Talk:Lean (proof assistant)
Add topic| This is the talk page for discussing improvements to the Lean (proof assistant) article. This is not a forum for general discussion of the subject of the article. |
Article policies
|
| Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL |
| This article is rated C-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | |||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||
Remove paragraph on features that differentiate lean?
[edit]As far as I can tell, none of these features are unique to lean:
- Compilation to javascript: Coq and F* both have this
- Native support for unicode symbols: Coq has this
- Typing symbols with LaTeX notations: this is an Emacs feature
- Own language for meta-programming: F* has this too
I think this paragraph could just be removed? 2603:400A:0:82C:FFF0:9101:2D27:7223 (talk) 00:52, 14 December 2020 (UTC)
Examples in Lean 4
[edit]As of October 2023, the examples section is in Lean 3, which the current version is not backwards-compatible with. We can borrow examples from the community's Mathematics in Lean book (which uses Lean 4 and Mathlib), to make this section of the article reflect the language's current syntax and features.
I think this is what each example should generally look like:
- The full text of the proof, in Latex/markdown. Should be readable to anyone familiar with math.
- The proof translated into Lean 4 code, with an explanation of what each "tactic" is doing, in comments. Tactics are basically functions built into Mathlib or core Lean that allow the language to understand steps in the proof. The most common one is "rw".
- The output of the code. This could also be a comment.
Lean is not a theorem prover. It does not "fill in the gaps" for you or generate a proof on its own. Rather, it is a proof assistant, meaning that you can take a draft of a proof, code it, and see that it is true. This the key idea that we want readers to take away from this section.
There should also be an explanation of what Tactics are, either in the Example or Overview sections.
Add A Fact: "Lean is a research project, not a product"
[edit]I found a fact that might belong in this article. See the quote below
It is a research project and not a product. Things change rapidly, and we constantly break backward compatibility.
The fact comes from the following source:
Here is a wikitext snippet to use as a reference:
{{Cite web |title=Frequently Asked Questions - Lean Manual |url=https://docs.lean-lang.org/lean4/doc/faq.html#should-i-use-lean |website=docs.lean-lang.org |access-date=2024-10-08 |quote=It is a research project and not a product. Things change rapidly, and we constantly break backward compatibility.}}
Additional comments from user: It's not only that Lean 4 is not backwards compatible to Lean 3; it's that the general ethos is that backward compatibility is constantly broken.
This post was generated using the Add A Fact browser extension.
Latest Release version and date are consistently off
[edit]There is no longer a link to the github latest release version in the infobox, just a chain of numbers and letters, like "". I believe this change was intended to automatically get the latest release, but it's not doing that. The latest release was 4.17, but the infobox currently says 4.16. RedrickSchu (talk) 15:48, 21 March 2025 (UTC)
Wiki Education assignment: Proofs and Problem Solving
[edit]
This article was the subject of a Wiki Education Foundation–supported course assignment, between 13 January 2026 and 17 April 2026. Further details are available on the course page. Student editor(s): Jevanniln02 (article contribs).
— Assignment last updated by Jevanniln02 (talk) 03:55, 1 February 2026 (UTC)
Rename from Lean (proof assistant) to Lean (programming language)?
[edit]Arguments in favor:
- The Lean website is lean-lang.org, not lean-prover.org and the title of that page is "Lean Programming Language".
- Technically any statically typed programming language can be used as a proof assistant, but not all proof assistants can be used for general-purpose programming, for instance Metamath.
- Several notable projects have used Lean as a programming language, such as Lean itself, Verso, SciLean, etc. There's also an official book Functional Programming in Lean about using Lean for programming.
- There's already Agda (programming language) and Idris (programming language).
Arguments against:
- The Lean GitHub org is "leanprover" and the Lean Zulip is leanprover.zulipchat.com, not "lean-lang".
- The vast majority of people reading the page are probably interested in Lean as a proof assistant, not as a programming language.
Jon Daker (talk) 23:35, 17 May 2026 (UTC)
- So I tried doing this move, but it didn't work because Talk:Lean (programming language) already exists. Oh well... Jon Daker (talk) 00:04, 1 July 2026 (UTC)
- Leonardo de Moura himself calls Lean a programming language. Also IIRC Terence Tao remarked that he inspects Lean programs before running them because a Lean program can do anything, including e.g. deleting local files. That would put it squarely in the category of programming languages if judged strictly by capabilities. ~2026-43329-95 (talk) 22:58, 13 August 2026 (UTC)
Recent fake proof of Collatz, verified by Lean
[edit]I added a small section about the recent disproof of the Collatz conjecture, as verified by Lean. Unsurprisingly, the proof relied on a bug in Lean, since fixed. This seems (to me anyway) more than notable enough, but most of the references I could find are to personal web sites. Better references would be appropriate, if anyone knows of any. LouScheffer (talk) 00:24, 5 August 2026 (UTC)
- From what I've heard the Collatz disproof was intended to be a troll way of announcing the soundness bug, not an actual claim of a proof of Collatz. Jon Daker (talk) 23:19, 13 August 2026 (UTC)
- I've seen this speculation as well. That's why I said by "perhaps related", to cover this possibility. Feel free to make this connection more directly stated, if you can find a reference. LouScheffer (talk) 23:53, 13 August 2026 (UTC)
Is the DeepSeek prover stale or dead?
[edit]One commit a year ago; nothing since.
https://github.com/deepseek-ai/DeepSeek-Prover-V2 J.B. Shirk (talk) 20:19, 11 October 2026 (UTC)
Nested inductive types
[edit]I gather, for example at https://www.lesswrong.com/posts/zHhGa3iQBQcdSifFL/a-critical-look-at-lean4-1, that use of nested inductive types is a possibly problematic area in Lean. Could knowledgeable people perhaps add an explanation here and the article (or redirect) Nested inductive type? PJTraill (talk) 12:28, 2 November 2026 (UTC)
- I think that could be added to the inductive type article. There's some more info about this topic in the Lean docs too: https://lean-lang.org/doc/reference/latest/The-Type-System/Inductive-Types/#nested-inductive-types Jon Daker (talk) 19:24, 22 November 2026 (UTC)
Acid Test
[edit]Is Lean the acid test, or the initial acid test for proofs? First hurdle, or what? I'm thinking here of its October 2026 formalized proof of the Reimann hypothesis, and I don't see any overview of Lean's impact on mathematics from its get-go. When did it become a standard feature, and so on. kencf0618 (talk) 17:43, 21 November 2026 (UTC)
- C-Class mathematics articles
- Low-priority mathematics articles
- C-Class Computing articles
- Mid-importance Computing articles
- C-Class software articles
- Low-importance software articles
- C-Class software articles of Low-importance
- All Software articles
- C-Class Computer science articles
- Low-importance Computer science articles
- C-Class Free and open-source software articles
- Low-importance Free and open-source software articles
- C-Class Free and open-source software articles of Low-importance
- All Free and open-source software articles
- All Computing articles


