Consider a function f and a point a.

There are two numbers that may be associated with f and a.

◦ The value of f at a (if f is defined at a, if a belongs to the domain of f ).

This number is denoted by f (a).

◦ The limit of f (x) as x approaches a (if this limit exists, as will be explained).
This number is denoted by lim f (x).
x→a

Notice that we do not speak of “the limit of f (x) at x = a”, we say “as x
approaches a” (but not equal to a). We write lim f (x), NOT lim f (x).
x→a x=a
The number lim f (x) has nothing to do with f (a).
x→a
In fact, the limit lim f (x) may exist even if f is not defined at a.
x→a
Even if f is defined at a, the limit lim f (x) may not exist, need not exist.
x→a
If both f (a) and lim f (x) exist, they may be equal, or unequal.
x→a

Let f be a function, let a be a point on the real line, let L be a real number.
Crude idea of limit.
If, for every x near a (but not equal to a), f (x) is close to L, we say that L is
the limit of f (x) as x approaches a, and write lim f (x) = L.
x→a
The equation lim f (x) = L encodes information about the values of f at points
x→a
near a, except a.
Better idea of limit.
If, for every x sufficiently close to a (but not equal to a), f (x) is close to L to
within a given degree of accuracy, we say that L is the limit of f (x) as x approaches
a, and write lim f (x) = L.
x→a
Even better.
If, for every prescribed tolerance of error, it is possible to restrict x to a small
open interval about a in such a way that, for every x in that open interval except
a, f (x) − L remains within the prescribed tolerance of error.

For any prescribed tolerance of error, represented by a positive real number ϵ,

there is a numerical measure of “being sufficiently close to a”, represented by a
positive real number δ, so that

for all x ∈ (a − δ, a + δ), x ̸= a: |f (x) − L| < ϵ.

Idea of limit.
For every positive real number ϵ, there is a positive real number δ with this
property:
if x ̸= a and if the difference between x and a is less than δ, then the difference
between f (x) and L is less than ϵ.

1
Definition. Let I be an open interval, let f be a function defined on I, except
perhaps at a. Let L be a real number.
If, for all ϵ > 0, there is a δ > 0 such that

x ∈ I, 0 < |x − a| < δ =⇒ |f (x) − L| < ϵ,

then L is called a limit of f (x) as x approaches a.

1 Theorem. Let I be an open interval, let f be a function defined on I, except

perhaps at a.
Suppose that L1 and L2 are real numbers with the property:
∀ϵ > 0, ∃δ > 0 such that

x ∈ I, 0 < |x − a| < δ =⇒ |f (x) − L1 | < ϵ

and ∀ϵ > 0, ∃δ > 0 such that

x ∈ I, 0 < |x − a| < δ =⇒ |f (x) − L2 | < ϵ.

Then L1 = L2 .

Proof: Suppose L1 ̸= L2 . Suppose L1 < L2 .

Then L2 − L1 > 0. Also 13 (L2 − L1 ) > 0.
There is a δ1 > 0 such that
1
x ∈ I, 0 < |x − a| < δ1 =⇒ |f (x) − L1 | < (L2 − L1 ),
3
and there is a δ2 > 0 such that
1
x ∈ I, 0 < |x − a| < δ2 =⇒ |f (x) − L2 | < (L2 − L1 ).
3

Pick an x0 ∈ I with 0 < |x0 − a| < min{δ1 , δ2 }. We now have

L2 − L1 = |L2 − L1 | = |L2 − f (x0 ) + f (x0 ) − L1 |

1 1
⩽ |L2 − f (x0 )| + |f (x0 ) − L1 | < (L2 − L1 ) + (L2 − L1 )
3 3
2
= (L2 − L1 )
3
< L2 − L1 ,

contradicting the trichotomy law. So L1 ̸= L2 is impossible. We conclude L1 = L2 .

□
The last theorem enables us to speak of the limit of f (x) as x approaches a.

Limit game.
Gameboard for the game lim f (x) = L.
x→a

2
 0 < |x − a| < δ =⇒ |f (x) − L| < ϵ

Limit game on a coordinate system.

L+ϵ

L−ϵ
x
a−δ a a+δ

Uniqueness of the limit

L2

L1

x
a

Example. f (x) = 3x, a = 5.

Guess L = 15.

3
 1
Make f (x) within ϵ = 10 of 15.

1
|f (x) − L| = |3x − 15| < ϵ =
10
1 1 1
|3x − 15| <
3 3 10
1
|x − 5| < ;
30
1 1
|x − 5| < =⇒ |f (x) − 15| < .
30 10
Similarly
1 1
|x − 5| < =⇒ |f (x) − 15| <
300 100
1 1
|x − 5| < =⇒ |f (x) − 15| <
30 000 10 000
ϵ
|x − 5| < =⇒ |f (x) − 15| < ϵ.
3

Example. Let f (x) = x2 . Consider a = 3.

