Stan distributions

A web app to visualize distributions in Stan. It uses Stan Math C++ code to evaluate the log probability density/mass functions. The Stan Math C++ is compiled to Webassembly using Emscripten. d3.js is used for visualizations.

This was inspired by the distribution zoo. The main differences of this approach are:

• it runs entirely in the user's browser and does not require a server (very easy to fork and deploy your own custom version),
• it uses the actual Stan C++ for all density/mass functions (Distribution zoo is based on R distribution implementations).

This currently supports the majority of Stan's univariate continuous distributions. Missing discrete distributions will be added shortly.

Links

Unbounded Continuous Distributions

• cauchy

• double_exponential

• exp_mod_normal

• gumbel

• logistic

• normal

• skew_double_exponential

• skew_normal

• std_normal

• student_t

Positive Continuous Distributions

• chi_square

• exponential

• frechet

• gamma

• inv_chi_square

• inv_gamma

• lognormal

• rayleigh

• scaled_inv_chi_square

• weibull

Positive Lower-Bounded Distributions

• pareto

• pareto_type_2

• wiener

Continuous Distributions on [0, 1]

• beta

• beta_proportion

Circular Distributions

• von_mises

Bounded Continuous Distributions

• uniform

Unbounded Discrete Distributions

• neg_binomial

• neg_binomial_2

• neg_binomial_2_log

• poisson

• poisson_log

Custom distributions

• beta_unbounded

Contributed by @noejn2.

Running it locally

1. Clone the repository

git clone https://github.com/rok-cesnovar/stan-distributions/

2. Start a simple Python server

This is required in order to serve the .wasm files. The actual computation still happens in the browser.

python3 -m http.server 8000

3. Open localhost:8000 in your favorite browser. Change the port number in step 2 if needed.