Calculus II: For Biology and Medicine

Calculus II: For Biology and Medicine
Yi Li
Department of Mathematics, Johns Hopkins University, 3400 North
Charles Street, Baltimore, MD 21218
E-mail address: yli@math.jhu.edu; yilicms@gmail.com
Abstract. This is the lecture note on Calculus for biology and medicine. The
content contains additional topics beyond the textbook and lots of interesting
problems.
• Please inform me typos and other error (e.g., grammar) via
yilicms@gmail.com!
• I will keep to update material.
• All figures come from Google Image and Wiki.
Contents
Chapter 1. Numbers and functions 5

1.1. Real and complex numbers 5
1.2. Elementary functions 11
Chapter 2. Sequences and difference equations 17
2.1. Sequences 17
2.2. Population models 22
Chapter 3. Limits and continuity 27
3.1. Limits 27
3.2. Continuity 31
Chapter 4. Differentiation 35
4.1. Derivatives 35
4.2. Basic rules and chain rule 36
Chapter 5. Applications of differentiation 39
5.1. Extrema and the mean-value theorem 39
5.2. Monotonicity, concavity, and inflection points 43
5.3. L’Hospital’s rule 45
5.4. Antiderivative 47
Chapter 6. Integrations 49
6.1. Definite integrals 49
6.2. The fundamental theorem of calculus 51
6.3. Applications of integration 54
Chapter 7. Integration techniques 59
7.1. Basic rules 59
7.2. Integration by parts 62
7.3. Rational functions and partial fractions 65
7.4. Improper integrals 68
7.5. Taylor approximation 74
3
CHAPTER 1
Numbers and functions
This chapter deals with the preliminaries used in the following

chapters, including the systematical description of real and com-
plex numbers, and elementary functions. An important class of
elementary functions is the class of polynomial functions, since
we will see that any continuous function on a compact set (e.g.,
on a closed interval [0, 1]) can be approximated by polynomials.
1.1. Real and complex numbers

Let N be the set of all natural numbers, i.e.,
N := {n : n = 1, 2, · · · },
and N≥0 := N ∪ {0} the set of whole numbers. Let us denote Z and Q the set of
all integers and rational numbers, respectively. The relations of N, Z, Q are related
in Figure1 1.1.
1.1.1. The real numbers. Let R be the set of all real numbers. For any real
numbers a, b ∈ R with a < b, define
open interval (a, b) := {x ∈ R : a < x < b},
closed interval [a, b] := {x ∈ R : a ≤ x ≤ b}.
We also use half-open intervals:
[a, b) := {x ∈ R : a ≤ x < b}, (a, b] := {x ∈ R : a < x ≤ b}.
1http : //546834391620409041.weebly.com/
Figure 1.1. Numbers

5
6 1. NUMBERS AND FUNCTIONS
Figure 1.2. Real numbers
Figure 1.3. Absolute value
The above three types of intervals are clearly bounded; unbounded intervals are
sets of the forms {x ∈ R : x > a}. Here are the possible cases:
[a, ∞) := {x ∈ R : x ≥ a},
(−∞, a] := {x ∈ R : x ≤ a},
(a, ∞) := {x ∈ R : x > a},
(−∞, a) := {x ∈ R : x < a}.
The symbols “∞” and “−∞” “plus infinity” and “minus infinity”, respectively. In
particular,
R = (−∞, ∞).
Definition 1.1.1. The absolute value of a real number a, denoted by |a|, is

a, a ≥ 0,
(1.1.1) |a| :=
−a, a < 0.
Looking at Figure 1.3, we see the x = 0 is the “singular point” of the blue
line; a geometric interpretation says that the function y = x has no tangent line at
x = 0. The second picture in Figure 1.3 locally looks like the first one, at points
x = b, c, d.
1.1.2. Lines in the plane. The linear equation in the plane is given by
(1.1.2) Ax + By + C = 0,
1.1. REAL AND COMPLEX NUMBERS 7
Figure 1.4. Linear equation
where A, B, and C are some constants. When A = B = 0 in (1.1.2), we must have

C = 0. Hence, we may assume that A and B are not both equal to 0. Geometrically,
the graph of a linear equation is a straight line.
If the two points (x1 , y1 ) and (x2 , y2 ) lie on a straight line (see Figure 1.4), then
the slope of the line is
y2 − y1
(1.1.3) m := .
x2 − x1
Two points (or one point and the slope) are sufficient to determine the equation of
a straight line.
(i) If we are given one point (x0 , y0 ) and the slope m, we can use the point-
slope form of a straight line to write its equation:
(1.1.4) y − y0 = m(x − x0 ).
(ii) The equation (1.1.4) can be equivalently written as the slop-intercept
form
(1.1.5) y = mx + b,
where b is the y-intercept, which is the point of intersection of the line
with the y-axis.
We say y is proportional to x, if
y = mx
for some m ∈ R; in this case we call m the constant of proportionality and we
write
y ∝ x.
In particular, y − b ∝ x from (1.1.5).
Given two lines l1 and l2 in the plane.
(i) When l1 and l2 have no points in common or are identical, they are called
parallel, denoted by l1 k l2 . Note that two noncoincident lines are parallel
if and only if their slopes are identical. In particular, for two noncoinci-
dent, nonvertical lines l1 and l2 with slopes m1 and m2 , respectively, we
have
l1 k l2 ⇐⇒ m1 = m2 .
Figure 1.5. Circle
Figure 1.6. Trigonometric functions
(ii) Two l1 and l2 are called perpendicular, denoted l1 ⊥ l2 , if their inter-

section forms an angle of 90◦ . Two nonvertical lines are perpendicular if
and only if their slopes are negative reciprocals; that is, if l1 and l2 are
nonvertical lines with slopes m1 and m2 , then
l1 ⊥ l2 ⇐⇒ m1 m2 = −1.
1.1.3. Equation of the circle. A circle is the set of all points at a given
distance, called the radius, from a given point, called the center.
If r is the distance from (x0 , y0 ) to (x, y), then we have
(1.1.6) (x − x0 )2 + (y − y0 )2 = r2 .
If r = 1 and (x0 , y0 ) the circle is called the unit circle, denoted S1 (that is, S1 is
the sphere of dimension 1).
1.1.4. Trigonometry. Consider a point B = (x, y) on the unit circle S1 (see

Figure 1.6). The point B and O = (0, 0) form an angle θ. Define
1
sin θ = y, csc θ = ,
sin θ
1
cos θ = x, sec θ = ,
cos θ
y 1
tan θ = , cot θ = .
x tan θ
1.1. REAL AND COMPLEX NUMBERS 9
From the definition, we have
(1.1.7) sin2 θ + cos2 θ = 1,

2
(1.1.8) tan θ + 1 = sec2 θ.
Furthermore,
(i) Symmetry:
sin(−θ) = − sin θ, cos(−θ) = cos θ,

π π
sin −θ = cos θ, cos −θ = sin θ,
2 2
sin(π − θ) = sin θ, cos(π − θ) = − cos θ.
(ii) Shifts:
π π
sin θ + = cos θ, cos θ + = − sin θ,
2 2
sin(θ + π) = − sin θ, cos(θ + π) = − cos θ,
sin(θ + 2π) = sin θ, cos(θ + 2π) = cos θ.
(iii) Angle sum and difference identities:
sin(α ± β) = sin α · cos β ± cos α · sin β, cos(α ± β) = cos α · cos β ± sin α · sin β.
(iv) Double-angle formula:

2 tan θ
sin(2θ) = 2 sin θ · cos θ = ,
1 + tan2 θ
cos(2θ) = cos2 θ − sin2 θ = 2 cos2 θ − 1 = 1 − 2 sin2 θ
1 − tan2 θ
= ,
1 + tan2 θ
2 tan θ
tan(2θ) = .
1 − tan2 θ
(v) Triple-angle formula:
3 tan θ − tan3 θ
sin(3θ) = −4 sin3 θ+3 sin θ, cos(3θ) = 4 cos3 θ−4 cos θ, tan(3θ) = .
1 − 3 tan2 θ
1.1.5. Exponentials and logarithms. An exponential is an expression of

the form
ar
where a is called the base and r the exponent. Unless r is an integer or unless r
is a rational number of the form p/q where p is an integer and q is an odd integer,
we will assume that a is positive. Some basic properties are
ar as = ar+s , (ab)r = ar br ,
ar a r ar
= ar−s , = ,
as b br
1
a−r = , (ar )s = ars .
ar
Figure 1.7. Complex numbers
If y = ax > 0 with a > 0 and a 6= 1, then we formally write x := loga y.

Consequently,
loga (xy) = loga x + loga y,

x
loga = loga x − loga y,
y
loga (xr ) = r loga x,
loga (ax ) = x,
aloga x = x.
When a = e, we write loga as ln, where e is the constant and approximated equal2
to 2.7182818.
1.1.6. Complex numbers and quadratic equations. Observe that the
square of a real number is always nonnegative. To extend the real numbers, we
introduce a symbol i satisfying
(1.1.9) i2 = −1.
√ √
Formally, the equation has two solutions −1 and − −1. We usually set
√
(1.1.10) i := −1
and call i the imaginary unit.
Definition 1.1.2. A complex number is a number of the form

√
z = a + b −1, a, b ∈ R.
The real number a =: <(z) is the real part of z, and the real number b =: =(z) is
the imaginary part. (See
√ Figure 1.7) The set of all complex numbers is denoted
C. When a = 0 we call b −1 an purely imaginary number.
√
A complex number z = a + b −1 corresponds to a point (a, b) in the plane,
conversely,
√ a point (a, b) in the plane corresponds to a complex number z = a +
b −1. Consequently,
√ √
a + b −1 = c + d −1 ⇐⇒ a = c and b = d.
2This number is an irrational number and can be defined as (see later) lim 1 n
n→∞ (1 + n ) .
1.2. ELEMENTARY FUNCTIONS 11
√ √
Given two complex numbers a + b −1 and c + d −1, we have
√ √ √
(a + b −1) + (c + d −1) = (a + c) + (b + d) −1,
√ √ √ √ √
(a + b −1)(c + d −1) = ac + ad −1 + bc −1 + bd( −1)2
√
= (ac − bd) + (ad + bc) −1.
√
Definition 1.1.3. If z = a + b −1 is a complex number, its conjugate, denoted
by z̄, is defined as
√
(1.1.11) z̄ := a − b −1.
Proposition 1.1.4. For any z, w ∈ C we have

(1.1.12) (z̄) = z, z + w = z̄ + w̄, zw = z̄ w̄.
√
Furthermore, if z = a + b −1 then
(1.1.13) |z|2 := z z̄ = a2 + b2 .
p
We call |z| := |z|2 the norm of z.
√
If z = a + b −1, then
p
|z| = r, r := a2 + b2 .
Thus z lies in the circle at the origin with radius r.
1.2. Elementary functions

A function f is a rule that assigns each element x in the set A exactly one
element y in the set B. The element y is called the image (or value) of x under
f and is denoted by f (x). The set A is called the domain of f , the set B is called
the codomain of f , and the set f (A) = {y ∈ B : y = f (x) for some x ∈ A} is
called the range of f .
Definition 1.2.1. Two functions f and g are equal if and only if

(1) f and g are defined on the same domain, and
(2) f (x) = g(x) for all x in the domain.
By this definition, we see that two functions

f : R → R, x 7→ x2 , g : [0, 1] → R, x 7→ x2 ,
do not coincide with each other.
Definition 1.2.2. A function f : A → B is called

(1) even if f (x) = f (−x) for all x ∈ A, and
(2) odd if f (x) = −f (−x) for all x ∈ A.
By the definition, if f is even or odd, the domain A must be symmetric.
Definition 1.2.3. The composite function f ◦ g (also called the composition

of f and g) is defined as
(f ◦ g)(x) := f [g(x)]
for each x in the domain of g for which g(x) is in the domain of f .
1.2.1. Polynomial functions. A polynomial function is a function of the

form
f (x) = a0 + a1 x + a2 x2 + · · · + an xn ,
where n is the nonnegative integer and a0 , a1 , · · · , an are real-valued constants with
an 6= 0. The coefficient an is called the leading coefficient, and n is called the
degree of the polynomial function. The largest possible domain of f is R.
A polynomial function f (x) of degree n may not have n roots. For example,
consider the polynomial function
2
1 3
f (x) = x2 + x + 1 = x + + ;
2 4
this function is always strictly positive and hence has no roots. However
Theorem 1.2.4. Any nonconstant polynomial function with complex-valued coeffi-

cients has, counted with multiplicity, exactly n complex roots, where n is the degree
of this polynomial function.
This is the fundamental theorem of algebra. For example, the polynomial

function f (x) = x2 + x + 1 has two complex roots
r
1 3√
− + −1.
2 4
1.2.2. Rational functions. A rational function is the quotient of two poly-
nomial functions p(x) and q(x):
p(x)
f (x) = , q(x) 6= 0.
q(x)
1.2.3. Power functions. A power function is of the form
f (x) = xr , r ∈ R.
1.2.4. Exponential functions. The function f is an exponential function

with base a if
f (x) = ax
where a is a positive constant other than 1. The largest possible domain of f is R.
When a = e, the exponential function is called the natural exponential
function
exp(x) := ex .
1.2.5. Inverse functions. A function f (x) is called one-to-one if f (x1 ) =

f (x2 ) implies x1 = x2 .
Definition 1.2.5. Let f : A → B be a one-to-one function with range f (A). The

inverse function f −1 had domain f (A) and range A and is defined by
f −1 (y) = x ⇐⇒ y = f (x)
for all y ∈ f (x).
Note that if f : A → B has an inverse f −1 , then

f −1 [f (x)] = x, x ∈ A,
f [f −1 (x)] = x, x ∈ f (A).
1.2.6. Logarithmic functions. The inverse of f (x) = ax is called the loga-
rithm to base a and is written f −1 (x) = loga x.
aloga x = x, x > 0,
x
loga (a ) = x, x ∈ R,
ax =
exp[x ln a],
ln x
loga x = , x > 0.
ln a
1.2.7. Trigonometric functions. A function f (x) is periodic if there is a
positive constant a such that
f (x + a) = f (x)
for all x in the domain of f . If a is the smallest number with this property, we call
it the period of f .
The trigonometric functions are examples of periodic functions. Observe that
the period of sin(x) or cos(x) is 2π.
Example 1.2.6. Consider the function

f (x) = a sin(kx), x ∈ R,
where a ∈ R and k 6= 0. Since f (x) takes on values between −a and a, we cal |a|
the amplitude. Observe that the period of f (x) is 2π/|k|.
Remark 1.2.7. (Fourier series of periodic functions) suppose that f (x) is a periodic
function of period 2π. Then we may assume that f (x) is defined on [−π, π]. If f (x)
is integrable (for definition, see Chapter 6), we define
1 π 1 π
Z Z
(1.2.1) an := f (x) cos(nx) dx, bn := f (x) sin(nx) dx, n ∈ N≥0 .
π −π π −π
which are called the Fourier coefficients of f . The formal infinite sum
a0 X
(1.2.2) (Sf )(x) := + an cos(nx) + bn sin(nx)
2
n∈N
Figure 1.8. Graphs of the six basic trigonometric functions
is called the Fourier series of f . If f ∈ C 1 ([−π, π]), then a theorem in Fourier

theory asserts that Sf = f for each x ∈ [−π, π]; that is,
(1.2.3) f (x) = (Sf )(x), x ∈ [−π, π], if f ∈ C 1 ([−π, π]).
By choosing a suitable function f , we can prove
π2 X 1 1 1 1
(1.2.4) = = 2 + 2 + 2 + ··· .
6 n2 1 2 3
n∈N
Definition 1.2.8. A number is called algebraic if√it is the solution of a polynomial

equation with rational coefficients. For example, 2 is algebraic, as it satisfies the
equation x2 − 2 = 0. Numbers that are not algebraic are called transcendental.
Example 1.2.9. (π is transcendental) Swiss scientist Johann Heinrich Lambert (26

