ! 4 `, which implies that ` = 4 `. It follows that
` = 1.
Similarly, if a > 1, we can show that an > 1 for all n 2 N, that the sequence
is decreasing, so that
lim an = ` 2 [1; a] :
n!1

As before, we conclude that ` = 1.

7 Powers with Real Exponents

If x 2 R and n 2 N, then
xn := x x:
n tim es
p
But what does it mean x 2 ? Or more generally, xa if a 2 R? To de…ne this, we
will assume that x > 0 (this is needed to preserve the properties of powers). If
n
a is positive and rational, say a = m , where m; n 2 N, then we de…ne
n
m
p n
x m := x :
n p
m
p
Remark 88 Note that x m = xn . Indeed, let y = m
x. Then
m n
(y n ) = (y m ) = xn ;
p p n p
and so y n = m xn , that is, ( m x) = m xn .
n
If a is rational and negative, say a = m, where m; n 2 N, then we de…ne
n n
1 m
x m := x :

Exercise 89 Prove that if x > 0 and r; q 2 Q, then

xr xs = xr+s ;
s r
(xr ) = (xs ) = xrs :

Exercise 90 Let x > 1 and r; q 2 Q.

(i) Prove that if r > 0, then xr > 1.

(ii) Prove that if r < s, then xr < xs .

33
(iii) Prove that if ftn g Q and tn ! 0, then xtn ! 1. Hint: Use Example 59.

We are now ready to de…ne xa for a real. Assume that x > 1. We want to
construct an increasing sequence frn g Q such that rn ! a. We proceed as
follows. By the density of the rational numbers, given a 1 and a there exists
r1 2 Q such that a 1 < r1 < a. Again by the density of the rational numbers,
given max a 21 ; r1 and a there exists r2 2 Q such that

1
max a ; r1 < r2 < a:
2

Inductively, assume that rational numbers r1 < r2 < < rn < a have been
constructed in such a way that
1
a < ri < a
i
for all i = 1; : : : ; n. We neednto constructo rn+1 . Again by the density of the
1
rational numbers, given max a n+1 ; rn and a there exists rn+1 2 Q such
that
1
max a ; rn < rn+1 < a:
n+1
Hence, by induction we have constructed an increasing sequence frn g of rational
numbers with
1
a < rn < a:
n
Letting n ! 1 in the previous inequality, it follows by the squeeze theorem
that rn ! a as n ! 1.
Since frn g is increasing, by part (ii) of the previous exercise, it follows that
the sequence fxrn g is also increasing, and thus by Theorem 85, there exists

lim xrn = ` 2 ( 1; 1] :
n!1

Since rn a bac + 1, again by part (ii) of the previous exercise, we have that
x rn xbac+1 , which implies that the sequence fxrn g is bounded from above.
Hence, ` 2 R.
Next let fsn g Q be such that sn ! a. Then by Exercise 89,

xsn x rn = x rn x s n rn
1 :

Since xrn ! ` 2 R and xsn rn ! 1 by part (iii) of the previous exercise, it

follows that xsn xrn ! 0. Thus, we have shown that there exists ` 2 R with
the property that for every sequence fsn g Q such that sn ! a, there exists

lim xsn = `:
n!1

Hence, we de…ne xa := `.

34
 If 0 < x < 1, we set
a
xa := x 1
:
Note that if x > 0 and a; b 2 R, then

xa xb = xa+b ;

Indeed, let frn g Q and fsn g Q be such that rn ! a and sn ! b. Then by

Exercise 89
xrn xsn = xrn +sn ;
and it is enough to let n ! 1.

Exercise 91 Let x > 1 and a; b 2 R.

(i) Prove that if a > 0, then xa > 1.

(ii) Prove that if a < b, then xa < xb .
(iii) Prove that if fan g Q and an ! 0, then xan ! 1.
(iv) Prove that if fan g Q and an ! a, then xan ! xa .

Exercise 92 Let x > 0 and a; b 2 R. Prove that

b a
(xa ) = xb = xab :
b
Hint: It is enough to show (xa ) = xab . Consider …rst the case in which a is
real and b is rational.

Exercise 93 Let x > 1 and a 2 R. Prove that

jxa 1j xjaj 1:

Monday, October 1, 2012

8 Limsup and Liminf

Proposition 94 If a sequence fxn g of real numbers is not bounded from above,
then there exists a subsequence fxnk g of fxn g such that

lim xnk = 1:
k!1

Proof. Consider the number M1 = 1. Since fxn g is not bounded from

above, there exists n1 2 N such that xn1 > 1. Consider the number

M2 = max f2; x1 ; : : : ; xn1 g < 1:

Since fxn g is not bounded from above, there exists n2 2 N such that xn2 >
M2 2. Note that, necessarily, n2 > n1 .

35
 Inductively, assume that k positive integers n1 < < nk have been chosen
in such a way that xni > i for all i = 1; : : : ; k and let’s choose nk+1 . Consider
the number
Mk = max fk + 1; x1 ; : : : ; xnk g < 1:
Since fxn g is not bounded from above, there exists nk+1 2 N such that xnk+1 >
Mk+1 . Note that, necessarily, nk+1 > nk . Thus we have constructed a subse-
quence fxnk g of the original sequence fxn g such that xnk > k for all k 2 N.
Letting k ! 1 and using the squeeze theorem, we get that lim xnk = 1.
k!1
Similarly, we have

Proposition 95 If a sequence fxn g of real numbers is not bounded from below,

then there exists a subsequence fxnk g of fxn g such that

lim xnk = 1:
k!1

Finally, we consider the case of bounded sequences.

Proposition 96 If a sequence fxn g of real numbers is bounded, then there exist

a subsequence fxnk g of fxn g and ` 2 R such that

lim xnk = `:
k!1

Proof. Let F := fxn : n 2 Ng. We distinguish two cases.

Case 1: If the set F has only a …nite number of di¤erent elements, then this
means that there exists ` 2 R such that xn = ` for in…nitely many n. Let fnk g,
k 2 N, be the sequence of all natural numbers such that xnk = `. Then

lim xnk = lim ` = `

k!1 k!1

Thus, we have constructed a subsequence converging to `.

Case 2: The sequence fxn g is bounded. Then the set F is bounded and it has
in…nitely many di¤erent elements. By the Bolzano–Weierstrass theorem, F has
an accumulation point ` 2 R. Then by Remark 50 there exists a sequence in the
set F that converges to `, namely there exists a subsequence fxnk g converging
to `.
On the extended real line [ 1; 1] we can consider a total order relation
by setting 1 x 1 for all x 2 R and by keeping the usual order relation
in R. Given a nonempty set E [ 1; 1], we can de…ne the supremum of a
set as for the real line. Let’s prove that the supremum of a set always exist.

Theorem 97 Every set E [ 1; 1] has a supremum and an in…mum.

36
 Proof. If 1 2 E, then
sup E = max E = 1:
If 1 2= E, then there are two cases. If E = f 1g, then sup E = max E = 1.
If E contains at least two elements, consider the set F = E n f 1g. This set is
contained in R and so it has a supremum L 2 [ 1; 1]. Since 1 x for all
x 2 R, it follows that sup F = sup E.
The existence of the in…mum can be proved in a similar way and we omit it.

Given a sequence fxn g of real numbers, consider the set

E := L 2 [ 1; 1] : there is a subsequence fxnk g such that lim xnk = L :

k!1

We de…ne the limit superior of the sequence fxn g to be

lim sup xn := sup E
n!1

and we de…ne the limit inferior of the sequence fxn g to be

lim inf xn := inf E:
n!1
n
Example 98 Consider the sequence xn = ( 1) . Then E = f 1; 1g, and so
n
lim inf ( 1) = min E = 1;
n!1
n
lim sup ( 1) = max E = +1:
n!1

Theorem 99 Given a sequence fxn g of real numbers, the set

E := L 2 [ 1; 1] : there is a subsequence fxnk g such that lim xnk = L

k!1

is nonempty.
Proof. This follows from Propositions 94, 95, and 96.
Wednesday, October 3, 2012
The next theorem is important for the exercises.
Theorem 100 Let fxn g be a sequence bounded from above and let ` 2 R. Then
the following are equivalent:
(a) ` = lim sup xn ;
n!1

(b) for every " > 0 there exists n" 2 N such that
xn `+" (13)
for all n n" , and
xn ` " (14)
for in…nitely many n 2 N.

37
 n 2n
Example 101 Consider the sequence xn = ( 1) n+1 . To prove that

n 2n
2 = lim sup ( 1)
n!1 n+1
…x " > 0. We want to prove that

n 2n
( 1) 2+"
n+1
for all n su¢ ciently large. If n is odd, then there is nothing to prove, since a
negative number is less than a positive. If n is even, we have that
2n
2 + ";
n+1
that is, 2n (2 + ") (n + 1), which gives 0 " + n" + 2. This is true for all
n 2 N. Thus we can take n" = 1.
Next we want to prove that

n 2n
2 " ( 1) (15)
n+1
for in…nitely many n. Note that if n is odd, then the previous inequality is false.
Thus assume that n is even. Then
2n
2 " ;
n+1
that is (2 ") (n + 1) 2n, which gives 2n " n" + 2 2n, that is, 2 "
2 "
n". Hence, the inequality (15) holds for all n even with n " . There are
in…nitely many such n.

We now turn to the proof of the theorem.

Proof. Assume that lim sup xn = ` 2 R. We need to prove (13) and (14). If
n!1
(13) fails then there exist in…nitely many many n 2 N such that xn ` + ". So
we can …nd a subsequence fxnk g such that xnk ` + " for all k 2 N. Applying
either Proposition 94 or Proposition 96 to the sequence
n ofxnk g, we have that
the sequence fxnk gk admits a further subsequence xnkj such that xnkj has
j
a limit, xnkj ! L. Note that L 2 E. Since xnkj ` + " for all j 2 N letting
j ! 1 we get that L ` + ", which contradicts the fact that ` = sup E. Hence
(13) holds.
To prove (14) note that, since ` " is not an upper bound of the set E there
exist L 2 E such that L > ` ". By the de…nition of E there exists a subsequence
fxnk g such that xnk ! L as k ! 1. Hence, taking "1 = L (` ") > 0 there
exists K 2 N such that

jxnk Lj "1 = L (` ")

38
for all k K, that is,

L + (` ") x nk L L (` ")

for all k K. In particular, xnk ` " for all k K. Hence (14) holds.
Conversely assume that (13) and (14) hold. Let’s prove that `+" is an upper
bound of E. Let L 2 E. By the de…nition of E there exists a subsequence fxnk g
such that xnk ! L as k ! 1. Since

xn `+"

for all n N it follows that

x nk `+"
for all k such that nk N . Letting k ! 1 we conclude that L ` + ". But
since this is true for every " > 0, letting " ! 0+ we get that L `. Hence L `
for all L 2 E, which shows that ` is an upper bound of E.
Let’s prove that ` " is not an upper bound of E. Take ` 2" . By (14)
"
xn `
2
for in…nitely many n 2 N. So we can …nd a subsequence fxnk g such that
x nk ` 2" for all k 2 N. Applying the previous theorem
n to
o fxnk g, we have
that the sequence fxnk gk admits a further subsequence xnkj such that xnkj
j
"
has a limit, xnkj ! L. Note that L 2 E. Since xnkj ` for all j 2 N, 2
letting j ! 1 we get that L ` 2" > ` ", which shows ` " is not an upper
bound of E. Hence ` = sup E.

Corollary 102 Given a sequence fxn g of real numbers, then

lim sup xn = max E;

n!1

that is, there exists a subsequence fxnk g of fxn g such that lim xnk = lim sup xn .
k!1 n!1

Proof. Exercise
A similar theorem holds for sequences which are not bounded from above.

Theorem 103 Given a sequence fxn g of real numbers, the following are equiv-
alent:

(a) lim sup xn = 1;

n!1

(b) The sequence fxn g is not bounded from above.

(c) For every M > 0 there exist in…nitely many n 2 N such that

xn M:

39
 Proof. We have already proved that (a) and (b) are equivalent. On the
other hand, (b) and (c) are clearly equivalent, and so we are done.
We have similar theorems for the limit inferior.

Theorem 104 Given a sequence fxn g of real numbers bounded from below and
a number ` 2 R, the following are equivalent:

(a) lim inf xn = `;

n!1

(b) for every " > 0 there exists a positive integer N = N (") 2 N such that

` " xn for all n N

and there exist in…nitely many n 2 N such that

xn ` + ":

Corollary 105 Given a sequence fxn g of real numbers, then

lim inf xn = min E;

n!1

that is, there exists a subsequence fxnk g of fxn g such that lim xnk = lim inf xn .
k!1 n!1

Theorem 106 Given a sequence fxn g of real numbers, the following are equiv-
alent:

(a) lim inf xn = 1;

n!1

(b) The sequence fxn g is not bounded from below.

(c) For every M > 0 there exist in…nitely many n 2 N such that

xn M:

Friday, October 5, 2012

First midterm.
Monday, October 8, 2012
The relation between limit, limit superior and limit inferior is given by the
following theorem.

Theorem 107 Given a sequence fxn g of real numbers, then

lim inf xn lim sup xn : (16)

n!1 n!1

Moreover there exists lim xn if and only if equality holds in (16), and in this
n!1
case the limit coincides with the common value in (16).

40
 Proof. We have

lim inf xn = inf E sup E = lim sup xn :

n!1 n!1

To prove the second part of the theorem assume that there exists lim xn = `.
n!1
I will consider only the case ` 2 R and leave the cases ` = 1 and ` = 1 as
an exercise. By de…nition of limit, for every " > 0 there exists N = N (") 2 N
such that
` " xn ` + "
for all n N . Since properties (13) and (14) are satis…ed, it follows from
Theorem 100 that ` = lim sup xn . Similarly, by Theorem 104, we have that
n!1
` = lim inf xn .
n!1
Conversely, assume that

lim inf xn = lim sup xn = L

n!1 n!1

for some L 2 [ 1; 1]. Again we consider the case L 2 R and leave the cases
L = 1 and L = 1 as an exercise. Fix " > 0. By Theorems 100 there exists
an integer N1 = N1 (") such that

xn L+"

for all n N1 while by and Theorem 104 there exists an integer N2 = N2 (")
such that
L " xn
for all n N2 . Then for n max fN1 ; N2 g, we have that

L " xn L + ";

which implies that there exists lim xn = L.

n!1
As a corollary of this theorem, we can …nally prove Theorem 62.
Proof of Theorem 62. Let fxn g be a sequence of real numbers. Assume
that there exists
lim xn = ` 2 [ 1; 1] :
n!1

We want to prove that every subsequence fxnk g converges to `. Consider the

set

E := fL 2 [ 1; 1] : there is a subsequence of fxn g converging to Lg :

Since xn ! `, by the previous theorem,

lim inf xn = lim sup xn = `:

n!1 n!1

Hence,
inf E = lim inf xn = lim sup xn = sup E;
n!1 n!1

41
and so inf E = sup E = `. But this implies that E = f`g (if E had two or
more elements the in…mum and the supremum would not coincide). By the
de…nition of E and the previous theorem, we have that every subsequence must
convergence to `.
Conversely, if every subsequence of fxn g goes to ` 2 [ 1; 1], then E = f`g,
and so inf E = sup E = `, which implies that
lim inf xn = lim sup xn = `:
n!1 n!1

Again by the previous theorem, it follows that

lim xn = `:
n!1

Exercise 108 Consider a sequence fxn g of real numbers. Prove that

lim inf ( xn ) = lim sup xn ;
n!1 n!1
lim sup ( xn ) = lim inf xn :
n!1 n!1

The next two theorems are very important to calculate liminf and limsup in
exercises.
Theorem 109 Consider two sequences fxn g and fyn g of real numbers. Then
lim inf xn + lim inf yn lim inf (xn + yn ) lim sup xn + lim inf yn
n!1 n!1 n!1 n!1 n!1

lim sup (xn + yn ) lim sup xn + lim sup yn ;

n!1 n!1 n!1

provided we exclude the case 1 1. Moreover, all inequalities may be strict in

general.
Finally, if one of the two sequences has a limit, then
lim inf xn + lim inf yn = lim inf (xn + yn )
n!1 n!1 n!1

and
lim sup (xn + yn ) = lim sup xn + lim sup yn ;
n!1 n!1 n!1

Wednesday, October 10, 2012

Proof. We only consider the case in which the sequences are bounded and
leave the other cases as an exercise. We prove that
lim inf xn + lim inf yn lim inf (xn + yn ) : (17)
n!1 n!1 n!1

Let
lim inf xn = `1 ; lim inf yn = `2 ;
n!1 n!1
lim inf (xn + yn ) = `3 :
n!1

42
For every " > 0 there exist two positive integer N1 ; N2 2 N such that

`1 " xn for all n N1 ; `2 " yn for all n N2 :

Summing the two inequalities we obtain that

`1 + `2 2" xn + yn for all n max fN1 ; N2 g :

On the other hand, there exist in…nitely many n 2 N such that

xn + yn `3 + ":

By combining the last two inequalities, it follows that for in…nitely many n
max fN1 ; N2 g,
`1 + `2 2" xn + yn `3 + ":
Hence,
`1 + `2 2" `3 + "
but this is true for every " > 0. Letting " ! 0, we conclude that `1 + `2 `3 .
Hence (17) is true.
Now we prove that

lim inf (xn + yn ) lim sup xn + lim inf yn : (18)

n!1 n!1 n!1

Let
lim sup xn = L1 :
n!1
For every " > 0 there exists a positive integer N3 2 N such that

xn L1 + " for all n N3

and there exist in…nitely many n 2 N for which

yn `2 + ":

Summing the two inequalities we obtain that

xn + yn L1 + `2 + 2" for in…nitely many n N3 :

On the other hand, there exists a positive integer N4 2 N such that

`3 " xn + yn for all n N4 :

By combining the last two inequalities, it follows that for in…nitely many n
max fN3 ; N4 g,
`3 " xn + yn L1 + `2 + 2":
and so
`3 " L1 + `2 + 2":
but this is true for every " > 0. Letting " ! 0, we obtain that (18) holds. The
other inequalities are very similar and so we omit them.

