Single Variable Calculus: A Summary

William Boyles
Contents

0 Background & Review 1

0.1 Algebra and Pre-Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.1.1 Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.1.2 Factoring Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . 2
0.1.3 Trig Functions & The Unit Circle . . . . . . . . . . . . . . . . . . . . 3
0.1.4 Trig Identities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
0.1.5 Exponentials & Logarithms . . . . . . . . . . . . . . . . . . . . . . . 5
0.1.6 Partial Fractions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Linear Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Repeated Linear Factors . . . . . . . . . . . . . . . . . . . . . . . . . 7
Quadratic Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Repeated Quadratic Factors . . . . . . . . . . . . . . . . . . . . . . . 9
Improper Fractions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1 Limits & Continuity 11

1.1 Limit Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 Limit Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.1 “Substitution Rule” . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Left & Right Hand Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 Sandwich Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.5 Infinite Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.1 End Behavior Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.2 Horizontal Asymptotes . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.6 Continuity Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.7 Discontinuity Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.8 Continuity Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.9 Intermediate Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2 Derivatives 21
2.1 Rates of Change & Tangent Lines . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Definition of the Derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2.1 Derivative at a Point . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2.2 Left & Right Hand Derivatives . . . . . . . . . . . . . . . . . . . . . 24

i
 2.3 Differentiability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.1 Differentiability Implies Local Linearity . . . . . . . . . . . . . . . . . 25
Numerical Approximation: Symmetric Difference Quotient . . . . . . 25
2.3.2 Differentiability Implies Continuity . . . . . . . . . . . . . . . . . . . 26
2.4 Intermediate Value Theorem for Derivatives . . . . . . . . . . . . . . . . . . 26
2.5 Derivative Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.5.1 Basic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.5.2 Power Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.3 Higher Order Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.4 Product Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5.5 Quotient Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5.6 Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
u Substitutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.5.7 Exponential Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5.8 Implicit Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Logarithmic Differentiation . . . . . . . . . . . . . . . . . . . . . . . 34
Combining Power and Exponential Rule . . . . . . . . . . . . . . . . 34
2.6 Derivatives of Trig Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.7 Derivatives of Inverse Trig Functions . . . . . . . . . . . . . . . . . . . . . . 36
2.7.1 arcsin, arctan, and arcsec . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.7.2 arccos, arccot, and arccsc . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.8 Physical Interpretations of the Derivative . . . . . . . . . . . . . . . . . . . . 38
2.8.1 Displacement, Velocity, and Acceleration . . . . . . . . . . . . . . . . 38

3 Applications of the Derivative 40

3.1 Extreme Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2 Mean Value Theorem for Derivatives . . . . . . . . . . . . . . . . . . . . . . 41
3.3 First and Second Derivative Tests . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3.1 First Derivative Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3.2 Second Derivative Test . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Modeling & Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.5 Related Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.6 Linearization & Newton’s Method . . . . . . . . . . . . . . . . . . . . . . . . 49
3.6.1 Linearization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.6.2 Newton’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4 Integrals 52
4.1 Estimating with Sums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Left Endpoint Approximation . . . . . . . . . . . . . . . . . . . . . . 53
Trapezoidal Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 Definite Integrals & Antiderivatives . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.1 Definition of a Definite Integral . . . . . . . . . . . . . . . . . . . . . 55
4.2.2 Basic Properties of Definite Integrals . . . . . . . . . . . . . . . . . . 56

ii
 Mean Value Theorem for Definite Integrals . . . . . . . . . . . . . . . 56
4.2.3 Fundamental Theorem of Calculus . . . . . . . . . . . . . . . . . . . 57
4.3 Indefinite Integrals & Ways to Solve Them . . . . . . . . . . . . . . . . . . . 59
4.3.1 Definition of an Indefinite Integral . . . . . . . . . . . . . . . . . . . . 59
4.3.2 Indefinite Integrals Properties & Strategies . . . . . . . . . . . . . . . 60
Power Rule & Integral of 1/x . . . . . . . . . . . . . . . . . . . . . . 60
Exponential Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.3.3 u Substitution (Reversing the Chain Rule) . . . . . . . . . . . . . . . 60
Trig Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.3.4 Integration by Parts (Reversing the Product Rule) . . . . . . . . . . 63
LIPET (How to choose u) . . . . . . . . . . . . . . . . . . . . . . . . 64
Inverse Trig Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Tabular Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

5 Applications of Integrals 69
5.1 Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.1.1 Position, Velocity, & Acceleration . . . . . . . . . . . . . . . . . . . . 69
5.1.2 Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.2 Lengths of Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.3 Areas in the Plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.3.1 Subregions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.3.2 Integrating with Respect to y . . . . . . . . . . . . . . . . . . . . . . 75
5.4 Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.4.1 Volumes from Cross Sections . . . . . . . . . . . . . . . . . . . . . . . 76
5.4.2 Solids of Revolution . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Washer Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Cylindrical Shells Method . . . . . . . . . . . . . . . . . . . . . . . . 79
Other Axes of Rotation . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Surface Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.5 Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.5.1 Separable Differential Equations . . . . . . . . . . . . . . . . . . . . . 82
Exponential Growth & Decay . . . . . . . . . . . . . . . . . . . . . . 84
Logistic Growth & Decay . . . . . . . . . . . . . . . . . . . . . . . . 85
5.5.2 Slope Fields & Euler’s Method . . . . . . . . . . . . . . . . . . . . . . 87
Slope Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Euler’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

6 Parametric, Vector, & Polar Functions 89

6.1 Parametric & Vector Functions . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.1.1 Vector Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.1.2 Slope & Concavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.1.3 Arc Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.2 Polar Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

iii
 6.2.1 Polar Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.2.2 Polar Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Area Enclosed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Area Between Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Arc Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

7 Sequences, L’Hôpital’s Rule, & Improper Integrals 100

7.1 Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.1.1 Common Types of Sequences . . . . . . . . . . . . . . . . . . . . . . 101
Arithmetic Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Geometric Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.1.2 Limits of a Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
The Sandwich Theorem for Sequences . . . . . . . . . . . . . . . . . . 103
Absolute Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 104
7.2 L’Hôpital’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
7.2.1 Indeterminate Form 0/0 . . . . . . . . . . . . . . . . . . . . . . . . . 104
7.2.2 Indeterminate Forms ∞/∞, ∞ · 0, & ∞ − ∞ . . . . . . . . . . . . . 106
∞/∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
∞·0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
∞−∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.2.3 Indeterminate Forms 1∞ , 00 , & ∞0 . . . . . . . . . . . . . . . . . . . 107
1∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
00 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
∞0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.3 Relative Growth Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.3.1 Transitive Grow Rates . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.3.2 Growth Rate Hierarchy (nn FEPL) . . . . . . . . . . . . . . . . . . . 110
7.4 Improper Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7.4.1 Infinite Integration Limits . . . . . . . . . . . . . . . . . . . . . . . . 111
7.4.2 Infinite Discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.4.3 Convergence Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
P-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Direct Comparison Test . . . . . . . . . . . . . . . . . . . . . . . . . 114
Limit Comparison Test . . . . . . . . . . . . . . . . . . . . . . . . . . 115

8 Infinite Series 117

8.1 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
8.1.1 Geometric Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
8.1.2 Functions from Geometric Series . . . . . . . . . . . . . . . . . . . . . 119
Term-By-Term Differentiation . . . . . . . . . . . . . . . . . . . . . . 119
Term-By-Term Integration . . . . . . . . . . . . . . . . . . . . . . . . 120
8.2 Taylor Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
8.2.1 Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

iv
 8.2.2 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
8.2.3 Common Maclaurin Series . . . . . . . . . . . . . . . . . . . . . . . . 123
Euler’s Identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.3 Approximation Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.3.1 Alternating Series Estimation Theorem . . . . . . . . . . . . . . . . . 124
8.3.2 Taylor’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Remainder Estimation Theorem . . . . . . . . . . . . . . . . . . . . . 126
8.4 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
8.4.1 nth Term Test for Divergence
  . . . . . . . . . . . . . . . . . . . . . . 127
8.4.2 Geometric Series with r < 1 . . . . .
  . . . . . . . . . . . . . . . . . 128
8.4.3 P-Series Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
8.4.4 Direct Comparison Test . . . . . . . . . . . . . . . . . . . . . . . . . 129
8.4.5 Limit Comparison Test . . . . . . . . . . . . . . . . . . . . . . . . . . 129
8.4.6 Integral Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
8.4.7 Ratio Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8.4.8 nth Root Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
8.4.9 Alternating Series Test . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Riemann Rearrangement Theorem . . . . . . . . . . . . . . . . . . . 133
8.5 Radius of Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
8.5.1 Convergence Theorem for Power Series . . . . . . . . . . . . . . . . . 133
8.5.2 Convergence at Endpoints . . . . . . . . . . . . . . . . . . . . . . . . 135

9 Additional Materials 137

9.1 Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
9.1.1 2018 Free-Response Questions . . . . . . . . . . . . . . . . . . . . . . 137
9.1.2 2018 Free-Response Answers . . . . . . . . . . . . . . . . . . . . . . . 141
9.1.3 2017 Free-Response Questions . . . . . . . . . . . . . . . . . . . . . . 146
9.1.4 2017 Free-Response Answers . . . . . . . . . . . . . . . . . . . . . . . 149
9.1.5 2016 Free-Response Questions . . . . . . . . . . . . . . . . . . . . . . 154
9.1.6 2016 Free-Response Answers . . . . . . . . . . . . . . . . . . . . . . . 158
9.2 Online Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

v
Chapter 0

Background & Review

Everything mentioned in this chapter should already be familiar to you from other math
classes. These topics span two major areas: algebra/pre-calculus and limits. Ideas from
these topics will often be used implicitly or without a passing reference.

If you are unfamiliar with anything mentioned, you can use man of the great online resources
like Khan Academy, to familiarize yourself before moving forward.

0.1 Algebra and Pre-Calculus

0.1.1 Complex Numbers
Definition. i is called the imaginary unit. It’s defined by i2 = −1.
Complex numbers (C) have the form z = α + βi, where α and β are real numbers. The α
part of z is called the real part, so <(z) = α. The β part of z is called the imaginary part,
so =(z) = βi.

Often, complex numbers are visualized as points or vectors in a 2D plane, called the com-
plex plane, where α is the x-component, and β is the y-component. Thinking of complex
numbers like points helps us definepthe magnitude of complex numbers and compare them.
a point (x, y) has a distance x2 + y 2 from the origin, we can say the magnitude of z,
Since p
|z| is α2 + β 2 . Thinking of complex numbers like vectors helps us understand adding two
complex numbers, since you just add the components like vectors.

A common operation on complex numbers is the complex conjugate. The complex conjugate
of z = α + βi is z = α − βi. z and z are called a conjugate pair.

Conjugate pairs have the following properties.

Let z, w ∈ C.
z±w =z±w

1
 zw = z w

z=z⇔z∈R

zz = |z|2 = |z|2

z=z

zn = zn

z −1 = z
|z|2

0.1.2 Factoring Polynomials

We want to break up a polynomial like f (x) = a0 + a1 x1 + . . . an xn into linear factors so
that f (x) = c(x − b1 ) · . . . · (x − bn ). This form makes it simple to see that the roots of f ,
solutions to f (x) = 0, are x = b1 . . . bn .

For quadratics, f (x) = ax2 + bx + c, there exists a simple formula that will give us both
roots, the quadratic formula √
−b ± b2 − 4ac
x= .
2a
We can see that when b2 − 4ac < 0, like for f (x) = x2 + 5x + 10, we will get complex roots
α ± βi. For any polynomial, these roots come in pairs, so if α + βi is a root, then so is
α − βi. This means that every conjugate pair α ± βi has a quadratic equation with those
roots. Sometimes we will not factor quadratics with complex roots into linear terms.

Although there do exist explicit formulas for finding roots for cubic (degree 3) and quartic
(degree 4) equations, they are too long and not useful enough to memorize. When working
by hand, we instead use other tricks to find roots.

There are a few useful tricks that can help. If the polynomial doesn’t have a constant term,
then 0 is a root. If all the coefficients sum to 0, then 1 is a root. For certain polynomials with
an even number of terms, like all cubics of the form ax3 + bx2 + cax + cb we can factor out a
term from the first two and last two terms to get x2 (ax + b) + c(ax + b) = (ax + b)(x2 + c).
For other polynomials, we might just try guessing and checking values. However, we need a
more efficient way that works in general.

Since we are looking to find linear factors f (x) = c(x − b1 ) · . . . · (x − bn ), we can see that the
constant term in the polynomial is the product of the roots b1 . . . bn . In fact, since the co-
efficients of polynomials are completely determined by the roots and the leading coefficient,
all the coefficients are sums and products of roots. You might remember when factoring
quadratics that the coefficient of x term is the sum of the two roots. These rules are called

2
Vieta’s formulas.

So, if we have the constant term, we can check all of its integer factors to see if any are roots.
For each root, we can divide, using a technique like synthetic division, to continue finding
the rest of the roots. This method is especially useful on tests because the roots tend to be
integers.

Example. Factor the polynomial x5 + x4 − 2x3 + 4x2 − 24x.

We can immediately see that there is no constant term, so x = 0 is a root. Now we need
to work on factoring x4 + x3 − 2x2 + 4x − 24.
The factors of -24 are: -24, -12, -8, -6, -4, -3, -2, -1, 1, 2, 3, 4, 6, 8, 12, and 24. Starting
from roots close to 0 and working outwards, we find that x = 2 is a root. So, we synthetic
divide like so

x = 2 | 1 1 -2 4 -24
↓ 2 6 8 24
1 3 4 12 | 0

to see that now we need to work on factoring x3 + 3x2 + 4x + 12. x3 + 3x2 + 4x + 12 =

x2 (x + 3) + 4(x + 3) = (x + 3)(x2 + 4), so x = −3 is a root, and we need to work on
factoring x2 + 4. x2 + 4 has two complex roots x = ±2i, so we’ll leave it as a quadratic.

x5 + x4 − 2x3 + 4x2 − 24x = x(x − 2)(x − 3)(x2 + 4)

0.1.3 Trig Functions & The Unit Circle

Imagine a circle of radius 1 centered at the origin that we’ll call the unit circle. The x and y
coordinates of a point on the unit circle are completely determined by the angle θ in radians
between the x-axis and a line from the origin to the point.

The function cos θ tells us x-coordinate of the point, while sin θ tells us the y-coordinate of
sin θ
the point. The function tan θ = cos θ
tells us the slope of the line from the origin to the
point. Most of the trig functions have geometric interpretations as shown below. The most
sin
used ones are sin, cos, tan = cos , cot = cos
sin
1
, csc = sin 1
, and sec = cos .

3
 Figure 1: Wikipedia - Unit circle

We can also think about the inverses of these trig functions. These are either notated with
a -1 exponent on the function, or the prefix arc in front of the function name. Many of these
functions are only defined on a part of the domain [0, 2π]. Below is a table of the inverse
trig functions and their domains.

Function Domain
arcsin [−1, 1]
arccos [−1, 1]
arctan (−∞, ∞)
arccot (−∞, ∞)
arccsc (−∞, −1] ∪ [1, ∞)
arcsec (−∞, −1] ∪ [1, ∞)

0.1.4 Trig Identities

As we could see in Figure 0.1.3, sin and cos form a right triangle with a hypotenuse of length
1. So, using the Pythagorean Theorem,
sin2 θ + cos2 θ = 1.
By dividing by sin2 or cos2 , we can also get
1 + cot2 θ = csc2 θ and tan2 θ + 1 = sec2 θ.
Together, these 3 identities are called the Pythagorean Identities.

4
 We can also relate functions and co-functions.
π 
xxx(θ) = cxx −θ
2
Some of the most useful and used identities are the sum and difference.
sin (α ± β) = sin α cos β ± cos α sin β
cos (α ± β) = cos α cos β ∓ sin α sin β
tan α ± tan β
tan (α ± β) =
1 ∓ tan α tan β
   
α±β α∓β
sin α ± sin β = 2 sin cos
2 2
   
α+β α−β
cos α + cos β = 2 cos cos
2 2
   
α+β α−β
cos α − cos β = −2 sin sin
2 2

0.1.5 Exponentials & Logarithms

Definition. e is the base of the natural logarithm. It’s definied by the limit
 n
1
e = lim 1 + .
n→∞ n
exp x = ex and ln x are inverse functions of each other such that
eln x = x, ln ex = x.
Just like other exponents, the normal rules for adding, subtracting, and multiplying powers
apply.
ex
ex ey = ex+y , y = ex−y , and (ex )k = exk .
e
Similar rules apply for logarithms.
 
x
, and ln ab = b ln a.


ln x + ln y = ln xy, ln x − ln y = ln
y
You can also change a logarithm of any base to a natural logarithm.
ln a
logb a =.
ln b
e is also unique in that it is the only real number a satisfying the equation
d x
a = ax .
dx
meaning ex is its own derivative.1
1
Don’t worry if you don’t know what a derivative is yet. It’s one of the first topics we’ll cover in calculus.

5
0.1.6 Partial Fractions
P (x)
If we have a function that is the quotient of two polynomials f (x) = Q(x) , it’s often easier to
break this quotient into a sum of parts where the denominator is a linear or quadratic factor
and the numerator is always a smaller polynomial degree than the denominator.
Example.
2x − 1 1/2 −3 5/2
= + + .
x3 2
− 6x + 11x − 6 x−1 x−2 x−3
One natural way to find these small denominators comes from the linear factors of the
denominator where we keep complex roots in quadratic form. This way, when making a
common denominator, we get back the original big denominator. However, there are a few
special cases we have to take care of.

Linear Factors
This is the the most basic case where the degree of the numerator is less than the degree
of the denominator and the denominator factors into all linear factors (ie no complex roots)
with no repeated roots. In this case we can write
P (x) A1 An
= + ... +
Q(x) (x − a1 ) (x − an )
Multiplying each side by Q(x),

P (x) = A1 (x − a2 ) . . . (x − an ) + . . . + An (x − a1 ) . . . (x − an−1 )

We can then find each Ai by evaluating both sides at x = ai , since every term except the ith
one has an (x − ai ) factor, all terms except the ith one will go to 0. So,
P (ai )
Ai =
(x − ai ) . . . (x − ai−1 )(x − ai+1 ) . . . (x − an )
2x−1
Example. Find the partial fraction decomposition of x3 −6x2 +11x−6

x3 − 6x2 + 11x − 6 = (x − 1)(x − 2)(x − 3).

So,
2x − 1 A1 A2 A3
= + + .
x3 − 6x2 + 11x − 6 x−1 x−2 x−3
Multiplying each side by the denominator,

2x − 1 = A1 (x − 2)(x − 3) + A2 (x − 1)(x − 3) + A3 (x − 1)(x − 2).

At x = 1,
1
1 = A1 (1 − 2)(1 − 3) =⇒ A1 = .
2
6
 At x = 2,
3 = A2 (2 − 1)(2 − 3) =⇒ A2 = −3.
At x = 3,
5
5 = A3 (3 − 1)(3 − 2) =⇒ A3 = .
2
So,
2x − 1 1/2 −3 5/2
= + + ,
x3 2
− 6x + 11x − 6 x−1 x−2 x−3
just as was shown in the previous example.

Repeated Linear Factors

If Q(x) has repeated roots, it factors into

Q(x) = R(x)(x − a)k , k ≥ 2 and R(a) 6= 0

When making the common denominator for each repeated root of multiplicity k (i.e the root
appears k times), we do

P (x) A1 Ak
k
= (Decomposition of R(x)) + + ... +
R(x)(x − a) x−a (x − a)k

You would then multiply each side by the denominator like in the linear factors case and
solve for the coefficients. The only additional difficulty is that you might have to use previous
results or solve a system of linear equations to get some of the constants.
x2 +5x−6
Example. Find the partial fraction of x3 −7x2 +16x−12
.

x3 − 7x2 + 16x − 12 = (x − 3)(x − 2)2 .

So,
x2 + 5x − 6 A1 A2 A3
= + + .
x3 − 7x2 + 16x − 12 x − 3 x − 2 (x − 2)2
Multiplying each side by the denominator,

x2 + 5x − 6 = A1 (x − 2)2 + A2 (x − 2)(x − 3) + A3 (x − 3).

At x = 2,
8 = A3 (2 − 3) =⇒ A3 = −8.
At x = 3,
18 = A1 (3 − 2)2 =⇒ A1 = 18.

7
 Now we’ll use our results for A1 and A3 to find A2 using a value for x that isn’t 2 or 3 so
the A2 term doesn’t become 0. A good choice is x = 0.
At x = 0,
−6 = 18(0 − 2)2 + A2 (0 − 2)(0 − 3) + −8(0 − 3) =⇒ A2 = −17.
So,
x2 + 5x − 6 18 17 8
3 2
= − − .
x − 7x + 16x − 12 x − 3 x − 2 (x − 2)2

Quadratic Factors
If a polynomial has complex roots, then it has a quadratic factors. Here, we’ll assume that
the quadratic factor isn’t repeated. So,
Q(x) = R(x)(ax2 + bx + c), b2 − 4ac < 0, and R(x) is not divisible by ax2 + bx + c
In this case, we say
P (x) A1 x + B1
2
= (Decomposition of R(x)) + 2
R(x)(ax + bx + c) ax + bx + c
and then solve for the constants in the numerator, possibly having to solve a system of
equations or using previous results and less convenient values for x.
6x2 +21x+11
Example. Find the partial fraction decomposition of x3 +5x2 +3x+15
.

x2 + 5x2 + 3x + 15 = (x + 5)(x2 + 3).

So,
6x2 + 21x + 11 A1 A2 x + B2
3 2
= + .
x + 5x + 3x + 15 x+5 x2 + 3
Multiplying each side by the denominator,
6x2 + 21x + 11 = A1 (x2 + 3) + (A2 x + B2 )(x + 5).
At x = −5,
56 = 28A1 =⇒ A1 = 2.
Now we’ll use the previous result and another value for x. We can use x = 0 to not have
to worry about the A2 term.
At x = 0,
11 = 2(3) + (B2 )(5) =⇒ B2 = 1.
Now we’ll use the previous 2 results to find A2 . x = 1 is a good choice to keep the numbers
small.
At x = 1,
38 = 2(1 + 3) + (A2 + 1)(6) =⇒ A2 = 4.
So,
6x2 + 21x + 11 2 4x + 1
3 2
= + 2
x + 5x + 3x + 15 x+5 x +3

8
Repeated Quadratic Factors
If a polynomial has repeated complex roots, then we can write

Q(x) = R(x)(ax2 + bx + c)k , b2 − 4ac < 0, k ≥ 0, and R(x) is not divisible by (ax2 + bx + c)k .

Now we have to do a combination of what we did for repeated linear factors and quadratic
factors. We say

P (x) A1 x + B1 Ak x + Bk
2 k
= (Decomposition of R(x)) + 2 + ... +
R(x)(ax + bx + c) ax + bx + c (ax2 + bx + c)k

and then solve for the coefficients in the numerator.

3x4 −2x3 +6x2 −3x+3
Example. Find the partial fraction decomposition of x5 +3x4 +4x3 +12x2 +4x+12
.

x5 + 3x4 + 4x3 + 12x2 + 4x + 12 = (x + 3)(x2 + 2)2 .

So,
3x4 − 2x3 + 6x2 − 3x + 3 A1 A2 x + B2 A3 x + B3
5 4 3 2
= + + 2 .
x + 3x + 4x + 12x + 4x + 12 x+3 x2 + 2 (x + 2)2
Multiplying each side by the denominator,

3x4 − 2x3 + 6x2 − 3x + 3 = A1 (x2 + 2)2 + (A2 x + B2 )(x2 + 2)(x + 3) + (A3 x + B3 )(x + 3).

At x = −3,
363 = 121A1 =⇒ A1 = 3.
Now, we’ll use our result for A1 and pick a value for x that minimizes the number of
things we need to solve for. We have 4 unknowns, so we’ll likely need 4 different values
for x.
At x = 0,
3 = 3(2)2 + B2 (2)(3) + B3 (3) =⇒ 2B2 + B3 = −3.
At x = 1,

7 = 3(3)2 + (A2 + B2 )(3)(4) + (A3 + B3 )(4) =⇒ 3A2 + A3 + 3B2 + B3 = −5.

At x = −1,

17 = 3(3)2 + (−A2 + B2 )(3)(2) + (−A3 + B3 )(2) =⇒ −3A2 − A3 + 3B2 + B3 = −5.

At x = 2,

53 = 3(6)2 + (2A2 + B2 )(6)(5) + (2A3 + B3 )(5) =⇒ 12A2 + 2A3 + 6B2 + B3 = −11.

9
 Now we have a system of equations


 0A2 + 0A3 + 2B2 + B3 = −3

3A + A + 3B + B = −5
2 3 2 3
=⇒ A2 = 0, A3 = 0, B2 = −2, and B3 = 1.


 −3A2 − A3 + 3B2 + B3 = −5
12A2 + 2A3 + 6B2 + B3 = −11


So,
3x4 − 2x3 + 6x2 − 3x + 3 3 2 1
5 4 3 2
= − 2 + 2
x + 3x + 4x + 12x + 4x + 12 x + 3 x + 2 (x + 2)2

Improper Fractions
If the degree of the numerator is greater than or equal to the degree of the denominator, we
have a case of improper fractions. In this case, we have to do polynomial long division to
get a quotient and remainder and then decompose the remainder if necessary. So,
P (x) S(x)
= R(x) +
Q(x) Q(x)
x3 +3
Example. Find the partial fraction decomposition of x2 −2x−3
.
First we do polynomial long division to find that
x3 + 3 7x + 9
2
=x+2+ 2 .
x − 2x − 3 x − 2x − 3
Now that the numerator is of a lesser degree than the denominator, we can decompose it
normally.
x2 − 2x − 3 = (x − 3)(x + 1).
So,
7x + 9 A1 A2
= + .
x2− 2x − 3 x−3 x+1
Multiplying each side by the denominator,
7x + 9 = A1 (x + 1) + A2 (x − 3).
At x = −1,
−1
2 = −4A2 =⇒ A2 = .
2
At x = 3,
15
30 = 4A1 =⇒ A1 = .
2
So,
x3 + 3 15/2 −1/2
2
=x+2+ +
x − 2x − 3 x−3 x+1

10
Chapter 1

Limits & Continuity

Limits are a way of describing what happens to a function f (x) as x gets arbitrarily close
to a value from some direction (positive or negative). This allows us not only to deal with
”holes” in some functions but describe some of the building blocks of calculus, namely the
derivative.

1.1 Limit Definition

Definition. Let f : D ⊆ R → R. Let c ∈ R be a limit point (ie c ∈ D or c is on the
boundary of D). f has a limit L as x approaches c if for any given positive real number ,
there is a positive real number δ such that for all x ∈ D,
   
0 < x − c < δ =⇒ f (x) − L < . (1.1)

We write this as
lim f (x) = L.
x→c

11
 Figure 1.1: Wikipedia - (, δ)-definition of limit

Visually, what this means is that for any ”error bound” of y values , I can give you a
corresponding error bound of x values δ such that all values of f (z) for z ∈ (c − δ, c + δ)
bound are between L −  and L + .
We don’t use this definition of the limit very often because it’s a bit cumbersome. However,
it’s important to know that when we use the limit, this is the formal definition making things
work.
Example. Use the (, δ) definition of the limit to show that
1
lim x sin = 0.
x→0 x
Letting  > 0, we need to find corresponding δ > 0 that satisfies the definition for L = 0.
Knowing that sin is bounded between -1 and 1,

x sin 1 − 0 = x sin 1  = x sin 1  ≤ x.

        
x x x
     
Letting δ = , if 0 < x − 0 < delta, then x sin x1 − 0 ≤ x < , as required by the
definition.

1.2 Limit Properties

Limit have many nice properties all allow us to make useful simplifications when evaluating
a limit. Let
lim f (x) = L and lim g(x) = M.
x→c x→c

12
 Sum and Difference Rule: lim (f (x) ± g(x)) = L ± M
x→c
Product Rule: lim (f (x)g(x)) = LM
x→c
Constant Multiple Rule: lim k · f (x) = k lim f (x) = kL
x→c x→c
f (x) limx→x f (x) L
Quotient Rule: lim = = , ifM 6= 0
x→c g(x) limx→c g(x) M
 n
Power Rule: If n 6=∈ R, lim (f (x))n = lim f (x) = Ln
x→c x→c

1.2.1 “Substitution Rule”

Although it may seem obvious from our idea that limits describe behavior at a point that
if f (x) is defined at x = c, then limx→c f (x) = f (c). However, this is not always the case.
Remember that our definition of a limit required these  and δ neighborhoods around the
limit point. If f (x) is defined at x = c, but (c, f (c)) is not a point in these neighborhoods
for any  > 0, then the limit will not evaluate to f (c).

