Contact: Raphaël Fua
Probabilistic proofs repeatedly rely on the same tools. Mathlib, Lean’s mathematical library, contains some of them, such as martingales, concentration inequalities and the central limit theorem, but has little on Markov chains, random walks or couplings.
LeanInfoTheory, developed by Serhat Emre Coban in 2026, builds reusable foundations for information theory on top of Mathlib. The goal of this project is to develop something similar for probability in order to allow other probabilists to more easily formalize their results. We will first decide on the scope of the framework we plan to develop, then work out how to express it in Lean. If time allows, we will then use the resulting library to formalize one of our papers on random graphs and branching processes.
No experience with Lean is assumed. The project requires strong fundamentals in probability and stochastic processes.
How to Apply: send your CV and transcript to raphael.fua@epfl.ch.