43
 n n 1
Example 110 By taking xn = ( 1) and yn = ( 1) , we have that
lim inf xn = 1; lim inf yn = 1;
n!1 n!1
lim (xn + yn ) = 0; lim sup xn = 1; lim sup yn = 1;
n!1 n!1 n!1

which shows that the inequalities in the previous theorem may be strict.
Theorem 111 Let fxn g and fyn g be two sequences of nonnegative numbers.
Then
lim inf xn lim inf yn lim inf (xn yn )
n!1 n!1 n!1
lim inf xn lim sup yn lim sup (xn yn )
n!1 n!1 n!1
lim sup xn lim sup yn ;
n!1 n!1

provided we exclude the case 0 1 and 1 0. Moreover, all inequalities can be

strict.
n n 1
Example 112 By taking xn = 3 + ( 1) and yn = 4 + ( 1) , we have that
lim inf xn = 2; lim inf yn = 3; lim inf (xn yn ) = 10;
n!1 n!1 n!1
lim sup (xn yn ) = 12; lim sup xn = 4; lim sup yn = 5;
n!1 n!1 n!1

so all inequalities are strict in the previous theorem.

n n 1
Example 113 By taking xn = ( 1) and yn = ( 1) , we have that
lim inf xn = 1; lim inf yn = 1; lim (xn yn ) = 1;
n!1 n!1 n!1
lim sup xn = 1; lim sup yn = 1;
n!1 n!1

so Theorem 111 fails is the sequences can take both positive and negative values.
Theorem 114 Consider two sequences fxn g and fyn g of real numbers. Prove
that if there exists lim xn = ` 2 (0; 1), then
n!1

lim xn lim inf yn = lim inf (xn yn ) ;

n!1 n!1 n!1
lim sup (xn yn ) = lim xn lim sup yn :
n!1 n!1 n!1

Remark 115 Note that Theorem 114 does not follow from Theorem 111 since
in Theorem 114 we are not assuming that yn 0.
n 2n 2n
Example 116 Consider the sequence xn = ( 1) n+1 . Since there exists limn!1 n+1 =
2, by the previous theorem,
2n
n n 2n
lim inf ( 1) = lim inf ( 1) lim = 1 2 = 2;
n!1 n+1 n!1 n!1 n + 1
n 2n n 2n
lim sup ( 1) = lim sup ( 1) lim = 1 2 = 2:
n!1 n+1 n!1 n!1 n + 1

44
Exercise 117 Let fxn g be a sequence of real numbers, with xn > 0 for all
n 2 N. Prove that
xn+1 p p xn+1
lim inf lim inf n
xn lim sup n
xn lim sup :
n!1 xn n!1 n!1 n!1 xn
Example 118 The inequalities in the previous exercise can be strict. Take
1
2n if n is even,
xn = 1
3n if n is odd.

Then
xn+1 2n
3n+1 if n is even,
= 3n
xn 2n+1 if n is odd,
and so
xn+1 22k 32k+1
lim inf = lim inf 2k+1 = 0; lim sup = 1;
n!1 xn k!1 3 n!1 22k+2
while
p 1
2 if n is even,
n
xn = 1
3 if n is odd,
and so
xn+1 p 1
lim inf = 0 < lim inf n xn =
n!1 xn n!1 3
1 p xn+1
< = lim sup n xn < lim sup = 1:
2 n!1 n!1 xn

Next we prove some important limits.

n
1
Theorem 119 lim 1+ = e.
n!1 n

Proof. Let
n
X n
1 1
sn := ; tn := 1+ :
k! n
k=0

By the binomial theorem,

n
X n 1 n k
tn = 1 :
k nk
k=0

Note that
n 1 n! 1 (n k)! n (n 1) (n k + 1)
k
= k
=
k n k! (n k)! n k! (n k)! n n
1 1 2 k+1 1
= 1 1 1 :
k! n n n k!

45
Hence, tn sn , and so

lim sup tn lim sup sn = lim sn = e:

n!1 n!1 n!1

On the other hand, if n m,

n
X m
X
n 1 n 1
tn =
k nk k nk
k=0 k=0
Xm
1 1 2 k+1
= 1 1 1 :
k! n n n
k=0

Fixing m and letting n ! 1 in the previous inequality gives

m
X 1 1 2 k+1
lim inf tn lim inf 1 1 1
n!1 n!1 k! n n n
k=0
m
X 1 1 2 k+1
lim inf 1 1 1
n!1 k! n n n
k=0
Xm
1
= :
k!
k=0
Pm
Note that inPthe last inequality it was important to have a …nite sum k=0
n
rather than k=0 . Finally, letting m ! 1 gives
m
X 1
lim inf tn lim = e:
n!1 m!1 k!
k=0

Thus, we have shown that

e lim inf tn lim sup tn e;

n!1 n!1

and so by Theorem 107, there exists limn!1 tn = e.

Friday, October 12, 2012

9 Metric Spaces and Completeness

De…nition 120 A metric on a set X is a map d : X X ! [0; 1) such that

(i) d (x; y) d (x; z) + d (z; y) for all x; y; z 2 X (triangle inequality),

(ii) d (x; y) = d (y; x) for all x; y 2 X (symmetry),
(iii) d (x; y) = 0 if and only if x = y.

A metric space (X; d) is a set X endowed with a metric d. When there is no

possibility of confusion, we abbreviate by saying that X is a metric space.

46
Example 121 Here are some of the most important examples of metric spaces.

(i) In R we have that d (x; y) := jx yj is a metric.

(ii) In RN , N 1, for x = (x1 ; : : : ; xN ) and y = (y1 ; : : : ; yN ),
q
2 2
d (x; y) := (x1 y1 ) + : : : + (xN yN )

is a metric.
(iii) Taking X to be the set of all bounded functions f : [a; b] ! R, it can be
shown that
d (f; g) := sup jf (x) g (x)j
x2[a;b]

is a metric.
(iv) In R we have that

x y
d1 (x; y) := (19)
1 + jxj 1 + jyj
is a metric.

De…nition 122 Given a metric space (X; d) and a sequence fxn g X, we say
that

(i) fxn g is a Cauchy sequence if for every " > 0 there exists N" 2 N such
that
d (xn ; xm ) "
for all n; m N" ,
(ii) fxn g converges to x 2 X if for every " > 0 there exists N" 2 N such that

d (xn ; x) "

for all n N" .

Remark 123 The previous de…nition does not change if we replace with <.
Thus, we can say that fxn g converges to x 2 X if for every " > 0 there exists
N" 2 N such that
xn 2 B (x; ")
for all n N" , where B (x; ") := fy 2 X : d (x; y) < "g is the ball centered at x
and radius " > 0.
Similarly, if (X; ) is a topological space, we can say that fxn g converges to
x 2 X if for every open set U 2 containing x there exists NU 2 N such that

xn 2 U

for all n NU .

47
Proposition 124 Given a metric space (X; d) and a sequence fxn g X, if
fxn g converges to x 2 X, then fxn g is a Cauchy sequence.
"
Proof. Since fxn g converges to x 2 X, given " > 0, consider 2 in the
de…nition of convergence. Then there exists N" 2 N such that
"
d (xn ; x)
2
for all n N" . Hence, by the triangle inequality and symmetry of d, if n; m
N" ,
" "
d (xn ; xm ) d (xn ; x) + d (x; xm ) = d (xn ; x) + d (xm ; x) + :
2 2

The opposite is not true, that is, there are Cauchy sequences that do not
have a limit.

Example 125 Consider X = (0; 1] with the metric d (x; y) = jx yj and con-
sider the sequence xn = n1 . Then xn ! 0 which does not belong to X = (0; 1],
but fxn g is a Cauchy (just applied the previous proposition in the metric space
R).

Exercise 126 Let fxn g be a sequence of real numbers.

(i) Prove that if fxn g is a Cauchy sequence and if a subsequence fxnk g of

fxn g converges to some x 2 R, then fxn g converges to x.
(ii) Prove that if there exists x 2 R such that
n for o
every subsequence fxnk g of
fxn g there exists a further subsequence xnkj that converges to x, then
fxn g converges to x.

A metric space (X; d) is said to be complete if every Cauchy sequence is

convergent.

Example 127 Let X = (0; 1) with the metric d (x; y) = jx yj. The sequence
1
n n2N converges to 0 in R, and so it is a Cauchy sequence in R. In particular,
it is a Cauchy sequence in X. However, it does not converge to an element of
X, since 0 2
= X.

Theorem 128 (R; d) is a complete metric space.

Proof. Let fxn g be a Cauchy sequence.

Step 1: We claim that fxn g is bounded. Fix " = 1. By the de…nition of Cauchy
sequence, there is exists N1 2 N such that

jxn xm j 1

for all n; m N1 . In particular, taking m = N1 , we have that

jxn xN1 j 1;

48
for all n N1 , and so
jxn j = jxn xN1 j jxn xN1 j + jxN1 j 1 + jxN1 j = M1
for all n N1 . On the other hand, if n < N1 , then
jxn j jx1 j + jx2 j + + jxN1 j = M2 :
Hence, for every n 2 N, we have that
jxn j max fM1 ; M2 g :
This shows that fxn g is bounded.
Step 2: Since fxn g is bounded, by Proposition 96, there exist a subsequence
fxnk g of fxn g and ` 2 R such that lim xnk = `. Let’s prove that the entire
k!1
sequence converges to `.
Given " > 0, by the de…nition of Cauchy sequence, there is exists N" 2 N
such that
"
jxn xm j
2
for all n; m N" . On the other hand, since lim xnk = `, there is exists K" 2 N
k!1
such that
"
jxnk `j
2
for all k K" . Since fnk g is strictly increasing, nk ! 1 as k ! 1 and so
nk N" for all k large, say k K1 . Fix k max fK" ; K1 g. Then for all
n N" ,
" "
jxn `j = jxn ` xnk j jxn xnk j + jxnk `j + ;
2 2
where we have used the facts that n; nk N" and that k K" .
This implies that fxn g converges to `.
Remark 129 (Important) The previous theorem relies on Proposition 96,
which in turn is based on the Bolzano–Weierstrass theorem. Note that in in…-
nite dimensional spaces (such as the space of bounded functions), the Bolzano–
Weierstrass theorem fails in general.
Exercise 130 Prove that the function d1 de…ned in (19) is a metric and that
(R; d1 ) is not complete.
The completeness of the real line together with the Archimedean property
is equivalent to Dedekind axiom.
Theorem 131 The following are equivalent.
(D) Dedekind axiom.
(S) The supremum property.
(M ) Every increasing sequence bounded from above has a limit in R.
(C) Every Cauchy sequence has a limit and the Archimedean property holds.
Monday, October 15, 2012

49
10 Sequences and Topology
A set C R is sequentially closed if for every sequence fxn g C such that
fxn g converges to some x 2 R, then x belongs to C.

Proposition 132 Let C R. Then C is closed if and only if C is sequentially

closed.

Proof. Step 1: Assume that C is closed and let fxn g C be such that
fxn g converges to some x 2 R. We need to show that x belongs to C. If not,
then x 2 RnC. Since RnC is open, there exists r > 0 such that B (x; r) RnC.
But then, taking " = r there exists nr 2 N such that jxn xj < r for all n nr ,
which implies that xn 2 B (x; r) R n C for all n nr . This contradicts the
fact that fxn g C.
Step 2: Assume that C is sequentially closed. We need to show that R n C is
open. Let x 2 RnC. We claim that there exists r > 0 such that B (x; r) RnC.
If not, then for every r > 0 we can …nd y 2 C such that y 2 B (x; r). Taking
r = n1 we can …nd xn 2 C such that jxn xj < n1 ! 0, which shows that fxn g
converges to x. Since C is sequentially closed, it follows that x 2 C, which is a
contradiction.

De…nition 133 A set K R is sequentially compact if for every sequence

fxn g K, there exist a subsequence fxnk g of fxn g and x 2 K such that
xnk ! x as k ! 1.

Example 134 A …nite set K R is sequentially compact. The set E = [0; 1)

is not sequentially compact, since the sequence 1 n1 converges to 1, which
does not belong to E. The problem here is that E is not closed.
The set F = [0; 1) is closed but not sequentially compact, since the sequence
fng converges to 1, which does not belong to F . The problem here is that F is
not bounded.

The following theorem is one of the main results of section.

Theorem 135 Let K R. Then the following are equivalent.

(i) K is sequentially compact.

(ii) K is closed and bounded.

Proof. Assume that K is sequentially compact. We claim that K is sequen-

tially closed. To see this, let fxn g K be a sequence such that xn ! x for
some x 2 R. Since K is sequentially compact, there exist a subsequence fxnk g
of fxn g and y 2 K such that xnk ! y as k ! 1. By the uniqueness of limits,
it follows that x = y 2 K, which shows that K is sequentially closed.
Next we prove that K is bounded from above. Indeed, if not, then for every
n 2 N we would …nd xn 2 K such that xn n, but then xn ! 1, and so

50
every subsequence of fxn g converges to 1, which contradicts the fact that K
is sequentially compact.
Conversely, assume that K is closed and bounded and let fxn g K. Since
K is bounded, so is fxn g. Hence,by Proposition 96 there exists a subsequence
fxnk g of fxn g such that lim xnk = `. Since K is closed, it is sequentially
k!1
closed, and so ` 2 K. This shows that K is sequentially compact.

De…nition 136 A set K R is compact if forS every open cover of K, i.e.,

for every collection fUi g of open sets such that i Ui K, there exists a …nite
subcover (i.e., a …nite subcollection of fUi g whose union still contains K).

Example 137 The set E = N is not compact. To see this, consider the family
of balls B n; 21 . These balls cover E but since they are disjoint, if we remove
even one of them, we cannot cover E anymore. The problem here is that E is
unbounded.

Example 138 The set E = [0; 1) is not compact. Consider the family of open
sets Un = 1; 1 n1 . Then
1
[ 1
1; 1 = ( 1; 1) ;
n=1
n

but if we consider a …nite number of the sets Un , say, Un1 , . . . , Un` , letting
N = max fn1 ; : : : ; n` g, we have that
`
[ 1 1
1; 1 = 1; 1 ;
nk N
k=1

which does not cover E. The problem is that E is not bounded.

Exercise 139 Prove that if K is closed and bounded, then K is compact.

S Hint:
Assume that there exists a collection fUi g of open sets such that i Ui K for
which there is no …nite subcover. Repeat the proof of the Bolzano–Weierstrass
theorem.

Exercise 140 Prove that if K is compact, then K is closed and bounded.

Example 141 In view of the previous theorem, the interval [a; b] R is se-
quentially compact.

Wednesday, October 17, 2012

11 Functions
Consider a function f : E ! R, where E R. The set E is called the domain
of f . If E is not speci…ed, then E should be taken to be the largest set of x for
which f (x) makes sense. This means that:

51
 If there are even roots, their arguments should be nonnegative. If there are
logarithms, their arguments should be strictly positive. Denominators should
be di¤erent from zero. If a function is raised to an irrational number, then the
function should be nonnegative.
Given a set F E, the set f (F ) = fy 2 R : y = f (x) for some x 2 F g is
called the image of F through f . The function f is said to be bounded from
above in F , bounded from below in F , bounded in F if the set f (F ) is bounded
from above, bounded from below, bounded, respectively.
Given a set G R, the set f 1 (G) = fx 2 E : f (x) 2 Gg is called the
inverse image of F through f . It has NOTHING to do with the inverse function.
It is just one of those unfortunate cases in which we use the same symbol for
two di¤erent objects.
The graph of a function is the set of R2 de…ned by
gr f = f(x; f (x)) : x 2 Eg :
A function f is said to be
increasing if f (x) f (y) for all x; y 2 E with x < y,
strictly increasing if f (x) < f (y) for all x; y 2 E with x < y,
decreasing if f (x) f (y) for all x; y 2 E with x < y,
strictly decreasing if f (x) > f (y) for all x; y 2 E with x < y,
monotone if one of the four property above holds,
one-to-one or injective if f (x) 6= f (y) for all x; y 2 E with x 6= y.
If f : E ! F , where E; F R, then f is said to be
onto or surjective if f (E) = F ,
bijective or invertible if it is one-to-one and onto. The function f =1 : F !
E, which assigns to each y 2 F = f (E) the unique x 2 E such that
f (x) = y, is called the inverse function of f .

12 Limits of Functions
De…nition 142 Let E R, let x0 2 R be an accumulation point of E, and let
f : E ! R. We say that
a number ` 2 R is the limit of f (x) as x approaches x0 if for every " > 0
there exists a real number = ("; x0 ) > 0 with the property that
jf (x) `j "
for all x 2 E with 0 < jx x0 j . We write
lim f (x) = ` or f (x) ! ` as x ! x0 :
x!x0

52
 1 is the limit of f (x) as x approaches x0 if for every M > 0 there exists
a real number = (M; x0 ) > 0 with the property that

f (x) M

for all x 2 E with 0 < jx x0 j . We write

lim f (x) = 1 or f (x) ! 1 as x ! x0 :

x!x0

1 is the limit of f (x) as x approaches x0 if for every M > 0 there

exists a real number = (M; x0 ) > 0 with the property that

f (x) M

for all x 2 E with 0 < jx x0 j . We write

lim f (x) = 1 or f (x) ! 1 as x ! x0 :

x!x0

Note that even when x0 2 E, we cannot take x = x0 since in the de…nition

we require 0 < jx x0 j.

Theorem 143 Let E R and let x0 2 R be an accumulation point of E. Given

a function f : E ! R there exists

lim f (x) = ` 2 [ 1; 1]
x!x0

if and only if
lim f (xn ) = `
n!1

for every sequence fxn g E n fx0 g converging to x0 . In particular, if the limit

exists, it is unique.

