Download as pdf or txt
Download as pdf or txt
You are on page 1of 43

REAL ANALYSIS II NOTES

J. A. VIRTANEN

Contents
1. The Riemann integral 3
1.1. Definition 3
1.2. Riemann sums 6
1.3. Uniform continuity 8
1.4. Integrability of continuous functions 9
1.5. Basic properties of the Riemann integral 10
1.6. Fundamental theorem of calculus 12
2. Sequences and series of functions 15
2.1. Definitions 15
2.2. Uniform convergence of sequences 16
2.3. Uniform convergence of series 21
2.4. Power series 26
3. Exponential and logarithmic functions 28
Appendix A. Preliminaries 31
A.1. Real numbers 31
A.2. Sequences of numbers 32
A.3. Series of numbers 34
A.4. Functions 35
A.5. Limits of functions 35
A.6. Continuity 36
A.7. Differentiability 37
Exercises 39
References 43

1
These are notes for Real Analysis II given at the University of Reading in Fall 2015.
The approach is based on standard analysis textbooks, such as [1, 2, 3].
You should attempt to solve all assigned exercises before each tutorial, where selected,
more challenging ones are discussed. There will be two assignments, one midterm, and
the other one toward the end of the course.

2
1. The Riemann integral
We start with an example that serves as preparation for the definition of the the
Riemann integral.
Example 1.1. Consider the region enclosed by the lines x = 0, x = 1, y = 0, and the
curve y = x2 ; that is,
A = {(x, y) ∈ R2 : 0 ≤ x ≤ 1, 0 ≤ y ≤ x2 }
How can we determine the area of A? If we assume that we only know how to compute
the area of some simple shapes in the plane, such as rectangles, we must make use of
limits.
We first divide [0, 1] into n subintervals (segments)
[0, 1/n], [1/n, 2/n], . . . , [n − 1/n, 1]
of length 1/n. Note that
2
k2

2 2 k+1
min x = 2 , max x = .
k
n
≤x≤ k+1
n
n k
n
≤x≤ k+1
n
n
Clearly, (
n−1  2 )
[ k k+1 k+1
A⊂ (x, y) : ≤ x ≤ , 0≤y≤ = Kn
k=0
n n n
and (
n−1  2 )
[ k k+1 k
A⊃ (x, y) : ≤ x ≤ , 0≤y≤ = kn .
k=0
n n n
We can easily determine the areas |Kn | and |kn | of Kn and kn . Using Exercise 1, we get
1 1 1 22 1 n2 1 + 22 + . . . + n2 n(n + 1)(2n + 1) 1
|Kn | = 2
+ 2
+ . . . + 2
= 3
= 3

nn nn nn n 6n 3
as n → ∞, and similarly
3n3 − 3n2 + n 1
|kn | = 3

6n 3
as n → ∞. Because kn ⊂ A ⊂ Kn , it is natural to require that the area |A| of A satisfies
|kn | ≤ |A| ≤ |Kn |
for all n ∈ N. Thus, |A| = 1/3.
1.1. Definition. Using the idea of the previous example, we start building the definition
of the Riemann integral. We first define lower and upper sums of a bounded function.
The order properties of the real line and the concept of the least upper bound of a set will
play an important role in defining the integral.
Definition 1.2. Let a, b ∈ R with a < b. A partition P of the interval [a, b] is a set of
points x0 , x1 , . . . , xn−1 , xn , where
a = x0 < x1 < x2 < . . . < xn−1 < xn = b.
For i = 1, . . . , n, set
∆xi = xi − xi−1 ,
which is the length of the segment [xi−1 , xi ]. Note that ∆xi = 1/n for each i in Exam-
ple 1.1.
3
Let f : [a, b] → R be a bounded function. Set
mi = inf f (x), Mi = sup f (x)
xi−1 ≤x≤xi xi−1 ≤x≤xi

(cf. the height of the rectangles in Example 1.1) and define the upper and lower sums by
Xn Xn
U (f, P ) = Mi ∆xi , L(f, P ) = mi ∆xi
i=1 i=1

(cf. the areas of the rectangles in Example 1.1). Also set


IM = inf U (f, P ), Im = sup L(f, P ),
P P

where P varies over all partitions of [a, b]. Finally, we say that f is Riemann integrable
on [a, b] if
IM = Im ,
and write f ∈ R[a, b] and
Z b Z b
f= f (x)dx = IM = Im .
a a

Remark 1.3. Let f : [a, b] → R be bounded. In order for IM and Im to exist, we need
to show that the set of numbers
{U (f, P ) : P is a partition of [a, b]}
is bounded below and
{L(f, P ) : P is a partition of [a, b]}
is bounded above, which we verify below in (1.1).
Example 1.4. Let c ∈ R, and define f (x) = c for x ∈ [a, b]. If P is a partition of [a, b],
then
U (f, P ) = c(b − a) = L(f, P ).
Rb
Thus, IM = Im , and so f is integrable on [a, b] and a f = c(b − a).
Definition 1.5. Let P be a partition of [a, b]. We say a partition Q of [a, b] is a refinement
of P if P ⊂ Q. Given two partitions P1 and P2 of [a, b], we say that a partition Q of [a, b]
is their common refinement if P1 ∪ P2 ⊂ Q.
Lemma 1.6. Let f : [a, b] → R be bounded, and let Q be a refinement of a partition P
of [a, b]. Then
L(f, P ) ≤ L(f, Q) and U (f, Q) ≤ U (f, P ).
Proof. Let P = {x0 , x1 , . . . , xn }. We may assume that Q = P ∪ {x0 }, where x0 ∈ (xi−1 , xi )
for some 1 ≤ i ≤ n. Since
m0i = inf f (x) ≥ mi and m00i = inf f (x) ≥ mi ,
xi−1 ≤x≤x0 x0 ≤x≤xi

we have
mi (xi − xi−1 ) = mi (xi − x0 ) + mi (x0 − xi−1 ) ≤ m0i (xi − x0 ) + m00i (x0 − xi−1 ),
and so L(f, P ) ≤ L(f, Q) because the other terms of L(f, P ) and L(f, Q) are equal.
The other case is similar (see Exercise 6). 
The following lemma shows that any lower sum can never exceed any upper sum related
to the same function.
4
Lemma 1.7. Let f : [a, b] → R be bounded, and let P1 and P2 be two partitions of [a, b].
Then
L(f, P1 ) ≤ U (f, P2 ).
Proof. Let Q be a common refinement of both P1 and P2 . By Lemma 1.6,
L(f, P1 ) ≤ L(f, Q) ≤ U (f, Q) ≤ U (f, P2 ).

Let f : [a, b] → R be a bounded function and let Q be a partition of [a, b]. Then
L(f, P ) ≤ U (f, Q) and L(f, Q) ≤ U (f, P ) (1.1)
for all partitions P of [a, b], and hence Im and IM exist (see Theorem A.4 and its corollary).
While we always have Im ≤ IM according to Lemma 1.7, it is way more difficult to
determine whether Im = IM ; that is, whether f is integrable.
Example 1.8. Define f : [−1, 1] → R by setting
(
−1 if − 1 ≤ x < 0,
f (x) =
1 if 0 ≤ x ≤ 1.
Let 0 <  < 1, and put P = {−1, −/2, /2, 1}. Then
L(f, P ) = −1(−/2 + 1) − 1(/2 + /2) + 1(1 − /2) = −
and
U (f, P ) = −1(−/2 + 1) + 1(/2 + /2) + 1(1 − /2) = .
R1
Therefore, − ≤ Im ≤ IM ≤ , and so Im = IM = 0; that is, f is integrable and −1
f = 0.
The following result, which we refer to as Riemann’s criterion for integrability, gives
us a useful criterion for a function to be integrable.
Theorem 1.9. A bounded function f is integrable if and only if for every  > 0 there is
a partition P such that
U (f, P ) − L(f, P ) < . (1.2)
Proof. We prove necessity first. Suppose Im = IM and let  > 0. Since
Im = sup{L(f, P ) : P is a partition of [a, b]},
there is a partition P1 such that
Im − /2 < L(f, P1 ).
Similarly, there is a partition P2 such that
IM + /2 > U (f, P2 ).
Let P be a common refinement of P1 and P2 . Then Lemma 1.6 implies
L(f, P ) ≥ L(f, P1 ) > Im − /2
and
U (f, P ) ≤ U (f, P2 ) < IM + /2.
Thus, U (f, P ) − L(f, P ) < IM + /2 − Im + /2 = .
For sufficiency, if  > 0, then there is a partition P such that (1.2) holds. Thus,
0 ≤ IM − Im ≤ U (f, P ) − L(f, P ) < , and it follows that IM = Im ; that is, f is
integrable. 
5
Corollary 1.10. Let f : [a, b] → R be bounded. If there are partitions Pn such that
U (f, Pn ) − L(f, Pn ) → 0
as n → ∞, then f is integrable on [a, b].
Proof. Exercise 11. 
Proposition 1.11. Let f : [a, b] → R be monotonic. Then f is integrable on [a, b].
Proof. Suppose that f is increasing. Then f (a) ≤ f (x) ≤ f (b) for all x ∈ [a, b], so f is
bounded. For n ∈ N, set Pn = {x0 , x1 , . . . , xn } with ∆xk = b−a
n
for k = 1, . . . , n. Then
n n
! n
X X b−a b−aX
U (f, Pn ) = Mk ∆k = sup f (x) = f (xk )
k=1 k=1
xk−1 ≤x≤xk n n k=1
Pn
and similarly L(f, Pn ) = b−a
n k=1 f (xk−1 ). Thus,
n
b−aX b−a
U (f, Pn ) − L(f, Pn ) = (f (xk ) − f (xk−1 )) = (f (b) − f (a)) → 0
n k=1 n
as n → ∞. By Corollary 1.10, f is integrable. Exercise 12 completes the proof. 
Using the previous proposition, many familiar functions can easily be seen to be in-
tegrable. For example, x 7→ sin x on [0, π/2]; x 7→ ex on [a, b]; x 7→ log x on [a, b] with
a > 0 are all increasing and hence integrable on the given intervals.
The following example shows that there are bounded functions that are not integrable.
Example 1.12. Define f : [0, 1] → R by
(
0 if x ∈ (R \ Q) ∩ [0, 1],
f (x) =
1 if x ∈ Q ∩ [0, 1].
Then f is bounded but not integrable on [0, 1].
Proof. For any partition P = {x0 , x1 , . . . , xn }, we have
mk = inf{f (x) : xk ≤ x ≤ xk−1 } = 0
because there is an irrational number in each [xk , xk−1 ] (see Theorem A.8). Similarly,
Mk = 1 (see Theorem A.7). Thus, L(f, P ) = 0 and U (f, P ) = 1, and so U (f, P ) −
L(f, P ) = 1 for all partitions P of [0, 1]. Therefore, by Theorem 1.9, the function f
cannot be integrable. 
1.2. Riemann sums. For a function f : [a, b] → R and a partition P = {x0 , . . . , xn } of
[a, b], pick ξk ∈ [xk−1 , xk ] and define the Riemann sum SP (f, ξ) by setting
n
X
SP (f, ξ) = f (ξk )(xk − xk−1 ).
k=1

Note, since mk ≤ f (ξk ) ≤ Mk for all k, we have


L(f, P ) ≤ SP (f, ξ) ≤ U (f, P ). (1.3)
Given a partition P = {x0 , . . . , xn }, we define its norm |P | by setting
|P | = max (xk − xk−1 ) < δ (1.4)
1≤k≤n

which is the length of the largest subinterval.