August 1728– 25 September 1777) in 1761 proved that π is irrational; French math-
ematician Adrien-Marie Legendre (18 September 1752– 10 January 1833) proved in
1794 that π is also irrational. In 1882, German mathematician Carl Louis Ferdinand
von Lindemann (12 April 1852– 6 March 1939) proved that π is transcendental, con-
firming a conjecture made by both Adrien-Marie Legendre and Leonhard Euler (15
April 1707– 18 September 1783).
Figure 1.9. Lambert, Legendre, and Euler
Figure 1.10. Lindemann and Gauss
Remark 1.2.10. (Gauss-Legendre algorithm) The Gauss-Legendre algorithm is

an algorithm to compute the digits of π. It is notable for being rapidly convergent,
with only 25 iterations producing 45 million correct digits of π.
√
(1) Choose a0 = 1, b0 = 1/ 2, t0 = 1/4, and p0 = 2.
(2) Consider
an + bn p
an+1 = , bn+1 = an bn , tn+1 = tn − pn (an − an+1 )2 , pn+1 = 2pn .
2
(3) π is then approximated as
(an + bn )2
π = lim .
n→∞ 4tn
The first three iterations give
3.140 · · · , 3.14159264 · · · , 3.14159265358979332382 · · · .
We call a function y = f (x) algebraic if it is the solution of an equation of the

form
Pn (x)y n + · · · + P1 (x)y + p0 y = 0
in which the coefficients are polynomial functions in x with rational coefficients. For
1
example, the function y = 1+x is algebraic, as it satisfies the equation (x+1)y −1 =
0. Functions that are not algebraic are called transcendental. All the trigonometric,
exponential, and logarithmic functions are transcendental functions.
Remark 1.2.11. 3The name “transcendental” comes from Leibniz in his 1682
paper where he proved sin x is not an algebraic function of x, and Euler was probably
the first person to define transcendental numbers in the modern sense.
(1) Joseph Liouville first proved the existence of transcendental numbers in
1844, and in 1851 gave the first decimal examples such as the Liouville
constant X 1
.
10k!
k∈N
(2) Johann Heinrich Lambert conjectured that e and π were both transcen-
dental numbers in his 1761 paper proving the number π is irrational.
(3) Charles Hermit in 1873 proved that e is transcendental.
(4) In 1874, Georg Cantor proved that the algebraic numbers are countable
and the real numbers are uncountable. In 1878, Cantor showed that there
are as many transcendental numbers as there are real numbers.
(5) In 1882, Ferdinand von Lindemann published a proof that the number π
is transcendental. He first showed
√
that e to any nonzero
√ algebraic power
is transcendental, and since e −1π = −1 is algebraic, −1π and therefore
π must be transcendental.
(6) In 1900, David Hilbert posed the Hilbert’s seventh problem: if a is an
algebraic number which is not zero or one, and b is an irrational algebraic
number, is ab necessarily transcendental? The affirmative answer was
provided in 1934 by the Helfond-Schneider theorem.
1.2.8. Graphing. The graph of

y = f (x) + a
is a vertical translation of the graph of y = f (x). If a > 0, the graph of y = f (x)
is shifted up a units; if a < 0, the graph of y = f (x) is shifted down |a| units.
The graph of
y = f (x − c)
is a horizontal translation of the graph y = f (x). If c > 0, the graph of y = f (x)
is shifted c units to the right; if c < 0, the graph of y = f (x) is shifted |c| units to
the left.
CHAPTER 2
Sequences and difference equations
In his chapter, we discuss models for populations that reproduce

at discrete times and we develop some of the theory needed to
analyze this type of model.
2.1. Sequences
A sequence if the function
f : N −→ R, n 7−→ f (n)
as a list of numbers a0 , a1 , a2 , · · · where an = f (n). We will write {an : n ∈ N≥0 }
or {an }n∈N≥0 if we mean the entire sequence.
Example 2.1.1. (1) an = (−1)n , n = 0, 1, 2, · · · .

(2) an = n2 , n = 0, 1, 2, · · · .
(3) an = (−1)n /(n + 1)2 , n = 0, 1, 2, · · · .
2.1.1. kth order recursion. From Example 2.1.1 (2), we see that
√
an+1 = (n + 1)2 = n2 + 2n + 1 = an + 2 an + 1;
thus an+1 depends only on an . In general, we have the following
Definition 2.1.2. We say a sequence {an }n∈N≥0 is kth order recursive if

an+1 = g(an , an−1 , · · · , an−k+1 )
for some function g. In particular, the 1th order recursion is of the form
an+1 = g(an );
the 2th order recursion is of the form
an+1 = g(an , an−1 ).
2.1.2. Limits. Looking at Example 2.1.1, we see that, as n goes to infinity,

(1) an can not be determined;
(2) an tends to infinity;
(3) an tends to zero.
17
18 2. SEQUENCES AND DIFFERENCE EQUATIONS
Figure 2.1. Limits
Definition 2.1.3. The sequence {an }n∈N or {an }n∈N≥0 has limit a < ∞, written
as limn∈∞ an = a, if, for every > 0, there exists an integer N (depending on )
such that
|an − a| <
whenever n > N . If the limit exists, the sequence is called convergent and we say
that an converges to a as n tends to infinity. If the sequence has no limit, it is
called divergent.
Proposition 2.1.4. If the sequence {an }n∈N or {an }n∈N≥0 is convergent, then the
limit of this sequence is unique.
Proof. Suppose we have two limits a and a0 . By the definition, for every
> 0, there exist integers N and N 0 such that
|an − a| < n > N, and |an − a0 | < n > N 0.
Consequently, for n > N + N 0 , we have
|a − a0 | = |(an − a) + (an − a0 )| ≤ |an − a| + |an − a0 | < 2.
Since the left-hand side does not depend on n and , letting → 0, we obtain
a = a0 .
1
Exercise 2.1.5. Show that limn→∞ n = 0.
Exercise 2.1.6. Show that the sequence {an }n∈N , where an = n, is divergent.
In order to calculus, we need the following fundamental results.

2.1. SEQUENCES 19
Theorem 2.1.7. If limn→∞ an and limn→∞ bn exist and c is a constant, then

(2.1.1) lim (an + bn ) = lim an + lim bn ,
n→∞ n→∞ n→∞
(2.1.2) lim (can ) = c lim an ,
n→∞ n→∞
(2.1.3) lim (an bn ) = lim an · lim bn ,
n→∞ n→∞ n→∞
an limn→∞ an
(2.1.4) lim = , provided lim bn 6= 0.
n→∞ bn limn→∞ bn n→∞
For example,

n+1 1 1
lim = lim 1 + = 1 + lim =1+0=1
n→∞ n n→∞ n n→∞ n
by (2.1.1) and Exercise 2.1.5.
Remark 2.1.8. The assumption in Theorem 2.1.7 that limn→∞ an and limn→∞ bn
exists is necessary. For example, consider an = n and bn = −n; then (2.1.1) does
not hold.
2.1.3. Recursions. We consider the sequences of the 1th order recursion

an+1 = g(an ), a0 is a given real number.
We start from several elementary special cases.
1. an+1 = can for some nonzero constant c. In this case, it is easily to see that
(2.1.5) an = can−1 = c2 an−2 = · · · = a0 cn .
2. an+1 = an + c for some constant c. In this case, we have
(2.1.6) an = a0 + cn.
3. an+1 = c1 an + c2 for sone constants c1 6= 1 and c2 . When c1 = the result
follows from 2. Consider the constant b such that
an+1 − b
= c1 ;
an − b
then
an+1 = c1 an + b(1 − c1 ), c2 = b(1 − c1 ).
c2
So b = 1−c1 and

c2 c2
an+1 − = c1 an − .
1 − c1 1 − c1
Applying (2.1.5) to this case, we arrive at

c2 c2
(2.1.7) an = + a0 − cn1 .
1 − c1 1 − c1
Example 2.1.9. Let an+1 = 4 − 2an with a0 = 43 . By (2.1.7) we have

4 4 4 4
an = + − (−2)n = .
1 − (−2) 3 1 − (−2) 3
In this case limn→∞ an = 34 .
In Example 2.1.9, we can give another short proof. Suppose the limit limn→∞ an =
a exists, by taking the limit of n, we have
a = lim an+1 = 4 − 2 lim an = 4 − 2a
n→∞ n→∞
which gives us a = 4/3. Thus approach shows that if we know the limit exists,
then we can take the limits on both-sides of an+1 = g(an ) to get the desired limit.
However, in order to applying this argument, we should know the existence at first.
This can be seen from the following fundamental theorem.
Theorem 2.1.10. Any bounded increasing or decreasing sequence has finite limit.
We say that a sequence {an }n∈N≥0 is bounded if |an | ≤ M for all n, where M
is some positive constant. A sequence {an }n∈N≥0 is increasing (or decreasing)
if an ≤ an+1 (or an ≥ an+1 ).
√
Example 2.1.11. Consider an+1 = 3an with a0 = 2. If limn→∞ an = a exists,
then √
a = 3a =⇒ a = 0 or a = 3.
We now prove that the limit of an exists. Since
√ √ √
a1 = 6 < 3, a2 = 3a1 < 3 × 3 = 3,
we claim that an ∈ (0, 3). If this claim holds for n, then
√ √
an+1 = 3an < 3 × 3 = 3, an+1 > 0,
by inductive hypothesis. Hence an ∈ (0, 3) for all n. From
r
an+1 3
= > 1,
an an
it follows that the sequence {an }n∈N≥0 is bounded and increasing, and therefore
the limit limn→∞ an exists.
Returning to the 1th order recursion

an+1 = g(an ).
We say that a is a fixed point of the sequence {an }n∈N≥0 if
a0 = a =⇒ an = a for all n.
For example, 4/3 is a fixed point of the sequence {an }n∈N≥0 . If a is a fixed point,
then
a = an+1 = g(an ) = g(a).
2.1. SEQUENCES 21
Figure 2.2. Fixed points of f (x)
Figure 2.3. Rudolf Otto Sigismund Lipschitz (1832–1903)
Definition 2.1.12. Given a function f : A → A. A number a ∈ A is called a fixed

point of f if a = f (a). Thus a is an intersection point of the functions y = x and
y = f (x). (see Figure 2.2)
Observe that if a is a fixed point of the sequence an+1 = g(an ), then a is a fixed
point of the function g : R → R. Conversely, if a is a fixed point of the function
f : R → R, then we can construct a sequence {an }n∈N≥0 of which a is a fixed
point.
Definition 2.1.13. A function f : A → B is called Lipschitz continuous if there

exists a nonnegative constant L such that
|f (x) − f (y)| ≤ L|x − y|
whenever x, y ∈ A.
√
For example f (x) = x2 + 5/2, x ∈ R, is a Lipschitz continuous function with
L ≤ 1/2.
Proposition 2.1.14. Let f : R → R be a Lipschitz continuous function with

constant L ∈ (0, 1). If a is a fixed point of f , then there exists a sequence {xn }n∈N≥0
of which a is a fixed point and limn→∞ an = a.
Proof. Given a point a0 ∈ R, consider

a1 := f (a0 ), a2 := f (a1 ) = f (f (a0 )) := f [2] (a0 ), an := f [n] (a0 ).
Then
|a1 − a| = |f (a0 ) − f (a)| ≤ L|a0 − a|;
in general, we have
|an − a| = f (an−1 ) − f (a)| ≤ L|an−1 − a| ≤ · · · ≤ Ln |a0 − a|
for all n. Letting n → ∞ and noting that L ∈ (0, 1) imply limn→∞ an = a. In
particular, when a0 = a, we have |an − a| ≤ Ln |a − a| = 0 so that an = a.
2.2. Population models

We denote by the population size at time t by Nt , t = 0, 1, 2, · · · , and consider
the model
(2.2.1) Nt+1 = f (Nt ),
where the function f describes the density dependence of the population dynamics.
A simple model is
Nt 1
(2.2.2) Nt+1 = RNt or = .
Nt+1 R
When R > 1, the population size will grow indefinitely, provided that N0 > 0.
2.2.1. Beverton-Holt model. The Beverton-Holt model is a classic discrete-
time population model which is a nonlinear version of (2.2.2)
Nt 1 1 − R1 RNt
(2.2.3) = + Nt or Nt+1 = ,
Nt+1 R K 1 + R−1
K Nt
where R > 1 and K > 0 is the carrying capacity.
If N0 > 0, then Nt > 0 for all n; moreover

Nt+1 Nt 1 K
= = 1− ,
KR K + (R − 1)Nt R−1 K + (R − 1)Nt
R − 1 + R−1
K Nt K + Nt
Nt+1 − Nt = Nt = (R − 1)Nt ,
1 + R−1
K Nt
K + (R − 1)Nt
R
which imply 0 < Nt+1 < R−1 K and Nt+1 ≥ Nt . Thus the sequence {Nt }t∈N≥0 is
bounded and increasing. By Theorem 2.1.10, the limit
N := lim Nt
t→∞
exists and is nonzero. Taking t → ∞ in (2.2.3) yields

1 1 − R1
1= + N =⇒ N = K.
R K
2.2. POPULATION MODELS 23
2.2.2. Discrete logistic equation. The most popular discrete-time single-

species model is the discrete logistic equation

Nt
(2.2.4) Nt+1 = Nt 1 + R 1 − ,
K
where R and K are positive constants. R is called the growth parameter and K
is called the carrying capacity.
Defining
R
(2.2.5) xt := Nt ,
K(1 + R)
we then obtain
(2.2.6) xt+1 = (1 + R)xt (1 − xt ).
Exercise 2.2.1. Prove (2.2.6).
The long-term behavior of the discrete logistic equation is very complicated,

but, we can compute its fixed points. If x is a nonzero fixed point of (2.2.6), then
R
x = (1 + R)x(1 − x) =⇒ x = ;
1+R
If N is the fixed point corresponding to x, then
K(1 + R) K(1 + R) R
N= x= = K.
R R 1+R
2.2.3. Ricker’s curve. The Ricker’s curve is given by

