Integration Techniques and Tricks

University of Miami Mathematics Union

Trevor Birenbaum
[email protected]

Part 1 2

1 The Substitution Rule 2

1.1 The Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Powers of Sine and Cosine . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Trigonometric Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 The Weierstrass Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 The Bounds Trick . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6 Inversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 Integration by Partial Fractions 12

2.1 The Normal Way . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 The Best Way . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 The Method of Partial Fractions . . . . . . . . . . . . . . . . . . . . . . . . 13

3 Integration by Parts 18
3.1 ex as Its Own Derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Multiplying by One . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.3 Tic-Tac-Toe, or Tabular Integration . . . . . . . . . . . . . . . . . . . . . . 21

Part 2 23

4 Integration of Parametric Equations 23

4.1 The Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2 Arc Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.3 Polar Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

5 Differentiation Under the Integral 30

6 Complex Numbers and Euler’s Formula 35

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
6.2 Integrals with Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . 36
6.3 Using Real Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

7 Just for Fun 38

8 Conclusion 43
Integration Techniques and Tricks University of Miami Mathematics Union

This packet contains the materials to accompany Parts 2 and 3 of the University of
Miami Mathematics Union Lecture Series, covering Integration Techniques. A foundational
understanding of Calculus I and II is recommended for these lectures. Material from those
classes will be reviewed, but at an accelerated pace.

1 The Substitution Rule

1.1 The Basics
Much how the process of taking the integral of a function is the reverse of taking the derivative
of that function, useful formulas in differentiation can also be reversed and re-written as
integration techniques. The first of these that we will cover is the Substitution Rule, which
is the opposite of the Chain Rule: dxd
f (g(x)) = g 0 (x)f 0 (g(x)).
R
Consider the expression x cos(x2 ). We are trying to find a function I(x) whose derivative is
i(x) = x cos(x2 ). Observing that the derivative of sin x is cos x and spotting x2 as an argument
of the trigonometric function, we may want to guess that sin(x2 ) is an antiderivative of
d
x cos(x2 ). Deriving and using the Chain Rule, we find that dx sin(x2 ) = 2x cos(x2 ), which is
really close to what we’re trying to find! It’s just off by a factor of 2. Thus, we can say this:
sin(x2 )
Z
x cos(x2 ) dx = +C
2
But we can’t solve every integral by making an educated guess and then deriving it to check
if we’re right. We have to generalize this process a bit. Observe that we can break the
integrand down into the components of a result obtained from the Chain Rule. Observe that
x2 and its derivative 2x both appear in the integrand. x2 is the argument of cosine, and
half its derivative is one of the
0
two terms being multiplied to produce the function. Setting
g(x) = x2 , we have i(x) = g (x) 2
cos(g(x)). Set f (x) = cos(x), and then i(x) = 12 g 0 (x)f (g(x))
and I(x) = f (g(x)) + C. Succinctly put,
Z
g 0 (x)f 0 (g(x)) dx = f (g(x)) + C (1)
Z b
g 0 (x)f 0 (g(x)) dx = f (g(b)) − f (g(a)) (2)
a
In the process of solving an integral with the Substitution Rule, it is helpful to use u and
du
R (shorthand for du
dx
dx) to represent g(x) and g 0 (x) respectively. Consider the integral
x
x2 +1
dx. We can set u = x2 + 1 and du = 2x to obtain the following:
Z Z
x u=x2 +1 du
2
dx =
x +1 2u
ln |u|
= +C
2
ln(x2 + 1)
= +C
2

Trevor Birenbaum Page 2

Integration Techniques and Tricks University of Miami Mathematics Union

Not all u-substitutions are quite as clear-cut. An oft-taught way to integrate secant is:
Z Z
tan θ + sec θ
sec θ dθ = sec θ · dθ
sec θ + tan θ
tan θ sec θ + sec2 θ
Z
= dθ
sec θ + tan θ
Z
u=sec θ+tan θ du
=
u
= ln |u| + C
= ln | sec θ + tan θ| + C

How unmotivated is that? Don’t worry. We’ll have something much more fun soon.

The following are left as exercises to the reader.

Z
dx
1.
x ln x
Z 2√
2. x3 − x2 dx
1
Z √
x2 2x3 − 7 dx
4
3.
Z
x+ex
4. ex+e dx

1.2 Powers of Sine and Cosine

The following identities are known results from trigonometry:

sin(2θ) = 2 sin θ cos θ (3)

cos(2θ) = cos2 θ − sin2 θ (4)
= 2 cos2 θ − 1 (5)
= 1 − 2 sin2 θ (6)
2 tan θ
tan(2θ) = (7)
1 − tan2 θ
1 − cos(2θ)
sin2 θ = (8)
2
1 + cos(2θ)
cos2 θ = (9)
2
The sine, cosine, and tangent addition formulas can be used to obtain values of trig(nθ)
for higher values of n. These formulas can be used to reduce the power of a trigonometric
function in an integral to something manageable and easy to integrate, as seen here:

Trevor Birenbaum Page 3

Integration Techniques and Tricks University of Miami Mathematics Union

Z Z
4
sin θ dθ = (sin2 θ)2 dθ
Z  2
1 − cos(2θ)
= dθ
2
Z
1
= (1 − 2 cos(2θ) + cos2 (2θ)) dθ
4
Z  
1 1 + cos(4θ)
= 1 − 2 cos(2θ) + dθ
4 2
Z  
1 3 cos(4θ)
= − 2 cos(2θ) + dθ
4 2 2
3x sin(2x) sin(4x)
= − + +C
8 4 32
The following identities are also known results from trigonometry:

sin2 θ + cos2 θ = 1 (10)

tan2 θ + 1 = sec2 θ (11)
cot2 θ + 1 = csc2 θ (12)

These can be used to reduce larger powers of trigonometric functions by introducing a mixture
of functions that can be manipulated into nice u-substitutions.
Z Z
sin θ cos θ dθ = sin5 θ(1 − sin2 θ) dθ
5 2

Z
= (sin5 θ − sin7 θ) dθ
Z
= (sin4 θ sin θ − sin6 θ sin θ) dθ
Z
= ((1 − cos2 θ)2 sin θ − (1 − cos2 θ)3 sin θ) dθ
Z
u=cos θ
= − ((1 − u2 )2 − (1 − u2 )3 ) du
Z
= − ((u4 − 2u2 + 1) − (−u6 + 3u4 − 3u2 − 1)) du
Z
= (−u6 + 2u4 − u2 − 2) du
u7 2u5 u3
=− + − − 2u + C
7 5 3
cos7 θ 2 cos5 θ cos3 θ
=− + − − 2 cos θ + C
7 5 3

Trevor Birenbaum Page 4

Integration Techniques and Tricks University of Miami Mathematics Union

These two methods can be combined in more exotic-looking integrals such as this one:
Z π Z π
2 2
sin θ sin(2θ) sin(4θ) dθ = sin θ(2 sin θ cos θ)(2 sin(2θ) cos(2θ)) dθ
0 0
Z π
2
= sin θ(2 sin θ cos θ)(2 · 2 sin θ cos θ(2 cos2 θ − 1)) dθ
0
Z π
2
=8 (2 sin3 θ cos4 θ − sin3 θ cos2 θ) dθ
0
Z π
2
=8 sin3 θ(2 cos4 θ − cos2 θ) dθ
0
Z π
2
=8 sin θ(1 − cos2 θ)(2 cos4 θ − cos2 θ) dθ
0
Z 0
u=cos θ
= −8 (1 − u2 )(2u4 − u2 ) du
1
Z 1
=8 (−2u6 + 3u4 − u2 ) du
0
1
16u7 24u5 8u3
=− + −
7 5 3 0
16 24 8
=− + −
7 5 3
16
=−
105
The following are left as exercises to the reader.
Z
1. sec2 θ cot θ dθ
Z √
2. cos3 θ sin θ dθ
Z
3. sin2020 θ cos3 θ dθ

Z π
2
4. sin θ sin(2θ) sin(3θ) dθ
0
Z
5. sin2a−1 θ dθ, where a is a positive integer
Z  
cos θ + sin θ
6. Use another trigonometric identity to find arctan dθ
cos θ − sin θ

Trevor Birenbaum Page 5

Integration Techniques and Tricks University of Miami Mathematics Union

1.3 Trigonometric Substitution

The goal of a trigonometric substitution is to replace expressions containing square roots
with expressions that contain trigonometric functions, which are generally easier to solve.
2 2
Most commonly, we can take advantage of the expression √ sin θ + cos θ = 1 and the other
2
p come from√it. Consider the expression 1 − x . If we set x = sin θ, then
identities that
√
1 − x2 = 1 − sin2 θ = cos2 θ = | cos θ|. If we are careful with the domain of θ, this can
usually simply be equal to cos θ. dx is very simple to calculate as well, merely being the
derivative of sin θ: cos θ dθ.
Z √ Z
x=sin θ
1 − x2 dx = cos2 θ dθ
Z
1 + cos(2θ)
= dθ
2
θ sin(2θ)
= + +C
2 4
θ + sin θ cos θ
= +C
2 √
arcsin x + x 1 − x2
= +C
2
√ √
Similar processes can be used for the expressions 1 + x2 and x2 − 1 with the substitutions
x = tan θ and x = sec θ respectively. In fact, u-substitutions can be used to set up the
following substitutions:
√
a2 − x 2 x = a sin θ dx = a cos θ dθ (13)
√ 2
a2 + x 2 x = a tan θ dx = a sec θ dθ (14)
√
x 2 − a2 x = a sec θ dx = a sec θ tan θ dθ (15)

These substitutions can all be applied as follows:

Z 1r Z 1
1+x 1+x
dx = √ dx
0 1−x 0 1 − x2
Z π
2 1 + sin θ
x=sin θ
= cos θ dθ
0 cos θ
Z π
2
= (1 + sin θ) dθ
0
π
= θ − cos θ|02
π
=1+
2

Trevor Birenbaum Page 6

Integration Techniques and Tricks University of Miami Mathematics Union

π
Z 3 Z
4
− 23 x=3 tan θ 3
2
(x + 9) dx = (9 tan2 θ + 9)− 2 · 3 sec2 θ dθ
0 0
π
sec2 θ
Z
4
= dθ
0 9 sec3 θ
Z π
1 4
= cos θ dθ
9 0
π
sin θ 4
=
9 0
√
2
=
18

√ π
Z 2 2 Z
dx x=2 sec θ 4 2 sec θ tan θ
√ = √ dθ
2 x2 − 4 0 4 sec2 θ − 4
Z π
4 2 sec θ tan θ
= dθ
0 2 tan θ
Z π
4
= sec θ dθ
0
π
= ln | sec θ + tan θ|04
√
= ln(1 + 2)

The following are left as exercises to the reader.