6
Definition 1.13. We say
lim SP (f, ξ) = I ∈ R,
|P |→0

if for every  > 0 there is a δ > 0 such that


|SP (f, ξ) − I| <  whenever |P | < δ
independently of ξ.
We prove another convenient criterion for integrability in terms of Riemann sums. We
start with a technical lemma that explains what happens to lower and upper sums when
the corresponding partitions get sufficiently fine.
Lemma 1.14. If f : [a, b] → R is bounded, then for every  > 0 there is a δ > 0 such
that
U (f, P ) < IM +  and L(f, P ) > Im −  (1.5)
whenever |P | < δ.
Proof. (∗) TBA. 
Rb
Theorem 1.15. Let f : [a, b] → R be bounded. Then f is integrable on [a, b] and a
f =I
if and only if
lim SP (f, ξ) = I.
|P |→0
R
Proof. For necessity, suppose f is integrable and f = I. Let  > 0. By Lemma 1.14,
there is a δ > 0 such that
U (f, P ) −  < I < L(f, P ) + 
if |P | < δ. Thus, independently of ξ (see (1.3)),
I −  < L(f, P ) ≤ SP (f, ξ) ≤ U (f, P ) < I + 
if |P | < δ. Therefore, |SP (f, ξ) − I| <  if |P | < δ.
For sufficiency, suppose lim|P |→0 SP (f, ξ) = I. Then, for every  > 0, there is a δ > 0
such that
I −  < SP (f, ξ) < I +  (1.6)
if |P | < δ. Choose ξk such that

f (ξk ) > Mk −
b−a
for k = 1, . . . , n. Then
n
X n
X
SP (f, ξ) = f (ξk )(xk − xk−1 ) > Mk (xk − xk−1 ) −  = U (f, P ) − .
k=1 k=1

Thus, by (1.6),
I +  > SP (f, ξ) > U (f, P ) −  ≥ IM − ;
that is, I > IM − 2 for all  >), so I ≥ IM . Similarly, we can show that I ≤ Im (see
Exercise 18). Therefore,
R I ≤ Im ≤ IM ≤ I, and hence I = Im = IM , which means f is
integrable and f = I. 
7
Example 1.16. Let
1 1 1
Sn = + + ... +
n+1 n+2 n+n
for n ∈ N. We compute limn→∞ Sn using the preceding theorem. Define f : [1, 2] → R
by f (x) = 1/x. For n ∈ N, let
k k−1 k
Pn = {1, 1 + 1/n, 1 + 2/n, . . . , 1 + n/n}, ξ(n, k) = 1 + ∈ [1 + , 1 + ],
n n n
k = 1, . . . , n. Note |Dn | → 0. Set ξ(n) = (ξ(n, 1), ξ(x, 2), . . . , ξ(n, n)). Then
n n n  
X 1 X n 1 X n+k 1
Sn = = = f
k=1
n + k k=1
n + k n k=1
n n
n Z 2
X 1
= f (ξ(n, k)) = SDn (f, ξ(n)) → f
k=1
n 1

according to Theorem 1.15 because f is monotonic R 2 and hence integrable (see Proposi-
tion 1.11). Later in the course, we will see that 1 f = log 2, and so Sn → log 2.
1.3. Uniform continuity. In addition to the concept of continuity, there is a related,
stronger property of continuity, known as uniform continuity. We first compare the two
concepts of continuity and then apply the theory of the latter to the study of further
properties of the Riemann integral.
Let us recall the definition of continuous functions. A function f defined on an interval
E is said to be continuous at x ∈ E if for every  > 0 there is a δ > 0 such that
|f (x) − f (y)| <  whenever |x − y| < δ and y ∈ E. (1.7)
If f is continuous at every point of E, then f is said to be continuous on E.
Definition 1.17. A function f : E → R is said to be uniformly continuous on E if for
all  > 0 there is a δ > 0 such that
|f (x) − f (y)| <  for all x, y ∈ E for which |x − y| < δ. (1.8)
Uniform continuity is a property of a function on a set while continuity can be consid-
ered at a point of the set.
Note that if f is continuous on E and if  > 0 is given, then δ in (1.7) may have to
be changed when x is changed. However, if f is uniformly continuous, then, for a given
 > 0, we must find a δ that works for all x ∈ E.
Example 1.18. Define f : [−1, 1] → R by

−2
 if − 1 ≤ x < 0,
f (x) = 1 if 0 ≤ x ≤ 1/2,

x + 1/2 if 1/2 < x ≤ 1.

Then f is continuous on [−1, 1] \ {0}.


Example 1.19. Let f (x) = x2 . Then f is uniformly continuous on [0, 1] (this can be seen
directly using the definition or applying Theorem 1.20), but it is not uniformly continuous
on [0, ∞) = {x ∈ R : x ≥ 0}. Indeed, let δ > 0. If x > 0 and y = x + δ/2, then
|x2 − y 2 | = |x + y||x − y| ≥ 2x|x − y| = xδ > 1
for x > 1/δ. Thus, there is no δ > 0 such that |x − y| < δ implies |f (x) − f (y)| < 1 for
all x, y ≥ 0.
8
It is obvious that every uniformly continuous function on E is continuous on E. The
converse is also true provided that E is a closed interval. We give two proofs of this
important result, the first (direct) one using the Heine-Borel Theorem A.12, and another
proof (by contradiction) using Theorem A.17.
Theorem 1.20. If f is continuous on a closed interval [a, b], then f is uniformly con-
tinuous on [a, b].
Proof 1. (∗) Suppose f is continuous on [a, b]. Let  > 0. If x ∈ [a, b], then there is a
δx > 0 such that
f (B(x, δx )) ⊂ B(f (x), /2). (1.9)
Clearly,
[
B(x, δx /2)
x∈[a,b]

is a cover of [a, b], where each B(x, δx /2) is an open interval. Since [a, b] is closed, the
Heine-Borel theorem A.12 implies that there are x1 , . . . xn such that
n
[
[a, b] ⊂ B(xk , δk /2). (1.10)
k=1

Observe that (1.9) implies


f (B(xk , δk )) ⊂ B(f (xk ), /2). (1.11)
Put δ = min{δ1 /2, . . . , δn /2}. Let x, y ∈ [a, b] and |x − y| < δ. We show that |f (x) −
f (y)| < . Indeed, x ∈ B(xk , δk /2) for some k according to (1.10), and so
|y − xk | = |y − x + x − xk | ≤ |y − x| + |x − xk | ≤ δ + δk /2 ≤ δk /2 + δk /2 = δk ,
which implies that y ∈ B(xk , δk ). By (1.11), f (x), f (y) ∈ B(f (xk ), /2), and so
|f (x) − f (y)| ≤ |f (x) − f (xk )| + |f (xk ) − f (y)| < /2 + /2 = .


Proof 2. Suppose that f is continuous but not uniformly continuous on [a, b]. Then there
exists an  > 0 such that for each n ∈ N there are xn , yn ∈ [a, b] for which |xn − yn | < 1/n
but |f (xn ) − f (yn )| ≥ ; cf. (1.8). By Theorem A.17, (xn ) has a convergent subsequence
xnk . Denote the limit of (xnk ) by x0 . Since a ≤ xnk ≤ b, we have a ≤ x0 ≤ b. Also observe
|xnk − ynk | < 1/nk ≤ 1/k for all k, and so
xnk − 1/k < ynk < xnk + 1/k.
Thus, ynk → x0 . Because f is continuous at x0 , it follows that
|f (xnk ) − f (ynk )| ≤ |f (xnk ) − f (x0 )| + |f (x0 ) − f (ynk )| → 0
as k → ∞ (see Theorem A.40), which is a contradiction. 

1.4. Integrability of continuous functions. We can use the previous theorem to prove
that each continuous function on a closed interval is integrable. This gives us a large class
of functions for which the integral exists.
Theorem 1.21. Every continuous function on a closed interval is integrable.
9
Proof. Suppose f is continuous on [a, b]. Let  > 0. By Theorem 1.20, there is a δ > 0
such that |f (x) − f (y)| < /(b − a) if x, y ∈ [a, b] and |x − y| < δ. Let P = {x0 , . . . , xn }
be a partition of [a, b] such that |P | < δ (see (1.4)). Then
n n
X X 
U (f, P ) − L(f, P ) = (Mk − mk )(xk − xk−1 ) < (xk − xk−1 ) = ,
k=1 k=1
b−a

where the inequality follows from the fact that the continuous function f attains the values
Mk , mk on [xk−1 , xk ] (see Theorem A.48). By Riemann’s criterion 1.9, f is integrable on
[a, b]. 
Example 1.22. Let f : [a, b] → R be bounded, and suppose f is continuous on [a, b]\{c},
where a < c < b, then f is integrable on [a, b].
Proof. Let  > 0. Since f is bounded, there is an M > 0 such that f (x) ≤ M for all
x ∈ [a, b], and so
 
sup f (x) − inf f (x) · 2h ≤ (M + M ) · 2h = 4M h < /2,
c−h≤x≤c+h c−h≤x≤c+h


where h < 8M . Since f is continuous on [a, c − h] and on [c + h, b], there are partitions
0 00
P and P of these intervals such that
U (f, P 0 ) − L(f, P 0 ) < /4, U (f, P 00 ) − L(f, P 00 ) < /4
(see Theorem 1.9). Let P = P 0 ∪ P 00 . Then P is a partition of [a, b] and
 
0
U (f, P ) = U (f, P ) + sup f (x) (c + h − c + h) + U (f, P 00 )
c−h≤x≤c+h

and  
0
L(f, P ) = L(f, P ) + inf f (x) (c + h − c + h) + L(f, P 00 ).
c−h≤x≤c+h

Therefore,
 
U (f, P ) − L(f, P ) ≤ /4 + sup f (x) − inf f (x) · 2h + /4
c−h≤x≤c+h c−h≤x≤c+h

≤ /4 + /2 + /4 = .


Thus, by Theorem 1.9, f is integrable on [a, b]. 
1.5. Basic properties of the Riemann integral. We list and prove several useful
basic results on the Riemann integral. These results will be applied several times in what
follows.
Theorem P1.23. Suppose that f1 , f2 , . . . , fn are integrable on [a, b] and ck ∈ R. Then the
n
function k=1 ck fk is integrable on [a, b] and
Z bXn Xn Z b
ck f k = ck fk .
a k=1 k=1 a

Proof. By Theorem 1.15,


Z b
lim SP (fk , ξ) = fk ,
|P |→0 a
10
and so
n
! n n
X X X
lim SP ck f k , ξ = lim SP (ck fk , ξ) = lim ck SP (fk , ξ)
|P |→0 |P |→0 |P |→0
k=1 k=1 k=1
n
X n
X Z b
= ck lim SP (fk , ξ) = ck fk .
|P |→0 a
k=1 k=1

Another application of Theorem 1.15 completes the proof. 


Theorem 1.24. If f is integrable on [a, b], then
Z b
m(b − a) ≤ f ≤ M (b − a),
a
where
m = inf f (x) and M = sup f (x).
a≤x≤b a≤x≤b

Proof. If Q = {a, b}, then Q is a partition of [a, b] and


U (f, Q) = M (b − a) and L(f, Q) = m(b − a).
Thus,
Z b
m(b − a) ≤ sup{L(f, P )} ≤ f ≤ inf {U (f, P )} ≤ M (b − a).
P a P


Theorem 1.25. Let f, g be Riemann integrable on [a, b], and suppose that
f (x) ≤ g(x) for all x ∈ [a, b].
Then Z b Z b
f≤ g.
a a

Proof. See Exercise 24. 


