New arXiv paper introduces a self-improvement loop for reasoning models by training them jointly to predict, reverse-engineer, and apply solution ideas, using hindsight from supplied solutions. Applied to interactive…

#AI #MachineLearning #TheoremProving #Lean
https://arxiv.org/abs/2610.12168

NanoProof releases its training data, extraction tooling, pipeline, and weights for fully open automated theorem proving in Lean 4, reporting 50.8% pass@16 on MiniF2F-Test while using far less compute than…

#OpenSourceAI #TheoremProving #Lean4 #MachineLearning
https://arxiv.org/abs/2610.11605

Researchers introduce CoCo-Prover, which frames program verification as cost-efficient decision-making across interconnected proof obligations rather than chasing single-proof pass rates. The approach uses…

#TheoremProving #ProgramVerification #AIagents #DevTools
https://arxiv.org/abs/2610.09681

What Does Lean-Verified Actually Mean?

In OpenAI’s October 2026 mathematics release, 722 manuscripts are grouped into 372 families of related results. Many, but not all, come with Lean formalizations.

#lean #formalverification #mathematics #theoremproving

New benchmark TCSAlgBench evaluates LLMs on research-level theoretical computer science proofs, featuring 398 theorem-level challenges drawn from 138 STOC and COLT 2026 papers.

#TCSAlgBench #AIResearch #TheoremProving #LLMs
https://arxiv.org/abs/2609.35606

Intriguing 🤔:

“We’re Gonna Need A Lot More Mathematicians”, Amit Sahai via Terence Tao (terrytao.wordpress.com/2026/09/24/w...).

Via HN: news.ycombinator.com/item?id=4985...

#AI #LLMs #Mathematics #Math #Research #TheoremProving #Mathematicians #Vetting #ArtificialIntelligence

I recently graduated with a BSc in Computer Science and I’m interested in getting into formal methods and formal verification.
For those working or studying in the field, what books, courses, or other resources would you recommend to get started?
#FormalMethods #FormalVerification #TheoremProving