Example. Find the limit of f (x) as x approaches 2 for the following function.
(
x2 x 6= 2
f (x) = .
0 x=2

We can clearly see that f (2) = 0, but for  = 0.1, for example, there is no δ that can
satisfy our definition, as points like (2 − δ, 4 − 2δ + δ 2 ) would outside the neighborhood
around (2, 0). In fact, the correct limit value is 4, the same as if f (x) = x2 for all x.
There are some more nuances we’ll need to describe before we can say when it’s OK to
substitute to evaluate a limit.

1.3 Left & Right Hand Limits

Our definition of the limit requires that the function get arbitrarily close to the limit value
when approaching from both the left and right hand sides. However, we can evaluate limits
by specifying that we only approach from one side. We usually notate this with a superscript
+ or − next to the x limit value. So,

lim f (x)
x→0+

would mean “the limit of f (x) as x approaches 0 from the right”, while

lim f (x)
x→0−

would mean “the limit of f (x) as x approaches 0 from the left.”

13
Our definition of the limit from both sides requires the left and right sides to be the same.
If they are different, the the limit does not exist.

lim f (x) = lim− f (x) =⇒ lim+ f (x) = lim− f (x) = lim f (x)
x→c+ x→c x→c x→c x→c

lim f (x) 6= lim− f (x) =⇒ lim f (x) does not exist (DNE).
x→c+ x→c x→c

1.4 Sandwich Theorem

We can use the Sandwich Theorem to indirectly find limits by ”sandwiching” the function in
question between two functions we do know the limit of. If these two sandwiching functions
go to the same value in the limit, then so to must the function in question.
Theorem (The Sandwich Theorem). If g(x) ≤ f (x) ≤ h(x) and limx→c g(x) = limx→c h(x) =
L, then limx→c f (x) = L.
Example. Evaluate the following limit
sin θ
lim
θ→0 θ

We’ll need to use some geometric ideas to solve this limit. Consider the following on a
unit circle.

Figure 1.2: Triangle with internal angle θ inside a unit circle.

We can see that the area of the swept arc is between the two triangle with base of length
1 and heights of sin θ and tan θ. So, we can write the following inequality.
1 1 1 sin θ
sin θ ≤ θ ≤
2 2 2 cos θ
sin θ
sin θ ≤ θ ≤
cos θ

14
 Taking the reciprocal of each part and multiplying by sin θ,
sin θ
1≥ ≥ cos θ.
θ
Taking the limit of as θ approaches 0 of each term,
sin θ
1 ≥ lim ≥ lim cos θ
θ→0 θ θ→0
sin θ
1 ≥ lim ≥ 1.
θ→0 θ

So, by the Sandwich Theorem,

sin θ
lim = 1.
θ→0 θ

1.5 Infinite Limits

Although our limit definition works for finite values of c, it’s also useful to think about what
happens as c goes to ±∞. We’ll need to add to our limit definition to incorporate infinite
values, since it doesn’t make sense to talk about neighborhoods at infinity.

Definition. Let f be a real-valued function defined on some subset D ⊆ R that contains

arbitrarily large values.
lim f (x) = L
x→∞

if for every real  > 0, there is a real number N > 0 such that for all x ∈ D,
 
x > N =⇒ f (x) − L < . (1.2)

All the same properties that we described for finite limits, like the Sum and Difference Rule,
still hold for infinite limits.

1.5.1 End Behavior Model

When x is numerically large, we can often model the behavior of a complicated function with
a simplier one that behaves roughly the same for numerically large input values and is the
same in the limit. There are a few rules that these follow.

1. For a polynomial, the end-behavior is highest-degree term.

2. For a rational function, like a ratio of polynomials, the end behavior is the ratio of the
highest degree terms.

3. For more complicated functions, we may need to use some reasoning about the graph
of the function and limit properties to determine end-behavior.

15
1.5.2 Horizontal Asymptotes
Horizontal Asymptotes are a special type of end-behavior model.
Definition. The line y = b is a horizontal asymptote of y = f (x) if limx→∞ f (x) = b or
limx→−∞ f (x) = b.
We can determine horizontal asymptotes for rational functions (usually quotient of polyno-
mials). There are a few cases to consider
1. If the numerator is a higher degree than the denominator, there is no horizontal asymp-
tote, so we’ll need a different method to calculate what happens at ±∞.
2. If the denominator is a higher degree than the numerator, then there is a horizontal
asymptote at y = 0.
3. If the numerator and denominator have the same degree, there is a horizontal asymp-
tote at y = k where k is the ratio of the highest degree terms.
Example. Find the following limits, if they exist.

x3 − 6x + 1 x−9 6x2 − 4x5 + 7x − 1

1. lim 2. lim 3. lim
x→∞ x2 + 2x − 3 x→−∞ 2x − x2 x→∞ 12x5 − 3x2 + 2
3x + 1
4. lim   5. lim x + e−x 6. lim x + e−x
x→∞ x + 2 x→∞ x→−∞

1. Since the numerator degree is bigger than the denominator degree, we’ll need to use
the end behavior model. The end behavior model tells us that the numerator term
dominates and has positive values, so the limit evaluates to ∞.
2. Since the denominator has higher degree than the numerator, there is a HA at y = 0,
so the limit evaluates to 0.
3. Since the numerator and denominator have the same degree, the limit is the ratio of
the highest-degree coefficients, −1
3
.
 
4. The numerator and denominator have the same degree. For x > 0, x + 2 = x + 2,
so the limit is the ratio of highest-degree coefficients, 3.
5. Looking at the two terms, we can see that as x gets large, e−x gets very small,
contributing less and less to the overall value. So, we can say that this function as
a right end behavior model of x, so the limit is ∞.
6. Looking at the two terms, we that that as x gets very large and negative, e−x changes
much faster than x. That is, e−x contributes more and more to the overall value of
the function compared to x. So, we can say that this function has a left end behavior
model of e−x , so the limit is ∞.

16
 1.6 Continuity Definition
When we were looking at limits, we noticed that we can’t always substitute to find the limit,
even if the function is defined there. In the example given to show that substitution and
the limit can give different results, we saw a special type of function that seemed to have
a ”hole” at the point we were interested in finding the limit of. This function is said to
be discontinuous at this point, and in this section we’ll define when a function is or isn’t
continuous at a point based on this idea of the limit and substitution giving different values.

Definition. Let f (x) be a real-valued function defined over D ⊆ R. f (x) is continuous at

some point x = c if all of the following hold.

1. limx→c f (x) exists

2. f (c) is defined

3. limx→c f (x) = f (c) (substitution works)

Otherwise, f (x) is discontinuous at c.1

We say that a function is continuous on an interval if it’s continuous on every point in that
interval.2

Example. Find the points of continuity and discontinuity of the following functions

1
1. f (x) = 2. g(x) = e1/x
x2 +1

1. There are no points where f (x) or its limit are undefined. Further, there are no
points where f and its limit at that point are different. So, f is continuous on
(−∞, ∞) and discontinuous on ∅.
2. Since 1/x is undefined at x = 0, g(x) is also undefined at x = 0. At every other
point, g and its limit are defined and are equal. So, g is continuous on (∞, 0)∪(0, ∞)
and discontinuous on [0].

1.7 Discontinuity Types

There are four major types of discontinuity.

Removable: If f is discontinuous at c but we can remove the discontinuity by setting

f equal to its limit at c, then f has a removable discontinuity at c.
1
Note that it’s not necessary for c ∈ D.
2
If the interval is closed on one or both sides, we check continuity on the open interval. Then, we check the
closed endpoints by looking at the limit from only one side.

17
 Jump: If f is discontinuous at c, and both of the one-sided limits exist but are
different, then f has a jump discontinuity at c.

Infinite: If f has a vertical asymptote at c, meaning one or both sides go to ±∞,

then f has an infinite discontinuity at c.

Oscillating: If f oscillates without limit at c, then f has an oscillating discontinuity

at c. An example of such a function would be sin x1 at x = 0.

It might seem strange that sin x1 has an oscillating discontinuity at x = 0 because we were
able to find the limit as x approaches of 0 of x sin x1 , a very similar function. However,
remembering how we applied the Ham Sandwich Theorem to find this limit, we see that the
x term bounds the amplitude of the oscillations, allowing the limit to be 0.

Example. For the following function state the following: its domain, any discontinuities and
their types, what values should redefine the function to remove any removable discontinuities
(give the extended function).
x3 − 7x − 6
f (x) =
x2 − 9
Polynomials are continuous on their entire domain of all real numbers. So, rational
functions like f can only be discontinuous when the denominator is equal to 0. This
happens in two places: x = 3 and x = −3. We’ll check the limits from each side at each
of these points to determine the type of discontinuity. For x = 3,

(x + 2)(x + 1)(x − 3) (x + 2)(x + 1) 20 10

lim+ f (x) = lim− f (x) = lim f (x) = lim = lim = = .
x→3 x→3 x→3 x→3 (x + 3)(x − 3) x→3 (x + 3) 6 3

So, f has a removable discontinuity at x = 3 because the left and right limits are the
same. For x = 3,
lim + f (x) = −∞ and lim + f (x) = ∞.
x→−3 x→−3

So, f has an infinite discontinuity at x = −3 because both of the left and right limits go
to ±∞. The value we got from the limits at x = 3 gives us the value we need to redefine
f as to remove the discontinuity. The extended function is therefore
(
f (x) x 6= 3
fe (x) = 10
3
x=3

18
1.8 Continuity Properties
These properties should look very similar to the properties of limits. Let f and g be contin-
uous functions at c.

Sum and Difference Rule: f ± g is continuous at c.

Product Rule: f · g is continuous at c.
Constant Multiple Rule: kf is continous at c for all real k.
f
Quotient Rule: is continuous at c as long as the value of the extended function of
g
g at c is not 0.
Composition Rule: f ◦ g is continuous at c if f is continuous at g(c).
 
Absolute Value Rule: f  is continuous at c.

The following types of functions are continuous on their domains

• Polynomials

• Rational functions, except where the denominator is 0

• Trigonometric functions where defined

Example. Show that the following function is continuous.

x2
 
f (x) = tan .
x2 + 4
2
We can write f as the composition of tan x and x2x+4 . tan x is continuous on its domain
because it is a trigonometric function. The only points not in its domain are (2n + 1) π2 ,
2
where n is an integer. x2x+4 is a rational function, but it’s denominator is never 0, so it
is continuous over all real numbers. Now, we just need to check that all points in the
2 2
range of x2x+4 are in the domain of tan x. The range of x2x+4 is [0, 1). None of the points
in this interval are not in the domain of tan x, so the composition is continuous over all
real numbers.

1.9 Intermediate Value Theorem

Theorem (Intermediate Value Theorem (IVT)). If f is continuous on the closed interval
[a, b], then for all c ∈ [f (a), f (b)], there exists x ∈ [a, b] such that f (x) = c.

That is, if f is continuous on [a, b], then f must take on every value between f (a) and
f (b). This encapsulates the idea that if a function is continuous on some interval, then it is
”connected” on that interval.

19
Example. Use the IVT to show that e−x = x has at least one solution.

Let f (x) = e−x − x. We are looking for x where f (x) = 0. f (0) = 1 and f (1) = 1e − 1.
Since f is continuous on the closed interval [0, 1], it must take on every value between 1
and 1e − 1. Since 1 is positive and 1e − 1 is negative, 0 is between these two values. Thus,
by the IVT, there must exist a solution between x = 0 and x = 1.

20
 Chapter 2

Derivatives

Derivatives describe how a function is changing. They are one of the two fundamental
operations in calculus.

2.1 Rates of Change & Tangent Lines

You might already be familiar from physics with the idea of an average rate of change over
some interval (often a time interval in physics). It is simply the amount of change that
occurred in the interval, divided by the length of the interval. This is exactly the same idea
as the slope of a line being ”rise over run”

f (b) − f (a)
∆fa,b = .
b−a
This is also known as the “secant slope”, which gets its name from secant lines on circles.

Example. Find the average rate of change of x2 − 1 over the interval [1, 4].

Applying the formula,

f (4) − f (1) 15 − 0
∆f1,4 = = = 5.
4−1 3
As we decrease the size of the interval, the secant line becomes closer and closer to a tangent
line. In the limit, as the size of the interval approaches 0, we get a tangent line, representing
an instantaneous rate of change.

f (a + h) − f (a) 1
∆fa = lim .
h→0 h

1
You may recognize this from pre-calculus as the “difference quotient”.

21
 Figure 2.1: Secant and Tangent Line

Example. Find the equation of the tangent line to f (x) = x2 − 4x at x = 1.

First, we need to find the instantaneous rate of change of f at x = 1. We do this by
evaluating the limit.
f (1 + h) − f (1)
∆f1 = lim
h→0 h
(1 + h) − 3(1 + h) − (1)2 + 3(1)
2
= lim
h→0 h
h2 + 2h + 1 − 3 − 3h − 1 + 3
= lim
h→0 h
2
h −h
= lim
h→0 h
= lim h − 1
h→0
= −1.
Now that we have the slope of the tangent line, we can write the equation of the line in
point-slope form. We can also rearrange to standard for is needed.
y − f (1) = −1(x − 1)
y + 2 = −x + 1
y = −x − 1.

2.2 Definition of the Derivative

Definition. The derivative of a function f (x), notated f 0 (x), is the slope of f at any point
along f . This is also the slope of the tangent line to f at this point, which is also the

22
 instantaneous rate of change of f at this point.

f (x + h) − f (x)
f 0 (x) = lim (assuming the limit exists). (2.1)
h→0 h
Example. Find the derivative of f (x) = x2 + 4.

Applying the definition,

f (x + h) − f (x)
f 0 (x) = lim
h→0 h
(x + h)2 + 4 − x2 − 4
= lim
h→0 h
x + 2xh + h2 + 4 − x2 − 4
2
= lim
h→0 h
2xh + h2
= lim
h→0 h
= lim 2x + h
h→0
2
= 2x.

There are a couple different notations that all mean the derivative of y = f (x). You should
be familiar with all of them.
dy
y0
dx
df d
f (x)
dx dx

2.2.1 Derivative at a Point

The formula given in the definition is useful because it gives a function that can give the
derivative at any point, but it may not be useful or feasible to use this formula. Instead, we
can use a formula to just give us the derivative at one point.

f (x) − f (a)
f 0 (a) = lim . (2.2)
x→a x−a
1
Example. Find the derivative of f (x) = x
at x = 2.

2
Note that the constant term 4 didn’t contribute anything to the outcome. It was canceled immediately when
subtracting f (x).

23
 Applying the formula for the derivative at a point,
f (x) − f (2)
f 0 (2) = lim
x→2 x−2
1/x − 1/2
= lim
x→2 x−2
−(x − 2)
= lim
x→2 2x(x − 2)
−1
= lim
x→2 2x
−1
= .
4

2.2.2 Left & Right Hand Derivatives

The normal derivative is defined in terms of a two-sided limit, meaning that the left and
right hand limits are equal. However, we can also calculate left and right hand derivatives
at every point along the function’s domain, which may be useful if the left and right hand
derivatives are not equal or the point in question is on the boundary of the domain.
Example. Find the left and high hand derivatives of the following function at x = 1. Say if
the derivative at this point exists and why.
(
x2 + x x ≤ 1
f (x) = .
x+1 x>1
Evaluating the left hand limit,
f (x) − f (1)
f 0 (1− ) = lim−
x→1 x−1
2
x +x−2
= lim−
x→1 x−1
(x − 1)(x + 2)
= lim−
x→1 x−1
= lim− x + 2
x→1
= 3.
Evaluating the right hand limit,
f (x) − f (1)
f 0 (1+ ) = lim+
x→1 x−1
x+1−2
= lim+
x→1 x−1
x−1
= lim+
x→1 x − 1
= 1.

24
 Since f 0 (1− ) 6= f 0 (1+ ), the derivative does not exist at x = 1.

2.3 Differentiability
As we saw when looking at left and right hand derivatives, a derivative will fail to exist
when the left and right hand derivatives differ. However, there are several other cases where
a function might not be differentiable at a point.
1. Corner: This is the simplest case where both the one   sided derivatives exist, are
finite, but are different. An example would be f (x) = x at x = 0.

2. Cusp: This is sort of an extreme example of a corner where the one of the one
sides derivatives
√ approaches −∞ and the other approaches ∞. An example would be
3 2
f (x) = x at x = 0.

3. Vertical Tangent: This is where the one sided√derivatives exist and agree, but
approach −∞ or ∞. An example would be f (x) = 3 x at x = 0.

4. Discontinuity: This is where one of the one sided derivatives doesn’t exist. An
example would be the unit step function at x = 0, which is 1 for x ≥ 0 and -1 for
x < 0.

2.3.1 Differentiability Implies Local Linearity

A function is locally linear at a point a when it is differentiable at a and closely resembles
its own tangent line at a. Visually, what this means is that as you ”zoom in” on a locally
linear point, the curve will look more and more like a line.

Figure 2.2: Zooming in on a cos curve

Numerical Approximation: Symmetric Difference Quotient

We can use the derivative at a point and a very small value of h to approximate the tangent
slope. However, we can get a better estimate with the same value of h by using the symmetric
difference quotient.
f (a + h) − f (a − h)
.
2h
25
 A calculator might prefer this formula because it tends to give better approximations.

2.3.2 Differentiability Implies Continuity

Theorem. If f 0 (a) exists, then f is continuous at a.
The converse is not necessarily true: a function can be continuous but not differentiable.
Example. Determine whether or not the follow function is differentiable at x = 2.
(
x2 + 1 x ≤ 2
f (x) = .
4x − 4 x > 2
Looking at the graph of this function, we see there is a jump discontinuity at x = 2. If
the function was differentiable at x = 2, it would be continuous at x = 2. Since it is not
continuous at x = 2, it must not be differentiable at x = 2.

2.4 Intermediate Value Theorem for Derivatives

Theorem (IVT for Derivatives). If f is differentiable on the closed interval [a, b], then f 0
takes on every value between f 0 (a) and f 0 (b).
Example. Show that any function that has the following function as it derivative cannot be
differentiable on the interval −1 ≤ x ≤ 1.
(
0 −1 ≤ x < 0
f (x) = .
1 0≤x≤1

f does not take all values between f (−1) = 0 and f (1) = 1. So, by the contrapositive of
the IVT for Derivatives, any function that has f as its derivative will not be differentiable
on [−1, 1].

2.5 Derivative Rules

2.5.1 Basic Properties
All of these properties should look familiar from properties of limits and continuity. Let f
and g be differentiable functions of x. Let c be some real constant.
Sum and Difference Rule: (f ± g)0 = f 0 ± g 0
Constant Multiple Rule: (cf )0 = cf 0
Constant Rule: (c)0 = 0

2
This follows from using the sum and difference rule with the power rule.

26
2.5.2 Power Rule
Lemma. Let f (x) = xn where n is a positive integer. Then

f 0 (x) = nxn−1 . (2.3)

Proof. Applying the limit definition of the derivative,

(x + h)n − xn
f 0 (x) = lim .
h→0 h
Applying the binomial theorem (think Pascal’s Triangle),

xn + nhxn−1 + . . . + nk hk xn−k + . . . + hn − xn


0
f (x) = lim
h→0
  h
n k−1 n−k
= lim nxn−1 + . . . + h x + . . . + hn−1
h→0 k
= nxn−1 .


We’ll revisit the power rule after we discover some more detailed rules to show that n can
be any real number, not just a positive integer.

2.5.3 Higher Order Derivatives

Definition. Let n be a positive integer. The nth derivative of f is
d (n−1)
f (n) (x) = f (x),
dx
where f (n−1) (x) is the n − 1th derivative of f . Using equivalent notation,
dn dn−1
f = f.
dxn dxn−1
Example. Find the third derivative of f (x) = x3 + x2 .
Taking the derivative once using the power rule and sum and difference rule,

f 0 (x) = 3x2 + 2x.

Taking the derivative a second time using the power, constant multiple rules, and sum
and difference rules,
f 00 (x) = 6x + 2.
Taking the derivative a final time using the power, constant multiple, constant, and sum
and difference rules,
f 000 (x) = 6.

27
2.5.4 Product Rule
Lemma. Let f and g be differentiable functions. Then

(f g)0 = f g 0 + gf 0 . (2.4)

Proof. Applying the definition of the derivative and limit properties,

f (x + h)g(x + h) − f (x)g(x)
(f g)0 = lim
h→0 h
f (x + h)g(x + h) − f (x + h)g(x) + f (x + h)g(x) − f (x)g(x)
= lim
h→0 h
f (x + h) (g(x + h) − g(x)) g(x) (f (x + h) − f (x))
= lim + lim
h→0 h h→0 h
g(x + h) − g(x) f (x + h) − f (x)
= lim f (x + h) lim + g(x) lim
h→0 h→0 h h→0 h
= f g 0 + gf 0

. 

Example. Given the the derivative of sin (x) is cos (x), find the derivative of x2 sin (x).

Using the product rule and power rule,

f 0 (x) = x2 cos (x) + sin (x)2x.

2.5.5 Quotient Rule

Lemma. Let f and g be differentiable functions where g 6= 0. Then
 0
f gf 0 − f g 0
= . (2.5)
g g2

28
Proof. Using the definition of the derivative and limit properties,
 0 f (x+h) f (x)
f g(x+h)
− g(x)
= lim
g h→0 h
1 f (x + h)g(x) − f (x)g(x + h)
= lim
h→0 h g(x + h)g(x)
1 f (x + h)g(x) − f (x)g(x) + f (x)g(x) − f (x)g(x + h)
= lim
h→0 h g(x + h)g(x)
1 f (x + h)g(x) − f (x)g(x) + f (x)g(x) − f (x)g(x + h)
= lim
h→0 g(x + h)g(x) h
 
1 f (x + h)g(x) − f (x)g(x) f (x)g(x) − f (x)g(x + h)
= lim +
h→0 g(x + h)g(x) h h
 
1 f (x + h) − f (x) g(x + h) − g(x)
= lim g(x) lim − f (x) lim
h→0 g(x + h)g(x) h→0 h h→0 h
1
= 2 (g(x)f 0 (x) − f (x)g 0 (x))
g (x)
gf 0 − f g 0
= .
g2

sin (x)
Example. Given that the derivative of sin (x) is cos (x), find the derivative of x2
.

Applying the quotient and power rules,

0 x2 cos (x) − 2x sin (x) x cos (x) − 2 sin (x)

f (x) = 4
= .
x x3

2.5.6 Chain Rule

Lemma. Let f and g be differentiable functions. Then
d
f (g(x)) = f 0 (g(x))g 0 (x). (2.6)
dx
Equivalently, if f is a function of g and g is a function of x,
df df dg
= . (2.7)
dx dg dx

29
Proof. Applying the limit definitions of the derivative,
df f (g(x + h)) − f (g(x))
= lim
dx h→0 h
f (g(x + h)) − f (g(x)) g(x + h) − g(x)
= lim
h→0 g(x + h) − g(x) h
f (g(x + h)) − f (g(x)) g(x + h) − g(x)
= lim lim
h→0 g(x + h) − g(x) h→0 h
 
f (g(x + h)) − f (g(x)) dg
= lim
h→0 g(x + h) − g(x) dx
df dg
= .
dg dx

Example. Find the derivative of y = (x2 + 1)5 using the chain rule.
y is a composition of the two functions x5 and x2 + 1. Applying the chain rule,
d d 2
y0 = (x2 + 1)5 (x + 1)
d(x2+ 1) dx
Making the substitution u = x2 + 1,
d 5 d 2
y0 = u (x + 1)
du dx
= 5u4 2x

Substituting back,

y 0 = 5(x2 + 1)4 2x
= 10x(x2 + 1)4 .

u Substitutions
As we did in the example, we can substitute a variable, usually called u when applying the
chain rule.
Example. Given that the derivative of sin x is cos x, find the derivative of f (x) = sin5 x.
f is a composition of x5 and sin. Substituting u = sin x, we can rewrite f as u5 . Applying
the chain rule,
df df du
=
dx du dx
= 5u4 cos x
= 5 sin4 x cos x.

30
2.5.7 Exponential Rule
Let’s find the derivative of the most natural exponential function: f (x) = ex . Using the
limit definition of the derivative,
ex+h − ex
f 0 (x) = lim
h→0 h

e eh − 1
x
= lim
h→0 h
h
e −1
= ex lim
h→0 h
Remembering the following definition of e,
 n
1
e = lim 1+ .
n→∞ n
Substituting h = 1/n,
e = lim (1 + h)1/h .
h→0
Putting substituting this definition for e into our work,
 h
1/h
(1 + h) −1
f 0 (x) = ex lim
h→0 h
(1 + h) − 1
= ex lim
h→0 h
h
= ex lim
h→0 h
x
=e .
Amazingly, this function is equal to it’s own derivative. In fact, aside from the trivial exam-
ple of 0, ex is the only function with this property.

We can apply the chain rule to find the derivative of bx for real, positive values of b.
f (x) = bx = ex ln b .
Let u(x) = x ln b.
f (x) = bx = eu(x) .
Applying the chain rule,
d u d
f 0 (x) = e x ln b.
du dx
= eu ln b
= ex ln b ln b
= bx ln b.

31
 2
Example. Find the derivative of f (x) = 2x .

Let u(x) = x2 .
f (x) = eu (x).
Using the exponential and chain rules,
df du
f 0 (x) =
du dx
d u d 2
= e x
du dx
= 2u ln (2)2x
2
= 2x ln (2)2x
2
= 2x ln (2)2x .

2.5.8 Implicit Differentiation

Although we normally have equations of the form y = f (x), some equations might be given
in are are more convenient to write in different forms. Using our understanding of the chain
rule, we can still work with these forms and find y 0 . There are generally three steps to finding
y 0 when equations are given in these different forms.

1. Get all terms involving y to one side of the equation. This step is technically optional
but usually makes the step 3 more convenient.

2. Take the derivatives of both sides of the equation, using the chain rule.

3. Rearrange to solve for y 0 , making substitutions for y to get the answer in terms of
input parameters (e.g x).

It’s possible that you can get multiple solutions for y 0 . You’ll need to check if

Example. Given that y 2 = x, find y 0 using implicit differentiation.

Step 1 is already complete by what’s given. Taking the derivative of both sides, remem-
bering to use the chain rule,
2yy 0 = 1.
Solving for y 0 ,
1
y0 = .
2y
√
Looking back at our original equation, we see y = ± x. Substituting back into our work,
1
y0 = ± √ .
2 x

32
Sometimes, it’s not possible or is not necessary for what you’re working on to complete step
3, meaning you leave your answer for y 0 in terms of y and input parameters.

Example. Find the derivative for y with respect to x at (0, 1) if y 4 = x3 + x + y.

Rearranging to complete step 1,

y 4 − y = x3 + x.

Taking the derivative of both sides,

(4y 3 − 1)y 0 = 3x2 + 1.

Solving for y 0 and completing step 2,

3x2 + 1
y0 = .
4y 3 − 1
Since we have both the x and y coordinates of where we’re looking for the slope, we don’t
need to complete step 3. We can simply substitute to find our numerical answer for y 0 .

0 0+1 1
y(0,1) = = .
4−1 3
Sometimes, we might already be given a formula for y, but rearranging and then doing
implicit differentiation is easier.

Example. Find the derivative with respect to x of y = ln x.

We don’t have any way to find the derivative of ln directly, but we do know how ln relates
to exponential functions, a form we do know how to differentiate. Exponentiating both
sides with base e,
ey = eln x = x.
Implicitly differentiating,
ey y 0 = 1.
Solving for y 0 ,
1
y0 = .
ey
Substituting what we were given for y,
1 1
y0 = = .
eln x x

33
Logarithmic Differentiation
Logarithmic differentiation is a certain type of differentiation where you take to natural log
of both sides and then implicitly differentiate. It’s especially useful when input parameters
appear in both the base and exponent.
Example. Find the derivative with respect to x of y = xln x .
Taking the natural log of both sides,
ln y = ln2 x.
Implicitly differentiating,
1 0 2 ln x
y =
y x
2 ln x
y0 = y
x
2 ln x
= xln x .
x
We now have the tools to prove the power rule for all real exponents.
Proof. Let n be an real number.
y = xn
ln y = n ln x
1 0 1
y =n
y x
ny
y0 =
x
nxn
=
x
= nxn−1 .


Combining Power and Exponential Rule

We can also find the derivative of f (x)g(x) , which combines the power and exponential rules
and is something you’ll likely won’t see in a standard calculus course.
y = fg
ln y = g ln f
1 0 1
y = g f 0 + ln f g 0
y f
 
0 g 1 0 0
y = f g f + ln f g
f
=f g−1
(gf 0 + f g 0 ln f ) .

