Lean (theorem prover)

Lean (theorem prover)
Name	Lean
Developer	Microsoft Research, Félix Boulanger, Leonardo de Moura, Daniel Selsam
Released	2013
Programming language	C++, Lean 4, Haskell
Operating system	Linux, Microsoft Windows, macOS
License	Apache License

Lean (theorem prover) Lean is an interactive theorem prover and programming language developed for formal proof, verification, and mathematics, combining automated tactics, dependent type theory, and metaprogramming to support formalization projects in logic and computer science. Its design links research from Microsoft Research, influences from Coq, Agda, and Isabelle, and applications in projects associated with Kantian ethics, Langlands program, and Fermat's Last Theorem-adjacent formal efforts. The system has fostered collaborations with institutions such as Imperial College London, Princeton University, Harvard University, and Stanford University.

History

Lean originated in a research group at Microsoft Research led by Leonardo de Moura, with early development beginning around 2013 and public releases following prototypes influenced by work at Carnegie Mellon University and INRIA. The project drew inspiration from proof assistants like Coq, Agda, Isabelle/HOL, HOL Light, and precedents set by logicians at Cambridge University and Princeton University. Early formalizations and tutorials connected Lean to formal efforts at Imperial College London, University of Pittsburgh, and University of Washington; these collaborations intensified after major workshops at International Congress of Mathematicians, Conference on Computer Aided Verification, and Automated Deduction in Geometry.

Design and Architecture

Lean's architecture integrates a kernel, tactic framework, and metaprogramming layer, reflecting design choices comparable to Coq's kernel model, Isabelle's proof object approach, and Agda's type-checker. The kernel enforces consistency similar to mechanisms used at Princeton University and Harvard University's formal verification projects, while the tactic language supports automation strategies influenced by tools from Carnegie Mellon University and Stanford University. The runtime and compilation strategies exhibit engineering patterns seen at Microsoft Research and Google Research, with an eye toward integration with editors developed at MIT and EPFL.

Language and Type Theory

Lean implements a dependent type theory rooted in the calculus of constructions and influenced by work from theorists at University of Cambridge, University of Oxford, and ETH Zurich. Its type system incorporates inductive types, universes, and quotient constructions with theoretical antecedents in papers by researchers at Princeton University and INRIA. The language supports tactics and metaprogramming comparable to extensions proposed in research from Carnegie Mellon University and practical ideas used at Microsoft Research and Stanford University.

Libraries and Formalized Mathematics

Lean's core libraries include a standard library and the mathematical library developed by contributors across Imperial College London, Harvard University, and Princeton University; these repositories host formalizations echoing classic results from Euclid, Euler, Gauss, Galois, and modern theorems studied at University of Cambridge and University of Oxford. Large-scale formalization projects in the ecosystem have addressed topics linked to Algebraic topology-adjacent work at ETH Zurich, Number theory inspired by research at Institute for Advanced Study, and analytic results pursued at Université Paris-Saclay and Max Planck Institute for Mathematics. The libraries interoperate conceptually with libraries from Coq and Isabelle communities while maintaining distinct organization influenced by contributors at Princeton University and Stanford University.

Tools and Ecosystem

An ecosystem of editors, linters, and automation arises from integrations with Visual Studio Code, plugins developed by teams at Microsoft Research and GitHub, continuous integration used at Travis CI and GitLab CI, and package infrastructure similar to systems used at NPM and PyPI. Automation and tactic development have been advanced by projects at Carnegie Mellon University and ETH Zurich, while verification workflows mirror practices at Google Research and Microsoft Research in industrial proof engineering. Educational materials and MOOCs referencing Lean have been hosted by Harvard University, MIT, and Imperial College London.

Adoption and Applications

Lean has been adopted for research at Stanford University, formal verification projects at Microsoft Research, and mathematical formalization initiatives at Princeton University and Imperial College London. Applications include mechanized proofs connected to work pursued at Institute for Advanced Study and verification tasks comparable to those tackled by teams at Google Research and Amazon Web Services. The system has seen use in pedagogy at Harvard University and MIT and in collaborations with institutions such as ETH Zurich and Université Paris-Saclay for large formal libraries.

Development and Community

Development is driven by core contributors from Microsoft Research and an active community spanning Harvard University, Princeton University, Imperial College London, Stanford University, ETH Zurich, and Carnegie Mellon University. The community coordinates via platforms hosted on GitHub and events at conferences such as International Conference on Functional Programming and Conference on Automated Deduction, with workshops held at University of Cambridge and University of Oxford. Outreach and governance resemble collaborative models used by projects at Linux Foundation and Apache Software Foundation.

Category:Theorem provers