ProbabilisticModeling.jl

NOTICE

This package is unmaintained. Its reliability is not guaranteed.

Introduction

This package will eventually offer a DSL for probabilistic modeling that's built around Julia. Inspired by probabilistic programming languages like BUGS, the DSL supported by this package will allow users to articulate mathematical models in a purely descriptive fashion. The macros provided by this package will parse the model and generate code for a simple sampler that returns a DataFrame in which rows correspond to independent samples and columns correspond to independent columns.

An example of a model that gets compiled into a sampler:

include("utils.jl")

@generate_sampler begin
    mu ~ Normal(0, 100)
    sigma ~ Gamma(10, 1 / 10)
    for i in 1:3
        x[i] ~ Normal(mu, sigma)
    end
end

sampler(4)

To Do

• Expand the sublanguage accepted by @generate_sampler
• Implement @fit_mle and @fit_map