WebPPL Unofficial Reference Manual

This documentation style is aimed at being Coq/Rocq-like in one specific way: full of snippets, and examples.

Contents

• Philosophy: docs that don’t lie
  • How we keep examples working
  • Determinism
• Data shapes and call forms
  • Arrays (a.k.a. “lists”)
  • Options objects (named arguments)
  • Tensors: dims + flat array
  • Inference call form (always options object)
• Distributions
  • Sampling and scoring
  • Flip
  • Bernoulli
  • Discrete vs. Categorical
  • Binomial
  • Multinomial
  • Negative Binomial
  • Normal (Gaussian) Distribution
  • Beta
• Functional Programming
  • Functional programming basics
  • Cache (memoization)
• Cookbook (patterns and “when to use what”)
  • Options objects
  • Arrays vs tensors