34
2.6 Derivatives of Trig Functions
Let’s find the derivative of f (x) = sin x. Using the limit definition of the derivative,
sin (x + h) − sin x
f 0 (x) = lim
h→0 h
sin x cos h + cos x sin h − sin x
= lim
h→0 h
sin x (cos h − 1) + cos x sin h
= lim
h→0 h
cos h − 1 sin h
= sin x lim + cos x lim
h→0 h h→0 h
= sin x · 0 + cos x · 1
= cos x.

We can now use the chain rule to get the derivative of cos x.

f (x) = cos x
π 
= sin −x
 2π 
f 0 (x) = cos − x · −1
2
= − sin x.

The rest of the derivatives of the common trig functions follow from these results and the
quotient rule.

d d d
sin x = cos x sec x = sec x tan x tan x = sec2 x
dx dx dx
d d d
cos x = − sin x csc x = − csc x cot x cot x = − csc2 x.
dx dx dx

Example. Use the identity cos 2x = cos2 x − sin2 x to find the derivative of cos 2x. Express
your answer in terms of sin 2x. Does this answer the same as what you’d expect by finding
the derivative using the chain rule?
Taking the derivative of cos2 x,
d
cos2 x = cos x · − sin x + cos x · − sin x = −2 sin x cos x = − sin 2x.
dx
Taking the derivative of sin2 x,
d
sin2 x = sin x cos x + sin x cos x = 2 sin x cos x = sin 2x.
dx
35
 Combining these results to get the derivative of cos 2x,
d d d
cos 2x = cos2 x − sin2 x = − sin 2x − sin 2x = −2 sin 2x.
dx dx dx
This answer is indeed the same one we’d expect using the chain rule.

2.7 Derivatives of Inverse Trig Functions

Inverse trig functions are defined so that composing them with their corresponding trig
function gives you the identity. For example arcsin (sin x) = x. The inverse of a function f
can be obtained by reflecting the graph of f across the line y = x, effectively swapping the x
and y coordinates of each point along the graph. For functions that repeat y values (i.e are
not injective), we have to limit the domain of the inverse functions so they don’t repeat x
values. Since the trig functions don’t have any cusps or corners, neither will the inverse trig
functions. The slope f −1 will the the reciprocal of the slope of f , since the change in x in f
becomes the change in y of f −1 and the change in y of f becomes the change in x of f −1 .

df −1 

1
= df 
dx f (a)
 
da a

This leads us to a theorem that will help us derive the derivatives of the inverse trig functions.
Theorem. If f is differentiable at every point along an interval I and its derivative is never
0 along I, then f −1 exists and is differentiable on every point in f (I).

2.7.1 arcsin, arctan, and arcsec

Let’s apply this theorem and our differentiation rules to find the derivative of y = arcsin x.
From − π2 < x < π2 , the derivative of f (x) = sin x, f 0 (x) = cos x is never 0. So, we know
by the previous theorem that f −1 (x) = arcsin x exists and is differentiable along every point
in sin(− π2 < x < π2 ) = −1 < x < 1. So, we can rearrange the equation and implicitly
differentiate to find the derivative of arcsin x in this interval.

y = arcsin x
sin y = x
cos (y)y 0 = 1
1
y0 =
cos y
1
=p
1 − sin2 y
1
=√ , −1 < x < 1.
1 − x2

36
We can use the same method to find the derivative of arctan x, which is differentiable for all
real numbers.
y = arctan x
tan y = x
sec2 (y)y 0 = 1
1
y0 =
sec2 y
1
=
1 + tan2 y
1
= .
1 + x2
 
Although we generally use the same method for arcsec, which is differentiable for x > 1,
we have to be careful with out trig identities.
y = arcsec x
sec y = x
sec (y) tan (y)y 0 = 1
1
y0 =
sec (y) tan (y)
1
=
x tan (y)
1
= p
±x sec2 y − 1
1
= √
±x x2 − 1
(
√1 x>1
= x 1−x2
√−1 x < −1
x 1−x2
1  
=  √ , x > 1.
 x 1 − x2

2.7.2 arccos, arccot, and arccsc

Now that we’ve derived derivatives of arcsin, arctan, and arcsec, finding the derivatives of the
arc-co-functions is much easier. Since trig functions and their co-functions are π/2 radians
different on the x-axis, arc-trig functions will be π/2 radians different on the y-axis.
π
arccxx(x) = − arcxxx(x).
2
Using our constant and sum and difference derivative rules, we can derive that the derivatives
of the arc-co-functions are simply the opposite of the derivatives of their corresponding arc-
functions.

37
 d −1 d −1 d −1  
arccos x = √ , −1 < x < 1 arccot x = arccsc x =  √ , x > 1.
dx 1 − x2 dx 1 + x2 dx  x 1 − x2

2.8 Physical Interpretations of the Derivative

As we’ve seen, the idea derivative is fundamentally about continuous change. This idea
makes the derivative very useful for describing physical situations.
Example. Find the rate of change of the area of a circle with respect to its radius in meters.
Find the rate of change of the volume of a sphere with respect to its radius in meters. What
are these quantities (with appropriate units) when r = 5m?
Starting with the area of a circle,
dA
A = πr2 = 2πr.
dr
You might recognize this as the formula for the circumference of a circle.

Starting with the volume of a sphere,

4
V = πr3
3
dV
= 4πr2 .
dr
You might recognize this as the formula for the surface area of a sphere.

When r = 5m,

dA 
= 2π (5m) = 10πm.
dr r=5m

dV 
= 4π (5m)2 = 100πm2 .
dr  r=5m

2.8.1 Displacement, Velocity, and Acceleration

If we have an object whose position is determined by a single variable, like time t, then we
can model the position as a function.

s = f (t).

The displacement over some interval of length ∆t, would be

∆s = f (t + ∆t) − f (t).

38
The average velocity over this interval would be
f (t + ∆t) − f (t)
v̄ = .
∆t
As ∆t approaches 0, we see that v̄ is exactly the definition of the derivative of f with respect
to t: the instantaneous velocity.
ds f (t + ∆t) − f (t)
v(t) = = lim .
dt ∆t→0 ∆t
We can go through the same steps to derive that instantaneous acceleration is the derivative
of instantaneous velocity with respect to t.
dv d2 s v(t + ∆t) − v(t)
a(t) = = 2 = lim .
dt dt ∆t→0 ∆t
Although in everyday language we might use the terms speed and velocity interchangeably,
speed is defined as the absolute value of velocity, meaning it is a scalar quantity while velocity
is a vector quantity.  
   ds 
Speed = v(t) = .
 
dt
Modeling the motion of free-falling bodies was one of the earliest motivations for discovering
calculus. Through experiments and applications of physics theory, we know that the height
of a falling body when dropped from initial height h0 meters is modeled by
1
s(t) = h0 − gt2 ,
2
where t is the time in seconds since the object was released and g = 9.81m/s2 is the accel-
eration due to gravity near Earth.
Example. A ball is dropped from an initial height of 100 meters. How long does it take for
the ball to hit the ground? What speed is the ball traveling when it hits the ground?
In this case, h0 = 100m. So,
1
s(t) = 100m − gt2 .
2
We want to solve for t where s(t) = 0.
r
200m
t= ≈ 4.515s.
g
We take take the derivative of s with respect to t and then take the absolute value to get
the speed.
d
v(t) = s(t) = −gt.
dt 
Speed(t) = v(t) = gt.
Plugging in the time we got when the ball hits the ground,
SpeedGround ≈ g(4.515s) ≈ 44.294m/s.

39
Chapter 3

Applications of the Derivative

3.1 Extreme Values

Extreme values of a function are the minimum and maximum values it attains on an interval.
The absolute extreme values would the the extreme values accross the function’s entire
domain.
Example. Find the extreme values of x2 over the following intervals.
1. (−∞, ∞)

2. [0, 2]

3. (0, 2]

4. (0, 2)
1. For the max, there is no maximum value because we can keep increasing x to get a
larger output. The min is 0 when x = 0.
2. The max is 4 when x = 2. The min is 0 when x = 0.
3. The max is 4 when x = 2. Although the function approaches 0 in the limit as x
approaches 0, there is no min because 0 itself is not a value x2 can take on the
interval.
4. Although the function approaches 4 as x approaches 2, there is no max because 4
itself is not a value x2 can take on the interval. Like in the previous question, there
is no min.
We see that a function can fail to have a max or min value, but this cannot happen on a
finite, closed interval.
Theorem (Extreme Value Theorem). If f is continuous on some finite, closed interval [a, b],
then f must have both a minimum and maximum value on the interval.
We can find these extreme values by following these steps.

40
 1. Find any relative/local minima and maxima.

2. Find the function values for these local minima and maxima.

3. Find the function values at the endpoints of the interval, a and b.

4. The smallest such function value will be the absolute minima, while the largest such
function value will be the absolute maxima.

Theorem. If a function f has a local extrema at a point c interior to its domain and f 0
exists at c, then
f 0 (c) = 0.

So, the only candidate values we need to check are the endpoints and where the derivative
is 0. Such points are called critical points.

Example. Find the absolute extrema of y = x3 + x2 − 8x + 5 on the interval [−3, 2].

1. We’ll take the derivative to find all the critical points. y 0 = 3x2 + 2x − 8, which is 0
when x = −2 and x = 4/3.
2. (−2)3 + (−2)2 − 8(−2) + 5 = 17 and (4/3)3 + (4/3)2 − 8(4/3) + 5 = −41/27.
3. (−3)3 + (−3)2 − 8(−3) + 5 = 11 and (2)3 + (2)2 − 8(2) + 5 = 1.
4. Of these values, (−2, 17) is the absolute maxima and (4/3, −41/27) is the absolute
minima.

3.2 Mean Value Theorem for Derivatives

Theorem (Mean Value Theorem for Derivatives). If f is continuous on the interval [a, b]
and differentiable on the interval (a, b), then there exists at least one point in (a, b) such that

f (b) − f (a)
f 0 (c) = .
b−a
That is, there’s at least one point where the instantaneous rate of change and average rate
of change are equal. Another way of visualizing this is that there’s al least one point where
the tangent and secant lines are parallel.

41
 Figure 3.1: Wikipedia - Mean Value Theorem

Example. A trucker drives 150 miles of a route in 2 hours. The speed limit along the route
is 65 miles per hour. Show that at at least one point, the trucker must have been speeding.
We can model the trucker’s position along the route s as a function of time t where s(0) = 0
and s(2) = 150. We know that velocity is the derivative of position, so v(t) = s0 (t). It’s
reasonable to assume that s is differentiable on the interval [0, 2]. So, by the Mean Value
Theorem, there must exist a point c where
s(2) − s(0) 150 − 0
v(t) = = = 75mph.
2−0 2
At this point, the trucker was exceeding the speed limit of 65 miles per hour.
Definition. Let f be defined on an interval I. Let a and b be any two different points in I.

f increases on I if a < b =⇒ f (a) < f (b).

f decreases on I if a < b =⇒ f (a) > f (b).

Corollary. Let f be continuous of [a, b] and differentiable on (a, b).

If f 0 > 0 at every point on (a, b) then f increases on [a, b].

If f 0 < 0 at every point on (a, b) then f decreases on [a, b].

This should make sense given our theorem about local extrema. If f could still increase/decrease
while its derivative was negative/positive, then we couldn’t be sure that f is at a local max-
ima/minima when f 0 = 0.
Corollary. If f 0 (x) = 0 at all points in an interval I, then there is some constant C such
that f (x) = C for all points in I.

42
This follows from the Mean Value Theorem. Since f 0 = 0, the numerator in the Mean Value
Theorem, f (b) − f (a), must also be 0, meaning f (b) = f (a) = C.
Corollary. If f 0 (x) = g 0 (x) at ever point in some interval I, then there is come constant C
such that f (x) = g(x) + C.
That is, functions with the same derivative differ by a constant. This should make sense
given our constant and sum and difference derivative rules. If we let h0 (x) = f 0 (x)−g 0 (x) = 0
and apply the previous corollary, we get C.
Definition. A function F (x) is the antiderivative of f (x) if F 0 (x) = f (x) for all points in
the domain of f .
As we saw in the previous corollary, a function will have infinitely many antiderivatives that
differ by a constant.

3.3 First and Second Derivative Tests

3.3.1 First Derivative Test
As we saw in previous sections, we can use the first derivative to find critical values, which
allow us to find extrema.
Theorem (First Derivative Test). Let f (x) be a continuous function. At a critical point c,
1. If f 0 changes sign from positive to negative (f 0 (x) > 0 for x < c and f 0 (x) < 0 for
x > c), then f has a local maximum at c.
2. If f 0 changes sign from negative to positive (f 0 (x) < 0 for x < c and f 0 (x) > 0 for
x > c), then f has a local minimum at c.
3. If f 0 does not change sign at c, then f does not have a local extrema at c.
4. At a left endpoint a, if (f 0 < 0 / f 0 > 0), then f has a local (maximum / minimum)
at a.
5. At a right endpoint b, if (f 0 < 0 / f 0 > 0), then f has a local (minimum / maximum)
at b.
Example. Find the local extrema of f (x) = x3 − 12x − 5. Identify any absolute extrema.
Taking the derivative,
f 0 (x) = 3x2 − 12 = 3(x + 2)(x − 2).
So, the critical values are x = −2 and x = 2. f 0 is negative between these values and
positive outside of them, so x = −2 is a local maximum, while x = 2 is a local minimum.
f (−2) = 11 and f (2) = −21.

43
 As x approaches ∞ (the right endpoint), f also approaches ∞, so there is no absolute
maximum. Similarly, as x approaches −∞ (the left endpoint), f also approaches −∞, so
there is no absolute minimum.

3.3.2 Second Derivative Test

The second derivative tells us how the derivative is changing. Whether the derivative is
increasing or decreasing describes the concavity.

Definition. On some open interval I, the graph of a twice differentiable function f (x) is

1. Concave up if f 00 > 0 on I.

2. Concave down if f 00 < 0 on I.

Concave up portions of a graph tend to look like valleys, while concave down portions tend to
look like hills. Unlike the first derivative, we can tell if a function is increasing or decreasing
using concavity. We can however say if the function is increasing or decreasing more or less
rapidly.

Figure 3.2: Paul’s Online Notes - The Shape of a Graph, Part II

The points where a function changes concavity are called ”inflection points”. At these
points, the function is increasing or decreasing most rapidly, depending on the sign of the
first derivative.
Rather than looking at the sign of f 0 around critical values, we can look at the concavity at
the critical point.

Theorem (Second Derivative Test). Let c be a critical value of f .

1. If f 00 (c) < 0, then c is a local maximum.

2. If f 00 (c) > 0, then c is a local minimum.

44
 3. If f 00 (c) = 0, then the test is inclusive.

In other words, if we’re at a critical point that’s turning into a valley, then the critical point
must be the top of a hill. If we’re at a critical point that’s turning into a hill, then the
critical point must be the bottom of a valley. This test is particularly useful because you
only need to know f 00 at c rather than an entire interval1 .

Example. Use the second derivative test to find the local extreme values of f (x) = x3 −
12x − 5.

Finding critical values,

f 0 (x) = 3x2 − 12 = 3(x + 2)(x − 2).

f 0 (x) = 0 at x = −2, x = 2.

Taking the second derivative,

f 00 (x) = 6x.
Evaluating the second derivative at the critical values,

f 00 (−2) = −12 and f 00 (2) = 12.

So, x = −2 is a local maximum and x = 2 is a local minimum.

3.4 Modeling & Optimization

Now that we know some ways to find extreme values, we can apply them to answer opti-
mization problems. The general steps needed to solve an optimization problem are:

1. Write an equation that represents what you’re trying to maximize/minimize. This is

called your primary equation.

2. Use additional information to eliminate excess variables.

3. Find extreme values.

4. Select the extreme values that fit the problem’s constraints. Make sure your answer is
what the problem is asking for.

Example. A farmer has 1000 linear feet of fence and wants to create a rectangular pasture.
The pasture borders a river, which doesn’t need a fence. What is the maximum area he can
enclose?

1
It also extends to higher dimensions much better than the first derivative test.

45
 1. Since the pasture is rectangular, we know the two lengths of fence perpendicular to
the river will be the same length, which we’ll call x. We’ll say the remaining side
parallel to the river has length y. So, the area enclosed is

A(x, y) = xy.

2. There is no reason for the farmer not to use all 1000 linear feet of fence, so we’d expect
the sum of the lengths of the 3 fenced sides to be equal to 1000 feet: 2x + y = 1000,
or y = 1000 − 2x. We can then substitute back into our primary equation to get it
in terms of just x.
A(y) = x(1000 − 2x) = 1000x − 2x2 .
3. We’ll do a first derivative test to find extreme values.

A0 (y) = 1000 − 4x
A0 (y) = 0 at x = 250
A0 (200) = 200 > 0
A0 (300) = −200 < 0.

4. So, x = 250 =⇒ y = 500 is a local maximum. This corresponds to an area of

A = 250 · 500 = 125000ft2 .
Example. What is the maximum area of a rectangle that has two vertices on the x − axis
and two vertices on the portion of the graph y = 8 − x2 where y > 0?
1. We can define any such trapeziod (which include all such rectangles) by the x values
of the vertices on the x axis. We’ll call these two values x1 and x2 . We’ll say
arbitrarily that x1 < x2 . So, the area enclosed is
f (x1 ) + f (x2 )
A(x1 , x2 ) = (x2 − x1 ) .
2
2. However, the problem specifically restricts us to a rectangle, not a trapezoid. So,
the heights of each side must be equal.

f (x1 ) = f (x2 )
8 − x21 = 8 − x22
x21 = x22
±x1 ± x2
−x1 = x2 because x1 6= x2 .

Substituting back into our primary equation,

f (x2 ) + f (−x2 )
A(x2 ) = (x2 − (−x2 )) = 2x2 f (x2 ) = 16x2 − 2x32 .
2
46
 3. We’ll do a second derivative test for fin extreme values.

A0 (x2 ) = 16 − 6x22
4
A0 (x2 ) = 0 at x = ± √
6
00
A (x2 ) = −12x2
√
 
00 4
A −√ =8 6>0
6
√
 
00 4
A √ = −8 6 < 0.
6
√ √
4. So, x2 = 4/ 6 is a local maximum, and x2 = −4/ √6 is a local minimum. The
problem asks for a maximum, so we select x2 = 4/ 6. The problem asks for a
maximum area, so
   3
4 4 128
Amax = 16 √ −2 √ = √ ≈ 17.419.
6 6 3 6

3.5 Related Rates

In related rates problems, we generally have two related functions and want to know and
answer questions about the rate of change of one function given that we know the rate of
change of the other. The same problem solving steps as in modeling and optimization apply,
but we’ll usually be taking derivatives using implicit differentiation with respect to some
variable like time.
Example. Let A be the area of a square with side length x. Assume that x varies with time.
How are dA dt
and dx
dt
related? At a certain instant, the sides are 3 feet and growing at a rate
of 2 feet per minute. How quickly is the area changing at this instant.
Starting with the area of the square and implicitly differentiating with respect to t,

A = x2
dA dx
= 2x .
dt dt
dx
When x = 3ft and dt
= 3ft/min,
dA
= 2(3ft)(3ft/min) = 12ft2 /min.
dt
Example. The top of a 13 foot ladder propped against a vertical wall begins falling towards
the ground at 12ft/s. When the top of the ladder is 5 feet off the ground, how quickly is the
bottom of the ladder moving away from the wall? How how is the angle between the ladder
and the ground changing?

47
 Let h be the height of the top of the ladder of the ground. Then dh dt
= −12ft/s. Let b
the the distance from the base of the ladder to the wall. We can relate b and h using the
Pythagorean Theorem, where the 13-foot long ladder is the hypotenuse.
b2 + h2 = 132 .
We can also use this relationship to see that when h = 5ft, b = 12ft. Implicitly differen-
tiating,
db dh
2b + 2h = 0.
dt dt
Plugging in what we know and solving for db dt
,
db
2(12ft) + 2(5ft)(−12ft/s) = 0
dt
db
24ft = 120ft2 /s
dt
db
= 5ft/s.
dt
So, the base of the ladder is moving away from the wall at a rate of 5ft/s. Let θ be the
angle between the ladder and the ground. We can use sin to relate θ to b.
13 sin θ = b.
Implicitly differentiating,
dθ db
13ft cos (θ)
= .
dt dt
When cos is adjacent divided by hypotenuse, so cos θ = 12/13. Plugging in what we know
and solving for dθ
dt
,
dθ
13ft(12/13) = −12ft/s
dt
dθ
= −1/s.
dt
So, the angle between the ladder and ground is decreasing at at rate of 1 rad/s.
Example. Grain is is poured at a rate of 10ft3 /min and falls into a cone-shaped pile whose
bottom radius is half its altitude. How fast will the circumference of the base be increasing
when the pile is 8 ft tall?
Let h the the altitude of the cone. At the instant we care about h = 8ft. Let r be the
bottom radius of the cone. At the instant we care about, r = h/2 = 4ft. Let V be the
volume of the cone. We know that dV dt
= 10ft3 /s. We can relate these three quantites
using the formula for the volume of a cone.
1
V = πr2 h.
3
48
 Since we know that 2r = h, we can simplify to get rid of h.
2
V = πr3
3
Implicitly differentiating,
dV dr
= 2πr2 .
dt dt
We know the formula for the circumference C of the circular base.

C = 2πr.

Implicitly differentiating,
dC dr
= 2π .
dt dt
dV
We can substitute into our equation involving dt .

dV dC
= r2 .
dt dt
Plugging in what we know,
dC
10ft3 /s = (4ft)2
dt
dC 5
= ft/s.
dt 8

3.6 Linearization & Newton’s Method

3.6.1 Linearization
As we’ve seen, tangent lines intersect their function at most once: at the point of tangency.
However, we know that differentiable functions are locally linear, so we’d expect the tangent
line to be a decent approximation of the function near the point of tangency.

Definition. If f is differentiable at a, then the approximating function

L(x) = f 0 (a)(x − a) + f (a)

is the linearization of f at a.

Example. Find the linearization of f (x) = ln (x + 1) at x = 0. How accurate is this

approximation at x = 0.1?

49
 Following the definition,

f (0) = ln (0 + 1) = 0
1
f 0 (x) =
x+1
1
f 0 (0) = =1
0+1
L(x) = 1(x − 0) + 0 = x.

Calculating the error at x = 0.1,

L(0.1) = 0.1
f (0.1) ≈ .0953
 
L(0.1) − f (0.1)
% error = 100% ≈ 4.7%.
L(0.1)

So, we can see the linear approximation is pretty good.

We call the difference in x between the point of tangency and the point we’re trying to
approximate the differential.

Definition. Let y = f (x) be a differentiable function. The differential dx is an independent

variable. The differential dy is dy = f 0 (x)dx.

dy is the approximated change in y expected by the linearization for some given change in
x, dx.
2x
Example. Find dy for y = 1+x2
, x = −2, and dx = 0.1.

−4x2 2
y0 = 2 +
2
(1 + x ) 1 + x2
y 0 (−2) = −6/25
dy = (−6/25)(0.1) = −0.024.

3.6.2 Newton’s Method

We can use the fact that the tangent line approximates the function to find the zeroes of
functions. Starting with an initial guess x0 for the x value of the zero, we look at the the

50
 tangent line at x0 and find where it intersects the x − axis.

L0 (x) = f 0 (x0 )(x − x0 ) + f (x0 )

0 = f 0 (x0 )(x − x0 ) + f (x0 )
−f 0 (x0 )(x − x0 ) = f (x0 )
f (x0 )
x − x0 = − 0
f (x0 )
f (x0 )
x = x0 − 0 .
f (x0 )

This x value serves as our next guess for the zero. We repeat this process, until we find the
zero or are satisfied with our error2 . This yields a recursive formula

f (xn )
xn+1 = xn − .
f 0 (xn )

2
For most well-behaved functions, Newton’s Method can get within a small margin of error or a zero relatively
quickly. There is also a generalized, sometimes faster version of Newton’s Method that approximates the
function with higher-order polynomials than just lines.

51
Chapter 4

Integrals

4.1 Estimating with Sums

If we know how something changes, we can use sums to estimate, or sometimes even know
exactly the net change.

Example. A train moves at 80 miles per hour for 3 hours. How far does it travel?

This is the kind of simple problem you might see in a physics class.
80mi
3hr = 80 · 3mi = 240mi.
hr
However, if we take a look at a graph of this situation, with speed in miles per hour on
the y axis and time in hours on the x axis, we see that the area underneath the curve
from x = 0hr to x = 3hr is exactly our answer of 240 miles. This is not a coincidence, as
what we effectively did mathematically is find the area of this rectangle.

Figure 4.1: Our answer is the area under the curve.

52
The same idea of finding the area under the curve works not just for constant speeds.

Example. A particle moves at velocity v(t) = 3t + 3 meters per second for time t ≥ 0. What
is the position of the particle at t = 2 seconds?

Figure 4.2: Our answer is still the area under the curve.

1
A = h(b1 + b2 )
2
1
= (2s − 0s)(v(0) + v(2))
2
1
= (2s)(3m/s + 9m/s)
2
= 12m.

Left Endpoint Approximation

In fact, the area under the curve even works for more complicated, non-linear curves like
v(t) = t2 + 1. We just need a way to find the area underneath these curves. One idea
that was used to find areas as far back as Archimedes was to estimate the complex shape
using easier shapes like rectangles. The narrower the width of each rectangle, the better the
estimate becomes.

Figure 4.3: Left Endpoint: Block widths of 1/3, 1/10, and 1/100.

Although above we’ve used a left endpoint approximation, we could have also used the right
endpoint or midpoint, which might give better approximations for certain types of curves.
No matter the approximation type, the estimate tends to get closer to the true area as the

53
 rectangle width is decreased. The formulas for estimating the area of f from a to b with n
rectangles are
n−1
X
Aleft = f (a + k∆x) ∆x
k=0
n−1
X
Aright = f (a + (k + 1)∆x) ∆x
k=0
n−1  
X 2k + 1
Amid = f a+ ∆x ∆x
k=0
2
b−a
where ∆x = .
n

Trapezoidal Rule
Another common shape to use rather than rectangles is trapezoids. Trapezoids allow us to
find the exact area of functions made of straight lines1 and like the left endpoint approxi-
mation, get better the narrower the width of each trapezoid. Starting from the formula, we
can make some simplifications.
1
A = (b1 + b2 )h
2
Now summing each of these trapezoids’ areas to approximate our function f from a to b,
n−1
X 1
Atrap = (f (a + i∆a) + f (a + (i + 1)∆x)) ∆x
k=0
2
n−1
∆x X
= f (a + i∆x) + f (a + (i + 1)∆x)
2 k=0
∆x
= ((f (a) + f (a + ∆x)) + (f (a + ∆x) + f (a + 2∆x)) + . . . + (f (a + (n − 1)∆x) + f (a + n∆x)))
2
∆x
= (f (a) + 2f (a + ∆x) + 2f (a + 2∆x) + . . . + 2f (a + (n − 1)∆x) + f (a + n∆x))
2
Aleft + Aright
= .
2
The trapezoidal rule overestimates areas when the curve is concave up and underestimates
when the curve is concave down.

1
There are higher-order, more accurate estimations than the trapezoidal rule. The most common is Simpson’s
Rule. A = 2M3+T where M is the midpoint formula area and T is the trapezoidal rule area. It can exactly
give the area of quadratics because it corresponds to estimating areas using parabolas that intersect the
curve at the left, middle, and right of each ”strip”.

54
 4.2 Definite Integrals & Antiderivatives
4.2.1 Definition of a Definite Integral
We saw several approximations for the areas under any type of curve. We also saw how these
approximations get better the narrower our “strips”. In the limit, we get exactly the area
under the curve, which defines the definite integral.
Definition. Let f be continuous on the closed interval [a, b]. Let this interval be partitioned
into n equal2 sub-intervals, each of length ∆x = b−a
n
. The definite integral of f over [a, b] is
given by
Z b n−1
X
f (x)dx = lim f (ck )∆x
a n→∞
k=0
where ck is any arbitrary value in the kth sub-interval. This particular type of limit of a sum
is called a Riemann Sum.
Note that the left, right, and midpoint approximations are all different ways of choosing ci ,
all of which in the limit give the area under the curve and definite integral from a to b.
Example. Find the value of the following definite integral using Riemann sums.
Z 1
x2 dx.
0
We’ll use a Riemann sum with a left-endpoint approximation.
Z 1 n−1  
2
X k 1−0
x dx = lim f 0+
0 n→∞
k=0
n n
n−1  2
X k 1
= lim
n→∞
k=0
n n
n−1 2
X k
= lim
n3
n→∞
k=0
 
1 n(n − 1)(2n − 1)
= lim 3
n→∞ n 6
 
1 3 1
= lim 2− + 2
n→∞ 6 n n
1
= .
3
Note that area below the x-axis is counted as negative area. In general
Z b
f (x)dx = area above x-axis − area below y-axis.
a
2
Technically, the intervals don’t need to be of equal size, as long as the width of all intervals goes to 0 in the
limit.