1. Prove that the area of a circle with radius a is πa2 .

Z z√ 2
x −1
2.
0 x
Z z √
x2 − 1
3.
1 x2
Z
dx
4.
(4 + 9x2 )2
Z 2
dx
5. √
0 x2 + 6x
r
b
c − 2a
Z
dx 1 u b
6. Prove that = arctan + C, where u = x + and v =
ax2 + bx + c av v 2a a

Trevor Birenbaum Page 7

Integration Techniques and Tricks University of Miami Mathematics Union

1.4 The Weierstrass Substitution

This brings us to one of the coolest u-substitutions, the Weierstrass substitution, also known
as the tangent half-angle formula. Set u = tan 2θ . It can be shown that sin x = 1+u 2u
2,
1−u2 2u 2
cos x = 1+u2 , tan x = 1−u2 , and dx = 1+u2 . These substitutions can be used in place
of algebraic manipulation with trigonometric functions, as seen in the two methods for
1
integrating 1+sin θ
seen here:

1 − sin θ
Z Z
dθ
= 2 dθ
Z 1 − sin θ
1 + sin θ
1 − sin θ
= dθ
cos2 θ
Z
= (sec2 θ − sec θ tan θ) dθ

= tan θ − sec θ + C

Z Z
dθ WS 1 2
= 2u · dx
1 + sin θ 1 + 1+u2 1 + u2
Z
2
= dx
1 + u2 + 2u
Z
2
= dx
(u + 1)2
2
=− +C
u+1
2
=− +C
tan 2θ + 1

Recall how the original derivation of the integral of secant (and for tangent and cosecant)
was unintuitive, boring, and uninspiring. The Weierstrass substitution can be used to prove
the definitions much more logically. The derivation of the integral of cosecant is shown here:
Z Z
dθ
csc θ dθ =
sin θ
1 + u2
Z
WS 2
= · du
2u 1 + u2
Z
du
=
u
= ln |u| + C
θ
= ln tan + C
2
Nice and easy! A similar strategy for integrating secant involves integration techniques from
Section 2 of this packet.

Trevor Birenbaum Page 8

Integration Techniques and Tricks University of Miami Mathematics Union

Even some algebraic functions can have Weierstrass-motivated solutions:

Z 1 Z 1 √
dx 1 − x2
√ = √ √ dx
0 x+ 1 − x2 + 1 0 x 1−x +
2 1 − x2 + 1 − x2
Z π p
x=sin θ 2 1 − sin2 θ
= p p dθ
0 sin θ 1 − sin2 θ + 1 − sin2 θ + 1 − sin2 θ
Z π
2 cos θ
= 2
· cos θ dθ
0 sin θ cos θ + cos θ + cos θ
Z π
2 cos θ
= dθ
0 sin θ + 1 + cos θ
Z 1 1−u2
WS 1+u2 2
= 2 · du
2u
0 1+u2 + 1 + 1+u2
1−u 1 + u2
Z 1
1 − u2 2
= 2 2
· du
0 2u + 1 + u + 1 − u 1 + u2
Z 1
1 − u2
=2 2
du
0 (2u + 2)(1 + u )
Z 1
1−u
= 2
du
0 1+u
Z 1 Z 1
du u
= 2
− 2
du
0 1+u 0 1+u
Z 1
1 2 dv
Z
v=1+u2 du
= 2
−
0 1+u 2 1 v
2
ln v
= arctan u|10 −
2 1
π ln 2
= −
4 2
The following are left as exercises to the reader.
Z
dθ
1.
sin θ + tan θ
Z
dθ
2.
5 sin θ + 12 cos θ
Z π
2 dθ
3. Generalize the value of for positive integer a.
0 a + cos θ
Z 1 2x
arcsin 1+x 2
4. Use a reverse Weierstrass substitution to find 2
dx.
0 1+x
Z 1 4
x (1 − x2 )5
5. dx
0 (1 + x2 )10

Trevor Birenbaum Page 9

Integration Techniques and Tricks University of Miami Mathematics Union

1.5 The Bounds Trick

The Bounds Trick is a very underrated substitution
Rb that does a very good job at simplifying
some integrals. Consider the definite integral a f (x) dx. When constructing a Riemann sum
for this integral, the rectangles are drawn from left to right, starting at a and moving in
increments of ∆x towards b. Well, what if we flipped the function over the line x = b and
did the same thing, starting at a + b and moving in increments of ∆x towards b? This would
be equivalent to a u-substitution of u = a + b − x and would maintain the same bounds as
the original definite integral. The following identity follows, and is important:
Z b Z b
f (x) dx = f (a + b − x) dx (16)
a a
1 b
Z
= (f (x) + f (a + b − x)) dx (17)
2 a
Algebraic manipulation may be required to express an integral in a form where the Bounds
Trick may be a viable option:
Z 7 Z 7
ln(x + 2) ln(x + 2)
2
dx = dx
3 ln(24 + 10x − x ) 3 ln(x + 2) + ln(12 − x)
Z 7
BT 1 ln(x + 2) + ln(12 − x)
= dx
2 3 ln(x + 2) + ln(12 − x)
1 7
Z
= 1 dx
2 3
=2
π
The Bounds
Trick often appears in

trigonometric integrals from 0 to 2 , using the identities
π π
sin 2 − θ = cos θ and cos 2 − θ = sin θ.
Z π Z π
2 sin2020 θ BT 1 2 sin2020 θ + cos2020 θ

2020 dθ = dθ
0 sin θ + cos2020 θ 2 0 sin2020 θ + cos2020 θ
Z π
1 2
= 1 dθ
2 0
π
=
4
The following are left as exercises to the reader.
Z π
2
1. Reevaluate cos2 θ dθ
0
Z π
2 dθ
2. √
2
0 1 + tan θ
Z π
2
3. ln(sin θ) dθ
0
k k 1
1 + x2
Z Z Z
4. Use the similar identity f (x) dx = (f (x) + f (−x)) dx to evaluate dx
−k 0 −1 1 + 2x

Trevor Birenbaum Page 10

Integration Techniques and Tricks University of Miami Mathematics Union

1.6 Inversion
The last u-substitution of note is u = x1 , which lends itself useful in certain scenarios according
to the following identities.
Z ∞ Z ∞
f x1


f (x) = dx (18)
0 0 x2

!
1 ∞ f x1
Z
= f (x) + dx (19)
2 0 x2

∞
1 ∞ ln(2x) 1 ∞ ln(2x−1 ) 1
Z Z Z
ln(2x)
dx = dx + · 2 dx
0 1 + x2 2 0 1 + x2 2 0 1 + x12 x
Z ∞ Z ∞ 2 −1
1 ln(2x) 1 x ln(2x ) 1
= 2
dx + · 2 dx
2 0 1+x 2 0 x2 + 1 x
Z ∞ −1
1 ln(2x) + ln(2x )
= dx
2 0 1 + x2
2 ln 2 ∞ dx
Z
=
2 0 1 + x2
π ln 2
=
2

Z ∞ Z ∞
−(x2 +
e
1
x2
) dx = 2 1
e−(x −2+ x2 ) dx
2

2
e 0
−∞
2 ∞ −(x− x1 )2
Z
= 2 e dx
e 0
1 2
1 0 e−( u −u)
Z ∞ Z
IV 1
2
−(x− x1 )
= 2 e dx + 2 − du
e 0 e ∞ u2
1 2
1 ∞ −(x− x1 )2 1 ∞ e−(u− u )
Z Z
= 2 e dx + 2 du
e 0 e 0 u2
1 ∞
Z  
1 1 2
= 2 x + 2 e−(x− x ) dx
e 0 x
Z ∞
u=x− x 1
1
2
= 2
e−u du
e −∞
√
π
= 2
e
We’ll prove that last step later. The following are left as exercises to the reader.
Z ∞ Z ∞
ln x 17x5 + 16x3 + 17x
1. dx 2. dx
0 x2 + 2x + 4 0 16x8 + 136x6 + 321x4 + 136x2 + 16

Trevor Birenbaum Page 11

Integration Techniques and Tricks University of Miami Mathematics Union

2 Integration by Partial Fractions

When integrating a quotient of polynomials, large denominators can appear scary and difficult
to integrate. However, if those large denominators can be broken up and factored, it may
be possible to express the quotient in terms of linear and quadratic factors, for which the
methods of integrating have already been discussed. For this, the Method of Partial Fractions
should be used.
P (x)
Consider the rational expression Q(x) , where the degree of P (x) is less than the degree of
Q(x) and Q(x) can be factored into polynomials q0 (x), q1 (x), . . . that multiply together to
obtain Q(x). Then it can be written as the sum of multiple rational functions, each of the
form aqkk(x)
(x)
for some polynomials ak (x). Each ak (x) will have degree at most one less than the
degree of its corresponding qk (x). Succinctly,
k
P (x) P (x) X ai (x)
= Qk = (20)
Q(x) i=1 qi (x)
q (x)
i=1 i

3x+1 3x+1 A B
Consider the rational expression x2 +x
. We know x2 +x = x(x+1), so we can say x2 +x
= x
+ x+1
for some constants A and B.