Nt
(2.2.7) Nt+1 = Nt exp R 1 −
K
where R and K are positive parameters. As in the discrete logistic model, R is
called growth parameter and K is the carrying capacity. If N is a nonzero fixed
point then

N
N = N exp R 1 − =⇒ N = K.
R
2.2.4. Harvesting model. The harvesting model is given by

(1 − c)Nt
(2.2.8) Nt+1 = (1 − c)Nt exp R 1 −
K
where R and K are positive constant, and c ∈ (0, 1). If N is a nonzero fixed point
then

K 1
N= 1 + ln(1 − c) .
1−c R
Figure 2.4. Fibonacci sequences
Figure 2.5. Fibonacci (1175–1250) and a page of Fibonacci’s

Liber Abaci from the Biblioteca Nazionale di Firenze showing (in
box on right) the Fibonacci sequence with the position in the se-
quence labeled in Roman numerals and the value in Hindu-Arabic
numerals.
2.2.5. Fibonacci sequences. A famous example of the 2th order recursions

is the Fibonacci sequence:
(2.2.9) Nt+1 = Nt + Nt−1 , N0 = 1 and N1 = 1.
The equation comes from the following problem posed in 1202 of his Liber Abaci
(Book of Calculation) by Leonardo Pisano Bigollo (also known as Leonardo of Pisa
or most commonly Fibonacci):
How many pairs of rabbits are produced if each pair reproduces
one pair of rabbits at age one month and another pair of rabbits
at age two months and initially there is one pair of newborn
rabbits?
The Fibonacci sequence (2.2.9) has a closed-form solution
ϕt − ψ t
(2.2.10) Nt = ,
ϕ−ψ
2.2. POPULATION MODELS 25
where
√
1+ 5
(2.2.11) ϕ := ≈ 1.6180339887 · · ·
2
is the golden ratio or golden mean, and
√
1− 5 1
(2.2.12) ψ := = 1 − ϕ = − ≈ −0.6180339887 · · · .
2 ϕ
Exercise 2.2.2. Using (2.2.10) to show

Nt
(2.2.13) lim = ϕ.
t→∞ Nt+1
CHAPTER 3
Limits and continuity
In this chapter we give the formal definition of limits and then

continuities. The key description is the “-δ” technique.
3.1. Limits
The statement
(3.1.1) lim f (x) = L

x→c
where c, L are finite, means that, for every > 0, there exists a number δ > 0 such
that
|f (x) − L| <
whenever 0 < |x − c| < δ. In this case, we say that f (x) converges to L. If this
limit does not exist, we say that f (x) diverges as x tends to c.
Remark 3.1.1. In the above definition, we exclude the value x = c from the
statement. (This is done in the inequality 0 < |x − c|.) A reason for that is the
function f (x) may not be defined at x = c. For example,

x, x 6= 0,
f (x) =
1, x = 0.
In this case, we have limx→0 f (x) = 0 6= f (0).
According to our definition of limits, we may consider one-sided limits:

(1) limx→c+ f (x) = L means that for every > 0 there exists a number δ > 0
such that
|f (x) − L| < whenever 0 < x − c < δ.
(2) limx→c− f (x) = L means that for every > 0 there exists a number δ > 0
such that
|f (x) − L| < whenever 0 < c − x < δ.
Observe that
lim f (x) = L ⇐⇒ lim f (x) = L and lim f (x) = L.

x→c x→c+ x→c−
27
28 3. LIMITS AND CONTINUITY
Figure 3.1. The -δ definition of limits
Example 3.1.2. Consider

f (x) = |x|, x ∈ R.
Then
lim f (x) = 1, lim f (x0 = −1,
x→0+ x→0−
however, the limit limx→0 f (x) does not exist.
The statement
(3.1.2) lim f (x) = ∞ (or − ∞)
x→c
means that, for every M > 0, there exists a δ > 0 such that
f (x) > M (or < −M )
whenever 0 < |x − c| < δ. Similarly, we can define
lim f (x) = ∞, lim f (x) = −∞, lim f (x) = ∞, lim f (x) = −∞.
x→c+ x→c+ x→c− x→c−
1 1
Example 3.1.3. limx→0+ x = ∞ and limx→0− x = −∞.
The statement
(3.1.3) lim f (x) = L
x→∞
means that, for every > 0, there exists an x0 > 0 such that
|f (x) − L| <
whenever x > x0 . Similarly, we can define limx→−∞ f (x) = L.
3.1. LIMITS 29
3.1.1. Limit laws. As Theorem 2.1.7, we have
Theorem 3.1.4. If limx→c f (x) (c can be infinity) and limx→c g(x) exist and a is
a constant, then
(3.1.4) lim [af (x)] = a lim f (x),
x→c x→c
(3.1.5) lim [f (x) + g(x)] = lim f (x) + lim g(x),
x→c x→c x→c
(3.1.6) lim [f (x)g(x)] = lim f (x) · lim g(x),
x→c x→c x→c
f (x) limx→c f (x)
(3.1.7) lim = , provided lim g(x) 6= 0.
x→c g(x) limx→c g(x) x→c
Example 3.1.5. Using (3.1.7) we have


xn  ∞, n > m,
lim m = 1, n = m,
x→∞ x
0, n < m.

Exercise 3.1.6. Show that


∞, deg(p) > deg(q),
p(x) 
(3.1.8) lim = L 6= 0, deg(p) = deg(q),
x→∞ q(x)
0, deg(p) < deg(q).

Here p(x) and q(x) are two polynomials of degrees deg(p) and deg(q), respectively,
and L is the ratio of the coefficients of the leading terms in the numerator and
denominator.
Another useful and simple fact is

1
(3.1.9) = 0. lim
ex x→∞
3.1.2. Sandwich theorem. To calculate limits, we need the following
Theorem 3.1.7. If f (x) ≤ g(x) ≤ h(x) for all x in an open interval that contains
c (except possibly at c) and
lim f (x) = lim h(x) = L,
x→c x→c
then
lim g(x) = L.
x→c
Proof. Given an > 0. By the definition, we can find a positive number δ

(depending on ) such that
|f (x) − L| < , |h(x) − L| <
Figure 3.2. Sandwich theorem
sin x sin x
Figure 3.3. x and the inequality cos x ≤ x ≤1
whenever 0 < |x − c| < δ. Hence

g(x) − L ≤ h(x) − L < , g(x) − L > f (x) − L > −;
thus |g(x) − L| < . Consequently, limx→c g(x) = L.
Remark 3.1.8. If f (x) ≤ g(x) ≤ h(x) for all x > x0 , where x0 is a positive fixed
number, and
lim f (x) = lim h(x) = L,
x→∞ x→∞
then
lim g(x) = L.
x→∞
Looking at the Figure 3.3, we have

area of the triangle OAP ≤ area of the sector OBP ≤ area of the triangle OBC;
1
therefore 2 sin x ≤ 21 x ≤ 1 sin x
2 cos x . Thus
sin x π
(3.1.10) cos x ≤ ≤ 1, 0<x< .
x 2
Theorem 3.1.9. We have

sin x
(3.1.11) lim = 1.
x→0 x
3.2. CONTINUITY 31
Since 1 − cos x = 2 sin2 x, it follows that

1 − cos x 2 sin2 x
2 sin x2 x
= = x · sin .
x x 2 2
The first factor tends to 1 while the second factor tends to 0. So the limit of
(1 − cos x)/x should be 0. In general,
Proposition 3.1.10. If f (x) is bounded (i.e., |f (x)| ≤ M for some positive con-
stant M ) and limx→c g(x) = 0, then
lim f (x)g(x) = 0.
x→c
Proof. Given > 0. There exists a positive number δ such that

|g(x)| <
M
whenever 0 < |x − c| < δ. Hence

|f (x)g(x)| ≤ M |g(x)| < M · =
M
whenever 0 < |x − c| < δ. Thus limx→c f (x)g(x) = 0.
Exercise 3.1.11. Prove

sin x(1 − cos x) sin(πx)
lim = 0, lim = π.
x→0 x2 x→0 x
3.2. Continuity
A function f is said to be continuous at x = c if
(3.2.1) lim f (x) = f (c).
x→c
To check whether a function is continuous at x = c, we need to check the following
three conditions:
(1) f (x) is defined at x = c,
(2) limx→c f (x) exists,
(3) limx→c f (x) = f (c).
If any of these three conditions fails, the function is discontinuous at x = c.
Example 3.2.1. The function defined in Remark 3.1.1 is discontinuous.
A function f is said to be continuous from the right at x = c if

(3.2.2) lim f (x) = f (c)
x→c+
and continuous from the left at x = c if

(3.2.3) lim f (x) = f (c).
x→c−
Example 3.2.2. Consider the function


 1, x > 0,
f (x) = 0, x = 0,
−1, x < 0.

Then limx→0+ f (x) = 1 and limx→0− f (x) = −1.
3.2.1. Properties of continuous functions. Using the limit law, Theorem

3.1.4, we can prove
Theorem 3.2.3. Suppose that a is a constant and the functions f and g are con-
tinuous at x = c. Then the following functions are continuous at x = c:
(1) a · f ,
(2) f + g,
(3) f · g,
(4) f /g provided that g(c) 6= 0.
The following functions are continuous wherever they are defined:

(1) polynomial functions
(2) rational functions
(3) power functions
(4) trigonometric functions
(5) exponential functions of the form ax , a > 0 and a 6= 1
(6) logarithmic functions of the form loga x, a > 0 and a 6= 1
Theorem 3.2.4. If g(x) is continuous at x = c with g(c) = L and f (x) is contin-

uous at x = L, then (f ◦ g)(x) is continuous at x = c. In particular
h i
(3.2.4) lim (f ◦ g)(x) = lim f [g(x)] = f lim g(x) = f [g(c)] = (f ◦ g)(c).
x→c x→c x→c
Proof. Note that (f ◦ g)(c) = f [g(c)] exists. Given an > 0. There exists a
positive number δ such that
|f (y) − f (L)| <
whenever |y − L| < δ. For such δ, we can find another positive number δ 0 such that
|g(x) − g(c)| < δ
whenever |x − c| < δ 0 . In particular,
|(f ◦ g)(x) − (f ◦ g)(c)| <
whenever |x − c| < δ 0 . Thus limx→c (f ◦ g)(x) = (f ◦ g)(c).

3.2. CONTINUITY 33
3.2.2. Cauchy’s functional equations. Cauchy’s functional equation

is given by
(3.2.5) f (x + y) = f (x) + f (y), x, y ∈ R.
Observe that f (x) = cx satisfies this functional equation (3.2.5). Conversely, any
continuous function f (x) satisfying (3.2.5) is of the form f (x) = cx for some con-
stant c.
Suppose f (x) is a function (not necessarily continuous) satisfying (3.2.5).
(1) Letting x = y = 0, we have f (0) = 0.
(2) Letting y = −x we have f (−x) = −f (x) for all x ∈ R.
(3) For any integer n we have f (n) = f (1)n.
(4) For any nonzero integer n we have f (1/n) = f (1)/n.
(5) If p/q ∈ Q then f (p/q) = pf (1/q) = f (1) · pq .
Hence f (x) = f (1)x for any x ∈ Q.
Now we in addition assume f is continuous. Recall that any irrational number
is the limit of a sequence of rational numbers. For example, for
√
2 = 1.414 · · · ,
√
we can take a0 = 1, a1 = 1.4, a2 = 1.41, · · · and then limn→∞ an = 2. If x ∈ R\Q,
then x = limn→∞ xn for a sequence of rational numbers xn . Since f is continuous,
we have

f (x) = f lim xn = lim f (xn ) = lim f (1)xn = f (1)x;
n→∞ n→∞ n→∞
thus f (x) = f (1)x still holds for x ∈ R \ Q. Consequently, f (x) = cx where

c := f (1).
Proposition 3.2.5. A continuous function f satisfying (3.2.5) must be of the form

f (x) = f (1)x.
Exercise 3.2.6. Any continuous function f satisfying

(3.2.6) f (x + y) = f (x)f (y), x, y ∈ R,
f (0) 6= 0 and f (1) > 0, must be of the form f (x) = ax where a = f (1).
3.2.3. Intermediate-value theorem. An important property of continuous

functions is the intermediate-value theorem. (see Figure 3.4)
Theorem 3.2.7. Suppose that f is continuous on the closed interval [a, b]. If M
is any real number with f (a) < M < f (b) or f (b) < M < f (a), then there exists at
least one number c on the open interval (a, b) such that f (c) = M .
Figure 3.4. Intermediate-value theorem
Remark 3.2.8. Consider the function f (x) = x. Then for each M ∈ (a, b), we
have a unique M ∈ (a, b) with f (M ) = M . Hence we can not weaken the open
interval (a, b) in Theorem 3.2.7 to the closed interval [a, b].
Remark 3.2.9. Suppose that f is continuous function defined on [a, b]. If f (a) <
0 < f (b), then Theorem 3.2.7 shows that f (x) = 0 has at least one solution in (a, b).
CHAPTER 4
Differentiation
In this chapter, we introduce derivatives and chain rule.
4.1. Derivatives
The derivative of a function f at x, denoted f 0 (x), is
f (x + h) − f (x)
(4.1.1) f 0 (x) := lim
h→0 h
provided that the limit exists. If the limit exists, then we say that f is differen-
tiable at x. (see Figure 4.1) The quotient
f (x + h) − f (x)
h
is called the difference quotient, and we denote it by ∆f ∆x . We use the Leibniz
notation
dy df d
(4.1.2) y0 = = f 0 (x) = = f (x).
dx dx dx
For particular value c, w can write

df
(4.1.3) = f 0 (c).
dx x=c
4.1.1. Geometric interpretation. If the derivative of a function f exists at

x = c, then the tangent line at x = c is the line going through the point (c, f (c))
with slope f 0 (c). The equation of the tangent line is given by
(4.1.4) y − f (c) = f 0 (c)(x − c).
Figure 4.1. Derivative and tangent line

35
36 4. DIFFERENTIATION
Remark 4.1.1. If the derivative of a function f exists at x = c, then

f (c + h) − f (c)
f 0 (c) = lim .
h→0 h
In particular,

f (c + h) − f (c − h) 1 f (c + h) − f (c) f (c − h) − f (c)
lim = lim + = f 0 (c).
h→0 2h 2 h→0 h −h
However, the converse is not true. Consider
f (x) = |x|, c = 0.
Then
0 1, x > 0,
f (x) =
−1, x < 0.
Hence the derivative of f (x) does not exist. But
f (c + h) − f (c − h)
= 0.
2h
4.1.2. Differentiability and continuity. A continuous function may not be

differentiable, see f (x) = |x|. Conversely, we have
Theorem 4.1.2. If f is differentiable at x = c, then f is also continuous at x = c.
Proof. From
f (x) − f (c)
f (x) − f (c) = · (x − c),
x−c
we have
f (x) − f (c)
lim [f (x) − f (c)] = lim · (x − c) = f 0 (c) · 0 = 0
x→c x→c x−c
by Proposition 3.1.10.
4.2. Basic rules and chain rule