Rb
We have defined a f only when a < b. We now extend this to the case when b < a.
Ra
Definition 1.26. If a < b and f is integrable on [a, b], then we define b f by
Z a Z b
f =− f.
b a
Ra
We also set a
f = 0.
The following basic property is frequently used in algebraic computations of integrals,
and we often use it without reference.
Theorem 1.27. Let ∆ be an interval and f be integrable on ∆. Suppose a, b, c ∈ ∆.
Then Z b Z c Z c
f, f, f
a a b
are all well defined. Also
Z c Z b Z c
f= f+ f. (1.12)
a a b
11
Proof. Suppose a < b. Let  > 0. Since f is integrable on ∆, there is a partition P of ∆
such that U (f, P ) − L(f, P ) <  (see Riemann’s criterion 1.9). Choose Q = {a, b} ∪ (P ∩
[a, b]). Then Q is a partition of [a, b] and U (f, Q) − L(f, Q) ≤ U (f, P ) − L(f, P ) < .
Thus, by Riemann’s criterion, f is integrable on [a, b]. For the other cases, see Exercise 25.
To show (1.12), suppose first a < b < c. Let P 0 and P 00 be partitions of [a, b] and [b, c],
respectively. Then P = P 0 ∪ P 00 is a partition of [a, c], and
SP (f, ξ) = SP 0 (f, ξ) + SP 00 (f, ξ), (1.13)
which implies (1.12) according to Theorem 1.15.
Suppose next that a < c < b (the remaining cases can be reduced to this and the first
case above; see Exercise 25). The first part implies that
Z b Z c Z b
f= f+ f,
a a c
and so Z c Z b Z b Z b Z c
f= f− f= f+ f.
a a c a b

1.6. Fundamental theorem of calculus. If f is integrable on [a, b], then
Z x
F (x) = f (t)dt (1.14)
a
is well defined for all x ∈ [a, b], and so we have a function F : [a, b] → R. By looking
at the graphs of some simple functions f (e.g., piecewise constant functions) and the
corresponding functions F , it seems that F behaves better than f . In fact, the next
theorem shows that F is always continuous provided that f is integrable.
Theorem 1.28. Let f be integrable on [a, b]. Then the function F defined in (1.14) is
continuous on [a, b].
Proof. Since f is integrable, it is bounded and hence there is an M > 0 such that
|f (x)| ≤ M for all x ∈ [a, b]. Let c ∈ [a, b). For h > 0,
Z c+h Z c Z c+h Z a Z c+h
F (c + h) − F (c) = f− f= + f= f.
a a a c c
By Theorem 1.24,
Z c+h
−M h ≤ f ≤ M h,
c
and so −M h ≤ F (c + h) − F (c) ≤ M h, which shows limh→0+ F (c + h) = F (c); that
is, limx→c+ F (x) = F (c). Similarly, if c ∈ (a, b], we can show limx→0− F (x) = F (c) (see
Exercise 26). Therefore, F is continuous on [a, b]. 
If, in addition, we know that f is continuous at some point, then F is differentiable at
that point.
Theorem 1.29. If f is integrable on [a, b] and if f is continuous at some x0 ∈ (a, b),
then the function F : [a, b] → R, defined by
Z x
F (x) = f (t)dt,
a
0
is differentiable at x0 and F (x0 ) = f (x0 ).
12
Proof. Let h > 0 be such that x0 + h ∈ [a, b], and write
mh = inf f (x), Mh = sup f (x).
[x0 ,x0 +h] [x0 ,x0 +h]

Then Z x0 +h Z x0 Z x0 +h
F (x0 + h) − F (x0 ) = f− f= f.
a a x0
By Theorem 1.24,
mh h ≤ F (x0 + h) − F (x0 ) ≤ Mh h.
Since h > 0, we have
F (x0 + h) − F (x0 )
mh ≤ ≤ Mh . (1.15)
h
We show limh→0 mh = limh→0 Mh = f (x0 ). Let  > 0. Because f is continuous at x0 ,
there is a δ > 0 such that
|f (x) − f (x0 )| < /2
if x0 < x < x0 + δ; that is,
f (x0 ) − /2 < f (x) < f (x0 ) + /2
if x0 < x < x0 + δ. Thus,
f (x0 ) −  < f (x0 ) − /2 ≤ gh ≤ Gh ≤ x(x0 ) + /2 < f (x0 ) + 
if 0 < h < δ, which implies limh→0 mh = limh→0 Mh = f (x0 ). Therefore, (1.15) implies
that F 0 (x0 +) = f (x0 ). Similarly, F 0 (x0 −) = f (x0 ). It follows that F is differentiable at
x0 and that F 0 (x0 ) = f (x0 ). 
In the next result, which is often used to evaluate integrals, the second assertion implies
the first one. However, we prove the first one independently of the second one because
the proof provides a nice application of the preceding theorem.
Theorem 1.30 (Fundamental theorem of calculus). (a) If f is continuous on [a, b] and
if F is a differentiable function on [a, b] such that F 0 = f , then
Z b
f = F (b) − F (a).
a
(b) If f is integrable on [a, b] and if F is a differentiable function on [a, b] such that
0
F = f , then
Z b
f = F (b) − F (a).
a

Proof. (a) Let Z x


G(x) = f (t)dt, x ∈ [a, b].
a
By Theorem 1.29, G0 = f = F 0 on [a, b]. Thus, there is a constant c ∈ R such that
G = F + c (see Theorem A.57). Note that 0 = G(a) = F (a) + c so that c = −F (a).
Therefore,
Z b
f = G(b) = F (b) + c = F (b) − F (a).
a
(b) Here we cannot apply Theorem 1.29 because f need not be continuous. Let P =
{x0 , . . . , xn } be a partition of [a, b]. For each k, there is a ξk ∈ [xk−1 , xk ] such that
F (xk ) − F (xk−1 ) = F 0 (ξk )(xk − xk−1 ) = f (ξk )(xk − xk−1 )
13
(see Theorem A.56). Therefore,
n
X n
X
F (b) − F (a) = F (xk ) − F (xk−1 ) = f (ξk )(xk − xk−1 ),
k=1 k=1

and so
U (f, P ) ≤ F (b) − F (a) ≤ L(f, P ),
Rb
which implies that I = Im ≤ F (b) − F (a) ≤ IM = I, where I = a
f . Consequently,
I = F (b) − F (a). 
Remark 1.31. In Theorem 1.30 (b), the condition that f be integrable on [a, b] cannot
be dropped because there are differentiable functions F on [a, b] whose derivative F 0 is
not integrable. Indeed, define
(
x2 sin x12 if x 6= 0,
F (x) =
0 if x = 0.
Then (
2x sin x12 − x2 cos x12 if x 6= 0,
F 0 (x) =
0 if x = 0.
However, F 0 is not bounded on any neighborhood of 0 and hence cannot be integrable on
any interval containing 0.
Example 1.32. If f (x) = 0 for x 6= 1 and f (1) = 1, then f is integrable but there is no
function F such that F 0 = f . This shows that there are functions whose integral cannot
be evaluated by means of Theorem 1.30.
The following useful result is an application of the chain rule.
Theorem 1.33 (Change of variable). Let f be continuous on [a, b] and let ϕ : [c, d] →
[a, b] be a function with a continuous derivative ϕ0 , and ϕ(α) = a, ϕ(β) = b for some
α, β ∈ [c, d]. Then
Z b Z β
f ϕ(t) ϕ0 (t)dt.

f=
a α
Rx
Proof. Define F (x) = a f (t)dt. By Theorem 1.30,
Z b Z β Z β
0
 0
f ϕ(t) ϕ0 (t)dt
  
f = F (b) − F (a) = F ϕ(α) − F ϕ(β) = F ϕ(t) ϕ (t)dt =
a α α

because (F ◦ ϕ)0 (t) = F 0 ϕ(t) ϕ0 (t) according to the chain rule (see Theorem A.53).


Example 1.34. Suppose f is continuous on [−h, h], where h > 0.
(a) If f is even; that is, f (x) = f (−x) for all x, then
Z a Z a
f =2 f,
−a 0

where 0 < a ≤ h.
(b) If f is odd; that is, f (−x) = −f (x) for all x, then
Z a
f = 0,
−a

where 0 < a ≤ h.
14
Proof. (a) Put ϕ(t) = −t. Then ϕ0 (t) = −1 and
Z 0 Z 0 Z 0 Z a
 0
f= f ϕ(t) ϕ (t)dt = − f (t)dt = f.
−a a a 0
Ra Ra
Thus, −a f = 2 0 f .
(b) Exercise 29. 
Another useful way of evaluating integrals is given in the following theorem.
Theorem 1.35 (Integration by parts). Suppose F, G are differentiable on [a, b], and both
F 0 = f and G0 = g are integrable on [a, b]. Then
Z b Z b
F g = F (b)G(b) − F (a)G(a) − f G.
a a
Rb
Proof. We show that a (F g + f G) = F (b)G(b) − F (a)G(a). Indeed, since (F G)0 =
F G0 + F 0 G = F g + f G, the fundamental theorem of calculus implies
Z b Z b
(F g + f G) = (F G)0 = F (b)G(b) − F (a)G(a).
a a
Notice that in order to use the fundamental theorem of calculus (see Theorem 1.30),
we need to know that the function F G = F g + f G is integrable. By assumption, all
the functions F, G, f, g are integrable. Also it can be showed that the product of two
functions is integrable (see [3, Chapter 13, Exercise 38] or [2, Theorem 6.12]), and so
F g + f G is integrable by Theorem 1.23. 

2. Sequences and series of functions


In this section we study sequences and series of functions with emphasis on uniform
convergence and how the properties of functions in a sequence (or series) relate to the
properties of the limit function.
2.1. Definitions.
Definition 2.1. Let (fn ) be a sequence of functions fn : E → R, n = 1, 2, . . ., where
E ⊂ R is a set. We say that (fn ) converges to a function f : E → R on E if
lim fn (x) = f (x) (2.1)
n→∞

for all x ∈ E. The function f is called the (pointwise) limit, or the (pointwise) limit
function, of the sequence (fn ).
The main difficulty is to find out which properties of functions fn are preserved under
the above limit operations; for example, if fn are continuous, does it follow that the
function f defined in (2.1) is also continuous? Similarly, what if fn are integrable or
differentiable?
Recall that f is continuous at x0 if
lim f (x) = f (x0 ),
x→x0

that is,
lim lim fn (x) = lim lim fn (x),
x→x0 n→∞ n→∞ x→x0
which is the same as asking whether we can change the order in which we take the
limits.The following examples demonstrate these considerations further.
15
Example 2.2. (a) Define
(
xn if 0 ≤ x ≤ 1,
fn (x) =
1 if x ≥ 1.
Then fn are all continuous, but
(
0 if 0 ≤ x < 1,
f (x) = lim fn (x) =
n→∞ 1 if x ≥ 1
is not continuous.
(b) Define

1
−1 if x ≤ − n ,

fn (x) = nx if − n1 ≤ x ≤ n1 ,
if x ≥ n1 .

1

Obviously, each fn is continuous, but



−1 if x < 0,

f (x) = lim fn (x) = 0 if x = 0
n→∞ 
1 if x > 0.
is not continuous.
What these examples show is that, for a given  > 0, if x ∈ E, then there is an Nx ∈ N
such that
|fn (x) − f (x)| <  if n ≥ Nx . (2.2)
What we lack in these examples is a “uniform” N which works for all x ∈ E; cf. uniform
continuity. For example, in Example 2.2 (a), when x is getting closer to 1, we need to
increase N in order for (2.2) to hold. However, if there is an N for which (2.2) holds
for all x ∈ E, then the graphs of fn will lie close to the graph of f provided that n is
sufficiently large; that is, n > N —of course, N still depends on .

2.2. Uniform convergence of sequences.


Definition 2.3. A sequence of functions fn is said to convergence uniformly on E to a
function f if for all  > 0 there is an N ∈ N such that n ≥ N implies
|fn (x) − f (x)| < 
for all x ∈ E.
Example 2.4. Recall the sequence of functions (fn ) defined in Example 2.2; that is,
(
xn if 0 ≤ x ≤ 1,
fn (x) =
1 if x ≥ 1.