2.1 The Normal Way

The most common way partial fraction decomposition is taught is to set up a system of
equations that correspond with each term in the expansion of g(x). When all of the terms
are summed together, there are known values on the left hand side and unknown values on
the right hand side; solving this system of equations yields the unknown values and thus the
partial fraction decomposition.

It is known x3x+1 A B
2 +x = x + x+1 for some constants A and B. Combining the two terms on the
A(x+1)+Bx
right hand side yields x3x+1
2 +x = x2 +x
= x(A+B)+A
x2 +1
. This yields the system of equations
A + B = 3 and A = 1, which gives the solution {A, B} = {1, 2}, so x3x+1 1 2
2 +x = x + x+1 . This

is a fine method to perform partial fraction decomposition, but it can get quite messy with
polynomial multiplication and larger systems of equations that have less obvious answers.

2.2 The Best Way

The best way to find a partial fraction decomposition, like the traditional way, involves
splitting the denominator into its components and making each of them correspond with its
own term. However, both sides of the equation can be multiplied by the (equal) denominators,
allowing for the plugging in of notable values of x to get an easier set of equations. In fact, if
all factors are linear, then this gives the solution for each variable outright!
A(x+1)+Bx
We know x3x+1
2 +x = x2 +x
. Multiplying both sides by x2 + x yields 3x + 1 = A(x + 1) + Bx.
Plugging in x = 0 yields 0 + 1 = A + 0 ⇒ A = 1, and plugging in x = −1 yields
−3 + 1 = 0 − B ⇒ B = 2, so x3x+1 1 2
2 +x = x + x+1 . Nice and simple.

Trevor Birenbaum Page 12

Integration Techniques and Tricks University of Miami Mathematics Union

2.3 The Method of Partial Fractions

Now that we understand the logic behind partial fraction decomposition, we can present an
P (x)
algorithm for breaking up any rational function Q(x) into its component partial fractions

1. Use polynomial long division to ensure that the degree of P (x) is less than the degree
of Q(x). The resulting polynomial can be integrated easily.

2. Factor Q(x) as much as possible.

P (x)
3. Correctly set up the expression of Q(x) according to the following rules and solve for
the constants by plugging in values of x.

Q(x) is a product of non-repeating linear factors

If Q(x) = (a1 x + b1 )(a2 x + b2 ) · · · (an x + bn ), then we have:

P (x) A1 A2 An
= + + ··· + (21)
Q(x) a1 x + b 1 a2 x + b 2 an x + b n

6x2 − 13x − 35
Z
Find dx
x3 − 7x − 6
2
Observe Q(x) = x3 − 7x − 6 = (x − 3)(x + 1)(x + 2), so 6xx3−13x−35
−7x−6
A
= x−3 B
+ x+1 + C
x+2
=
A(x+1)(x+2)+B(x−3)(x+2)+C(x−3)(x+1) 3
x3 −7x−6
. Multiplying by x − 7x − 6 on both sides yields:

A(x + 1)(x + 2) + B(x − 3)(x + 2) + C(x − 3)(x + 1) = 6x2 − 13x − 35.

• x = −1: −4B = −16 ⇒ B = 4.

• x = −2: 5C = 15 ⇒ C = 3.

• x = 3: 20A = −20 ⇒ A = −1.

6x2 − 13x − 35
Z Z  
1 4 3
dx = − + + dx
x3 − 7x − 6 x−3 x+1 x+2
= − ln |x − 3| + 4 ln |x + 1| + 3 ln |x + 2| + C
(x + 1)4 (x + 2)3
= ln +C
x−3

Trevor Birenbaum Page 13

Integration Techniques and Tricks University of Miami Mathematics Union

Q(x) is a product of linear factors, some of which may be repeated

If a term (ax + b) is a multiplicative factor of Q(x) k times then it is represented as the sum
of constants divided by each power of (ax + b) from 1 to k:

P (x) A1 A2 A3 Ak
k
= + 2
+ 3
+ ··· + (22)
(ax + b) ax + b (ax + b) (ax + b) (ax + b)k

All other non-multiplicative terms should be handled as previous.

P (x) A1 A2 Ak Ak+1
k
= + 2
+ ··· + k
+ (23)
(a1 x + b1 ) (a2 x + b2 ) a1 x + b1 (a1 x + b1 ) (a1 x + b1 ) a2 x + b 2

3x2 + 23x + 29
Z
Find dx
x3 + x2 − 8x − 12
2 A(x+2)2 +B(x−3)(x+2)+C(x−3)
Observe Q(x) = (x−3)(x+2)2 , so x3x +23x+29 A B C
3 +x2 −8x−12 = x−3 + x+2 + (x+2)2 = x3 +x2 −8x−12
.
Multiplying by x3 + x2 − 8x − 12 on both sides yields:

A(x + 2)2 + B(x − 3)(x + 2) + C(x − 3) = 3x2 + 23x + 29

• x = −2: −5C = −5 ⇒ C = 1.

• x = 3: 25A = 125 ⇒ A = 5.

• There aren’t any values of x that we can plug in that give a function of one variable
on the left hand side, but since we know the values of the other two variables, we
should just choose a value that makes the numbers easier to work with; plugging in
x = 0 yields 4A − 6B − 3C = 29. Since A = 5 and C = 1, we have 20 − 6B − 3 = 29.
17 − 6B = 29, so B = −2.

3x2 + 23x + 29
Z Z  
5 2 1
dx = − + dx
x3 + x2 − 8x − 12 x − 3 x + 2 (x + 2)2
1
= 5 ln |x − 3| − 2 ln |x + 2| − +C
x+2
(x − 3)5 1
= ln − +C
x+2 x+2

Trevor Birenbaum Page 14

Integration Techniques and Tricks University of Miami Mathematics Union

Q(x) is contains one or more non-repeating quadratic factors

A linear function must appear in the numerator of a term with a quadratic denominator.

P (x) A1 x + B1 A2 x + B2
= + (24)
(a1 x2 2
+ b1 x + c1 )(a2 x + b2 x + c2 ) a1 x + b 1 x + c 1 a2 x 2 + b 2 x + c 2
2

Z
14
Find dx
x3 + 2x2+ 3x + 6
A(x2 +3)+(Bx+C)(x+2)
Observe Q(x) = (x + 2)(x2 + 3), so x3 +2x14
2 +3x+6 =
A
x+2
+ Bx+C
x2 +3
= x3 +2x2 +3x+6
.
3 2
Multiplying by x + 2x + 3x + 6 on both sides yields:

A(x2 + 3) + (Bx + C)(x + 2) = 14

• x = −2: 7A = 14 ⇒ A = 2.

• x = 0: 3A + 2C = 14 ⇒ 6 + 2C = 14 ⇒ C = 4.

• x = −1: 4A − B + C = 14 ⇒ 8 − B + 4 = 14 ⇒ B = −2.

Z  
−2x + 4
Z
14 2
dx = + 2 dx
x3 + 2x2 + 3x + 6 x+2 x +3
Z  
2 2x 4
= − + dx
x + 2 x2 + 3 x2 + 3
u=x2√
+3 Z Z √
v=x/ 3 du 4 3
= ln |x + 2| − + dv
u 3v 2 + 3
Z
4 dv
= 2 ln |x + 2| − ln |u| + √ 2
3 v +1
4 arctan v
= 2 ln |x + 2| − ln |u| + √ +C
3
 
4 arctan √x3
= 2 ln |x + 2| − ln(x2 + 3) + √ +C
3
 
x
(x + 2)2 4 arctan √3
= ln 2 + √ +C
x +3 3

Trevor Birenbaum Page 15

Integration Techniques and Tricks University of Miami Mathematics Union

Q(x) contains at least one repeated quadratic factor

Similar to previous, a repeated quadratic factor of Q(x) with multiplicity k must have terms
for every power of the quadratic factor from 1 to k.

P (x) A1 x + B1 A2 x + B2 Ak x + Bk
= + + ··· + (25)
(a1 x2 + b1 x + c 1 )k 2 2
a1 x + b1 x + c1 (a2 x + b2 x + c2 )2 (ak x2 + bk x + ck )k

Z
4
Find dx
x5 − x4 + 2x3− 2x2 + x − 1
4 A
Observe Q(x) = (x − 1)(x2 + 1)2 , so x5 −x4 +2x3 −2x2 +x−1
= x−1
+ Bx+C
x2 +1
+ Dx+E
(x2 +1)2
=
A(x2 +1)2 +(Bx+C)(x−1)(x2 +1)+(Dx+E)(x−1) 5 4 3 2
x5 −x4 +2x3 −2x2 +x−1
. Multiplying by x − x + 2x − 2x + x − 1 on both
sides yields:

A(x2 + 1)2 + (Bx + C)(x − 1)(x2 + 1) + (Dx + E)(x − 1) = 4

• x = 1: 4A = 4 ⇒ A = 1.

• x = 0: A − C − E = 4 ⇒ C + E = −3. (1)

• x = −1: 4A + 4B − 4C + 2D − 2E = 4 ⇒ 2B − 2C + D − E = 0. (2)

• x = 2: 25A + 10B + 5C + 2D + E = 4 ⇒ 10B + 5C + 2D + E = −21. (3)

• x = −2: 25A + 30B − 15C + 6D − 3E = 4 ⇒ 10B − 5C + 2D − E = −7 (4)

Subtracting (4) from (3) yields 10C + 2E = −14, or 5C + E = −7. Subtracting (1) from this
yields 4C = −4, or C = −1, which results in E = −2. (2) becomes 2B + 2 + D + 2 = 0, or
2B + D = −4. Adding (3) to (4) yields 20B + 4D = −28, or 5B + D = −7. subtracting the
new (2) from this yields 3B = −3, or B = −1, which results in D = −2.

