Dicey

Tip

You may be viewing documentation for an older (or newer) version of the gem than intended. Look at Changelog to see all versions, including unreleased changes.

---

The premier solution in total paradigm shift for resolving dicey problems of tomorrow, today, used by industry-leading professionals around the world!

In seriousness, this program is mainly useful for calculating total frequency (probability) distributions of all possible dice rolls for a given set of dice. Dice in such a set can be different or even have arbitrary numbers on the sides. It can also be used to roll any dice that it supports.

Table of contents

• Online (no installation)
  • Recommended
  • For those who want the full command line experience
• Installation
  • Requirements
• Usage: CLI (command line)
  • Example 1: Basic distribution
  • Example 2: Complex distribution with different dice
  • Example 3: Custom dice
  • Example 4: Rolling even more custom dice
  • All ways to define dice
• Usage: API
  • Dice
  • Rolling
  • Calculators
• Diving deeper
• Development
• Contributing
• License

Online (no installation)

Recommended

Use online version of Dicey on its own website: dicey.bulancov.tech!

It does not provide quite all features, but is much easier and quicker to use.

For those who want the full command line experience

Thanks to the efforts of Ruby developers, you can run full Dicey online!

1. Head over to the prepared RunRuby page.
2. Make sure that "-main.rb" is open.
3. Input arguments between "ARGUMENTS" lines, separated by spaces. Refer to Usage / CLI section.
4. Click "Run code" button below the editor.
5. Results will be printed to the "Logs" tab.

If familiar with Ruby, you can also use RunRuby to explore the API. Refer to Usage / API section for documentation.

Installation

Install manually via gem:

gem install dicey

Or, if using Bundler, add it to your Gemfile:

gem "dicey", "~> 0.14"

Tip

Versions upto 0.12.1 were packaged as a single executable file. You can still download it from the release.

Note

dicey 0.0.1 was a completely separate project by Adam Rogers. Big thanks for transfering the name!

Requirements

Dicey is tested to work on CRuby 3.0+, latest JRuby and TruffleRuby. Compatible implementations should work too.

• JSON and YAML formatting require json and yaml.
• Decimal dice require bigdecimal.

Otherwise, there are no direct dependencies.

Usage: CLI (command line)

Following examples assume that dicey (or dicey-to-gnuplot) is executable and is in $PATH. You can also run it with ruby dicey instead.

Note

💡 Run dicey --help to get a list of all possible options.

Example 1: Basic distribution

Let's start with something simple. Imagine that your Bard character has Vicious Mockery cantrip with 2d4 damage, and you would like to know the distribution of possible damage rolls. Run Dicey with two 4s as arguments:

$ dicey 4 4

It should output the following:

# ⚃;⚃
2 => 1
3 => 2
4 => 3
5 => 4
6 => 3
7 => 2
8 => 1

First line is a comment telling you that calculation ran for two D4s. Every line after that has the form roll sum => frequency, where frequency is the number of different rolls which result in this sum. As can be seen, 5 is the most common result with 4 possible different rolls.

If probability is preferred, there is an option for that:

$ dicey 4 4 --result probabilities # or -r p for short
# ⚃;⚃
2 => 0.0625
3 => 0.125
4 => 0.1875
5 => 0.25
6 => 0.1875
7 => 0.125
8 => 0.0625

This shows that 5 will probably be rolled a quarter of the time.

Example 2: Complex distribution with different dice

During your quest to end all ends you find a cool Burning Sword which deals 1d8 slashing damage and 2d4 fire damage on attack. You run Dicey with these dice:

# Note the shorthand notation for two dice!
$ dicey 8 2d4
# [8];⚃;⚃
3 => 1
4 => 3
5 => 6
6 => 10
7 => 13
8 => 15
9 => 16
10 => 16
11 => 15
12 => 13
13 => 10
14 => 6
15 => 3
16 => 1

Results show that while the total range is 3–16, it is much more likely to roll numbers in the 6–13 range. That's pretty fire, huh?

Example 2.1: Graph

If you downloaded dicey-to-gnuplot and have gnuplot installed, it is possible to turn these results into a graph with a somewhat clunky command:

$ dicey 8 2d4 --format gnuplot | dicey-to-gnuplot
# `--format gnuplot` can be abbreviated as `-f g`

This will create a PNG image named [8];⚃;⚃.png:

Example 2.2: JSON and YAML

If you find that you need to export results for further processing, it would be great if a common data interchange format was used. Dicey supports output as JSON and YAML with --format json (or -f j) and --format yaml (or -f y) respectively.

JSON via dicey 8 2d4 --format json:

{"description":"[8];⚃;⚃","results":{"3":1,"4":3,"5":6,"6":10,"7":13,"8":15,"9":16,"10":16,"11":15,"12":13,"13":10,"14":6,"15":3,"16":1}}

YAML via dicey 8 2d4 --format yaml:

---
description: "[8];⚃;⚃"
results:
  3: 1
  4: 3
  5: 6
  6: 10
  7: 13
  8: 15
  9: 16
  10: 16
  11: 15
  12: 13
  13: 10
  14: 6
  15: 3
  16: 1

Example 3: Custom dice

While walking home from work you decide to take a shortcut through a dark alleyway. Suddenly, you notice a die lying on the ground. Looking closer, it turns out to be a D4, but its 3 side was erased from reality. You just have to learn what impact this has on a roll together with a normal D4. Thankfully, you know just the program for the job.