We guess L = 9.
Can we ensure that x2 − 9 is small by requiring that x − 3 is small and different
from 0?

f (x) − L = x2 − 9 = (x + 3)(x − 3)
x − 3 is easier to handle, what about x + 3?
|x + 3| = |x − 3 + 6| ⩽ |x − 3| + 6
|x + 3| < 1 + 6 = 7 if |x − 3| < 1.
Choose x so that 0 < |x − 3| < 1. Then |x + 3| < 7.
Then |f (x) − L| = |x + 3| · |x − 3| < 7|x − 3|.
Then |x − 3| < 7ϵ implies |f (x) − L| < ϵ.
So |x − 3| < min{1, 7ϵ } =⇒ |f (x) − L| < ϵ.

Fundamental example.
Let f (x) = x. Let a ∈ R. Does f (x) have a limit as x approaches a?
Think of the limit game. We have

0 < |x − a| < ϵ =⇒ |x − a| < ϵ.

For any ϵ, there is a δ > 0, namely δ = ϵ such that

0 < |x − a| < δ =⇒ |x − a| < ϵ.

Therefore limx→a x = a.
Recall the sum, difference, . . . , of functions.
Let f , g be functions.
◦ Sum s(x) = f (x) + g(x). s = f + g.

4
 ◦ Difference d(x) = f (x) − g(x). d = f − g.

◦ Product p(x) = f (x)g(x). p = f g.

f (x)
◦ Quotient. q(x) = g(x) (g(x) ̸= 0). q = fg .

◦ Let c be a constant. (cf )(x) = c · f (x). cf .

Rules for limits. Let f , g be functions.

Suppose that lim f (x) exists and is equal to L.
x→a
Suppose that lim g(x) exists and is equal to M .
x→a
Then

◦ lim s(x) exists and is equal to L + M . lim [f (x) + g(x)] = L + M .

x→a x→a

◦ lim d(x) exists and is equal to L − M . lim [f (x) − g(x)] = L − M .

x→a x→a

◦ lim p(x) exists and is equal to LM . lim [f (x)g(x)] = LM .

x→a x→a

f (x) L
◦ lim q(x) exists and is equal to L
M . x→a
lim = (provided M ̸= 0).
x→a g(x) M
◦ Let c be a constant. lim (cf )(x) exists and is equal to cL. lim cf (x) = cL.
x→a x→a

The rules can be written as follows.

◦ lim [f (x) + g(x)] = lim f (x) + lim g(x)

x→a x→a x→a

◦ lim [f (x) − g(x)] = lim f (x) − lim g(x)

x→a x→a x→a

◦ lim [f (x)g(x)] = lim f (x) lim g(x)

x→a x→a x→a

f (x) limx→a f (x)

◦ lim = (provided lim g(x) ̸= 0)
x→a g(x) limx→a g(x) x→a

◦ lim cf (x) = c lim f (x) (if c is a constant)

x→a x→a

if the limits on the right sides exist!

If the limits on the right sides exist!
If the limits on the right sides exist!
Limit of a sum.
2 Theorem. Let I be an open interval and let a ∈ I. Let f , g be functions
defined on I, except perhaps at a.
If limx→a f (x) = L and limx→a g(x) = M , then

lim [f (x) + g(x)] = L + M.

x→a

5
 Proof: We have

|(f (x) + g(x)) − (L + M )| = |(f (x) − L) + (g(x) − M )| ⩽ |f (x) − L| + |g(x) − M |.

Let ϵ > 0 be an arbitrary positive real number.

There is a δ1 > 0 such that

0 < |x − a| < δ1 =⇒ |f (x) − L| < 12 ϵ,

There is a δ2 > 0 such that

0 < |x − a| < δ2 =⇒ |g(x) − M | < 12 ϵ.

Put δ = min{δ1 , δ2 }. Then 0 < |x − a| < δ implies

|(f (x) + g(x)) − (L + M )| ⩽ |f (x) − L| + |g(x) − M | < 12 ϵ + 12 ϵ = ϵ.

For any given ϵ > 0, we are able to find a δ > 0 such that

0 < |x − a| < δ =⇒ |(f (x) + g(x)) − (L + M )| < ϵ,

therefore limx→a [f (x) + g(x)] = L + M . □

Strategy to win the game lim [f (x) + g(x)] = L + M .

x→a

0 < |x − a| < =⇒ |f (x) + g(x) − (L + M )| < ϵ

1 1
Take ϵ, take its half 2 ϵ. With 2 ϵ, play the game lim f (x) = L against an
x→a
expert of that game.