This system has the solution {A, B, C, D, E} = {1, −1, −1, −2, −2}.

Trevor Birenbaum Page 16

Integration Techniques and Tricks University of Miami Mathematics Union

Z Z  
4 1 x+1 2x + 2
dx = − − dx
x − x + 2x − 2x2 + x − 1
5 4 3 x − 1 x2 + 1 (x2 + 1)2
Z
dx
I1 =
x−1
= ln |x − 1| + C
Z Z
x dx
I2 =− 2
dx − 2
x +1 x +1
Z Z
2
u=x +1 1 du dx
= − − 2
2 u x +1
ln |u|
=− − arctan x + C
2
ln(x2 + 1)
=− − arctan x + C
Z 2 Z
2x 2
I3 =− 2 2
dx − dx
(x + 1) (x + 1)2
2

u=x2 +1
2 sec2 θ
Z Z
x=tan θ du
= − − dθ
u2 (tan2 θ + 1)2
2 sec2 θ
Z
1
= − dθ
u sec4 θ
Z
1
= − 2 cos2 θ dθ
u
Z
1
= − (1 + cos(2θ)) dθ
u
1 sin(2θ)
= 2 −θ− +C
x +1 2
1
= 2 − θ − sin θ cos θ + C
x +1
1 x
= 2 − arctan x − 2 +C
x +1 x +1
1−x x−1
I1 + I2 + I3 = 2 − ln √ − 2 arctan x + C
x +1 x2 + 1
The following are left as exercises to the reader.
Z Z
26x + 36 dx
1. 3 2
dx 4.
x − x − 12x 4
x +4
3x4 + 4x3 − 4x2 + 24
Z 5
x + 3x3 + 8x2 − 45x − 58
Z
2. dx 5. dx
x3 + x2 − 4x − 4 x5 + 3x4 − 4x2
Z
dx
Z
dx
3. 6. Generalize
x3 + 2x2 + 2x + 4 (ax2 + bx + c)2

7. Derive the antiderivative of secant using a Weierstrass substitution.

Trevor Birenbaum Page 17

Integration Techniques and Tricks University of Miami Mathematics Union

3 Integration by Parts
We have already discussed the reverse of the Chain Rule when we discussed the Substitution
Rule. Now we can discuss the reverse of the Product Rule, which is known as Integration
by Parts. We know Rdxd
(u(x) · v(x)) =Ru0 (x)v(x) + u(x)v 0 (x). Integrating both sides of this
yields u(x) · v(x) = u0 (x)v(x) dx + u(x)v 0 (x) dx. It’s some basic algebra to get to the
usual statement of Integration by Parts:
Z Z
u dv = uv − v du (26)

This also applies to definite integrals:

Z b Z b
b
u dv = uv|a − v du (27)
a a

Integration by Parts is very useful for integrating the products of functions. A general rule
of thumb for picking u is to follow ‘‘LIATE” order, in decreasing priority (everything else
would be made dv):

1. Logarithmic functions

2. Inverse trigonometric functions

3. Algebraic functions

4. Trigonometric functions

5. Exponential functions

Z u=x Z
dv=sin x
x sin x dx = −x cos x − (− cos x) dx
Z
= −x cos x + (cos x) dx

= sin x − x cos x + C

Integration by Parts may need to be applied multiple times in the solving of one integral.
Z u=x2 Z
dv=ex
x e dx = x e − 2xex dx
2 x 2 x

Z
= x e − 2 xex dx
2 x

u=x Z
dv=ex 2 x
= x e − 2xe + 2 ex dx
x

= x2 ex − 2xex + 2ex + C

Trevor Birenbaum Page 18

Integration Techniques and Tricks University of Miami Mathematics Union

Z u=sinxx Z
x dv=e x
e sin x dx = ex cos x dx
e sin x −
u=cosxx Z
dv=e
= e sin x − e cos x − ex sin x dx
x x

Observe that the original integral has come up again, and can be added to both sides to
obtain:
ex (sin x − cos x)
Z
ex sin x dx = +C
2
The Beta function B(m, n) is defined as follows:
Z 1
B(x, y) = xm−1 (1 − x)n−1 dt (28)
0

Observe that the Bounds Trick can be used to show that the Beta function is symmetric;
B(m, n) = B(n, m). We will now prove the value of the Beta function for positive integers
m and n. Observe that B(0, n) = n1 . Suppose n is fixed.

1 u=xm−1 1 1
(1 − x)n m−1
Z Z
m−1 n−1 dv=(1−x)n−1 m−1
x (1 − x) dx = x · − xm−2 (−(1 − x)n ) dx
0 −n 0 n 0
1
m−1
Z
=0+ xm−2 (1 − x)n dx
n 0
m−1
= B(m − 1, n + 1)
n

m−1 m−2 1 (m − 1)!(n − 1)!

B(m, n) = · · ··· · B(0, n + m) = (29)
n n+1 n+m−2 (m + n − 1)!
The following are left as exercises to the reader.
Z
1. (4x + 3)(x − 2)5 dx
Z
2
2. (2x2 − 1)e−x dx
Z
3. sec3 θ dθ
Z
x
4. esin 2 sin x dx
Z 1
5. (1 − x2 )9 x9 dx
0
∞
X 1
6. 2n


n=1
n n

Trevor Birenbaum Page 19

Integration Techniques and Tricks University of Miami Mathematics Union

3.1 ex as Its Own Derivative

It is important to recall that ex is its own derivative, so many integrals that contain ex as
a factor of the integrand may be susceptible to Integration by Parts when it may not be
immediately obvious. In fact, polynomials multiplied by ex can be integrated in one iteration
of Integration by Parts as opposed to the multiple iterations shown previously.
Z
ex (f + f 0 ) dx = ex f + C (30)

Z Z
x 3 2
e (x + 4x − 6x + 9) dx = ex ((x3 + 3x2 ) + (x2 + 2x) + (−8x − 8) + 17) dx

= ex (x3 + x2 − 8x + 17) + C

Z Z
x 2 2 sin x cos x + 2
e sec x(sin(2x) + 2) dx = ex · dx
cos2 x
Z
= 2 ex (tan x + sec2 x) dx

= 2ex tan x + C

ex (2 − x2 ) ex ex (1 − x2 )
Z Z Z
√ dx = √ dx + √ dx
(1 − x) 1 − x2 (1 − x) 1 − x2 (1 − x) 1 − x2
r
ex
Z Z
x 1+x
= √ dx + e dx
(1 − x) 1 − x2 1−x
r
x 1+x
=e +C
1−x

3.2 Multiplying by One

In addition to the ex trick mentioned previously, it is also important to recognize that u = x
is a function that results in u0 = 1, so the following can occur:
Z u=f (x) Z
dv=1
f (x) dx = xf (x) − xf 0 (x) dx (31)

This is how we can obtain certain antiderivatives such as the antiderivative of the natural
log or of arctangent.
Z u=ln x Z
dv=1 1
ln x dx = x ln x − x · dx
x
Z
= x ln x − 1 dx

= x ln x − x + C

Trevor Birenbaum Page 20

Integration Techniques and Tricks University of Miami Mathematics Union

Z u=arctan x Z
dv=1 x
arctan x dx = x arctan x − dx
1 + x2
Z
u=1+x2 1 du
= x arctan x −
2 u
ln u
= x arctan x − +C
2
ln(1 + x2 )
= x arctan x − +C
2
The following are left as exercises to the reader.
Z
1. (arcsin x + arccos x) dx
Z
2. ln(x2 + 1) dx
Z Z
3. Prove that if f (x) dx = F (x), then f −1 (x) dx = xf −1 (x) − F (f −1 (x)).

3.3 Tic-Tac-Toe, or Tabular Integration

Integration by parts can become extremely tedious if you have to do it over and over again.
It may also become difficult to keep track of the signs of the integrals. So when multiple
integrations by parts are needed to solve an integral, we can construct a table that gives
the final answer. The left column of the table contains the function that you want to
differentiate, and its derivatives. The middle column of the table contains the function you
will be repeatedly integrating, and its antiderivatives (with no constant of integration). The
rightmost column alternates between − and +. Terms in the antiderivative of the function
of interest will be comprised of multiplying an element from each column together. Let one
of these elements be the nth element of the left column. Then the other two elements are the
(n + 1)th elements of the middle and right columns. All of these terms can be added together
to obtain the antiderivative of the original function.

Z x2 sin x
2x − cos x ⊕
x2 sin x dx
2 − sin x
0 cos x ⊕
Z
x2 sin x dx = −x2 cos x + 2x sin x + 2 cos x + C

Trevor Birenbaum Page 21

Integration Techniques and Tricks University of Miami Mathematics Union

(x − 2)3 e2x
Z 3(x − 2)2 e2x /2 ⊕
(x − 2)3 e2x dx 6(x − 2) e2x /4
6 e2x /8 ⊕
2x
0 e /16

(x − 2)3 e2x 3(x − 2)2 e2x 6(x − 2)e2x 6e2x

Z
(x − 2)3 e2x dx = − + − +C
2 4 8 16

x10 sin x
10x9 − cos x ⊕
90x8 − sin x
720x7 cos x ⊕
Z 5040x6 sin x
x10 sin x dx 30240x5 − cos x ⊕
151200x4 − sin x
604800x3 cos x ⊕
1814400x2 sin x
3628800x − cos x ⊕
3628800 − sin x
0 cos x ⊕
Z
x10 sin x = −x10 cos(x) + 10x9 sin(x) + 90x8 cos(x) − 720x7 sin(x) − 5040x6 cos(x)

+ 30240x5 sin(x) + 151200x4 cos(x) − 604800x3 sin(x) − 1814400x2 cos(x)

+ 3628800x sin(x) + 3628800 cos(x) + C

That’s probably a lot easier than doing 10 integrations by parts...

