Pre: Demo of creating a random number generator¶
Corresponding lab sheet:¶
Objectives¶
- Motivate the use of lists and functions by creating a random number generator.
- Describe list manipulation;
- Describe functions and start discussing writing good code;
- Give insight about random numbers.
Notes¶
Explain to students that we are going to use programming to generate random numbers.
Ask students to discuss in groups how they would generate a random number?
Bring this discussion together, perhaps some students will talk about using dice, flipping a coin, etc…
Ask how would we be able to check that a number is being generated randomly?
Lead the conversation to the notion of being able to predict a number no better than by “chance”.
Discuss how this could be done by a computer, there are actually only very few true random number generators:
- Atmospheric noise.
- Thermal noise.
- Cosmic background radiation measured over a short amount of time.
We are going to look at something called pseudo-random number generators.
There are a number we could choose:
We will consider the last one (LCG) which was considered for a little while to be state of the art (before being proved to be cyrptographically unsafe: ie predictable).
This generator takes the form:
Where \(a, c, m, s\) are some parameters.
In groups choose some parameters and ask students to generate some random numbers.
Here is an example using \(a=2\), \(c=1\), \(s=0\) and \(m=4\):
n | X_n |
---|---|
0 | 0 |
1 | 1 |
2 | 3 |
3 | 3 |
4 | 3 |
5 | 3 |
Is this random? Did any other group come up with more randomness?
Now let us code this, to do so we will make use of two new programming concepts:
- Lists
- Functions
First let us write a function that represents the definition of the random number generator:
>>> def random_number_generator(previous_term,
... modulus=4,
... multiplier_a=2,
... multiplier_c=1):
... """
... Generate a random number using
... the linear congruential generator
... """
... return (previous_term * multiplier_a + multiplier_c) % modulus
Let us confirm the table above:
>>> random_number_generator(previous_term=0)
1
>>> random_number_generator(previous_term=1)
3
>>> random_number_generator(previous_term=3)
3
>>> random_number_generator(previous_term=3)
3
This becomes quickly tedious: it would be much easier to be able to “hold” the calculated numbers in a container of some sort. In python these are called lists:
>>> seed = 0
>>> random_numbers = [seed]
>>> number_of_numbers = 10
>>> for _ in range(number_of_numbers):
... random_numbers.append(random_number_generator(random_numbers[-1]))
>>> random_numbers
[0, 1, 3, 3, 3, 3, 3, 3, 3, 3, 3]
If the numbers we were generating were truly random what would the mean of our list be?:
>>> sum(random_numbers) / len(random_numbers)
2.545...
>>> sum(range(4)) / 4
1.5
Our choice of parameters is clearly poor, here is a more common choice:
>>> modulus = 2 ** 32
>>> multiplier_a = 1664525
>>> multiplier_c = 1013904223
I am going to use the code I wrote previously so I will wrap it in a function:
>>> def generate_random_numbers(number_of_numbers, seed, modulus, multiplier_a, multiplier_c):
... """Generate N random numbers"""
... random_numbers = [seed]
... for repetition in range(number_of_numbers):
... random_numbers.append(random_number_generator(random_numbers[-1],
... modulus=modulus,
... multiplier_a=multiplier_a,
... multiplier_c=multiplier_c,))
... return random_numbers
Let us generate a thousand random numbers:
>>> random_numbers = generate_random_numbers(number_of_numbers=10 ** 3,
... seed=0,
... modulus=modulus,
... multiplier_a=multiplier_a,
... multiplier_c=multiplier_c)
>>> sum(random_numbers) / len(random_numbers)
2114463563.02497...
We will see in a few weeks time how to plot with python but here’s a quick example:
>>> import matplotlib.pyplot as plt
>>> plt.plot(random_numbers)
[<matplotlib.lines...
Lab sheet¶
Show how these two things will be gone over in the lab sheet. Potentially discuss how the previous demo could be improved.
Highlight that this is just a demo of using lists and functions: not a course on random number generation: in fact this algorithm is known to not be sufficient and also python and most programming languages all have random number generators that can be used directly.