Proof. We consider only the case ` 2 R and leave the cases ` = 1 and
` = 1 as an exercise. Assume that there exists ` = limx!x0 f (x) and let
fxn g E n fx0 g converge to x0 . Fix " > 0 and …nd = ("; x0 ) > 0 such that

jf (x) `j "

for all x 2 E with 0 < jx x0 j . Since xn ! x0 , there exists n" 2 N such

that jxn x0 j for all n n" and so

jf (xn ) `j "

for all n n" , which shows that f (xn ) ! `.

Conversely, assume that limn!1 f (xn ) = ` for every sequence fxn g E n
fx0 g converging to x0 . If either the limit limx!x0 f (x) does not exist or it exists

53
but it it di¤erent from `, then there exists " > 0 with the property that for every
there exists x 2 E n fx0 g with 0 < jx x0 j < such that

jf (x) `j > ":

1 1
Take = n for n 2 N and …nd xn 2 E n fx0 g with jxn x0 j < n such that

jf (xn ) `j > ": (20)

Then x0 n1 xn x0 + n1 for every n, and so by the squeeze theorem, xn ! x0 .

But then, by hypothesis f (xn ) ! ` as n ! 1, which contradicts (20).

Remark 144 In view of the previous theorem, in order to show that the limit
limx!x0 f (x) does not exist, it is enough to …nd two sequences fxn g E n fx0 g
and fyn g E n fx0 g both converging to x0 and such that

lim f (xn ) 6= lim f (yn ) :

n!1 n!1

Example 145 Consider the function f (x) = cos x1 de…ned in E = R n f0g.

Note that 0 is an accumulation point of E. To prove that the limit
1
lim cos
x!0 x
1 1
does not exist, consider the sequences xn = 2n ! 0 and xn = n ! 0. Then

lim f (xn ) = lim cos (2n ) = 1 6= lim f (yn ) = lim cos (n ) = 1:

n!1 n!1 n!1 n!1

We now list some important operations for limits.

Theorem 146 Let E R and let x0 2 R be an accumulation point of E. Given

three functions f; g; h : E ! R, assume that there exist

lim f (x) = `1 2 R; lim g (x) = `2 2 R:

x!x0 x!x0

Then

(i) there exists lim (f + g) (x) = `1 + `2 ,

x!x0

(ii) there exists lim (f g) (x) = `1 `2 ,

x!x0

(iii) if `2 6= 0 and F := fx 2 E : g (x) 6= 0g, then x0 is an accumulation point

f `1
of F and there exists lim (x) = ,
x!x0 g F `2
(iv) (Squeeze Theorem) if `1 = `2 and f (x) h (x) g (x) for every x 2 E,
then there exists lim h (x) = `1 .
x!x0

54
 Proof. Parts (i) and (ii) follow from Theorems 70 and 143, part (iv) from
Theorems 77 and 143. The fact that x0 is an accumulation point of F is left as
an exercise.

Remark 147 As in the case of sequences, the previous theorem continues to

hold if `1 ; `1 2 [ 1; 1], provided we avoid the cases 1 1, 01, 00 , 1
1.

Theorem 148 Let E R and let x0 2 R be an accumulation point of E. Given

two functions f; g : E ! R, assume that there exists

lim f (x) = 0;
x!x0

and that g is bounded. Then there exists lim (f g) (x) = 0.

x!x0

Proof. This follows from Corollary 78 and Theorem 143.

Example 149 The previous theorem can be used for example to show that for
a>0
1
lim xa sin = 0:
x!0 x
Example 150 We list below some important limits.
sin x 1 cos x 1 log (1 + x)
lim = 1; lim = ; lim = 1;
x!0 x x!0 x2 2 x!0 x
a x
(1 + x) 1 e 1
lim = a for a 2 R; lim = 1:
x!0 x x!0 x
Friday, October 19, 2012
Midsemester break.
Monday, October 22, 2012
sin x
Theorem 151 lim = 1.
x!0 x
Proof. Let 0 < x < 2 and consider the triangle of vertices O = (0; 0),
A = (1; 0), and B = (cos x; sin x). Consider the point D = (1; tan x). Then
the area of the triangle OAB is less than the area of the circular sector1 that
contains the triangle, and in turn the area of the circular sector is less than the
area of the triangle OAD. Note that the base of the triangle OAB is one and
the height is sin x. Thus, we have
1 x 1
0< sin x < < tan x:
2 2 2
1 A circular sector is the portion of a circle enclosed by two radii and an arc. It is the
angle in radians formed by the two radii and r is the radius of the circle, the area of the
circular sector is given by
1
A = r2 :
2

55
 1
Dividing everything by 2 sin x and inverting the inequalities, we get

sin x
cos x < < 1;
x
and so
sin x x x 2
0<1 < 1 cos x = 2 sin2 2 :
x 2 2
By the squeeze theorem we get that
sin x
lim = 1:
x!0+ x
Since sin ( x) = sin x, considering the change of variables y = x, we have
that
sin y sin ( x) sin x
lim+ = lim+ = lim+ = 1;
y!0 y x!0 x x!0 x
sin x
and so there exists lim = 1.
x!0 x
Remark 152 Note that we have also shown that

jsin xj jxj

for every 0 < jxj < 2. On the other hand, if jxj 2, then

jsin xj 1< jxj

2
and so the inequality jsin xj jxj holds for every x. In particular, by the squeeze
theorem,
lim sin x = 0:
x!0

1 cos x
Theorem 153 limx!0 x2 = 12 .

Proof. We have
x
1 cos x = 2 sin2
2
and so
2
1 cos x sin2 x2 2 sin x2 1
2
= 2 2
= x ! 1
x x 4 2 2
as x ! 0.

Exercise 154 Prove that for every x0 2 R,

lim cos x = cos x0 :

x!x0

We next study the limit of composite functions.

56
Theorem 155 Let E; F R and let x0 2 R be an accumulation point of E.
Given two functions f : E ! F and g : F ! R, assume that there exist

lim f (x) = ` 2 R;
x!x0

that ` is an accumulation point of F , and that there exists

lim g (y) = L 2 [ 1; 1] :
y!`

Assume that either there exists 1 > 0 such that f (x) 6= ` for all x 2 E with
0 < jx x0 j 1 , or that ` 2 F , L 2 R and g (`) = L. Then there exists
lim g (f (x)) = L.
x!x0

Proof. We consider only the case L 2 R and leave the cases L = 1 and
L = 1 as an exercise. Fix " > 0 and …nd = ("; `) > 0 such that

jg (y) Lj " (21)

for all y 2 F with 0 < jy `j .

Since limx!x0 f (x) = `, there exists 2 = 2 (x0 ; ) > 0 such that

jf (x) `j

for all x 2 E with 0 < jx x0 j 1.

We now distinguish two cases.
Case 1: Assume that f (x) 6= ` for all x 2 E with 0 < jx x0 j 1. Then
taking = min f 1 ; 2 g, we have that for all x 2 E with 0 < jx x0 j ,

0 < jf (x) `j :

Hence, taking y = f (x), by (21), it follows that

jg (f (x)) Lj "

for all x 2 E with 0 < jx x0 j . This shows that there exists lim g (f (x)) =
x!x0
L.
Case 2: Assume that ` 2 F and g (`) = L. Let x 2 E with 0 < jx x0 j 1.
If f (x) = `, then g (f (x)) = L, and so

jg (f (x)) Lj = 0 ";

while if f (x) 6= `, then taking y = f (x), by (21), it follows that

jg (f (x)) Lj ":

57
Example 156 Let’s prove that the previous theorem fails without the hypotheses
that either f (x) 6= ` for all x 2 E near x0 or ` 2 F , L 2 R and g (`) = L.
Consider the function
1 if y 6= 0;
g (y) :=
2 if y = 0:
Then there exists
lim g (y) = 1:
y!0

So L = 1. Consider the function f (x) := 0 for all x 2 R. Then for every

x0 2 R, we have that
lim f (x) = 0:
x!x0

So ` = 0. However, g (f (x)) = g (0) = 2 for all x 2 R. Hence,

lim g (f (x)) = lim 2 = 2 6= 1;

x!x0 x!x0

which shows that the conclusion of the theorem is violated.

Note that the previous theorem can be used to change variables in limits.

Example 157 Let’s try to calculate

log (1 + sin x)
lim :
x!0 x
For sin x 6= 0, we have
log (1 + sin x) log (1 + sin x) sin x log (1 + sin x) sin x
= = :
x x sin x sin x x
sin x
Since limx!0 x = 1, it remains to study

log (1 + sin x)
lim :
x!0 sin x

Consider the function g (y) = log(1+y)

y and the function f (x) = sin x. As x ! 0,
we have that sin x ! 0 = `, while
log (1 + y)
lim = 1:
y!0 y

Moreover sin x 6= 0 for all x 2 E := 2; 2 n f0g. Hence, we can apply the

previous theorem to conclude that
log (1 + sin x)
lim = 1:
x!0 sin x
In turn,
log (1 + sin x)
lim = 1:
x!0 x

58
 Wednesday, October 24, 2012

De…nition 158 Let E R and let f : E ! R be a function. Given a subset

F E we denote by f jF the restriction of the function f to the set F , that is
the function f : F ! R.

Theorem 159 Let E R be such that E = F [G and let x0 2 R be an accumu-

lation point of both F and G. Consider a function f : E ! R. Then there exists
limx!x0 f (x) if and only if there exist limx!x0 f jF (x) and limx!x0 f jG (x) and
they are equal.

Proof. Assume that there exist limx!x0 f jF (x) and limx!x0 f jG (x) and
that their equal. Let ` be the common value. Assume that ` 2 R (the cases ` =
1 and ` = 1 are left as an exercise). Fix " > 0. Since limx!x0 f jF (x) = `,
there exists 1 = 1 ("; x0 ) > 0 with the property that

jf (x) `j "

for all x 2 F with 0 < jx x0 j 1 . Since limx!x0 f jG (x) = `, there exists

2 = 2 ("; x0 ) > 0 with the property that

jf (x) `j "

for all x 2 G with 0 < jx x0 j 2 . Take = min f 1 ; 2 g > 0.

If x 2 E, with 0 < jx x0 j , then, since E = F [ G, we have that either
x 2 F or x 2 G (or both), and so

jf (x) `j ":

Conversely, if the limit limx!x0 f (x) = ` exists, assume that ` 2 R (the

cases ` = 1 and ` = 1 are left as an exercise). Fix " > 0. Then there exists
= ("; x0 ) > 0 with the property that

jf (x) `j "

for all x 2 E with 0 < jx x0 j . By restricting x to F , we get that there

exists limx!x0 f jF (x) = `, while by restricting x to G, we get that there exists
limx!x0 f jF (x) = `.
An important special case is obtained by taking as sets F and G,

E := E \ ( 1; x0 ] ; E + := E \ (x0 ; 1) :

Whenever they exist, the limits limx!x0 f jE (x) and limx!x0 f jE + (x) are
called the left and right limit of f as x ! x0 and they are denoted, respec-
tively, by
lim f (x) and lim f (x) :
x!x0 x!x+
0

59
Exercise 160 Prove that there exists
1
lim (1 + x) x :
x!0

Hint: Calculate the left and right limits separately.

Theorem 161 Let E R, let x0 2 R be an accumulation point of E, and
let f : E ! R be a monotone function. If x0 is an accumulation point of
E := E \ ( 1; x0 ], then there exists
lim f (x) = ` 2 [ 1; 1] ;
x!x0

while if x0 is an accumulation point of E + := E \ (x0 ; 1), then there exists

lim f (x) = L 2 [ 1; 1] :
x!x+
0

Moreover, if f is increasing and x0 is an accumulation point of E and E + ,

then ` L, while if f is decreasing and x0 is an accumulation point of E and
E + , then ` L.
Proof. Assume that f is increasing and let
` := sup ff (x) : x 2 E; x < x0 g :
Assume that ` 2 R (the case ` = 1 is left as an exercise). Fix " > 0. We need
to …nd = (x0 ; ") > 0 such that
` " f (x) `+"
for all x 2 E with x0 x < x0 . By the de…nition of supremumn, we have
that f (x) ` for all x 2 E with x < x0 . On the other hand, since ` " is not an
upper bound, there exists x1 2 E, with x1 < x0 such that ` " < f (x1 ). Take
:= x0 x1 . If x 2 E with x0 = x1 x < x0 , then, since f is increasing,
` " < f (x1 ) f (x) ;
which gives the other inequality.
Exercise 162 Prove that
lim xa logb x = 0
x!0+

for a > 0 and b 2 R. Hint: Use Exercise 82.

Another important theorem is the following.
Exercise 163 Let E R, let x0 2 R be an accumulation point of E, and let
f : E ! R be a function. Assume that there exist
lim f (x) = ` 2 (0; 1] :
x!x0

Prove that there exists = ("; x0 ) > 0 such that for all x 2 E with jx x0 j <
we have
f (x) > 0:
Remark 164 A similar result continues to hold if ` < 0.

60
13 Limits at In…nity
De…nition 165 Let E R be unbounded from above and let f : E ! R. We
say that

a number ` 2 R is the limit of f (x) as x diverges to 1 if for every " > 0

there exists a real number M = M (") > 0 with the property that

jf (x) `j "

for all x 2 E with x M . We write

lim f (x) = ` or f (x) ! ` as x ! 1:

x!1

1 is the limit of f (x) as x diverges to 1 if for every L > 0 there exists

a real number M = M (L) > 0 with the property that

f (x) L

for all x 2 E with x M . We write

lim f (x) = 1 or f (x) ! 1 as x ! 1:

x!1

1 is the limit of f (x) as x diverges to 1 if for every L > 0 there exists

a real number M = M (L) > 0 with the property that

f (x) L

for all x 2 E with x M . We write

lim f (x) = 1 or f (x) ! 1 as x ! 1:

x!1

Note that if E R be unbounded from above, then there exists a sequence

fxn g E such that xn ! 1, so the previous de…nition makes sense, since for
every M > 0 there will always be in…nitely many x 2 E with x M .
Similarly, we have

De…nition 166 Let E R be unbounded from below and let f : E ! R. We

say that

a number ` 2 R is the limit of f (x) as x diverges to 1 if for every " > 0

there exists a real number M = M (") > 0 with the property that

jf (x) `j "

for all x 2 E with x M . We write

lim f (x) = ` or f (x) ! ` as x ! 1:

x! 1

61
 1 is the limit of f (x) as x diverges to 1 if for every L > 0 there exists
a real number M = M (L) > 0 with the property that

f (x) L

for all x 2 E with x M . We write

lim f (x) = 1 or f (x) ! 1 as x ! 1:

x! 1

1 is the limit of f (x) as x diverges to 1 if for every L > 0 there

exists a real number M = M (L) > 0 with the property that

f (x) L

for all x 2 E with x M . We write

lim f (x) = 1 or f (x) ! 1 as x ! 1:

x! 1

Theorems 143, 146, 148, 155 continue to hold if we replace x0 2 R with 1

or 1. We omit the details.

Exercise 167 Calculate the following limits

x
1
1. lim 1+ ,
x!1 x
x
1
2. lim 1+ ,
x! 1 x
1
3. lim+ (1 + x) x .
x!0

Exercise 168 Using the previous exercise, prove that

log (1 + x)
lim =1
x!0 x
and that
ex 1
lim = 1:
x!0 x
Friday, October 26, 2012

14 Continuity
De…nition 169 Let E R. A point x0 2 E is called an isolated point of E if
there exists > 0 such that

(x0 ; x0 + ) \ E = fx0 g :

62
 It is clear that if a point of the set E is not an isolated point of E then it is
an accumulation point of E.

De…nition 170 Let E R and let x0 2 E. Given a function f : E ! R we

say that f is continuous at x0 if for every " > 0 there exists a real number
= ("; x0 ) > 0 such that for all x 2 E with jx x0 j < we have

jf (x) f (x0 )j < ":

If f is continuous at every point of E we say that f is continuous on E and we

write f 2 C (E) or f 2 C 0 (E).

Theorem 171 Let E R and let x0 2 E. Given a function f : E ! R,

(i) if x0 is an isolated point of E then f is continuous at x0 ;

(ii) if x0 is an accumulation point of E then f is continuous at x0 if and only

if there exists lim f (x) = f (x0 ).
x!x0

Proof of part (i). If x0 is an isolated point of E then there exists 0 >0

such that
(x0 0 ; x0 + 0 ) \ E = fx0 g :

Fix " > 0 and take := 0 in the de…nition of continuity. Clearly if x 2 E and
jx x0 j < then necessarily x = x0 so that we have jf (x) f (x0 )j = 0.

Exercise 172 Prove that the functions sin x, cos x, xn , where n 2 N, are con-
tinuous.

The following theorems follows from the analogous results for limits.

Theorem 173 Let E R and let x0 2 E. Given two functions f , g : E ! R

assume that f and g are continuous at x0 . Then

(i) f + g and f g are continuous at x0 ;

f
(ii) if g (x0 ) 6= 0 then restricted to the set F := fx 2 E : g (x) 6= 0g is
g
continuous at x0 .

Example 174 In view of Exercise 172 and the previous theorem, the functions
sin x cos x
tan x = cos x and cot x = sin x are continuous in their domain of de…nition.

Theorem 175 Let E, F R and let x0 2 R be an accumulation point of E.

Given two functions f : E ! F and g : F ! R assume that f is continuous at
x0 and that g is continuous at f (x0 ). Then g f : E ! R is continuous at x0 .v

63
15 Discontinuities
Let E R and let f : E ! R. Given x0 2 E, what happens when f is
discontinuous at x0 ? Then x0 is an accumulation point of E. The following
situations can arise. It can happen that there exists

lim f (x) = ` 2 R
x!x0

but ` 6= x0 . In this case, we say that x0 is a removable discontinuity. Indeed,

consider the function g : E ! R de…ned by
f (x) if x 6= x0 ;
g (x) :=
` if x = x0 :
Then there exists
lim g (x) = ` = g (x0 ) ;
x!x0

and so the new function g is continuous at x0 .

Another type of discontinuity is when x0 is an accumulation point of E :=
E \ ( 1; x0 ] and of E + := E \ (x0 ; 1) and there exist

lim f (x) = ` 2 R; lim f (x) = L 2 R

x!x0 x!x+
0

but ` 6= L. In this case the point x0 is called a jump discontinuity of f .