The following are left as exercises to the reader.

Z Z
6 −x
1. x e dx 3. x arctan2 x dx
Z Z
2. 4
ln x dx 4. sin(x)e−x dx
Z
5. Generalize xa sin(bx) dx for positive integer a.
Z
6. Generalize eax sin(bx) dx.

Trevor Birenbaum Page 22

Integration Techniques and Tricks University of Miami Mathematics Union

4 Integration of Parametric Equations

To this point, we have only worked with functions of y in terms of x, where y can be written
out explicitly as y = f (x). However, we can’t always represent a function like this. Both x
and y may need to be written out as functions of another variable, usually t or θ, as x = f (t)
and y = g(t) for some functions f and g. Note that trivially, for the function y = f (x), we
can use the parametrization x = t and y = f (t).

4.1 The Basics

To find the definite integral of y = F (x) from a to b, we will assume a = f (α) and b = f (β).
Z b Z β
x=f (t)
F (x) dx = F (f (t))f 0 (t) dt
a α
Z β
= g(t)f 0 (t) dt (32)
α
Z β
= y dx (33)
α

The values of α and β must be carefully picked to satisfy the conditions given by the problem.
For example, to find the area of the curve bounded by the equations x = 4 − t2 and y = t2 + 3t
in the first quadrant, we have to find where the two curves cross the coordinate axes. For the
given equations, x = 0 at t = ±2, and between those values x is positive. y = 0 at t = −3
and t = 0, between which y is negative and outside of which y is positive. Thus, values of t
to integrate at are the solutions to the simultaneous inequalities −2 ≤ t ≤ 2 and t ≤ −3 OR
t ≥ 0, which is 0 ≤ t ≤ 2. Note that to integrate from left to right to make a positive-valued
integral, x must be increasing t must be decreasing. α must equal 2, and β must equal 0.
Using the formula,
Z 0 Z 0
y dx = (t2 + 3t)(−2t) dt
2
Z2 0
= (−2t3 − 6t) dt
Z2 2
= (2t3 + 6t) dt
0
4 2
t
= + 3t2
2 0
= 20
It is important to note that the area of the curve traced out by two parameters is always
positive, so α and β may need to be switched. Additionally, the curve must only be traced
around the figure once. For example, the unit may be parametrized by x = cos θ and y = sin θ
from 0 to 8π, but this would represent 4 revolutions around the circle, and the area of the
circle from the above formula would be quadrupled if β were set to 8π instead of the correct 2π.

Trevor Birenbaum Page 23

Integration Techniques and Tricks University of Miami Mathematics Union

A cycloid may be parametrized as x = a(θ − sin θ) and y = a(1 − cos θ). Observe that this
makes a complete revolution between θ = 0 and θ = 2π with θ increasing. We can find the
area of a cycloid like this.
Z 2π Z 2π
y dx = a(1 − cos θ) · a(1 − cos θ) dθ
0 0
Z 2π
2
=a (cos2 θ − 2 cos θ + 1) dθ
Z0 2π  
2 cos(2θ) 3
=a − 2 cos θ + dθ
0 2 2
  2π
2 sin(2θ) 3θ
=a − 2 sin θ +
2 2 0
= 3a2 π
It also may help to parametrize a non-parametric function in the integrand. This can be
recognized by the presence of the composition of some function with another invertible
function; let the invertible function be x−1 (t) and solve for x(t), and let the outside function
be y(t). For example, a parametrization makes is much easier than Integration by Parts.
√
Z 2 Z 1
arctan {t3 + 1}−1 dt
3


arctan( x − 1) dx =
1 0
y=arctan t Z 1
x=t3 +1
= arctan t · 3t2 dt
0
u=arctan t 1
t3
Z
dv=3t2 3 1
= t arctan t − 0 2
dt
0 t +1
Z 1 
π t
= − t− 2 dt
4 0 t +1
Z 1
1 2 du
Z
u=t2 +1 π
= − t dt +
4 0 2 1 u
! !
1 2
π t2 ln u
= − +
4 2 0 2 1
π 1 ln 2
= − +
4 2 2
The following are left as exercises to the reader. For the first 3 questions, find the area of
the curve defined by the given parametrizations.
√
1. x = ln t and y = t + t in the first quadrant
2. x = 4t + sin t and y = 2 sin t in the first quadrant
3. x = θ − 4 sin θ and y = 1 − 2 cos θ above the x-axis
√
Z 26  
4. ln 3 x + 1 + 1 dx
0

Trevor Birenbaum Page 24

Integration Techniques and Tricks University of Miami Mathematics Union

4.2 Arc Length

The following is the way to find the length of a function y = f (x) where x is between a and b.
s  2
Z b
dy
1+ dx (34)
a dx

The Chain Rule can be used to find the length of a parametrized curve x = f (t) and y = g(t)
where t is between α and β.
s  2 s
Z b Z b  2  2  
dy dx dy dx dt
1+ dx = + dx
a dx a dt dx dt dx
s
Z β  2  2
dx dy
= + dt (35)
α dt dt
Z βp
= (f 0 (t))2 + (g 0 (t))2 dt (36)
α

Note that trivially, for the function y = f (x), we can use the parametrization x = t and
y = f (t).

Like area, arc length is always positive. To find the length of the perimeter of a cycloid, we
integrate over one revolution from 0 to 2π.
Z 2π p Z 2π p
0 2 0 2
(f (t)) + (g (t)) dt = (a(1 − cos θ))2 + (a sin θ)2 dθ
0 0
Z 2π q
=a (1 − 2 cos θ + cos2 θ) + sin2 θ dθ
Z0 2π
√
=a 2 − 2 cos θ dθ
0
Z 2π r
θ
=a 4 sin2 dθ
2
Z0 2π
= 2a | sin θ| dθ
0
Z π
2
= 8a sin θ dθ
0
π
= −8a cos θ|02
= 8a

Trevor Birenbaum Page 25

Integration Techniques and Tricks University of Miami Mathematics Union

The length of the curve defined by x = 3t2 and y = 2t3 between (3, 2) and (12, 16) is
Z 2p Z 2p
0 2 0 2
(f (t)) + (g (t)) dt = (6t)2 + (6t2 )2 dt
1 1
Z 2 √
=6 t 1 + t2 dt
1
Z 5
u=1+t2 √
= 3 u du
2
3
= 2u 2 |52
√ √
= 10 5 − 4 2
The following are left as exercises to the reader. Find the length of the curve defined by the
given parametrizations.
1. The arcs on the curve y 2 = x3 from (1, −1) to (1, 1)
2. The hyper cycloid defined by x = 4 cos t − cos(4t) and y = 4 sin t − sin(4t)

4.3 Polar Equations

The polar coordinate system is an alternative to the Cartesian coordinate system, where
instead of points being represented by their horizontal and vertical distance from the origin,
points are represented by the angle of the ray that begins at the origin and extends through
the point makes with the horizontal, and the distance from the origin to the point on that ray.
Points are written as (r, θ), where θ is the angle with the horizontal and r is the distance along
the ray. One full turn around the origin is 2π radians. Conversion between the Cartesian
and polar coordinate systems can be done using the following formula:
(x, y) = (r cos θ, r sin θ) (37)
p
The unit circle is represented by r2 = 1. r is taken to be equal to x2 + y 2 . There are two
basic equations whose behavior is important to know in the polar coordinate system. r = r0
represents a circle centered at the origin with radius r0 . θ = θ0 represents a line passing
through the origin with angle θ0 to the horizontal.

The area of the region bounded by the lines θ = α and θ = β and the curve r = f (θ) is
1 β 2
Z
r dθ (38)
2 α
The area bounded by r = eθ between θ = 0 and θ = π is
Z π π
2θ e2θ
e dθ =
0 2 0
2π
e −1
=
2

Trevor Birenbaum Page 26

Integration Techniques and Tricks University of Miami Mathematics Union

The following image represents the major types of curves seen in the polar coordinate system.

The one missing type of curve from the above image is the dimpled limaçon, which is convex
and has equation r = a ± b cos θ or r = a ± b sin θ with a ≥ 2b.