Let 0 < s < 1. We show that (fn ) converges uniformly on E = [0, s]. Indeed, if x ∈ E,
|fn (x) − f (x)| = xn → 0
as n → ∞. Thus, given  > 0, there is an N ∈ N such that if x ∈ E and n ≥ N , then
|fn (x) − f (x)| < .
16
Theorem 2.5. Suppose (fn ) converges to f on E. Then fn → f uniformly on E if and
only if
sup |fn (x) − f (x)| → 0
x∈E
as n → ∞.
Proof. Exercise. 
n
Example 2.6. (a) Define fn (x) = xn for x ∈ E = [−1, 1]. For each x ∈ E, clearly
fn (x) → 0. We show that the convergence is uniform. Since
sup |fn (x)| = sup |xn /n| = 1/n → 0
x∈E −1≤x≤1

as n → ∞, the sequence (fn ) converges uniformly on E by Theorem 2.5.


(b) Define fn (x) = n1 sin(nx) for x ∈ R and let f (x) = 0 for all x ∈ R. Then
1
sup |fn (x)| ≤ →0
x∈R n
as n → ∞. Thus, (fn ) converges uniformly on R to the function f .
(c) Again, consider fn (x) = xn . Recall fn (x) → 0 for all x ∈ (−1, 1). Since
sup |fn (x)| = 1
−1<x<1

for all n, Theorem 2.5 implies (fn ) does not converge uniformly on (−1, 1).
n
(d) Define fn (x) = xn! . If x ∈ R,
lim fn (x) = 0
n→∞

because fn+1 (x)/fn (x) → 0 as n → ∞.


Suppose that E1 is a (bounded) interval. Then there is an M > 0 such that |x| ≤ M
for all x ∈ E1 . Since n
x Mn
sup ≤ → 0,
x∈E1 n! n!
the sequence (fn ) converges uniformly on E1 to f .
Let E2 = (a, ∞). Then n
x nn
sup ≥ ≥ 1,
x∈E2 n! n!
and so (fn ) does not converge uniformly on E2 .
We now answer the question whether continuity is preserved when the convergence is
uniform.
Theorem 2.7. Let (fn ) be a sequence of continuous functions on E which converges
uniformly on E to a function f . Then f is continuous on E.
Proof. Let  > 0. Then there is an N ∈ N such that
sup |fn (x) − f (x)| < /3
x∈E

for n ≥ N . If x, x + h ∈ E,
|f (x + h) − f (x)| ≤ |fN (x + h) − fN (x)| + |f (x + h) − fN (x + h)| + |fN (x) − f (x)|
2
≤ |fN (x + h) − fN (x)| + .
3
17
Since fN is continuous at x, there is a δ > 0 such that
|fN (x + h) − fN (x)| < /3
if |h| < δ. Therefore, |f (x + h) − f (x)| <  if |h| < δ; that is, f is continuous at x. 
The following immediate corollary can often be used to determine whether convergence
can be uniform.
Corollary 2.8. Let fn be continuous on E ⊂ R and fn → f pointwise on E. If f is not
continuous on E, then the sequence (fn ) cannot converge uniformly on E.
Example 2.9. The preceding corollary implies that the convergence of the sequences
given in (a) and (b) of Example 2.2 is not uniform.
The juxtaposition of Theorem 2.7 and the next example illustrates a general principle
that uniform convergence preserves good behavior, not bad behavior.
Example 2.10. Define
(
1
n
if x ∈ Q,
fn (x) =
0 if x ∈ R \ Q.
Then fn (x) → 0 for all x ∈ R and the convergence is uniform. Observe that each fn is
everywhere discontinuous while the limit function is everywhere continuous.
Our next goal is to consider what role uniform convergence plays in integration of
sequences of functions.
Example 2.11. Define fn (x) = n2 xn (1 − x) for x ∈ [0, 1]. Then fn (0) = fn (1) = 0, and
if 0 < x < 1, we have
lim fn (x) = 0;
n→∞
see Theorem A.26. Thus, f (x) := limn→∞ fn (x) = 0 for all x ∈ [0, 1]. Observe that
Z 1
n2
fn (x)dx = → 1,
0 (n + 1)(n + 2)
but Z 1 Z 1 Z 1
f= lim fn = 0 6= 1 = lim fn (x).
0 0 n→∞ n→∞ 0
It is natural to ask whether fn → f uniformly. Since
n 
n2
 
2 n 2 n n 1
sup |fn (x)| = sup n x (1 − x) ≥ n 1− = → ∞,
x∈[0,1] 0≤x≤1 n+1 n+1 n + 1 (1 + n1 )n
the convergence is not uniform.
However, limit processes can be interchanged without affecting the result if fn → f
uniformly. The next result also shows that the limit function of a sequence of integrable
functions is integrable provided that the convergence is uniform.
Theorem 2.12. Suppose (fn ) is a sequence of integrable functions which converges uni-
formly on [a, b] to a function f . Then f is integrable on [a, b] and
Z b Z b
lim fn = f. (2.3)
n→∞ a a
18
Proof. Note that f is bounded because uniform convergence implies that there is some
n1 ∈ N such that |fn1 (x) − f (x)| < 1 for all x ∈ [a, b], and so |f (x)| < |fn1 (x)| + 1 for all
x ∈ [a, b], where fn1 is bounded.
We use Riemann’s criterion 1.9 to show that f is is integrable. Let  > 0. Put
0 
 = 3(b−a) . Uniform convergence implies there is an N ∈ N such that |fN (x) − f (x)| < 0
for all x ∈ [a, b]. Since fN is integrable, there is a partition P = {x0 , . . . , xm } of [a, b]
such that

U (fN , P ) − L(fN , P ) < .
3
0
Since f (x) < fN (x) +  , we have
sup f (x) < sup fN (x) + 0 ,
xk−1 ≤x≤xk xk−1 ≤x≤xk

and so

U (f, P ) ≤ U (fN , P ) + .
3

Similarly, L(f, P ) ≥ L(fN , P ) − 3 . Therefore,
U (f, P ) − L(f, P ) ≤ U (fN , P ) + /3 − L(fN , P ) + /3 < .
By Riemann’s criterion, f is integrable on [a, b].
It remains to verify (2.3). Let  > 0. Then there is an N ∈ N such that if n ≥ N ,

|f (x) − fn (x)| <
b−a
for all x ∈ [a, b]. When n ≥ N , we have
Z b Z b Z b Z b Z b

f− fn = (f − fn ) ≤
|f − fn | ≤ = .
a b−a

a a a a

Thus, (2.3) holds. 


Remark 2.13. Even if the convergence is not uniform, it may well happen that
Z b Z b
lim fn = f;
n→∞ a a
Rb Rb
for example, consider fn (x) = nx(1 − x)n for x ∈ [0, 1]. Then limn→∞ a
fn = a
f but
(fn ) does not converge uniformly.
Next we consider uniform convergence and differentiation. We start with an example
that shows that (fn0 ) may not converge to f 0 even if fn → f uniformly.
Example 2.14. (a) Define fn : R → R by
sin nx
fn (x) = √ .
n
Put f (x) = 0√for all x ∈ R. Then fn → f uniformly, but (fn0 ) does not converge to f 0 ;
e.g. fn0 (0) = n → ∞ while f 0 (0) = 0.
2
(b) Define fn (x) = xe−nx for x ∈ R. Then
2
f 0 (x) = (1 − 2nx)e−nx
for x ∈ R. Since  
1 1 1
|fn (x)| ≤ fn √ = √ e− 2 → 0
2n 2n
19
it follows that fn → f uniformly, where f (x) = 0 for all x ∈ R. However,
lim fn0 (0) = 1 6= 0 = f 0 (0).
n→∞

The preceding examples show that stronger hypotheses are required.


Theorem 2.15. Let fn be differentiable on [a, b] such that fn0 is integrable on [a, b] for
all n. Suppose that
(i) fn → f pointwise, and
(ii) (fn0 ) converges uniformly on [a, b] to some continuous function g.
Then f is differentiable and
f 0 (x) = lim fn0 (x)
n→∞
for all x ∈ [a, b].
Proof. By Theorem 2.12 and the fundamental theorem of calculus,
Z x Z x Z x
0
g= lim fn = lim fn0 = lim (fn (x) − fn (a)) = f (x) − f (a).
a a n→∞ n→∞ a n→∞
Rx
Thus, f (x) = f (a) + a g for all x ∈ [a, b]. By Theorem 1.29, the function f is differen-
tiable and f 0 = g on [a, b]. 
A stronger version of the previous theorem holds. Its proof is optional material as
indicated by (∗) below. We need the following criterion for uniform convergence, which
is mainly of theoretical importance.
Theorem 2.16 (∗). Let (fn ) be a sequence of functions defined on E. Then (fn ) converges
uniformly on E if and only if for every  > 0 there is an N ∈ N such that n ≥ N , m ≥ N
implies
|fn (x) − fm (x)| ≤ 
for all x ∈ E.
Proof. (∗) See [2, Theorem 7.8]. 
Theorem  2.17. Suppose (fn ) is a sequence of0 differentiable functions on [a, b] such that
fn (x0 ) converges for some x0 ∈ [a, b]. If (fn ) converges uniformly on [a, b], then (fn )
converges uniformly on [a, b] to some function f and
f 0 (x) = lim fn0 (x) (2.4)
n→∞

for all x ∈ [a, b].


Proof. (∗) Let  > 0. Then there is an N ∈ N such that if n ≥ N, m ≥ N , we have

|fn (x0 ) − fm (x0 )| <
2
(see Theorem A.24) and

|fn0 (x) − fm 0
(x)| < (2.5)
2(b − a)
for all x ∈ [a, b] (see Theorem 2.16). The mean value theorem A.56 implies

fn (x) − fm (x) − fn (y) − fm (y) ≤
  
|x − y| ≤ (2.6)
2(b − a) 2
for all x, y ∈ [a, b] if n ≥ N, m ≥ M . Now
|fn (x) − fm (x)| ≤ |fn (x) − fm (x) − fn (x0 ) + fm (x0 )| + |fn (x0 ) − fm (x0 )| ≤ 
if n, m ≥ N and x ∈ [a, b]. By Theorem 2.16, (fn ) converges uniformly.
20
Let f (x) = limn→∞ fn (x) for x ∈ [a, b]. Let x ∈ [a, b] and define
fn (t) − fn (x) f (t) − f (x)
ϕn (t) = , ϕ(t) =
t−x t−x
for t ∈ [a, b] \ {x}, and ϕn (x) = fn0 (x). Then
lim ϕn (t) = fn0 (x)
t→x

for all n ∈ N, so that ϕn are continuous on [a, b]. By (2.5) and (2.6),
 
|ϕn (t) − ϕm (t)| ≤ and |ϕn (x) − ϕm (x)| <
2(b − a) 2(b − a)
if n, m ≥ N and t ∈ [a, b] \ {x}. Thus, (ϕn ) converges uniformly by Theorem 2.16. Since
ϕn are continuous (see above), we conclude that limn→∞ ϕn is continuous and we have
 
lim ϕ(t) = lim lim ϕn (t) = lim ϕn (x) = lim fn0 (x).
t→x t→x n→∞ n→∞ n→∞

Therefore, f is differentiable at x and f (x) = limt→x ϕ(t) = limn→∞ fn0 (x).