Example 176 The integer and fractional part of x have jump discontinuity at
every integer.

An important class of functions that exhibit only jump discontinuities are

monotone functions.

De…nition 177 A set E R is countable if there exists an injective function

f : E ! N.

Example 178 The following sets are countable.

1. A …nite set E is countable. Let E = fx1 ; x2 ; : : : ; xk g and de…ne the func-

tion f (xi ) := i. Then f is injective.
2. The set of integers Z is countable. De…ne
2k if k 1;
f1 (k) :=
3 k if k < 0:

Then f2 : Z ! N is injective (exercise).

3. The set of integers N N is countable. De…ne

f2 (n; m) := 2n 3m

Then f2 : N N ! N is injective (exercise).

64
 4. The set of integers Z N is countable. De…ne

2k 5m if k 1;
f3 (k; m) :=
3 k 5m if k < 0:

Then f3 : Z N ! N is injective (exercise).

k
5. The set of rationals Q is countable. Given r 2 Q write r = m where
m 2 N and k 2 Z, where r = 0 take k = 0 and m = 1, while if r 6= 0, take
m and n with no common divisors. The function f4 : Q ! Z N de…ned
by
f4 (k; m) := (k; m)
Since composition of injective functions is injective the composition of f4
and f3 is injective (exercise) and so Q is countable.

Exercise 179 Prove that if En R, n 2 N, is countable, then

1
[
E= En
n=1

is countable.

Remark 180 It can be shown that R is NOT countable.

Example 181 The set of irrationals R n Q is not countable. Indeed, if R n Q

were countable, then since Q is countable, it would follows that R = (R n Q) [ Q
and so R would be countable, since union of two countable sets.

Example 182 The set 2; 2 is not countable. Indeed,

De…nition 183 A set I R is an interval if for every x; y 2 I, with x < y,

we have that the interval [x; y] is contained in I.

Theorem 184 Let I R be an interval and let f : I ! R be a monotone

function. Then f has at most countably many discontinuity points.

Proof. Step 1: Assume that I = [a; b] and, without loss of generality, that
f is increasing. For every x 2 (a; b) there exist

lim f (y) =: f+ (x) ; lim f (y) =: f (x) :

y!x+ y!x

Let
E := fx 2 (a; b) : f is discontinuous at xg :
Let S (x) := f+ (x) f (x) 0 be the jump of f at x. Then f is continuous
at x if and only if S (x) = 0. Hence,

E = fx 2 (a; b) : S (x) > 0g :

65
For each n 2 N de…ne
1
En := x 2 (a; b) : S (x) :
n

Fix n 2 N. We claim that En has at most ` := bn (f (b) f (a))c elements.

To see this, assume by contradiction that En has more than ` elements. Let
x1 ; : : : ; x`+1 2 E. By reordering the lements, we can assume that

x1 < x2 < < x`+1 :

Since f is increasing, we have that

f (a) f (x1 ) f+ (x1 ) f (x2 ) f+ (x2 )

f (x`+1 ) f+ (x`+1 ) f (b) ;

and so, also using the de…nition of En ,

`+1
X `+1
X
`+1
S (xi ) = (f+ (xi ) f (xi )) f (b) f (a) ;
n i=1 i=1

which implies that ` + 1 n (f (b) f (a)). This contradicts the fact that
` := bn (f (b) f (a))c. Hence, En has …nitely elements, and so it is countable.
Next we claim that
1
[
E= En : (22)
n=1
S1
Since En E for every n 2 N, we have that n=1 En E. To prove the
opposite inclusion, let x 2 E. Then S (x) > 0. By the Archimedean property,
1
there exists n0 2 N such that n0 > S(x) and so S (x) > n10 . This means
S1
that S x 2 En0 , and, in turn, that x 2 n=1 En . Thus, we have proved that
1
E n=1 En .
Since (22) holds and each En is countable, it follows by Exercise 179 that E
is countable.
Step 2: If I is an arbitrary interval, construct an increasing sequence of intervals
[an ; bn ] such that
an & inf I; bn % sup I:
Since the union of countable sets is countable by Exercise 179 and on each
interval [an ; bn ] the set of discontinuity points of f is at most countable, by the
previous step it follows that the set of discontinuity points of f in I is at most
countable.
Finally, the last type of discontinuity is when at least one of the limits
limx!x f (x) and limx!x+ f (x) is not …nite or does not exist. In this case, the
0 0
point x0 is called an essential discontinuity of f .

66
Example 185 The function

sin x1 if x 6= 0;
f (x) :=
1 if x = 0;

and
log x if x > 0;
g (x) :=
1 if x = 0;
have an essential discontinuity at x = 0.

Monday, October 29, 2012

16 Important Theorems on Continuity

In this section we study some important consequences of continuity. The next
theorem shows that continuity preserves the sign of a function.

Theorem 186 Let E R ,and let f : E ! R be continuous at some x0 2 E

with f (x0 ) > 0. Then there exists = ("; x0 ) > 0 such that for all x 2 E with
jx x0 j < we have
f (x) > 0:

Proof. This follows from 186.

Remark 187 A similar result continues to hold if f (x0 ) < 0.

Example 188 The previous theorem implies in particular that sets of the form

fx 2 R : 4 sin x log (1 + jxj) > 0g

are open. We used this in the exercises.

More generally, we have the following theorem.

De…nition 189 Given a set E R, a set F E is said to be relatively open

in E if there exists an open set U R such that F = U \ E. A set G E
is said to be relatively closed in E if there exists a closed set C R such that
G = C \ E.

Theorem 190 Let E R and let f : E ! R.

1
(i) The function f is continuous if and only if f (U ) is relatively open for
every open set U R.
1
(ii) The function ,f is continuous if and only if f (C) is relatively closed for
every closed set C R.

Proof. Exercise.

67
Remark 191 The previous characterization of continuous functions is useful
to de…ne continuity in a topological space.
Another important theorem on continuity is the following.
Theorem 192 (Zeros of a continuous function) Let I R be an interval
and let f : I ! R be a continuous function. If there exist x1 ; x2 2 I such that
f (x1 ) < 0 and f (x2 ) > 0, then there exists x0 2 I in the interval of endpoints
x1 ; x2 such that f (x0 ) = 0.
Proof. Without loss of generality, we may assume that x1 < x2 , the case
x1 > x2 is similar. Then [x1 ; x2 ] I, and so we may de…ne the set
E := fx 2 [x1 ; x2 ] : f (x) < 0g :
Let x0 := sup E. Then x0 2 [x1 ; x2 ] and f (x) 0 for all x 2 (x0 ; x2 ].
We claim that x1 < x0 < x2 . Indeed, since f (x1 ) < 0, by Theorem 186,
we can …nd 1 > 0 such that f (x) < 0 for all x 2 I \ (x1 1 ; x1 + 1 ). Note
that this imply that 1 x2 x1 . It follows that x0 x1 + > x1 . Similarly,
since f (x2 ) > 0, by Theorem 186, we can …nd 2 > 0 such that f (x) > 0 for all
x 2 I \ (x2 2 ; x2 + 2 ), so that 2 x2 x1 and x0 x2 2 < x2 .
Next, we claim that f (x0 ) = 0. Indeed, if f (x0 ) < 0, then by Theorem 186,
we can …nd > 0 such that f (x) < 0 for all x 2 I \ (x0 0 ; x0 + 0 ). Since
x0 < x2 , taking 0 < x2 x0 , this implies that f (x) < 0 for all x 2 [x0 ; x0 + 0 ),
which contradicts the fact that x0 is the supremum of E.
On the other hand, if f (x0 ) > 0, then again by Theorem 186, we can …nd
> 0 such that f (x) > 0 for all x 2 I \(x0 0 ; x0 + 0 ). Since x0 < x2 , taking
0 < x0 x 1 , this implies that f (x) > 0 for all x 2 (x0 0 ; x0 ]. Together with
the fact that f (x) 0 for all x 2 (x0 ; x2 ], it follows that E [x1 ; x0 0 ], and
so its supremum cannot be x0 .This shows that f (x0 ) = 0.
Remark 193 By applying the previous theorem to the function g (x) = f (x) t,
we can show that if I R is an interval and f : I ! R a continuous function
such that f (x1 ) < t and f (x2 ) > t for some x1 ; x2 2 I, then there exists x0 2 I
in the interval of endpoints x1 ; x2 such that f (x0 ) = t.
Corollary 194 Let I R be an interval and let f : I ! R be a continuous
function. Then f takes all the values between inf I f and supI f . Moreover, f (I)
is an interval (possibly degenerate).
Proof. If inf I f = supI f , then f is constant. In this case f (I) is a singleton
and there is nothing to prove.
Thus, in what follows we assume that inf I f < supI f and let inf I f < t <
supI f . By the de…nition of in…mum and of supremum, there exist x1 ; x2 2 I
such that f (x1 ) < t and f (x2 ) > t. By the previous remark, there exists
x0 2 I in the interval of endpoints x1 ; x2 such that f (x0 ) = t. This shows
that f (I) (inf I f; supI f ). It remains to show that f (I) is an interval. Let
y1 ; y2 2 f (I) with y1 < y2 and let y1 < y < y2 . Since inf I f y1 < y2 supI f ,
by what we just proved, there exists x 2 I such that f (y) = x.

68
Corollary 195 Let I R be an interval and let f : I ! R be a monotone
function. Then f is continuous if and only if f (I) is an interval.

Proof. If f is continuous, then the result follows from the previous corollary.
Assume that f (I) is an interval. We claim that f is continuous. Let x0 2 I and
assume that x0 is an interior point (the case in which x0 is an endpoint of I is
similar). Assume that f is not continuous at x0 . Then by Theorem 161 there
exist
lim f (x) = ` 2 [ 1; 1] < lim+ f (x) = L 2 [ 1; 1] :
x!x0 x!x0

Since f is increasing, we have that

` = sup f (x) ; L = inf f (x) ;

x<x0 x>x0

and so f (x) ` for all x < x0 , while f (x) L for all x0 . This implies that
f (I) does not contain the set (`; L) n ff (x0 )g, which contradicts the fact that
f (I) is an interval because f takes values less than or equal to ` and values
greater than or equal to L.
Wednesday, October 31, 2012
Next we show that continuous functions preserve compactness.

Proposition 196 Consider a continuous function f : E ! R, where E R.

Then f (K) is sequentially compact for every sequentially compact set K E.

Proof. Let K E be sequentially compact. By Theorem 135, it is enough

to show that f (K) is sequentially compact. Let fyn g f (K). Then for
every n 2 N there exists xn 2 K such that f (xn ) = yn . Then there exist a
subsequence fxnk g and x0 2 K such that xnk ! x0 . By the continuity of f ,

ynk = f (xnk ) ! f (x0 ) 2 f (K) :

The following theorem is important.

Theorem 197 (Weierstrass) Let K R be sequentially compact and let f :

K ! R be a continuous function. Then there exists x0 ; x1 2 K such that

f (x0 ) = min f (x) ; f (x1 ) = max f (x) :

x2K x2K

Proof. Let
t := inf f (x) :
x2K

By the previous proposition, t is a real number. For every n 2 N, t + n1 is not

a lower bound, and so by the de…nition of in…mum, we may …nd xn 2 K such
that
1
t < f (xn ) < t + :
n

69
Letting n ! 1, by the squeeze theorem, we get
lim f (xn ) = t: (23)
n!1

Since fxn g K, and K is sequentially compact (see Theorem 135), there exist
a subsequence fxnk g of fxn g and x 2 K such that xnk ! x as k ! 1. Using
the continuity of f and (23), we get
t = lim f (xn ) = lim f (xnk ) = f (x) ;
n!1 k!1

which shows that the in…mum is a minimum.

The sequence fxn g constructed in the previous proof is called a minimizing
sequence.
Remark 198 Note that to prove the existence of a minimum, we only used a
weaker form of continuity, namely that the set
lim inf f (xj ) f (x)
j!1

for all sequences fxj g converging to x 2 R. A function satisfying this property

is called sequentially lower semicontinuous.
Remark 199 A typical application of the Weierstrass theorem is the following.
Let f : R ! R be a continuous function. Assume that f is bounded from below,
so that
` = inf f (x) > 1
x2R
and that
lim f (x) = 1:
jxj!1

By the de…nition of limit, we can …nd R > 0 such that f (x) > ` for all x 2 R
such that jxj R. Thus,
` = inf f (x) = inf f (x) :
x2R x2[ R;R]

By the Weierstrass theorem, f has a minimum in [ R; R] and we are done.

We now discuss the continuity of inverse functions and of composite func-
tions. If a continuous function f is invertible its inverse function f 1 may not
be continuous.
Example 200 Let
x if 0 x 1;
f (x) :=
x 1 if 2 < x 3:
1
Then f : [0; 2] ! R is given by

1 x if 0 x 1;
f (x) :=
x+1 if 1 < x 2;
which is not continuous at x = 1.

70
 We will see that this cannot happen if E is an interval or a sequentially
compact set.

Theorem 201 Let K R be a sequentially compact set and let f : K ! R

be one-to-one and continuous. Then the inverse function f 1 : f (K) ! R is
continuous.

Proof. Let y0 2 f (K) and let yn 2 f (K) be such that yn ! y0 . To prove

that f 1 is continuous at y0 , we need to show that there exists
1 1
lim f (yn ) = f (y0 ) :
n!1

Assume by contradiction that this does not happen. Then there exists " > 0
such that
f 1 (yn ) f 1 (y0 ) > "
for in…nitely many n, so we can …nd a subsequence f 1 (ynk ) k for which the
previous inequality holds. Let xk := f 1 (ynk ) 2 K. Since K is sequentia;;y
compact, there exist a subsequence fxki g and x 2 K such that xki ! x. Note
that, since
xki f 1 (y0 ) = f 1 ynki f 1 (y0 ) > ";

letting i ! 1 gives
1
x f (y0 ) ";
1
and so x 6= f (y0 ). On the other hand, since f is continuous, it follows that

lim f (xki ) = f (x) :

i!1

But f (xki ) = f f 1 ynki = ynki ! y0 , and so y0 = f (x), which is a

contradiction since f is injective.
Friday, November 02, 2012

Theorem 202 Let I R be an interval and let f : I ! R be one-to-one and

continuous. Then the inverse function f 1 : f (I) ! R is continuous.

Proof. Step 1: Let’s prove that if x0 2 I then for every x 2 I with

x > x0 , f (x) f (x0 ) always has the same sign. Fix x1 2 I with x1 > x0 and
say that f (x1 ) f (x0 ) > 0 (the case f (x1 ) f (x0 ) < 0 is similar). We claim
that f (x) f (x0 ) > 0 for all x 2 I with x > x0 . If not, then there exists x2 2 I
with x2 > x0 such that f (x2 ) f (x0 ) < 0. But then by Remark 193 (with
t = f (x0 )) we would …nd x3 2 I in the interval of endpoints x1 and x2 such
that f (x3 ) = f (x0 ), which contradicts the fact that f is one-to-one.
In a similar way we can show that for every x 2 I with x < x0 , f (x) f (x0 )
always has the same sign.
Step 2: Let’s prove that f is monotone. Fix a; b 2 I with a < b and assume
that f (b) > f (a) (the case f (b) < f (a) is similar). We claim that f is strictly

71
increasing. Let x1 ; x2 2 I with x1 < x2 and let b1 = max fx2 ; bg. Since
f (b) > f (a), by Step 1 (with x0 = a), we have that f (x) > f (a) for all x 2 I
with x > a. In particular, f (b1 ) > f (a). Again by Step 1 (with x0 = b1 ), we
have that f (b1 ) > f (x) for all x 2 I with x < b1 . In particular, f (b1 ) > f (x1 ).
By Step 1, once more, (with x0 = x1 ), we have that f (x) > f (x1 ) for all x 2 I
with x > x1 . In particular, f (x2 ) > f (x1 ), which proves that f is strictly
increasing.
Step 3: Since f is continuous and I is an interval, by Corollary 194, f (I) is an
interval. Moreover, since f is monotone, it follows (exercise) that f 1 : f (I) !
R is also monotone. By Corollary 195 (applied to the function f 1 ), it follows
that f 1 is continuous.

Example 203 In view of the previous theorem and of Exercise 172, the func-
tions arccos x, arcsin x, arctan x are continuous.
Given a > 0, the function loga x is continuous for x > 0, since it is the
inverse of ax . p
Given n 2 N, the function 2n+1 p x, x 2 R, is continuous, since it is the
inverse of x2n+1 . The function 2n x, x 2 [0; 1), is continuous, since it is the
inverse of x2n .
Given a > 0, since ex and log x are continuous in (0; 1), by writing
a
xa = elog x = ea log x ;
x
xx = elog x = ex log x ;

it follows from Theorems 173 and 175, that xa and xx are continuous in (0; 1).

17 Uniform Continuity
Next we introduce the notion of uniform continuity.

De…nition 204 Consider a function f : E ! R, where E R. The function

f is said to be uniformly continuous if for every " > 0 there exists = (") > 0
such that
jf (x) f (y)j "
for all x; y 2 E with jx yj .

Remark 205 To negate uniform continuity it is enough to …ns two sequences

fxn g ; fyn g E such that

lim jxn yn j = 0
n!1

and jf (xn ) f (yn )j 9 0 (so either the limit does not exist or it exists but it is
not zero).

