Indefinite Integration : Substitution Method

Aritra Mondal
Contents
1 Introduction II
1.1 Form of this content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II
1.2 Who shall be the reader? . . . . . . . . . . . . . . . . . . . . . . . . . . . II
1.3 Notations and other information . . . . . . . . . . . . . . . . . . . . . . . II

2 Some tips and tricks 1

3 Indefinite Integration : Substitution Method 2

3.1 Nonnerd overview introduction . . . . . . . . . . . . . . . . . . . . . . . 2
3.2 Starting: The nerd definition . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3 Standard Integrals Level 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.4 Power technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.4.1 Setting t2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.4.2 Setting tn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.5 Multiple Roots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.6 Standard Trigonometric Substitutions . . . . . . . . . . . . . . . . . . . . 9
3.6.1 Some Methodologies . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.7 Standard Integrals Level 2(R-Quadratic) . . . . . . . . . . . . . . . . . . 12
3.7.1 Bringing the form . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.8 R-Some standard substitution . . . . . . . . . . . . . . . . . . . . . . . . 14
3.8.1 R-Linear and Root Linear . . . . . . . . . . . . . . . . . . . . . . 14
3.8.2 R-Linear and Root Quadratic . . . . . . . . . . . . . . . . . . . . 15
3.8.3 R-Quadratic and Root Quadratic . . . . . . . . . . . . . . . . . . 15
3.8.4 Multiple Powers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4 Exercises 17
~1 Introduction

Thanks to Kaixin Wang, i3435, Pranav and Starchan for proofreading this article and providing
me with valuable suggestions.

~1.1 Form of this content

I have tried to discuss things lucidly and let’s hope it would help the readers to some extent.

~1.2 Who shall be the reader?

Since we are discussing the substitution method, I am not going to start from the very first,
the prerequisite to the article is,
• The anti-derivative concept
• Basic integration rules
• The prelim standard formulas
If you have learned these, this handout is intended for both beginners and intermediate. This
is the handout of a series of handouts on integration, so many things are not covered here in
the substitution method but in later articles.
This is an introductory part though, many things are not covered yet here, those will be done
in later articles.

~1.3 Notations and other information

Sometimes I used some names such as

• Power technique
• Multiple Roots

and some more, those are not any standard names, those names are just which I gave on my
wish :) . By R-something, I usually mean reciprocal of that something.

II
~2 Some tips and tricks

• Please be fluent with use of trigonometry. This will be of great help while solving
problems here and not having a fluent knowledge of them can sometimes ruin your life.
Eventually, trigonometry has a big influence here, so a casual warning.
• Try to wisely apply theorems, techniques, and methods stated in this handout or in
general. Not everything is going to work by those methods easily, but there will always
exist the ’good way’. Don’t forget to check their conditions before applying them.
• Although you can always come to a conclusion of a standard formula of your own self,
it’s advised that you remember the standard formulas to use them quickly wherever
required since you won’t get plenty of time. Moreover, standard formulas might just play
a small role within actual contest problems, so deriving the standard ones might just be
an inefficient use of your time
• Here, there are numerous methods to solve a problem, and upon doing a large amount of
practice only, you would be understanding gradually where to use what, what to try at
first, what we should not do, and pre-thinking of the steps, that’s where our step can
lead to.

1
~3 Indefinite Integration : Substitution Method

A very nice method that is pretty useful in integration. I will show an example of how to do it
rather than explaining to you the theory and stuff.

~3.1 Nonnerd overview introduction

Classic Use — Integrate Z

2x dx
I=
1 + x2

What to do here? Notice that derivative of 1 + x2 is 2x which is our numerator, so we will try
to make the form simpler. Set 1 + x2 = t, differentiating both sides gives

2x dx = dt

Hence we can rewrite I as now Z

dt
= log |t| + c
t
Now substitute t back to in terms of x, i.e

I = log |x2 + 1| + c

Well there could be multiple ways and it’s based upon what observation you can make quickly,
anyone pretty fair with the use of trigonometry may do Trigonometric Substitution who
can see through the form easily. Just set

x = tan θ =⇒ dx = sec2 θ dθ
2 tan θ
Since we know that sin 2θ = 1+tan2 θ
, so the question reduces to
Z Z Z
2
sin 2θ sec θ dθ = 2 sin θ sec θ dθ = 2 tan θ dθ = log | sec2 θ| = log |1 + tan2 θ| + c

So well x2 = tan2 θ, and hence we get the same thing.