55
4.2.2 Basic Properties of Definite Integrals
The following are properties of definite integrals. Many should look familiar from properties
of limits and derivatives. Let f and g be continuous functions of x. Let a, b, and c be real
constants where a 6= b. Let f[a,b] be the values of f on the interval [a, b].
Z b Z a
Order of Integration Rule: f (x)dx = − f (x)dx
a b
Z b Z b Z b
Sum and Difference Rule: (f (x) ± g(x))dx = f (x)dx ± g(x)dx
a a a
Z a
Zero Rule: f (x)dx = 0
a
Z b Z b
Constant Multiple Rule: cf (x)dx = c f (x)dx
a a
Z b Z c Z c
Additivity Rule: f (x)dx + f (x)dx = f (x)dx
a b a
Z b
Max-Min Rule: (b − a) min f[a,b] ≤ f (x)dx ≤ (b − a) max f[a,b]
a
Z b Z b
Domination Rule: min f[a,b] ≥ max g[a,b] =⇒ f (x)dx ≥ g(x)dx.
a a

Mean Value Theorem for Definite Integrals

The mean value of f over the interval [a, b] is given by
Z b
1
f (x)dx.
b−a a

You can effectively think of this as the definition of the average: adding up all the values
and dividing by the number of values. This is exactly what happens if you work through
the Riemann sums. Much like the Mean Value Theorem for Derivatives, since the function
is continuous, it will take on its average value somewhere in the interval.

Theorem (Mean Value Theorem for Definite Integrals). If f is continuous on the interval
[a, b], then there is some point c in the interval such that
Z b
1
f (c) = f (x)dx.
b−a a

56
 Figure 4.4: A continuous function always takes on its mean value

Example. A plane’s airspeed is given by v(t) = t/8, t ≥ 0. What what time t does the Mean
Value Theorem guarantee that the plane’s speed was 1? When is this time?
Rather than go through the trouble of evaluating a limit to find the definite integral, we
can notice that the graph of the plane’s airspeed is a triangle, which we know how to find
the area of geometrically.
1
Area(t) = tv(t).
2
Dividing by the length of the interval to get the average airspeed,
1 1 1
Avg(t) = tv(t) = v(t).
t−0 2 2
Solving for t when Avg(t) = 1,
1 t
= 1 =⇒ t = 16.
2 8
So, at time t = 16, the Mean Value Theorem tells us that the plane’s speed was 1. Looking
at the graph, we can see that at time t = 8 the plane’s speed was indeed 1.

4.2.3 Fundamental Theorem of Calculus

Although it’s nice to be able to evaluate definite integrals for numerical bounds, it’d be more
convenient if we only had to do the work of integrating once and could then have a function
that would tell us the area like so:
Z x
F (x) = f (t)dt.
a

57
Let’s try taking the derivative of F using the limit definition.
F (x + ∆x) − F (x)
F 0 (x) = lim
∆x→0 ∆x
R x+∆x Rx
a
f (t)dt − a f (t)dt
= lim
∆x→0 ∆x
R x+∆x
f (t)dt
= lim x (by Integral Rules)
∆x→0 ∆x
f (k)∆x
= lim , x ≤ k ≤ x + ∆x (by Mean Value Theorem)
∆x→0 ∆x
= lim f (k)
∆x→0
= f (x) (by Sandwich Theorem).
Taking the derivative seems to undo the integration. The same fact applies if we take the
integral of a derivative. Starting with what we’ve just shown,
Z x
d
f (t)dt = f (x).
dx 0
d
Let g(x) = dx
f (x). Z x
d d
g(t)dt = g = f (x).
dx 0 dx
Since the derivatives are equal, we know the functions must differ by at most a constant C.
Z x
g(t)dt = f (x) + C
0
Z x 
d
f (t) dt = f (x) + C.
0 dx
So, the integral of the derivative gets us a function that differs from the original by at most
a constant.
This idea that the integral and derivative undo each other is captured by the Fundamental
Theorem of Calculus.
Theorem (Fundamental Theorem of Calculus). Let f be a continuous function on the in-
terval [a, b]. Then Z x
F (x) = f (t)dt
a
has a derivative at every point in [a, b], and
Z x
dF d
= f (t)dt = f (x).
dx dx a
Further, if F is the antiderivative of f on [a, b], then
Z b
f (x)dx = F (b) − F (a).
a

58
This last equation is especially useful for calculating definite integrals. Rather than evalu-
ating a limit of a Riemann sum, if we know the antiderivative, we can just evaluate at two
points.
Example. Evaluate the following definite integral using antiderivatives.
Z 1
dx
2
.
0 1+x
1
We previously derived that the derivative of arctan x is 1+x 2. So, arctan x is the an-
1
tiderivative of 1+x2 . Applying the Fundamental Theorem of Calculus (FTC)
Z 1
dx
2
= arctan 1 − arctan 0
0 1+x
π
= −0
4
π
= .
4

4.3 Indefinite Integrals & Ways to Solve Them

4.3.1 Definition of an Indefinite Integral
We’ve seen in the previous section from the Fundamental Theorem of Calculus how we can
let the upper bound of a definite integral be a variable, giving us a new function that is
an antiderivative of the function being integrated. Anti-differentiation is a useful operation,
so notationally we drop the bounds from the definite integral, which we call an indefinite
integral.
Definition. The indefinite integral of some continuous function f is
Z Z x
f (x)dx = f (t)dt = F (x) − F (a) = F (x) + C,
a

where F (x) is the antiderivative of f .

Rather than choose a particular value for C, we just write “+C” when integrating to capture
all antiderivatives. Note that since we defined an indefinite integral in terms of a definite
integral, all the basic properties of definite integrals still apply.
Example. Solve the following indefinite integral:
Z
cos xdx.

We derived previously that the derivative of sin x is cos x. So, for any real constant C,
sin x + C is the antiderivative of cos x.
Z
cos xdx = sin x + C.

59
4.3.2 Indefinite Integrals Properties & Strategies
Knowing that the indefinite integral give us the antiderivative, we can simply apply the rules
we derived for derivatives in reverse.

Power Rule & Integral of 1/x

Lemma. Let n 6= −1 be a real number. Then

xn+1
Z
xn dx = + C.
n+1
n = −1 has to be a special case, otherwise the denominator would be 0.

We derived earlier that that the derivative of ln x was 1/x. However, we have to be a little
more careful before we write the integral of 1/x. 1/x is defined for all real values except 0, but
ln x is only defined for positive values. So, we need a different antiderivative, which would  
have to be equal to ln x for positive x, that’s defined everywhere that 1/x is defined. ln x
works nicely because it’s defined everywhere that 1/x is defined, it’s the antiderivative for
positive x, and its slope for negative values of x corresponds exactly to how 1/x is negative
for negative values of x.

Lemma. Z
1  
dx = ln x + C.
x

Exponential Rule
Since ex is its own derivative, it’s also its own antiderivative. A similar idea applies for any
base.

Lemma.
bx
Z
bx dx = + C.
ln b

4.3.3 u Substitution (Reversing the Chain Rule)

d
The chain rule tells us that dx f (g(x)) = f 0 (g(x))g 0 (x). So, if we see that we are integrating a
function that contains within it a function and its derivative, we might want to try running
the chain rule in reverse.

Example. Solve the following indefinite integral:

Z
x2 · 2xdx.

60
 2
Since we know that dxdx
= 2x, or in differential form dx2 = 2xdx, we can make a substi-
tution. Letting u = x2 ,
Z Z
2
x · 2xdx = udu
u2
= +C
2
x4
= + C.
2
Sometimes the derivative itself won’t be there, but a constant multiple of the derivative.
Example. Solve the following indefinite integral:
x2
Z
dx.
1 − x3
A good way to spot that there might be the possibility of a u-substitution here is the fact
that the denominator is a polynomial of one higher degree than the numerator. So, if we
took the derivative of the denominator, we’d get a polynomial of the same degree as the
numerator. Let u = 1 − x3 . Then du = −3x2 dx; x2 dx = −du/3.
x2 −du
Z Z
3
dx =
1−x u
 
= − ln u + C

 
= − ln 1 − x3  + C.
We can also apply u-substitutions to definite integrals. We’ll need to change the bounds of
integration to be in terms of u before we evaluate.
Example. Solve the following definite integral:
Z 1 √
15x2 5x3 + 4dx.
0

Let u = 5x + 4. Then du = 15x2 dx. When x = 1, u = 5(0)3 + 4 = 4; when x = 1,

3

u = 5(1)3 + 4 = 9.
Z 1
2
√ Z 9
√
3
15x 5x + 4dx = udu
0 4
√ 9
2u u 
=
3 4
√ √
2(9) 9 2(4) 4
= −
3 3
54 16
= −
3 3
38
= .
3
61
Trig Functions
We can get the integrals of some of the trig functions by simply applying the derivatives for
trig functions in reverse.
Z Z
sin xdx = − cos x + C cos xdx = sin x + C

We can get the integrals of tan x and cot x using basic u-substitutions.
Z Z
sin x
tan xdx = dx
cos x
u = cos x and du = − sin xdx
−du
Z
=
u
 
= − ln u + C

 
= − ln  cos x + C.

Z Z
cos x
cot xdx = dx
sin x
u = sin x and du = cos xdx
Z
du
=
u
 
= ln u + C

 
= ln  sin x + C.

We can also get the integrals of sec x and csc x using u-substitutions, but we have to be a
bit more clever and multiply the integrand by a strange-looking fraction that equals 1.
Z Z
sec x + tan x
sec xdx = sec x dx
tan x + sec x
sec2 x + sec x tan x
Z
=
tan x + sec x
u = tan x + sec x and du = (sec2 x + sec x tan x)dx
Z
du
=
u
 
= ln u + C

 
= ln  tan x + sec x + C.

62
 Z Z
csc x + cot x
csc xdx = csc x dx
cot x + csc x
csc2 x + csc x cot x
Z
= dx
cot x + csc x
u = cot x + csc x and du = −(csc2 x + csc x cot x)dx
−du
Z
=
u
 
= − ln u + C

 
= − ln  cot x + csc x + C.

4.3.4 Integration by Parts (Reversing the Product Rule)

When a u-substitution won’t work because the integrand is the product of two functions
that are not related by a derivative, integration by parts may be able to help us. Starting
from the product rule3 ,
d dv du
(uv) = u +v
Z dx Z dx dx Z
d dv du
(uv)dx = u dx + v dx
dx dx dx
Z Z Z
d(uv) = udv + vdu
Z Z
uv = udv + vdu
Z Z
udv = uv − vdu.

Essentially, if we are integrating a function that is a product of something we can differentiate

(u) and something we can integrate (dv), then we can do integration by parts.

Example. Solve the following indefinite integral:

Z
x sin xdx.

Since x and sin x don’t have any parts that are related by a derivative, a u-substitution
wouldn’t be helpful. However, we know how to differentiate x and integrate sin x, so

3
This justification isn’t completely rigorous because you can’t always treat differentials like fractions, but the
overall result is correct.

63
 integration by parts is a good strategy to try.
u = x and du = dx
dv = sin xdx and v = 4 − cos x
Z Z
x sin xdx = −x cos x − − cos xdx

= −x cos x + sin x + C.

LIPET (How to choose u)

The tricky part of integration by parts is choosing what should be u and what should be
R If in the previous
dv.
x2
R xexample
2
we had instead chosen u = sin x, we would have gotten that
x sin xdx = 2 − 2 cos xdx, which although true, doesn’t give us an easier function to
integrate. Generally, we want to choose u such that du is a ”simplier” function to integrate.
This can mean a lower-degree polynomial, or getting rid of functions like ln x and replacing
them with polynomials. You can remember the acronym LIPET as a guide to help you.
Logarithms
Inverse Trig Functions
Polynomials
Exponentials
Trig Functions
Example. Solve the following indefinite integral:
Z
x3 ln xdx.

Following LIPET, we select u = ln x.

dx
u = ln x and du =
x
3 4
dv = x dx and v = x /4
x4 ln x
Z Z 3
3 x
x ln xdx = − dx
4 4
x4 ln x x4
= − + C.
4 16
Sometimes, we might have to do integration by parts multiple times. It might seem like
integration by parts is leading us in circles, but this can actually be a good thing, allowing
us to solve the integral if we’re observant.
4
We don’t include the +C when integrating dv. We’ll still include one at the end.

64
 Example. Solve the following indefinite integral:
Z
e−x cos xdx.

Following LIPET, u = e−x .

u = e−x and du = −e−x dx

dv = cos xdx and v = sin x
Z Z
e cos xdx = e sin x + e−x sin xdx.
−x −x

Following LIPET, u = e−x again.

u = e−x and du = −e−x dx

dv = sin xdx and v = − cos x
Z Z
e sin x + e cos xdx = e sin x − e cos x − e−x cos xdx
−x −x −x −x

It might seem like we’ve hit a dead end, since we’re back trying to integrate the exact
same function we started with. However, if we rearrange our equations a little bit, we’ll
see that we’re actually really close to a solution.
Z Z
e cos xdx = e sin x − e cos x − e−x cos xdx
−x −x −x

Z
2 e−x cos xdx = 5 e−x sin x − e−x cos x + C
e−x sin x − e−x cos x
Z
e−x cos xdx = + C.
2

5
We need to remember to include a +C. Usually it would come from completely solving an integral, but here
we just did some rearranging.

65
Inverse Trig Functions
We can use tabular integration to integrate the inverse trig functions. We’ll also end up ap-
plying two different strategies: Integration by parts and u-substitution to solve one integral.
1
u = arcsin x and du = √
1 − x2
dv = dx and v = x
Z Z
x
arcsin xdx = x arcsin x − √ dx
1 − x2
−1
u = 1 − x2 and xdx = du (this is a different u than before)
2 Z
1 du
= x arcsin x + √
2 u
√ 
1 u
= x arcsin x + +C
2 1/2
√
= x arcsin x + u + C
√
= x arcsin x + 1 − x2 + C.

We can do the exact same thing to find the integral of arccos x.

Z √
arccos xdx = x arccos x − 1 − x2 + C.

The integration by parts into u-substitution idea even works for arctan x.
1
u = arctan x and du =
1 + x2
dv = dx and v = x
Z Z
x
arctan xdx = x arctan x − dx
1 + x2
1
u = 1 + x2 and xdx = du (this is a different u than before)
2 Z
1 du
= x arctan x −
2 u
1  
= x arctan x − ln u + C
2
1  
= x arctan x − ln 1 + x2  + C.
2
We can do the same thing to find the integral of arccot x.
Z
1  
arccot xdx = x arccot x + ln 1 + x2  + C.
2

66
The integrals of arcsec x and arccsc x involve a special type of function you probably haven’t
seen before, arccoshx. However, if you’re familiar with this function and it’s derivative, you
can still apply integration by parts to find their antiderivatives.
Z
arcsec xdx = x arcsec x − arccoshx + C
Z
arccsc xdx = x arccsc x + arccoshx + C.

Tabular Integration
If we’re trying to integrate a function like x4 e−x using integration by parts, we quickly realize
that we’ll need to perform integration by parts 4 times, choosing the same value for dv every
time. Since this would be rather tedious and repetitive, we can extract the essential elements
of integration by parts into a table and perform integration by parts many times without
needing to repeat ourselves.
In integration by parts, we essentially choose one function (u) to continually differentiate
and another (dv) to continually integrate. We then alternate signs with each iteration of
integration by parts, due to negative signs canceling each other out. So, in our table, we’ll
have three columns, one where we’ll continually differentiate a function, usually until its
a constant, a column to keep track of the sign, and a final column that we’ll continually
integrate. Then, all we’ll have to do is multiply across columns and sum across rows to get
our result from repeated integration by parts.

Example. Solve the following indefinite integral:

Z
x4 e−x dx.

Following LIPET, x4 is our function to differentiate and e−x is our function to integrate.
Setting up our table,

67
Multiplying across each arrow and summing,
Z
x4 e−x dx = x4 −e−x + 4x3 (−1)e−x + 12x2 −e−x + 24x(−1)e−x + 24 −e−x + C

= −x4 e−x − 4x3 e−x − 12x2 e−x − 24xe−x − 24e−x + C

= −e−x x4 + 4x3 + 12x2 + 24x + 24 + C.

68
 Chapter 5

Applications of Integrals

5.1 Physics
5.1.1 Position, Velocity, & Acceleration
Since we know by the FTC that integration is the opposite of differentiation, we can also
interpret integrals in a similar physical sense as derivatives.
Z Z
v(t)dt = x(t) + C a(t)dt = v(t)

Example. A particle starts at t = 0 with an initial velocity of 5m/s and accelerates for 8
seconds. It’s acceleration is given by a(t) = 2.4t m/s2 . What is the particle’s velocity after
the 8 seconds pass? What is the particle’s displacement after the 8 seconds pass?
We can integrate acceleration to get velocity.
Z
v(t) = 2.4tdt = 1.2t2 + C.

We know that v(0) = 5m/s so we can solve1 for C.

1.2(0) + C = 5
C=5
2
v(t) = 1.2t + 5.
So, at t = 8, v(8) = 1.2(8)2 + 5 = 81.8m/s. We can now integrate velocity to get
displacement Z
x(t) = 1.2t2 + 5dt = 0.4t3 + 5t + C.

1
When we solve for C like this, we’re solving what’s called an ”inital value problem,” which we’ll do more of
when talking about differential equations.

69
 We know that x(0) = 0m, so C = 0.

x(t) = 0.4t3 + 5t.

So, at t = 8, x(t) = 0.4(8)3 + 5(8) = 244.8m. We could have also tackled this problem
with definite integrals.
Z 8
∆x = v(t)dt = 244.8m =⇒ Net Displacement = x0 + ∆x = 244.8m
0
Z 8
∆v = a(t)dt = 76.8m/s =⇒ Net Velocity = v0 + ∆v = 81.8m/s
0

5.1.2 Work
Work is defined as
W = Fd
where F is force and d is displacement. The force applied by stretching or compressing a
spring beyond its natural length is given by Hooke’s Law

F = kx

where k is some spring constant and x is the displacement beyond the spring’s natural length.
We can apply ideas as integrals representing net change to find the work needed to compress
or stretch a spring.
Example. It takes 10N of force to stretch a spring 2m beyond its natural length. How much
work is done stretching the spring 4m beyond its natural length?
We can use the first bit of information to get the spring constant.

F = kx
10N = k(2m)
k = 5N/m.

So, F (x) = 5xN. If we stretch the spring by ∆x, the work done over this interval is
approximately ∆W = F (x)∆x = 5x∆x. In the limit, dW = 5xdx. Integrating both sides
from x = 0 to x = 4, Z 4 Z 4
W = dW = 5xdx = 40Nm.
0 0
We can even bring in other concepts to these problems, like related rates.
Example. An inverted conical tank with a height of 10ft and a base radius of 5ft is filled to
within 2ft of the top with a liquid with a density of 57lbs/ft3 . How much work does it take
to fill the remaining 2ft of the tank with liquid, assuming you only have to pump the liquid
to the current liquid level in the tank?

70
 Let V be the volume of the tank and x the height of the liquid. Imagine we pump in
some liquid that changes the height of the liquid in the tank by ∆x. Then
∆V = πr2 ∆x
dV = πr2 dx.
Since the height of the tank is 10ft and the base radius 5ft, the radius of the liquid level
will always be half the liquid depth.
r = x/2
dV = π(x/2)2 dx.
Since weight in pounds is already a unit of force, we can multiply dV by the density of
the liquid to get F .
F = 57π(x/2)2 dx.
The displacement of the liquid is the current height of the liquid x, getting us dW .
dW = 57πx(x/2)2 dx
Z 10
W = 57πx(x/2)2 dx
8
= 21033πft lbs.

5.2 Lengths of Curves

If we’re at some point (x, f (x)) and move dx units over, we’ll be at a new point (x+dx, f (x+
dx)). This point is dx units horizontally and dy = f (x + dx) − f (x)p vertically away from
(x, f (x)). So, by the Pythagorean Theorem, the new point is ds = (dx)2 + (dy)2 units
away.

Figure 5.1: Secant length approaches arc length

71
As dx approaches 0, ds, the length of the secant line, approaches the length of the curve.
If we summed all of these ds’s in some interval we’d have the length of the curve on that
interval.
Z b
s= ds
a
Z bp
= (dx)2 + (dy)2
a
v
Z bu  2 !
t(dx)2 1 + dy
u
=
a dx
s  2
Z b
dy
= 1+ dx.
a dx

Example. Show that the circumference of a circle with radius r is C = 2πr.

Starting with the equation of a circle of radius r and implicitly differentiating,

x2 + y 2 = r2
dy
2x + 2y =0
dx
dy −x
=
dx y
 2
dy x2 x2
= 2 = 2
dx y r − x2
Z r r
x2
C=2 1+ 2 dx
−r r − x2
(2 b/c we need to count upper and lower half)
Z rr
r2
=2 dx
−r r 2 − x2
Z r r
1
=2 r dx
−r r − x2
2
 x r
= 2r arcsin 
r −r
= 2πr.

Note that when deriving the arc length formula, we could have just as easily have divided
by (dy)2 . This would give us an equivalent arc length formula that’s applicable when x is a
function of y.
Example. Find the length of the curve y = x1/3 from (−8, 2) to (8, 2).

72
 Rather than tediously integrate the square root of a cube root if we set our bounds in
terms of x, we can rewrite our equation and set out bounds in terms of y.

x = y3
dx
= 3y 2
dy
 2
dx
= 9y 4
dy
Z 2p
s= 1 + 9y 4 dy
−2
2
≈ 17.261.

5.3 Areas in the Plane

Lemma. If f (x) ≥ g(x) on [a, b] then the area between f (x) and g(x) on [a, b] is given by
Z b
A= (f (x) − g(x)) dx.
a

You can easily visualize why this is true. We can break the integral into two parts: adding
the area under f and the other subtracting the area under g. The first integral will over
count the area between f and g, counting all area between f and the x-axis. The second
integral will subtract exactly the amount of area that is over counted: the area that is also
between g and the x-axis.

2
You shouldn’t be expected to evaluate this integral analytically. Using a calculator to get a numerical answer
is fine.

73
 Figure 5.2: Subtracting two areas to get the area between them

Even if both of the curves has more area below the x-axis, the same idea applies. f will
give a smaller-magnitude negative area, while g will give a larger magnitude negative area.
Subtracting a larger-magnitude negative number from a smaller magnitude negative number
will give a positive area.

Example. Find the area enclosed between the two curves y = 2 − x2 and y = −x.

These two curves intersect at x = −1 and x = 2. Applying the formula,

Z 2
A= ((2 − x2 ) − (−x))dx
−1
2
x3 x2 
= 2x − + 
3 2 −1
9
= .
2

5.3.1 Subregions
Sometimes it might be useful to break the enclosed regions into subregions and find the area
of each separately.
√
Example. Find the area above the x-axis, below y = x, and above y = x − 2.

74
 If we simply took the area between the two curves, we’d count some area we don’t want
3
beneath the x-axis
√ . Instead, we can break the area√into two subregions: one between
the x-axis and x and the other between x − 2 and x.

Figure 5.3: Break complicated areas into simpler subregions

Finding the area of the blue region,

2 √ 2 √
√
Z
2x x  4 2
A1 = xdx = = .
0 3 0 3

Finding the area of the green region,

4 √ 4 √
√ 2x x x2
Z
 10 4 2
A2 = ( x − (x − 2))dx = − + 2x = − .
2 3 2 2 3 3

Adding the areas of the two regions,

10
A = A1 + A2 = .
3

5.3.2 Integrating with Respect to y

Another strategy when finding the area of more complex regions is to see if they become
easier to deal with if we were to instead integrate with respect to y. In the above example,
it is indeed easier to work with respect to y because both bounding curves are between the
same y values.
√
Example. Find the area above the x-axis, below y = x, and above y = x − 2.

We need to rearrange our equations to be of the form x = . . . instead of y = . . . by simply

solving for x.
x = y 2 , y ≥ 0 and x = y + 2.

3
Although this extra region is just a triangle and not too hard to find the area of, you’d effectively be applying
the same strategy of two subregions, they’d just overlap. However, using geometry to your advantage is
certainly a valid approach.

75
 Since x = y + 2 is further from the x-axis, it becomes our top curve.
Z 2
A= ((y + 2) − y 2 )dy
0
2
y2 y 3 
= + 2y − 
2 3 0
10
= .
3

5.4 Volumes
5.4.1 Volumes from Cross Sections
Imagine we want to find the volume of some complex object. One way we could approximate
it is by slicing it into narrow cross sections. The volume of each cross section would roughly
be the area of the cross sectional face, times the width of the cross section.

Figure 5.4: Paul’s Online Notes - Area and Volume Formulas

Adding the volumes of these cross sections up, we’d get our approximation, which would get
better and better the narrower the width of each cross section. In the limit, this is exactly
the definition of an integral.

Definition. The volume of a solid with integrable corss sectional area A(x) from x = a to
x = b is given by Z b
V = A(x)dx.
a

Let’s start by finding the volume of a solid we already know.

Example. Find the volume of a cube with sidelength a.

76
 For any slice of a cube, the cross section is a square with sidelength a.

A(x) = a2
Z a
V = a2 dx
0
a
2 

= a x
0
3
=a .

Now let’s find the volume of a slightly more complicated shape.

Example. Find the volume of a square pyramid with base sidelength a and height h.

Like the cube, any slice of the square pyramid is a square. However, the sidelength of the
square depends on the height of your slice. A slice at the very tip of the pyramid would
have sidelength 0, while a slice at the very bottom of the pyramid would have sidelength
a. The sidelength grows linearly from x = 0 to x = h, so it must be ha x.
 a 2
A(x) = x
h
Z h 2
a 2
V = 2
x dx
0 h
h
a2 3 
= 2x 
3h 0
a2 h
=
3
1 2
= a h.
3

5.4.2 Solids of Revolution

You can think of solids of revolution as a special case of volumes from cross sections. We’ll
tend to be on the lookout for a function that defines the radius, and we can then use the
formula for the area of a circle to get the cross sectional area.

A(x) = πr2 (x).

Example. Find the volume of the cone formed by rotating the line y = x/3 about the x-axis
for 0 ≤ x ≤ 6.

The radius is simply the distance from the x-axis, which is just another name for y. So,
 x 2
A(x) = π .
3

77
 Integrating,
Z 6  x 2
V = π dx
0 3
Z 6
π
= x2 dx
9 0
6 !
π x3 
=
9 3 0
= 8π.

This is the volume of a cone with radius 2 and height 6: 31 π(2)2 (6) = 8π.
Let’s try a more complicated solid.

Example. Find the volume of the solid of revolution bounded by y = 2 + x cos x from
− π2 ≤ x ≤ π2 .

Again, the radius is simply the distance from the x-axis, which is y.

A(x) = π (2 + x cos x)2

= π 4 + 4x cos x + x2 cos2 x


Z π/2
π 4 + 4x cos x + x2 cos2 x dx


V =
−π/2
!
Z π/2 Z π/2 Z π/2
=π 4dx + 4x cos xdx + x2 cos2 xdx
−π/2 −π/2 −π/2
 
1 2


= π 4π + 0 + π π − 6 (use integration by parts)
24
π 4 15π 2
= + .
24 4

Washer Method
Imagine now we want to find the volume of a solid of revolution with a “hole.” If we were
to take a cross section of such a solid, it’d look like a washer with an outer radius R(x) and
inner radius r(x).

78
 Figure 5.5: She Loves Math - Applications of Integration: Area and Volume

We can think of this in a similar way as the area between curves in 2D. We’ll find the
volume swept by the outer radius and the subtract the volume swept and removed by the
inner radius. Z b
V =π (R2 (x) − r2 (x))dx.
a

Example. Find the volume of shape formed by revolving the area enclosed by the y-axis,
y = cos x, and y = sin x around the x-axis.

In the first quadrant, cos x ≥ sin x for x ≤ π4 , so our bounds are 0 ≤ x ≤ π4 , R(x) = cos x,
and r(x) = sin x.
Z π/4
V =π (cos2 x − sin2 x)dx
0
Z π/4
=π cos (2x)dx
0
 π/4
sin 2x 
=π
2 
0
π
= .
2

Cylindrical Shells Method

All the previous methods for finding the volumes of solids of rotation have relied on summing
the volumes of thin cross sectional slices that are perpendicular to the axis of rotation.
However, we can instead sum the volume of thin cylindrical shells that grow outwards from
and parallel to the axis of revolution.

