## Subscribe to

# Ruby Magic

Magicians never share their secrets. But we do. Sign up for our Ruby Magic email series and receive deep insights about garbage collection, memory allocation, concurrency and much more.

Random numbers are useful for a variety of purposes such as in gaming, encryption and building simulations. Technically, computers cannot generate random numbers purely by computation. It is fundamentally impossible to produce truly random numbers on any deterministic device. The best you can hope for is pseudorandom numbers, a stream of numbers that appear as if they were generated randomly.

In this article, we’ll look at the various ways you can generate random numbers in Ruby.

To start off, let’s generate random numbers with the `rand`

method. When the method is called with no arguments, it returns a float that is greater than or equal to 0.0 and less than 1.0.

1 2 | rand() > 0.7308136972953823 |

To get an integer, you pass an integer to the function. The function will return a random integer value that is greater than or equal to 0 and less than the integer passed to the function. Each time the following is run, you will get a number that is between 0 and 7.

1 2 | rand(8) > 5 |

For a random number within a particular range, pass the Range to `rand`

.

The following uses an *inclusive* Range to generate random numbers from a lower limit (`1`

), up to (and including) the upper limit (`10`

).

1 2 | rand(1..10) > 6 |

The next example uses a *non-inclusive* Range to generate random numbers from a lower limit, up to (but not including) the upper limit.

1 2 | rand(1...10) > 9 |

The range can also be between floating point values.

1 2 | rand(1.5..3.0) > 1.7494305393711571 |

You can also use negative range limits with `rand`

.

1 2 | rand(-5..-1) > -5 |

Passing in single negative numbers may give surprising results, as shown below.

1 2 3 4 5 | rand(-100) > 94 rand(-0.5) > 0.7692627344737486 |

This is because for an argument `n`

passed into `rand`

, `rand`

returns random numbers from 0 up to (but not including) `n.to_i.abs`

. For the above example `(-100).to_i.abs`

is `100`

and `(-0.5).to_i.abs`

is `0`

, thus the resulting random numbers.

Calling `rand(0)`

is similar to calling `rand()`

. You will get random numbers between 0.0 and 1.0 (not inclusive).

Before moving on to the next method of generating random numbers, let’s first look at the `srand`

function.

`Kernel#srand`

sets the seed for `Kernel#rand`

. We can use it to generate repeatable sequences of random numbers between different runs of the program.

To understand what this means, we first need to understand how random numbers are generated.

As stated earlier, computers don’t generate truly random numbers purely from computation. What they do is generate a sequence of numbers that seem random. To do this, the computer starts with a seed number, which it runs through some algorithm and then spits out a seemingly random output.

The seed number is generated by the computer using a combination of different elements, e.g. timestamp, the process ID of the program, e.t.c. Because these elements vary for each request to generate a random number, the seed number will always be different, which would produce a different sequence of numbers, thus the resulting random number output. If you ran the algorithm with the same seed, then you would get the same sequence of numbers each time. This is what `Kernel#srand`

allows us to do.

`srand`

is usually used in testing. It could be handy for testing code in your app that deals with randomness, with values that are random but still predictable enough to test. It could also help in isolating or reproducing bugs.

Below we use `srand`

to set the seed and then call `rand`

first to produce a couple of individual random numbers and then to produce a couple sequences of random numbers.

1 2 3 4 5 6 7 8 9 10 11 12 13 | srand(777) rand() > 0.152663734901322 rand() > 0.3023566097075212 10.times.map { rand(10) } > [7, 1, 7, 4, 7, 9, 8, 7, 2, 0] 10.times.map { rand(10) } > [1, 2, 4, 5, 7, 1, 7, 2, 2, 7] |

If you run `srand`

again with the same seed and make the same calls we made previously, you will see that we get the same random numbers.