Having ran to a computer as fast as you can, you sic Dicey on the problem:

$ dicey 1,2,4 4
# (1,2,4);⚃
2 => 1
3 => 2
4 => 2
5 => 3
6 => 2
7 => 1
8 => 1

Hmm, this looks normal, doesn't it? But wait, why are there two 2s in a row? Turns out that not having one of the sides just causes the roll frequencies to slightly dip in the middle. Good to know.

Tip

💡 A single positive integer argument N practically is a shorthand for listing every side from 1 to N.

But what if you had TWO weird D4s?

$ dicey 2d1,2,4
# (1,2,4);(1,2,4)
2 => 1
3 => 2
4 => 1
5 => 2
6 => 2
8 => 1

Hah, now this is a properly cursed distribution!

Example 4: Rolling even more custom dice

You have a sudden urge to roll dice while only having boring integer dice at home. Where to find the cool dice though?

Look no further than roll mode introduced in Dicey 0.12:

$ dicey 0.5,1.5,2.5 4 --mode roll # As always, can be abbreviated to -m r
# (0.5e0,0.15e1,0.25e1);⚃
roll => 0.35e1 # You probably will get a different value here.

Note

💡 Roll mode is compatible with --format, but not --result.

All ways to define dice

There are three main ways to define dice:

• "5", "25", or "525": a single positive integer makes a regular die (like a D20).
• "3-6", "-5..5", "(0-1)": a pair of integers with a separator, possibly in round brackets, makes a numeric die with integers in the range.
  • Accepted separators: "-", "..", "...", "–" (en dash), "—" (em dash), "…" (ellipsis).
• "1,2,4", "(-1.5,0,1.5)", or "2,": a list of any numbers separated by commas, possibly in round brackets, makes an arbitrary numeric die.
  • Lists can end in a comma, allowing single-number lists.

"D6", "d(-1,3)", or "d2..4": any definitions can be prefixed with "d" or "D". While this doesn't do anything on its own, it can be useful to not start a definition with "-".

"2D6", "5d-1,3", or "277D(2..4)": any definitions can be prefixed with "Nd" or "ND", where N is a positive integer. This creates N copies of the die.

Usage: API

Note

• Latest API documentation from main branch is automatically deployed to GitHub Pages.
• Documentation for published versions is available on RubyDoc.

Dice

There are 3 classes of dice currently:

• Dicey::AbstractDie is the base class for other dice, but can be used on its own. It has no restrictions on values of sides. For now, it is only useful for rolling and can't be used for distribution calculations.
• Dicey::NumericDie behaves much the same as Dicey::AbstractDie (being its subclass), except for checking that all values are instances of Numeric. It can be initialized with an Array or Range.
• Dicey::RegularDie is a specialized subclass of Dicey::NumericDie. It is defined by a single integer which is expanded to range (1..N).

All dice classes have constructor methods aside from .new:

• .from_list takes a list of definitions and calls .new with each one;
• .from_count takes a count and a definition and calls .new with it specified number of times.

See Diving deeper for more theoretical information.

Note

💡 Using Float values is liable to cause precision issues. Due to in-built result verification, this will raise errors. Use Rational or BigDecimal instead.

DieFoundry

Dicey::DieFoundry#call provides the interface for creating dice from Strings as available in CLI:

Dicey::DieFoundry.new.call("100")
  # same as Dicey::RegularDie.new(100)
Dicey::DieFoundry.new.call("2d6")
  # same as Dicey::RegularDie.from_count(2, 6)
Dicey::DieFoundry.new.call("1d1,2,4")
  # same as Dicey::NumericDie.from_list([1,2,4])

It only takes a single argument and may return both an array of dice and a single die. You will probably want to use Enumerable#flat_map:

foundry = Dicey::DieFoundry.new
%w[8 2d4].flat_map { foundry.call(_1) }
  # same as [Dicey::RegularDie.new(8), *Dicey::RegularDie.from_count(2, 4)]

Rolling

Dicey::AbstractDie#roll implements the rolling:

Dicey::AbstractDie.new([0, 1, 5, "10"]).roll
  # almost same as [0, 1, 5, "10"].sample
Dicey::RegularDie.new(6).roll
  # almost same as rand(1..6)

Dice retain their roll state, with #current returning the last roll (or initial side if never rolled):

die = Dicey::RegularDie.new(6)
die.current
  # => 1
die.roll
  # => 3
die.current
  # => 3

Rolls can be reproducible if a specific seed is set:

Dicey::AbstractDie.srand(493_525)
die = Dicey::RegularDie.new(6)
die.roll
  # => 4
die.roll
  # => 1
# Repeat:
Dicey::AbstractDie.srand(493_525)
die = Dicey::RegularDie.new(6)
die.roll
  # => 4
die.roll
  # => 1

Note

💡 Randomness source is global, shared between all dice and probably not thread-safe.