79
 Figure 5.6: Wikipedia - Shell Integration

Each cylindrical shell will have a radius r(x), a height h(x), and a thickness dx, meaning a
volume of 2πr(x)h(x)dx.
Z b
V = 2π r(x)h(x)dx.
a
Example. Find the volume of the area bounded by the y-axis, y = 4 − x2 , and y = x revolved
around the y-axis.
Each shell’s height is parallel to the axis of rotation. In this case that’s the distance
between the two curves.
h(x) = 4 − x2 − x and r(x) = x.
√
−1+ 17
The two curves intersect at x = 2
. Finding the volume,
√
Z −1+ 17
2
V = 2π x(4 − x2 − x)dx
0
 −1+2√17
x4 x3


= 2π − − + 2x2 
4 3 0
π  √ 
= 121 − 17 17 .
12
Example. Find the volume of the area bounded by the x-axis and the curve y = (x − 1)2 (x −
2)2 rotated about the y-axis.
The height of the shell is parallel to the y-axis. In this case it’s exactly equal to the value
of the bounding curve.
h(x) = (x − 1)2 (x − 2)2 and r(x) = x.
Finding the volume,
Z 2
V = 2π x(x − 1)2 (x − 2)2 dx
1
2
x6 6x5 13x4

3

2 
= 2π − + − 4x + 2x 
6 5 4 1
π
= .
10
80
Other Axes of Rotation
Although rotating about the x and y axes are the most common, the methods we have here
apply to rotating about any axis parallel to the x or y axis. The includes any lines of the
form y = k or x = k where k is some constant.

One valid approach is simply to rewrite the equations for your bounding curves by shifting
them such that your axis of rotation is the x or y axis. However, it’s often more convenient
to not rewrite the equations and simply apply the methods.
Example. Find the volume of the solid generated by rotating the region in the first quadrant
bounded by y = x2 and y = 2x about the line x = −2.
We can use the washer method (although shells also works). Since we’re rotating about
a line parallel to the y-axis, we’ll have to rewrite our equations in the form x = . . ..
√ y
x= y and x = .
2
The inner radius is given by the distance from x = −2 to x = y/2, and the outer radius is
√
given by the distance from x = −2 to x = y. The curves intersect at y = 0 and y = 4.
y √
r(x) = 2 + and R(x) = 2 + y
2 Z 4
√ 2  y 2
V =π (2 + y) − 2 + dy
0 2
Z 4
y2

√
=π 4 y−y− dy
0 4
= 8π.

Example. Find the volume of the solid generated by rotating the region bounded by y = x2
and y = x + 2 about the line x = 3.
We can use the shells method. Our height is simply the difference in y values between
the two curves. The radius is the distance between the lines x = 3 and x + 2.

h(x) = x + 2 − x2 and r(x) = 3 − x.

The curves intersect at x = −1 and x = 2. Finding the volume,

Z 2
V = 2π (3 − x)(x + 2 − x2 )dx
−1
4
2
4x3 x2

x 
= 2π − + + 6x 
4 3 2 −1
45π
= .
2
81
Surface Area
Can apply the idea behind shell integration to derive a formula for surface area of a solid
of rotation. A cylindrical shell would have height ds, which is given to us by the arc length
formula, and radius x or y, if the axis of rotation is the y or x axis respectively.
( R
2π yds Rotation about x-axis
S= R
2π xdx Rotation about the y-axis
where
q
dy 2


 1 + dx
 dx y = f (x)
ds = r  2 .
 1 + dx

dy
dy x = g(y)

Example. Find the surface area of a sphere with radius r.

√
We can obtain a sphere by rotating y = r2 − x2 , −r ≤ x ≤ r about the x-axis.
dy −x
=√
dx r 2 − x2
 2
dy x2
= 2
dx r − x2
Z r r
x2
S = 2π y 1+ 2 dx
−r r − x2
Z r√
r
= 2π r 2 − x2 √ dx
−r r − x2
2
Z r
= 2π rdx
−r
2
= 4πr .

5.5 Differential Equations

Differential equations are equations, or sometimes systems of equations where we are given
how a function relates to its derivatives and would like to find satisfying functions or families
of functions. Differential equations land themselves well to modeling real-world phenomena
and are still an area of active mathematical study. We’ll see some very basic differential
equations and how what we’ve learned about integrals can be used to solve them.

5.5.1 Separable Differential Equations

If we can break up a first-order (just involving a first derivative) ordinary differential equation
into the following form,
dy
= f (x)g(y)
dx
82
then we say the differential equation is separable and may be able to be solved using a
technique called separation of variables. The steps to solve a separable differential equation
are:
dy
1. Write the equation in differential form (i.e using dx
)
dy
2. Separate the variables ( g(y) = f (x)dx)

3. Integrate both sides

4. Solve for y in terms of x, if possible

5. Find the general solution

6. Find a particular solution, given any initial conditions

Example. Solve the following differential equation.

dy
= (xy)2 , y(1) = 1.
dx
Separating and integrating,
dy
= (xy)2 = x2 y 2
dx
dy
= x2 dx
y2
1 x3
= + C.
y 3

Solving for C,

−1 13
= +C
1 3
−4
C= .
3
Solving for y,

−1 x3 4
= −
y 3 3
−3
y= 3 .
x −4

83
Exponential Growth & Decay
We can use differential equations to model growth and decay.

Example. Imagine we have some money in a bank account earning interest. The more
money in the bank, the more the more the account will receive in interest. So, the rate of
growth of the account value is proportional to the current account value. In the language of
differential equations,
dy
= ky.
dt
where k is some constant of proportionality.

This equation is separable, so we’ll try to solve it using separation of variables.

dy
= kdx
t 
ln y  = kt + C
 
y  = Cekt
y = Cekt
y(0) = Cek·0 = C
y = y0 ekt .

This is exactly the equation for continually compounding interest: y0 is the principle and k
is an interest rate.
Note that k could theoretically be negative, meaning the amount would decrease propor-
tionally to the remaining amount.

Example. The half-life of Pu-239 is 24360 years. Suppose that 10g of Pu-239 were released
in a nuclear accident, how long would it take to decay to 1g?

Our ”principle” is 10g. We know that half-life follows the exponential decay differential
equation, so it’s modeled by y = 10ekt . We also know that y(24360) = 5, which should
give us enough information to solve for k.

5 = 10ek·24360
1
= ek·24360 (see the “half” in half-life?)
2
1
ln = 24360k
2
− ln 2 = 24360k
− ln 2
k= .
24360

84
 We now can plug k back into our equation to get the full model.
− ln 2
y = 10e 24360 t

−t
= 10 eln 2 24360
−t
= 10 · 2− 24360 .

We can now plug in 1 for y and solve for t

t
1 = 10 · 2− 24360
1 t
= 2− 24360
10
1 t
log2 =−
10 24360
t
log2 10 =
24360
t = 24360 log2 10 ≈ 80922yr.
N0
If you’re familiar with half-life equations, this is exactly t = t1/2 log2 Nf
.

Logistic Growth & Decay

Although our differential equations assuming that growth is proportional to amount work
well for things like bank accounts, bacteria or radioactive particles that can grow and decay
without limits, that model is a little too simplistic to model populations that are limited by
resources.

Example. Imagine that we have a population of animals in a forest. If the forest has
lots of resources to support to animal population, then they can grow basically like normal.
However, as the population grows, resources become more scarce, so population growth would
slow down, or some of the population would starve. We can model this with the following
differential equation.
dP
= kP (M − P )
dt
where k is some constant of proportionality and M is some maximum population before
growth starts to decline.

85
 This differential equation is also separable, so let’s try to solve it.
dP
= kP (M − P )
dt
dP
= kdt
P (M − P )
 
1/M 1/M
dP + = kdt (using partial fractions)
P M −P
1    

ln P  − ln M − P  = kt + C
M
 P 
ln   = M kt + C
M −P
P
= CeM kt (b/c P ≥ 0)
M −P
P = CM eM kt − CP eM kt
P 1 + CeM kt = CM eM kt

CM eM kt
P =
1 + CeM kt
M
= .
1 + Ce−M kt

Figure 5.7: Logistic Growth

Looking at a graph of this function, we can see that it starts growing like an exponential
curve but begins to flatten, obtaining a maximum value of M , which is called the carrying
capacity. The population is growing the fastest when P = M/2, which you can verify by
finding the global maxima of dPdt
using a first derivative test.

86
5.5.2 Slope Fields & Euler’s Method
Unfortunately, not all differential equations are as easy to solve as separable differential
equations. In fact, some are impossible to get nice, closed-form solutions.

Slope Fields
We may still be able to visualize what the graph of a solution might look like by drawing
lines that have the same slope as a solution. Any solution that passes through the points
where we draw the sloped lines must be tangent to these lines, meaning a solution will follow
the “flow” of these lines.

dy
Figure 5.8: Slope field of dx
= −x/y with a possible solution

Euler’s Method
If your differential equation also has an initial condition, you can start at the initial point
and follow the slope field to find an approximate solution. This is what Euler’s Method tries
to accomplish. It can approximate the value of a solution at some x value by starting at
some point, usually given by the initial condition, and iteratively taking small steps of size
∆x in the direction determined by the slope field.

xn+1 = xn + ∆x
dy
yn+1 = yn + ∆x .
dx (xn ,yn )

87
 The smaller the steps, the more accurate the approximation.

Figure 5.9: Calc Workshop - Euler’s Method; Wikipedia - Euler Method

dy
Example. Given that dx = 3 − x and y(4) = 2, approximate the value of y(5) using Euler’s
method with increments of ∆x = 0.25.

dy dy
(x, y) dx
∆x ∆y = ∆x dx (x + ∆x, y + ∆y)
(4, 2) −1 0.25 −0.25 (4.25, 1.75)
(4.25, 1.75) −1.25 0.25 −0.3125 (4.5, 1.4375)
(4.5, 1.4375) −1.5 0.25 −0.375 (4.75, 1.0625)
(4.75, 1.0625) −1.75 0.25 −0.4375 (5, 0.625)

So, Euler’s Method yields an approximate4 value of (5, 0.625).

4
The actual value of the solution is (5, 0.5), so not too far off.

88
Chapter 6

Parametric, Vector, & Polar

Functions

6.1 Parametric & Vector Functions

Up to this point, almost all the graphs we have worked with have been of the form y = f (x),
defining the y coordinate in terms of the x coordinate. These sorts of functions are limited in
the types of graphs they can draw. If we instead let both the x and y coordinates be defined
in terms of another variable t, like (x(t), y(t)), then we can draw much more interesting
graphs. For example, a unit circle, which can’t be defined with a single function y = f (x),
would be (cos t, sin t).

We are always able to translate a function of the form y = f (x) into a parametric function
as (t, f (t)). Sometimes, but not always, we are also able to translate parametric functions
into y as a function of x.

Example. Given the following parametric function, find y as a function of x.

√
( t, t − 2).

Squaring both sides of the x equation,

x2 = t.

Substituting our expressing for t in terms of x into the y equation,

y = x2 − 2.

6.1.1 Vector Functions

Vector and parametric functions are essentially the same thing. In fact, in multivariable
calculus, we drop the idea of parametric functions almost completely and exclusively talk

89
about vector-valued functions. Both can graph the exact same functions. Visually, you might
imagine an arrow rooted at the origin tracing out the graph of a vector function. You’re
more likely to see vector functions written in the following form

~r(t) = hx(t), y(t)i.

All the normal vector operations, like addition and subtraction, scalar multiplication, and
dot products work exactly the same. If we think of ~r(t) as a position function,

Velocity: ~v (t) = r~0 (t) = hx0 (t), y 0 (t)i

 q
Speed: ~v (t) = (x0 (t))2 + (y 0 (t))2


Acceleration: ~a(t) = v~0 (t) = hx00 (t), y 00 (t)i

~t(t)
Direction:  
~v (t)

6.1.2 Slope & Concavity

Just like with functions like y = f (x), we can find the slope and concavity of parametric
functions using first and second derivatives respectively. We just apply the chain rule.
dy
dy dt
= dx
dx dt
d2 y dy 0 dy 0 /dt
= = .
dx2 dx dx/dt

Example. Consider the following parametric function:

(t2 − 5, 2 sin t), 0 ≤ t ≤ π.

Find the first and second derivatives of y with respect to x.

Differentiating both x and y with respect to t,

x0 (t) = 2t
y 0 (t) = 2 cos t
dy 2 cos t cos t
= = .
dx 2t t

90
 Finding the derivative of y 0 with respect to t,
d 0 d cos t
y =
dt dt t
−t sin t − cos t
=
t2
2 0
dy dy /dt
2
=
dx dx/dt
−t sin t−cos t
t2
=
2t
t sin t + cos t
=− .
2t3

6.1.3 Arc Length

Remember that we had the following formula for ds when deriving arc length.
q
ds = (dx)2 + (dy)2 .

Since we now have x and y as functions of t, we can rewrite this formula to get a formula
for arc length of a parametric function.
s 
2  2
dx dy
ds = + dt
dt dt
s
Z b  2  2
dx dy
s= + dt.
a dt dt

When talking about vector-valued functions or working in a more physics-based context, you
might hear the term “distance traveled” instead of arc length and see the following formula.
They are equivalent ideas. Z b
 
s= ~v (t)dt.
a

Example. A circle of radius r is defined parametrically as

(r cos t, r sin t), 0 ≤ t ≤ 2π.

Use this definition to find its circumference.

91
 dx
= −r sin t
dt
 2
dx
= r2 sin2 t
dt
dy
= r cos t
dt
 2
dy
= r2 cos2 t
dt
Z 2π p
C= r2 sin2 t + r2 cos2 tdt
Z0 2π p
= r sin2 t + cos2 tdt
Z0
= 02π rdt

= 2πr.

6.2 Polar Functions

6.2.1 Polar Coordinates
Up to this point, we’ve mostly described points in the plane by listing two numbers: the
distance along the x-axis and the distance along the y-axis. This description, called “rect-
angular coordinates” is pretty simple and has the advantage that every pair of coordinates
describes a unique point on the plane.

However, we could instead describe points in the plane with two numbers (r, θ), where r is
the point’s distance from the origin and θ is the point’s angle of inclination. This system is
more suited to describing
 √ √ points related to trig functions. For example the point at x = cos π4
and y = sin π4 is 22 , 22 in rectangular coordinates but 1, π4 in polar coordinates.

Note that unlike rectangular coordinates, multiple pairs of numbers can describe the same
point. For example (1, 0) is the same point as (1, 2π) is the same point as (1, −2π) is the
same point as (1, 4π).

We can easily convert between polar and rectangular coordinates.

(r, θ) polar = (r cos θ, r sin θ) rectangular

p
2 2
y
(x, y) rectangular = x + y , arctan polar.
x

92
6.2.2 Polar Functions
Polar functions are written in the form r = f (θ). Using our polar coordinate conversion
formulas, we can convert any polar function to a parametric function.

x(θ) = r cos θ = f (θ) cos θ

y(θ) = r sin θ = f (θ) sin θ.
dy
Now we can use our parametric function formulas to get dx
.

dy dy/dθ
= .
dx dx/dθ
dy
Example. A cardioid is defined by r = 1 − cos θ, 0 ≤ θ ≤ 2π. Find dx
.

x(θ) = (1 − cos θ) cos θ

dx
= sin (2θ) − sin θ
dθ
y(θ) = (1 − cos θ) sin θ
dy
= cos θ − cos (2θ)
dθ
dy cos θ − cos (2θ)
=
dx sin (2θ) − sin θ
3θ
= tan .
2

Area Enclosed
When a polar function is changed by dθ, it sweeps out an circular sector.

93
 Figure 6.1: Polar Area

dθ 2
This circular sector has area dA = 2
r . Integrating dA for α ≤ θ ≤ β,
Z β Z β
1 2 1 2
A= r dθ = f (θ)dθ.
α 2 α 2
Example. Find the area inside the smaller loop of the limaçon r = 2 cos θ + 1.

First, we need to find our bounds on θ. We know that the small loop begins and ends
when r = 0.

0 = 2 cos θ + 1
1
− = cos θ
2
π 4π
θ= , .
3 3

94
 Now that we have our bounds, we can integrate.
Z 4π/3
1
A= (2 cos θ + 1)2 dθ
2π/3 2
Z 4π Z 4π Z 4π
!
1 3 3 3
= 4 cos2 (θ) dθ + 4 cos (θ) dθ + dθ
2 2π
3
2π
3
2π
3
Z 4π Z 4π Z 4π
3 3 1 3
= (1 + cos (2θ)) dθ + 2 cos (θ) dθ + dθ
2π
3
2π
3
2 2π
3
Z 4π Z 4π Z 4π
3 3 3 3
= cos (2θ) dθ + 2 cos (θ) dθ + dθ
2π
3
2π
3
2 2π3
4π/3
1 3 
= sin (2θ) + 2 sin (θ) + θ
2 2 2π/3
           
1 8π 4π 3 4π 1 4π 2π 3 2π
= sin + 2 sin + − sin + 2 sin +
2 3 3 2 3 2 3 3 2 3
√ ! √ !
3 √ 3 √
= − 3 + 2π − − + 3+π
4 4
√
3 √
= −2 3+π
2 √
3 3
=π− .
2

Area Between Curves

The area between r1 (θ) and r2 (θ) is simply the difference between the areas.
Z β Z β Z β
1 1 1
A= r12 (θ)dθ − r22 (θ)dθ = (r12 (θ) − r22 (θ))dθ.
2 α 2 α 2 α

Example. Find the area that lies inside the circle r = 1 and outside the cardioid r = 1−cos θ.

To find the bounds, we need to find where these curves intersect.

1 = 1 − cos θ
cos θ = 0
π −π
θ= , .
2 2
Since we want the area inside of the circle and outside of the cardioid, our bounds are
−π
2
≤ θ ≤ π2 . We’ll also have r1 (θ) be the circle and r2 (θ) be the cardioid, since we are

95
 effectively finding the area inside the circle and subtracting away the area that is also in
the cardioid.
1 π/2 2
Z
1 − (1 − cos θ)2 dθ


A=
2 −π/2
Z π
1 2
2 cos (θ) − cos2 θ dθ


=
2 −π2
Z π 
1 2 1 + cos (2θ)
= 2 cos (θ) − dθ
2 −π2
2
  π
1 θ sin (2θ)  2
= 2 sin (θ) − −
2 2 4  −π
2
  π  π sin (π)     
1 −π π sin (−π)
= 2 sin − − − 2 sin + −
2 2 4 4 2 4 4
  π  π sin (π) 
= 2 sin − −
2 4 4
π
=2− .
4
Example. Find the area that lies outside the circle r = 1 and inside the cardioid r = 1−cos θ.
We need to make sure our bounds are sweeping out the correct area. If we did −π 2
≤ θ ≤ π2 ,
we’d get the area inside the circle and outside the cardioid, which isn’t what we want
here. We know that for polar coordinates, −π 2
≡ 3π
2
. So, out bounds are π2 ≤ θ ≤ 3π2
.
Since we want the area inside the cardioid and outside the circle, effectively taking the
cardioid and subtracting away the intersection, so r1 (θ) is the cardioid, and r2 (θ) is the
circle.
1 3π/2
Z
(1 − cos θ)2 − 12 dθ


A=
2 π/2
Z 3π
1 2
cos2 (θ) − 2 cos (θ) dθ


=
2 π2
Z 3π  
1 2 1 + cos (2θ)
= − 2 cos (θ) dθ
2 π2 2
 3π/2
1 θ sin (2θ) 
= + − 2 sin (θ) 
2 2 4 π/2
     π 
1 3π sin (3π) 3π π sin (π)
= + − 2 sin − + − 2 sin
2 4 4 2 4 4 2
π
= + 2.
4
Example. Find the area inside both the circle r = 1 and the cardioid r = 1 − cos θ.

96
 This area isn’t between two polar curves like the previous examples in the sense that we
can’t define it as one region minus another. We’ll prove it it two ways: logically and by
using two regions. Logically, we know that the total area of the circle is π. We also know
the area inside the circle but outside the cardioid is 2 − π4 . So,
 π  5π
Aboth = π − 2 − = − 2.
4 4
We can break the region inside both curves into two parts: a half circle for x ≤ 0 and the
two cardioid bulges for x ≥ 0. The area of the half-circle is π/2. We can find the area of
the two cardioid bulges.

Abulges = 2Abulge
Z π/2
= (1 − cos θ)2 dθ
0
Z π
2
cos2 θ − 2 cos (θ) + 1 dθ


=
0
Z π 
2 1 + cos (2θ)
= − 2 cos (θ) + 1 dθ
0 2
π
3θ sin (2θ) 2
= + − 2 sin (θ)
2 4 0
3π sin (π) π 
= + − 2 sin
4 4 2
3π
= −2
4

Adding in the area of the half-circle,

Aboth = Ahalf + Abulges

π 3π
= + −2
2 4
5π
= − 2.
4
We see that we get the same answer either way.

Arc Length
Since we know how to convert polar functions to parametric, we can simply adapt the
parametric arc length formula.
s
Z β  2  2
dx dy
s= + dθ.
α dθ dθ

97
However, there is an alternate form that works just for polar functions. For some small
change dθ, we see a corresponding small changes dr and rdθ.

Figure 6.2: Polar Arc Length

We see that these changes for a right triangle with hypotenuse ds.

(ds)2 = (rdθ)2 + (dr)2

 2 !
dr
= r2 + (dθ)2
dθ
s  2
dr
ds = r2 + dθ
dθ
s  2
Z β
2
dr
s= r + dθ.
α dθ

Giving us an alternate formula for polar arc length.

Example. Find the arc length of the cardioid r = 1 − cos θ.

98
The bounds are 0 ≤ θ ≤ 2π. Using the polar arc length formula,
dr
= sin θ
dθ
 2
dr
= sin2 θ
dθ
Z 2π q
s= (1 − cos θ)2 + sin2 θdθ
Z0 2π √
= 2 + 2 cos θdθ
0
Z 2π  
θ
= 2 sin dθ
0 2
 2π
θ 
= −4 cos
2 0
= 8.

99
Chapter 7

Sequences, L’Hôpital’s Rule, &

Improper Integrals

7.1 Sequences
Definition. A sequence {an } = {a1 , a2 , . . . , an } is an ordered list of numbers. Each element
of a sequences is called a term and is identified by its index in the sequence. Sequences can
be finite or infinite.
Example. The nth term of a sequence is defined by the following formula:
(−1)n
an = .
n2 + 1
Find the 1st, 2nd, and 100th terms of the sequence.

(−1)1 −1
a1 = =
12 + 1 2
2
(−1) 1
a2 = 2 =
2 +1 5
100
(−1) 1
a100 = 2
= .
100 + 1 10001
The above sequence was defined explicitly, meaning that we have a formula for the nth term
of the sequence only in terms of n. However, sequences can also be defined recursively,
meaning the formula for subsequent terms of the sequence contains previous terms. For a
recursive sequence to be properly defined, there need to be one or more base terms that
aren’t defined recursively. For example, the Fibonacci sequence, one of the most famous
recursive sequences, as a1 and a2 as base terms.
(
1 n = 1, 2
an = .
an−1 + an−2 n ≥ 3

100
7.1.1 Common Types of Sequences
There are some common types of sequences that you should be familiar with. You might
recognize these types of sequences and some of the formulas surrounding them from previous
math classes.

Arithmetic Sequences
Definition. An arithmetic sequence is one where an+1 − an = d, a common difference, for
all terms.

That is, to get the next term, we simply add some number d (which could be negative) to
the previous term. Arithmetic sequences can be defined either explicitly or recursively. Let
a0 be the starting term of the sequence.

an = dn + a0
(
a0 n=0
= .
d + an−1 n≥1

As we can see from the explicit formula, if we graphed values of an arithmetic sequence on
in the plane with x coordinate n and y coordinate an , all points would lie on a line with
slope d and y intercept a0 .

Example. Write an explicit formula for the following arithmetic sequence.

{ln 2, ln 6, ln 18, . . .} .

Since we are given that this sequence is arithmetic, we’ll find the common difference.
6
d = ln 6 − ln 2 = ln = ln 3.
2
So, applying the explicit formula for an arithmetic sequence with starting term ln 2 and
common difference ln 3,
an = ln (3)n + ln 2, n ≥ 0.
We might have also noticed that each term inside the ln is triple the previous one, meaning
we can write an explicit formula and then simplify to the same answer as before.

an = ln (3n · 2), n ≥ 0
= ln 3n + ln 2, n ≥ 0
= ln (3)n + ln 2, n ≥ 0.

101
Geometric Sequences
an+1
Definition. A geometric sequence is one where an
= r, a common ratio, for all terms.

That is, to get the next term, we simply multiply some number r (which could be negative)
by the previous term. Geometric sequences can also be defined explicitly or recursively. Let
a0 be the starting term of the sequence.

an = a0 (r)n
(
a0 n=0
= .
ran−1 n≥1

As we can see with the explicit formula, if we graphed terms of a geometric sequence for
positive r, the points would lie on an exponential curve with y intercept a0 and exponential
base r.

Example. Write an explicit formula for the following geometric sequence.

{2, −6, 18, −54, . . .} .

Since we are given that this sequence is geometric, we’ll find the common ratio.
−6
r= = 3.
2
So, applying the explicit formula for a geometric sequence with starting term 2 and
common ratio -3,
an = 2(−3)n .

7.1.2 Limits of a Sequence

Once we have a formula for a sequence, we might be interested to know if an tends towards
some value as n gets large.

Definition. Let L be a real number, the sequence {an } as limit L as n approaches infinity
if given any real  > 0, there is some index m such that for all n > m
 
an − L < .

We notate this as
lim an = L
n→∞

and say the sequence converges to L. If the sequence does not have a limit, then we say the
sequences diverges.

102
The following rules we gave for limits of a function: Sum and Difference Rule, Product Rule,
Constant Multiple Rule, and Quotient Rule, all still apply to limits of sequences. The only
rule that doesn’t still hold is the Power Rule because of the following sort of problem:

an = (−1)n
lim a2n = 1
n→∞
 2
lim an = DNE
n→∞
1 6= DNE.

The Sandwich Theorem for Sequences

Theorem (Sandwich Theorem for Sequences). If limn→∞ an = limn→∞ cn = L and there is
an integer m such that an ≤ bn ≤ cn for all n > m, then limn→∞ bn = L.

This is essentially the same as the Sandwich Theorem for limits of a function. The only
added caveat is that we have to find some index m for which the sandwiching inequality
always holds for terms after the mth.

Example. Determine if the following sequence converges or diverges. If it converges, find

its limit.
n−1
an = (−1)n , n ≥ 1.
2
 
We see that as n grows large an  approaches 1. However, an bounces between 1 and -1
depending on whether n is even or odd. Thus, we could let  = 1/2, which would show
that the limit diverges.

Example. Determine if the following sequences converges or diverges. If it converges, find

its limit.
cos n
an = , n ≥ 1.
n
It might at first seem that this limit diverges because cos bounces between -1 and 1.
However, we can use the Sandwich Theorem to show that the limit converges.
−1 cos n 1
≤ ≤ ,n ≥ 1
n n n
−1 cos n 1
lim ≤ lim ≤ lim
n→∞ n n→∞ n n→∞ n
cos n
0 ≤ lim ≤0
n→∞ n
cos n
lim = 0.
n→∞ n

103
Absolute Value Theorem
We can apply the Sandwich Theorem to show that sequences whose absolute value converges
to 0 must also converge to 0.
 
Theorem (Absolute Value Theorem). If limn→∞ an  = 0, then limn→∞ an = 0.

Proof. For all n,    

−an  ≤ an ≤ an .
Applying the Sandwich Theorem and limit properties,
   
− lim an  ≤ lim an ≤ lim an 
n→∞ n→∞ n→∞
−0 ≤ lim an ≤ 0
n→∞
lim an = 0.
n→∞

7.2 L’Hôpital’s Rule

7.2.1 Indeterminate Form 0/0
If both f (x) and g(x) are 0 at x = a, then the limit

f (x)
lim .
x→a g(x)

is an indeterminate form of 0/0, meaning we can’t substitute x = a to evaluate the limit.

However, L’ôpital’s Rule allows us to modify this limit to get another limit, which might not
have this indeterminate form but is guaranteed to have the same limit.

