Lean (proof assistant)

Lean (proof assistant)
Name	Lean
Developer	Microsoft Research, Carnegie Mellon University, Imperial College London
Released	2013
Programming language	C++, Lean, OCaml
Operating system	Unix-like, Microsoft Windows, macOS
Genre	Proof assistant, Interactive theorem prover
License	Apache License 2.0

Lean (proof assistant) is an interactive theorem prover and proof assistant developed to support formalization of mathematics and verification of software and hardware. It was initiated in an academic setting and later maintained by research groups associated with Microsoft Research, Carnegie Mellon University, and Imperial College London; it emphasizes a dependently typed core, automation, and a large community-driven mathematical library. The system has been used alongside projects involving Andrew Wiles, Grigori Perelman, Alan Turing, Ada Lovelace, and institutions such as Princeton University, Harvard University, Massachusetts Institute of Technology, University of Cambridge, and École Normale Supérieure through collaborations or comparative formalization efforts.

History

Lean was started in the early 2010s by a team led by researchers at Microsoft Research and later developed with contributors at Carnegie Mellon University and Imperial College London. Its development occurred in the context of preceding systems like Coq, Isabelle/HOL, HOL Light, Agda, and NuPRL, reflecting influence from projects at University of Oxford, Cornell University, and Stanford University. Early public announcements and workshops took place at venues including ICFP, CADE, CPP (conference), and ICLR where formal methods and proof assistants were discussed alongside work from Tony Hoare and Robin Milner. Over time Lean gained traction through formalization efforts associated with researchers at University of Chicago, University of Pennsylvania, and collaborative efforts tied to the Simons Foundation and Royal Society.

Design and features

Lean is designed around a dependently typed lambda calculus with a hierarchy of universes, drawing on type theory traditions found in work by Per Martin-Löf and influenced by type-checking research at Harvard University and Princeton University. The system supports tactics, automated decision procedures, and a term elaboration mechanism informed by techniques from Robert Harper and Benjamin Pierce. Key features include an interactive tactic framework similar to that used in Coq and Isabelle/HOL, a meta-programming environment used in projects affiliated with Carnegie Mellon University and Microsoft Research, and integration with editors such as Visual Studio Code and experimentations in Emacs. Lean's syntax and elaborator support notation and coercions comparable to practices at University of Edinburgh and University of Cambridge, while automation draws on algorithms studied at ETH Zurich and University of Tokyo.

Mathematical library (mathlib)

The community-maintained library, mathlib, provides large-scale formalizations of mathematics and has become a central resource comparable in ambition to libraries maintained in Coq and Isabelle. Mathlib hosts contributions from researchers at Princeton University, Imperial College London, University of Chicago, University of California, Berkeley, and University of Oxford. It contains formal proofs and definitions spanning areas associated with mathematicians and works such as Euclid's Elements, Évariste Galois-related algebra, results in the style of Carl Friedrich Gauss, and theorems influenced by modern work at Institute for Advanced Study. The library's governance and development model reflects collaborative workflows seen in projects supported by the Simons Foundation and institutions like Stanford University.

Formalization projects and applications

Lean has been used to formalize a variety of significant theorems and projects analogous to efforts carried out in other systems by teams at Princeton University and Massachusetts Institute of Technology. Notable formalizations include results in number theory inspired by work of Andrew Wiles and structural developments in topology and analysis related to themes explored by Henri Poincaré and John Milnor. Applications extend to software verification projects at Microsoft Research and hardware verification work similar to projects at ARM Holdings and Intel Corporation. Lean has also been adopted in educational settings at Carnegie Mellon University, Imperial College London, and University of Cambridge for courses that parallel pedagogical uses seen at University of Oxford and Harvard University.

Implementation and performance

The core of Lean is implemented in a mix of native code and its own language layers, with performance engineering influenced by compiler research from groups at Google and Microsoft Research and runtime considerations discussed at conferences like POPL and PLDI. Its kernel implements a trusted type checker and normalization procedures comparable in role to kernels in Coq and Isabelle/HOL. Benchmarks and performance comparisons have been reported informally in workshops attended by researchers from ETH Zurich and University of Tokyo, showing trade-offs in elaboration and proof checking relative to systems such as Agda and Lean 3 predecessors. Ongoing optimization efforts involve contributors at Carnegie Mellon University and industrial partners including Microsoft Research.

Community and development ecosystem

The Lean ecosystem comprises contributors from academic institutions including Princeton University, Imperial College London, Carnegie Mellon University, University of Cambridge, and organizations such as Microsoft Research and the Simons Foundation. Community activities occur on platforms patterned after collaboration models used by projects at GitHub and mailing lists styled like those at Stack Overflow and forums associated with arXiv preprint discussions. Workshops, tutorials, and summer schools have been hosted at venues like Institute for Advanced Study, University of Oxford, and ETH Zurich, engaging participants familiar with formal methods communities at Cornell University and Stanford University. The project attracts contributions from mathematicians, computer scientists, and engineers connected to institutions including Harvard University, University of Chicago, and Princeton University.

Category:Proof assistants