Summary

Latest Version	2.0.0
License	Apache-2.0
CI Status	Failing
Downloads	0
Last Indexed	2026-09-01 07:19

Authors

ayman albaz

Installation

nimble install distributions

choosenim install distributions

git clone https://github.com/ayman-albaz/distributions

OS Compatibility

Platform	Linux	macOS	Windows	FreeBSD	OpenBSD	NetBSD	Android	iOS	WASM	Embedded
distributions	✓	✓	✓	-	-	-	-	-	-	-

Dependencies

Package	Version	Optional
nim >=	2.0.0	No
special_functions >=	1.0.0	No
results >=	0.5.1	No

Source

Repository	https://github.com/ayman-albaz/distributions
Homepage	https://github.com/ayman-albaz/distributions
Registry Source	nimble_official

README

Distributions

Distributions is a Nim library for probability distributions and their functions. Generic over float32 and float64.

Supported Distributions

Distribution	Nim Command
Bernoulli	`initBernoulliDistribution(p = 0.5)`
Beta	`initBetaDistribution(alpha = 2.0, beta = 2.0)`
Binomial	`initBinomialDistribution(n = 10, p = 0.5)`
Chisquare	`initChi2Distribution[float64](df = 1)`
F	`initFDistribution[float64](df1 = 3, df2 = 5)`
Gamma	`initGammaDistribution(k = 2.0, theta = 1.0)`
Negative Binomial	`initNegativeBinomialDistribution(r = 10, p = 0.5)`
Normal	`initNormalDistribution(mu = 0.0, sigma = 1.0)`
Poisson	`initPoissonDistribution(lambda = 10.0)`
t	`initTDistribution(df = 3.0)`
Uniform Continuous	`initUniformContinuousDistribution(a = 0.0, b = 1.0)`
Uniform Discrete	`initUniformDiscreteDistribution[float64](a = 0, b = 1)`

Type parameter defaults to float64 and is inferred from float arguments. Distributions with only int parameters (Chi2, F, UniformDiscrete) require an explicit type, e.g. initChi2Distribution[float64](1). For float32, pass float32 literals: initNormalDistribution(0.0'f32, 1.0'f32). The t-distribution accepts a non-integer df: T parameter, so type inference works: initTDistribution(3.5) infers float64.

Distribution objects expose their parameters as public fields (d.p, d.n, d.df, d.mu, d.sigma, d.lambda, d.alpha, d.beta, d.k, d.theta, d.r, d.a, d.b), so d.p is the success probability of a Bernoulli/Binomial/NegativeBinomial while the quantile argument of ppf is q.

Supported Functions

Constructors return Result — use .get() (or ? propagation) with import std/results. Fallible moments (F.mean/variance/mode, t.mean/variance, Beta.mode, Bernoulli.mode) also return Result. Discrete ppf (Binomial, NegativeBinomial, Poisson, UniformDiscrete) and the NegativeBinomial/Poisson median return Result too (the non-convergence safety net is err(DistError.NotConverged), unreachable for valid parameters).

import std/results

let d = initNormalDistribution(0.0, 1.0).get()
discard d.mean()     # Mean
discard d.median()   # Median
discard d.mode()     # Mode
discard d.pdf(x)     # Probability density function
discard d.pmf(k)     # Probability mass function (discrete distributions)
discard d.cdf(x)     # Cumulative distribution function
discard d.sf(x)      # Survival function (1 - cdf)
discard d.ppf(q)     # Percent point function (quantile, inverse CDF)

# Sampling (all distributions)
import std/random
var r = initRand(0xDEADBEEF)
discard initNormalDistribution(0.0, 1.0).get().sample(r)  # continuous → float64
discard initPoissonDistribution(5.0).get().sample(r)      # discrete → int

Accuracy

~1e-13 relative where convergent (float64); ~1e-5 (float32).

Requirements

Nim >= 2.0.0
special_functions >= 1.0.0

Install

nimble install distributions

TODO

Add more distributions on an as-needed basis.
Add fit, CF, skewness functions.

Performance, feature, and documentation PR's are always welcome.

Contact

I can be reached at aymanalbaz98@gmail.com

distributions