0


2.3. Uniform convergence of series. For the theory of series of numbers, see Sec-
P∞ A.3. Let fn be a sequence of functions on some set E ⊂ R. We say that
tion Pthe

series
n=1 fn converges pointwise on E if for every x ∈ E, the series of numbers n=1 fn (x)
converges. If the series converges, then we can define a function f : E → R by

X
f (x) = fn (x).
n=1

As in the previous section, we want to study how the properties of fn relate to the
function f . For example, if fn are continuous (or integrable), then does it follow that f is
continuous (or integrable)? The following example shows that this need not be the case,
and, again, we need some stronger condition on convergence—uniform convergence.
x 2
Example 2.18. For x ∈ N, define fn : R → R by fn (x) = (1+x 2 )n for n ∈ N. The
P∞
geometric series n=0 fn (x) converges pointwise on R. If x 6= 0,

X 1
fn (x) = x2 1 = 1 + x2
n=0
1− 1+x2

(see Theorem A.30). For all n ∈ N, fn (0) = 0, so the series is not continuous on R even
though each fn is a continuous function on R.
P
Definition 2.19. Let (fn ) be a sequence of functions defined on E. The series fn is
said to converge uniformly on E if the sequence (sn ) of partial sums converges uniformly
on E, where
n
X
sn (x) = fn (x)
k=1

for x ∈ E. We also write



X
s(x) = fn (x) = lim sn (x).
n→∞
k=1
21
Example 2.20. Consider the geometric series ∞ n
P
n=0 x and its partial sums
n
X 1 − xn+1
sn (x) = x k = 1 + x + x2 + . . . + xn =
k=0
1−x
1
for x 6= 1. If |x| < 1, then xn → 0 (see Theorem A.26) and so sn → 1−x
. Thus, the
geometric series converges pointwise on (−1, 1) and

X 1
xn = = s(x).
n=0
1−x
Since
1 n+1

|x|n+1 2
sup |s(x) − sn (x)| = ≤ 1 → 0,
− 1 ≤x≤ 1 |1 − x| 2
2 2

P∞ n 2.5 implies that sn → s uniformly, which means that the geometric series
Theorem
n=0 x converges uniformly on [−1/2, 1/2].
However, the geometric series does not converge uniformly on (−1, 1) because
n+1
|x|n+1 |1 − 1/n|n+1

1
sup ≥ =n 1− → ∞,
−1<x<1 1 − x 1 − 1 + 1/n n
where the inequality is obtained by taking x = 1 − 1/n ∈ (−1, 1).
Using the results on uniform converges of sequences, we can easily prove analogous
results for series of functions. We start with continuity.
P
Theorem
P 2.21. Suppose f n are continuous and fn converges uniformly on E. Then
fn is continuous on E.
Proof. Note that the finite sum snP= f1 + f2 + . . . + fn is continuous because each function
in the sum is continuous.
P Since fn converges uniformly, sn → s uniformly and hence
the series s = fn is continuous by Theorem 2.7. 
The following result shows that a series converging uniformly can be integrated term
by term.
P
Theorem 2.22.P Suppose fk are integrable on [a, b] and fk converges uniformly on
[a, b]. Then fk is integrable on [a, b] and
Z bX ∞ ∞ Z b
X
fk (x)dx = fk (x)dx.
a k=1 k=1 a

Proof. See Exercise 42. 


P∞ 1
Example 2.23. Recall k=0 xk = 1−x for all x ∈ (−1, 1) and the convergence is uniform
P∞ k
on [−a, a], where 0 < a < 1. Let t ∈ (−1, 1). Then k=1 x converges uniformly
k
on ∆ := [−|t|, |t|] and each function x 7→ x is continuous and hence integrable on ∆.
Therefore, by Theorem 2.22, we get
Z t Z tX ∞ ∞ Z t ∞ k
dx k
X
k
X t
log(1 − t) = − =− x dx = − x dx = − .
0 1−x 0 k=0 k=0 0 k=1
k
Replacing t by −t, we get

X tk
log(1 + t) = (−1)k−1
k=1
k
for all t ∈ (−1, 1).
22
2.24. Suppose fk are differentiable on [a, b], ∞ 0
P
Theorem P k=1 fk converges
P∞ uniformly on

[a, b], and f (x
k=1 k 0 ) converges for some x 0 ∈ [a, b]. Then s = k=1 fk converges
uniformly on [a, b] and
X∞
0
s (x) = fk0 (x)
k=1
for all x ∈ [a, b].
Proof. Apply Theorem 2.17. 
The following simple result gives a powerful criterion for uniform convergence of series.
Theorem 2.25 (Weierstrass). Let (fn ) be a sequence of functions on E and suppose
|fn (x)| ≤ Mn
P P
for all x ∈ E and n ∈ N. If Mn converges, then fn converges uniformly on E.
P
Proof. By Theorem A.29, |fn (x)| converges for all x ∈ E. We have

X ∞ X∞ X ∞
sup |s(x) − sn (x)| = sup fn (x) ≤ sup |fk (x)| ≤ Mk → 0

x∈E x∈E x∈E
k=n+1 k=n+1 k=n+1

because ∞
P P
n=1 M n converges. Thus, s n → s uniformly on E; that is, the series fn
converges uniformly on E. 
Example 2.26. (a) The series

X xn
k=1
n2
converges uniformly on [−1, 1] because
n
x
≤ 1
n2 n2
1
P
for all x ∈ [−1, 1] and k2
converges.
(b) Define
x
fk (x) =
k(1 + k 2 x2 )
for x ∈ R. Let k ∈ N. It is not difficult to show that fk0 (x) = 0 if and only if x = ± k1 . It
follows that
1
|fk (x)| ≤ 2 .
P 2k
By Theorem 2.25, the series fn converges uniformly on R.
In order to show that there exists an everywhere continuous function on R that is
nowhere differentiable, we consider a preliminary lemma.
Lemma 2.27 (∗). Suppose that f : (a, b) → R is differentiable at x ∈ (a, b). If
a < αk ≤ x < βk < b
for k ∈ N and if αk → x, βk → x, then
f (βk ) − f (αk )
→ f 0 (x)
βk − αk
as k → ∞.
23
Proof. Put
βk − x
λk =
βk − α k
for k ∈ N. Then
x − αk
1 − λk = ,
βk − αk
and so 0 < λk < 1 and 0 < 1 − λk < 1 for all k ∈ N. Now
   
f (βk ) − f (αk ) 0 f (βk ) − f (x) 0 f (αk ) − f (x) 0
− f (x) = λk − f (x) + (1 − λk ) − f (x)
βk − αk βk − x αk − x
for all k ∈ N. Thus,
f (βk ) − f (αk )
→ f 0 (x)
βk − αk
as k → ∞. 
Theorem 2.28 (∗). There is a continuous function on R which is nowhere differentiable.
Proof. Define h : R → R by
(
x if 0 ≤ x ≤ 1,
h(x) =
2 − x if 1 ≤ x ≤ 2
and h(x + 2) = h(x) for all x ∈ R. Observe h is 2-periodic; more precisely, if m ∈ Z,
(
x − 2m if 2m ≤ x ≤ 2m + 1,
h(x) =
2m + 2 − x if 2m + 1 ≤ x ≤ 2m + 2.
Then h is continuous on R.
For k ∈ N ∪ {0}, define
 k
3
fk (x) = h(4k x)
4
∈ R. Note that |fk (x)| ≤ ( 34 )k for all x ∈ R and ∞ 3 k
P
for x P k=0 ( 4 ) converges, which implies
that ∞ k=0 fk converges uniformly on R by Theorem 2.25.
Define f : R → R by
∞ ∞  k
X X 3
f (x) = fk (x) = h(4k x).
k=0 k=0
4
By Theorem 2.21, the function f is continuous on R.
Let x ∈ R. We show that f is not differentiable at x. For k ∈ N, choose pk ∈ Z such
that
pk ≤ 4k x < pk + 1.
Put
αk = 4−k pk , βk = 4−k (pk + 1)
for k ∈ N. Consider the difference
h(4k βn ) − h(4k αn ),
where n ∈ N, k ∈ N ∪ {0}. Now
(1) if k > n, then 4k βn − 4k αn = 4k−n ∈ Z, and so h(4k βn ) − h(4k αn ) = 0 by periodicity;

(2) if k = n, then 4k βn − 4k αn = 1, and so h(4k βn ) − h(4k αn ) = ±1;

24
(3) if k < n, then there is no q ∈ Z such that 4k αn < q < 4k βn , and so |h(4k βn ) −
h(4k αn )| = |4k βn − 4k αn | = 4k−n .
Therefore,
(
0 if k > n,
|h(4k βn ) − h(4k αn )| =
4k−n ifk ≤ n.
If n ∈ N, we get
n  k
X 3
h(4k βn ) − h(4k αn )

f (βn ) − f (αn ) =
k=0
4
 n X n−1  k
3 3
≥ − |h(4k βn ) − h(4k αn )|
4 k=0
4
 n X n−1  k  n n−1
3 3 k−n 3 1 X k
= − 4 = − n 3
4 k=0
4 4 4 k=0
 n  n  n  n
3 1 3n − 1 3 1 3 1 3
= − n > − = .
4 4 3−1 4 2 4 2 4
Thus,
1 3 n
> 2 ( 4 ) = 1 3n → ∞.
f (βn ) − f (αn )

βn − α n 4−n 2
Since
0 ≤ x − αn = x − 4−n pn < 4−n pn + 4−n − 4−n pn = 4−n
and
0 < βn − x = 4−n (pn + 1) − x ≤ 4−n (pn + 1) − 4−n pn = 4−n ,
we have αn → x and βn → x. By Lemma 2.27, f is not differentiable at x. 

The next result is extremely useful because it allows us to approximate continuous


functions by polynomials on closed intervals.
Theorem 2.29 (Stone-Weierstrass). If f is continuous on [a, b], there is a sequence of
polynomials Pn such that Pn → f uniformly on [a, b].
Proof. Without loss of generality, we may assume that [a, b] = [0, 1] (apply the linear
mapping x 7→ x−a
b−a
) and that f (0) = f (1) = 0 (consider g(x) = f (x)−f (0)−x(f (1)−f (0))
so that g(0) = g(1) = 0, and if g is a uniform limit of polynomials, then so is f , too, since
f − g is a polynomial).
So f : [0, 1] → R is continuous and f (0) = f (1) = 0. We extend f to a continuous
function on R by setting f (x) = 0 for all x ∈ R \ [0, 1]. Then f is uniformly continuous
on R.
For k ∈ N, define a polynomial
Qk (x) = ck (1 − x2 )k ,
where ck are chosen such that
Z 1
Qk = 1 for k ∈ N. (2.7)
−1
25
Since

Z 1 Z 1 Z 1/ n
2 n 2 n
(1 − x ) dx = 2 (1 − x ) dx ≥ 2 (1 − x2 )n dx
−1 0 0

Z 1/ n
4 1
≥2 (1 − nx2 )dx = √ > √
0 3 n n
R1 √
and since −1 cn (1 − x2 )dx = 1, we have c1n > √1n , that is, cn < n.
If 0 < δ < 1, then, for x ∈ [−1, −δ] ∪ [δ, 1], we get

Qn (x) = cn (1 − x2 )n ≤ n(1 − δ 2 )n → 0 (2.8)
as n → ∞ (see Theorem A.26).
Define Z 1
Pn (x) = f (x + t)Qn (t)dt
−1
for x ∈ [0, 1]. Since f (x) = 0 for all x ∈ R \ [0, 1], using Theorem 1.33, we have
Z 1−x Z 1
Pn (x) = f (x + t)Qn (t)dt = f (t)Qn (t − x)dt,
−x 0

and so Pn is a polynomial in x (just write Qn (s) = a0 + a1 s + a2 s2 + . . . + am sm , so that