Now suppose you are given a bit different question
Z
x + 2 dx
I= √
1 + 4x + x2

2
 Remark. Here before doing anything, please try to think about your method can lead to
where or whether things become simpler, our main objective should be always to make
the structure simple and not complicate it further.

So, let’s see how we can approach this problem, lets looks on the derivative, I will set

1 + 4x + x2 = t2 =⇒ 2x + 4 dx = 2t dt =⇒ x + 2 dx = t dt

Exercise 3.1. Check if setting t works here.

So we have x + 2 in the numerator,

Z Z
x + 2 dx t dt p
√ = = t = 1 + 4x + x2 + c
1 + 4x + x2 t

So we did raise a question earlier in exercise 3.1 that what if would put, would it then work?
Let us check, Set 1 + 4x + x2 = t =⇒ 4 + 2x dx = dt, well now we need to take out a half to
bring 4 + 2x because we have 2(2 + x) so we rewrite the problem as,

1 √
Z Z
1 2x + 4 dx 1 dt p
√ = √ = · 2 t = 1 + 4x + x2 + c.
2 1 + 4x + x2 2 t 2

So we see this too work, so it might be possible that there are multiple ways but you should
choose the best way. Like here, well due to this problem, both methods are easy, however, our
notion should be to keep things out of the root when we have this type of problem. So for
square root, we used t2 to do our job, so when we would write out the final expression or any
further step of the problem, we can avoid the root. Let me show you a problem where setting
t2 shall be the best thing, Z
x + 2 dx
I= √
n + 1 + 4x + x2
Let’s try the ’silly’ way by setting 1 + 4x + x2 = t =⇒ 4 + 2x dx = dt So like the previous time
Z Z
1 2x + 4 dx 1 dt
√ = √
2 n + 1 + 4x + x 2 2 n+ t
√
So we are at the same place back again, having k + j like before, so this process does not
work. Well, we can proceed from here though by setting t = k 2 . Well, that’s similar to setting
to t2 . So we must try to escape out of the roots.
This is the correct way of doing it, set 1 + 4x + x2 = t2 =⇒ x + 2 dx = tdt, so we can now
write it as Z Z Z
t 1
dt = dt − n = t − n log |t + n| + c
t+n t+n
√ √
Now set t back in the terms of x to conclude the result that is 1 + 4x + x2 −n log | 1 + 4x + x2 +
n| + c.

3
If you want to cross-check, we will proceed from where we left the problem (setting with t one)

Z
1 dt
√
2 n+ t
Now we set t = k 2 =⇒ dt = 2k dk, so,
Z
k dk
n+k
well that’s the same form now. So I think it’s pretty clear now.
So what are we doing here basically? Let’s get modern now, it’s called the U − substitution.
Well in general people do with u, but that’s your wish what variable you want to use.

~3.2 Starting: The nerd definition

U -substitution — Suppose for two functions f, g we have

Z
f (g(x))g ′ (x)dx

Then we can transform the equation, if we let g(x) = u and differentiate both sides, we
would have g ′ (x)dx = du. Now see here
Z
f (g(x))g ′ (x)dx

Note that g(x) = u, and g ′ (x)dx = du, so we can transform it to this,

Z
f (u)du

Since things are clear to you now, prove that

Z ′
f (x)
dx = log |f (x)| + c
f (x)

The proof must be quite easy for you now, checking the above, and if not, we have already done
this form at the very start of this content. Here is another one,
Prove that
f (x)n+1
Z
f ′ (x)f (x)n dx = +c
n+1