Finding the area of one of these curves requires usage of trigonometric identities and the
general formula for polar areas. Since they are very prevalent in finding polar areas, it is
worth knowing the following integrals offhand.
Z 2π Z 2π
cos θ dθ = sin θ dθ = 0 (39)
0 0
Z π Z π
2 2
cos θ dθ = 2 cos θ dθ = 2 (40)
− π2 0
π
Z π Z
2
sin θ dθ = 2 sin θ dθ = 2 (41)
0 0
Z 2π Z π
2
2
cos θ dθ = 4 cos2 θ dθ = π (42)
0 0
Z 2π Z π
2
sin2 θ dθ = 4 sin2 θ dθ = π (43)
0 0

Trevor Birenbaum Page 27

Integration Techniques and Tricks University of Miami Mathematics Union

To find the area enclosed by the limaçon r = 2 + cos θ, we do this:

Z 2π Z 2π
2
(2 + cos θ) dθ = (4 + 4 cos θ + cos2 θ) dθ
0 0
= 4 · 2π + 4 · 0 + π
= 9π

It is important to ensure the region being integrated over does not intersect itself (consider
the lemniscate at θ = π4 ) or double-count areas (consider the inner loop of some limaçons).
Bounds must be chosen correctly to avoid mistakes. For example, to find the area enclosed
by the inner loop of the limaçon r = 2 + 4 cos θ, we note that r = 0 at θ = π3 and 2π
3
and do
this:
Z 2π Z 2π
1 3
2 1 3
(2 + 4 cos θ) dθ = (4 + 16 cos θ + 16 cos2 θ) dθ
2 π3 2 π3
Z 2π
1 3
= (4 + 16 cos θ + 8 + 8 cos(2θ)) dθ
2 π3
Z 2π
1 3
= (12 + 16 cos θ + 8 cos(2θ)) dθ
2 π3
2π
= 6θ + 8 sin θ + 2 sin(2θ)| π3
3
√
= 2π − 2 3

Similar to in the Cartesian plane, it is also possible to find the area between two polar curves.
The same procedure of finding the values of θ where the curves intersect and using them
as the integrating bounds is followed. The regions defined by r = 3 + 2 sin θ and r = 2
intersect at θ = 7π
6
and θ = 11π
6
. Note that for a continuous single integral, this second bound
π
is equivalent to − 6 . The area outside of r = 2 but inside of r = 3 + 2 sin θ is equal to
Z 7π Z 7π
1 6
2 2 1 6
((3 + 2 sin θ) − 2 ) dθ = (9 + 12 sin θ + 4 sin2 θ − 4) dθ
2 − π6 2 − π6
Z 7π
1 6
= (5 + 12 sin θ + 2 − 2 cos(2θ)) dθ
2 − π6
Z 7π
1 6
= (7 + 12 sin θ − 2 cos(2θ)) dθ
2 − π6
1 7π
= (7θ − 12 cos θ − sin(2θ))|−6 π
2 √ 6

11 3 14π
= +
2 3

Trevor Birenbaum Page 28

Integration Techniques and Tricks University of Miami Mathematics Union

An interesting application of polar integration is that it can be used to prove the area under
2
the error function erf(x) = e−x . The error function actually has no elementary antiderivative,
but the area bounded by it and the x-axis over R (and similarly, over R/− and R/+ ) is known.
Note that double integrals must be multiplied by r and integrated dr dθ (this is proved in
Multivariable Calculus).
Z ∞ Z ∞
−x2 2
e dx = 2 e−x dx
−∞ 0
 Z ∞ Z ∞  12
−x2 −y 2
= 4 e dx e dy
0 −∞
 Z ∞  21
−(x2 +y 2 )
= 4 e dx
0
π
! 12
Z
2
Z ∞
−r2
= 4 re dr dθ
0 0

π
! 12
Z
2
Z −∞
= −2 eu du dθ
0 0

π
! 21
Z
2
Z 0
= 2 eu du dθ
0 −∞

Z π
! 21
2
= 2 (eu |0−∞ ) dθ
0

Z π
! 21
2
= 2 1 dθ
0
 π  12
= 2·
√ 2
= π
Converting to polar coordinates also allows for finding the area of some figures in the Cartesian
plane. Consider the equation x4 + y 3 = x2 y. This graph consists of loops in the first and
second quadrant and trends down towards −∞ in the third and fourth quadrant. Suppose
we want to find the area of one of these loops. Setting x = r cos θ and y = r sin θ yields this:
x4 + y 3 = x2 y
r4 cos4 θ + r3 sin3 θ = (r2 cos2 θ)(r sin θ)
r4 cos4 θ + r3 sin3 θ = r3 cos2 θ sin θ
r cos4 θ + sin3 θ = cos2 θ sin θ
r cos4 θ = cos2 θ sin θ − sin3 θ
r = tan θ sec θ − tan3 θ sec θ
r = tan θ sec θ(1 − tan2 θ)

Trevor Birenbaum Page 29

Integration Techniques and Tricks University of Miami Mathematics Union

r is equal to 0 when θ = 0 and θ = π4 . This can now be integrated like a polar curve.
π
1 1 2
Z Z
1 4
2 2 2 2
tan θ sec θ(1 − tan θ) dθ = u (1 − u2 )2 du
2 0 2 0
1 1 6
Z
= (u − 2u4 + u2 ) du
2 0
1
u7 u5 u3
= − +
14 5 6 0
4
=
105
The following are left as exercises to the reader.

1. Find the area bounded by the region formed by r = θ between θ = 2π, and θ = 4π.

2. Find the area of r = 1 + 4 cos θ to the left of the y-axis.

3. Find the area inside r = 4 − 2 cos θ and outside r = 6 + 2 cos θ.

4. Find the area inside both r = 1 − sin θ and r = 2 + sin θ.

5. Find general forms for the area of all nine named polar curves.

6. Use polar parametrization to derive the integral representing polar curve arc length.

7. Find the arc length of r = θ cos θ between θ = 0 and θ = 2π.

5 Differentiation Under the Integral

Differentiation under the integral is a technique used to evaluate certain integrals by
introducing a second variable (usually a or t) to the integrand and taking the derivative
of the integrand with respect to it. The integrand must be continuous and continuously
differentiable on its bounds. By first converting an integrand f (x) to f (x, t) and finding the
partial derivative of f (x, t) with respect to t, we are able to integrate with respect to x and
then with respect to t (undoing the initial partial derivative) to obtain an answer. Succinctly,
Z b Z b
d ∂
f (x, t) dx = f (x, t) dx (44)
dt a a ∂t
Z b Z Z b 
∂
f (x) dx = f (x, t) dx dt (45)
a a ∂t

Differentiation under the integral as shown in the second formula is accurate to a difference
of a constant, which can be found by choosing a convenient value of t and integrating the
original function with it, or comparison with known integrals and identities.

Trevor Birenbaum Page 30

Integration Techniques and Tricks University of Miami Mathematics Union

Consider the function f (x, t) = (2x + t3 )2 . Two ways to find the integral of this with respect
to x from x = 0 to x = 1 are shown here.
Z 1 Z 2+t3
3 2 u=2x+t3 1
(2x + t ) dx = u2 du
0 2 t3

2+t3
u3
=
6 t3
4
= t6 + 2t3 +
3
Alternatively, differentiating under the integral,
Z 1 Z  Z 1 
3 2 d 3 2
(2x + t ) dx = (2x + t ) dx dt
0 dt 0
Z Z 1 
∂ 3 2
= (2x + t ) dx dt
0 ∂t
Z Z 1 
2 3
= 6t (2x + t ) dx dt
0
Z
6t2 x2 + 6t5 x|10 dt


=
Z
= (6t2 + 6t5 ) dt

= 2t3 + t6 + C

We can observe that when t = 0,

Z 1 Z 1
3 2
(2x + t ) dx = (2x)2 dx
0
Z0 1
= 4x2 dx
0
1
4x3
=
3 0
4
=
3
Thus, C = 43 and the value of the integral is t6 + 2t3 + 43 . Although differentiating under the
integral produced a much longer solution for that integral, differentiating under the integral
makes other integrals significantly easier and shorter if not making them possible when they
were previously impossible.

Trevor Birenbaum Page 31

Integration Techniques and Tricks University of Miami Mathematics Union

Now consider Euler’s integral for the factorial, written out as

Z ∞
n! = xn e−x dx
0

Values of n! can be obtained with this integral by repeated integrating e−x by parts (or using
Tic-Tac-Toe), but this can become tedious and error prone for larger values of n. Instead, we
can introduce a parameter t such that x := tx and differentiate under the integral.
Z ∞
e−x dx = 1
0
Z ∞
te−tx dx = 1
Z0 ∞
1
e−tx dx =
t
Z ∞0
∂ −tx d1
e dx =
∂t dt t
Z 0∞
1
−xe−tx du = − 2
0 t
Z ∞
1
xe−tx du = 2
t
Z ∞0
∂ −tx d 1
xe du =
∂t dt t2
Z0∞
2
−x2 e−tx du = − 3
0 t
Z ∞
2
x2 e−tx du = 3
t
Z0 ∞
6
x3 e−tx du = 4
t
Z0 ∞
24
x4 e−tx du = 5
t
Z0 ∞
120
x5 e−tx du = 6
t
Z0 ∞
720
x6 e−tx du = 7
0 t
..
.
Z ∞
n!
xn e−tx du = n+1
0 t

Setting t = 1 gives Euler’s integral for the factorial.

Trevor Birenbaum Page 32

Integration Techniques and Tricks University of Miami Mathematics Union

It is very helpful to have a knowledge of constants that appear in the derivatives of elementary
functions. For example, the following integral makes use of the fact that the derivative of xt
with respect to t is xt ln x.
Z 1 3 Z 1 t
x −1 x −1
dx = dx (t = 3)
0 ln x 0 ln x
Z Z 1
∂ xt − 1

= dx dt
0 ∂t ln x
Z Z 1 t 
x ln x
= dx dt
0 ln x
Z Z 1 
t
= x dx dt
0
!
1
xt+1
Z
= dt
t+1 0
Z
dt
=
t+1
= ln |t + 1| + C

When t = 0, the numerator of the integral simplifies to 0., to the integral is equal to 0. Thus,
C = 0 and the parametrized integral is equal to ln |t + 1|. Setting t = 3 yields the value of
the original integral: 2 ln 2.

sin x
Another integral that differentiation under the integral allows us to find is that of x
over
the positive reals. The following may be tempting.
Z ∞ Z ∞
sin x sin(tx)
dx = dx (t = 1)
0 x 0 x
Z Z ∞ 
∂ sin tx
= dx dt
0 ∂t x
Z Z ∞ 
= cos(tx) dx dt
0
Z
= ??? dt

This results in an undefined integral, which does not make any sense. What failed here
was that sin(tx)
x
is not continuously differentiable, as can be seen by examining the partial
derivative with respect to t when x = 0. However, there is another way to introduce a
parameter to the integral so that differentiation under the integral becomes possible, by
multiplying by e−tx .