Theorem (L’Hôpital’s Rule, Weaker Form). If f (a) = g(a) = 0; f 0 (a) and g 0 (a) 6= 0 exist,
then
f (x) f 0 (a)
lim = 0 .
x→a g(x) g (a)

104
Proof.
f (x) f (x) − 0
lim = lim
x→a g(x) x→a g(x) − 0

f (x) − f (a)
= lim
x→a g(x) − g(a)
f (x)−f (a)
x−a
= lim
x→a g(x)−g(a)
x−a

limx→a f (x)−f
x−a
(a)
=
limx→a g(x)−g(a)
x−a
0
f (a)
= .
g 0 (a)

x2 −4
Example. Find limx→2 x−2
using L’Hôpital’s Rule.

x2 − 4 2(2)
lim = = 4.
x→2 x − 2 1
Note that we get the same answer if we didn’t use L’Hôpital’s Rule and instead factored.
x2 − 4 (x + 2)(x − 2)
lim = lim lim x + 2 = 4.
x→2 x − 2 x→2 x−2 x→2

It’s possible that f 0 (a) = g 0 (a) = 0, meaning we’re still left with the indeterminate form
0/0. However, we can use a stringer form of L’Hôpital’s Rule that allows us to not have to
immediate substitute x = a and allows us to apply the rule multiple times if needed.
Theorem (L’Hôpital’s Rule, Stronger Form). If f (a) = g(a) = 0; f and g are differentiable
on an open interval I that contains a; g 0 (x) 6= 0 if x 6= a, then
f (x) f 0 (x)
lim = lim 0
x→a g(x) x→a g (x)

if the right-hand limit exists.

Example. Find the following limit or show that it doesn’t exist.
cos x − 1
lim x .
x→0 e − x − 1

cos 0 − 1 = 0 = e0 − 0 − 1; ex − x − 1 6= 0 if x 6= 0 on all real numbers.

cos x − 1 − sin x
lim x
= lim x .
x→0 e − x − 1 x→0 e − 1

− sin 0 = 0 = e0 − 1; ex − 1 6= 0 if x 6= 0 on all real numbers.

− sin x − cos x −1
lim x
= lim x
= = −1.
x→0 e − 1 x→0 e 1

105
7.2.2 Indeterminate Forms ∞/∞, ∞ · 0, & ∞ − ∞
∞/∞
L’Hôpital’s still applies as written for the indeterminate form ∞/∞.
Example. Find the following limit or show that it doesn’t exist.
tan x
lim .
x→π/2 1 + tan x

Applying L’Hôpital’s Rule,

tan x sec2 x
lim = lim = 1.
x→π/2 1 + tan x x→π/2 sec2 x

∞·0
We need to rearrange the limit into a 0/0 or ∞/∞ indeterminate form.
Example. Find the following limit or show that it doesn’t exist.
1
lim x sin .
x→∞ x
Rearranging,
1 sin 1
lim x sin = lim 1 x .
x→∞ x x→∞ x
Since the limit now has indeterminate form 0/0, we can apply L’Hôpital’s Rule.
−1
sin x1 x2
cos x1
lim 1 = lim −1 = cos(0) = 1.
x→∞ x→∞
x x2

∞−∞
We need to rearrange the limit into a 0/0 or ∞/∞ indeterminate form.
Example. Find the following limit or show that it doesn’t exist.
1 1
lim − .
x→1 ln x x−1
Rearranging,
1 1 x − 1 − ln x
lim − = lim .
x→1 ln x x − 1 x→1 (x − 1) ln x
Since the limit now has indeterminate form 0/0, we can apply L’Hôpital’s Rule.

x − 1 − ln x 1 − x1 1
x2 1 1
lim = lim = lim = = .
x→1 (x − 1) ln x x→1 (x − 1) 1 + ln x x→1 1
− (x − 1) x12 + 1
1−0+1 2
x x x

106
7.2.3 Indeterminate Forms 1∞ , 00 , & ∞0
For indeterminate forms with exponents, we should take the natural log of the limit, solve
that limit, and then exponentiate.

1∞
Example. Find the following limit or show that it doesn’t exist.
 x
1
lim 1 + .
x→∞ x

Let L be the value of the limit.

 x
1
L = lim 1 +
x→∞ x
 x 
1
ln L = lim ln 1+
x→∞ x
 
1
= lim x ln 1 + (indeterminate form ∞ · 0)
x→∞ x
ln 1 + x1


= lim 1 (indeterminate form 0/0)
x→∞
x
1 −1
1+ x1 x2
= lim −1
x→∞
x2
1
= lim 1
x→∞ 1 +
x
=1
ln L
e = e1
L = e.

00
Example. Find the following limit or show that it doesn’t exist.

lim+ xx .
x→0

107
 Let L be the value of the limit.

L = lim+ xx
x→0
ln L = lim+ x ln x (indeterminate form 0 · −∞)
x→0
ln x
= lim+ 1 (indeterminate form −∞/∞)
x→0
x
1
x
= lim+ −1
x→0
x2
1
= lim+ −1
x→0
x
= lim+ −x
x→0
=0
ln L
e = e0
L = 1.

∞0
Example. Find the following limit or show that it doesn’t exist.
1
lim x x .
x→∞

Let L be the value of the limit.

1
L = lim x x
x→∞
1
ln L = lim ln x (indeterminate form 0 · ∞)
x→∞ x
ln x
= lim (indeterminate form ∞/∞)
x→∞ x
1
x
= lim
x→∞ 1
=0
eln L = e0
L = 1.

7.3 Relative Growth Rates

In many practical applications, like the run times of computer algorithms for example, we
often want to know if a function grows slower, the same, or faster than another.

108
Definition. If
f (x) g(x)
lim = ∞ ⇔ lim = 0,
x→∞ g(x) x→∞ f (x)
then f grows faster than g. If

f (x) g(x) 1
lim = c ⇔ lim =
x→∞ g(x) x→∞ f (x) c

for some non-zero constant c, then f and g grow at the same rate.

Example. Compare ex and x100 . Does one grow faster than the other, or do they grow at
the same rate?

ex ex
lim = lim
x→∞ x100 x→∞ 100x99
.
= .. (after many applications of L’Hôpital’s Rule)
ex
= lim = ∞.
x→∞ 100!

So, ex grows faster than x100 . In fact, any exponential bx grows faster than any polynomial,
as long as b > 1.

7.3.1 Transitive Grow Rates

For sufficiently large x, growth rates are transitive. That is, if f grows the same/faster/slower/
than g, and g grows the same/slower/faster than h, then f also grows the same/faster/slower
than h.
√ √ 2
Example. Show that f (x) = x2 + 5 and g(x) = (2 x − 1) grow at the same rate.

We’ll show that both f and g grow at the same rate as h(x) = x. Starting with f and h,
√
f (x) x2 + 5
lim = lim
x→∞ h(x) x→∞ x
r
x2 + 5
= lim
x→∞
r x2
x2 + 5
= lim 2
(by the Power Rule)
√
x→∞ x
= 1
= 1.

109
 So, f and h grow at the same rate. Moving on to g and h,
√ 2
g(x) (2 x − 1)
lim = lim
x→∞ h(x) x→∞ x
 √ 2
2 x−1
= lim √
x→∞ x
 √ 2
2 x−1
= lim √ (by the Power Rule)
x→∞ x
= (2)2
= 4.

So, g and h grow at the same rate. Since f and g both grow at the same rate as h, f and
g must grow at the same rate as each other.

7.3.2 Growth Rate Hierarchy (nn FEPL)

For most of the common types of functions we see, we can establish families of functions
and rank these families by their growth rates from fastest-growing to slowest-growing. If two
functions are in different families, we can be sure that one grows faster than the other. If
two functions are in the same family, we’ll have to do more work to compare them. These
families are summarized by the acronym nn FEPL1 .
nn These are functions that have a variable both in the base and exponent.
Factorials These are functions that have an n! term.
Exponentials These are functions that have a constant base and a variable exponent.
Note that if the variable base if less than 1, the function actually gets smaller for
larger n.
Polynomials These are functions with a variable base and constant exponent. Certain
2
polynomials can still grow
√faster than others. For example, x grows faster than x,
which grows faster than x.
Logarithms These are functions that have a log of a polynomial.
Although these rules are indeed true, don’t just apply them blindly. You should try to sim-
plify a function first before figuring out to which family it belongs. For example, although
ln xx contains an xx and a ln, it’s neither in the nn family nor in the logarithms family. In
fact, although this function is not a polynomial, it grows faster than x but slower than x2 .

A function belongs to the family of its fastest-growing positive term. Negative terms can
either be ignored or used to simplify other terms. For example, although x3 + ex contains
1
You might recognize these families as a sort of Big-O family from computer science.

110
a polynomial x3 term, for very large x, the ex term dominates the growth, meaning this
function is part of the exponentials family.

7.4 Improper Integrals

Now that we’ve developed the tools to deal with limits as they approach infinity and the
possible indeterminate forms that may arise, we can apply these ideas to integrals, allowing
us to have −∞ and ∞ as limits of integration. We call these integrals with ±∞ as limits of
integration, and functions that become ±∞ somewhere within the interval we’re integrating
on improper integrals.

7.4.1 Infinite Integration Limits

Definition. If f is continuous on [a, ∞), then
Z ∞ Z b
f (x)dx = lim f (x)dx.
a b→∞ a

If f is continuous on (−∞, b], then

Z b Z b
f (x)dx = lim f (x)dx.
∞ a→−∞ a

If f is continuous on (−∞, ∞), then

Z ∞ Z c Z ∞
f (x)dx = f (x)dx + f (x)dx
−∞ −∞ c

for any real constant c. If these limits exist, then the integral converges and has a value.
Otherwise, the integral diverges and does not have a value.

Example. Evaluate the following integral or state that it diverges.

Z ∞
3
2
dx.
2 x −x

111
 Applying the definition,
Z ∞ Z b
3 3
2
dx = lim dx
2 x −x b→∞ 2 x2−x
Z b 
1 1
= lim 3 − dx
b→∞ 2 x−1 x
 b
 x − 1 
= lim 3 ln  
b→∞ x 
  2  
b − 1 1
= lim 3 ln    − 3 ln  
b→∞ b  2
= 0 + 3 ln 2
= 3 ln 2.
Example. Evaluate the following integral or state that it diverges.
Z ∞
dx
√ .
1
4
x
Applying the definition,
Z ∞ Z b
dx dx
√ = lim √
1
4
x b→∞ 1
4
x
4√
b
4

= lim x3 
b→∞ 3
1
4√ 4 4√4
= lim b3 − 13
b→∞ 3 3
4
=∞−
3
= diverges.

7.4.2 Infinite Discontinuities

An infinite discontinuity occurs when a function takes on a value of ±∞ on the interval we’re
integrating on. In this case, we’ll need to split the integral into pieces, evaluating the limit
as we approach this infinite discontinuity from both sides.
Definition. If f is continuous on (a, b], then
Z b Z b
f (x)dx = lim+ f (x)dx.
a c→a c

If f is continuous on [a, b), then

Z b Z c
f (x)dx = lim− f (x)dx.
a c→b a

112
If f is continuous on [a, c) ∪ (c, b], then
Z b Z c Z b
f (x)dx = f (x)dx + f (x)dx.
a a c

If these limits exist, then the integral converges and has a value. Otherwise, the integral
diverges and does not have a value.
Example. Evaluate the following integral or state that it diverges.
Z 1
dx
2
.
0 x

We see that we have an infinite discontinuity at x = 0. Applying the definition,

Z 1 Z 1
dx dx
2
= lim+ 2
0 x c→0 c x
1
−1 
= lim+
c→0 x c
−1 1
= + lim+
1 c→0 c
= −1 + ∞
= diverges.
Example. Evaluate the following integral or state that it diverges.
Z 1
dx
1/2
.
0 x

We see that we have an infinite discontinuity at x = 0. Applying the definition,

Z 1 Z 1
dx dx
1/2
= lim+ 1/2
0 x c→0 c x
1

= lim+ 2x1/2 
c→0 c
1/2
= 2 − lim+ 2c
c→0
=2−0
= 2.

7.4.3 Convergence Tests

P-Test
Lemma. The following integral will converge when p > 1 and diverge if 0 < p ≤ 1.
Z ∞
dx
.
1 xp

113
Example. Evaluate the following integral or state that it diverges.
Z ∞
dx
.
1 x
We have an integral where p = 1. So, by the P-Test, the integral diverges.
Example. Evaluate the following integral or state that it diverges.
Z ∞
dx
.
1 x1.001
We have an integral where p = 1.001. So, by the P-Test, the integral converges.
Z ∞ Z b
dx dx
= lim
1 x1.001 b→∞ 1 x1.001
b
−0.001 

= lim −1000x
b→∞ 
1
−0.001
= lim −1000b + 1000(1)−0.001
b→∞
= 0 + 1000
= 1000.

Direct Comparison Test

Lemma. Let f and g be continuous on [a, ∞) with 0 ≤ f (x) ≤ g(x) for all x ≥ a.
Z ∞ Z ∞
f (x)dx converges if g(x)dx converges.
a a
Z ∞ Z ∞
g(x)dx diverges if f (x)dx diverges.
a a

That is, if a larger function converges, then so will a smaller funtion; if a smaller function
diverges, then so will a larger function.

The hardest part of the Direct Comparison Test is deciding what function you should com-
pare to. A general rule is to pick a function that is similar to, but simpler than then given
function.
Example. Evaluate the following integral or state that it diverges.
Z ∞
dx
2
.
1 x − 0.1
If the 0.1 wasn’t inside the square root, the function would simplify to 1/x. Since the 0.1
is subtracted, the denominator is smaller than 1/x. So, 1/x is a function that is smaller
on [1, ∞), meaning if it diverges, then so will the original function. We know by the
P-Test that the integral of 1/x from 1 to ∞ will diverge, so the original function also
diverges.

114
Example. Evaluate the following integral or state that it diverges.
Z ∞
2
e−x dx.
1

Since the exponent is negative, a smaller exponent would mean a larger value. So, e−x is
a larger function on [1, ∞). Z ∞
1
e−x dx = ,
1 e
meaning it converges, so the original function also converges.

Limit Comparison Test

Lemma. If positive functions f and g are continuous on [a, ∞) and

f (x)
lim
x→∞ g(x)

converges to a positive real number, then

Z ∞ Z ∞
f (x)dx and g(x)dx
a a

both converge or both diverge.

Many functions to which you can apply the Limit Comparison Test you can also apply
the Direct Comparison Test. The practical use of the limit comparison test is to take an
uglier function, that may be tedious to integrate and compare it to a function that is easy
to determine whether it diverges using something like the P-Test. A common strategy,
especially for rational functions, is to look at their end behavior model.

Example. Evaluate the following integral or state that it diverges.

Z ∞
dx
.
1 1 + x2
Although you might recognize this as the derivative of arctan, let’s continue with the
Direct Comparison Test. This function looks very similar to 1/x2 , which we know by the
P-Test will converge on [1, ∞).
1
x2 1 + x2
lim 1 = lim = 1.
x→∞
1+x2
x→∞ x2

Since 1 is a positive real constant and the integral of 1/x2 converges, then the original
integral also converges by the Limit Comparison Test.

115
Example. Evaluate the following integral or state that it diverges.
Z ∞
3x + 6
dx.
1 1 − 5x + 7x2
Looking at this rational function, we see a degree 1 polynomial in the numerator and a
degree 2 polynomial in the denominator. So, we’d expect this rational function to have
the same end behavior model as 1/x, which we know by the P-Test diverges.
3x+6
1−5x+7x2 3x2 + 6x 3
lim 1 = lim = .
x→∞ x→∞ 7x2 − 5x + 1 7
x

Since 3/7 is a positive real constant and the integral of 1/x diverges, the the original
integral also diverges by the Limit Comparison Test.

116
Chapter 8

Infinite Series

We previously discussed some special types of series and formulas for their nth term. Now,
we’ll look at sums of infinitely many terms and eventually see how summing variables rather
than just numbers allows us to approximate functions.

8.1 Power Series

8.1.1 Geometric Series
First, we need to define what we mean by an infinite series.

Definition. An infinite series is of the form

∞
X
a1 + a2 + . . . + an + . . . or equivalently, ak .
k=1

Just like with finite series, each ai is a term, and an is the nth term.

We can describe the behavior of an infinite series by looking at how its value behaves after
summing a finite number of terms. We can define what it means for an infinite sum to have
a value by looking at the limit of the partial sums as n grows large.

Definition. The nth partial sum of an infinite series is

n
X
sn = ak .
k=1

The infinite series converges to value L if

lim sn = L.
n→∞

Otherwise, the series diverges and does not have a value.

117
Example. State if the following infinite series converges or diverges.
3 3 3
+ + ... + n + ....
10 100 10
Looking at the partial sums,

s1 = 0.3
s2 = 0.33
..
.
sn = 0. 33333
| {z . .}.
n total 3’s

So, it seems the limit of the partial sums tends towards a decimal with an infinite number
of 3’s. This value corresponds to the decimal expansion of 1/3, which clearly is real and
finite, so the series converges.
The above series is a geometric series since each subsequent term is 10 times smaller than
the previous one (i.e r = 1/10).

Lemma. The geometric series

∞
X
a0 (r)k
k=0
 
converges to a value of a0 /(1 − r) if r < 1 and diverges otherwise.

Proof. We’ll first find a formula for the partial sums and then find the limit of the partial
sums for 1 < r < 1.
n
X
sn = a0 (r)k
k=0
= a0 + a0 r + a0 r 2 + . . . + a0 r n
rsn = a0 r + a0 r2 + a0 r3 + . . . + a0 rn + a0 rn+1
= −a0 + sn + a0 rn+1
sn (r − 1) = a0 rn+1 − 1

rn+1 − 1
s n = a0 .
r−1
This formula for partial sums holds for all values of r. Now we’ll take the limit of sn and see

118
for what values of r the limit exists.
∞
X
a0 (r)n = lim sn
n→∞
k=0
rn+1 − 1
= lim a0
n→∞ r−1
−a0  
= , r <1
r−1
a0  
= , r < 1.
1−r


8.1.2 Functions from Geometric Series

What happens if rather than letting r be some fixed value we know beforehand, we let r be
some variable x? Applying the formula,
a0  
a0 + a0 x + a0 x 2 + . . . = , x < 1.
1−x
 
This sort of sum of powers of x is called a power series, and the condition that x < 1 is
called the interval of convergence. Right now, this power series is centered at x = 0, but we
can generalize it a bit to be centered at x = h.
a0  
a0 + a0 (x − h) + a0 (x − h)2 + . . . = , x − h < 1.
1 − (x − h)
Note that this formula allows us to find the power series of any function a0 /(mx + b).
a0 a0  
= = a0 + a0 (1 − mx + b) + a0 (1 − mx + b)2 + . . . , 1 − mx − b < 1.
mx + b 1 − (1 − mx − b)
So,
1  
= 1 + (1 − x) + (1 − x)2 + . . . , 1 − x < 1
x
1  
= 1 + x + x 2 + x3 + . . . ,  x < 1
1−x
1  
= 1 − x + x2 − x3 + . . . , x < 1.
1+x

Term-By-Term Differentiation
Theorem. If the power series
∞
X
f (x) = ck (x − a)k = c0 + c1 (x − a) + c2 (x − a)2 + . . .
k=0

119
  
converges for x − a < R, including R = ∞, then the power series
∞
X
kck (x − a)k−1 = c1 + 2c2 (x − a) + 3c3 (x − a)2 + . . .
k=1
 
also converges for x − a < R and is equal to f 0 (x) on that interval.
Applying the theorem,
−1 2
 
= −1 − 2(1 − x) − 3(1 − x) − . . . ,  1 − x < 1
x2
1 2
 
= 1 + 2x + 3x + . . . ,  x < 1
(1 − x)2
−1 2
 
= −1 + 2x − 3x + . . . , x < 1.
(1 + x)2

Term-By-Term Integration
Theorem. If the power series
∞
X
f (x) = ck (x − a)k = c0 + c1 (x − a) + c2 (x − a)2 + . . .
k=0
 
converges for x − a < R, including R = ∞, then the power series
∞
X (x − a)k+1 (x − a)2 (x − a)3
ck = c0 (x − a) + c1 + c2 + ...
k=0
k+1 2 3
 
also converges for x − a < R and represents the antiderivative of f on that interval.
Applying the theorem,
  (1 − x)2 (1 − x)3  
ln x = 1 + + + . . . ,  1 − x < 1
2 3
2 3
  x x  
− ln 1 − x = x + + + . . . ,  x < 1
2 3
2 3
  x x x4  
ln 1 + x = x − + − + . . . , x < 1.
2 3 4
This is pretty impressive: we now have an formula for ln x in terms of polynomials. This
also includes some pretty surprising identities. For example1 ,
1 1 (−1)n+1
ln 2 = 1 − + − ... + + ....
2 3 n
1
Technically, we’re substituting x = 1 which isn’t in the interval of convergence. However, since we are dealing
with an alternating sum, the series will also converge for x = −1 and x = 1.

120
 Example. Find a power series for arctan x and state the interval of convergence.
We know that
d 1
arctan x = .
dx 1 + x2
So, if we can find a power series that represents 1/(1 + x2 ), we can integrate term-by-term
to get a power series for arctan x. We also already know the following power series.
1  
= 1 − u + u2 − u3 + . . . , u < 1.
1+u
Letting u = x2 ,
1  
2
= 1 − x2 + x4 − x6 + . . . ,  x2  < 1
1+x  
= 1 − x2 + x4 − x6 + . . . , x < 1.

Integrating term-by-term,
x3 x5 x2n+1  
arctan x = x − + + . . . + (−1)n + . . . , x < 1.
3 5 2n + 1
This power series also gives rise to a pretty interesting identity2 .
π 1 1 1 x2n+1
arctan 1 = = 1 − + − + . . . + (−1)n ....
4 3 5 7 2n + 1

8.2 Taylor Series

8.2.1 Construction
Although we can approximate lots of functions using power series derived from the geometric
series, it’d be nice if we had a more general way to approximate any function using a power
series. We already have an approximation using a tangent line: the 0th and 1st derivatives
of the function and line are equal at the point of tangency. We could extend this idea of
derivatives being equal to higher-order derivatives and higher degree polynomials.
Example. Construct polynomial P (x) = a0 + a1 x + a2 x2 + a3 x3 + a4 x4 with the following
behavior at x = 0:

P (0) = 1
P 0 (0) = 2
P 00 (0) = 3
P 000 (0) = 4
P (4) (0) = 5.
2
See footnote 1.

121
 Plugging in x = 0 and solving for a0 ,
a0 + a1 (0) + a2 (0)2 + a3 (0)4 + a4 (0)4 = 1
a0 = 1.
Differentiating once, plugging in x = 0, and solvinf for a1 ,
P 0 (x) = a1 + 2a2 x + 3a3 x2 + 4a4 x3
2 = a1 + 2a2 (0) + 3a3 (0)2 + 4a4 (0)3
a1 = 2.
Continuing with differentiating and plugging in x = 0, we get a2 = 23 , a3 = 23 , and a4 = 5
25
.
So,
3 2 5
P (x) = 1 + 2x + x2 + x3 + x4 .
2 3 24
Example. Construct a degree 4 polynomial that approximates ln (1 + x) at x = 0.

f (x) P (x) an
f (x) = ln (1 + x) P (x) = a0 + a1 x + a2 x2 + a3 x3 + a4 x4
f (0) = ln (1 + 0) = 0 P (0) = a0 a0 = 0
0 1 0
f (x) = 1+x P (x) = a1 + 2a2 x + 3a3 x2 + 4a4 x3
f 0 (0) = 1+01
=1 P 0 (0) = a1 a1 = 1
−1
f 00 (x) = (1+x)2 P 00 (x) = 2a2 + 6a3 x + 12a4 x2
−1 −1
f 00 (0) = (1+0)2 = −1 P 00 (x) = 2a2 a2 = 2
2
f (3) (x) = (1+x) 3 P (3) (x) = 6a3 + 24a4 x
2 1
f (3) (0) = (1+0) 3 = 2 P (3) (0) = 6a3 a3 = 3
−6
f (4) (x) = (1+x) 4 P (4) (x) = 24a4
−6 −1
f (4) (x) = (1+0) 4 = −6 P (4) (0) = 24a4 a4 = 4

So, our polynomial is

x2 x3 x4
P (x) = x −+ − .
2 3 4
Note that this polynomial exactly matches the power series we derived for ln (1 + x).
This polynomial is called the 4th order Taylor polynomial of ln (1 + x) at x = 0. The
series created from all order Taylor polynomials is called the Taylor series of ln (1 + x) at
x = 0.

8.2.2 Definition
Definition. Let f be a n times differentiable function where all derivatives exist at x = a.
The Taylor series for f at x = a is
∞
X f (k) (x) f 00 (a) f (n) (a)
(x − a)k = f (a) + f 0 (a)(x − a) + (x − a)2 + . . . + (x − a)n + . . . .
k=0
k! 2! n!

122
The nth partial sum of the Taylor Series,
n
X f (k) (x)
Pn (x) = (x − a)k
k=0
k!

is the Taylor polynomial of order n for f at x = a.

When a = 0 you might also hear Taylor series referred to as Maclaurin series. Like power
series, Taylor series have intervals of convergence.
Example. Find the Taylor series for ex at x = 0. Verify using term-by-term differentiation
that ex is its own derivative.
We know that ex is its own derivative, so f (k) (0) = e0 = 1 for all i. Applying the definition,
e0 e0 e0
ex = e0 + e0 (x − 0) + (x − 0)2 + (x − 0)3 + . . . + (x − 0)n + . . .
2! 3! n!
2 3 n
x x x
=1+x+ + + ... + + ....
2! 3! n!
Differentiating term-by-term,
d x x2 xn−1
e =1+x+ + ... + + ...
dx 2! (n − 1)!
we see that we get the same series, confirming that ex is its own derivative.

8.2.3 Common Maclaurin Series

∞
1 X  
= 1 + x + x2 + . . . = xk ,  x < 1
1−x k=0
∞
1 X  
= 1 − x + x2 − . . . = (−1)k xk , x < 1
1+x k=0
∞
x x2 X xk
e =1+x+ + ... = , all real x
2! k=0
k!
∞
x3 x5 X
k x
2k+1
sin x = x − + − ... = (−1) , all real x
3! 5! k=0
(2k + 1)!
∞
x2 x4 X 22k
cos x = 1 − + − ... = (−1)k , all real x
2! 4! k=0
(2k)!
∞
x2 x3 X xk+1  
ln (1 + x) = x − + − ... = (−1)k , x ≤1
2 3 k=0
k + 1
∞
x3 x5 X x2k+1  
arctan x = x − + − ... = (−1)k , x ≤ 1.
3 5 k=0
2k + 1

123
Euler’s Identity
You might have seen the identity eiπ + 1 = 0 or even eix = cos x + i sin x. Using our common
Taylor Series, we can derive this famous identity.

Starting with the Taylor series for eix ,

∞
ix (ix)2 (ix)3 (ix)4 (ix)5 X (ix)k
e = 1 + (ix) + + + + + ... = , all real x
2! 3! 4! 5! k=0
k!
∞
x2 x3 x4 x5 X xk
= 1 + ix + i2 + i3 + i4 + i5 + . . . = ik , all real x
2! 3! 4! 5! k=0
k!
x2 x3 x4 x5
= 1 + ix − −i + + i − . . . , all real x
2! 3! 4!  5!
x2 x4 x3 x5
 
= 1− + + ... + i x − + + ...
2! 4! 3! 5!
= cos x + i sin x.

Plugging in x = π,

eiπ = cos π + i sin π

= −1 + 0
eiπ + 1 = 0.

8.3 Approximation Error

Although knowing these infinite Taylor series is nice, if we want to use them to find function
values, we’ll need to approximate. This means only using some finite number of terms in the
Taylor series for our approximation. Because we’re not using all the terms, we’ll naturally
have some truncation error. We’d like to be able to say how big this truncation error is so
we can be sure that our approximation is good enough.

8.3.1 Alternating Series Estimation Theorem

Theorem (Alternating Series Estimation Theorem). Let s be a convergent, alternating (i.e
the sign of each term alternates) series where the terms of s are strictly decreasing. Then
the nth term truncation error is the same sign as and less than in absolute value the (n+1)th
term.

Example. Give a bound for the truncation error of using the first 10 terms of the Maclaurin
series of ln (1 + x) to approximate ln 2.

124
 The Maclaurin series for ln (1 + x) is
∞
x2 x3 X xk+1  
ln (1 + x) = x − + − ... = (−1)k , x ≤ 1,
2 3 k=0
k + 1

which is an alternating series. We see that to approximate ln 2, we’d use x = 1, which is

in the interval of convergence, so the series converges. The first missing term of the series
is 1/11. Let s10 be the partial sum of the first 10 terms when x = 1. By the Alternating
Series Estimation Theorem,
1
0 < ln 2 − s10 < .
11

8.3.2 Taylor’s Theorem

Although the Alternating Series Estimation Theorem is useful for quickly bounding the error
of alternating series, we’d like something that can apply more generally to all Taylor series.
Theorem (Taylor’s Theorem). Let f be a k + 1 times differentiable function on an open
interval I containing a. Then for all x in I,
f 00 (a) f (n) (a)
f (x) = f (a) + f 0 (a)(x − a) + (x − a)2 + . . . + (x − a)n + Rn (x).
2! n!
where  
 f (n+1) (c) 
x − an+1
 
Rn (x) =
(n + 1)!
for some c between x and a.
 
