A narrative introduction to probability

Probability is usually taught either as measure theory or as a bag of formulas. This tutorial takes a third path: a narrative that starts from probability over sets, and builds — without skipping the interesting parts — to models a researcher would actually want to use.

Chapters currently cover set-based probability and its instantiation in GenJAX; and probability over continuous values together with inference for continuous models, going from a bimodal Gaussian mixture, to K-component clustering, to a working Bayesian nonparametric model. Six Jupyter notebooks let readers explore the concepts interactively rather than only reading about them.

A Japanese machine translation is also being maintained alongside the English — fitting for a tutorial written in Tokyo, and a small experiment in whether technical exposition survives translation.

The tutorial is a work in progress and feedback is very welcome: josephausterweil.github.io/probintro