Qn (t − x) = b0 (t) + b1 (t)x + b2 (t)x2 + . . . + bl (t)xl , and the integral of Qn (t − x) with
respect to t is a polynomial in x).
Since f is uniformly continuous on R, for  > 0, there is a δ ∈ (0, 12 ) such that
|f (x) − f (y)| < /2 if |x − y| < δ. Let M = supx∈R f (x). Then, for x ∈ [0, 1], we get
Z 1 Z 1

|Pn (x) − f (x)| = f (x + t)Qn (t)dt − Qn (t)dtf (x)
−1 −1
Z 1 Z 1

= (f (x + t) − f (x)) Qn (t)dt ≤ |f (x + t) − f (x)|Qn (t)dt
−1 −1
Z −δ Z δ Z 1

≤ 2M Qn (t)dt + Qn (t)dt + 2M Qn (t)dt
−1 2 −δ δ
√ 
≤ 4M n(1 − δ 2 )n + < 
2
if n is sufficiently large. Thus, Pn → f uniformly on [0, 1]. 
2.4. Power series. Let ak ∈ R for k ∈ N ∪ {0}, and let x0 ∈ R. Consider

X
f (x) = ak (x − x0 )k , (2.9)
k=0
which is referred to as a function represented by a power series.
Theorem 2.30 (Abel). (a) If (2.9) converges for some x1 6= x0 , then it converges abso-
lutely for |x − x0 | < |x1 − x0 |.
(b) If (2.9) diverges for some x2 6= x0 , then it diverges for |x − x0 | > |x2 − x0 |.
Proof. (a) If k ak (x1 − x0 )k converges, then there is an M > 0 such that for k ∈ N ∪ {0}
P

|ak ||x1 − x0 |k ≤ M,
and so
M
|ak | ≤ .
|x1 − x0 |k
26
Let |x − x0 | < |x1 − x0 |. Then
x − x0 k

k
|ak (x − x0 )| ≤ M

x1 − x0
k
for all k ∈ N ∪ {0}. Therefore, since the geometric series k M xx−x
P
converges, the
0

1 −x0

series k |ak (x − x0 )k | converges by the comparison test A.29. Thus, k ak (x − x0 )k


P P
converges absolutely.
(b) Suppose that |x − x0 | > |x2 − x0 |. If k ak (x − x0 )k converges, then k ak (x − x2 )k
P P
converges by (a). 
Example 2.31. If x = 1, the series

X xk
k=1
k
diverges (see Theorem A.31). By Theorem 2.30, the power series diverges if |x| > 1. For
x = −1, the series is alternating and hence converges because 1/k → 0. Thus, using
Theorem 2.30 again, we see that the series converges for |x| < 1. Therefore, the series
converges if and only if x ∈ [−1, 1).
Theorem 2.32. Suppose that

an+1
lim = R ∈ R.
n→∞ an

If R = 0, then ∞
P n
P∞ n
n=0 an x converges for all x ∈ R. If R > 0, then n=0 an x converges
1 1
for |x| < R and diverges for |x| > R .
Proof. Let R > 0. According to the ratio test, ∞ n
P
n=0 an x converges if

an+1 xn+1

= |x| an+1 → |x|R < 1;


an x n an

that is, |x| < 1/R. Similarly, by the ratio test, the series diverges if |x|R > 1. If R = 0,
then the above limit is less than one for all x ∈ R. 
Theorem 2.33. Suppose

X
ak x k (2.10)
k=0

converges for |x| < R and define



X
f (x) = ak x k (2.11)
k=1

for |x| < R. Then (2.10) converges uniformly on [−R + , R − ], where  > 0. Also, the
function f is continuous and differentiable in (−R, R), and

X
0
f (x) = kak xk−1 (2.12)
k=1

for −R < x < R.


27
an xn0 converges because |x0 | < R.
P
Proof. Let  > 0. Choose x0 ∈ (R − , R). Then
n
Thus, there is an M > 0 such that |an x0 | ≤ M for all n ∈ N. If |x| ≤ R − , then
n n
x R − 
|an xn | = |an ||x0 |n ≤ M
x0 x0
 n
and the geometric series ∞ R−
converges, so the series ∞ n
P P
n=0 M x0 n=0 an x converges
uniformly on [−R + , R − ]. Since the series converges uniformly and each term is
continuous, the function f represented by the series is continuous by Theorem 2.21.
Next we consider the series given in (2.12). We can show that it also converges
an x n
P
uniformly on [−R + , R − ] by a similar argument as used for the series
above. Thus, we can apply Theorem 2.24 to conclude P that f is differentiable and the
0
derivative
P∞ f is obtained by differentiating the series an xn term by term; that is,
0 n−1
f (x) = n=1 nan x . 
P∞
Corollary 2.34. Suppose f (x) = k=0 ak xk converges for |x| < R. Then f has deriva-
tives of all orders in (−R, R), which are given by

X
(n)
f (x) = k(k − 1) · · · (k − n + 1)ak xk−n
k=n

for all n ∈ N. In particular,


f (n) (0) = n!an .
Proof. Apply Theorem 2.33 repeatedly. 

3. Exponential and logarithmic functions


Define

x
X xk
e = exp(x) = (3.1)
k=0
k!
for x ∈ R. Note that the power series converges for all x ∈ R, and so, by Theorem 2.33,
exp is continuous and differentiable on R, and

0
X 1 n−1
exp (x) = n x = exp(x)
n=1
n!
for all x ∈ R.
We need the following result to prove that exp(x) exp(y) = exp(x + y) for all x, y ∈ R.
Theorem 3.1 (∗). If

X
ak
k=0
converges absolutely, then

X ∞
X ∞
X ∞ X
X n
ak bk = cn = ak bn−k ,
k=0 k=0 n=0 n=0 k=0
Pn
where cn = k=0 ak bn−k .
Proof. (∗) See [2, Theorem 3.50]. 
28
For x, y ∈ R, using the previous theorem and binomial formula, we have
∞ ∞ ∞ n
X xk X y k X X xk y n−k
exp(x) exp(y) = =
k=0
k! k=0
k! n=0 k=0
k! (n − k)!
∞ n   ∞ (3.2)
X 1 X n k n−k X (x + y)n
= x y = = exp(x + y).
n=0
n! k=0 k n=0
n!
Therefore,
exp(x) exp(−x) = exp(0) = 1,
so exp(x) 6= 0 for any x ∈ R and exp(x) > 0 for all x ∈ R.
By (3.1), exp(x) → ∞ as x → ∞, and
1
exp(x) = →0
exp(−x)
as x → −∞. Again, by (3.1), if 0 < x < y, then
1 1
exp(x) < exp(y) and exp(−x) = > = exp(−y),
exp(x) exp(y)
so exp is strictly increasing on R.
Theorem 3.2. (a) The exponential function ex is continuous and differentiable on R.
(b) We have exp0 (x) = exp(x) for all x ∈ R.
(c) The function ex is strictly increasing and positive for all x ∈ R.
(d) For x, y ∈ R, ex+y = ex ey ; and ex → ∞ as x → ∞, and ex → 0 as x → −∞.
n
(e) For n ∈ N, limx→∞ xex = 0.
Proof. Properties (a)-(d) were proved above. Let n ∈ N. Then

x
X xk xn+1
e = > ,
k=0
k! (n + 1)!
so that
xn (n + 1)! (n + 1)!
x
< xn n+1 = →0
e x x
as x → ∞. 
Since exp : R → (0, ∞) is strictly increasing and differentiable on R, it has an inverse
function log : (0, ∞) → R that is also strictly increasing and differentiable on (0, ∞). For
all y > 0,
exp (log(y)) = y,
or, equivalently, log (exp(x)) = x for all x ∈ R. Therefore,
log0 (exp(x)) exp(x) = 1,
so that log0 (y) = y1 , where y = exp(x). Since ran exp = (0, ∞), we have
1
log0 (y) = for all y > 0.
y
Thus, by Theorem 1.30, Z y
1
log(y) = dx.
1 x
If u = exp(x) and v = exp(y), then
log(uv) = log (exp(x) exp(y)) = log (exp(x + y)) = x + y = log(u) + log(v).
29
Using the properties of the exponential function, we can see that
log x → −∞ as x → 0
and
log x → ∞ as x → ∞.
The trigonometric functions sin and cos will be defined in Complex Analysis I in spring
2016.

30
Appendix A. Preliminaries
We recall some of the most important definitions and results from earlier parts of
analysis that are all necessary prerequisites for the main text. The proofs of all these
results can be found in any of the standard textbooks on analysis, e.g., in [1, 2, 3].
A.1. Real numbers. We list some basic properties of the set of all real numbers R. For
a construction of the real numbers from the rational numbers, see [2, Chapter 1, Real
numbers] or [3, Chapter 29]. Note that the symbols ∞ and −∞ are not real numbers.
Let E ⊂ R be nonempty. If there is a number M such that x ≤ M for all x ∈ E, then
we say E is bounded above and call M an upper bound of E. If there is an m ∈ R such
that m ≤ x for all x ∈ E, then we say E is bounded below and call m a lower bound of E.
Definition A.1. If G is an upper bound of a set E such that G ≤ M for any upper
bound M of E, then G is called the least upper bound of E and we write sup E = G. In
other words, sup E = G if
(a) x ≤ G for all x ∈ E;
(b) if x < G, then x is not an upper bound of E.
The greatest lower bound of a set E that is bounded below is defined similarly and
denoted by inf E; that is, we say g = inf E if
(a) g ≤ x for all x ∈ E (so g is a lower bound of E);
(b) if x > g, then x is not a lower bound of E (so g is the greatest lower bound).
These definitions are fundamental to the construction of the integral and many other
parts of basic analysis. To make sure you have understood them, try to prove the following
two theorems.
Theorem A.2. If G = sup E and  > 0, then there is an x ∈ E such that x > G − .
Conversely, if G is an upper bound of E and if for every  > 0 there is an x ∈ E such
that x > G − , then G = sup E.
Theorem A.3. If g = inf E and  > 0, then there is an x ∈ E such that x < G + .
Conversely, if g is a lower bound of E and if for every  > 0 there is an x ∈ E such
that x < G + , then g = inf E.
It is easy to show that sup E is unique, but a lot more difficult to show that it exists.
Theorem A.4. If E ⊂ R is nonempty and bounded above, then sup E exists in R.
Proof. See [2, Theorem 1.36]. 
Corollary A.5. If E ⊂ R is nonempty and bounded below, then inf E exists in R.
Theorem A.6 (Archimedean Property). If x, y ∈ R and x > 0, then there is an n ∈ N
such that nx > y.
Proof. See [2, Theorem 1.20]. 
The next theorem shows that the set of all rational numbers is dense in R.
Theorem A.7. If x, y ∈ R with x < y, then there is a q ∈ Q such that x < q < y.
Proof. See [2, Theorem 1.20]. 
Similarly, using the Archimedean property, we can show that between two rational
numbers there is an irrational number.
Theorem A.8. If p, q ∈ Q with p < q, then there is an x ∈ R \ Q such that p < x < q.
31

Proof. Exercise. Use Theorem A.6 and the fact that 2 ∈ R \ Q. 
Let x ∈ R and a > 0. Recall that
(
x, x ≥ 0,
|x| =
−x, x < 0,
and |x| < a if and only if −a < x < a.
The following simple properties are used repeatedly in the main text of these notes.
Theorem A.9. Let a, b, c ∈ R. Then
(a) |a| ≥ 0;
(b) |a| = 0 if and only if a = 0;
(c) |ab| = |a||b|;
(d) |a + b| ≤ |a| + |b|;
(e) |a − c| ≤ |a − b| + |b − c|.
Proof. See [2, Theorem 1.64]. 
Definition A.10 (∗). If  > 0, then the open interval
B(x0 , ) = {x ∈ R : |x − x0 | < }
is called the -neighborhood of x0 , and
B 0 (x0 , ) = B(x0 , ) \ {x0 } = {x ∈ R : 0 < |x − x0 | < }
is called the punctured -neighborhood of x0 .
A set A ⊂ R is open if for each x ∈ A there is an  > 0 such that B(x0 , ) ⊂ A. A set
S ⊂ R is called closed if its complement R \ S is open.
Definition A.11 (∗). Let E ⊂ R. A collection {Ai }i∈I of subsets of R such that E ⊂
∪i∈I Ai is said to be a cover of E. If each Ai is open, then we call {Ai }i∈I an open cover
of E.
The following Heine-Borel theorem is one of the most important results on the topo-
logical properties of the field of all real numbers.
Theorem A.12 (∗). Suppose that {Ai }i∈I is a cover of a closed interval [a, b] and that
each Ai is an open interval. Then {Ai }i∈I has a finite subcover.
Proof. See [2, Theorem 2.40]. 
A.2. Sequences of numbers. We say that a sequence of numbers (an ) converges to a
if for every  > 0 there is an N ∈ N such that
|an − a| <  for n ≥ N.
In this case we also say that a is the limit of (an ), and write
an → a or lim an = a.
n→∞

If (an ) does not converge, we say that it diverges.


