Conversation
fmontesi
requested review from
SamuelSchlesinger,
arademaker,
chenson2018,
kim-em and
sorrachai
as code owners
September 23, 2026 17:07
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds a cut elimination procedure for CLL. The definition requires a list representation of the sequent (since that's a multiset) for computability reasons. The PR also adds a multiset grind set for multiset rewritings, which has been very useful to formulate a series of utility lemmas for improving the efficiency of several pieces of the cut elimination/admissibility proofs.
AI disclosure and discussion:
This PR is also an AI-assisted formalisation milestone. Cut elimination is a very tedious proofs with lots of cases, rewritings, and nontrivial termination arguments. I've used cut elimination as a test for AI agents for a while, and until recently they failed badly. The same AI agents now managed to formulate appropriate induction principles and carry out the proof. I've reviewed the most important pieces and refactored/polished a lot of the code.
It seems that one of the biggest difference makers has been that the repository has grown enough to give useful context. In particular all the developments about logic, the various READMEs, and the introduction of the
modalgrind set seem to have been important to 'inspire' the AI on how to work. I suspect CSLib is starting to make a difference in AI performance.Depends on #941 and #938.