Chapter 5: Central limit theorems

Nguyen Minh Tri

University of Information Technology

April 1, 2025

5.1 Law of large numbers

Theorem 5.1 Let X , . . . , X be a set of mutually independent random variables with a
common distribution with E(X ) = µ, and var(X ) = . Then for any " > 0

P( X µ ") 1
n"

1
and as n the probability approaches 1, where X = X.
n
Equivalently, for every " > 0,
P X µ " 1

as n , where S = X.

We can start with a random experiment whose outcome cannot be predicted with certainty,
and by taking averages, we can obtain an experiment in which the outcome can be predicted
with a high degree of accuracy.
5.2 Central limit theorem
Theorem 5.2 If X , . . . , X be a set of mutually independent random variables with a
common distribution with E(X ) = µ < and var(X ) = < , then the limiting
X µ
distribution of Z = as n is the standard normal probability distribution.
/ n
If X , X , . . . , X satisfies the assumptions of the central limit theorem, then for large n,
X +X + +X N (nµ, n )
X +X + +X
N (µ, /n)
n
Example 5.3 Customers at a popular restaurant are waiting to be served. Waiting times
are independent with mean 30 minutes and standard deviation 30. If 16 customers are
waiting what is the probability that their average wait is less than 25 minutes?
Solution. Let X be the time that the ith customer waits to be served. The average
X + ... + X
waiting time of 16 customers is X = . The mean and standard deviation
16
of each X : µ = 30, = 30. By the central limit theorem, X N (µ; /n) and
X µ 25 30
P (X < 25) = P <
/ n 30/ 16
= P (Z < 0.67) = 0.251.
Example 5.4 A hard disk has 330 megabytes of free space. Given 300 independent images,
the average size of each image is 1 megabyte with a standard deviation of 0.5 megabytes.
What is the probability that this hard drive can store 300 images?

Solution. Let X be the file size of the ith photo and S = X . We have n =
300, E(X ) = µ = 1 and var(X ) = = 0.5 . It follows from the CLT, S N (nµ; n )
and then
S nµ 330 nµ
P (S 330) = P
n n
S 300 330 300
=P
300.0, 5 300.0, 5
= P (Z 3, 46) = 0, 9997.
Example 5.5 An electrical firm manufactures light bulbs that have a length of life that is
approximately normally distributed, with mean equal to 800 hours and a standard deviation
of 40 hours. Find the probability that a random sample of 16 bulbs will have an average
life of less than 775 hours.
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
5.3 Approximations for discrete distributions
Let X , X , . . . , X be a random sample from a Bernoulli distribution with mean
µ = p and variance = p(1 p), where 0 < p < 1. Then Y = X B(n; p).

The central limit theorem states that the distribution of

Y np X p
Z= =
np(1 p) p(1 p)/n

is N (0, 1) in the limit as n .

Thus, if n is “sufficiently large,” the distribution of Y is approximately
N (np, np(1 p)), and probabilities for the binomial distribution B(n, p) can be
approximated with this normal distribution.
A rule often stated is that n is sufficiently large if np 5 and n(1 p) 5.
Note that we shall be approximating probabilities for a discrete distribution with
probabilities for a continuous distribution.
Recall that the probability P (X = x) may be positive if X is discrete, whereas it is
always 0 for continuous X. Thus, a direct use of Y Z will always approximate this
probability by 0.
This is resolved by introducing a continuity correction. Expand the interval by 0.5
units in each direction, then use the Normal approximation.

Note that if X is a discrete random variable, then

P (X = a) = P (a 0.5 < X < a + 0.5).

therefore, the continuity correction does not change the event and preserves its
probability.
It makes a di↵erence for the Normal distribution, so every time when we
approximate some discrete distribution with some continuous distribution, we should
be using a continuity correction.
Example 5.6 A new computer virus attacks a folder consisting of 200 files. Each file gets
damaged with probability 0.2 independently of other files. What is the probability that
fewer than 50 files get damaged?
Solution.
X : The number of damaged files. Then X has Binomial distribution with
n = 200, p = 0.2 (or X B(n = 200; p = 0.2))
Check to see whether a normal approximation can be used

np = 40 > 5 and n(1 p) = 160 > 5.

X N µ; , where

µ = np = 40; = np(1 p) = 5.5657

Using the continuity correction

 P (X < 50) = P (X 49) = P (X < 49 + 0.5)
X 40 49.5 40
P <
5.5657 5.657
= P (Z < 1.68)
= (1.68) = 0.9535
Example 5.7 A data center uses a type of hard drive that has a 0.85 probability of func-
tioning properly for 5 years without failure. Suppose that a company installs 600 such hard
drives in its servers. Determine the probability that:
a. exactly 500 hard drives will still be functioning after 5 years.
b. Between 510 and 550 hard drives (inclusive) will still be functioning after 5 years.
Solution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................

....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................
....................................................................................