72
Example 206 The function f (x) = x, x 2 R, is uniformly continuous, while
the function g (x) = x2 , x 2 R, is not. To see this, take " = for the function
f . To prove that g is not uniformly continuous, consider the two sequences
xn = n + n1 and yn = n. Then xn yn = n1 ! 0, while
2
1 1
f (xn ) f (yn ) = n+ n2 = 2 + ! 2 6= 0;
n n
which implies that g is not uniformly continuous, by the previous remark.
Example 207 The continuous function f (x) = sin x1 is not uniformly contin-
1
uous, by the previous remark. Consider the two sequences xn = +2n and
2
1
yn = 3
+2n
. Then xn ! 0, yn ! 0, so xn yn ! 0 while
2

f (xn ) f (yn ) = 1 ( 1) ! 2 6= 0;
Example 208 The continuous function f (x) = x1 is not uniformly continuous,
by the previous remark. Consider the two sequences xn = n1 and yn = n+1 1
.
Then xn ! 0, yn ! 0, so xn yn ! 0 while
f (xn ) f (yn ) = n (n + 1) ! 1 6= 0;
Monday, November 05, 2012
Solutions of Exercises 3 and 4 in Homework #6
Wednesday, November 07, 2012
Simple examples of uniformly continuous functions are Lipschitz continuous
functions.
De…nition 209 Consider a function f : E ! R, where E R. The function
f is said to be Lipschitz continuous if there exists L > 0 such that
jf (x) f (y)j L jx yj
for all x; y 2 E.
Remark 210 To negate Lipschitz continuity it is enough to …ns two sequences
fxn g ; fyn g E with xn 6= yn such that
f (xn ) f (yn )
! 1:
xn yn
Proposition 211 Let E R and let f : E ! R be Lipschitz continuous. Then
f is uniformly continuous.
Proof. Since f is Lipschitz continuous, there exists L > 0 such that
jf (x) f (y)j L jx yj
for all x; y 2 E.
"
Fix " > 0, and take = L. If x; y 2 E with jx yj , then
"
jf (x) f (y)j L jx yj L =L = ":
L

73
 p
Example 212 The function f (x) = x, x 2 [0; 1], is uniformly continuous
(we will see this later), but not Lipschitz. Indeed, let xn = 0 and yn = n1 . Then
q
1
f (xn ) f (yn ) 0 n p
= 1 = n ! 1;
xn yn 0 n

which shows that f is not Lipschitz continuous.

Proposition 213 If I R is an interval and f : I ! R is di¤ erentiable in I

and the derivative f 0 is bounded, then f is Lipschitz continuous.

Proof. By hypothesis, there exists M > 0 such that jf 0 (x)j M for all
x 2 I. Let x; y 2 I, with, say y < x. By the mean value theorem (we will prove
this later), there exists z 2 (x; y) such that

f (x) f (y) = f 0 (z) (x y) :

Hence,
jf (x) f (y)j = jf 0 (z)j jx yj M jx yj ;
which shows that f is Lipschitz continuous.

Example 214 The function f (x) = jxj is Lipschitz continuous, since

jf (x) f (y)j = jjxj jyjj jx yj ;

but it is not di¤ erentiable in x = 0.

Remark 215 Thus we have shown that if I R is an interval and f : I ! R,

then

f di¤ erentiable with bounded derivative =) f Lipschitz continuous

=) f uniformly continuous =) f continuous

but none of the opposite implications is true.

Example 216 The functions f (x) = cos x, f (x) = sin x, f (x) = arctan x all
have bounded derivatives, and so they are Lipschitz continuous, and, in turn,
uniformly continuous.
p
To prove that the function f (x) = x, x 2 [0; 1], is uniformly continuous in
[0; 1], we apply the following theorem:

Theorem 217 Let K R be sequentially compact and let f : K ! R be a

continuous function. Then f is uniformly continuous.

74
 Proof. Assume by contradiction that f is not uniformly continuous. Then
there exist "0 > 0 and two sequences fxn g ; fyn g K such that
lim jxn yn j = 0
n!1

and jf (xn ) f (yn )j > "0 . Since K is sequentially compact, there exist a sub-
sequence fxnk gk of fxn gn and x 2 K such that xnk ! x0 as k ! 1. It follows
that
jynk x0 j = jynk xnk x0 j jynk xnk j + jxnk x0 j ! 0
and so by the squeeze theorem ynk ! x0 . By continuity of f at x0 , we have that
f (xnk ) ! f (x0 ) and f (ynk ) ! f (x0 ), which implies that f (xnk ) f (ynk ) !
f (x0 ) f (x0 ) = 0. Hence, for all k su¢ ciently large
"0
jf (xnk ) f (ynk ) 0j ;
2
which contradicts the fact that jf (xnk ) f (ynk )j > "0 for all k.
p
Example 218 Since [0; 1] is sequentially compact and the function f (x) = x,
x 2 [0; 1], is continuous, it follows by the previous theorem that it is uniformly
continuous.
Example 219 We have seen that the continuous function f (x) = x2 , x 2 R, is
not uniformly continuous in R. However, by the previous theorem it is uniformly
continuous in every set [a; b].
Example 220 Next we study continuous functions of the type f (x) = sin x1 .
Consider the set E = (0; 1]. This set is not sequentially compact (it is bounded
but not closed), thus we cannot apply the previous theorem.
Example 221 Consider the function f (x) = sinx x in (0; 1]. Let’s prove that f
is uniformly continuous. Consider the function
sin x
x if 0 < x 1;
g (x) =
1 if x = 0:
Since limx!0 sinx x = 1, we have that g is continuous in [0; 1] and since [0; 1]
is sequentially compact, g is uniformly continuous. If we restrict g to (0; 1], it
remains uniformly continuous, and so f is uniformly continuous.
The previous example can be generalized.
Proposition 222 Let f : E ! R be uniformly continuous, where E R. Then
for every x 2 E n E there exists the limit
lim f (y) 2 R
y!x

and the function g : E ! R, de…ned by

f (x) if x 2 E;
g (x) :=
limy!x f (y) if x 2 E n E;
is uniformly continuous in E.

75
 Proof. Exercise.

Exercise 223 Given a function f : [0; 1) ! R, prove that if f is uniformly

continuous, then there exist A, B 2 R such that

jf (x)j A + Bx

for all x 0.

Exercise 224 Let f : [a; 1) ! R be a continuous function.

1. Prove that if g : [a; 1) ! R is uniformly continuous and

lim (f (x) g (x)) = 0

x!1

then f is uniformly continuous.

2. Prove that if there exist A, B 2 R such that

lim (f (x) A Bx) = 0

x!1

then f is uniformly continuous.

Friday, November 09, 2012

Second midterm.
Monday, November 12, 2012

18 Di¤erentiation
De…nition 225 Let E R and let x0 2 E be an accumulation point of E.
Given a function f : E ! R we say that f is di¤erentiable at x0 if there exists

f (x) f (x0 )
lim 2R
x!x0 x x0

In this case the limit is called derivative of f at x0 and is denoted f 0 (x0 ) or

df
(x0 ). If f is di¤ erentiable at every point of E \ acc E, we say that f is
dx
di¤erentiable on E.

De…nition 226 Given two functions f : E ! R and g : E ! R and a point

x0 2 acc E, we say that the function f is a little o of g as x ! x0 , and we write
f = o (g), if g 6= 0 in E and

f (x)
lim = 0:
x!x0 g (x)

76
 Hence, a little o of g is simply a function that goes to zero faster than g as
x ! x0 .
If f is di¤erentiable at x0 , then by setting x := x0 + h we may write

f (x0 + h) f (x0 ) = f 0 (x0 ) h + o (h) : (24)

Note that (24) expresses f (x0 + h) f (x0 ) as the sum of the linear functional

L:R!R
h 7! f 0 (x0 ) h

plus a small reminder. We can therefore regard the derivatives of f at x0 , not

as a real number, but as a linear operator on R that takes h to f 0 (x0 ) h. So an
alternative de…nition of di¤erentiability is the following

De…nition 227 Let E R and let x0 2 E be an accumulation point of E.

Given a function f : E ! R we say that f is di¤erentiable at x0 if there exists
a linear functional Lx0 : R ! R such that

f (x) f (x0 ) Lx0 (x x0 )

lim = 0:
x!x0 x x0
This alternative form will be useful when we de…ne di¤erentiability for func-
tions of several variables.

Theorem 228 Let E R and let x0 2 E be an accumulation point of E. If a

function f : E ! R is di¤ erentiable at x0 , then it is continuous at x0 .

Proof. For x 2 E, x 6= x0 , write

f (x) f (x0 )
f (x) f (x0 ) = (x x0 ) :
x x0
Then by the product of limits

f (x) f (x0 )
f (x) f (x0 ) = (x x0 ) ! f 0 (x0 ) 0
x x0
as x ! x0 .
The converse of this theorem is not true.
p
Example 229 (a) The functions f (x) := jxj and g (x) := x are continuous
but not di¤ erentiable at x = 0.
(b) By direct application of the de…nitions one easily proves that the derivative
of any constant is clearly zero and that if f (x) := x then f 0 (x) = 1. The
basic elementary functions sin x, cos x, ex are known to satisfy
0 0 0
(sin x) = cos x; (cos x) = sin x; (ex ) = ex :

77
 0
To prove that (ex ) = ex we need a few preliminary results.

Lemma 230
log (1 + x)
lim = 1:
x!0 x
1
Proof. If x ! 0+ , then y = x ! 1, and so changing variables,
y
1 1
lim (1 + x) x = lim 1+ = e;
x!0+ y!1 y
1
while if x ! 0 , then y = x ! 1, and so changing variables,
y
1 1
lim+ (1 + x) x = lim 1+ = e:
x!0 y! 1 y

Thus, by the continuity of log x,

log (1 + x) 1
= log (1 + x) x ! log e = 1:
x

ex 1
Lemma 231 limx!0 x = 1:

Proof. Put y = ex 1. Then y + 1 = ex and so log (y + 1) = x. If x ! 0,

then y ! 0, and so changing variables,
ex 1 y
lim = lim =1
x!0 x y!0 log (y + 1)

0
Theorem 232 (ex ) = ex .

Proof. Let x0 2 R. Then for x 6= x0 ,

f (x) f (x0 ) ex ex0 ex x0 1

= = ex0 :
x x0 x x0 x x0
Let t = x x0 . When x ! x0 , we have that t ! 0, and so

ex ex0 et 1
lim = lim ex0 = ex0 1;
x!x0 x x0 t!0 t
where we have used the previous lemma.
0
Theorem 233 (sin x) = cos x.

78
 Proof. Let x0 2 R. Then for x 6= x0 ,
x x0
f (x) f (x0 ) sin x sin x0 sin 2 x + x0
= =2 cos :
x x0 x x0 x x0 2
x x0
Let t = 2 . When x ! x0 , we have that t ! 0, and so
x x0
sin 2 sin t
lim x x0 = lim = 1:
x!x0 t!0 t
2

Hence, as x ! x0 ,

f (x) f (x0 ) sin x 2x0 x + x0 x0 + x0

= x x0 cos ! 1 cos = cos x0 ;
x x0 2
2 2

where we have used Theorem 151 and Exercise 154.

Theorem 234 (Weierstrass) Let 0 < a < 1 and let b 2 N be an odd integer
such that ab > 1 + 32 . Then the function
1
X
f (x) := an cos (bn x) ; x 2 R;
n=0

is continuous, but nowhere di¤ erentiable.

We now list some elementary operations for derivatives.

Theorem 235 Let E R and let x0 2 E be an accumulation point of E.

Given two functions f , g : E ! R assume that f and g are di¤ erentiable at x0 .
Then
0
(a) f + g is di¤ erentiable at x0 and (f + g) (x0 ) = f 0 (x0 ) + g 0 (x0 );
0
(b) f g is di¤ erentiable at x0 and (f g) (x0 ) = f 0 (x0 ) g (x0 ) + f (x0 ) g 0 (x0 );
f
(c) if g (x0 ) 6= 0 then restricted to the set F := fx 2 E : g (x) 6= 0g di¤ er-
g
entiable at x0 and
0
f f 0 (x0 ) g (x0 ) f (x0 ) g 0 (x0 )
(x0 ) = :
g g 2 (x0 )

Example 236 Repeated application of (b) and (c) shows that if k 2 Z then the
0
function f (x) = xk is di¤ erentiable with f 0 (x) = kxk 1 and that (tan x) = Of course, if
2
tan x + 1. k < 0 we have
to restrict our-
We now study the di¤erentiation of inverse and composite functions. We selves to x 6= 0
begin with composite functions.

79
Theorem 237 (Chain rule) Let E, F R and let x0 2 E be an accumulation
point of E. Given two functions f : E ! F and g : F ! R assume that f is
di¤ erentiable at x0 , that f (x0 ) is an accumulation point of F and that g is
di¤ erentiable at f (x0 ). Then g f : E ! R is di¤ erentiable at x0 and
0
(g f ) (x0 ) = g 0 (f (x0 )) f 0 (x0 ) :

Proof. Consider the function

(
g(y) g(f (x0 ))
y f (x0 ) if y 6= f (x0 ) ;
h (y) :=
g 0 (f (x0 )) if y = f (x0 ) :

Since g is di¤erentiable at x0 , we have that limy!f (x0 ) h (y) = h (f (x0 )) =

g 0 (f (x0 )). For x 2 E, x 6= x0 , write

g (f (x)) g (f (x0 )) f (x) f (x0 )

= h (f (x)) :
x x0 x x0
Since f is di¤erentiable at x0 , it is continuous at x0 . But h is continuous
at f (x0 ), thus the composition h f is continuous at x0 . It follows that
limx!x0 h (f (x)) = h (f (x0 )) = g 0 (f (x0 )). Hence, by the product of limits

g (f (x)) g (f (x0 )) f (x) f (x0 )

= h (f (x)) ! g 0 (f (x0 )) f 0 (x0 )
x x0 x x0
as x ! x0 .
Wednesday, November 14, 2012

Example 238 The function

(
1
x2 sin if x 6= 0;
f (x) := x
0 if x = 0

is continuous in R since by Example 149

1
lim f (x) = lim x2 sin = 0 = f (0) :
x!0 x!0 x
Using Theorem 233 and Theorems 235 and 237 for x 6= 0 we have

d 1 1 1
x2 sin = 2x sin cos :
dx x x x

For x = 0 we cannot apply the previous theorems and so we have to use the
de…nition. We have
1
f (x) f (0) x2 sin 0 1
= x = x sin ! 0
x 0 x 0 x

80
as x ! 0 again by Example 149 and so f 0 (0) = 0. Note that

2x sin x1 cos x1 if x 6= 0;
f 0 (x) =
0 if x = 0

is not continuous at x = 0 since the limit

1 1
lim 2x sin cos
x!0 x x

does not exist (exercise).

Remark 239 (Important) In the exercises, when checking the di¤ erentiabil-
ity of a function f , one has to distinguish between “good” points in the domain,
where one can apply Theorems 235, 237, and ??, and “bad” points in the do-
main, where one has to use the de…nition. Examples of bad points are points
where an absolute vaue is zero (for example, for f (x) = jsin xj all points where
sin x = 0, that is, x = 2k , are badppoints), or where the argument of an n-root
3
is zero (for example, for f (x) = sin2 x, all points where sin2 x = 0, that is,
x = 2k , are bad points), or points where the law of the function changes as
x = 0 in the previous example.

Theorem 240 Let E R and let x0 2 E be an accumulation point of E.

Given a function f : E ! R assume that f is one-to-one and di¤ erentiable at
x0 . Suppose also that the inverse function f 1 : f (E) ! R is continuous at
f (x0 ). Then f (x0 ) is an accumulation point of f (E) and f 1 is di¤ erentiable
at f (x0 ) if and only if f 0 (x0 ) 6= 0 and in this case

1 0 1
f (f (x0 )) = :
f 0 (x0 )

Proof. We …rst prove that f (x0 ) is an accumulation point of the set f (E).
Since x0 is an accumulation point of E by Remark 50 there exists a sequence
fxn g E n fx0 g which converges to x0 . De…ne yn := f (xn ). Since f is one-
to-one and fxn g E n fx0 g we have that fyn g f (E) n ff (x0 )g. By the
continuity of f in x0 (which follows from Theorem 228) and Theorems 171 and
143 we have that yn = f (xn ) ! f (x0 ). Hence f (x0 ) is an accumulation point
of the set f (E).
Next we show that f 1 : f (E) ! R is di¤erentiable at f (x0 ) if and only if
0
f (x0 ) 6= 0. Consider the quotient
1
f (y) f 1 (f (x0 ))
y f (x0 )

for y 2 f (E). For every y 2 f (E) there exists a unique x 2 E such that
f (x) = y and so we may write
1
f (y) f 1 (f (x0 )) x x0
= :
y f (x0 ) f (x) f (x0 )

81
By assumption the function f 1 is continuous at f (x0 ) and so x = f 1 (y) !
x0 = f 1 (f (x0 )) as y ! f (x0 ). Hence since f is di¤erentiable at x0 we have
that
f 1 (y) f 1 (f (x0 )) 1 1
= ! 0
y f (x0 ) f (x) f (x0 ) f (x0 )
x x0
as y ! f (x0 ).

Remark 241 Note that by writing y0 = f (x0 ), the previous theorem says that

df 1 1
(y0 ) = 0 :
dy f (f 1 (y0 ))

Exercise 242 Using the previous theorem prove that

0 1 0 1
(arccos x) = p ; (arcsin x) = p ;
1 x2 1 x2
0 1 0 1
(arctan x) = 2 ; (log x) = :
x +1 x
Next we study local minima and maxima.

De…nition 243 Let f : E ! R, where E R, and let x0 2 E. We say that

(i) f attains a local minimum at x0 if there exists r > 0 such that f (x)
f (x0 ) for all x 2 E \ B (x0 ; r),
(ii) f attains a global minimum at x0 if f (x) f (x0 ) for all x 2 E,
(iii) f attains a local maximum at x0 if there exists r > 0 such that f (x)
f (x0 ) for all x 2 E \ B (x0 ; r),
(iv) f attains a global maximum at x0 if f (x) f (x0 ) for all x 2 E.

Theorem 244 Let f : E ! R, where E R. If f attains a local minimum (or

maximum) at some interior point x0 2 E and if f is di¤ erentiable at x0 , then
f 0 (x0 ) = 0.