4
~3.3 Standard Integrals Level 1

We will know of some standard integrals now, I will show you the proof of one of them,
• sec x dx = log | sec x + tan x| + c = log | tan π4 + x2 | + c
R

Proof. Let me show how to integrate sec x, we will try the substitution method, so we know
that derivative of sec x is sec x tan x, hence we will try to bring a tan x term. However
we have tan x, derivative of tan x is sec2 x. So (sec x + tan x)′ = sec x tan x + sec2 x =
sec x(sec x + tan x), if we can bring this form, we are done. So now just multiply by
sec x + tan x
sec x + tan x
′
with sec x, then we will get a ff (x)
(x)
form inside integral(Why?). I hope that solves the
problem after using substitution method.
• cosec x dx = − log |cosec x + cot x| + c = log |cosec x − cot x| + c = log | tan x2 | + c
R

• tan x dx = log | sec x| + c

R

Let’s show you the proof for this one, then you will be able to prove the below one
too, don’t ignore the proof, since the methodology used for solving them serves as the
technique for many other problems of the certain type.
Now, since we have tan x, to try our substitution method, we need a sec2 x, so simply
multiply and divide by sec2 x i.e

tan x sec2 x dx
Z

sec2 x

Setting tan x = t gives sec2 xdx = dt, this gives that

Z Z
t dt 1 2t dt
2
=
1+t 2 1 + t2

Why 1 + t2 , because sec2 x = 1 + tan2 x = 1 + t2 . So now again set 1 + t2 = u =⇒

2t dt = du, this gives the final form as
1
Z
du 1 √
= log |u| + c = log | sec2 x| + c = log | sec x| + c
2 u 2

•
R
cot x dx = log | sin x| + c
• ex [f (x) + f ′ (x)] dx = ex f (x) + c
R

A direct example sayR evaluating ex (cos x − sin x) dx, so cos x derivative is − sin x, hence
R

we can directly say ex (cos x − sin x) dx = ex cos x + c

Let me categorize them as types, the previous part was merely an infant introduction to this
topic.

5
~3.4 Power technique

Type 1(Power technique) — Something we learned just now, is when we have things in
roots in any place which seems to create problems in solving the problem, we try to
escape out of the roots. We ideally set things according to the question. Let us name it
as ”Power technique”.

I will bring out a few cases, but the integration problems are not limited to it. There are many
ways to do a certain problem,

3.4.1 Setting t2

Case I(Setting t2 ). We have already seen an example of setting t2 , but still let’s look at one
more. We do this usually when we have some term in square root.

√
sec3 θ sin 2θ dθ
R
Example 3.1. Integrate the following

When you don’t see any direct substitution, just try doing some trigonometry here!

√ √
r
Z √ Z √ Z
sin θ
sec3 θ sin 2θ dθ = 2 sec3 θ sin θ cos θ dθ = 2 sec2 θ dθ
cos θ
√ Z √
= 2 sec2 θ tan θ dθ

Now set tan θ = t2 , this gives sec2 θ dθ = 2t dt.

Exercise 3.2. Finish the problem.

Exercise 3.3. Try checking if this works by setting tan θ = t also.

3.4.2 Setting tn

Setting tn . We do this when we have a nth root thing. Well you see, that’s from where you
can understand our concept of using t2 for square root. The idea as I said from the start, it’s
all about escaping out of roots. Now let’s try a nice problem.

6
 Example 3.2. Integrate Z
dx
1
xn (1 + xn ) n

This has a new technique along the technique we were about to use now, setting directly won’t
be quite nice(why?), Let’s say why, as said before don’t use any technique blindly, we want to
substitute to make our terms easier if you check the derivative of the term(nxn−1 ) in the root,
it won’t help us to get the x−n to make our substitution work.
Thought Process. So in order to get derivative as this type of form j −k , what should be our
step? We would try to rewrite the term in the way such the term comes something like of form
1
c + j k−1 ,
We will try now, see we can write
   1  1
n 1
n 1 n 1 n
(1 + x ) = x
n +1 =x +1
xn xn

Let’s rewrite the question now Z

dx
 1
n+1 1 n
x +1
xn
NOW, set
1 n 1
n
+ 1 = tn =⇒ − n+1 dx = ntn−1 dt =⇒ n+1 dx = −tn−1 dt
x x x

Now try finishing the question.