If f (x) is the constant function f (x) ≡ a, then
d
f (x) = 0.
dx
4.2.1. Power rule. In general, we have

d n
(4.2.1) (x ) = nxn−1 .
dx
4.2. BASIC RULES AND CHAIN RULE 37
Theorem 4.2.1. Suppose a is a constant and f (x) and g(x) are differentiable at
x. Then
d d
(4.2.2) [af (x)] = a f (x),
dx dx
d d d
(4.2.3) [f (x) + g(x)] = f (x) + g(x).
dx dx dx
4.2.2. Product and quotient rules. From

d 5 d 3 d 2
x = 5x4 6= 6x3 = x x ,
dx dx dx
we see that (f (x)g(x))0 6= f 0 (x) · g 0 (x).
Theorem 4.2.2. If f (x) and g(x) are differentiable at x, then

d
(4.2.4) [f (x)g(x)] = f 0 (x)g(x) + f (x)g 0 (x).
dx
Theorem 4.2.3. If f (x) and g(x) are differentiable at x, and g(x) 6= 0, then
f 0 (x)g(x) − f (x)g 0 (x)

d f (x)
(4.2.5) = .
dx g(x) [g(x)]2
By Theorem 4.2.3, we have
d −n
(4.2.6) (x ) = −nx−n−1 .
dx
In general,
d r
(4.2.7) (x ) = rxr−1 .
dx
4.2.3. Chain rule. To find the derivative of composite functions, we need the
chain rule.
Theorem 4.2.4. (Chain rule) If g is differentiable at x and f is differentiable at

y = g(x), then the composite function (f ◦ g)(x) = f [g(x)] is differentiable at x,
and
(4.2.8) (f ◦ g)0 (x) = f 0 [g(x)]g 0 (x).
38 4. DIFFERENTIATION
Example 4.2.5. We have

d d
(4.2.9) sin x = cos x, cos x = − sin x,
dx dx
d d
(4.2.10) tan x = sec2 x, cot x = − csc2 x,
dx dx
d d
(4.2.11) sec x = sec x · tan x, csc x = − csc x · cot x,
dx dx
d x
(4.2.12) e = ex ,
dx
d x
(4.2.13) a = ax · ln a,
dx
d 1
(4.2.14) ln x = ,
dx x
d 1
(4.2.15) loga x = .
dx (ln a)x
Theorem 4.2.6. (Derivative of the inverse function) If f (x) is one-to-one and

differentiable with inverse function f −1 (x) and f 0 [f −1 (x)] 6= 0, then
d −1 1
(4.2.16) f (x) = 0 −1 .
dx f [f (x)]
For example,
d d 1
(4.2.17) arcsin x = sin−1 x = √ ,
dx dx 1 − x2
d d 1
(4.2.18) arctan x = tan−1 x = .
dx dx 1 + x2
4.2.4. Linear approximation and error propagation. Assume that y =
f (x) is differentiable at x = a; then
(4.2.19) L(x) := f (a) + f 0 (a)(x − a)
is the tangent line approximation or the linearization of f at x = a. If |x − a|
is sufficiently small, we have
(4.2.20) f (x) ≈ L(x) = f (a) + f 0 (a)(x − a).
CHAPTER 5
Applications of differentiation
In this chapter, we discuss the behavior of functions by using

differentiations.
5.1. Extrema and the mean-value theorem

This section defines a extrema, gives conditions that guarantee extrema (via
the extreme-value theorem), and provides a characterization of extrema (Fermat’s
theorem).
5.1.1. Extreme-value theorem. Let f be a function defined on the set D

that contains the number c. Then f has a global (or absolute) maximum at
x = c if
f (c) ≥ f (x)
for all x ∈ D, and f has a global (or absolute) minimum at x = c if
f (c) ≤ f (x)
for all x ∈ D. (see Figure 5.1)
Theorem 5.1.1. (Extreme-value theorem) If f is continuous on a closed interval

[a, b], −∞ < a < b < ∞, then f has a global maximum and a global minimum in
[a, b].
The proof of Theorem 5.1.1 is beyond the scope of this course and will be
omitted.
Figure 5.1. Absolute maximum and minimim

39
40 5. APPLICATIONS OF DIFFERENTIATION
Figure 5.2. Relative maximum and minimim
Remark 5.1.2. (1) Theorem 5.1.1 only tells the existence of global (or absolute)
extrema (global maximum or global minimum). We can find functions that have
more than one global extrema; for example,

1, 0 ≤ x ≤ 1,
f (x) =
2 − x, 1 ≤ x ≤ 2.
(2) If f is continuous on an open interval (a, b), then f may not have a global
extrema. For example, f (x) = 1/x with x ∈ (0, 1).
5.1.2. Local extrema. A function f defined on a set D has a local (or rel-
ative) maximum at a point c if there exists a δ > 0 such that
f (c) ≥ f (x)
for all x ∈ (c − δ, c + δ) ∩ D. A function f defined on a set D has a local (or
relative) minimum at a point c if there exists a δ > 0 such that
f (c) ≤ f (x)
for all x ∈ (c − δ, c + δ) ∩ D. Local maxima and local minima are collectively called
local (or relative) extrema.
Theorem 5.1.3. (Fermat’s theorem) If f has a local extremum at an interior

point c and f 0 (c) exists, then f 0 (c) = 0.
Proof. We only prove the case that f has a local minimum at c. Then there
exists a δ > 0 such that
f (c) ≤ f (x)
for all x ∈ (c − δ, c + δ) ∩ D. For x ∈ (c, c + δ) ∩ D, we have
f (c) − f (x) f (x) − f (c)
≤0 or ≥ 0;
x−c x−c
thus
f (x) − f (c)
lim ≥ 0.
x→c+ x−c
Similarly, we can show that
f (x) − f (c)
lim ≤ 0.
x→c− x−c
5.1. EXTREMA AND THE MEAN-VALUE THEOREM 41
Figure 5.3. Rolle’s theorem
Since f 0 (c) exists, we must have f 0 (c) = 0.
However, the converse of Theorem 5.1.3 is not true. For example, consider thee
function f (x) = x, x ∈ [−1, 1]; we see that f 0 (0) = 0 but f has no local extremums
in (−1, 1).
5.1.3. Mean-value theorem. For the function f (x) = x2 , x ∈ [−1, 1], we

easily see that f 0 (0) = 0. This result also holds for general differentiable functions.
Theorem 5.1.4. (Rolle’s theorem) If f is continuous on the closed interval [a, b]

and differentiable on the open interval (a, b), and if f (a) = f (b), then there exists
a number c ∈ (a, b) such that f 0 (c) = 0.
Proof. If f is a constant function, then the assertion is obvious. Now we

suppose that f is nonconstant. Since f is continuous on the closed interval [a, b],
according to Theorem 5.1.1, f has a global maximum and a global minimum in
[a, b]. For example, we assume that f has a global maximum at x0 ∈ [a, b]. If
x0 = a, then
f (a) = f (x0 ) ≥ f (a) =⇒ f (x0 ) = f (a) = f (b).
Hence f has a global maximum at a and b. By Theorem 5.1.1 again, f has a global
minimum at c ∈ [a, b]. Since f is nonconstant, we must have c ∈ (a, b). By Theorem
5.1.3, we get f 0 (c) = 0.
Theorem 5.1.5. (Mean-value theorem (MVT)) If f is continnuous on the closed

interval [a, b] and differentiable on the open interval (a, b), then there exists at least
one number c ∈ (a, b) such that
f (b) − f (a)
= f 0 (c).
b−a
Figure 5.4. Mean-value theorem
Proof. Define
f (b) − f (a)
F (x) := f (x) − (x − a), x ∈ [a, b].
b−a
Then F is continuous on [a, b] and differentiable on (a, b); furthermore F (a) =
f (a) = F (b). By Theorem 5.1.4, F 0 (c) = 0 for some c ∈ (a, b). Since
f (b) − f (a)
F 0 (x) = f 0 (x) − ,
b−a
f (b)−f (a)
it follows that 0 = F 0 (c) = f 0 (c) − b−a .
Corollary 5.1.6. If f is continuous on the closed interval [a, b] and differentiable

on the open interval (a, b) such that
m ≤ f 0 (x) ≤ M
for all x ∈ (a, b), then
m(b − a) ≤ f (b) − f (a) ≤ M (b − a).
Exercise 5.1.7. Prove Corollary 5.1.6.
If f is a constant, then f 0 (x) = 0 for all x ∈ R. Conversely we have
Corollary 5.1.8. If f is continuous on the closed interval [a, b] and differentiable

on the open interval (a, b), with f 0 (x) = 0 for all x ∈ (a, b), then f is constant on
[a, b].
Exercise 5.1.9. Prove Corollary 5.1.8.

5.2. MONOTONICITY, CONCAVITY, AND INFLECTION POINTS 43
Exercise 5.1.10. If f is continuous on the closed interval [a, b] and differentiable

on the open interval (a, b), then there exists a number c ∈ (a, b) such that
f (a) + f (b)
f 0 (c) = 0 or f (c) = .
2
(Hint: Consider the function F (x) = [f (x) − f (a)][f (x) − f (b)].)
5.2. Monotonicity, concavity, and inflection points

This section discusses the important concepts of monotonicity– whether a func-
tion is decreasing or increasing–and concavity–whether a function bends upwards
or downward.
5.2.1. Monotonicity. A function f defined on an interval I is called (strictly)
increasing on I if
f (x1 ) < f (x2 )
whenever x1 < x2 in I, and is called (strictly) decreasing on I if
f (x1 ) > f (x2 )
whenever x1 < x2 in I. An increasing or decreasing function is called monotonic.
Remark 5.2.1. The monotonicity of a function f defined on the interval I. For

example, consider the function f (x) = |x|, x ∈ [−1, 1]. Then f is increasing on
[0, 1] while is decreasing on [−1, 0].
Proposition 5.2.2. (First-derivative test for monotonicity) Suppose f is continu-

ous on [a, b] and differentiable on (a, b).
(a) If f 0 (x) > 0 for all x ∈ (a, b), then f is increasing on [a, b].
(b) If f 0 (x) < 0 for all x ∈ (a, b), then f is decreasing on [a, b].
Proof. We only prove part (a). Fixed x1 < x2 in (a, b). By Theorem 5.1.5,
we have
f (x2 ) − f (x1 )
= f 0 (c)
x2 − x1
for some c ∈ (x1 , x2 ). Since f 0 (x) > 0 for all x ∈ (a, b), we get f (x2 ) > f (x1 ).
5.2.2. Concavity. A differentiable function f (x) is concave up on an inter-
val I if the first derivative f 0 (x) is an increasing function on I. f (x) is concave
down on an interval I if the first derivative f 0 (x) os a decreasing function on I.
Theorem 5.2.3. (Second-derivative test for concavity) Suppose f is rwice differ-

entiable on an open interval I.
(a) If f 00 (x) > 0 for all x ∈ I, then f is concave up on I.
Figure 5.5. Inflection points
(b) If f 00 (x) < 0 for all x ∈ I, then f is concave down on I.
Exercise 5.2.4. Prove Theorem 5.2.3.
5.2.3. Extrema. A continuous function has a local minimum at c if the func-

tion is decreasing to the left of c and increasing to the right of c. A continuous
function has a local maximum at c if the function is increasing to the left of c and
decreasing to the right of c.
Theorem 5.2.5. (Second-derivative test for local extrema) Suppose f is twice

differentiable on an open interval containing c.
(a) If f 0 (c) = 0 and f 00 (c) < 0, then f has a local maximum at x = c.
(b) If f 0 (c) = 0 and f 00 (c) > 0, then f has a local minimum at x = c.
5.2.4. Inflection points. Inflection points are points where the concav-
ity of a function changes–that is, where the function changes from concave up to
concave down or from concave down to concave up.
Example 5.2.6. 0 is the only one inflection point of f (x) = x3 , x ∈ R.
If f (x) is twice differentiable and has an inflection point at x = c, then f 00 (c) =

0. (see Figure 5.5)
5.2.5. Asymptotes. A line y = b is a horizontal asymptote if either

lim f (x) = b or lim f (x) = b.
x→−∞ x→∞
5.3. L’HOSPITAL’S RULE 45
Figure 5.6. Guillaume François Antoine, Marquis de l’Hôpital

(1661–1704)
A line x = c is a vertical asymptote if

lim f (x) = +∞ or lim f (x) = −∞
x→c+ x→c+
or
lim f (x) = +∞ or lim f (x) = −∞.
x→c− x→c−
5.3. L’Hospital’s rule

The rule is named after 17th-century French mathematician Guillaume de
l’Hôpital, who published the rule in his book Analyse des Infiniment Petits pour
l’Intelligence des Lignes Courbes (literal translation: Analysis of the Infinitely Small
for the Understanding of Curved Lines) (1696), the first textbook on differential
calculus. However, it is believed that the rule was discovered by the Swiss mathe-
matician Johann Bernoulli.
Theorem 5.3.1. (L’Hôospital’s rule) Suppose that f and g are differentiable func-
tions and that
lim f (x) = lim g(x) = 0
x→a x→a
or
lim f (x) = lim g(x) = ∞.
x→a x→a
If
f 0 (x)
lim =L
x→a g 0 (x)
then
f (x)
lim = L.
x→a g(x)
Proof. We treat with the first case. Define

f (a) := lim f (x), g(a) := lim g(x).
x→a x→a
Then f (x) and g(x) are continuous at a. Fixed x > a. From Theorem 5.1.5, there
exist x1 , x2 ∈ (a, x) such that
f (x) − f (a) g(x) − g(a)
= f 0 (x1 ), = g 0 (x2 ).
x−a x−a
Observe that x1 , x2 → a+ as x → a+. Hence
f (x) f 0 (x1 )(x − a) f 0 (x1 )
lim = lim 0 = lim 0 = L.
x→a+ g(x) x→a+ g (x2 )(x − a) x→a+ g (x2 )
Similarly, we can show limx→a− f (x)/g(x) = L.
Example 5.3.2. (1)(∞/∞) Evaluate

ln x
lim .
x→∞ x
(2) (0 · ∞) Evaluate
lim x ln x.
x→0+
(3) (0/0) Evaluate
sin x
lim .
x→0 x
(4) (∞ − ∞) Evaluate p
lim (x − x2 + x).
x→∞
(5) (00 ) Evaluate
lim xx .
x→0+
Proof. For (1),

ln x 1/x
lim = lim = 0.
x→∞ x x→∞ 1
For (2), where y = 1/x,
− ln y
lim x ln x = lim = 0.
x→0+ y→∞ y
For (3),
sin x cos x
lim = lim = 1.
x→0 x x→0 1
For (4),
p −x −1 1
lim (x − x2 + x) = lim √ = lim q =− .
x→∞ x→∞ x+ x2 + x x→∞ 1 + 1 + 1 2
x
For (5),
lim xx = lim ex ln x = lim e0 = 1
x→0+ x→0+ x→0+
by (2).
5.4. ANTIDERIVATIVE 47
Figure 5.7. Antiderivatives
5.4. Antiderivative
A function F is called an antiderivative of f on an interval I if F 0 (x) = f (x)
for all x ∈ I. (see Figure 5.7)
Corollary 5.4.1. If F (x) and G(x) are antiderivatives of the continuous function
f (x) on an interval I, then there exists a constant C such that
G(x) = F (x) + C
for all x ∈ I.
Proof. By definition, we have (G − F )0 (x) = f (x) − f (x) = 0 for all x ∈ I.