0 < |x − a| < δ1 =⇒ |f (x) − L| < 12 ϵ

1
Record δ1 . With 2 ϵ, play the game lim g(x) = M against an expert of that
x→a
game.

0 < |x − a| < 6 > δ2 =⇒ |g(x) − M | < 12 ϵ

Record δ2 . Compare δ1 and δ2 . Define δ := min{δ1 , δ2 }. Then:

0 < |x − a| < δ =⇒ |f (x) + g(x) − (L + M )| < ϵ

Limit of a difference. Exercise.

Limit of a product.

6
3 Theorem. Let I be an open interval and let a ∈ I. Let f , g be functions
defined on I, except perhaps at a.
If limx→a f (x) = L and limx→a g(x) = M , then

lim f (x)g(x) = LM.

x→a

Proof: We have

f (x)g(x) − LM = f (x)g(x) − f (x)M + f (x)M − LM

= f (x)[g(x) − M ] + [f (x) − L]M
|f (x)g(x) − LM | ⩽ |f (x)||g(x) − M | + |f (x) − L||M |
= |f (x) − L + L||g(x) − M | + |f (x) − L||M |
⩽ (|f (x) − L| + |L|)|g(x) − M | + |f (x) − L||M |.

|f (x)g(x) − LM | ⩽ (|f (x) − L| + |L|)|g(x) − M | + |f (x) − L||M |.

Let ϵ > 0 be given. Consider 1 > 0, too.

There is a δ1 > 0 such that

0 < |x − a| < δ1 =⇒ |f (x) − L| < 1,

There is a δ2 > 0 such that

ϵ 1
0 < |x − a| < δ2 =⇒ |g(x) − M | < ,
2 1 + |L|
There is a δ3 > 0 such that
ϵ 1
0 < |x − a| < δ3 =⇒ |f (x) − L| < ,
2 1 + |M |
Write δ = min{δ1 , δ2 , δ3 }. Then 0 < |x − a| < δ implies

|f (x)g(x) − LM | ⩽ (|f (x) − L| + |L|)|g(x) − M | + |f (x) − L||M |

ϵ 1 ϵ 1
< (1 + |L|) + |M | < ϵ.
2 1 + |L| 2 1 + |M |

For any given ϵ > 0, we are able to find a δ > 0 such that

0 < |x − a| < δ =⇒ |f (x)g(x) − LM | < ϵ,

therefore limx→a f (x)g(x) = LM . □

4 Theorem. Let I be an open interval and let a ∈ I. Let g be a function defined
on I, except perhaps at a.
If limx→a g(x) = M ∈ R and if M ̸= 0, then
1 1
lim = .
x→a g(x) M

7
 −g(x)
1
Proof: We have g(x) −M 1
= Mg(x)M .
Corresponding to 2 |M | > 0, by the definition of the limit, there exists a δ1 > 0
1

such that
0 < |x − a| < δ1 =⇒ |g(x) − M | < 12 |M |.
Then, for all x satisfying 0 < |x − a| < δ1 we have

|g(x)| = | − M − (g(x) − M )| ⩾ | − M | − |g(x) − M | > |M | − 12 |M | = 12 |M |,

thus
1 2
0 < |x − a| < δ1 =⇒ < .
g(x) |M |

Now let ϵ > 0 be given.

Consider |M2| ϵ > 0.
2

There is a δ2 > 0 such that

|M |2
0 < |x − a| < δ2 =⇒ |g(x) − M | < ϵ.
2
Write δ = min{δ1 , δ2 }.
For all x with 0 < |x − a| < δ, we have
1 1 M − g(x)
− =
g(x) M M g(x)
1 1
= |M − g(x)| · ·
|M | g(x)
|M | 2
1 2
< ϵ· ·
2 |M | |M |
= ϵ.

This completes the proof. □

5 Theorem. Let I be an open interval and let a ∈ I. Let f , g be functions
defined on I, except perhaps at a.
If limx→a f (x) = L and limx→a g(x) = M and if M ̸= 0, then
f (x) L
lim = .
x→a g(x) M
1
Proof: This follows from the last two theorems: limx→a g(x) exists and is equal
1
to M , so the limit of f · g1 exist and equals L · M
1 L
=M .
□
Limit of a constant multiple.
6 Theorem. Let I be an open interval and let a ∈ I. Let f be a function defined
on I, except perhaps at a. Let c be a constant.
If limx→a f (x) = L, then

lim cf (x) = cL.

x→a

8
 Proof: Special case of the last theorem. □
Recall a form of the triangle inequality:

|a| − |b| ⩽ |a − b|.

7 Theorem. Let f : R −→ R be the absolute value function. Then

lim f (x) = lim |x| = |a|.

x→a x→a

Proof: We have
|x| − |a| ⩽ |x − a|.
Thus |x − a| < ϵ implies |x| − |a| < ϵ.
For any ϵ > 0, there is a δ > 0, namely δ = ϵ, such that

|x − a| < ϵ =⇒ |x| − |a| < ϵ.

Thus limx→a |x| = |a|. □