Outreach
Lean Workshop
This section contains material for the Lean workshop at the event UK welcomes Studienstiftung.
Schedule
- Introduction
- Lean Tutorial: The Natural Number Game
- Coffee Break
- Lean Tutorial Part II & Mini-Project
- Final discussion
Material for mini-projects
Project A: Beyond the Natural Number Game
What's your favourite (undergraduate-level) theorem in mathematics? The idea of this mini-project is to explore how a theorem of your choice is formalised in Lean. You can use the search bar in the Mathlib documentation and refer to search tools listed below. Probably, what you get out of this won't make a lot of sense at first. Take a look at the book Mathematics in Lean to learn more about how to do formalisation beyond the Natural Number Game.
- Mathematics in Lean: The book that teaches you to formalise more than just natural numbers
- Mathlib Documentation: Lean's fast-growing mathematical library
- Lean Online Editor: An easy way to get started writing Lean code without downloading or installing anything
- Loogle: a Lean search engine and LeanSearch, an AI-powered search engine
Project B: A New Understanding of Mathematical Proof
How does formalisation with tools like Lean change how we view mathematical proofs? Or does it not? The idea of this mini-project is to analyse how the adoption of formalisation in mathematical research can change the conception of what constitutes a proof. The material below focuses on perspectives from mathematicans and computer scientists. If you have a background in philosophy: how do various philosophers think about the concept of mathematical proof? Can you find literature on the topic and make a comparison?
- On Proof and Progress in Mathematics: Mathematical proof from the perspective of a renowned research mathematician, written long before Lean was born
- Proof Assistant Makes Jump to Big-League Math: Mathematicians using Lean have verified the accuracy of a difficult theorem at the cutting edge of research mathematics
- Mathematical Proof Between Generations: Researchers with various levels of involvement in formalisation tell their perspective on mathematical proof
- "Understanding in Mathematics is highly overrated": interview with Marijn Heule who works on encoding mathematical problems so they can be solved by computers
Project C: Lean meets AI
Recently, Lean has played an important role in the development of AI systems for mathematics. How do these AI systems work? What are their limitations? And how does Lean come into the picture? The mini-project is to shed light on these questions by analysing the examples linked to below. Note that, recently, a lot of work has focused on solving problems from the International Mathematical Olympiad (IMO). These problems can be explained to anyone with a high-school level understanding of mathematics but are very challenging to solve, requiring multi-step reasoning.
- AlphaProof: Google Deepmind's AI solves IMO problems at silver medal level (last year)
- AI wins Gold @ IMO 2025: a news report, you might also look up the official announcements from OpenAI and Google
- Mathematician's reaction: Terence Tao on Mastodon and a blog post by Kevin Buzzard
- Kimina Prover: Some of my own work (together with amazing collaborators at Numina) — not quite IMO gold level (yet)