Remark 245 In view of the previous theorem, when looking for minima and
maxima, we have to search among the following:

Interior points at which f is di¤ erentiable and f 0 (x) = 0, these are called
critical points. Note that if f 0 (x0 ) = 0, the function f may not attain a
local minimum or maximum at x0 . Indeed, consider the function f (x) =
x3 . Then f 0 (0) = 0, but f is strictly increasing, and so f does not attain
a local minimum or maximum at 0.
Interior points at which f is not di¤ erentiable. The function f (x) = jxj
attains a global minimum at x = 0, but f is not di¤ erentiable at x = 0.

82
 Boundary points.

Friday, November 16, 2012

Proof. Assume that f attains a local minimum (the case of a local maximum
is similar). Then there exists r > 0 such that f (x) f (x0 ) for all x 2 E \
B (x0 ; r). Since x0 2 E is an interior point, by taking r smaller, we can assume
that B (x0 ; r) E so that f (x) f (x0 ) for all x 2 B (x0 ; r). If x > x0 , then
f (x) f (x0 ) 0 and so
f (x) f (x0 )
0:
x x0
Letting x ! x+ 0 and using the fact that f is di¤erentiable at x0 , we get that
f 0 (x0 ) 0.
If x < x0 , then f (x) f (x0 ) 0 and so
f (x) f (x0 )
0:
x x0
Letting x ! x0 and using the fact that f is di¤erentiable at x0 , we get that
f 0 (x0 ) 0. This shows that f 0 (x0 ) = 0.

Remark 246 If x0 2 E is not an interior point, then it is on the boundary. In

this case it is either an isolated point or an accumulation point. In the …rst case
it makes not sense to talk about f 0 (x0 ), since we cannot take limits. On the
other end, it x0 2 acc E and f is di¤ erentiable at x0 , then If f attains a local
minimum at x0 we can adapt the previous proof to show that f 0 (x0 ) 0 if there
exists a sequence fxn g E n fx0 g such that xn ! x+ 0
0 , while f (x0 ) 0 if there
exists a sequence fxn g E n fx0 g such that xn ! x0 . Similar conclusions can
be reached for local maxima (reversing the inequalities).

An important application of the previous theorem is given by the following

result.

Theorem 247 (Rolle) Let f : [a; b] ! R be continuous in [a; b] and di¤ eren-
tiable in (a; b). If f (a) = f (b), then there exists c 2 (a; b) such that f 0 (c) = 0.

Proof. Since [a; b] is sequentially compact, by the Weierstrass theorem, f

has a global maximum and a global minimum in [a; b]. If

max f = min f;
[a;b] [a;b]

then f is constant, and so f 0 (x) = 0 for all x 2 (a; b). If max[a;b] f > min[a;b] f ,
then since f (a) = f (b), there f admits one of them at some interior point
c 2 (a; b). By the previous theorem, f 0 (c) = 0.

Theorem 248 (Lagrange or Mean Value Theorem) Let f : [a; b] ! R be

continuous in [a; b] and di¤ erentiable in (a; b). Then there exists c 2 (a; b) such
that
f (b) f (a) = f 0 (c) (b a) :

83
 Proof. The function
f (b) f (a)
g (x) = f (x) (x a)
b a
is continuous in [a; b], di¤erentiable in (a; b), and g (a) = g (b) = f (a). Hence, we
are in a position to apply Rolle’s theorem to …nd c 2 (a; b) such that g 0 (c) = 0,
or, equivalently,
f (b) f (a)
0 = f 0 (x) 1 :
b a

Corollary 249 Let f : [a; b] ! R be continuous in [a; b] and di¤ erentiable in

(a; b) with f 0 bounded. Then f is Lipschitz continuous.

Proof. Let L > 0 be such that jf 0 (x)j L for all x 2 (a; b). By the mean
value theorem, for all x; y 2 [a; b] with x < y, there exists z 2 (x; y) such that

f (y) f (x) = f 0 (z) (y x) :

Hence,
jf (y) f (x)j L (y x) ;
which shows that f is Lipschitz continuous.

Corollary 250 Let f : (a; b) ! R be di¤ erentiable in (a; b). If f 0 (x) = 0 for
all x 2 (a; b), then f is constant.

Proof. By the mean value theorem, for all x; y 2 [a; b] with x < y, there
exists z 2 (x; y) such that

f (y) f (x) = f 0 (z) (y x) = 0:

Hence, f (y) f (x) = 0, which shows that f is constant.

Remark 251 If A R is open and f : (a; b) ! R be di¤ erentiable in A with

f 0 (x) = 0 for all x 2 A, then we can decompose A into a countable number
of disjoint open intervals. By applying the previous result in each interval, we
conclude that f is constant in every interval (the constant may change from
interval to interval).

Corollary 252 Let f : (a; b) ! R be di¤ erentiable in (a; b). Then f is increas-
ing if and only if f 0 (x) 0 for all x 2 (a; b).

Proof. If f is increasing, then f (x) f (x0 ) for x > x0 , and so

f (x) f (x0 )
0:
x x0

Letting x ! x+ 0
0 , we get f (x0 ) 0.

84
 Conversely, if f 0 0, by the mean value theorem, for all x; y 2 [a; b] with
x < y, there exists z 2 (x; y) such that
f (y) f (x) = f 0 (z) (y x) 0:
Hence, f (y) f (x).
Remark 253 If f 0 > 0 is (a; b), then with the same proof we can show that f is
strictly increasing, but the opposite is not true. Indeed, the function f (x) = x3
is strictly increasing but f 0 (x) = 3x2 , which is zero for x = 0.
Remark 254 If f 0 (x0 ) > 0, we cannot conclude that f is increasing near x0
but only that f (x) < f (x0 ) for x0 < x < x0 and that f (x) > f (x0 ) for
x0 < x < x0 + for some small > 0. Indeed, consider the function
x + 2x2 sin x1 if 0 < x 1;
f (x) =
0 if x = 0:
Let’s look at the di¤ erentiability at x = 0. We have
f (x) f (0) x + 2x2 sin x1 0 1
= = 1 + 2x sin !1
x 0 x 0 x
as x ! 0, since 0 2x sin x1 j2xj ! 0. Hence,
1 + 4x sin x1 2 cos x12 if x 6= 0;
f 0 (x) =
0 if x = 0:
Hence f 0 (0) = 1 but f is not increasing near x = 0. Indeed, if it were, then
f 0 0 in ( ; ) for some small > 0. However, since 4x sin x1 ! 0 as x ! 0,
we have that
1 1
f 0 (x) = 1 + 4x sin 2 cos 2
x x
oscillates between 1 and 1 as x ! 0. Note that this is also an example of a
di¤ erentiable function, whose derivative is discontinuous.
Theorem 255 (Cauchy) Let f : [a; b] ! R and g : [a; b] ! R be continuous
in [a; b] and di¤ erentiable in (a; b) with g 0 (x) 6= 0 for all x 2 (a; b). Then there
exists c 2 (a; b) such that
f (b) f (a) f 0 (c)
= 0 :
g (b) g (a) g (c)
Proof. The function
h (x) = f (x) (g (b) g (a)) g (x) (f (b) f (a))
is continuous in [a; b], di¤erentiable in (a; b), and h (a) = h (b) = f (a) g (b)
g (a) f (b). Hence, we are in a position to apply Rolle’s theorem to …nd c 2 (a; b)
such that h0 (c) = 0, or, equivalently,
0 = f 0 (c) (g (b) g (a)) g 0 (c) (f (b) f (a)) :

85
Remark 256 Note that by Rolle’s theorem, g (b) 6= g (a).

Monday, November 19, 2012

Theorem 257 (De l’Hôpital) Let f : [a; b] ! R and g : [a; b] ! R be con-

tinuous in [a; b] and assume that f (x0 ) = g (x0 ) = 0 for some x0 2 [a; b].
Assume that f and g are di¤ erentiable in (a; b) n fx0 g with g (x) ; g 0 (x) 6= 0 for
all x 2 (a; b) n fx0 g. If there exists
f 0 (x)
lim = ` 2 [ 1; 1] ;
x!x0 g 0 (x)
then there exists
f (x)
lim = `:
x!x0 g (x)
Proof. Let fxn g (a; b) n fx0 g be such that xn ! x0 . Apply Cauchy’s
theorem in the interval of endpoints xn and x0 to …nd yn between xn and x0
such that
f (xn ) 0 f (xn ) f (x0 ) f 0 (yn )
= = 0 :
g (xn ) 0 g (xn ) g (x0 ) g (yn )
As xn ! x0 , we have that yn ! x0 , and so by hypothesis
f (xn ) f 0 (yn )
= 0 ! `:
g (xn ) g (yn )
Since this is true for every sequence fxn g converging to x0 , it follows that there
exists
f (x)
lim = `:
x!x0 g (x)

Example 258 Let’s calculate

sin x x
lim :
x!0 x3
Consider the functions f (x) = sin x x and g (x) = x3 . Then f (0) = g (0) = 0
and both functions are di¤ erentiable. We have that f 0 (x) = cos x 1 and
g 0 (x) = 3x2 . Let’s calculate
f 0 (x) cos x 1
lim = lim : (25)
x!0 g 0 (x) x!0 3x2

We get 00 . Consider the functions f1 (x) = cos x 1 and g1 (x) = 3x2 . Then
f1 (0) = g1 (0) = 0 and both functions are di¤ erentiable. We have that f10 (x) =
sin x and g10 (x) = 6x. Let’s calculate
f10 (x) sin x 1
lim 0 (x) = x!0
lim = : (26)
x!0 g1 6x 6

86
By (26) and De l’Hôpital’s theorem, there exists

f1 (x) cos x 1 1
lim = lim = :
x!0 g1 (x) x!0 3x2 6

Finally, by (25) and De l’Hôpital’s theorem, there exists

f (x) sin x x 1
lim = lim = :
x!0 g (x) x!0 x3 6

Remark 259 The converse of De l’Hôpital’s theorem does not hold, that is, if
0
there exists limx!x0 fg(x)
(x)
, we cannot conclude that there exists the limit limx!x0 fg0 (x)
(x)
.
To see this, take
x2 sin x1 if x 6= 0;
f (x) =
0 if x = 0;
and g (x) = x. Then f and g are continuous and for x 6= 0, have that f 0 (x) =
2x sin x1 x2 1 1 0
x2 cos x and g (x) = 1. Then

f (x) x2 sin x1 1
lim = lim = lim x sin = 0;
x!0 g (x) x!0 x x!0 x

since 0 x sin x1 jxj ! 0, but the limit

f 0 (x) 2x sin x1 + cos x1

lim = lim
x!0 g 0 (x) x!0 1

does not exist (it oscillates between 1 and 1).

Another version of De l’Hôpital’s theorem is the following:

Theorem 260 (De l’Hôpital) Let f : [a; b] n fx0 g ! R and g : [a; b] n fx0 g !
R be continuous in [a; b] n fx0 g and assume that

lim jf (x)j = lim jg (x)j = 1

x!x0 x!x0

and that f and g are di¤ erentiable in (a; b) n fx0 g with g (x) ; g 0 (x) 6= 0 for all
x 2 (a; b) n fx0 g. If there exists

f 0 (x)
lim = ` 2 [ 1; 1] ;
x!x0 g 0 (x)

then there exists

f (x)
lim = `:
x!x0 g (x)

The proof is left as an exercise.

87
Exercise 261 Calculate
lim xa log x;
x!0+

where a > 0.

Exercise 262 Calculate

lim xsin x :
x!0+

Another version of the previous theorem is the following.

Theorem 263 (De l’Hôpital) Let f : [a; 1) ! R and g : [a; 1) ! R be

continuous in [a; 1) and di¤ erentiable in (a; 1). Assume that

lim f (x) = lim g (x) = 0

x!1 x!1

and that g (x) ; g 0 (x) 6= 0 for all x 2 (a; 1). If there exists

f 0 (x)
lim = ` 2 [ 1; 1] ;
x!1 g 0 (x)

then there exists

f (x)
lim = `:
x!1 g (x)
The proof is similar and we omit it.

Remark 264 A similar result holds if limx!1 jf (x)j = limx!1 jg (x)j = 1.

Exercise 265 Prove the following:

xa
lim = 0;
x!1 logb x

where a > 0 and b 2 R,

xa
lim = 0;
x!1 bx

where a 2 R and b > 1.

Exercise 266 Let’s calculate

log x x
lim :
x!1 log (1 + ex )
Next we study Taylor’s formula.

De…nition 267 Given an open set U R and an integer n 2 N, a function

f : U ! R is said to be of class C n if f is di¤ erentiable up to order n with
continuous derivatives f 0 , f 00 , f 000 , f (4) , ,f (n) . The space of all functions of
class C n is denoted by C n (U ). A function of class C n for every n 2 N is said to
be of class C 1 and the space of all functions of class C 1 is denoted by C 1 (U ).

88
Theorem 268 (Taylor’s Formula) Let f 2 C (n) ((a; b)) and let x0 2 (a; b).
Then for every x 2 (a; b),

f 00 (x0 ) 2
f (x) = f (x0 ) + f 0 (x0 ) (x x0 ) + (x x0 )
2!
f (n) (x0 ) n
+ + (x x0 ) + Rn (x; x0 ) ;
n!
where the remainder Rn (x; x0 ) satis…es

Rn (x; x0 )
lim n = 0:
x!x0 (x x0 )

Lemma 269 Let g 2 C (n) ((a; b)) and let x0 2 (a; b). Then

g (x)
lim n =0 (27)
x!x0 (x x0 )

if and only if
g (x0 ) = g 0 (x0 ) = = g (n) (x0 ) = 0: (28)

Proof. Assume that (28) holds. By applying De l’Hôpital’s theorem several

times we get

g (x) g 0 (x) g (2) (x)

lim n = lim n 1 = lim n 2
x!x0 (x x0 ) x!x0 n (x x0 ) x!x0 n (n 1) (x x0 )
g (n 1) (x) g (n) (x) g (n) (x0 )
= = lim = lim = = 0:
x!x0 n! (x x0 ) x!x0 n!1 n!

Conversely, assume (27). If g (k) (x0 ) 6= 0 for some 0 k < n, then by what we
just proved (with k in place of n)

g (x) g (k) (x0 )

lim k
= 6= 0:
x!x0 (x x0 ) k!

On the other hand,

n k
g (x) g (x) (x x0 ) g (x) n k
k
= k n k
= n (x x0 ) !0
(x x0 ) (x x0 ) (x x0 ) (x x0 )

as x ! x0 , which is a contradiction.
We now turn to the proof of Theorem 268.
Proof of Theorem 268. Note that given a polynomial of degree n,
n
X
n i
p (x) = a0 + a1 (x x0 ) + + an (x x0 ) = ai (x x0 ) ;
i=0

89
we have that
n
X i k
p(k) (x) = i (i 1) (i k + 1) ai (x x0 ) ;
i=k

so that
p(k) (x0 ) = k!ak :
We apply the lemma to the function

g (x) := f (x) p (x)

to conclude that
f (x) p (x)
lim n =0
x!x0 (x x0 )
if and only if for all k = 0; : : : ; n,

0 = g (k) (x0 ) = f (k) (x0 ) p(k) (x0 ) = f (k) (x0 ) k!ak ;

that is,
f (k) (x0 )
ak = :
k!
Thus
f (n) (x0 ) n
g (x) = Rn (x; x0 ) = f (x) f (x0 ) + f 0 (x0 ) (x x0 ) + + (x x0 ) :
n!

De…nition 270 Given two functions f : E ! R and g : E ! R and a point

x0 2 acc E, we say that the function f is a little o of g as x ! x0 , and we write
f = o (g), if g 6= 0 in E and

f (x)
lim = 0:
x!x0 g (x)
Hence, a little o of g is simply a function that goes to zero faster than g as
x ! x0 . Hence, Taylor’s formula can be written as
1 00 2 1 000 3
f (x) = f (x0 ) + f 0 (x0 ) (x x0 ) + f (x0 ) (x x0 ) + f (x0 ) (x x0 )
2! 3!
1 (n) n n
+ + f (x0 ) (x x0 ) + o ((x x0 ) )
n!
as x ! x0 .
Wednesday, November 21, 2012
Thanksgiving, no classes
Friday, November 23, 2012
Thanksgiving, no classes
Monday, November 24, 2012

90
Corollary 271 Let f : (a; b) ! R be a function of class C n ((a; b)), n 2, and
let x0 2 (a; b) be such that

f 0 (x0 ) = = f (m 1)
(x0 ) = 0 (29)

for some 2 m n. Then the following hold:

(i) if m is even and f (m) (x0 ) > 0, then f has a local minimum at x0 ,
(ii) if m is even and f (m) (x0 ) < 0, then f has a local maximum at x0 ,
(iii) if m is odd and f (m) (x0 ) 6= 0, then f has neither a local minimum nor a
local maximum at x0 ,
(iv) if f 0 (x0 ) = = f (n) (x0 ) = 0, then anything can happen.

Proof. (i) For x close to x0 by Taylor’s formula of order m and center x0

1 00 2 1 000 3
f (x) = f (x0 ) + f 0 (x0 ) (x x0 ) + f (x0 ) (x x0 ) + f (x0 ) (x x0 )
2! 3!
1 (m) m
+ + f (x0 ) (x x0 ) + Rm (x) ;
m!
where the remainder Rm (x; x0 ) satis…es

Rm (x)
lim m = 0: (30)
x!x0 (x x0 )

By (29),
1 (m) m
f (x) = f (x0 ) + f (x0 ) (x x0 ) + Rm (x)
m!
m 1 (m) Rm (x)
= f (x0 ) + (x x0 ) f (x0 ) + m :
m! (x x0 )
1 1 (m)
Take " = 2 m! f (x0 ) > 0 then by (30) there exists > 0 such that

Rm (x)
m 0 ";
(x x0 )

that is
Rm (x)
" m 0 ";
(x x0 )
for all x 2 (a; b) with jx x0 j . Hence

1 (m) Rm (x) 1 (m)

f (x0 ) + m f (x0 ) "
m! (x x0 ) m!
1 (m) 1 1 (m) 1 1 (m)
= f (x0 ) f (x0 ) = f (x0 ) > 0
m! 2 m! 2 m!