Trevor Birenbaum Page 33

Integration Techniques and Tricks University of Miami Mathematics Union

Z ∞ Z ∞
sin x sin x
dx = e−tx dx
0 x 0 x
Z Z ∞ 
∂ −tx sin x
= e dx dt
0 ∂t x
Z Z ∞ 
−tx sin x
= −xe dx dt
0 x
Z  Z ∞ 
−tx
= − e sin x dx dt
0

This integral is a known result from earlier, obtained with two applications of Integration By
Parts.
Z  Z ∞  Z  ∞
−tx t sin x + cos x −tx
− e sin x dx dt = e dt
0 1 + t2 0
Z
1
= − dt
1 + t2
= − arctan t + C

As t approaches ∞, the parametrized integral approaches 0, so C = π2 and the integral is

equal to π2 − arctan t. Setting t = 0 yields the value of the original integral: π2 .

The parameter may need to be introduced in multiple places, as seen in this integral:
Z 2π Z 2π
cos θ
e cos(sin θ) dθ = ea cos θ cos(a sin θ) dθ (a = 1)
0
Z0 Z 2π 
∂ a cos θ
= e cos(a sin θ) dθ da
0 ∂a
Z Z 2π 
a cos θ a cos θ
= (cos θe cos(a sin θ) − e sin(a sin θ)) dθ da
0
Z Z 2π
∂ ea cos θ sin(a sin θ)

= dθ da
0 ∂θ a
2π
!
ea cos θ sin(a sin θ)
Z
= da
a 0
Z
= 0 da

=C

The value of the integral is the same no matter what the value of a is. Setting a equal to 0 is
just the integral of 1 from 0 to 2π, which is the answer: 2π.

Trevor Birenbaum Page 34

Integration Techniques and Tricks University of Miami Mathematics Union

The following are left as exercises to the reader.

Z ∞
dx
1. Evaluate with differentiation under the integral.
0 x 2 + a2
Z ∞
dx
2. 6
2 x ln x
Z 1
3. (x ln x)2020 dx
0
∞
1 − cos x
Z
4. e−x dx
0 x
∞
e−6x − e−9x
Z
5. dx
0 x
Z π
2 arctan(a tan x)
6. Evaluate x cot x dx by rewriting the integrand as where a = 1.
0 tan x
Z ∞ Z ∞
2
7. Evaluate the Fresnel integrals sin(θ ) dθ and cos(θ2 ) dθ by rewriting the
0 0
sin(x2 (1 + t2 )) cos(x2 (1 + t2 ))
integrands as and .
1 + t2 1 + t2

6 Complex Numbers and Euler’s Formula

The complex (Argand) plane is a geometric representation of the √ complex numbers, which
are all numbers of the form a + bi for real numbers a and b and i = −1. The horizontal axis
of the complex plane represents the real part, and the vertical axis represents the imaginary
part. Complex numbers are traditionally represented by the variable z = x + yi, which
corresponds on the Argand plane to the point (x, y) on the Cartesian plane.

6.1 Introduction
The following result was proved by Euler, and it results in one of the most beautiful equations
in mathematics.

eix = cos x + i sin x (46)

iπ
e +1=0 (47)

Substituting x := −x and solving the resulting system of equations leads to the following
representations for sine and cosine:
eix − e−ix
sin x = (48)
2i
eix + e−ix
cos x = (49)
2
Trevor Birenbaum Page 35
Integration Techniques and Tricks University of Miami Mathematics Union

6.2 Integrals with Complex Numbers

One approach to integrating cos2 θ has already been covered, but there is an easier one with
complex numbers.
2
e + e−iθ
Z Z  iθ
2
cos θ dθ = dθ
2
Z
1
= (e2iθ + e−2iθ + 2) dθ
4
1 e2iθ − e−2iθ
 
= + 2θ + C
4 2i
1
= (sin(2x) + 2x) + C
4
Other integrals of trigonometric functions are made easier with complex numbers.
2  4iθ
e − e−iθ e + e−4iθ
Z Z  iθ 
2
sin θ cos(4θ) dθ = dθ
2i 2
Z
1
=− (e2iθ + e−2iθ − 2)(e4iθ + e−4iθ ) dθ
8
Z
1
=− (e6iθ − 2e4iθ + e2iθ + e−2iθ − 2e−4iθ + e−6iθ ) dθ
8
e6iθ − e−6iθ e4iθ − e−4iθ e2iθ − e−2iθ
=− + − +C
48i 16i 16i
sin(6θ) sin(4θ) sin(2θ)
=− + − +C
24 8 8

6.3 Using Real Parts

It is important to note the following property about real numbers:
x = <(x + yi) (50)
Here, <(z) represents the real part of z. This is particularly useful in the following identity:
cos x = <(eix ) (51)
To demonstrate the usage of taking the real part in integration, consider the integral of
cosine:
Z Z
cos x dx = < eix dx
 ix 
e
=< +C
i
 
cos x + i sin x
=< +C
i
= <(−i cos x + sin x) + C
= sin x + C

Trevor Birenbaum Page 36

Integration Techniques and Tricks University of Miami Mathematics Union

The following integral can be solved without integration by parts!

Z Z
e cos x dx = < ex eix dx
x

Z
= < ex(1+i) dx
 x(1+i) 
e
=< +C
1+i
 ix 
x e (1 − i)
=e < +C
2
 
x (cos x + i sin x)(1 − i)
=e < +C
2
cos x + sin x
= ex +C
2

Let’s revisit integrating ecos θ cos(sin θ) with using the real part.

Z 2π Z 2π
cos θ
e cos(sin θ) dθ = <(ecos θ (cos(sin θ) + i sin(sin θ))) dθ
0 0
Z 2π
iθ
=< (ee ) dθ
Z0 2π
iθ
=< eae dθ (a = 1)
Z0 Z 2π 
∂ aeiθ
=< e dθ da
0 ∂a
Z Z 2π 
iθ aeiθ
=< e e dθ da
0
 
2π
Z aeiθ
e
=<   da
ai
0
Z
= < 0 da

=C

The rest of the integral progresses as before, and the answer is 2π.

The following are left as exercises to the reader.

1 + cos2 θ
Z
3 + 4 cos θ
Z
1. dθ 2. dθ
(3 cos θ + 4)2 cos θ + cos(3θ)

Trevor Birenbaum Page 37

Integration Techniques and Tricks University of Miami Mathematics Union

7 Just for Fun

The following integral is one of a family called Coxeter’s integrals, and is presented as an
example of a very difficult integral that can be solved using the techniques in this packet.
Z π  
2 cos x
arccos dx
0 1 + 2 cos x
Because we are working over the range 0, π2 , the following manipulations are possible.
 

cos(2θ) = 2 cos2 θ − 1
2θ = arccos(2 cos2 θ − 1)
t=cos θ
2θ = arccos(2t2 − 1)
2 arccos t = arccos(2t2 − 1)
z=2t2 −1
2 arccos t =
arccos(z)
r
z+1
arccos(z) = 2 arccos
2
cos x
Let z = 1+2 cos x
.
π π
r cos x
1+
Z   Z
2 cos x 2
1+2 cos x
arccos dx = 2 arccos dx
0 1 + 2 cos x 0 2
Z π r
2 1 + 3 cos x
=2 arccos dx
0 2 + 4 cos x

The following identity will be helpful.

r r
sin2 θ 1 − cos2 θ
tan θ = =
cos2 θ cos2 θ
π π
s
1+3 cos x
r
1 − 2+4
Z Z
2 1 + 3 cos x 2
cos x
2 arccos dx = 2 arctan 1+3 cos x dx
0 2 + 4 cos x 0 2+4 cos x
Z π r
2 1 + cos x
=2 arctan dx
0 1 + 3 cos x
Z π s
x=2t 4 1 + cos(2t)
= 4 arctan dt
0 1 + 3 cos(2t)
s
Z π
4 2 cos2 t
=4 arctan dt
0 1 + 3 − 6 sin2 t
s
Z π
4 cos2 t
=4 arctan dt
0 2 − 3 sin2 t
Z π
4 cos t
=4 arctan p dt
0 2 − 3 sin2 t

Trevor Birenbaum Page 38

Integration Techniques and Tricks University of Miami Mathematics Union

Consider the following identity:

Z 1
a x 1 1
2 2
dx = arctan = arctan
0 x +a a 0 a
We can use this to introduce a second integral, taking the argument of arctan in what we
have so far as the value of a.
Z π Z πZ 1p
4 cos t 4 2 − 3 sin2 t 1
4 arctan p dt = 4 · 2 t dx dt
0 2 − 3 sin2 t 0 0 cos t x2 + cossin
2−3
2t
Z πZ 1 p
4 cos t 2 − 3 sin2 t
=4 2 2 2 dx dt
0 0 x cos t + 2 − 3 sin t
Z πZ 1 p
4 cos t 2 − 3 sin2 t
=4 2 2 2 2 dx dt
0 0 x − x sin t + 2 − 3 sin t
Z πZ 1 p
4 cos t 2 − 3 sin2 t
=4 2 2 2 dx dt
0 0 x + 2 − (x + 3) sin t
q
√2 Z πZ 1 2
cos φ
p
2 − 2 sin2 φ
sin t= 3 sin φ 3 3
= 4 2 2 2 2 dx dφ
0 0 x + 3 − 3 (x + 3) sin φ
√
√ Z 3Z 1 cos2 φ
=8 3 2 2 2
dx dφ
0 0 x + 2(x + 3) cos φ
Z πZ 1 1
u=tan φ √ 3
1+u2 1
= 8 3 2 +3) · dx du
0 0 x2 + 1+u2
2(x 1 + u2
√ Z π3 Z 1 dx du
=8 3 2 2 2 2
0 0 (1 + u )(u x + 3x + 6)

This can be simplified with partial fraction decomposition.