Exercise 3.4. Complete the question.

Exercise 3.5. Integrate Z

dx
√
x xn + 1

Multiple Roots. By this I mean suppose we have a term in square root, and that term even
in cube root, and the term in 5th root too, or so on..
What do we do in this case? So we have to set it such that it escapes out of all roots at the
same time. Suppose in a question we have an expression inside a square root and the expression
in a cube root too, so we would set t6 , but why?
The reason, using t6 we can escape out from both cube root and square root. So in general
what would we do

7
~3.5 Multiple Roots

Multiple Roots General — Let Ω(k) mean that k th root on a particular term say z. Suppose
we have in question,
Ω(a), Ω(b), Ω(x), Ω(y), . . .
Then we set z as t(abxy... ) to make the problem simpler.

Let’s see an example

Example 3.3. Integrate Z

x dx
√ √
x+1+ 3x+1

Walk through.
• So we have the term x + 1 is a square root and cube root.
• Hence set x + 1 = t3·2 = t6
•
x t5 dt
Z
5
dx = 6t =⇒ 6
t3 + t2
and since x = t6 − 1, substitute in x. So our term inside the integral is now basically,
Z 11
t − t5 dt
Z 9
t − t3 dt
Z 3 3
t (t − 1)(t + 1)(t2 − t + 1) dt
Z
= = = t3 (t3 −1)(t2 −t+1) dt
t3 + t2 t+1 t+1
Hence our integration is basically now,
Z
6 −t3 + t4 − t5 + t6 − t7 + t8 dt

Which is now trivial. Again saying, don’t forget to substitute back the value in terms of x after
you get the answer.

8
~3.6 Standard Trigonometric Substitutions

Before doing them, here are some trig things things you should know of, you should have known
them before them, but I just get a place to write them in here,
• sin 2x = 2 sin x cos x
2 tan θ
• sin 2x =
1 + tan2 θ
• cos 2x = cos2 x − sin2 x = 2 cos2 x − 1 = 1 − 2 sin2 x
1 − tan2 θ
• cos 2x =
1 + tan2 θ
2 tan(x)
• tan 2x = 1−tan2 (x)

2
• 1 ± sin x = sin x2 ± cos x2 .
q
• sin x2 = ± 1−cos
2
x

q
• cos x2 = ± 1+cos
2
x

• tan x2 = sin x
1+cos x = 1−cos x
sin x

Try proving point 2, 4 and 6.

Now see, let me write it nicely as a table,

Trigonometric substitutions

Substitution Form Substitution Form

x = a sin θ a2 − x2 x = a sin2 θ a−x

x = a cos θ a2 − x2 x = a cos2 θ a−x
x = a tan θ x2 + a2 x = a tan2 θ x+a
x = a cot θ x2 + a2 x = a cot2 θ x+a
x = a sec θ x2 − a2 x = a sec2 θ x−a
x = acosec θ x2 − a2 x = a cosec 2 θ x−a

9
 R√
Exercise 3.6. Try solving this one 1 − x2 dx

I would like to discuss this exercise because I want to show something, here’s a hint to it, set
x = sin θ or cos θ, let me take sin θ, dx = cos θ dθ, hence the problem is now
Z Z
2 1 sin 2θ x
cos θ dθ = cos 2θ + 1 = +
2 4 2
√ √
Now sin θ =√x, then cos θ will be 1 − x2 . And sin 2θ = 2 sin θ cos θ = 2x 1 − x2 , hence our
answer is x2 1 − x2 + sin−1 x + c.
Now some points to note here, the reason why I wrote it,
• Always try to get rid of powers on the trig functions here, in general everywhere(why?)
and try them bringing in their simple forms, by simple form I generally mean something
like ϵ(nx) from their power form where ϵ denotes any random trig function.
The reason is simple, let me hold cos x, for example, to make you understand, we know
how to integrate cos nx, but we don’t know how would we integrate directly cosn x, for
example at the problem above, we reformed cos2 θ in terms of cos 2θ to solve our problem.
• Trigonometric identities should be at your foot nails to solve the problems efficiently.
Usually, there are many ways to solve(not always though), but you should go through
the best way.