According to Corollary 5.1.8, G(x) − F (x) =constant.
CHAPTER 6
Integrations
In this chapter, we give the definition on integrals and introduce

the fundamental theorem of calculus.
6.1. Definite integrals

Let f be a function on the interval [a, b]. We partition [a, b] into n subinterval
by choosing n − 1 numbers x1 , · · · , xn−1 in (a, b) such that
a = x0 < x1 < x2 < · · · < xn−1 < xn = b.
The n subintervals
[x0 , x1 ], [x1 , x2 ], · · · , [xn−1 , xn ]
form a partition of [a, b], which we denote by P = [x0 , x1 , x2 , · · · , xn ]. The par-
tition P depends on the number of subintervals n and on the choice of points
x0 , x1 , · · · , xn . The length of the kth subinterval [xk−1 , xk ] is denoted by ∆xk .
The length of the longest subinterval is called the norm of P and is denoted by
||P||; thus
||P|| = max{∆x1 , ∆x2 , · · · , ∆xn }.
In each subinterval [xk−1 , xk ]. we choose a point ck and construct a rectangle with
base ∆xk and height |f (ck )|. (see Figure 6.1) The sum of these products is denoted
by SP ; that is,
n
X
(6.1.1) SP := f (ck )∆xk .
k=1
The value of the sum depends on the choice of the partition P and the choice of
the points ck ∈ [xk−1 , xk ] and is called a Riemann sum for f on [a, b].
Figure 6.1. Riemann sum

49
50 6. INTEGRATIONS
6.1.1. Riemann integrals. Let f be a function on [a, b].
Definition 6.1.1. Let P = [x0 , x1 , x2 , · · · , xn ], n = 1, 2, · · · , be a sequence of

partition of [a, b] with ||P|| → 0. Set ∆xk = xk − xk−1 and ck ∈ [xk−1 , xk ]. The
definite integral of f from a to b is
Z b Xn
(6.1.2) f (x) dx := lim f (ck )∆xk
a ||P||→0
k=1
if the limit exists, in which case f is said to be (Riemann) integrable on the
interval [a, b]. Moreover, f (x) is called the integrand.
The phase “if the limit exists” means, in particular, that the value of lim||P||→0 SP
does nor depend on how we choose the partitions and the points ck ∈ [xk−1 , xk ] as
we rake the limit.
Theorem 6.1.2. All continuous functions are Riemann integrals; that is, if f (x)
is continuous on [a, b] then
Z b
f (x) dx
a
exists.
Remark 6.1.3. (1) We can find a function that is Riemann integral but not con-
tinuous. For example,

|x|, x ∈ [−1, 0) ∪ (0, 1],
f (x) =
1, x = 0.
(2) Consider the function

x, x ∈ Q ∩ [0, 1],
f (x) =
0, x ∈ (R/Q) ∩ [0, 1].
It can be showed that the function f (x) is not Riemann integrable.
(3) Another important concept is Lebesgue integrals, which can be think of
the Riemann integral in terms of the partition of y-axis. A Lebesgue integrable
function is Riemann integrable, but the converse is not true. For example, the
Lebesgue integral of f (x) defined in (2) equals 0.
6.1.2. Properties of the Riemann integrals. In this subsection, we collect

important properties that will help us to evaluate definite integrals.
Proposition 6.1.4. If f is integrable over [a, b], then

Z a Z a Z b
(6.1.3) f (x) dx = 0, f (x) dx = − f (x) dx.
a b a
6.2. THE FUNDAMENTAL THEOREM OF CALCULUS 51
Figure 6.2. Riemann integrals and Lebesgue integrals
Proposition 6.1.5. Assume that f and g are integrable over [a, b].
Z b Z b
(6.1.4) kf (x) dx = k f (x) dx, k is constant,
a a
Z b Z b Z b
(6.1.5) [f (x) + g(x)] dx = f (x) dx + g(x) dx,
a a a
Z b Z c Z b
(6.1.6) f (x) dx = f (x) dx + f (x) dx,
a a c
where c is any interior point of [a, b]. Moreover
(i) If f (x) ≥ 0 on [a, b], then
Z b
(6.1.7) f (x) dx ≥ 0.
a
(ii) If f (x) ≤ g(x) on [a, b], then
Z b Z b
(6.1.8) f (x) dx ≤ g(x) dx.
a a
(iii) If m ≤ f (x) ≤ M on [a, b], then
Z b
(6.1.9) m(b − a) ≤ f (x) dx ≤ M (b − a).
a
6.2. The fundamental theorem of calculus

Let f (x) be a continuous function on [a, b] and let
Z x
F (x) := f (u) du.
a
Then
"Z #
x+h x
F (x + h) − F (x)
Z
1
F 0 (x) = lim = lim f (u) du − f (u) du
h→0 h h→0 h a a
Z x+h
1
= lim f (u) du.
h→0 h x
52 6. INTEGRATIONS
6.2.1. The fundamental theorem of calculus I. The fundamental theorem

for calculus tells that F 0 (x) = f (x).
Theorem 6.2.1. (The fundamental theorem of calculus I) If f is continuous on

[a, b], then the function F defined by
Z x
(6.2.1) F (x) = f (u) du, a ≤ x ≤ b
a
is continuous on [a, b] and differentiable on (a, b) with
(6.2.2) F 0 (x) = f (x).
Proof. Since f is continuous on [a, b], by Theorem 5.1.1, m ≤ f (x) ≤ M on

[a, b] for some constants m and M . Consequently, by (6.1.9),
1 x+h
Z
m≤ f (u) du ≤ M.
h x
Define
1 x+h
Z
I := f (u) du, h ∈ [0, b − a].
h x
Then m ≤ I ≤ M . By Theorem 3.2.7, we have
1 x+h
Z
f (ch ) = I = f (u) du
h h
for some ch ∈ [x, x + h]. Hence
1 x+h
Z
f (x) = f lim ch = lim f (ch ) = lim f (u) du.
h→0 h→0 h→0 h x
Thus F 0 (x) = limh→0 I = f (x).
Proposition 6.2.2. If g(x) and h(x) are differentiable functions and f (u) is con-
tinuous for u between g(x) and h(x), then
Z h(x)
d
(6.2.3) f (u) du = f [h(x)]h0 (x) − f [g(x)]g 0 (x).
dx g(x)
Proof. By Propositions 6.1.4 and 6.1.5, it suffices to prove

Z h(x)
d
f (u) du = f [h(x)]h0 (x).
dx
Introduce a new variable
v := h(x).
Then Z h(x) Z v
d d dv
f (u) du = f (u) du · = h0 (x)f [h(x)]
dx 0 dv 0 dx
by Theorem 6.2.1.
6.2. THE FUNDAMENTAL THEOREM OF CALCULUS 53
6.2.2. Antiderivatives and indefinite integrals. Let f be a continuous

function on an interval I and choose two fixed numbers a, b ∈ I. Consider
Z x Z x
F (x) := f (u) du, G(x) := f (u) du.
a b
0 0
Then F (x) = f (x) = G (x) by Theorem 6.2.1. According to Corollary 5.4, we have
F (x) = G(x) + C
for some constant C; in particular,

Z x Z x Z b
C = F (x) − G(x) = f (u) du − f (u) du = f (u) du.
a b a
The above discussion tells us that any antiderivative can be obtained from a
special one by adding a constant. We denote by
Z
f (x) dx
any antiderivative of f , called an indefinite integral. Note that

Z Z x
f (x) dx = C + f (u) du
a
for some constant C.

For example
Z
1 4
x3 dx = x + C.
4
A collection of indefinite integrals:

xa+1
Z
(6.2.4) xa dx = + C, a 6= −1,
a+1
Z
1
(6.2.5) dx = ln |x| + C,
x
ax
Z Z
x x
(6.2.6) e dx = e + C, ax dx = + C,
ln a
Z Z
(6.2.7) cos x dx = sin x + C, sin x dx = − cos x + C,
Z Z
(6.2.8) sec2 x dx = tan x + C, csc2 x dx = − cot x + C,
Z Z
(6.2.9) sec x tan x dx = tan x + C, csc x cot x dx = − csc x + C,
Z Z
(6.2.10) tan x dx = ln | sec x| + C, cot x dx = − ln | csc x| + C,
Z Z
1 1
(6.2.11) dx = tan−1 x + C, √ dx = sin−1 x + C.
1 + x2 1 − x2
54 6. INTEGRATIONS
Figure 6.3. Area
6.2.3. The fundamental theorem of claculus II. Let f be a continuous

on [a, b] and set
Z x
G(x) = f (u) du, x ∈ [a, b].
a
Then G0 (x) = f (x) and G(x) is an antiderivative of f (x). If F (x) is any antideriv-
ative of f (x), then
G(x) = F (x) + C, x ∈ [a, b],
by Theorem 6.2.1. In particular,
Z b
f (u) du = G(b) = F (b) + C = F (b) + [G(a) − F (a)] = F (b) − F (a).
a
Theorem 6.2.3. (The fundamental theorem of calculus II) If f is continuous on

[a, b], then
Z b
(6.2.12) f (x) dx = F (b) − F (a),
a
where F (x) is an antiderivative of f (x).
6.3. Applications of integration

If f is a nonnegative, continuous function on [a, b], then
Z b
(6.3.1) A := f (x) dx
a
represents the area of the region bounded by the graph of f (x) between a and b,
the vertical lines x = a and x = b, and the x-axis between a and b. (see Figure 6.3)
6.3.1. Areas. From Figure 6.4, we have

6.3. APPLICATIONS OF INTEGRATION 55
Figure 6.4. Area of curves
If f and g are continuous on [a, b], with f (x) ≥ g(x) for all x ∈ [a, b], then the
area of the region between the curves y = f (x) and y = g(x) from a to b is equal to
Z b
(6.3.2) Area = [f (x) − g(x)] dx.
a
Suppose that a region is bounded by x = f (y) and x = g(y), with g(y) ≤ f (y)
for x ≤ y ≤ d; that is, f (y) is to the right of g(y) for all y ∈ [c, d]. Then the area
of the shaded region is given by
Z d
(6.3.3) Area = [f (y) − g(y)] dy.
c
6.3.2. Average values. If f (x) is a continuous function on [a, b]. The aver-
age value of f on [a, b] is
Z b
1
(6.3.4) favg := f (x) dx.
b−a a
Theorem 6.3.1. (Mean-value theorem for definite integrals) If f (x) is a continu-

ous function on [a, b], then
(6.3.5) f (c) = favg
for some number c ∈ (a, b).
Proof. Since f is continuous on [a, b], by Theorem 5.1.1, m ≤ f (x) ≤ M on

[a, b] for some constants m and M . Consequently, by (6.1.9),
Z b
1
m≤ f (x) dx ≤ M.
b−a a
By theorem 3.2.7, the number c ∈ (a, b) is obtained.
56 6. INTEGRATIONS
Figure 6.5. The length of a curve
6.3.3. Rectification of curves. Let y = f (x), a ≤ x ≤ b, be a differentiable

function whose derivative is continuous on [a, b]. (see Figure 6.5) We partition the
interval [a, b] into subintervals by using the partition P = [x0 , x1 , x2 , · · · , xn ], where
a = x0 < x1 < x2 < · · · < xn = b, and approximate the curve by a polygon that
consists of the straight-line segments connecting neighboring points on the curve.
Set
∆xk := xk − xk−1 , ∆yk := yk − yk−1 .
Then the length of the line segment is given by
p
(∆xk )2 + (∆yk )2 .
Using the partition P, we fins that the length of the polygon is
n p
X
LP = (∆xk )2 + (∆yk )2 .
i=1
By Theorem 5.1.5,
f (xk ) − f (xk−1 ) ∆yk
f 0 (ck ) = =
xk − xk−1 ∆xk
for some number ck ∈ (xk−1 , xk ). Consequently,
n p
X
LP = 1 + [f 0 (ck )]2 ∆xk
i=1
and the length of the curve y = f (x) is

Z b p
L := lim LP = 1 + [f 0 (x)]2 dx.
||P||→0 a
If f (x) is differentiable on (a, b) and f 0 (x) is continuous on [a, b], then the length
of the curve y = f (x) from a to b is given by
Z bp
(6.3.6) L= 1 + [f 0 (x)]2 dx.
a
6.3. APPLICATIONS OF INTEGRATION 57
Since
dy
= f 0 (x),
dx
it follows from (6.3.6) that
s 2
Z b Z bp
dy
(6.3.7) L= 1+ dx = (dx)2 + (dy)2 .
a dx a
p
We call the expression (dx)2 + (dy)2 the arc length differential and denote it
by ds.
6.3.4. Integral inequalities. For any real numbers a1 , · · · , an and b1 , · · · , bn ,

a basic inequality is
n
!2 n
! n !
X X X
2 2
(6.3.8) ai bi ≤ ai bi
i=1 i=1 i=1
called the Cauchy-Schwarz inequality.
Exercise 6.3.2. Verify (6.3.8) for n = 2.
Another basic inequality is the arithmetic-geometric inequality

r
1 √ a+b a2 + b2
(6.3.9) 1 1 ≤ ab ≤
2
≤
2
, a, b > 0.
a + b
Exercise 6.3.3. Use (6.3.9) to prove

1 2
(6.3.10) ab ≤ |ab| ≤ a2 + b
4
for any a, b ≥ 0 and > 0.
An integral version of (6.3.8) is the following
Theorem 6.3.4. (Hölder’s inequality) Suppose that p and q are two real numbers,
1 ≤ p, q ≤ ∞ with p1 + 1q = 1, and f and g are two Riemann integrable functions
on [a, b]. Then the functions |f (x)|p and |g(x)|q are also Riemann integrable and
Z !1/p Z !1/q
b Z b b
p q
(6.3.11) f (x)g(x) dx ≤ |f (x)| dx |g(x)| dx .

a a a
An integral version of (6.3.9) is the following

58 6. INTEGRATIONS
Figure 6.6. Young’s inequality
Theorem 6.3.5. (Minkowski inequality) Suppose that p ≥ 1 is a real number

and f (x), g(x) are Riemann integrable functions on [a, b]. Then |f (x)|p , |g(x)|p and
|f + g|p are also Riemann integrable and
! p1 ! p1 ! p1
Z b Z b Z b
(6.3.12) |f (x) + g(x)|p dx ≤ |f (x)|p dx + |g(x)|p dx .
a a a
The third basic inequality is Young’s inequality

ap bq 1 1
(6.3.13) ab ≤ + , a, b ≥ 0, + = 1, p, q > 1.
p q p q
Proof. We may assume that b ≤ ap−1 . Consider the function f (x) := xp−1
1
with x ∈ [0, a]. Then the inverse function g(x) of f (x) is equal to g(x) = x p−1 . By
Figure 6.6, we see that the area of the red region is given by
Z a Z a
ap
f (x) dx = xp−1 dx = ,
0 0 p
while the area of the yellow region is given by
Z b Z b Z b
1 p − 1 p−1
p bq
f −1 (y) dy = g(x) dx = x p−1 dx = b = .
0 0 0 p q
Since we assume that b ≤ ap−1 , from Figure 6.6 we have
ap bq
ab ≤ area of the red region + area of the yellow region = + .
p q
Similarly, we can deal with the case that b ≥ ap−1 .
Exercise 6.3.6. Verify (6.3.13) when b ≥ ap−1 .