1 2 3 4 5 6 7 8 9 10 11 12 13 | srand(777) rand() > 0.152663734901322 rand() > 0.3023566097075212 10.times.map { rand(10) } > [7, 1, 7, 4, 7, 9, 8, 7, 2, 0] 10.times.map { rand(10) } > [1, 2, 4, 5, 7, 1, 7, 2, 2, 7] |

You can also generate random numbers with the Random class.

The class method `rand`

provides the base functionality of `Kernel#rand`

along with better handling of floating point values.

1 2 | Random.rand(1...10) > 5 |

Unlike `Kernel#rand`

, if `Random.rand`

is given a negative or `0`

argument, it raises an ArgumentError.

In the real world, many things tend to follow a Normal Distribution. If you have a range of values that something falls under, rarely do you get an equal distribution of all the values. Mostly, a majority of the data tends to fall within a smaller range, with a smaller percentage falling within the larger range. Let’s take an adult man’s height as an example. The shortest height recorded is 54.6 cm (21.5 inches) while the tallest is 267 cm (8'9"). If you want to generate data to simulate the height of men in a population, you might not want to use `rand`

with these limits. You don’t want the probability of getting an 8'9" man to be the same as getting a 6’ man, because the latter is more common.

Other examples of things that follow a Normal Distribution are:

- Errors in measurements
- Blood pressure
- Test scores
- Weight of an adult man/woman

To generate better random numbers for such use cases, you can use the `rubystats`

gem.

1 | ```
$ gem install rubystats
``` |

1 2 3 4 5 6 | require 'rubystats' adult_male_height = Rubystats::NormalDistribution.new(178, 10) sample = 50.times.map { adult_male_height.rng.round(1) } > [183.2, 169.5, 189.7, 171.9, 176.0, 179.3, 189.3, 175.3, 188.3, 190.0, 185.5, 182.8, 187.2, 191.6, 185.4, 178.4, 187.1, 183.3, 189.6, 179.7, 172.7, 174.4, 153.8, 197.4, 176.0, 174.6, 181.1, 182.0, 204.7, 185.2, 175.9, 167.7, 160.6, 170.5, 169.3, 160.6, 165.6, 166.4, 182.6, 179.7, 183.1, 171.9, 185.4, 175.4, 179.7, 176.9, 160.6, 173.8, 181.9, 190.2] |

In the above, we pass the average height for men (178cm) and a standard deviation of 10cm to `NormalDistribution.new`

, before generating 50 values that fall in this normal distribution. If you are curious about the math, this article may interest you.

That brings us to the end of this discussion. We covered a few different ways of creating ‘random’ numbers in Ruby, with `rand`

, `srand`

, `Random`

and `Rubystats`

. We also briefly touched on how 'random’ numbers are created and looked at the reason why deterministic devices cannot create real random numbers.

You should note that the methods covered are not ideal for all use cases that call for randomness. The integer or floating point numbers generated by the methods might be ideal for producing chance in gaming or for creating simulations, but in situations that call for some security, for instance when generating a password reset token, you should consider using SecureRandom. With `SecureRandom`

, you can generate hexadecimal, base64, binary and UUID strings that are much harder to crack as compared to plain numbers.

We hope you found some of this interesting. If you have any comments or questions on what we covered, please reach out to us @AppSignal. You can also send us your requests for topics you want covered.

*This post is written by guest author Joyce Echessa. Joyce is a full stack developer, and loves words and literature.*

*We’ve updated this article on August 1st, 2018 to include a note on SecureRandom*

- ActiveRecord vs. Ecto Part Two
- Improved Navigation
- The Magic of Class-level Instance Variables
- ActiveRecord vs. Ecto Part One
- #to_s or #to_str? Explicitly casting vs. implicitly coercing types in Ruby
- How OTP Applications are structured
- Differences Between #nil?, #empty?, #blank?, and #present?
- Closures in Ruby: Blocks, Procs and Lambdas
- Push and Pub/Sub in Ruby
- Lists vs Tuples in Elixir