3.6.1 Some Methodologies

Form A — Z
dx
a cos x + b sin x
We will make use of trigonometry here, think of a triangle with height a and base b seeing
with the perspective of angle θ, then sin θ = √a2a+b2 and cos θ = √a2b+b2 Well so see that
we can rewrite Z
1 dx
√ a
2
a +b 2 √
a2 +b2
cos x + √a2b+b2 sin x
Now see this can be written as
Z
1 dx
√
a + b2
2 sin θ cos x + cos θ sin x

Well we can know that sin(x + θ) = sin x cos θ + cos x sin θ,

Z Z
dx
= cosec (x + θ) dx = log |cosec (x + θ) − cot(x + θ)| + c
sin(x + θ)

Try this problem

10
 Example 3.4. Z
sin x dx
√
1 + sin x

• We should first think about escaping out of roots. That’s simple right? We already saw
how we can write 1 + sin x as square.
• A thing to worry is, there sin x on numerator, hence we try getting rid of it, a simple
way is write sin x as sin x + 1 − 1, then it will break as two parts.
• Now we will have Z h Z
x xi dx
sin + cos dx −
2 2 sin x2 + cos x2
The first part is easily solvable now.
• However we will now do the second part, the form A mechanism, we would apply here.
Here a = b = 1, so √a2a+b2 = √a2b+b2 = √12 Now if we multiply by √12 at denominator like
these Z
1 dx
√ 1 x
sin √1 x
2 + 2 cos 2
2 √
2

So the denominator part can be written as sin( x2 + π4 )

Form B — Z
dx
a cos x + b sin x + c
In this case usually try to write everything in terms of tan x.

Example 3.5. Z
dx
5 + 4 cos x

1−tan2 x
So how can we write cos x in terms of tan x? Isn’t that simple? Since cos 2x = 1+tan2 x
that
means
1 − tan2 x2
cos x =
1 + tan2 x2
Do this replacement and you will see you get things to substitute(take tan x2 = t). However, in
the end, you will see you are getting the ”form” of something similar to
Z
dx
x ± a2
2

you will understand what form you are getting once you follow the walkthrough, so back to
point, we have not learned this yet, so check the next page to clear your reasoning further.

11
~3.7 Standard Integrals Level 2(R-Quadratic)

• dx 1
tan−1 x
R
x2 +a2
= a a +c

1 x−a
• dx
R
x2 −a2
= log +c
2a x+a

• dx 1 a+x
R
a2 −x2
= 2a log a−x +c
√
• √ dx
R
x2 ±a2
= log x + x2 ± a2 + c

• √ dx = sin−1 x
R
a2 −x2 a +c

3.7.1 Bringing the form

So when we have things like this Z

dx
ax2 + bx + c
So lets go, Z Z
dx dx
2
= √ 2 √ b2 b2
ax + bx + c ( ax) + 2 ax · 2√b a + 4a − 4a +c
Z
dx
√ 2
2
ax + 2√b a − b −4ac
4a

dx
R
So we have the x2 −a2
form!

Remark. As you can see this would work the same way even inside roots and you would
get the respective forms. This is a mechanism when you have things like this. However
here are some mistakes peoples do sometimes. Let me point it out in caution.

 
x2
dx dx 1
tan−1
R R
Caution. Suppose you get x4 +a2
and then you think it as (x2 )2 +a2
= a a .
This is totally wrong! Remember that there is a dx and not dx2 . So simply, think carefully
before you do something.