CHAPTER 7
Integration techniques
In this chapter we introduce and discuss several techniques on

computing (indefinite, definite, improper) integrals. Most meth-
ods are presented in the textbook, however, other approaches will
be illustrated by considering practical examples which mainly
come from mathematics research.
An elementary rule or axiom is that an indefinite integral
does not depend on the variables. For example, the following
two indefinite integrals
Z Z
u du, y dy
are the same. The above rule is the original stage of the following
several methods.
7.1. Basic rules

If we consider the function y = x2 , then
dy
= 2x, dy = 2x dx,
dx
and hence Z Z Z
2x3 dx = x2 2x dx = y dy.
If we consider another function u = x2 , then the same argument shows that

Z Z Z
u du = x 2x dx = 2x3 dx.
2
Hence Z Z Z
3
u du = 2x dx = y dy.
7.1.1. Substitution rule for indefinite integrals. In general, we can prove
Proposition 7.1.1. If u = g(x) for some C 1 -function g, then

Z Z
(7.1.1) f [g(x)]g 0 (x) dx = f (u) du.
We say a function g(x) is C 1 , if the derivative of g(x) exists and is continuous.

59
60 7. INTEGRATION TECHNIQUES
Proof. Let F (x) be an anti-derivative of f (x), i.e., F 0 (x) = f (x). Then

d
F [g(x)] = F 0 [g(x)]g 0 (x) = f [g(x)]g 0 (x)
dx
by the chain rule. Thus F [g(x)] is an anti-derivative of f [g(x)]g 0 (x) and hence
Z Z
f [g(x)]g 0 (x) dx = F [g(x)] + C = F (u) + C = f (u)du
showing the formula (7.1.1).
Example 7.1.2. Evaluate

√
Z p Z p Z
4x x2 + 1 dx, 4(x3 + x) x2 + 1 dx, x 2x − 1 dx.
Proof. For the first indefinite integral, we set u = x2 + 1 and hence

Z p Z
4 4
2 x2 + 1 d(x2 + 1) = 2u1/2 du = u3/2 + C = (x2 + 1)3/2 + C.
3 3
Similarly,
Z p Z p Z
4(x3 + x) x2 + 1 dx = 2(x2 + 1) x2 + 1 d(x2 + 1) = 2u3/2 du
4 5/2 4 2
= u +C = (x + 1)5/2 + C.
5 5
To evaluate the third one, we let u := 2x − 1. Then
√ u + 1 √ du
Z Z Z
1 3/2
x 2x − 1 dx = u = u + u1/2 du
2 2 4
1 5/2 1 3/2 1 1
= u + u +C = (2x − 1)5/2 + (2x − 1)3/2 + C.
10 6 10 6

7.1.2. Substitution rule for definite integrals. Proposition 7.1.1 can be
transformed to definite integrals.
Proposition 7.1.3. If u = g(x) for some C 1 -function g, then

Z b Z g(b)
(7.1.2) f [g(x)]g 0 (x) dx = f (u) du.
a g(a)
Proof. The proof is essentially similar as that of Proposition 7.1.1 Let F (x)
be an anti-derivative of f (x). Then F [g(x)] is an anti-derivative of f [g(x)]g 0 (x) and
hence, by the fundamental theorem of calculus, we have
Z b Z g(b)
f [g(x)]g 0 (x) dx = F [g(b)] − F [g(a)] = f (u) du,
a g(a)
where we use the definition of F .

7.1. BASIC RULES 61
Example 7.1.4. Evaluate limk→∞ Ik , where

Z π/2
Ik := sink x cos x dx, k ∈ N.
0
Proof. Letting u = sin x we have

Z 1 1
1 1
Ik = uk du = uk+1 = .

0 k+1 0 k+1
Hence limk→∞ Ik = 0.
7.1.3. Trigonomtric functions method. Recall trigonometric functions for-

mulas
2 tan x2 1 − tan2 x
2 2 tan x2
sin x = , cos x = , tan x = .
1 + tan2 x2 1 + tan2 x
2 1 − tan2 x2
Since
d d sin x 1 sin2 x + cos2 x
tan x = = = = 1 + tan2 x,
dx dx cos x cos2 x cos2 x
it follows that
x x
2d tan = 1 + tan2 dx
2 2
and hence
dx dx 1 + tan2 x2
= x = x dx
1 + cos x 1−tan2
1 + 1+tan2 2
x
1 + + (1 − ) tan2 2
2

Z π
1 π
dx = √ , ∈ (−1, 1).
0 1 + cos x 1 − 2
Proof. Letting u = tan x2 we have

r !
1−
Z Z Z
1 2 du 2 1
dx = 1+ =√ 2 d u .
1 + cos x 1− 1− + u
2
1 − 2 1+
q
1−
1+ 1+ u
Evaluating at π and 0 implies

!
π
r
1−
Z π
dx 2 x π
=√ tan−1 tan =√ ,

0 1 + cos x 1 − 2 1+ 2 0 1 − 2
because tan−1 (∞) = π/2.

7.2. Integration by parts

Let u = u(x) and v = v(x) be differentiable functions. Then
(uv)0 = u0 v + uv 0
or
(7.2.1) uv 0 = (uv)0 − u0 v.
Integrating both sides with respect to x, we obtain
Proposition 7.2.1. If u(x) and v(x) are C 1 -functions, then

Z Z
(7.2.2) u(x)v (x) dx = u(x)v(x) − u0 (x)v(x) dx
0
or, in short form,

Z Z
(7.2.3) u dv = uv − v du.
7.2.1. Recursion method. This method reduces the “power” of functions

by 1 each time.

Z Z
Ik = sink x dx, Jk = cosk x dx.
Proof. By Proposition 7.2.1 for any k ≥ 2, we have

Z Z
Ik = − sink−1 x d cos x = − sink−1 x cos x + (k − 1) cos2 x sink−2 x dx
Z
k−1
= − cos x sin x + (k − 1) sink−2 x(1 − sin2 x) dx
= − cos x sink−1 x + (k − 1)Ik−2 − (k − 1)Ik ;

thus
k−1 1
Ik = Ik−2 − cos x sink−1 x.
k k


Z π/2
Ik = sink x dx.
0
Proof. From Example 7.2.2 we have

k−1
Ik = Ik−2 , k ≥ 2.
k
7.2. INTEGRATION BY PARTS 63
On the other hand,

Z π/2 Z π/2 π/2
π
I0 = dx = , I1 = sin x dx = − cos x = 1.

0 2 0 0
When k = 2m is even, we have

2m − 1 2m − 1 2m − 3 (2m − 1)!! π
I2m = I2m−2 = I2m−4 = · · · = .
2m 2m 2m − 2 (2m)!! 2
When k = 2m + 1 is odd, we have
2m 2m 2m − 2 (2m)!!
I2m+1 = I2m−1 = I2m−3 = · · · = .
2m + 1 2m + 1 2m − 1 (2m + 1)!!

Example 7.2.4. Evaluate, for k ∈ N,

Z Z
Ik = x e dx, Ik (a) = xk eax dx
k x
(a 6= 0).
d ax
Proof. The basic idea to deal with Ik and Ik (a) is an observation that dx e =
aeax . In other words,
deax = aeax dx.
Since Ik = Ik (1), we suffice to consider the second integral Ik (a). Letting u = xk /a
and v = eax in (7.2.3), we have, for k ≥ 1,
Z Z Z
1 1 k ax
Ik (a) = xk eax dx = xk aeax dx = x de
a a
Z Z
1 k ax 1 1 k ax 1
= x e − eax dxk = x e − kxk−1 eax dx
a a a a
1 k ax k
= x e − Ik−1 (a).
a a
Note that Z
1
I0 (a) = eax dx = eax + C.
a
for example,
1 ax 1 x ax 1
I1 (a) = xe − I0 (a) = e − 2 eax + C,
a a a a
x2

1 2 ax 2 2x 2
I2 (a) = x e − I1 (a) = − 2 + 3 eax + C.
a a a a a
It is a good excise for you to find the general formula for Ik (a).
Example 7.2.5. Evaluate, for k ∈ N,

Z
Ik = (ln x)k dx.
Proof. Taking u = (ln x)k and v = x in (7.2.3) implies

Z
Ik = x(ln x) − k(ln x)k−1 dx = x(ln x)k − kIk−1 .
k
Since Z Z
I1 = ln x dx = x ln x − dx = x ln x − x + C,
we can calculate all Ik ’s by the above recursion formula and the initial value.
7.2.2. Gamma function and Beta function. The gamma function Γ(x)
is defined by
Z ∞
(7.2.4) Γ(x) := tx−1 e−t dt, x > 0.
0
This is an improper integral which is well-defined (see later). By (7.2.3), we have
1 ∞ −t x tx
Z Z Z
x−1 −t 1
t e dt = e dt = t + tx e−t dt.
x 0 xe x
Since
tx tx
lim t
= 0 = lim t ,
t→∞ xe t→0 xe
it follows that
(7.2.5) Γ(1 + x) = xΓ(x).
Using the recursion formula (7.2.5), we have
(7.2.6) Γ(n + 1) = n!, n ∈ N.
There are some fundamental properties of Γ(x):
(1) Euler’s reflection formula: for any 0 < x < 1, we have
π
(7.2.7) Γ(x)Γ(1 − x) = .
sin(πx)
In particular,
√
r
1 π
(7.2.8) Γ = = π.
2 sin(π/2)
(2) Duplication formula: for any x > 0, we have
√

1
(7.2.9) Γ(x)Γ x + = 21−2x πΓ(2x).
2
Another important and useful integral is the Beta function defined by
Z 1
(7.2.10) B(x, y) := tx−1 (1 − t)y−1 dt, x, y > 0.
0
This is also an improper integral. The beta function was studied by Euler and
Legendre and was given its name by Jscques Binet.
(3) B(x, y) = B(y, x).
(4) For any x, y > 0 we have
Γ(x)Γ(y)
(7.2.11) B(x, y) = .
Γ(x + y)
7.3. RATIONAL FUNCTIONS AND PARTIAL FRACTIONS 65
Example 7.2.6. (Example 7.2.3 revisited) We now use gamma/beta functions to

give another way to evaluate Ik defined in Example 7.2.3. Writing t = sin2 u in
(7.2.10) with u ∈ [0, π/2], we arrive at
Z π/2
B(x, y) = (sin u)2x−2 (cos u)2y−2 2 sin u · cos u du
0
Z π/2
= 2 (sin u)2x−1 (cos u)2y−1 du;
0
in particular,
Z π/2
k+1 1
B , =2 (sin u)k du = 2Ik .
2 2 0
Using (7.2.11), we have
√
1 Γ( k+1 1
2 )Γ( 2 ) π Γ( k+1
2 )
Ik = k+2
= .
2 Γ( 2 ) 2 Γ( k2 + 1)
If k = 2m is even, then
√ √
π Γ(m + 12 )

π 1 1
I2m = = Γ m+
2 Γ(m + 1) 2 m! 2
by (7.2.6); from (7.2.5), we see that

1 1 1 1
Γ m+ = Γ 1+m−1+ = m−1+ Γ m−1+
2 2 2 2

2m − 1 1 2m − 1 2m − 3 1
= Γ m−1+ = Γ m−2+
2 2 2 2 2
(2m − 1)!! √

(2m − 1)!! 1
= ··· = Γ = π.
2m 2 2m
Hence
π (2m − 1)!! π (2m − 1)!!
I2m = = .
2 2m m! 2 (2m)!!
The odd case is left to readers.
7.3. Rational functions and partial fractions

A rational function f is the quotient of two polynomials. That is
P (x)
(7.3.1) f (x) = ,
Q(x)
where P (x) and Q(x) are polynomials. A basic idea to treat with rational functions
is to write f (x) as a sum of a polynomial and simpler rational functions. Such a sum
is called a partial-fraction decomposition. These simpler rational functions are
of the form
A Bx + C
(7.3.2) n
, ,
(ax + b) (ax + bx + c)n
2
where A, B, C, a, b, and c are constants and n is a positive integer.

If the degree of P (x) in (7.3.1) is greater than or equal to the degree of Q(x),
then the first step in the partial-fraction decomposition is to use long division to
write f (x) as a sum of a polynomial and a rational function, where the rational
function is such that the degree of the polynomial in the numerator is less than the
degree of the polynomial in the denominator. (Such rational functions are called
proper.)
7.3.1. Partial-fraction decomposition I. We discuss the cases of distinct
and repeated linear factors.
If the linear factor ax + b is contained n times in the factorization of the

denominator of a proper rational function, then the partial-fraction decomposition
contains terms of the form
A1 A2 An
+ + ··· + ,
ax + b (ax + b)2 (ax + b)n
where A1 , · · · , An are constants.
Case 1a: Q(x) is a product of m distinct linear factors. Q(x) is thus of the
form
Q(x) = a(x − x1 ) · · · (x − xm )
where x1 , · · · , xm are the m distinct roots of Q(x). The rational function can then
be written as
P (x) 1 A1 Am
= + ··· + .
Q(x) a x − x1 x − xm
Case 1b: Q(x) is a product of repeated linear factors. Q(x) is thus of the form
Q(x) = a(x − x1 )n1 · · · (x − xm )nmm
where x1 , · · · , xm are the m distinct roots of Q(x), and n1 , · · · , nm are positive
integers. Then

P (x) 1 A1,1 A1,n1 Am,1 Am,nm
= + ··· + + ··· + + ··· + .
Q(x) a x − x1 (x − x1 )n1 x − xm (x − x1 )nm
For example,
Z
−1
Z
1 1
dx = + dx = ln |x − 1| − ln |x| + C,
x(x − 1) x x−1
and
Z
−1
Z
x 1 1
dx = + dx = ln |x + 1| + + C.
(x + 1)2 x + 1 (x + 1)2 x+1
7.3.2. Irreducible quadratic factors. We discuss the cases of distinct and
repeated irreducible quadratic factors.
If the irreducible quadratic factor ax2 + bx + c is contained n times in the

factorization denominator of a proper rational function, then the partial-fraction
decomposition contains temrs of the form
B1 x + C1 B2 x + C2 Bn x + C n
+ + ··· + ,
ax2 + bx + c (ax2 + bx + c)2 (ax2 + bx + c)n
7.3. RATIONAL FUNCTIONS AND PARTIAL FRACTIONS 67
where B1 , C1 , · · · , Bn , Cn are constants.