91
 m
for all x 2 (a; b) with jx x0 j . Since m is even, we have that (x x0 ) >0
for x 6= x0 so

m 1 (m) Rm (x)
f (x) = f (x0 ) + (x x0 ) f (x0 ) + m
m! (x x0 )
> f (x0 ) + 0

for all x 2 (a; b) with 0 < jx x0 j , which shows that f has a local minimum
at x0 ;
m
(iii) If m is odd then (x x0 ) > 0 if x > x0 and so we proceed exactly as
in the previous part to conclude that

f (x) > f (x0 )

m
for all x 2 (a; b) with if x0 < x < x0 + . On the other hand, (x x0 ) < 0 if
x < x0 and so
f (x) < f (x0 )
for all x 2 (a; b) with if x0 < x < x0 . Thus f has neither a local minimum
nor a local maximum at x0 .

Important Taylor’s formulas with center x = 0

1 2 1 3 1 4 1 n
ex = 1 + x + 2! x + 3! x + 4! x + + n! x + o (xn ), hence the …rst order
formula is
ex = 1 + x + o (x) ;
while the second order formula is
1 2
ex = 1 + x + x + o x2 :
2!
n+1 1 n
log (1 + x) = x 12 x2 + 31 x3 1 4
4x + + ( 1) n
n x + o (x ), hence the
…rst order formula is

log (1 + x) = x + o (x) ;

while the second order formula is

1 2
log (1 + x) = x x + o x2 :
2
a
(1 + x) = 1+ax+ 12 a (a 1) x2 + + n! 1
a (a 1) (a 2) (a 3) (a n + 1) xn +
n
o (x ), hence the …rst order formula is
a
(1 + x) = 1 + ax + o (x) ;

while the second order formula is

a 1
(1 + x) = 1 + ax + a (a 1) x2 + o x2 :
2

92
 1 1 n+1
= (1 + x) = 1 x + x2 x3 + x4 + + ( 1) xn + o (xn ) hence
1+x
the …rst order formula is
1 1
= (1 + x) =1 x + o (x) ;
1+x
while the second order formula is
1 1
= (1 + x) =1 x + x2 + o x2 :
1+x

1 2 1 4 1 6 1 8 k 1 2k
cos x = 1 2! x + 4! x 6! x + 8! x + ( 1) 2k! x + o x2k+1 , hence
the third order formula is
1 2
cos x = 1 x + o x3 ;
2!
while the …fth order formula is
1 2 1
cos x = 1 x + x4 + o x5 ;
2! 4!

1 3 1 5 1 7 1 9 k 1 2k+1
sin x = x 3! x + 5! x 7! x + 9! x + ( 1) (2k+1)! x + o x2k+2 ,
hence the second order formula is

sin x = x + o x2 ;

while the fourth order formula is

1 3
sin x = x x + o x4 ;
3!
k
arctan x = x 13 x3 + 51 x5 17 x7 + 19 x9 + ( 1) 1
(2k+1) x
2k+1
+o x2k+2 ,
hence the second order formula is

arctan x = x + o x2 ;

while the fourth order formula is

1 3
arctan x = x x + o x4 :
3

Example 272 Note that as x ! 0,

x4 = o (x) ; x 4 = o x2 ; x 4 = o x3 ; x4 = o (xn ) for every n < 4:

but

x4 is not o x4 ; x4 is not o x5 ; x4 is not o (xm ) for every m 5:

93
Example 273 Note that as x ! 0,
3x + x2 + o(x3 ) x4 + 6x7 = 3x + x2 + o(x3 )
since x4 = o(x3 ) and 6x7 = o(x3 ).
Example 274 Note that
2
1 3 1 6 2
x x + o x4 = x2 + x + o x4
6 36
1 3 2 3
+ 2x x + 2xo x4 x o x4
6 6
1 1 4 2
= x2 + x6 + o x8 x + 2o x5 o x7
36 3 6
1 1 4
= x2 + x6 + o x8 x + o x5 + o x7
36 3
1 4
= x2 x + o x5 :
3
Example 275 Let’s calculate
sin100 x x100
lim+ ;
x!0 xa
where a > 0. We have
1 3
sin x = x x + o x4 :
3!
Hence,
100 100
1 3 1 2
sin100 x = x x + o x4 = x100 1 x + o x3 :
3! 3!
On the other hand, as t ! 0,
100
(1 + t) = 1 + 100t + o (t) ;
1 2
and so taking t = 3! x + o x3 ! 0 as x ! 0, we get
100
1 2 1 2 1 2
1 x + o x3 = 1 + 100 x + o x3 +o x + o x3
3! 3! 3!
100 2
=1 x + o x2 :
3!
Hence,
100 100
1 3 1 2
sin100 x = x x + o x4 = x100 1 x + o x3
3! 3!
100 2 100 102
= x100 1 x + o x2 = x100 x + o x102 :
3! 3!

94
It follows that
100 102 100 102
sin100 x x100 = x100 x + o x102 x100 = x + o x102 ;
3! 3!
and so
!
100 102
sin100 x x100 3! x + o x102 x102 100 o x102
= = +
xa xa xa 3! x102
8 100
< 0 3! + 0 if a < 102
100
! 1 3! + 0 if a > 102
: 100
1 3! + 0 if a = 102
8
< 0 if a < 102
= 1 if a > 102
: 100
3! if a = 102

Wednesday, November 26, 2012

Example 276 Let’s calculate

1 2
2x + log cos x
lim :
x!0 x4
We have
1 2
2x+ log cos x 0 + log cos 0 0 log 1 0
lim = = = :
x!0 x4 0 0 0
Taylor’s formula of cos x of order …ve is given by
1 2 1
cos x = 1 x + x4 + o x5 :
2! 4!
Hence,
1 2 1
log cos x = log 1 x + x4 + o x5 :
2! 4!
Let’s use now Taylor’s formula
1 2
log (1 + t) = t t + o t2 ;
2

95
 1 2 1 4
where for us t = 2! x + 4! x + o x5 . We get

1 2 1
log 1 x + x4 + o x5
2! 4!
!
2 2
1 2 1 1 1 2 1 1 2 1
= x + x4 + o x5 x + x4 + o x5 +o x + x4 + o x5
2! 4! 2 2! 4! 2! 4!
1 2 1 4 1 1 4
= x + x + o x5 x + o x4
2 4! 2 4
1 2 1 1
= x + + x4 + o x4
2 8 4!
1 2 1 4
= x x + o x4 ;
2 12
where all the powers of order bigger than 4 have been absorbed by o x4 .
Hence,
!
1 2 1 2 1 2 1 4 4 4 4
x + log cos x x x x + o x x 1 o x
2
= 2 2 12
= 4 +
x4 x4 x 12 x4
1 o x4 1
= + ! +0
12 x4 12
as x ! 0.

Remark 277 (Important) Note that if in the exercises we end up with

n
o ((x x0 ) )
lim m
x!x0 (x x0 )

where n < m, then we cannot conclude anything. This means that we have go
back and use more terms in our Taylor’s formula.
In the previous example, if we had used
1 2
cos x = 1 x + o x3 :
2!
Then
1 2
log cos x = log 1 x + o x3 :
2!
Let’s use now Taylor’s formula
1 2
log (1 + t) = t t + o t2 ;
2

96
 1 2
where for us t = 1 2! x + o x3 . We get

1 2
log 1 x + o x3
2!
!
2 2
1 2 1 1 2 1 2
= x + o x3 x + o x3 +o x + o x3
2! 2 2! 2!
1 2
= x + o x3
2
Then
1 2 1 2 1 2
2x + log cos x 2x 2x + o x3 o x3
= =
x4 x4 x4
and we are stuck.

19 Integration
Given an interval [a; b] with a < b, by a partition P of [a; b] we mean a …nite set
of points
a = x0 < x1 < < xn 1 < xn = b:
Consider a bounded function f : [a; b] ! R and a partition P . We de…ne the
lower and upper sums of f for the partition P respectively by
n
X
L (f; P ) := (xi xi 1) inf f (x) ;
xi 1 x xi
i=1
Xn
U (f; P ) := (xi xi 1) sup f (x) :
i=1 xi 1 x xi

Since f is bounded, we have that

(b a) inf f (x) L (f; P ) U (f; P ) (b a) sup f (x) : (31)

a x b a x b

The lower and upper integrals of f over [a; b] are de…ned respectively by
Z b
f (x) dx := sup fL (f; P ) : P partition of [a; b]g ;
a
Z b
f (x) dx := inf fU (f; P ) : P partition of [a; b]g :
a

If a = b we set Z Z
a a
f (x) dx = f (x) dx := 0:
a a

We study some properties of the lower and upper integrals of f .

97
Proposition 278 Consider a bounded function f : [a; b] ! R. Then
Z b Z b
(b a) inf f (x) f (x) dx f (x) dx (b a) sup f (x) : (32)
a x b a a a x b

Proof. Step 1: Let P = fx0 ; : : : ; xn g be a partition and consider a point

y 2 [a; b] not in P . Then there exists i0 2 f1; : : : ; ng such that xi0 1 < y < xi0 .
Consider the new partition P 0 given by P [ fyg. Then

(xi0 xi0 1) inf f (x)

xi0 i x xi0

= (xi0 y) inf f (x) + (y xi0 1) inf f (x)

xi0 1 x xi0 xi0 x xi0

(xi0 y) inf f (x) + (xi0 y) inf f (x) :

y x xi0 y x xi0

Since L (f; P ) and L (f; P 0 ) only di¤er in the interval [xi0 i ; xi0 ], this shows that

L (f; P ) L (f; P 0 ) : (33)

A similar argument shows that

U (f; P 0 ) U (f; P ) :

Using an induction argument, we have proved that by adding a …nite number

of points to a partition, we increase the lower sum and decrease the upper sum.
Step 2: Let P = fx0 ; : : : ; xn g and Q = fy0 ; : : : ; ym g be two partitions of [a; b]
and consider the new partition P 0 = P [ Q. By what we just proved in Step 1,

L (f; P ) L (f; P [ Q) U (f; P [ Q) U (f; Q) : (34)

Hence, L (f; P ) U (f; Q) for all partitions P and Q of [a; b]. Taking the
supremum over all partitions P of [a; b], we get
Z b
f (x) dx = sup L (f; P ) U (f; Q)
a P partition of [a;b]

for all partitions Q of [a; b]. Taking the in…mum over all partitions Q of [a; b],
we get
Z b Z b
f (x) dx inf U (f; Q) = f (x) dx:
a Q partition of [a;b] a

The remaining inequalities in (32) follow from (31).

Friday, November 28, 2012

Remark 279 If f : [a; b] ! R is bounded, then there exists M > 0 such that
jf (x)j M for all x 2 [a; b]. In turn,

M f (x) M

98
for all x 2 [a; b], and so
M inf f (x) sup f (x) M:
a x b a x b

It follows by the previous proposition

Z b
(b a) M (b a) inf f (x) f (x) dx
a x b a
Z b
f (x) dx (b a) sup f (x) (b a) M;
a a x b

so that
Z b
f (x) dx (b a) M;
a
Z b
f (x) dx (b a) M:
a

We will use this property later on.

Exercise 280 Consider a bounded function f : [a; b] ! R. Prove that for every
constant C 2 R,
Z b Z b
(f (x) + C) dx = f (x) dx + C (b a) ;
a a
Z b Z b
(f (x) + C) dx = f (x) dx + C (b a) :
a a

Exercise 281 Consider a bounded function f : [a; b] ! R and let c 2 (a; b).
Prove that
Z b Z c Z b
f (x) dx = f (x) dx + f (x) dx; (35)
a a c
Z b Z c Z b
f (x) dx = f (x) dx + f (x) dx: (36)
a a c

The next theorem will be used to prove the fundamental theorem of calculus.
Theorem 282 Consider a bounded function f : [a; b] ! R and let
Z x Z x
F (x) := f (y) dy G (x) := f (y) dy
a a

for x 2 [a; b]. Then F and G are Lipschitz continuous and

F 0 (x0 ) = G0 (x0 ) = f (x0 )
at every point x0 2 [a; b] at which f is continuous.

99
 Proof. Step 1: We only study the function F . To prove that F is Lipschitz
continuous, let M > 0 be such that jf (x)j M for all x 2 [a; b]. Fix x; y 2 [a; b].
Without loss of generality, we may assume x < y. By (35) with x in place of c
and [a; y] in place of [a; b], we have
Z x Z y Z y
F (y) = f (y) dy + f (y) dy = F (x) + f (y) dy:
a x x

Hence,
Z y
jF (y) F (x)j f (y) dy M (y x)
x

by Remark 279, which shows that F is Lipschitz.

Step 2: Assume that f is continuous at x0 2 [a; b]. We consider the case
x0 2 (a; b) (the cases x0 = a and x0 = b) are simpler. We want to prove that
F (x) F (x0 )
lim = f (x0 ) :
x!x0 x x0
For h 6= 0 consider the di¤erent quotient
( 1 Rx
F (x) F (x0 ) x x0 x0 (f (y) f (x0 )) dy if x > x0 ;
f (x0 ) = 1
R x0
x x0 x x0 x (f (y) f (x0 )) dy if x < x0 ;

where we have used (35) and Exercise 280. Since f is continuous at x0 , given
" > 0 there exists = (x0 ; ") > 0 such that
jf (x) f (x0 )j "
for all x 2 [a; b] with jx x0 j . Take jx x0 j . Then for x > x0 (the case
x < x0 is similar), by Remark 279, we have
Z x
F (x) F (x0 ) 1
f (x0 ) = (f (y) f (x0 )) dy
x x0 x x0 x0
1
" (x x0 ) ;
x x0
which shows that
F (x) F (x0 )
f (x0 ) ":
x x0
This concludes the proof.
As a corollary of the previous theorem, we have the mean value theorem for
integrals.
Corollary 283 (Mean Value Theorem for Integrals) Let f : [a; b] ! R
be bounded. Assume that f is continuous in (a; b). Then there exists c 2 (a; b)
such that Z b
1
f (x) dx = f (c) :
b a a
A similar result holds for the upper integral.

100
 Proof. Consider the function F de…ned in the previous theorem. Since f
is continuous in (a; b), by the previous theorem, F is di¤erentiable in (a; b) and
F 0 (x) = f (x) for all x 2 (a; b). Moreover, F is Lipschitz continuous in [a; b] and
so it is continuous in [a; b]. By the mean value theorem applied to the function
F there exists c 2 (a; b) such that
Z b
f (x) dx 0 = F (b) F (a) = F 0 (c) (b a) = f (c) (b a) :
a

Given a bounded function f : [a; b] ! R, we say that f is Riemann integrable

over [a; b] if the lower and upper integral coincide. We call the common value
Rb
the Riemann integral of f over [a; b] and we denote it by a f (x) dx. Thus, for
a Riemann integrable function,
Z b Z b Z b
f (x) dx := f (x) dx = f (x) dx:
a a a

The family of all Riemann integrable functions over [a; b] is denoted R ([a; b]).
Example 284 The function
1 if x 2 [0; 1] \ Q;
f (x) :=
0 if x 2 [0; 1] n Q;
is called the Dirichlet function. By the density of the rationals and of the irra-
tionals we have that
inf f (x) = 0; sup f (x) = 1;
xi 1 x xi xi 1 x xi

and so
n
X
L (f; P ) = (xi xi 1) inf f (x) = 0;
xi 1 x xi
i=1
Xn
U (f; P ) = (xi xi 1) 1 =1 0 = 1:
i=1

Thus
Z 1 Z 1
f (x) dx = sup 0 = 0; f (x) dx = inf 1 = 1:
0 0

To determine when a function is Riemann integrable we need to introduce

the notion of sets of Lebesgue measure zero.
De…nition 285 A set E R has Lebesgue measure zero if for every " > 0
there exists a countable family of open intervals (an ; bn ) such that
1
[ 1
X
E (an ; bn ) and (bn an ) ":
n=1 n=1

101
Here
1
X X̀
(bn an ) := lim (bn an )
`!1
n=1 n=1
nP o
`
and this limit exists since the sequence n=1 (bn an ) is increasing. Indeed,
`

`+1
X X̀ X̀
(bn an ) = (bn an ) + (b`+1 a`+1 ) (bn an ) + 0:
n=1 n=1 n=1

Monday, December 03, 2012

Example 286 Let x 6= 1. By Exercise 17,

xn+1 1
1+x + xn =
x 1
for all n 2 N. Hence, if x > 0,
1
X X̀ x`+1 x 0 x
if 0 < x < 1;
xn = lim xn = lim = x 1
`!1 `!1 x 1 1 if x > 1:
n=1 n=1

Example 287 Let’s see some examples.

(i) A singleton E = fcg has Lebesgue measure zero. Given " > 0, take
I1 = c 2" ; c + 2" .
(ii) If a set E contains an open interval (a; b), then it cannot have Lebesgue
measure zero. Indeed, for any countable family of open intervals (an ; bn )
we have
1
[
(a; b) E (an ; bn ) ;
n=1

and so
1
X
b a (bn an ) :
n=1

Taking " < b a, we obtain a contradiction.

(iii) A countable set E = fxn gn has Lebesgue measure zero. Given " > 0, take
In = xn 2n" 1 ; xn + 2n" 1 . Then
1
X X1
1
length In = " n
":
n=1 n=1
2

In particular, N, Q, and Z all have Lebesgue measure zero.

102
(iv) If fEk gk is a countable family of sets, each with Lebesgue measure zero,
then their union
[1
E := Ek
k=1

has Lebesgue measure zero. Indeed, given " > 0 …x k. Since Ek has
Lebesgue measure zero, there exists a countable family of open intervals
(k)
In such that
1
[ 1
X "
Ek In(k) and length In(k) :
n=1 n=1
2k
n o
(k)
Consider the family In . It is still countable,
n;k

1
[ 1 [
[ 1
E= Ek In(k)
k=1 k=1 n=1

and X XX X "
length In(k) = length In(k) ":
n
2k
n;k k k

(v) There are uncountable sets that have Lebesgue measure zero. One such
example is given by the Cantor set.