Let’s try a problem,

12
 R√
Example 3.6. Integrate sec x − 1dx

Here is a walkthrough,
• First thing, don’t take the part as t2 inside the square root. Because we don’t have the
sec x tan x term anyway, so it’s won’t be helpful.
• Write sec x − 1 as 1−cos x
cos x , now see the thing is in the root, if you see careful if we multiply
a 1 + cos x with numerator then that part will come out from root as sin x, so multiply
and divide by 1 + cosx, hence our expression is now
Z
sin x
√
cos x + cos2 x

• Things are too simple now, we have sin x dx above which is the derivative (− cos x),
hence take cos x = t.
• Now you will see we are getting the form. Try from here now. Since the denominator
would be inside the root, either of these two forms may come to use depending upon
√
what you get as the form - √xdx = log x + x2 ± a2 +c or √adx = sin−1 xa +c
R R
2 ±a2 2 −x2

Exercise 3.7. Integrate

Z r
1+x
dx
x
Exercise 3.8. Integrate Z
sin x dx
sin 3x

Exercise 3.9. Integrate Z

dx
√
1 − x2 (1 + x2 )

13
~3.8 R-Some standard substitution

By linear I mean like lx + m and by quadratic I mean like ax2 + bx + c, just writing it because
though I know you know, it just might be hard to understand for someone(maybe?).

3.8.1 R-Linear and Root Linear

R-Linear and Root Linear — When we have this form

Z
dx
√
(lx + m) ax + b

In this case just set ax + b = t2 .

Well, here is an example,

Example 3.7. Integrate Z

3 dx
√
(x + 3) x + 7

Set x + 7 = t2 =⇒ dx = 2t dt, now the question turns to

t−2
Z Z
t dt dt 6
6 =6 = log +c
(t2 − 4)t t2 − 4 4 t+2

Substitute t back in terms of x. Done.

14
3.8.2 R-Linear and Root Quadratic

R-Linear and Root Quadratic — When we have this form

Z
dx
√
(lx + m) ax2 + b
1
In this case just set lx + m = .
t

Example 3.8. Try solving Z

dx
√
(x + 1) x2 − 1

• Set x + 1 = 1
t =⇒ dx = − t12 dt
• Substitute simply, the things we are doing now for so much time. You will see you would
get reliable things.
• Go on your own.

3.8.3 R-Quadratic and Root Quadratic

R-Quadratic and Root Quadratic — When we have this form

Z
dx
√
(lx + m) ax2 + b
2

√
Here set ax2 + b = xt

Try solving Z
dx
√
(3 + 4x2 ) 4 − 3x2

3.8.4 Multiple Powers

Multiple Powers — In this type of forms,

Z
dx
(x − a)m (x − b)n

15
 In this type of forms we set (x − a) = t(x − b).

Here is a simple problem

Example 3.9. Integrate Z

dx
(x − a)4 (x − b)3

This is a kind of direct application, so try it on your own.

We would end here for now. However there are yet many things left that could have been
discussed, however, I hope this is fine for starting with.

16
~4 Exercises

Problem 1. Integrate Z
tan 2x tan 3x tan 5x dx

A simple hints is thinking of tan(2x + 3x).

Problem 2. Integrate Z
dx
sin x + cos x

Did not we do a sin x + b cos x form?

Problem 3. Integrate
ex (1 + x2 ) dx
Z

(1 + x)2

Try to break them uniquely!

Problem 4. If lr (x) means log log . . . x, the log being repeated r times then find
Z
−1
xl(x)l2 (x)l3 (x) · · · lr (x)

dx

Possibly the easiest problem here, don’t go by its looks, instead think what would be the
derivative of lr (x)? Try the last three on own own without hints.

Problem 5. Integrate Z
dx
x log x[log(log x)]

Problem 6. Integrate Z
2x dx
√
(1 − x2 ) x4 − 1

Problem 7. Integrate Z
dx
3
(3x2 − 4x + 2) 2

17