Case 1a: Q(x) is a product of m distinct irreducible quadratic factors. Q(x)
is thus of the form
Q(x) = (a1 x2 + b1 x + c1 ) · · · (am x2 + bm x + cm )
where a1 , b1 , c1 , · · · , am , bm , cm are constants. Then
P (x) B1 x + C 1 Bm x + C m
= 2
+ ··· + .
Q(x) a1 x + b1 x + c1 am x2 + bm x + cm
Case 2a: Q(x) is a product of m repeated irreducible quadratic factors. Q(x)
is thus of the form
Q(x) = (a1 x2 + b1 x + c1 )n1 · · · (am x2 + bm x + cm )nm ,
where a1 , b1 , c1 , · · · , am , bm , cm are constants, and n1 , · · · , nm are positive integers.
Then
n1 nm
P (x) X B1,i x + C1,i X Bm,i x + Cm,i
= 2 i
+ · · · + .
Q(x) i=1
(a1 x + b1 x + c1 ) i=1
(am x2 + bm x + cm )nm
For example,
2x3 − x2 + 2x − 2
Z
−1
Z
2x
dx = + dx = ln(x2 + 2) − tan−1 x + C,
(x2 + 2)(x2 + 1) x2 + 2 x2 + 1
and
x2 + x + 1
Z Z
1 x 1
dx = + dx = tan−1 x − + C.
(x2 + 1)2 x2 + 1 (x2 + 1)2 2(x2 + 1)

1
x4 (1 − x)4
Z
22
2
dx = − π.
0 1+x 7
Proof. Since
x4 (1 − x)4 4
2
= x6 − 4x5 + 5x4 − 4x2 + 4 − ,
1+x 1 + x2
it follows that
1
x4 (1 − x)4
Z
22
dx = − π.
0 1 + x2 7
2
Since 1 ≤ 1 + x ≤ 2 for any x ∈ [0, 1], we have
Z 1 Z 1 4 Z 1
1 4 x (1 − x)4
x (1 − x)4 dx ≤ 2
dx ≤ x4 (1 − x)4 dx.
0 2 0 1 + x 0
Recall the Beta function (7.2.10). Hence

Z 1
Γ(5)Γ(5) 4!4! 1
x4 (1 − x)4 dx = B(5, 5) = = =
0 Γ(10) 9! 630
Figure 7.1. An improper integral
by (7.2.11) and (7.2.6). Consequently

1 22 1
< −π ≤
1260 7 630
which gives 3.140 ≤ π ≤ 3.142.
7.4. Improper integrals

Look at the gamma function defined in (7.2.4). At fist glance, the integration
1
interval is unbounded and the function t 2 −1 e−t is infinity as t → 0+.


 0, x > 1,
lim tx−1 e−t = 1, x = 1,
t→0+
∞, x < 1.

We call an integral is improper if

(1) one or both limits of integration are infinite (that is, the integration in-
terval is unbounded), or
(2) the integrand becomes infinite at one or more points of the interval of
integration.
7.4.1. Unbounded intervals. By cutting off the interval, we can define
Definition 7.4.2. If f (x) is a continuous function on [a, ∞) or (−∞, a], then we

define
Z ∞ Z z
(7.4.1) f (x) dx := lim f (x) dx,
a z→∞ a
and
Z a Z a
(7.4.2) f (x) dx := lim f (x) dx.
−∞ z→−∞ z
7.4. IMPROPER INTEGRALS 69
From the definition above, we know that for each z, the integrals
Z z Z a
f (x) dx, f (x) dx
a z
are finite, however, after taking the limits, the improper integrals may be infinity.
Example 7.4.3. A typical example is

Z ∞
1
(7.4.3) p
dx, p > 0.
1 x
For example,
Z ∞ Z z
1 1 1
dx = lim dx = lim 1 − = 1,
1 x2 z→∞ 1 xp z→∞ z
Z ∞ Z z
1 1 √
√ dx = lim √ dx = lim 2 z − 2 = ∞.
1 x z→∞ 1 x z→∞
In general, since Z z 1 1−p

1 1−p (z − 1), p 6= 1,
p
dx =
1 x ln z, p = 1,
we arrive at
Z ∞ 1
1 p−1 , p > 1,
(7.4.4) dx =
1 x p ∞, 0 < p ≤ 1.
As in the case of sequences, we can define
Definition 7.4.4. Let f (x) be continuous on the interval [a, ∞). If

Z z
lim f (x) dx
z→∞ a
exists and has a finite value, we say that the improper integral
Z ∞
f (x) dx
a
converges and define
Z ∞ Z z
f (x) dx := lim f (x) dx.
0 z→∞ a
Otherwise we say that the improper integral diverge.
From Example 7.4.3 and Definition 7.4.4, we see that the integral
Z ∞
1
p
dx
1 x
is convergent if p > while is divergent if 0 < p ≤ 1.

Example 7.4.5. Consider the integral

Z ∞
1
dx.
0 1 + x2
According to (6.2.11), we have
Z z z
1 −1
2
dx = tan x = tan−1 z
0 1+x 0
and then Z ∞
1 π
2
dx = lim tan−1 z = .
0 1 + x z→∞ 2
Definition 7.4.4 can be extend to the integrals

Z a
f (x) dx.
−∞
For example
Z 0
1 π
dx = lim (− tan−1 z) = .
−∞ 1 + x2 z→−∞ 2
If f (x) is continuous on (−∞, ∞), we define

Z ∞ Z a Z ∞
(7.4.5) f (x) dx := f (x) dx + f (x) dx,
−∞ −∞ a
where a is a real number. If both improper integrals on the right-hand side of
(7.4.5) are convergent, then the value of the impropert integral on the left-hand
side of (7.4.5) is the sum of the two limiting values on the right-hand side.
I give some remarks on the above definition.
Remark 7.4.6. (1) More precisely, if for any given a ∈ R, the integrals
Z a Z ∞
f (x) dx, f (x) dx
−∞ a
R∞
are convergent, then we define the integral −∞ f (x) dx via (7.4.5). In this case, we
can show that the definition (7.4.5) does not depend on the particular choice of a.
Indeed, for any given two real numbers a and b, say a < b, we have
Z ∞ Z z "Z #
z Z b
f (x) dx = lim f (x) dx = lim f (x) dx + f (x) dx
a z→∞ a z→∞ b a
Z ∞ Z b
= f (x) dx + f (x) dx.
b a
Similarly, we can show
Z b Z b Z a
f (x) dx = f (x) dx + f (x) dx.
−∞ a −∞
Since all integrals are finite, it follows that

Z ∞ Z a Z ∞ Z b
f (x) dx + f (x) dx = f (x) dx + f (x) dx.
a −∞ b −∞
Thus the definition (7.4.5) does not depend on the choice
R ∞ of a.
(2) Another important way to define the integral −∞ f (x) dx is the Cauchy’s
principal-value integral
Z ∞ Z z
(7.4.6) p.v. f (x) dx := lim f (x) dx.
−∞ z→∞ −z
R∞
(3) If −∞ f (x) dx exists in the sense of (7.4.5), then
Z ∞ Z a Z z Z ∞
f (x) dx = lim f (x) dx + f (x) dx = p.v. f (x) dx.
−∞ z→∞ −z a −∞
However, the converse is not true. Thus, we can find a continuous function f (x)
defined on (−∞, ∞), which has finite integral in the sense of (7.4.6) but not of
(7.4.5). For example,
f (x) := sin x, x ∈ (−∞, ∞).
Then
Z ∞ Z z
p.v. sin x dx = lim sin x dx = lim 0 = 0,
−∞ z→∞ −∞ z→∞
Z ∞
sin x dx = ∞.
−∞
Let
f (x) := ex − x, x ∈ [0, ∞).
Since f 0 (x) = ex − 1 ≥ e0 − 1 = 0, it follows from Proposition 5.2.2 that f (x) is
increasing on [0, b] for each fixed b and then on [0, ∞). Hence
(7.4.7) ex − x = f (x) ≥ f (0) = e0 − 0 = 1 =⇒ ex ≥ 1 + x, x ∈ [0, ∞).

x2
ex ≥ 1 + x + , x ∈ [0, ∞).
2
x2
(Hint: Consider the function f (x) = ex − 1 − x − 2 and use (7.4.7).)
In general, we have, for any given positive integer n,

n
X xi x2 xn 1
(7.4.8) ex ≥ =1+x+ + ··· + ≥ xn , x ∈ [0, ∞).
i=0
i! 2! n! n!
Actually, we can show that (beyond the scope of this course)
n ∞
X xi X xi
(7.4.9) ex = lim =: , x ∈ (−∞, ∞).
n→∞
i=0
i! i=0
i!
We call (7.4.9) the Taylor series of ex .
Exercise 7.4.8. Verify (7.4.8) for n = 3, 4.
We now use the inequality (7.4.8) to show that the improper integral (7.2.4) is
convergent at least for x ≥ 1.
Example 7.4.9. Recall the gamma function

Z ∞
Γ(x) := tx−1 e−t dt.
0
We now show that Γ(x) is convergent whenever x ≥ 1.
Case 1: x = 1. Then
Z ∞ ∞
e−t dt = −e−t = 1.

Γ(1) =
0 0
Case 2: x > 1. Divide Γ(x) into two terms
Z 1 Z ∞
x−1 −t
Γ(x) = t e dt + tx−1 e−t dt;
0 1
since x − 1 > 0 and 0 ≤ t ≤ 1, we have tx−1 ≤ 1 and
Z 1 Z 1 1
1
tx−1 e−t dt ≤ e−t dt = −e−t = 1 − .

0 0 0 e
Choose an integer n such that x − 1 ≤ n; for t ≥ 1, we then get tx−1 ≤ tn . Using
(7.4.8) implies n
1 t
≤ et/2 =⇒ tx−1 ≤ tn ≤ 2n n!et/2
n! 2
and hence
Z ∞ Z ∞ Z ∞
x−1 −t t/2 −t 2n+1 n!
t e dt ≤ n
2 n!e e dt = 2 n! n
e−t/2 dt = √ .
1 1 1 e
Therefore Γ(x) is convergent for x ≥ 1.
7.4.2. Unbounded integrands. A typical example is the integral

Z 1
1
√ dx.
0 x
√
The integrand f (x) = 1/ x is unbounded as x → 0+. By cutting off the point 0,
we can study
Z 1
1
I(c) := √ dx, 0 < c < 1.
c x
Now the function f (x) is well-defined and continuous on [c, 1]; moreover
√ 1 √

Ic = 2 x = 2(1 − c)
c
and
√
lim Ic = lim (2 − 2 c) = 2.
c→0+ c→0
This example suggests us the following
Definition 7.4.10. If f is continuous on (a, b] and limx→a+ f (x) = ±∞, we define

Z b Z b
(7.4.10) f (x) dx := lim f (x) dx,
a c→a+ c
provided that this limit exists. If the limit exists, we say that the improper integral
on the left-hand side of (7.4.10) converges; if the limit does not exist, we say that
the improper integral diverges.
Similarly, if f is continuous on [a, b) and limx→b− f (x) = ±∞, we define
Z b Z c
(7.4.11) f (x) dx := lim f (x) dx,
a c→b− a
provided that this limit exists.
If f is continuous on (a, b) and limx→a+ f (x) = ±∞ and limx→b− f (x) = ±∞,
we define
Z b Z b−
(7.4.12) f (x) dx := lim f (x) dx,
a ,η→0 a+η
provided that this limit exits. Note that the limit in (7.4.12) is the limit of multi-
variable which will be treat with in later.
For example
Z 1 1
dx 1/3

2/3
= 3(x − 1) = 3,
0 (x − 1) 0
Z 1 1

ln x dx = (x ln x − x) = −1 − lim x ln x = −1.
0 x→0+
0
If f (x) is discontinuous at a point c ∈ (a, b), then we define

Z b Z b Z c−η
(7.4.13) f (x) dx := lim f (x) dx + lim f (x) dx.
a →0+ c+ η→0+ a
By this definition, we have

Z 1
1
dx = ∞.
−1 x3
We can also define Cauchy’s principal-value integral
Z b "Z #
b Z c−
(7.4.14) p.v. f (x) dx := lim f (x) dx + f (x) dx .
a →0+ c+ a
For example,
Z 1 Z 1 Z −
1 dx dx
p.v. dx = lim + = 0.
−1 x3 →0+ x3 −1 x3
7.4.3. A comparison result for improper integrals. The inequality (6.1.8)

can be generalized to improper integrals.
Proposition 7.4.11. Suppose that fR(x) and g(x) are continuous onR[a, ∞).
∞ ∞
(1) If f (x) ≤ g(x) on [a, ∞ and a g(x) dx is convergent, then a f (x) dx is
also convergent and
Z ∞ Z ∞
(7.4.15) f (x) dx ≤ g(x) dx.
a a
R∞ R∞
(2) If f (x) ≥ g(x) on [a, ∞) and a g(x) dx is divergent, then a f (x) dx is
also divergent.
Proof. (1) In this case, for any fixed number z > a, we have
Z z Z z
f (x) dx ≤ g(x) dx;
a a
letting z →R∞∞ implies (7.4.15).

(2) If a f (x)dx is convergent, by (1), we must have that the improper integral
R∞
a
g(x) dx is convergent, a contradiction.

Z ∞
2
I := e−x dx.
0
Note that Z 1 Z ∞
2 2
I= e−x dx + e−x dx.
0 1
−x2
For any 0 ≤ x ≤ 1, we have 0 ≤ e ≤ 1 and then
Z 1 Z 1
2
e−x dx ≤ dx = 1.
0 0
2
When x > 1, we have then −x ≤ −x so that
Z ∞ Z ∞ ∞
2 1
e−x dx ≤ e−x dx = −e−x = .