We can use this theorem to find the maximum value of Rn (x) over some interval. We can
also be more precise about what it means for a Taylor series to converge to some function
over some interval.
Definition. Let Rn (x) be the remainder of truncating after the degree n term in the Taylor
series for f centered at x = a. If for all x in some interval I containing a,
lim Rn (x) = 0,
n→∞

then we say the Taylor series for f at x = a converges to f on I.

Example. Show that the Maclaurin series for sin x converges to sin x for all real x.
We need to find the remainder and show that in the limit it goes to 0 as n grows large.
By Taylor’s Theorem,
 
 f (n+1) (c) 
x − an+1
 
Rn (x) =
(n + 1)!
 n+1
 x
≤ .
(n + 1)!

125
 The numerator is an exponential function, while the denominator is a factorial function,
so using nn FEPL,
 n+1
 x
0 ≤ lim Rn (x) ≤ lim = 0.
n→∞ n→∞ (n + 1)!

lim Rn (x) = 0.
n→∞

So, the Maclaurin series converges to sin x for all real x.

Remainder Estimation Theorem

Notice that we didn’t have to actually find the value of f (n+1) (c). We just had to find a
suitable upper bound where the limit would still go to 0.
Theorem (Remainder Estimation Theorem). If there are positive constants M and r such
that  (n+1) 
f (t) ≤ M r(n+1)
for all t between a and x, then Rn (x) satisfies the inequality

n+1 
n+1
 
Rn (x) ≤ M r x − a 
.
(n + 1)!
 
Example. Give a maximum error bound for using ln (1 + x) = x − x2 /2 when x ≤ 0.1.
We are using the second-order Taylor polynomial, so we need to find R2 (x). By Taylor’s
Theorem,
   
  f (2+1) (c)  2+1 f (3) (c)  3
R2 (x) = x − 0 =  x
(2 + 1)! 3!
(3) d3 2
f (x) = 3 ln (1 + x) =
dx (1 + x)3

 
On −0.1 ≤ x ≤ 0.1, f (3) (x) is maximal at (−0.1, 2000
729
).
  2000  3
R2 (x) ≤ 729 
x .
6
 3 1
On −0.1 ≤ x ≤ 0.1, x is maximal at (−0.1, 1000 ).
2000
1 1
≈ 4.572 × 10−4 .
  729
R2 (x) ≤ =
6 1000 2187
So, the error of approximating ln (1 + x) with x − x2 /2 when −0.1 ≤ x ≤ 0.1 is at most
4.572 × 10−4 .

126
8.4 Convergence
We have described what it means for a Taylor series to converge to a function over some
interval. What if we’re given an infinite series that’s not a function, just an infinite sum of
numbers? Can we still check if the series converges or diverges? We’ll develop several tests
that we can apply to check for convergence or divergence.

8.4.1 nth Term Test for Divergence

Lemma. Let an be the nth term of a series s. If

lim an 6= 0,
n→∞

then the series diverges.

Proof. Assume not. Let an be the nth term and sn the nth partial sum of a convergent
series s whose terms do not tend to 0. Since s converges to some value L, there exists some
positive integer m such that for all n > m and  > 0,
 
sn − L < .

We can also say the same for sn+1 .

 
sn+1 − L < .

So, subtracting one inequality from the other,

 
sn+1 − sn  < 2.

The difference between the two partial sums is just an+1 . So,
 
an+1  < 2.

Since the terms of s don’t tend to 0, there exists some positive integer h and real value δ > 0
such that for all n > h,  
an  > δ.
We can also say the same for an+1 .  
an+1  > δ.
Combining the two inequalities involving an+1 , for all n > max (m, h),
 
δ < an+1  < 2.

However, for  ≤ δ/2, the inequality creates a contradiction. 

Although the proof has to be a bit specific to cover the case of alternating series, the idea
behind the test makes sense. If you’re adding on terms that don’t get smaller in absolute
value, then you can’t ”zero in” on a particular value and converge.

127
Example. Show that the following series diverges:
∞
X
(−1)k = 1 − 1 + 1 − 1 + . . . .
k=0

We see that
lim an = DNE 6= 0.
n→∞

So, by the nth Term Test for Divergence, the series diverges.
 
8.4.2 Geometric Series with r < 1
Lemma.
  If s is a geometric series with common ratio r, then s converges if and only if
r < 1.

We already proved this when talking about power series and gave a formula to find its value.

Example. Show that the following series diverges:

∞
X 2k
.
k=0
3
 
This is a geometric series with initial term 1/3 and common ratio r = 2. Since r > 1,
the series diverges.

8.4.3 P-Series Test

Lemma. The p-series
∞
X 1
k=1
kp
converges if and only if p > 1.

Example. Show that the Harmonic Series diverges:

∞
X 1
.
k=1
k

This is a p-series with p = 1. So, by the P-Series Test, the sum diverges.

128
8.4.4 Direct Comparison Test
Lemma. Let s = ∞
P
P∞ak be a series where all ak ≥ 0. s converges if there exists some
k=0
convergent series c = k=0 ck and positive integer m such that for all n > m,
c n ≥ an .
P∞
s diverges if there exists some divergent series d = k=0 dk and positive integer m such that
for all n > m,
an ≥ d n .
Example. Show that the following series converges:
∞
X 1
.
k=0
2 + 3k

We see that this series looks really similar to a geometric series with common ratio 1/3,
just with an extra 2 in the denominator. For all n > 0,
1 1
n
≤ n.
2+3 3
So, by the Direct Comparison Test, the series converges.
Example. Show that the following series diverges:
∞
X 1
√ .
k=0
2+ k
Although the sum looks like a p-series with p = 1/2, we can use that series because our
series has smaller terms. Instead, we can compare to a p-series where p = 1/2.
1 1
0≤ ≥ √
n 2+ n
√
0≤2+ n≤n
n ≥ 4.
So, for all n ≥ 4, our series has larger values than the p-series with p = 1. We know the
p-series diverges by the P-Test, so by the Direct Comparison Test, our series also diverges.

8.4.5 Limit Comparison Test

Lemma. Let an and bn be the nth terms of two series a and b that have all positive terms
after some point. If
an
lim
n→∞ bn

converges to a finite value greater than 0, then a and b either both converge or both diverge.
If the limit converges to 0 and b converges, then a also converges. If the limit goes to ∞ and
b diverges, then a also diverges.

129
A common tactic is to select one of a or b to be a geometric or p-series.
Example. Show that the following series diverges:
∞
X 2k
.
k=2
k2 −k+a

Looking at this rational function, we see a degree 1 term in the numerator and a degree
2 term in the denominator. So, we might expect that the series behaves similarly to 1/n.
2n
n2 −n+a 2n2
lim 1 = lim = 2.
n→∞ n→∞ n2 − n + a
n

Since the limit converges to a finite value greater than 0, and we know that 1/n diverges
by the P-Test, then the series must also diverge by the Limit Comparison Test.
Example. Show that the following series converges:
∞
X 1
.
i=1
2i − 1

This series looks similar to the geometric series 1/2n .

1
2n 2n − 1
lim = lim = 1.
n→∞ n1 n→∞ 2n
2 −1

Since the limit converges to a finite value greater than 0, and we know that 1/2n converges
because it’s a geometric series with r = 1/2, then the series must also converge the the
Limit Comparison Test.

8.4.6 Integral Test

Lemma. Let a be a sequence of positive terms where an = f (n). If f is continuous, positive
after some m, and decreasing, then the series
∞
X
ak
k=m

converges if and only if the integral Z ∞

f (x)dx
m
converges.
Example. Show that the following series diverges:
∞
X 2k
.
k=1
k2 +1

130
 The function f (x) = 2x/(x2 + 1) is continuous, positive for all x ≥ 1, and decreasing.
Z ∞ Z ∞ ∞
2x 2 du 
2+1
dx = (u = x + 1) = ln u  = diverges.
1 x 2 u 
2

So, by the Integral Test, the series also diverges.

Example. Show that the following series converges:

∞
X 1
.
k=1
k2 +1

The function f (x) = 1/(x2 + 1) is continuous, positive for all x ≥ 1, and decreasing.
Z ∞ ∞
dx  = π − π = π.

2
= arctan x
1 x +1 
1 2 4 4

So, by the Integral Test3 the series also converges.

8.4.7 Ratio Test

Lemma. Let a be a series with only positive terms after some index. If the limit
an+1
lim
n→∞ an
is less than 1, then the series converges. If the limit is greater than 1, then the series diverges.
If the limit is equal to 1, then the test is inconclusive.

Example. State whether the following series converges or diverges:

∞
X k ln k
.
k=1
2k

Taking the limit of the ratio of subsequent terms,

(n+1) ln (n+1)
2n+1 (n + 1) ln n + 1 n+1 1
lim n ln n
= lim = lim = .
n→∞
2n
n→∞ 2n ln n n→∞ 2n 2

Since the limit is less than 1, the series converges by the Ratio Test.

3
We omitted evaluating the improper integral with limits, but the value is what you would get from doing
that.

131
8.4.8 nth Root Test
Lemma. Let a be a series with all positive terms after some index. If the limit
√
lim n
an
n→∞

is less than 1, then the series converges. If the limit is greater than 1, the series diverges. If
the limit is equal to 1, the test is inconclusive.

This test is most useful for series that looks like geometric series, but the common ratio is
not a constant.

Example. State whether the following series converges or diverges:

∞  k
X k
.
k=1
2k − 1

Taking the limit of the nth root,

s n
n n n 1
lim = lim = .
n→∞ 2n − 1 n→∞ 2n − 1 2

Since the limit is less than 1, the series converges by the nth Root Test.

8.4.9 Alternating Series Test

All of the previous 8 tests have tested whether a series converges absolutely. That is, all
terms in the series could be made positive and the series would still converge or diverge.
There are some series that do converge but don’t converge absolutely. We say that these
series converge conditionally. You’ve already seen few, like the formula for π/4 using the
Maclaurin series for arctan x. All series that converge absolutely also converge conditionally.
Series that converge conditionally do so when they pass the following test.

Lemma. The series ∞

X
(−1)k uk
k=0

converges if all of the following conditions are satisfied.

1. All ui are positive.

2. There exists an integer m such that for all n > m, un+1 ≤ un .

3. The series of u’s pass the nth Term Test.

132
 Example. State whether the following series converges or diverges:
∞
X 1
(−1)k .
k=1
k

1. All the terms 1, 1/2, 1/3, . . . are positive.

2. All terms are less than the previous term.
3. The terms tend to 0, passing the nth Term Test.
So, by the Alternating Series Test, the series converges conditionally4 . Note that the
series does not converge absolutely because it fails the P-Test.

Riemann Rearrangement Theorem

Theorem (Riemann Rearrangement Theorem). If an infinite series is conditionally con-
vergent, but not absolutely convergent, then its terms can be rearranged to form a divergent
series or to converge to any constant.
This result may seem counter intuitive, given that we know addition is associative and
commutative. However, the result stems from the fact that we defined convergence of an
infinite series at the limit of partial sums. If we rearrange terms of the series, the partial
sums and their limits can change, meaning the limit of partial sums can diverge or converge
to some other value.

8.5 Radius of Convergence

We’ve established some tests to determine if an infinite series is or isn’t convergent, but like
we saw with geometric and Taylor series, it’s possible that a series only converges over some
interval. We’d like to be able to find this interval so we know if we can rightly use Taylor
series to approximate a function.

8.5.1 Convergence Theorem for Power Series

Theorem (Convergence Theorem for Power Series). There are three possibilities for any
power series of the form
X∞
ck (x − a)k
k=0

with respect to convergence.

1. The power series converges, but only on some finite interval centered at x = a. That
is, there is a positive real number r such that the series converges for x − a < r and
diverges otherwise. The series may or may not converge at the endpoints.
4
You’ve already seen this series too. It converges to ln 2.

133
 2. The power series converges for all real numbers.

3. The power series converge only at x = a and diverges elsewhere (i.e r = 0).

We call this value r the radius of convergence.

Generally, we start by applying the ratio test to determine where the series converges ab-
solutely. If we find the series converges for all real numbers of just at x = a, we are done.
Otherwise the series converges for some finite interval and we need to apply a different test
to determine convergence at endpoints.

Example. Determine the radius of convergence for the following series:

∞
X kxk
.
k=0
10k

Since this series isn’t quite geometric enough for an nth root test but does have all poitive
terms, it seems most suited for a ratio test.
(n+1)xn+1
n+1 (n + 1)xn+1 (n + 1)x x
lim 10nxn = lim n
= lim = .
n→∞
10n
n→∞ 10nx n→∞ 10n 10

x can be positive or negative. Recall that with a ratio test, the series converges if the
limit is less than 1.
x
 <1
10
 
x < 10.

So, the radius of convergence is 10. Note that this analysis doesn’t tell us for sure of the
series converges for x = ±10 (it happens to not converge in this case).

Example. Determine the radius of convergence for the following series:

∞
X
k!xk .
k=0

We can again use a ratio test.

(
(n + 1)!xn+1 ∞ x 6= 0
lim = lim (n + 1)x = .
n→∞ n!xn n→∞ 0 x=0

We see that the only time the limit is less than 1 is when x = 0. So, the radius of
convergence is 0.

134
 Example. Determine the radius of convergence for the following series5 :
∞
X xk
.
k=0
k!

Applying the ratio test,

xn+1
(n+1)! x
lim n = lim = 0.
n→∞ x n→∞ n + 1
n!
Since the limit is less than 1 everywhere, the series converges for all real x.

8.5.2 Convergence at Endpoints

If a series only converges over some finite interval, our method of applying the ratio test
is inconclusive. Instead, we plug in the endpoint value for x into our series and apply a
different test, which usually depends on the specific series we’re working with. The most
common tests to use are the direct comparison test, the limit comparison test, the integral
test, and the alternating series test.
 
Example. Given that the following series converges by the ratio test for x < 10, determine
convergence at the endpoints.
∞
X kxk
.
k=0
10k
Plugging in x = ±10,
∞ ∞
X k(±10)k X
= (±1)k k.
k=0
10k k=0

  by the nth term test for both x = 10 and x = −10. So the radius of
This series diverges
convergence is x < 10.
Example. For what values of x does the following series converge:
∞
X x2k
(−1)k+1 .
k=1
2k

Applying the ratio test to the absolute series,

x2n+2
2nx2
lim 2n+2
2n = lim = x2 .
n→∞ x n→∞ 2n + 2
2n
 
So, by the ratio test, the series converges absolutely when x < 1. The series is the same
at x = 1 and x = −1.
∞
X (−1)k+1
.
k=0
2k
5
You might recognize this as the Maclaurin series for ex .

135
 Since the series, disregarding the (−1)k+1 has all terms positive and decreasing and passes
the nth term test, it converges by the alternating series test. Therefore the interval of
convergence is [−1, 1].
It’s important to remember that although a series might converge over some interval it might
do so very slowly, especially at endpoints. That is why when estimating a function value
using a Taylor series we still need to estimate the error.

136
 Chapter 9

Additional Materials

9.1 Tests
The following are worked free-response questions from three years of AP Calculus BC tests1 .

9.1.1 2018 Free-Response Questions

Questions 1 and 2 are part of the same section and are allotted 30 minutes for completion
with the aid of a graphing calculator. Question 3 through 6 are part of the same section are
allotted 1 hour for completion without the aid of a graphing calculator.

1. People enter a line for an escalator at rate modeled by the function r given by
(
t 3 t 7



44 100 1 − 300 for 0 ≤ t ≤ 300
r(t) =
0 for t > 300,

where r(t) is measured in people per scond and t is measured in seconds. As people
get on to the escalator, they exit the line at a rate of 0.7 person per second, There are
20 people in line at time t = 0.
(a) How many people entered the line for the escalator during the time interval 0 ≤
t ≤ 300?
(b) During the time interval 0 ≤ t ≤ 300, there are always people in line for the
escalator. How many people are in line at time t = 300?
(c) For time t > 300, what is the first time t that there are no people in line for the
escalator?
(d) For time 0 ≤ t ≤ 300, at what time t is the number of people in line at a
minimum? To the nearest whole number, find the number of people in line at this
time. Justify your answer.
1
These tests questions are owned by the College Board. The questions and College Board’s sample responses
are available on their website.

137
2. Researchers on a boat are investigating plankton cells in a sea. At a depth of h
meters, the density of plankton cells, in millions of cells per cubic meter, is modeled
2
by p(h) = 0.2h2 e−0.0025h for 0 ≤ h ≤ 30 and is modeled by f (h) for h ≥ 30. The
continuous function f is not explicitly given.

(a) Find p0 (25). Using correct units, interpret the meaning of p0 (25) in the context of
the problem.
(b) Consider a vertical column of water in this sea with horizontal cross sections of
constant area 3 square meters. To the nearest million, how many plankton cells
are in this column of water between h = 0 and h = 30 meters?
R∞
(c) There is a function u such that 0 ≤ f (h) ≤ u(h) for all h ≥ 30 and 30 u(h)dh =
105. The column of water in part (b) is K meters deep, where K > 30. Write
an expression involving one or more integrals that gives the number of plankton
cells in the entire column. Explain why this number of plankton cells is less than
or equal to 2000 million.
(d) The boat is moving on the surface of the sea. At time t ≥ 0, the position of
the boat is (x(t), y(t)), where x0 (t) = 662 sin (5t) and y 0 (t) = 880 cos (6t). Time t
is measured in hours, and x(t) and y(t) are measured in meters. Find the total
distance traveled by the boat over the time interval 0 ≤ t ≤ 1.

Figure 9.1: AP Calculus BC 2018 Exam Free-Response Question 3, Graph of g

3. The graph of the continuous function g, the derivative of a function f , is shown above.
The function g is piecewise linear for −5 ≤ x < 3 and g(x) = 2(x − 4)2 for 3 ≤ x ≤ 6.

138
 (a) If f (1) = 3, what is the value of f (−5)?
R6
(b) Evaluate 1 g(x)dx.
(c) For −5 < x < 6, on what open intervals, if any, is the graph of f both increasing
and concave up? Give a reason for your answer.
(d) Find the x-coordinate of each point of inflection of the graph of f . Give a reason
for your answer.

t (years) 2 3 5 7 10
H(t) (meters) 1.5 2 6 11 15

4. The height of a tree at time t is given by a twice-differentiable function H, where H(t)

is measured in meters and t is measured in years. Selected values of H(t) are given in
the table above.

(a) Use the data in the table to estimate H 0 (6). Using correct units, interpret the
meaning of H 0 (6) in the context of the problem.
(b) Explain why there must be at least one time t, for 2 ≤ t ≤ 10 such that H 0 (t) = 2.
(c) Use a trapezoidal sum with 4 subintervals indicated by the data in the table to
approximate the average height of the tree over the time interval 2 ≤ 2 ≤ 10.
(d) The height of the tree, in meters, can also be modeled by the function G, given
by G(x) = 100x
1+x
, where x is the diameter of the base of the tree, in meters. When
the tree is 50 meters tall, the diameter of the base of the tree is increasing at a
rate of 0.03 meters per year. According to this model, what is the rate of the
change of the height of the tree with respect to time, in meters per year, at the
time when the tree is 50 meters tall?

139
 Figure 9.2: AP Calculus BC 2018 Exam Free-Response Question 5

5. The graph of the polar curves r = 4 and r = 3 + 2 cos θ are shown in the figure above.
The curves intersect at θ = π3 and θ = 5π
3
.

(a) Let R be the shaded region inside the graph of r = 4 and outside the graph of
r = 3 + 2 cos θ, as shown in the figure above. Write an expression involving an
integral for the area of R.
(b) Find the slope of the line tangent to the graph r = 3 + 2 cos θ at θ = π2 .
(c) A particle moves along the portion of the curve r = 3 + 2 cos θ for 0 < θ < π2 .
The particle moves in such a way that the distance between the particle and the
origin increases at a constant rate of 3 units per second. Find the rate at which
the angle θ changes with respect to time at the instant when the position of the
particle corresponds to θ = π3 . Indicate units of measure.

6. The Maclaurin series for ln (1 + x) is given by

x2 x3 x4 xn
x− + − + . . . + (−1)n+1 + . . . .
2 3 4 n
On its interval of convergence,

this series converges to ln (1 + x). Let f be a function
x
defined by f (x) = x ln 1 + 3 .

(a) Write the first four nonzero erms and the general term of the Maclaurin series for
f.
(b) Determine the interval of convergence for f . Show the work that leads to your
answer.

140
 (c) Let P4 (x) be the fourth-degree Taylor polynomial for f about x = 0. Use the
alternating series estimation bound to find an upper bound for P4 (2) − f (2).

9.1.2 2018 Free-Response Answers

1. (a) r(t) tells us the rate at which people enter the line. By integrating r(t) over the
given interval, we can find the total number of people who entered the line. Using
a calculator,
Z 300 Z 300  3  7
t t
r(t)dt = 44 1− dt = 270 people.
0 0 100 300
(b) We know that at t = 0, there are already 20 people in line. Combining this fact
with our answer from part (a), we know that 270 + 20 = 290 people entered the
line in the time interval. We also know that people leave the line at a constant
rate of 0.7 people per second. Integrating this rate over the time interval will give
us the number of people who left the line.
Z 300
0.7dt = 210 people.
0
So, there are 290 − 210 = 80 people in line at t = 30.
(c) We know from our answer in (b) that there are 80 people in line at t = 300. We
see from r(t) that no more people are entering the line when t > 300. So, the
only thing that contributes to changing the number of people in line is people
constantly leaving at a rate of 0.7 people per second.
80 people
300s + ≈ 414.286s.
0.7 people/s
(d) Combining the fact that there are initially 20 people in line, the inflow rate is
given by r(t) and the outflow rate is 0.7, we can get that the number of people in
line at time x is modeled by
Z x
20 + (r(t) − 0.7) dt.
0
Finding when the derivative is 0,
r(t) − 0.7 = 0 =⇒ t = 33.013 or 166.575.
Finding the number of people in line at these times and the endpoints 0 and 300,
people0 = 20
people33.013 = 3.803
people166.575 = 158.070
people300 = 80,
we see that the fewest number of people occurs when t ≈ 33.013s.

141
2. (a) Using a calculator, we see that p0 (25) = −1.17906. In the context of the problem,
this value means that at a depth of 25 meters, the density of plankton is decreasing
at a rate of 1.17906 million plankton per cubic meter per meter.
(b) The vertical density of the plankton in the column is given by p(h) ∗ 3m2 . In-
tegrating this density function for 0 ≤ h ≤ 30m, we’ll get the total number of
plankton on the column.
Z 30
3p(h)dh ≈ 1675.414.
0

So, there are 1675.414 million plankton in the column from 0 to 30 meters depth.
(c) The total number of plankton in the entire vertical column with cross section area
Am2 is given by Z ∞
P = Adensity(h)dh.
0

We can break this integral up, and we can use A = 3m2 for our particular column.
Z 30 Z ∞
P =3 p(h)dh + 3 f (h)dh.
0 30

Since u(h) ≥ f (h) for h ≥ 30, we can write the following inequality, the right side
of which we can numerically evaluate:
Z 30 Z ∞
P ≤3 p(h)dh + 3 u(h)dh
0 0
≤ 1675.414 + 3(105)
≤ 1990.414.

So, we see that the number of plankton in the entire vertical column is at most
1990.414 million, which is strictly less than 2000 million.
(d) Since the position of the boat is parametric, we can use the parametric arc length
formula to find the total distance traveled for 0 ≤ t ≤ 1hr.
Z 1p
D= (x0 (t))2 + (y 0 (t))2 dt
0
Z 1q
= (662 sin (5t))2 + (880 cos (6t))2 dt
0
≈ 757.456.

So, for 0 ≤ t ≤ 1hr, the boat travels a distance of 757.456 meters.

3. (a) Trying to treat this like an initial value problem would be too tedious with all
the piecewise parts of g. Plus, it would be doing more than the question asked

142
 because it only wants the value of f at a particular point, not an expression
for f . Instead, we can use the fact that g is the derivative of f and apply the
Fundamental Theorem of Calculus.
Z 1
g(x)dx = f (1) − f (−5) = 3 − f (−5).
−5

Evaluating the integral geometrically,

Z 1
3 −19
g(x)dx = −(9 + ) + 1 = .
−5 2 2

Solving for f (−5),

−19 25
= 3 − f (−5) =⇒ f (−5) = .
2 2
(b) Evaluating the integral,
Z 6 Z 3 Z 6
g(x)dx = 2(x − 4)2 dx
2dx +
1 1 3
3 6
 2 3

= 2x + (x − 4) 
3 1 3
18
=4+
3
= 10.

(c) f is increasing when its first derivative is positive and concave up when its second
derivative is positive. Since g is the derivative of f , we can also say that f is
increasing when g is positive and concave up when g is increasing. g is positive on
(0, 4) ∪ (4, 6]. g is increasing on (−2, −1) ∪ (0, 1) ∪ (4, 6). Taking the intersection
of these intervals, we see that f is increasing and concave up on (0, 1) ∪ (4, 6).
(d) f has an inflection point when its second derivative is equal to 0, changing sign.
Since g is the derivative of f , we can also say that f has an inflection point when
the derivative of g is equal to 0, changing sign. Although we see by looking at the
graph that the derivative of g is 0 over several intervals, it does not change sign.
The only place the derivative is 0 and changes sign is at x = 4.

4. (a) We can approximate H 0 (6) as the average rate of change between t = 5 and t = 7.

H(7) − H(5) 11 − 6 5
H 0 (6) ≈ = = .
7−5 7−5 2
In the context of this problem H 0 (6) ≈= 5
2
means that at 6 years, the tree is
growing at a rate of 5/2 meters per year.

143
 (b) Looking at the average rate of change between t = 3 and t = 5,
H(5) − H(3) 6−2
H̄ 0 3,5 = = = 2.
5−3 5−3
Since H is differentiable, it is also continuous. So, by the mean value theorem,
there is at least one point 3 ≤ c ≤ 5 (which is a sub-interval of 2 ≤ t ≤ 10) such
that H 0 (c) = 2.
(c) We know that the average value of a continous function f over some interval [a, b]
is Z b
1
f¯a,b = f (x)dx.
b−a a
So, over our interval of [2, 10], we need to approximate this integral using trape-
zoids. Because the sub-interval widths given by the table are not equally-sized,
we can’t apply the shortcut trapezoidal rule. However, we can still approximate
the area using trapezoids.
Z 10  
1 1 1.5 + 2 2+6 6 + 11 11 + 15 1
H(x)dx ≈ 1+ 2+ 2+ 3 = (65.75) = 8.21875.
10 − 2 2 8 2 2 2 2 8
So, the average height of the tree between 2 and 10 years is approximately 8.21875
meters.
(d) Finding the diameter of the base of the tree when it is 50 meters tall according
to G.
100x
50 = =⇒ x = 1.
1+x
Differentiating G, remembering the chain rule,
100(1 + x) dx − 100x dx
G0 (x) = dt dt
.
(1 + x)2
dx
Plugging in x = 1 and dt
= 0.03,
100(1 + 1)(0.03) − 100(1)(0.03) 6−3 3
G0 (1) = 2
= = .
(1 + 1) 4 4
So, according to G, the tree is growing at 3/4 meters per year when it is 50 meters
tall.
5. (a) We know the formula for the area between two polar curves is
1 β 2
Z
f (θ) − g 2 (θ) dθ.


A=
2 α
Since we want the area inside the circle and outside the limaçon for π/3 ≤ θ ≤
5π/3,
1 5π/3
Z
(4)2 − (3 + 2 cos θ)2 dθ.