Theorem A.13. If (xn ) is convergent, then {xn } is bounded above and below.
Proof. See [2, Theorem 3.2]. 
Theorem A.14. If an → a and bn → b and c ∈ R, then
(a) an + bn → a + b;
(b) can → ca;
32
(c) an bn = ab;
(d) a1n = a1 if there is an N ∈ N such that an 6= 0 for n ≥ N and a 6= 0.
Theorem A.15. If xn → a and zn → a, and if xn ≤ yn ≤ zn for all n ∈ N, then yn → a.
Let (an ) be a sequence of numbers. If nk ∈ N for every k and
n1 < n2 < n3 < . . . ,
then (ank ) is called a subsequence of (an ).
Theorem A.16. If (xn ) converges to some x ∈ R, then each subsequence (xnk ) also
converges to x.
Theorem A.17. Every bounded sequence in R contains a convergent subsequence.
Proof. See [2, Theorem 3.6]. 
Theorem A.18. If xn ≤ M for all n ∈ N and lim xn = x, then x ≤ M .
Proof. If x > M , then there is an N such that |x − xn | < x − M for n > N . Thus,
xn > M for n > N , which is a contradiction. 
Remark A.19. If xn < M for all n ∈ N and lim xn = x, then we may have x = M . For
example, consider xn = −1/n.
Theorem A.20. If (xn ) converges and (yn ) diverges, then (xn + yn ) diverges.
Proof. If (xn + yn ) is convergent, then (yn ) converges because yn = (xn + yn ) − xn . 
Remark A.21. A sequence (xn + yn ) may converge even if both (xn ) and (yn ) diverge.
For example, consider two sequences 0, 1, 0, 1, . . . and 1, 0, 1, 0, . . ..
Definition A.22. A sequence (xn ) of real numbers is called monotonically increasing
if xn ≤ xn+1 for all n ∈ N. The sequence (xn ) is called monotonically decreasing if
xn ≥ xn+1 for all n ∈ N. The class of monotonic sequences consists of the increasing and
the decreasing sequences.
Theorem A.23. Suppose (xn ) is monotonic. Then (xn ) converges if and only if it is
bounded.
Proof. See [2, Theorem 3.14]. 
Theorem A.24 (Cauchy criterion for convergence). A sequence (xn ) converges if and
only if it is a Cauchy sequence; that is, for every  > 0 there is an N ∈ N such that
|xn − xm | <  whenever n, m > N .
Proof. See [2, Theorem 3.11]. 
Definition A.25. Let (xn ) be a sequence of real numbers. If for every M ∈ R there is
an nM ∈ N such that xn > M whenever n > nM , then we write xn → ∞.
Similarly, if for each m ∈ R there is an nm ∈ N such that xn < m whenever n > nm ,
then we write xn → −∞.
The following special sequences are extremely important and useful.
Theorem A.26 (Special sequences). Let p > 0. Then
(a)
1
lim p = 0;
n→∞ n
33
(b)
1
lim p n = 1;
n→∞
(c)
1
lim n n = 1;
n→∞
(d) if α ∈ R,

lim = 0;
n→∞ (1 + p)n

(e) if |x| < 1,


lim xn = 0.
n→∞

Proof. See [2, Theorem 3.20]. 


A.3. Series of numbers. Let (an ) be a sequence of real numbers. We say the series
P ∞
n=1 an converges if the sequence of its partial sums (sn ) converges, where
Xn
sn = ak = a1 + a2 + . . . + an .
k=1
P∞
If sn → s, then we write n=1 an = s.
P
Theorem A.27 (Cauchy criterion). A series an converges if and only if for every
 > 0 there is an N ∈ N such that

Xm
ak ≤ 



k=n
whenever m ≥ n ≥ N .
Proof. See [2, Theorem 3.22] or [3, Chapter 23, Cauchy Criterion]. 
P
Theorem A.28. If an converges, then an → 0 as n → ∞.
Proof. See [2, Theorem 3.23] or [3, Chapter 23, Vanishing Condition]. 
P
PThe condition an → 0 is not sufficient for the series an to convergence; e.g. the series
∞ 1
n=1 n diverges.

Theorem
P A.29 (Comparison test).P(a) If |an | ≤ cn for n ≥ N0 , where N0 ∈ N, and if
cn converges, then the series an converges.
(b) If an ≥ dn ≥ 0 for n ≥ N0 and if dn diverges, then an diverges.
Proof. See [2, Theorem 3.25] or [3, Chapter 23, Theorem 1]. 
Theorem A.30 (Geometric series). If |x| < 1, then

X 1
xn = .
n=0
1 − x
If |x| ≥ 1, the series diverges.
Proof. See [2, Theorem 3.26] or [3, Chapter 23]. 
P 1
Theorem A.31. The series np
converges if p > 1 and diverges if p ≤ 1.
Proof. See [2, Theorem 3.28] or [3, Chapter 23, Application of Theorem 4]. 
n
Theorem A.32. limn→∞ 1 + n1 = e.
34
Proof. See [2, Theorem 3.31]. 
1 P
Theorem A.33 (Root test). Suppose that |an | n → α. Then an converges if α < 1,
and diverges if α > 1.
Proof. See [2, Theorem 3.33] or [3, Chapter 23, Exercise 7]. 

Theorem A.34 (Ratio test). Suppose that an+1
P
→ α. Then an converges if α < 1,

an
and diverges if α > 1.
Proof. See [2, Theorem 3.34] or [3, Chapter 23, Theorem 3]. 
P P
Definition A.35. We say an converges absolutely if |an | converges.
P P
Theorem A.36. If an converges absolutely, then an converges.
Proof. See [2, Theorem 3.45]. 
A.4. Functions.
Definition A.37. Let A, B ⊂ R. If with each x ∈ A, there is associated a number in
B, denoted by f (x), then f is called a function from A to B. The set A is called the
domain of f , and f (x) is referred to as the value of f at x. The set of all values is called
the range of f , and it is denoted by ran f .
Definition A.38. Let f be a function from A to B, and let E ⊂ A. We define f (E)
to be the set of all values f (x), for x ∈ E, and call it the image of E under f . Note
f (A) = ran f . We say f is onto if ran f = B; that is, f maps A onto B.
If E ⊂ B, then f −1 (B) is the set of all x ∈ A such that f (x) ∈ B. We call f −1 (B) the
pre-image of B under f . Note that if y ∈ B, then f −1 (y) is the set of all x ∈ A such that
f (x) = y. If for each y ∈ B, the set f −1 (y) consists of at most one element, then f is
said to be one-to-one; that is, f (x1 ) 6= f (x2 ) whenever x1 6= x2 for x1 , x2 ∈ A.
If E ⊂ R and f, g are functions on E, we define the functions f + g and f g on E by
(f + g)(x) = f (x) + g(x) and (f g)(x) = f (x)g(x) for x ∈ E. Also, if g(x) 6= 0, then we
define (f /g)(x) = f (x)/g(x).
A.5. Limits of functions.
Definition A.39. Let f be defined on (a, b) and c ∈ (a, b). We write f (x) → L as x → c,
or
lim f (x) = L,
x→c
if for every  > 0 there is a δ > 0 such that |f (x) − L| <  whenever 0 < |x − c| < δ.
Theorem A.40. Let f be a function. Then f (x) → L as x → c if and only if f (xn ) → L
as n → ∞ for every sequence (xn ) with the property that xn =
6 c and xn → c.
Proof. See [2, Theorem 4.2]. 
Definition A.41. We write
lim f (x) = L
x→∞
if, for every  > 0, there is an n > 0 such that
|f (x) − L| < 
whenever x > n .
35
Definition A.42. We write
lim f (x) = ∞
x→c
if, for every M ∈ R, there is a δM such that f (x) > M whenever 0 < |x − c| < δM .
Theorem A.43. Let f and g be functions on (a, b) and c ∈ (a, b). Suppose that
lim f (x) = L1 and lim g(x) = L2 .
x→c x→c

Then
(a) (f + g)(x) → L1 + L2 as x → c;
(b) (f g)(x) → L1 L2 as x → c;
(c) (f /g)(x) → L1 /L2 as x → c if L2 6= 0.
Proof. See [2, Theorem 4.4]. 
A.6. Continuity.
Definition A.44. Let f be a function defined on (a, b) and let c ∈ (a, b). We say f is
continuous at c if limx→c f (x) = f (c).
Using the definition of the limit, it is easy to see that this means that for every  > 0
there is a δ > 0 such that |f (x) − f (c)| <  whenever |x − c| < δ.
If f is not continuous at c, we say f is discontinuous at c; that is, the limit limx→c f (x)
does not exist or limx→c f (x) 6= f (c).
Definition A.45. A function f is said to be continuous on an open interval (a, b) if it is
continuous at every point of (a, b).
Let f : (a, b) → R be a function. We say limx→a+ f (x) = L1 if for every  > 0 there is
a δ > 0 such that
|f (x) − L1 | <  whenever a < x < a + δ.
Similarly, we say limx→b− f (x) = L2 if for every  > 0 there is a δ > 0 such that
|f (x) − L2 | <  whenever b − δ < x < b.
Definition A.46. A function f is said to be continuous on a closed interval [a, b] if it is
continuous on (a, b) and limx→a+ f (x) = f (a) and limx→b− f (x) = f (b).
Theorem A.47. If f is continuous on a closed interval [a, b], then there exist m, M ∈ R
such that
m ≤ f (x) ≤ M
for all x ∈ [a, b].
Proof. See [2, Theorem 4.15]. 
Theorem A.48. If f is continuous on a closed interval [a, b], then there are x1 , x2 ∈ [a, b]
such that
f (x1 ) = sup f (x), f (x2 ) = inf f (x);
x∈[a,b] x∈[a,b]

that is, f attains its maximum at x1 and its minimum at x2 .