1 1 1 e
Set
t = x2 , x ≥ 0.
Then ∞ √
√ 1 ∞ −1/2 −t
Z Z
1 1 π
I= e−t d t = t e dt = Γ =
0 2 0 2 2 2
by (7.2.8).
7.5. Taylor approximation

The Taylor polynomial of degree n about x = 0 for the function f (x) is given
by
f 00 (0) 2 f (n) (0) n
(7.5.1) P (x) := f (0) + f 0 (0)x + x + ··· + x ,
2! n!
7.5. TAYLOR APPROXIMATION 75
0 00 (n)
Pn fi (0), f (0), · · · , f (0) exist.
provided all derivatives
For example, i=0 x /i! in (7.4.8) is the Taylor polynomial of degree n about
x = 0 for ex .
More general, the Taylor polynomial of degree n about x = a for the function
f (x) is given by
f 00 (a) f (n) (a)

(7.5.2) Pn (x) = f (a) + f 0 (a)(x − a) + (x − a)2 + · · · + (x − a)n .
2! n!
Exercise 7.5.1. Compute the Taylor polynomial of degree 3 about x = 0 for sin x
and cos x.
7.5.1. Taylor’s formula. Let f be a continuous function whose continuously

derivatives exist up to second order. By Theorem 6.2.3, we have
Z x
f (x) = f (a) + f 0 (t) dt.
a
Applying Proposition 7.2.1 to u = −(x − t) and v = f 0 (t), we get

Z x Z x x Z x
0 0 0
(x − t)f 00 (t) dt

f (t) dt = f (t)d[−(x − t)] = −(x − t)f (t) +
a a a a
Z x
= (x − a)f 0 (a) + (x − t)f 00 (t) dt.
a
That is
Z x
f (x) = f (a) + f 0 (a)(x − a) + (x − t)f 00 (t) dt.
a
In general
Theorem 7.5.2. (Taylor’s formula) Suppose that f : I → R where I is an interval,

a ∈ I, and f and its first n+1 derivatives are continuous at a ∈ I. Then, for x ∈ I,
n
X f (i) (a)
(7.5.3) f (x) = f (a) + (x − a)i + Rn+1 (x),
i=1
i!
where
Z x
1
(7.5.4) Rn+1 (x) = (x − t)n f (n+1) (t) dt.
n! a
By Theorem 6.3.1, we have

Z x
f (x) = f (a) + f 0 (t) dt = f (a) + (x − a)f 0 (c)
a
for some c between a and x.

Corollary 7.5.3. Suppose that f : I → R where I is an interval, a ∈ I, and f and

its first n + 1 derivatives are continuous at a ∈ I. Then, for x ∈ I, there exists a c
between a and x such that the error term Rn+1 (x) is of the form
f (n+1) (c)
(7.5.5) Rn+1 (x) = (x − a)n+1 .
(n + 1)!
Proof. Since f (n+1) (t) is continuous, it follows that m ≤ f (n+1) (t) ≤ M for
all t ∈ I. In particular,
m x M x
Z Z
(x − t)n dt ≤ Rn+1 (x) ≤ (x − t)n dt,
n! a n! a
m M
which gives us (n+1)! (x − a)n+1 ≤ Rn+1 (x) ≤ (n+1)! (x − a)n+1 . Thus m ≤
(n+1)!Rn+1 (x) n+1
(x−a)m+1 ≤ M. By Theorem 3.2.7, (n + 1)!Rn+1 (x)/(x − a) = f (n+1) (c)
for some c between a and x.
For example,
n
x
X xi
(7.5.6) e = + Rn+1 (x), x ∈ (−∞, ∞),
i=0
i!
n
X x2i+1
(7.5.7) sin x = (−1)i + Rn+1 (x), x ∈ (−∞, ∞),
i=0
(2i + 1)!
n
X x2i
(7.5.8) cos x = (−1)i + Rn+1 (x), x ∈ (−∞, ∞),
i=0
(2i)!
n
X xi
(7.5.9) ln(1 + x) = (−1)i+1 + Rn+1 (x), x ∈ (−1, 1],
i=1
i
n
1 X
(7.5.10) = xi + Rn (x), x ∈ (−1, 1).
1−x i=0
7.5.2. Taylor’s series. Indeed, we can show that (7.5.6)–(7.5.10) become
∞
X xi
(7.5.11) ex = , x ∈ (−∞, ∞),
i=0
i!
∞
X x2i+1
(7.5.12) sin x = (−1)i , x ∈ (−∞, ∞),
i=0
(2i + 1)!
∞
X x2i
(7.5.13) cos x = (−1)i , x ∈ (−∞, ∞),
i=0
(2i)!
∞
X xi
(7.5.14) ln(1 + x) = (−1)i+1 , x ∈ (−1, 1),
i=1
i
∞
1 X
(7.5.15) = xi , x ∈ (−1, 1).
1−x i=0
A function f (x) defined on an open interval I is called smooth if all its higher
order derivatives are continuous. For example ex , sin x, cos x are smooth functions
on (−∞, ∞). The set of all smooth functions defined on I is denoted by C ∞ (I).
We say f ∈ C k (I) if all higher order derivative up to kth order are continuous.
Observe that
(7.5.16) C k (I) ⊂ C ∞ (I) for all k.
A function f (x) defined on an open interval I is called analytic if for any a ∈ I,

there exists an open interval Ia such that a ∈ Ia ⊂ I and
∞
X
f (x) = cn (x − a)n , x ∈ Ia .
n=0
The set of all analytic functions defined on I is denoted by C ω (I).
Example 7.5.4. ex ∈ C ω ((−∞, ∞)). For any a ∈ (−∞, ∞), we have

∞
X (x − a)i
ex = ea · ex−a = ea
i=0
i!
by (7.5.11).
dn x
Since dxn e = ex , it follows that ex is a smooth function. In general, we have
Proposition 7.5.5. For any open interval I, we have C ω (I) ⊂ C ∞ (I) and C ω (I) 6=
C ∞ (I). If f ∈ C ω (I), then, for any a ∈ I,
∞
X f (n) (a)
(7.5.17) f (x) = (x − a)n
n=0
n!
for any x near a.
I only give a smooth function on (−∞, ∞) which is not analytic.
Example 7.5.6. Consider

e−1/x , x > 0,

(7.5.18) f (x) =
0, x ≤ 0.
Since
1 1
lim e−1/x = lim = lim y = 0,
x→0+ e1/x
x→0+ y→∞ e
where y = 1/x. This shows that f (x) is continuous on (−∞, ∞). To prove f ∈
C ∞ ((−∞, ∞)), we suffice to check that limx→0+ f (n) (x) = f (0) = 0 for all n ∈ N.
Calculate
1 −1/x
f 0 (x) = e ,
x2
1 −1/x

00 −2x −1/x 1 2
f (x) = e + 4 = − 3 e−1/x ,
x4 x x4 x
3 2

000 −4x 6x 1 2 −1/x 1 6 6
f (x) = + 6 + 6− 5 e = − 5 + 4 e−1/x .
x8 x x x x6 x x
By induction on n, we can show that

1
(n)
f (x) = P2n e−1/x
x
where P2n (t) denotes the polynomial (without constant term) of degree 2n in terms
of t. Hence
P2n (y)
lim f (n) (x) = lim ;
x→0+ y→∞ ey
1
by (7.4.8), we have ey ≥ (2n+1)! y 2n+1 and then
(2n + 1)!P2n (y)
lim f (n) (x) ≤ lim =0
x→0+ y→∞ y 2n+1
by (3.1.8). Thus f (x) is smooth.
We now show that f (x) is not analytic. Otherwise,
∞
X
f (x) = cn xn
n=0
around 0. By (7.5.17), we must have f (x) ≡ 0 near 0. But, for any > 0, by the
definition, we get f (x) > 0, a contradiction. Therefore, f (x) ∈ C ∞ ((−∞, ∞)) but
/ C ω ((−∞, ∞)).
f (x) ∈
For any real number α and any positive integer n, define

α α(α − 1) · · · (α − n + 1)
(7.5.19) = .
n n!
When α is a positive integer greater than n, the definition (7.5.19) coincides with
the usual binomial coefficients.
For all |x| < 1 and all α ∈ R, we have

∞
X α
(7.5.20) (1 + x)α = 1 + xn .
n
n=1
Other important Taylor’s series are

We have
∞
X (2n)!
(7.5.21) sin−1 x = x2n+1 , |x| ≤ 1,
n=0
4n (n!)2 (2n + 1)
π
(7.5.22) cos−1 x = − sin−1 x, |x| ≤ 1,
2
∞
X (−1)n 2n+1
(7.5.23) tan−1 x = x , |x| ≤ 1.
n=0
2n + 1
In particular, taking x = 1 in (7.5.23) yields

∞
π X (−1)n 1 1 1
(7.5.24) = = 1 − + − + ··· .
4 n=0
2n + 1 3 5 7
Exercise 7.5.7. Verify

ex 2 1
= 1 + x + x2 + x3 + x4 + · · · ,
cos x 3 2
∞
X n+1 n
(1 + x)ex = x .
n=0
n!
The hyperbolic functions are

ex − e−x ex + e−x
(7.5.25) sinh x = , cosh x ,
2 2
sinh x cosh x
(7.5.26) tanh x = , coth x = .
cosh x sinh x
Exercise 7.5.8. Prove

sinh(−x) = − sinh x, cosh(−x) = cosh x,
tanh(−x) = − tanh x, coth(−x) = − coth x,
1 = cosh x − sinh2 x,
2
p p
sinh−1 x = ln x + 1 + x2 , cosh−1 x = ln x + x2 − 1 (x ≥ 1),

1 1+x 1 x+1
tanh−1 x = ln −1
(|x| < 1), coth x = ln (|x| > 1),
2 1−x 2 x−1
d d
sinh x = cosh x, cosh x = sinh x,
dx dx
d d
tanh x = 1 − tanh2 x, coth x = 1 − coth2 x,
dx dx
d 1 d 1
sinh−1 x = √ , cosh1 x = √ ,
dx 1+x 2 dx 2
x −1
d 1 d 1
tanh−1 x = , coth−1 x =
dx 1 − x2 dx 1 − x2
Figure 7.2. Hyperbolic functions
Figure 7.3. Geometric interpretation
The Taylor’s series of hyperbolic functions are

∞
X 1
(7.5.27) sinh x = x2n+1 ,
n=0
(2n + 1)!
∞
X 1
(7.5.28) cosh x = x2n .
n=0
(2n)!
Exercise 7.5.9. Evaluate the first two terms of their Taylor polynomials:
1
tanh x = x − x3 + · · · ,
3
1 1 1
coth x − = x − x3 + · · · .
x 3 45

Calculus II: For Biology and Medicine

Uploaded by

Copyright:

Available Formats

Calculus II: For Biology and Medicine

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Calculus II: For Biology and Medicine

Uploaded by

Copyright:

Available Formats

Calculus II: For Biology and Medicine

Chapter 1. Numbers and functions 5

Numbers and functions

This chapter deals with the preliminaries used in the following

1.1. Real and complex numbers

Figure 1.1. Numbers

Figure 1.2. Real numbers

Figure 1.3. Absolute value

Definition 1.1.1. The absolute value of a real number a, denoted by |a|, is

Figure 1.4. Linear equation

where A, B, and C are some constants. When A = B = 0 in (1.1.2), we must have

Figure 1.5. Circle

Figure 1.6. Trigonometric functions

(ii) Two l1 and l2 are called perpendicular, denoted l1 ⊥ l2 , if their inter-

1.1.4. Trigonometry. Consider a point B = (x, y) on the unit circle S1 (see

From the definition, we have

(1.1.7) sin2 θ + cos2 θ = 1,

sin(−θ) = − sin θ, cos(−θ) = cos θ,

(iii) Angle sum and difference identities:

(iv) Double-angle formula:

1.1.5. Exponentials and logarithms. An exponential is an expression of

Figure 1.7. Complex numbers

If y = ax > 0 with a > 0 and a 6= 1, then we formally write x := loga y.

Definition 1.1.2. A complex number is a number of the form

Proposition 1.1.4. For any z, w ∈ C we have

1.2. Elementary functions

Definition 1.2.1. Two functions f and g are equal if and only if

By this definition, we see that two functions

Definition 1.2.2. A function f : A → B is called

By the definition, if f is even or odd, the domain A must be symmetric.

Definition 1.2.3. The composite function f ◦ g (also called the composition

1.2.1. Polynomial functions. A polynomial function is a function of the

Theorem 1.2.4. Any nonconstant polynomial function with complex-valued coeffi-

This is the fundamental theorem of algebra. For example, the polynomial

1.2.4. Exponential functions. The function f is an exponential function

1.2.5. Inverse functions. A function f (x) is called one-to-one if f (x1 ) =

Definition 1.2.5. Let f : A → B be a one-to-one function with range f (A). The

Note that if f : A → B has an inverse f −1 , then

Example 1.2.6. Consider the function

Figure 1.8. Graphs of the six basic trigonometric functions

is called the Fourier series of f . If f ∈ C 1 ([−π, π]), then a theorem in Fourier

Definition 1.2.8. A number is called algebraic if√it is the solution of a polynomial

Example 1.2.9. (π is transcendental) Swiss scientist Johann Heinrich Lambert (26

Figure 1.9. Lambert, Legendre, and Euler

Figure 1.10. Lindemann and Gauss

Remark 1.2.10. (Gauss-Legendre algorithm) The Gauss-Legendre algorithm is

We call a function y = f (x) algebraic if it is the solution of an equation of the

1.2.8. Graphing. The graph of

Sequences and difference equations

In his chapter, we discuss models for populations that reproduce

Example 2.1.1. (1) an = (−1)n , n = 0, 1, 2, · · · .

Definition 2.1.2. We say a sequence {an }n∈N≥0 is kth order recursive if

2.1.2. Limits. Looking at Example 2.1.1, we see that, as n goes to infinity,

Figure 2.1. Limits

|an − a| <  n > N, and |an − a0 | <  n > N 0.

Consequently, for n > N + N 0 , we have

|a − a0 | = |(an − a) + (an − a0 )| ≤ |an − a| + |an − a0 | < 2.

In order to calculus, we need the following fundamental results.

Theorem 2.1.7. If limn→∞ an and limn→∞ bn exist and c is a constant, then

|an − a| < n > N, and |an − a0 | < n > N 0.

|a − a0 | = |(an − a) + (an − a0 )| ≤ |an − a| + |an − a0 | < 2.

|f (x) − L| < whenever 0 < x − c < δ.

|f (x) − L| < whenever 0 < c − x < δ.

Figure 3.1. The -δ definition of limits

Proof. Given an > 0. By the definition, we can find a positive number δ

Proof. Given > 0. There exists a positive number δ such that

|(f ◦ g)(x) − (f ◦ g)(c)| <