1 v=u2 1
=
(1 + u2 )(u2 x2 + 3x2 + 6) (1 + v)(vx2
+ 3x2 + 6)
A B
= + 2
1 + v vx + 3x2 + 6
A(vx2 + 3x2 + 6) + B(v + 1)
=
(1 + v)(vx2 + 3x2 + 6)
1 = A(vx2 + 3x2 + 6) + B(v + 1)

Setting v = −1 yields A(2x2 + 6) = 1, so A = 2x21+6 .

2
Setting v = −3 yields 6A − 2B = 1, so B = 3A − 21 = 3
2x2 +6
− 1
2
= − 2xx2 +6 .

Note that we are able to set v equal to negative numbers because this is solving a partial
fraction decomposition; there does not necessarily need for values of u to exist in the reals.

Trevor Birenbaum Page 39

Integration Techniques and Tricks University of Miami Mathematics Union

x2
π
!
√ Z 3
Z 1
dx du √ Z π3 Z 1 2x21+6 2
8 3 =8 3 − 2 2 2x +6 2 du dx
0 0 (1 + u2 )(u2 x2 + 3x2 + 6) 0 0 1 + u2 u x + 3x + 6
√ Z 1 Z √3 Z √3 !
1 du du
=8 3 2+6 2+1
− x2 dx
0 2x 0 u 0 u2 x2 + 2x2 + 6
√ Z 1 √
Z √3 !
1 du
=8 3 2+6
arctan u|0 3 − dx
0 2x 0 u2 + 2 + 6x2
√ Z 1 Z √3 !
1 π du
=8 3 2
− √ 2 dx
0 2x + 6 3 0 u2 + 2 + 6x2
√ !!
√ Z 1 3
1 π x ux
=8 3 2
− √ arctan √ dx
0 2x + 6 3 3x2 + 6 3x2 + 6 0
√ Z 1
 
1 π x x
=8 3 2
−√ arctan √ dx
0 2x + 6 3 3x2 + 6 x2 + 2
4π 1
Z Z 1  
dx x x
=√ √ −4 √ arctan √ dx
3 0 x2 + 32 2
0 (x + 3) x + 2
2 x2 + 2
x 1
4π arctan √3
Z 1  
x x
= √ · √ −4 √ arctan √ dx
2 2 x2 + 2
3 3 0 0 (x + 3) x + 2
Z 1
2π 2
 
x x
= −4 √ arctan √ dx
9 2
0 (x + 3) x + 2
2 x2 + 2
Now consider the integral in the second term. The first step of solving this is Integration By
Parts with u = arctan √xx2 +2 and dv = (x2 +3)x√x2 +2 . du can be found as follows:

d x
du = arctan √
dx 2
x +2
 
d x 1
= √ · 2
dx 2
x +2 1 + x2x+2
√ 2
x2 + 2 − √xx2 +2 x2 + 2
= · 2
x2 + 2 2x + 2
1
= √
(x2 + 1) x2 + 2
Z
x
v= √ dx
(x2 + 3) x2 + 2
Z
a=x2 +2 1 da
= √
2 (a + 1) a
√ Z
b= a db
=
b2 + 1
= arctan b
√
= arctan x2 + 2

Trevor Birenbaum Page 40

Integration Techniques and Tricks University of Miami Mathematics Union

 1 Z 1 arctan √x2 + 2
 
Z 1 x arctan √ x  √
x2 +2 x
√ dx = arctan √ arctan 2
x +1 − √ dx
0 (x2 + 3) x2 + 2 x2 + 2 0
2 2
0 (x + 1) x + 2
Z 1 √
π2 arctan x2 + 2
= − √ dx
18 2 2
0 (x + 1) x + 2

Thus,
1 Z 1 √ !
2π 2 2π 2 π2
 
arctan x2 + 2
Z
x x
−4 √ arctan √ dx = −4 − √ dx
9 0 (x + 3) x2 + 2
2 2
x +2 9 18 2 2
0 (x + 1) x + 2
Z 1 √
arctan x2 + 2
=4 √ dx
2 2
0 (x + 1) x + 2

This can be solved by differentiating under the integral.

Z 1 √ Z 1 √
arctan x2 + 2 arctan(t x2 + 2)
√ dx = √ dx (t = 1) (1)
2 2 2 2
0 (x + 1) x + 2 0 (x + 1) x + 2
Z Z 1 √ !
∂ arctan(t x2 + 2)
= √ dx dt
0 ∂t (x + 1) x + 2
2 2

Z Z 1 √ !
1 x2 + 2
= √ · 2 2
dx dt
0 (x + 1) x + 2 1 + t (x + 2)
2 2
Z Z 1 
dx
= 2 2 2 2
dt
0 (x + 1)(t x + 2t + 1)

This can be simplified with partial fraction decomposition.

1 y=x2 1
2 2 2 2
=
(x + 1)(t x + 2t + 1) (y + 1)(t y + 2t2 + 1)
2

A B
= + 2
y + 1 t y + 2t2 + 1
A(t2 y + 2t2 + 1) + B(y + 1)
=
(y + 1)(t2 y + 2t2 + 1)
1 = A(t2 y + 2t2 + 1) + B(y + 1)

Setting y = −1 yields A(t2 + 1) = 1, so A = t21+1 .

1 2
Setting y = −2 yields A − B = 1, so B = A − 1 = t2 +1
− 1 = − t2t+1 .

Note that we are able to set y equal to negative numbers because this is solving a partial
fraction decomposition; there does not necessarily need for values of x to exist in the reals.

Trevor Birenbaum Page 41

Integration Techniques and Tricks University of Miami Mathematics Union

t2
! !
Z Z 1  Z Z 1 1
dx t2 +1 t2 +1
dt = − dx dt
0 (x2 + 1)(t2 x2 + 2t2 + 1) 0 x2 + 1 t2 x2 + 2t2 + 1
Z Z 1 Z 1 
1 dx 2 dx
= −t dt
t2 + 1 2
0 x +1
2 2 2
0 t x + 2t + 1
Z  Z 1 
1 1 dx
= arctan x|0 − 1 dt
t2 + 1 2
0 x + 2 + t2
 
Z Z 1
1 π dx
= − 2  dt

t2 + 1 4
 q
0
x2 + 2 + t12
!
Z 1
1 π t tx
= − √ arctan √ dt
t2 + 1 4 2t2 + 1 2t2 + 1 0
Z  
1 π t t
= −√ arctan √ dt
t2 + 1 4 2t 2+1 2t2+1
Z Z
π dt t t
= 2
− √ arctan √ dt
4 t +1 (t2 + 1) 2t2 + 1 2t2 + 1

Let the parametrized integral in (1) be equal to I(t). We are looking for I = I(1). We can do
this by computing the following:
Z ∞
I 0 (t) dt = I(∞) − I(1)
1
∞
π ∞ dt
Z Z Z ∞
0 t t
I (t) dt = 2
− √ arctan √ dt
1 4 1 t +1 1
2 2
(t + 1) 2t + 1 2
2t + 1
Z ∞
π ∞ t t
= arctan t|1 − √ arctan √ dt
4 1 (t2 + 1) 2t2 + 1 2t2 + 1
Z ∞
π2 t t
= − √ arctan √ dt
16 1
2 2
(t + 1) 2t + 1 2
2t + 1

Let the integral in the second term of this be equal to J.

Z ∞
t t
J= √ arctan √ dt
2 2
(t + 1) 2t + 1 2
2t + 1
1
 
1
Z 0 1 1
u= t
u
= −
q 2 arctan  q u  du
1 u 2 1 2
u2
+1 u2
+1 u2
+1
 
1
Z 1 arctan √
2 + u2
= √ du
0 (1 + u2 ) 2 + u2

Trevor Birenbaum Page 42

Integration Techniques and Tricks University of Miami Mathematics Union

We can now make use of the identity arctan x + arctan x1 = π2 for x > 0.
 
1 √
Z 1 arctan √ Z 1 π
2 + u2 2
− arctan 2 + u2
√ du = √ du
0 (1 + u2 ) 2 + u2 0 (1 + u2 ) 2 + u2
Plugging this back in,
√
π2 π 1
Z 1
arctan x2 + 2
Z
dx
I(∞) − I(1) = − √ + √ dx
16 2 0 (x2 + 1) x2 + 2 2 2
0 (x + 1) x + 2
π2
I(∞) − I(1) = − I(∞) + I(1)
16
π2
I(1) = I(∞) −
32

π 1
Z
dx
I(∞) = √
2 0 (1 + x2 ) x2 + 2
√ Z arctan √1
x= 2 tan u π 2 sec u
= du
2 0 2 tan2 u + 1
1
π arctan √2 cos u
Z
= du
2 0 sin2 u + 1
Z √1
v=sin u π 3 dv
= 2
2 0 v +1
π √1
= arctan v|0 3
2
π2
=
12
π2 2 2 2
Thus, I(∞) = 12
and I(1) = π12 − π32 = 5π
96
. The original integral is 4 multiplied by this, so
Z π
5π 2
 
2 cos x
arccos dx =
0 1 + 2 cos x 24

8 Conclusion
There are more strategies to attacking integrals than are traditionally taught in a standard
Calculus I or Calculus II class. This packet showed a number of interesting ways to solve
integrals. With these techniques and tricks, and with knowledge of identities and Integration
By Ingenious Substitution (IBIS), a wide array of integrals can be solved.

University of Miami Mathematics Union

President Vice-President Treasurer Secretary

Sanjoy Kundu Trevor Birenbaum Chengzhang Fu Daniel Provder

Trevor Birenbaum Page 43