Proof. See [2, Theorem 4.16]. 
Theorem A.49. Let f be continuous on [a, b]. If f (a) < c < f (b), then there is an
x ∈ (a, b) such that f (x) = c. A similar result holds if f (b) < c < f (a).
Proof. See [2, Theorem 4.23]. 
36
A.7. Differentiability.
Definition A.50. Let f be a function on (a, b) and x ∈ (a, b). If the limit
f (x + h) − f (x)
lim (A.1)
h→0 h
exists, then we say that f is differentiable and denote the limit by f 0 (x). We call f 0 (x)
the derivative of f at x.
Let f be defined on [a, b]. We say f is differentiable at a if
f (a + h) − f (a)
lim .
h→0+ h
Similarly, f is said to be differentiable at b if
f (b + h) − f (b)
lim .
h→0− h
If f is differentiable at every point of E, then it is said to be differentiable on E.
Theorem A.51. Let f be defined on [a, b]. If f is differentiable at x ∈ [a, b], then f is
continuous at x. Thus, if f is differentiable on [a, b], then f is continuous on [a, b].
Proof. See [2, Theorem 5.2]. 
The converse of this theorem is not true. In fact, there exists a continuos function on
R that is nowhere differentiable—see TBA.
Theorem A.52. Let f and g be functions on [a, b] and differentiable at some x ∈ [a, b].
Then f + g, f g and f /g are differentiable at x, and
(a) (f + g)0 (x) = f 0 (x) + g 0 (x);
(b) (f g)0 (x) = f 0 (x)g(x) + f (x)g 0 (x);
(c) and if g(x) 6= 0, then
 0
f f 0 (x)g(x) − f (x)g 0 (x)
(x) =  .
g g(x)2
Theorem A.53 (Chain rule). If f is continuous on [a, b], if f 0 exists at some point
x ∈ [a, b], if g is a defined on an interval that contains ran f , and if g is differentiable at
f (x), then the function h : [a, b] → R defined by

h(t) = g f (t)
is differentiable at x and
h0 (x) = g 0 f (x) f 0 (x).


Proof. See [2, Theorem 5.5]. 


Definition A.54. Let f be a function on (a, b). We say f has a local maximum at c if
there is a δ > 0 such that
f (x) ≤ f (c) for all x ∈ (c − δ, c + δ).
Local minima are defined analogously.
Theorem A.55. If f : [a, b] → R has a local maximum at some x ∈ (a, b) and if f is
differentiable at x, then f 0 (x) = 0.
Theorem A.56 (Mean value theorem). If f is a continuous function on [a, b] and dif-
ferentiable in (a, b), then there is a point ξ ∈ (a, b) such that
f (b) − f (a) = f 0 (ξ)(b − a).
37
Proof. See [2, Theorem 5.10]. 
Theorem A.57. Let f be differentiable in (a, b).
(a) If f 0 (x) = 0 for all x ∈ (a, b), then f is constant.
(b) If f 0 (x) ≥ 0 for all x ∈ (a, b), then f is increasing.
(c) If f 0 (x) ≤ 0 for all x ∈ (a, b), then f is decreasing.
Proof. See [2, Theorem 5.11]. 

38
Exercises
Exercise 1. Prove by induction that
n
X n(n + 1)(2n + 1)
k2 = .
k=1
6

Exercise 2. Use the approach of Example 1.1 to determine the area of the region A
enclosed by the curve y = 2x and the lines y = 0, x = 1 and x = 2.
Exercise 3. Use the approach of Example 1.1 to determine the area of the region enclosed
by the lines x = 0, x = 1, y = 0, and the curve y = ex .
Exercise 4. Consider the partitions P1 = {0, 1}, P2 = {0, 1/2, 1}, P3 = {0, 1/2, 2/3, 1},
P4 = {0, 1/2, 2/3, 1, 2}, and determine which partitions Pi are refinements of some other
partitions Pj .
Exercise 5. Let P = {x0 , x1 , . . . , xn } be a partition of [a, b]. Compute nk=1 ∆xk , where
P
∆xk = xk − xk−1 .
Exercise 6. Let f be bounded on [a, b] and Q be a refinement of a partition P of [a, b].
Prove that U (f, Q) ≤ U (f, P ).
Exercise 7. Let J1 = {−4, −1, 0, 1, 4} and J2 = {−4, −2/3, 2/3, 4} be two partitions of
[−4, 4]. Let J = J1 ∪ J2 . Recall that ∆xi = xi − xi−1 .
(a) Determine |J| (= maxi ∆xi ).
(b) Find a refinement Q = {y0 , . . . , yn } of J which has the least number of points and
for which ∆yi is constant.
(c) Find a refinement A = {a0 , . . . , am } of J such that ∆xi 6= ∆xj for i 6= j.
Exercise 8. Let f (x) = sin x for x ∈ [0, π/2]. Find a partition P of [0, π/2] such that
U (f, P ) − L(f, P ) < 10−4 .
Exercise 9. Let (
2x + 1 if 0 ≤ x ≤ 1/2,
f (x) =
−x if 1/2 < x ≤ 1
and let P = {0, 1/4, 1/2, 3/4, 1}. Compute L(f, P ) and U (f, P ), and plot the rectangles.
Exercise 10. Define f (x) = x2 for x ∈ [0, 2]. Using the R 2 definition of the Riemann
integral, prove that f is integrable on [0, 2] and determine 0 f .
Exercise 11. Let f : [a, b] → R be bounded. Suppose that there are partitions Pn of
[a, b] such that
U (f, Pn ) − L(f, Pn ) → 0
as n → ∞. Prove that f is integrable on [a, b].
Exercise 12. Let f : [a, b] → R be decreasing. Prove that f is integrable on [a, b].
Exercise
R1 13. Let f (x) = x3 for x ∈ [−1, 1]. Prove that f is integrable on [−1, 1] and
−1
f = 0.
Exercise 14. Let f (x) = x3 for x ∈ [1, 2]. Find a partition P of [1, 2] for which U (f, P )−
L(f, P ) < 2.
39
Exercise 15. Let (
−3 if − 20 ≤ x ≤ 1/2,
f (x) =
8 if 1/2 < x ≤ 1.
R1
Prove that f is integrable on [−20, 1] and compute −20 f .
Exercise 16. Let (
x2 x ∈ [0, 1] \ { 12 },
if
f (x) =
0 x = 12 .
if
R1
Prove that f is integrable on [0, 1] and determine 0 f .
Exercise 17. Let f (x) = 2x2 for x ∈ [−1, 4]. Find a partition P of [−1, 4] for which the
corresponding subintervals are of equal length. Let ξk be the midpoint of each subinterval
[xk−1 , xk ]. Compute the Riemann sum SP (f, ξ).
Exercise 18. Complete the proof of Theorem 1.15; that is, suppose lim|P |→0 SP (f, ξ) = 0
and then show that I ≤ Im . (Hint: recall the proof of the inequality IM ≤ I.)
Exercise 19. Let f (x) = 1/x2 for x ∈ [1, 2]. Compute R 2 the Riemann sums SD (f, ξ) when

D = {x0 , x1 , . . . , xn } and ξk = xk−1 xk . Determine 1 f using Riemann sums.
Exercise 20. Using Theorem 1.15, show that
Z 1
n n n dx
lim 2 2
+ 2 2
+ ... + 2 = .
n→∞ n + 1 n +2 n + n2 0 1 + x2
Exercise 21. Verify that the set of all integrable functions f : [a, b] → R is a linear
space.
Exercise 22. Let f : [a, b] → R be bounded. Suppose there are partitions Pn of [a, b]
such that
n4 + 4n3 − 51n − 1
U (f, Pn ) − L(f, Pn ) =
−4n5 + 32n4 + n2 − n + 10
for n ∈ N. Is f integrable on [a, b]. Does the sequence (L(f, Pn )) converge?
Exercise 23. Prove that Z π/4
π π
≤ cos2 x dx ≤ .
4 −π/4 2
Exercise 24. Prove Theorem 1.25.
Exercise 25. Work out the details of the cases omitted in the proof of Theorem 1.27.
Exercise 26. Prove that limx→c− F (x) = F (c) in the proof of Theorem 1.28.
Exercise 27. Let f be integrable on [a, b]. Prove that the functions f + and f − are both
integrable on [a, b].
Exercise 28. Let f be integrable on [a, b]. Use the previous exercise to prove that |f | is
integrable on [a, b] and
Z b Z b

f ≤ |f |.

a a

Exercise 29. Prove the identity for the integral of odd functions in Example 1.34 (b).
40
Exercise 30. Let f be continuous on [a, b]. Show that there is a ξ ∈ (a, b) such that
Z b
f = f (ξ)(b − a).
a
(Hint: use the fundamental theorem of calculus and mean value theorem.)
Exercise 31. Evaluate Z 2x
1
lim e−t sin t dt.
x→∞ x x
(Hint: use the preceding exercise.)
Exercise 32. Let f be integrable on [a, b]. Prove that the function F : [a, b] → R defined
by Z x
F (x) = f (t)dt
a
is Lipschitz continuous.
Exercise 33. Is there anything wrong with the following computation? If F (x) = −1/x,
then F 0 (x) = 1/x2 and
Z 1
dx
2
= F (1) − F (−1) = −2.
−1 x

Exercise 34. Suppose that (fn ) converges uniformly on a set E ⊂ R. Show that fn
converges pointwise on E.
Exercise 35. For n ∈ N, define fn : [1, 2] → R by fn (x) = xn /n. Determine whether fn
converges pointwise. Does (fn ) converge uniformly on [1, 2]?
2
Exercise 36. For n ∈ N, define fn : [0, 1] → R by fn (x) = 2nxe−nx . Find the pointwise
limit f (x) = limn→∞ fn (x) for x ∈ [0, 1]. Does (fn ) converge uniformly on [0, 1]?
Exercise 37. Consider the following sequences (fn ) on E. Find the pointwise limit (if
it exists) on E and determine whether (fn ) converges uniformly on E.
x1/n and E = [0, 1].
(a) fn (x) = (
0, x ≤ n,
(b) fn (x) =
x − n, x≥n
and E = [a, b] or E = R.
2
(c) fn (x) = e−nx and E = [−1, 1].
−x2
(d) fn (x) = e n and E = R.
Exercise 38. For n ∈ N, define fn : R → R by
x2n
fn (x) = .
1 + x2n
Draw the graph of fn and show that (fn ) does not converge uniformly on R.
Exercise 39. For n ∈ N, define fn : [0, 1] → R by
(
nx, 0 ≤ x ≤ n1
fn (x) = n(1−x) 1
n−1
, n < x ≤ 1.
Prove that (fn ) does not converge uniformly on [0, 1]. Does (fn ) converge uniformly on
any [a, 1], where 0 < a < 1?
41
Exercise 40. Compute Z 1
1 2
lim ex+ n x dx.
n→∞ 0

Exercise 41. Compute Z 2


lim (1 + xn )1/n dx.
n→∞ 0

Exercise
P∞ 42. Let (fn ) be a sequence of integrablePfunctions on [a, b] and suppose that

n=1 fn converges uniformly on [a, b]. Prove that n=1 fn and

Z b X ! ∞
X Z b
fn (x) dx = fn (x)dx.
a n=1 n=1 a

Hint: use Theorem 2.12.


Exercise 43. Prove that the series

X sin nx
(−1)n−1
n=1
n3
converges uniformly on R.
P∞
Exercise 44. Let 0 < a < 1. Show that the geometric series n=1 xn converges uniformly
on [−a, a]. Hint: see Example 2.20.
Exercise 45. Let n ∈ N. Prove that

(n − 1)! X
= k(k − 1) · · · (k − n + 2)xk−n+1
(1 − x)n k=n−1
for |x| < 1. Hint: use Theorem 2.33.
Exercise 46. Using the previous exercise, prove that

X (x2 + x)
k 2 xk = 3
.
k=1
(1 − x)

Exercise 47. Let f be continuous on [0, 1] and suppose


Z 1
f (x)xn dx = 0
0
for all n ∈ N ∪ {0}. Prove that f (x) = 0 for all x ∈ [0, 1]. (Hint: verify that that the
integral of the product
R 1 2 of f with any polynomial is zero and use the Stone-Weierstrass
theorem to show 0 f = 0.)

42
References
[1] L. Myrberg, Differentaali- ja integraalilaskenta. Kirjayhtymä, Helsinki, 1994.
[2] W. Rudin, Principles of mathematical analysis. Third edition. McGraw-Hill, New York, 1976.
[3] M. Spivak, Calculus. Third edition. Cambridge University Press, New York, 1994.

43

You might also like