Calculators

Frequency calculators live in Dicey::SumFrequencyCalculators module. There are four calculators currently:

• Dicey::SumFrequencyCalculators::KroneckerSubstitution is the recommended calculator, able to handle all Dicey::RegularDie. It is very fast, calculating distribution for 100d6 in about 0.1 seconds on my laptop.
• Dicey::SumFrequencyCalculators::MultinomialCoefficients is specialized for repeated numeric dice, with performance only slightly worse. However, it is currently limited to dice with arithmetic sequences.
• Dicey::SumFrequencyCalculators::BruteForce is the most generic and slowest one, but can handle any dice. Currently, it is also limited to Dicey::NumericDie, as it's unclear how to handle other values.
• Dicey::SumFrequencyCalculators::Empirical. This is more of a tool than a calculator. It can be interesting to play around with and see how practical results compare to theoretical ones.

Calculators inherit from Dicey::SumFrequencyCalculators::BaseCalculator and provide the following public interface:

• #call(dice, result_type: {:frequencies | :probabilities}, **options) : Hash
• #valid_for?(dice) : Boolean

See next section for more details on limitations and complexity considerations.

Diving deeper

For a further discussion of calculations, it is important to understand which classes of dice exist.

• Regular die — a die with N sides with sequential integers from 1 to N, like a classic cubic D6, D20, or even a coin if you assume that it rolls 1 and 2. These are dice used for many tabletop games, including role-playing games. Most probably, you will only ever need these and not anything beyond.

Tip

💡 If you only need to roll regular dice, this section will not contain anything important.

• Natural die has sides with only positive integers or 0. For example, (1,2,3,4,5,6), (5,1,6,5), (1,10000), (1,1,1,1,1,1,1,0).
• Arithmetic die's sides form an arithmetic sequence. For example, (1,2,3,4,5,6), (1,0,-1), (2.6,2.1,1.6,1.1).
• Numeric die is limited by having sides confined to ℝ (or ℂ if you are feeling particularly adventurous).
• Abstract die is not limited by anything other than not having partial sides (and how would that work anyway?).

Note

💡 If your die starts with a negative number or only has a single natural side, brackets can be employed to force treating it as a sides list, e.g. dicey '(-1)' (quotation is required due to shell processing).

Dicey is in principle able to handle any numeric dice and some abstract dice with well-defined summation (tested on complex numbers), though not every possibility is exposed through command-line interface: that is limited to floating-point values.

Currently, three algorithms are implemented, with different possibilities and trade-offs.

Note

💡 Complexity is listed for n dice with at most m sides and has not been rigorously proven.

Kronecker substitution

An algorithm based on fast polynomial multiplication. This is the default algorithm, used for most reasonable dice.

• Limitations: only natural dice are allowed, including regular dice.
• Example: dicey 5 3,4,1 '(0)'
• Complexity: O(m⋅n) where m is the highest value

Multinomial coefficients

This algorithm is based on raising a univariate polynomial to a power and using the coefficients of the result, though certain restrictions are lifted as they don't actually matter for the calculation.

• Limitations: only equal arithmetic dice are allowed.
• Example: dicey 1.5,3,4.5,6 1.5,3,4.5,6 1.5,3,4.5,6
• Complexity: O(m⋅n²)

Brute force

As a last resort, there is a brute force algorithm which goes through every possible dice roll and adds results together. While quickly growing terrible in performace, it has the largest input space, allowing to work with completely nonsensical dice, including aforementioned dice with complex numbers.

• Limitations: objects on dice sides must be numbers.
• Example: dicey 5 1,0.1,2 1,-1,1,-1,0
• Complexity: O(mⁿ)

Development

After checking out the repo, run bundle install to install dependencies. Then, run rake spec to run the tests, rake rubocop to lint code and check style compliance, rake rbs to validate signatures or just rake to do everything above. There is also rake steep to check typing, and rake docs to generate YARD documentation.

You can also run bin/console for an interactive prompt that will allow you to experiment, or bin/benchmark to run a benchmark script and generate a StackProf flamegraph.

To install this gem onto your local machine, run rake install.

To release a new version, run rake version:{major|minor|patch}, and then run rake release, which will build the package and push the .gem file to rubygems.org. After that, push the release commit and tags to the repository with git push --follow-tags.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/trinistr/dicey.

Checklist for a new or updated feature

• Running rake spec reports 100% coverage (unless it's impossible to achieve in one run).
• Running rake rubocop reports no offenses.
• Running rake steep reports no new warnings or errors.
• Tests cover the behavior and its interactions. 100% coverage is not enough, as it does not guarantee that all code paths are tested.
• Documentation is up-to-date: generate it with rake docs and read it.
• "CHANGELOG.md" lists the change if it has impact on users.
• "README.md" is updated if the feature should be visible there, including the Kanban board.

License

This gem is available as open source under the terms of the MIT License, see LICENSE.txt.