Brussels / 3 & 4 February 2024


A murder party with Lea

Lea is a Python module helping to calculate probabilities on situations presenting uncertainties. It is meant to be open, intuitive, and easy.

Lea can deal with discrete variables, interdependent events, conditional probabilities, probabilistic programming (PP) and many more. Unlike most of other probability toolkits, it is open to different probability representations, including fractions and symbols (using the SymPy library). These features make Lea ideal for education, to learn the theory of probability, from basics to advanced concepts.

The talk will present classical or non-classical problems in probability, and how these can be solved easily using Python and Lea. These include games (dice roll, RPG combats, etc.) and funny paradoxes (non-transitive dice, boy or girl paradox, etc.). The “murder party” case will demonstrate how uncertain information can help to find out the most probable perpetrator of a murder. These examples will switch between probabilities as float numbers, fractions and symbols, i.e. returning formulas instead of numbers—and demonstrating practically Python’s “duck typing” at work.

The talk will conclude with a revival of the “bullshit generator”, a 15 years-old, non-serious precursor of ChatGPT (300 lines script, including data)!


Photo of Pierre Denis Pierre Denis