A=
2 π/3

144
 (b) Remembering that x = r cos θ and y = r sin θ,
x = (3 + 2 cos θ) cos θ, y = (3 + 2 cos θ) sin θ.
Differentiating x and y with respect to θ,
dx dy
= −(3 + 2 cos θ) sin θ + cos θ(−2 sin θ), = (3 + 2 cos θ) cos θ + sin θ(−2 sin θ).
dθ dθ
Dividing one by the other,
dy
dy dθ (3 + 2 cos θ) cos θ + sin θ(−2 sin θ)
= dx
= .
dx dθ
−(3 + 2 cos θ) sin θ + cos θ(−2 sin θ)
At θ = π/2,
dy (3 + 2 · 0) · 0 + 1(−2 · 1) −2 2
= = = .
dx θ=π/2 −(3 + 2 · 0)1 + 0(−2 · 1) −3 3
(c) r is a function of θ, and we’re being told that θ behaves as a function of t, due to
how the particle moves. So, we can apply the chain rule:
dr dr dθ
= ·
dt dθ dt
dθ
= −2 sin θ .
dt
dθ
Solving for dt
,
dθ dr 1
= .
dt dt −2 sin θ
Moving away from the origin at some rate tell us how r changes. Plugging in
dr
dt
= 3 and θ = π/3,
dθ 1 3 √
=3 = √ = − 3.
dt −2 sin (π/3) − 3
√
So, θ is changing at a rate of − 3 radians per second.
6. (a) Starting with the given Maclaurin series,
u2 u3 u4 un
ln (1 + u) = u − + − + . . . + (−1)n+1 + . . . .
2 3 4 n
Substituting u = x/3,
 x  x ( x3 )4 ( x3 )4 ( x3 )4 ( x )n
ln 1 + = − + − + . . . + (−1)n+1 3 + . . . .
3 3 2 3 4 n
Multiplying by x,
 x x ( x )4 ( x )4 ( x )4 ( x )n
x ln 1 + = x − x 3 + x 3 − x 3 + . . . + (−1)n+1 x 3 + . . . .
3 3 2 3 4 n
145
 (b) Applying the ratio test on the absolute terms,
( x3 )n+1
x n+1 n x3 x
lim ( x3 )n
= lim = .
n→∞ n→∞ n + 1 3
x n

The ratio test tells

  us the series converges when the limit is less than 1, so we can
safely say that x < 3, and we still need to test the endpoints. When x = −3 the
series becomes ∞ −3 n ∞


X
n+1 3
X 3
(−1) (−3) =
n=1
n n=1
n
which diverges by the P-Test. When x = 3, the series becomes
∞ 3 n ∞


X
n+1
X 3
(−1) 3 3
= (−1)n+1
n=1
n n=1
n

which converges by the Alternating Series Test. So, the interval of convergence is
−3 < x ≤ 3.
(c) The Alternating Series Estimation Theorem tells us that the error from an alter-
nating series is at most the value of the first term not included, and the error is
the same sign as the first not included term. For P4 (x),
2 4


P4 (2) − f (2) ≤ (−1)5 2 3  = 8 .
  
4 81
So, our upper bound on the error is 8/81.

9.1.3 2017 Free-Response Questions

Questions 1 and 2 are part of the same section and are allotted 30 minutes for completion
with the aid of a graphing calculator. Questions 3 through 6 are part of the same section
and are allotted 1 hour for completion without the aid of a graphing calculator.

h (feet) 0 2 5 10
A(h) (square feet) 50.3 14.4 6.5 2.9

1. A tank has a height of 10 feet. The are of the horizontal cross section of the tank at h
feet is given by the function A, where A(h) is measured in square feet. The function A
is continuous and decreases as h increases. Selected values for h are given in the table
above.

(a) Use a left Riemann sum with three subintervals indicated by the data to approx-
imate the volume of the tank. Indicate units of measure.

146
 (b) Does the approximate in part (a) overestimate or underestimate the volume of
the tank? Explain your reasoning.
(c) The area, in square feet, of the horizontal cross section at height h feet is modeled
50.3
by the function f given by f (h) = e0.2h +h
. Based on this model, find the volume
of the tank. Indicate units of measure.
(d) Water is pumped into the tank. When the height of the water is 5 feet, the height
is increasing at a rate of 0.26 foot per minute. Using the model from part (c),
find the rate at which the volume of water is changing with respect to time when
the height of the water is 5 feet. Indicate units of measure.

Figure 9.3: AP Calculus BC 2017 Exam Free-Response Question 2

2. The figure above show the polar curves r = f (θ) = 1 + sin θ cos (2θ) and r = g(θ) =
2 cos θ for 0 ≤ θ ≤ π2 . Let R be the region in the first quadrant bounded by the curve
r = f (θ) and the x-axis. Let S be the region in the first quadrant bounded by the
curve r = f (θ) the curve r = g(θ), and the x-axis.

(a) Find the area of R.

(b) The ray θ = k, where 0 < k < π2 , divides S into two regions of equal area. Write
out, but do not solve, an equation involving one or more integrals whose solution
gives the value of k.
(c) For each θ, 0 ≤ θ ≤ π2 , let w(θ) be the distance between the points with polar
coordinates (f (θ), θ) and (g(θ), θ). Write an expression for w(θ). Find wA , the
average value of w(θ) over the interval 0 ≤ θ ≤ π2 .
(d) Using the information given from part (c), find the value of θ for which w(θ) = wA .
Is the function w(θ) increasing or decreasing at that value of θ? Give a reason for
your answer.

147
 Figure 9.4: AP Calculus BC 2017 Exam Free-Response Question 3, Graph of f 0

3. The function f is differentiable on the closed interval [−6, 5] and satisfies f (−2) = 7.
The graph of f 0 , the derivative of f , consists of a semicircle and three line segments,
as shown in the figure above.

(a) Find the values of f (−6) and f (5).

(b) On what intervals is f increasing? Justify your answer.
(c) Find the absolute minimum value of f on the closed interval [−6, 5]. Justify your
answer.
(d) For each of f 00 (−5) and f 00 (3), find the value or explain why it doesn’t exist.

4. At time t = 0, a boiled potato is taken from a pot on a stove and left to cool in a
kitchen. The internal temperature of the potato is 91 degrees Celsius (◦ C) at time
t = 0, and the internal temperature of the potato is greater than 27◦ C for all time
t > 0. The internal temperature of the potato at time t minutes can be modeled by
a function H that satisfies the differential equation dH
dt
= − 14 (H − 27), where H(t) is
measured in degrees Celsius and H(0) = 91.

(a) Write an equation for the line tangent to the graph of H at t = 0. Use this
equation to approximate the internal temperature of the potato at time t = 3.
2
(b) Use ddtH2 to determine whether your answer in part (a) is an underestimate or
overestimate of the internal temperature of the potato at time t = 3.
(c) For t < 10, an alternate model for the internal temperature of the potato at time t
minutes is the function G that satisfies the differential equation dG
dt
= −(G−27)2/3 ,
where G(t) is measured in degrees Celsius and G(0) = 91. Based on this model,
what is the internal temperature of the potato at time t = 3?
3
5. Let f be the function defined by f (x) = 2x2 −7x+5
.

148
 (a) Find the slope of the tangent line of the graph of f at x = 3.
(b) Find the x-coordinate of each critical point of f in the interval 1 < x < 2.5. Clas-
sify each critical point as the location of a relative minimum, a relative maximum,
or neither. Justify your answer.
3 2 1
R∞
(c) Using the identity that 2x2 −7x+5 = 2x−5 − x−1 , evaluate 5 f (x)dx or show that
the integral diverges.
P∞ 3
(d) Determine whether the series n=5 2n2 −7n+5 converges or diverges. State the
conditions of the test used for determining convergence or divergence.

f (0) = 0
f 0 (0) = 1
f (n+1) (0) = −n · f (n) (0) for all n ≥ 1

6. A function f has derivatives of all order for −1 < x < 1. The derivatives
 of f satisfy
the conditions above. The Maclaurin Series for f converges to f (x) for x < 1.
2
(a) Show that the first four non-zero terms of the Maclaurin series for f are x − x2 +
x3 4
3
− x4 , and write the general term for the Maclaurin series for f .
(b) Determine whether the Maclaurin series described in part (a) converges absolutely,
converges conditionally, or diverges at x = 1. Explain your reasoning.
(c) Write the first
R x four nonzero terms and the general term for the Maclaurin series
for g(x) = 0 f (t)dt.
(d) Let Pn 21 represent the nth degree Taylor polynomial for g about x = 0 and

evaluated at x = 21 , where g is the function defined in part (c). Use the alternating
series error bound to show that
    
 P4 1 − g 1  < 1 .
 
 2 2  500

9.1.4 2017 Free-Response Answers

1. (a) Using a left Riemann sum,
Z 10
A(h)dh ≈ (2 − 0)50.3 + (5 − 2)14.4 + (10 − 5)6.5 = 100.6 + 43.2 + 32.5 = 176.3.
0

So, we approximate that the volume of the tank is 176.3 cubic feet.
(b) Since we are given that the function is decreasing over the interval, a left Riemann
sum will overestimate the volume.

149
 (c) Integrating f from h = 0 to h = 10,
Z 10
50.3
0.2h
dh ≈ 101.325.
0 e +h
So, the volume of the tank given by f is 101.325 cubic feet.
(d) We know from (c) that
Z h
V (h) = f (x)dx.
0
Differentiating with respect to t, and applying the chain rule,
dV dh
= f (h) · .
dt dt
dh
Plugging in h = 5 and dt
= 0.26,
dV 50.3
= · 0.26 ≈ 1.694.
dt h=5 e1 + 5
So, when h = 5 feet, the volume is changing at a rate of 1.694 cubic feet per
minute.
2. (a) Finding the area enclosed by f from θ = 0 to θ = π/2,
1 π/2
Z
A= (1 + sin θ cos (2θ))2 dθ ≈ 0.648.
2 0
(b) Our ray θ = k will represent a lower bound in one integral and an upper bound
in the other. Finding equal areas between g and f ,
1 k 1 π/2
Z Z
2 2

(2 cos θ)2 − (1 + sin θ cos (2θ))2 dθ.


(2 cos θ) − (1 + sin θ cos (2θ)) dθ =
2 0 2 k
(c) Since both f and g are polar functions evaluated at θ, the distance between then
will simply be the difference in their radii.
w(θ) = g(θ) − f (θ).
Finding the average of w,
Z π/2
1
wA = (2 cos θ − (1 + sin θ cos (2θ)))dθ ≈ 0.485.
π/2 − 0 0
(d) Solving w(θ) = wA ,
2 cos θ − (1 + sin θ cos (2θ)) = 0.485 =⇒ θ ≈ 0.518.
Evaluating w0 (0.518),
w0 (0.518) ≈ −0.581.
So, w is decreasing at this value.

150
 3. (a) Applying the Fundamental Theorem of Calculus,
Z −2
f 0 (x)dx = f (−2) − f (−6) = 7 − f (−6) = 4 =⇒ f (−6) = 3
−6
Z 5
f 0 (x)dx = f (5) − f (−2) = f (5) − 7 = −2π + 3 =⇒ f (5) = 10 − 2π.
−2

(b) f is increasing when its derivative is positive. Looking at the graph of f 0 , we see
it is positive on [−6, 2) ∪ (2, 5). So, f is increasing on [−6, 2] ∪ [2, 5]2 .
(c) We see that the critical points of f 0 are x = −2 and x = 2. A minimum can occur
at a left/right endpoint that will/was decreasing. So, the two points we need to
consider for the absolute minimum are x = −6 and x = 2. We know from (a)
that f (−6) = 3. Applying the Fundamental Theorem of Calculus again,
Z 2
f 0 (x)dx = f (2) − f (−2) = f (2) − 7 = 2π =⇒ f (2) = 7 − 2π.
−2

Since 7 − 2π < 3, the absolute minimum of f is at x = 2.

(d) f 00 (−5) = − 21 because f is continuous at -5, and the left and right hand limits are
both the slope of the line, which is − 12 . f 00 (3) = DNE. The left hand limit is the
slope of the left hand line, which is 2, and the right hand limit is the left of the
right hand line, which is -1. Since the limits do not agree, f 0 is not continuous,
and hence not differentiable at x = 3.
4. (a) Writing the line in point-slope form and then converting to standard form,
y − y0 = m(t − t0 )
dH
y − H(0) = (t − 0)
dt t=0
y − 91 = −16t
y = −16t + 91.
Using this line to approximate H(3),
y = −16(3) + 91 = 43.
So, the tangent line at t = 0 approximates H(3) to be 43◦ C.
dH
(b) Taking the derivative of dt
,
d2 H 1
2
=− .
dt 4
Since the graph is concave down at every point, the tangent line at t = 0 gives an
overestimate of H(3).
2
I personally don’t think f is “increasing” on the open endpoints where the derivative is 0. I think it’s neither
increasing nor decreasing. However, the test writers and graders feel differently, and I can’t say their view
is necessarily wrong.

151
 (c) This is a separable differential equation.
dG
= −(G − 27)2/3
dt
dG
= −1dt
(G − 27)2/3
Z Z
dG
= −1dt
(G − 27)2/3
3(G − 27)1/3 = −t + C
(G − 27)1/3 = −t/3 + C
G − 27 = (−t/3 + C)3
G = (−t/3 + C)3 + 27.

Solving for C using G(0) = 91,

91 = (−0/3 + C)2 + 27
64 = C 3
C = 4.

So, our overall solution for G is

G(t) = (4 − t/3)3 + 27.

Evaluating at t = 3,

G(3) = (4 − 3/3)3 + 27 = 33 + 27 = 54.

So, the internal temperature of the potato at t = 3 minutes is 54◦ C.

5. (a) Applying the quotient rule,
(2x2 − 7x + 5)(0) − 3(4x − 7) −12x + 21
f 0 (x) = 2 2
= .
(2x − 7x + 5) (2x2 − 7x + 5)2
Evaluating at x = 3,
−12(3) + 21 15 15
f 0 (3) = 2 2
=− 2
=− .
(2(3) − 7(3) + 5) (18 − 21 + 5) 4

(b) f 0 (x) = 0 only when the numerator is 0,

7
−12x + 21 = 0 =⇒ x = .
4
f 0 is negative to the left of 74 and positive to the right of it. Therefore, x = 7
4
is a
relative maximum by the first derivative test.

152
 (c) Evaluating the limit using the given partial fraction decomposition,
Z ∞ Z ∞ 
2 1
f (x)dx = − dx
5 5 2x − 5 x − 1
∞

= ln (2x − 5) − ln (x − 1)
5
 ∞
2x − 5 
= ln
x − 1 5
   
2b − 5 5
= lim ln − ln
b→∞ b−1 4
 
5
= ln 2 − ln
4
 
8
= ln .
5

(d) For n ≥ 5, f (n) is positive and decreasing. Using our work from part (c), we
know that the integral of f from 5 to ∞ converges. So, by the Integral test, the
series also converges.

6. (a) Calculating the first four derivatives and the general derivative,

f (0) = 0
f 0 (0) = 1
f 00 (0) = −1 · f 0 (0) = −1
f (3) (0) = −2 · f 00 (0) = 2
f (4) (0) = −3 · f (3) (0) = −6.f (n+1) (0) = −n · f (n) (0) = (−1)n+1 n!.

Applying the Maclaurin series formula,

f (0) f 0 (0) f 00 (0) 2 f (n) (0) n

Pn (x) = + x+ x + ... + x
0! 1! 2! n!
0 1 −1 2 2 3 −6 4 (−1)n+1 (n − 1)! n
= + x+ x + x + x + ... + x
1 1 2 6 24 n!
x2 x3 x4 xn
=x− + − + . . . + (−1)n+1 .
2 3 4 n

(b) At x = 1, the series is

∞
X 1
(−1)n+1 .
n=1
n
This series converges by the Alternating Series test, which determines conditional
converges. So, the series converges conditionally at x = 1.

153
 (c) Integrating term-by-term,
Z
g(x) = xf (t)dt
Z0 x 
t2 t3 t4 n

n+1 t
= t − + − + . . . + (−1) dt
0 2 3 4 n
x
t2 t3 t4 t5 n+1 tn+1 
= − + − + . . . + (−1)
2 6 12 20 n(n + 1) 0
x2 x3 x4 x5 xn+1
= − + − + . . . + (−1)n+1 .
2 6 12 20 n(n + 1)

(d) The Alternating Series Estimation Theorem tells us that the upper bound for the
5
error is the absolute value of the first not included term. For P4 , this term is − x20 .
So,
1 5 
     

 1 1   2 = 1 < 1 .
P 4 −g  < −
 2 2   20  640 500

9.1.5 2016 Free-Response Questions

Questions 1 and 2 part of the same section area are allotted 30 minutes from completion
with the aid of a graphing calculator. Questions 3 through 6 are part of the same section
and area allotted 1 hour from completion without the aid of a graphing calculator.

t (hours) 0 1 3 6 8
R(t) (liters / hour) 1340 1190 950 740 700

2
1. Water is pumped into a tank at a rate modeled by W (t) = 2000e−t /20 liters per hour
for 0 ≤ t ≤ 8, where t is measured in hours. Water is removed from the tank at a rate
modeled by R(t) liters per hour, where R is differentiable and decreasing on 0 ≤ t ≤ 8.
Selected values of R(t) are shown in the table above. At time t = 0, there are 50000
liters of water in the tank.

(a) Estimate R0 (2). Show the work that leads to your answer. Indicate units of
measure.
(b) Use a left Riemann sum with the four subintervals indicated by the table to
estimate the total amount of water removed from the tank during the 8 hours.
Is this an over estimate or underestimate of the total amount of water removed?
Give a reason for your answer.
(c) Use your answer from part (b) to find an estimate for the total amount of water
in the tank, to the nearest liters, at the end of the 8 hours.

154
 (d) For 0 ≤ t ≤ 8, is there a time t when the rate at which the water is pumped into
the tank is the same rate as the rate at which water is removed from the tank?
Explain why or why not.

Figure 9.5: AP Calculus BC 2016 Exam Free-Response Question 2

2. At time t, the position of a particle moving in the xy-plane is given by the parametric
functions (x(t), y(t)), where dx
dt
= t2 + sin (3t2 ). The graph of y consisting of three line
segments, is shown in the figure above. At t = 0, the particle is at position (5, 1).

(a) Find the position of the particle at t = 3.

(b) Find the slope of the line tangent to the part of the particle at t = 3.
(c) Find the speed of the particle at t = 3.
(d) Find the total distance traveled by the particle from t = 0 to t = 2.

155
 Figure 9.6: AP Calculus BC 2016 Exam Free-Response Question 3, Graph of f

3. The figure above shows the graph ofRthe piecewise-linear function f . For −4 ≤ x ≤ 12,
x
the function g is defined by g(x) = 2 f (t)dt.

(a) Does g have a relative minimum, a relative maximum, or neither at x = 10?

Justify your answer.
(b) Does the graph of g have a point of inflection at x = 4? Justify your answer.
(c) Find the absolute minimum value and the absolute maximum value of g on the
interval −4 ≤ x ≤ 12. Justify your answers.
(d) For −4 ≤ x ≤ 12, find all intervals for which g(x) ≤ 0.
dy
4. Consider the differential equation dx
= x2 − 12 y.
d2 y
(a) Find dx2
in terms of x and y.
(b) Let y = f (x) be the particular solution to the given differential equation whose
graph passes through the point (−2, 8). Does the graph of f have a relative
minimum, a relative maximum, or neither at the point (−2, 8)? Justify your
answer.
(c) Let y = g(x) be the particular
 solution
 to the given differential equation with
g(x)−2
g(−1) = 2. Find limx→−1 3(x+1)2 . Show the work that leads to your answer.
(d) Let y = h(x) be the particular solution to the given differential equation with
h(0) = 2. Use Euler’s method, starting at x = 0 with two steps of equal size, to
approximate h(1).

156
 Figure 9.7: AP Calculus BC 2016 Exam Free-Response Question 5

5. The inside of a funnel of height 10 inches has circular cross sections, as shown in the
1
figure above. At height h, the radius of the funnel is given by r = 20 (3 + h2 ), where
0 ≤ h ≤ 10. The units of r and h are inches.

(a) Find the average value of the radius of the funnel.

(b) Find the volume of the funnel.
(c) The funnel contains liquid that is draining from the bottom. At the instant when
the height of the liquid is h = 3 inches, the radius of the surface of the liquid
is decreasing at a rate of 15 inch per second. At this instant, what is the rate of
change of the height of the liquid with respect to time?

6. The function f has a Taylor series about x = 1 that converges to f (x) for all x in the
interval of convergence. It is known that f (1) = 1, f 0 (1) = − 21 , and the nth derivative
of f at x = 1 is given by f (n) (1) = (−1)n (n−1)!
2n
for n ≥ 2.

(a) Write the first four nonzero terms and the general term of the Taylor series for f
about x = 1.
(b) The Taylor series for f about x = 1 has a radius of convergence of 2. Find the
interval of convergence. Show the work that leads to your answer.
(c) The Taylor series for f about x = 1 can be used to represent f (1.2) as an al-
ternating series. Use the first three non-zero terms of the alternating series to
approximate f (1.2).

157
 (d) Show that the approximation found in part (c) is within 0.001 of the exact value
of f (1.2).

9.1.6 2016 Free-Response Answers

1. (a) We can approximate R0 (2) as the tangent slope between 1 and 3.
R(3) − R(1) 950 − 1190
R0 (2) ≈ = = −120.
3−1 2
So, we estimate R0 (3) to be -120 liters/hour2 .
(b) Using a left Riemann sum using the values in the table,
Z 8
R(t)dt ≈ (1−0)1340+(3−1)1190+(6−3)950+(8−6)740 = 1340+2380+2850+1480 = 8050.
0

So, the left Riemann sum approximates the total water removed to be 8050 liters.
(c) Using our answer from part (b) and integrating, we can find the total amount of
water added or removed.
Z
2
∆Water = −8050 + 82000e−t /20 dt ≈ −214.
0

Since we know there is 50000 liters of water at t = 0, we can add the net change
to find the amount of water at t = 8.
Water8 = Water0 + ∆Water = 50000 − 214 = 49768.
So, to the nearest liter, we approximate the amount of water in the tank at t = 8
to be 49768 liters.
(d) At t = 0, W (0) = 2000 > R(0) = 1340, so W (0) − R(0) > 0. At t = 8,
W (8) = 81.52 < R(8) = 700, so W (8) − R(8) < 0. Since both W and R are
continuous, so is W − R. So, by the Intermediate Value Theorem, there is some
0 < t < 8 such that W (t) − R(t) = 0, or W (t) = R(t).
2. (a) We can apply the Fundamental Theorem of Calculus to find x(3).
Z 3
dx
dt = x(3) − x(0) = x(3) − 5 = 9.377 =⇒ x(3) = 14.377.
0 dt

Looking at the graph, we see that y(3) = − 12 . So, at t = 3, the particle’s position
is (14.377, −0.5).
(b) Looking at the graph, we see that y 0 (3) = 21 . Evaluating dx
dt
at t = 3, we see that
x0 (3) = 9 + sin 27. So,
dy 1
dy dt 2
= dx
= ≈ 0.0502.
dx dt
9 + sin 27

158
 (c) Using our formula for parametric speed,
s 
2
1
+ (9 + sin 27)2 ≈ 9.969.
p
0 2 0 2
s = (x (t)) + (y (t)) =
2

(d) Starting with the formula for parametric arc length and splitting the integral into
two peices,
Z 2q Z 1p Z 2p
0 2 0 2 2 2 2 2
(y (t)) + (x (t)) dt = (−2) + (t + sin (3t )) dt + (0)2 + (t2 + sin (3t2 ))2 dt
0 0 1
≈ 2.237 + 2.112
= 4.439.

3. (a) Since we know by the Fundamental Theorem of Calculus that f is the derivative
of g, g has critical points where f is 0. x = 10 is one such critical point. Both to
the left and right of x = 10, f is negative. So, although x = 10 is a critical point,
it is neither a relative minimum or relative maximum of g because g is decreasing
both left and right of x = 10.
(b) Inflection points occur when the second derivative changes sign. Since f is the
derivative of g, inflection points of g occur when the derivative of f changes sign.
To the left of x = 4, the derivative of f is positive. To the right of x = 4, the
derivative of f is negative. So, x = 4 is indeed an inflection point for g.
(c) g has critical points where f is 0, which is at x = −2, x = 2, x = 6, and x = 10.
Of these, x = 2 and x = 10 do not change sign, so they cannot be absolute
extrema. Evaluating the remaining critical points and the endpoints using the
geometry of the graph of f ,

g(−4) = −4
g(−2) = −8
g(6) = 8
g(12) = −4.

So, the absolute minimum of g is at x = −2, and the absolute maximum of g is

at x = 6.
(d) To the left x = 2, g(x) is negative whenever there is more area between f and the
x-axis above the x-axis than below. So, all points [−4, 2] have g(x) ≤ 0. To the
left of x = 2, the opposite is true. So, all points [10, 12] have g(x) ≤ 0. Putting
these two results together, g(x) ≤ 0 on [−4, 2] ∪ [10, 12].
4. (a) Implicitly differentiating,
 
2 2 1 dy 1 2 1
d yx = 2x − = 2x − x − y .
2 dx 2 2

159
 (b) Since we have expression for the first and second derivatives, it makes the most
sense to apply a second derivative test. First, checking that (−2, 8) is indeed a
critical point,
dy 1
= (−2)2 − (8) = 0.
dx (−2,8) 2
Next, evaluating the second derivative,

d2 y
 
1 2 1
= 2(−2) − (−2) − (8) = −4 − 0 = −4.
dx2 (−2,8) 2 2

Since the second derivative is negative at this critical point, the second derivative
test tells us that this is a relative maximum.
(c) Since we know that g(−1) = 2, both the numerator and denominator of the limit
are in an indeterminate form of 0/0. So, we can apply L’Hôpital’s Rule.
   0 
g(x) − 2 g (x)
lim = lim .
x→−1 3(x + 1)2 x→−1 6(x + 1)

Using the given differential equation, we can see that at (−1, 2), g 0 (−1) = 0. So
again we have an indeterminate form of 0/0 and can apply L’Hôpital’s Rule.
 0
g 00 (x)

g (x)
lim = lim .
x→−1 6(x + 1) x→−1 6

Using our answer from part (b), we know that at (−1, 2), g 00 (−1) = −2. So,

g 00 (x) −2 1
lim = =− .
x→−1 6 6 3

(d) Applying two iterations of Euler’s method starting at (0, 2) with ∆x = 21 ,

dy dy
(x, y) dx ∆x ∆y = ∆x dx (x + ∆x, y + ∆y)
1
(0, 2) −1 2
− 21 ( 12 , 32 )
( 12 , 32 ) − 12 1
2
− 41 (1, 54 )

So, our application of Euler’s method approximates h(1) to be 5/4.

5. (a) Applying the formula for average value,

Z 10
h3
 
1 1 2

1 109
3 + h dh = 3h + = .
10 − 0 0 20 200 3 60
109
So, the average radius of the funnel is 60
inches.

160
 (b) Applying the volume formula for a solid of revolution,
Z 10  
1 2


V =π 3+h dh
0 20
 5 10
π h 3

= + 2h + 9h 
400 5 0
π
= (20000 + 2000 + 90)
400
2209π
= .
40

(c) Applying the chain rule,

dr dr dh
=
dt dh dt
1 dh
= h .
10 dt
Solving with the information given,
1 1 dh dh 2
− = (3) =⇒ =− .
5 10 dt dt 3
So, at this instant, the height of the liquid is decreasing at 2/3 inches per second.

6. (a) Finding several derivatives at x = 1,

f (1) = 1
1
f 0 (1) = −
2
1
f 00 (1) =
4
1
f (3) (1) = − .
4
Applying the Taylor Series formula centered at x = 1,
1 1 1 1
f (x) = 1 − (x − 1) + (x − 1)2 − (x − 1)3 + . . . + (−1)n n (x − 1)n .
2 8 24 n2

(b) Since we know that the series is centered at x = 1 and the radius of convergence
is 2, we know the series converges on (−1, 3) and need to check the endpoints.
When x = −1,
∞ ∞
n 1 1
X X
n
(−1) n
(−1 − 1) =
n=1
n2 n=1
n

161
 diverges by the P-Test. When x = 3,
∞ ∞
X
n1 n
X 1
(−1) n
(3 − 1) = (−1)n
n=1
n2 n=1
n

converges by the Alternating Series Test. So, the invterval of convergence is

(−1, 3].
(c) Using the first three terms of our series from part (a) with x = 1.2,
1 1 1 1
f (1.2) ≈ 1 − (1.2 − 1) + (1.2 − 1)2 = 1 − + = 0.905.
2 8 10 200

(d) Since this series is alternating, we can apply the Alternating Series Estimation
Theorem.
f (1.2) − P2 (1.2) ≤  −1 (.2)3  = 1 ≤ 0.001.
   
23 · 3 3000
So, by the Alternating Series Estimation Theorem, the error is certainly at most
0.001.

9.2 Online Resources

Below is a list of other useful resources for learning single variable calculus. Most are freely
available online.

• Paul’s Online Notes, Calculus I – Covers first “half” of single variable calculus, includ-
ing derivatives and definite integrals.

• Paul’s Online Notes, Calculus I – Covers second “half” of single variable calculus,
including improper integrals, parametric and polar functions, and series.

• Khan Academy, AP Calculus BC – Video lectures and practice problems that should
cover the same material as in this book. There is also resources for AP Calculus AB
which covers many of the same, but fewer topics than BC.

• MIT OCW 18.01 Complete series of lectures, recitations, assignments, practice prob-
lems, lecture notes, and exams needed for independent study.

• Finney, Demana, Waits, and Kennedy: Calculus: Graphical, Numerical, Algebraic –

Common textbook for AB and BC students. Contains explinations, practice problems,
and tips for the AP exams.

162