The following theorem characterizes Riemann integrable functions.

Theorem 288 A bounded function f : [a; b] ! R is Riemann integrable if and

only if the set of its discontinuity points has Lebesgue measure zero.

We will not prove this theorem.

In view of the previous theorem, we have the following.

Corollary 289 Given f : [a; b] ! R,

(i) if f is continuous, then f is Riemann integrable,

(ii) if f is monotone, then f is Riemann integrable.

Proof. If f is continuous, then by the Weierstrass theorem it is bounded,

and thus by the previous theorem it is Riemann integrable.
If f is monotone, then it is bounded from below by min ff (a) ; f (b)g and
from above by max ff (a) ; f (b)g. Moreover, by Theorem 184 its set of disconti-
nuity points is at most countable. Since a countable set has Lebesgue measure
zero, it follows from the previous theorem that f is Riemann integrable.

Example 290 Some examples of functions that are Riemann integrable and
others that are not.

103
 (i) The function
sin x1 if 0 < x 1;
f (x) =
0 if x = 0;
is Riemann integrable over [0; 1], since it is bounded and discontinuous
only at x = 0.
(ii) The function

1 if x = n1 for some n 2 N;
f (x) =
0 otherwise,

is Riemann integrable over [0; 1], since it is bounded and its set of discon-
tinuity points is
1
E= [ f0g ;
n n
which is countable.

(iii) The function

1 if x 2 Q;
f (x) =
0 otherwise,
is bounded but not Riemann integrable over [0; 1], since it is bounded and
its set of discontinuity points is [0; 1].

Exercise 291 Consider the function g : [0; 1] ! [0; 1] de…ned by

8
< 0 if x is irrational,
g (x) := 1
if x = pq with p; q 2 N relatively prime, 0 < p < q;
: p
1 if x = 0 or x = 1:

(i) Prove that g is discontinuous at every rational point of [0; 1].

(ii) Prove that g is continuous at every irrational point of [0; 1].
(iii) Prove that g is Riemann integrable.

Next we discuss some properties of Riemann integration.

Theorem 292 Let f; g : [a; b] ! R be bounded. Then

Z b Z b Z b
f (x) dx + g (x) dx (f (x) + g (x)) dx
a a a
Z b Z b Z b
(f (x) + g (x)) dx f (x) dx + g (x) dx:
a a a

104
 Proof. Let P and Q be two partitions of [a; b] and let P [ Q = fx0 ; : : : ; xn g.
Then

inf (f (x) + g (x)) inf f (x) + inf g (x) ;

xi 1 x xi xi 1 x xi xi 1 x xi

and so
n
X
L (f + g; P [ Q) = (xi xi 1) inf (f (x) + g (x))
xi 1 x xi
i=1
Xn n
X
(xi xi 1) inf f (x) + (xi xi 1) inf g (x)
xi 1 x xi xi 1 x xi
i=1 i=1
= L (f; P [ Q) + L (g; P [ Q) :

Hence,
Z b
(f (x) + g (x)) dx = sup L (f + g; R) L (f + g; P [ Q)
a R

L (f; P [ Q) + L (g; P [ Q)
L (f; P ) + L (g; Q) ;

where in the last inequality we have used (34). Hence,

Z b
(f (x) + g (x)) dx L (f; P ) + L (g; Q)
a

for all partitions P and Q of [a; b]. Now we …x Q. Then

Z b
(f (x) + g (x)) dx L (g; Q) L (f; P )
a

Rb
for all partitions P of [a; b]. Hence, the number a
(f (x) + g (x)) dx L (g; Q)
is an upper bound for the set

E := fL (f; P ) : P partition of [a; b]g :

It follows that
Z b Z b
(f (x) + g (x)) dx L (g; Q) sup E = f (x) dx:
a a

Thus, we have proved that

Z b Z b
(f (x) + g (x)) dx f (x) dx L (g; Q)
a a

105
 Rb
for all all partitions Q of [a; b]. Hence, the number a
(f (x) + g (x)) dx
Rb
a
f (x) dx is an upper bound for the set

F := fL (g; Q) : Q partition of [a; b]g :

It follows that
Z b Z b Z b
(f (x) + g (x)) dx f (x) dx sup F = g (x) dx:
a a a

This proves that

Z b Z b Z b
(f (x) + g (x)) dx f (x) dx + g (x) dx:
a a a

The proof with the upper integral is similar and relies on the fact that
sup (f (x) + g (x)) sup f (x) + sup g (x) :
xi 1 x xi xi 1 x xi xi 1 x xi

We omit the details.

Wednesday, December 05, 2012
Important properties of lower and upper integrals are the following.
Theorem 293 Let f : [a; b] ! R be bounded and let 2 R. If 0, then
Z b Z b Z b Z b
( f ) (x) dx = f (x) dx; ( f ) (x) dx = f (x) dx; (37)
a a a a

while if < 0,
Z b Z b Z b Z b
( f ) (x) dx = f (x) dx; ( f ) (x) dx = f (x) dx: (38)
a a a a

Proof. If = 0, then f = 0 and there is nothing to prove since we get

0 = 0 in (37) and (38). If > 0, then
n
X
L ( f; P ) = (xi xi 1) inf f (x)
xi 1 x xi
i=1
Xn
= (xi xi 1) inf f (x) = L (f; P ) ;
xi 1 x xi
i=1

so that
Z b
f (x) dx = sup fL ( f; P ) : P partition of [a; b]g
a

= sup f L (f; P ) : P partition of [a; b]g

= sup fL (f; P ) : P partition of [a; b]g
Z b Z b
= f (x) dx = f (x) dx;
a a

106
while if < 0, then
n
X
L ( f; P ) = (xi xi 1) inf f (x)
xi 1 x xi
i=1
Xn
= (xi xi 1) sup f (x) = U (f; P ) ;
i=1 xi 1 x xi

so that
Z b
f (x) dx = sup fL ( f; P ) : P partition of [a; b]g
a

= sup f U (f; P ) : P partition of [a; b]g

= inf fU (f; P ) : P partition of [a; b]g
Z b Z b
= f (x) dx = f (x) dx:
a a

The other cases are similar.

Theorem 294 Let f; g : [a; b] ! R be bounded with f (x) g (x) for all x 2
[a; b]. Then
Z b Z b Z b Z b
f (x) dx g (x) dx; f (x) dx g (x) dx:
a a a a

Proof. Assume f g. Then for every partition P ,

n
X
L (f; P ) = (xi xi 1) inf f (x)
xi 1 x xi
i=1
Xn
(xi xi 1) inf g (x) = L (g; P ) ;
xi 1 x xi
i=1

so that
Z b
f (x) dx = sup fL (f; P ) : P partition of [a; b]g
a
Z b
sup fL (g; P ) : P partition of [a; b]g = g (x) dx:
a

The proof for the upper integral is the same.

Proposition 295 Let f; g : [a; b] ! R be Riemann integrable.

107
 (i) If c 2 (a; b), then f is Riemann integrable over [a; c] and [c; b] and
Z b Z c Z b
f (x) dx = f (x) dx + f (x) dx: (39)
a a c

(ii) If 2 R, then f is Riemann integrable and

Z b Z b
f (x) dx = f (x) dx: (40)
a a

(iii) The functions f + g and f g are Riemann integrable and

Z b Z b Z b
(f (x) + g (x)) dx = f (x) dx + g (x) dx: (41)
a a a

(iv) If f g, then
Z b Z b
f (x) dx g (x) dx:
a a

(v) The function jf j is Riemann integrable and

Z b Z b
f (x) dx jf (x)j dx:
a a

Proof. (i) Since f is Riemann integrable, the set of discontinuity points in

[a; b] has Lebesgue measure zero. Hence, the set of discontinuity points in [a; c]
and [c; b] has Lebesgue measure zero. It follows that f is Riemann integrable
over [a; c] and [c; b]. Property (39) follows from (35).
(ii) If = 0, then f = 0 and both sides of (40) are zero. If 6= 0, then
the set of discontinuities points of f is the same of f . Hence, f is Riemann
integrable. Property (40) follows from Theorem 293.
(iii) Since the set of discontinuities points of f + g and f g is contained in
the union of the sets of discontinuities points of f and g, using the fact that the
…nite union of sets of Lebesgue measure zero still has measure zero, it follows
that the functions f + g and f g are Riemann integrable.
To prove (41), we apply Theorem 292.
(iv) Apply Theorem 294.
(v) Since the set of discontinuities points of jf j is contained in the set of
discontinuities points of f , it follows that jf j is Riemann integrable. Using
parts (ii) and (iv) and the facts that f jf j and that f jf j, we get
Z b Z b
f (x) dx jf (x)j dx;
a a
Z b Z b
f (x) dx jf (x)j dx;
a a

which give (v).

108
Exercise 296 Give an example of a bounded function f : [a; b] ! R such that
jf j is Riemann integrable over [a; b], but f is not.

Remark 297 If f : [a; b] ! R is Riemann integrable, recall that

Z a Z b
f (x) dx := f (x) dx:
b a

Then property (v) should be replaced by

Z a Z a
f (x) dx jf (x)j dx :
b b

The next theorem will be used to prove the fundamental theorem of calculus.

Theorem 298 If f : [a; b] ! R is Riemann integrable then for every " > 0 there
exists a partition P " of [a; b] such that for every partition P = fx0 ; : : : ; xn g that
contains P " and for every yi 2 [xi 1 ; xi ], i = 1; : : : ; n,
Z b n
X
f (x) dx (xi xi 1) f (yi ) ": (42)
a i=1

Proof. Using the de…nition of supremum and of in…mum, we may …nd a

partition P of [a; b] and a partition Q of [a; b] such that
Z b
"
L (f; P ) f (x) dx ;
a 2
Z b
"
U (f; Q) f (x) dx + :
a 2

Then P " := P [ Q is a partition of [a; b]. Then by (34), L (f; P " ) L (f; P )
and U (f; Q) U (f; P " ). Hence,

0 U (f; P " ) L (f; P " ) U (f; P ) L (f; Q)

Z b Z b !
" "
f (x) dx + f (x) dx = 0 + ";
a 2 a 2

Rb Rb
where we have used a f (x) dx = a f (x) dx. Write P " = fx0 ; : : : ; xn g be a
partition of [a; b] that contains P " .
For every x 2 [xi 1 ; xi ], we have that

jf (x) f (yi )j sup f inf f;

[xi 1 ;xi ]
[xi 1 ;xi ]

109
and so by Remark 279 and Proposition 295,
Z b n
X n Z
X xi n Z
X xi
f (x) dx (xi xi 1) f (yi ) = f (x) dx f (yi ) dx
a i=1 i=1 xi 1 i=1 xi 1

Xn Z xi
= (f (x) f (yi )) dx
i=1 xi 1

n
X Z xi
f (x) f (yi ) dx
i=1 xi 1

n
!
X
(xi xi 1) sup f inf f
[xi 1 ;xi ]
[xi 1 ;xi ]
i=1
= U (f; P ) L (f; P ) ":

This concludes the proof.

Theorem 299 (Fundamental Theorem of Calculus) Let F : [a; b] ! R be

a di¤ erentiable function. Assume that F 0 is Riemann integrable over [a; b]. Then
Z b
F (b) F (a) = F 0 (x) dx: (43)
a

Proof. Fix " > 0 and apply Theorem 298 to …nd a partition P " =
fx0 ; : : : ; xn g of [a; b] such that for every yi 2 [xi 1 ; xi ], i = 1; : : : ; n,
Z b n
X
F 0 (x) dx (xi xi 1) F
0
(yi ) ": (44)
a i=1

By the mean value theorem, for every i = 1; : : : ; n, there exists yi 2 (xi 1 ; xi )

such that
F (xi ) F (xi 1 ) = (xi xi 1 ) F 0 (yi ) :
Hence,
n
X n
X
0
F (b) F (a) = F (xn ) F (x0 ) = (F (xi ) F (xi 1 )) = (xi xi 1) F (yi ) :
i=1 i=1

By substituting this expression in (44), we …nd that

Z b
F 0 (x) dx (F (b) F (a)) ":
a

By letting " ! 0+ , we get (43).

As a corollary of the fundamental theorem of calculus, we have the formula
for integration by parts.

110
Corollary 300 (Integration by Parts) Let F; G : [a; b] ! R be a di¤ eren-
tiable functions. Assume that F 0 and G0 are Riemann integrable over [a; b].
Then
Z b Z b
F (x) G0 (x) dx = F (b) G (b) F (a) G (a) = F 0 (x) G (x) dx:
a a

Proof. G and F are di¤erentiable, so they are continuous. Hence, they are
Riemann integrable. By Proposition 295, the functions F G0 and F 0 G are also
Riemann integrable. Consider the function H = F G. Then H 0 = F 0 G + F G0 .
By the fundamental theorem of calculus applied to H, we have
Z b
F (b) G (b) F (a) G (a) = H (b) H (a) = H 0 (x) dx
a
Z b
= (F 0 (x) G (x) + F (x) G0 (x)) dx;
a

which concludes the proof.

Example 301 Let’s calculate

Z 2
x log2 x dx:
1

x2
Take F (x) = log2 x and G0 (x) = x. Then F 0 (x) = 2
x log x while G (x) = 2 .
Hence,
Z 2 x=2 Z 2
x2 x2 2
x log2 x dx = log2 x log x dx
1 2 x=1 1 2 x
Z 2
= 2 log2 2 0 x log x dx:
1

We integrate by parts once more, taking F (x) = log x and G0 (x) = x. Then
2
F 0 (x) = x1 while G (x) = x2 . Hence,
Z 2 x=2 Z 2
x2 x2 1
x log x dx = log x dx
1 2 x=1 1 2 x
Z 2
1 3
= 2 log 2 0 x dx = 2 log 2 :
2 1 4

In conclusion, Z 2
3
x log2 x dx = 2 log2 2 2 log 2 + :
1 4

Friday, December 07, 2012

111
 Next we discuss integration by substitution. The classical formula that you
see in calculus is given by
Z b Z G(b)
f (G (x)) G0 (x) dx = f (y) dy:
a G(a)

We will prove a change of variables formula that requires weaker hypotheses.

Theorem 302 (Change of Variables) Let g : [a; b] ! R be Riemann inte-

grable and let Z x
G (x) := G (a) + g (t) dt; x 2 [a; b] :
a

Assume that f : G ([a; b]) ! R is Riemann integrable. Then (f G) g is Rie-

mann integrable over [a; b] and the following change of variables formula holds
Z b Z G(b)
f (G (x)) g (x) dx = f (y) dy;
a G(a)

where, if G (a) G (b), we are using the notation (??) and (??).

Note that we are not assuming that G is di¤erentiable in [a; b].

Proof. We will give the proof in the very special case in which both f and
g are continuous.
Case 1: Assume …rst that G (a) G (b) and de…ne the function
Z t
F (t) := f (y) dy; t 2 [G (a) ; G (b)] :
G(a)

Since f and g are continuous, by Theorem ?? and Exercise , F and G are

di¤erentiable, with F 0 = f and G0 = g. Hence, by Theorem 237, the composite
function H := F G is di¤erentiable, with

H 0 (x) = F 0 (G (x)) G0 (x) = f (G (x)) g (x)

for all x 2 [a; b]. By the fundamental theorem of calculus applied to the function
H := F G, we have that
Z G(b) Z b
f (y) dy 0 = H (b) H (a) = H 0 (x) dx
G(a) a
Z b
= f (G (x)) g (x) dx:
a

Case 2: If G (a) > G (b), we de…ne the function

Z G(a)
F (t) := f (y) dy; t 2 [G (b) ; G (a)] :
t

112
By Exercise ??, F 0 = f . Hence, by Theorem 237, the composite function
H := F G is di¤erentiable, with

H 0 (x) = F 0 (G (x)) G0 (x) = f (G (x)) g (x)

for all x 2 [a; b]. By the fundamental theorem of calculus applied to the function
H := F G, we have that
Z G(a) Z b
f (y) dy 0 = H (b) H (a) = H 0 (x) dx
G(b) a
Z b
= f (G (x)) g (x) dx:
a

Using (??), we obtain

Z G(b) Z G(a) Z b
f (y) dy = f (y) dy = f (G (x)) g (x) dx:
G(a) G(b) a

Remark 303 Note that there are examples of functions f and g for which
f (G (x)) g (x) is integrable, but not f (G (x)).

Next we discuss the composition of Riemann integrable functions. The next

example shows that the composition of Riemann integrable functions may not
be Riemann integrable.

Example 304 Let f be the function de…ned in Exercise 291 and let g : [0; 1] !
[0; 1] be the function
0 if x = 0;
g (x) :=
1 if x > 0:
Then f and g are both Riemann integrable, but their composition g f is not.
Indeed, of x 2 [0; 1]

0 if x is irrational,
g (f (x)) =
1 if x is rational,

and we have seen that this function is not Riemann integrable.

The next proposition shows that if f is continuous, then the composition is

Riemann integrable.

Proposition 305 Let f : [a; b] ! R be Riemann integrable, let g : [c; d] ! R

be continuous, and let f ([a; b]) [c; d]. Then g f : [a; b] ! R is Riemann
integrable.

113
 Proof. We begin by observing that if f is continuous at some point x 2
[a; b], then since g is continuous, the function g will be continuous at f (x)
and so g f will be continuous at x. This shows that set of discontinuities
points of g f is contained in the set of discontinuities points of g, and since
g is Riemann integrable, this set has Lebesgue measure zero. In turn, g f is
Riemann integrable.
If f is continuous and g is Riemann integrable, then g f may not be Riemann
integrable.

Exercise 306 Since the rationals are countable, we can write [0; 1] \ Q as a
sequence frn gn . Consider the open set
1
[ 1 1
U := rn n
; rn + n
n=1
4 4

and let C := [0; 1] n U . Note that C is closed.

1. Prove that C has empty interior.

2. Prove that, C does not have Lebesgue measure zero.
3. Let f (x) = dist (x; C) and let

0 if x = 0;
g (x) :=
1 if x > 0:

Prove that g f is not Riemann integrable.

114