Lean
PulseAugur coverage of Lean — every cluster mentioning Lean across labs, papers, and developer communities, ranked by signal.
- used by Gotit.pub 90%
- used by Terence Tao 90%
- used by Kevin Buzzard 90%
- authored by Millennium Problems 90%
- uses DagsHub 90%
- developed by Millennium Problems 90%
- used by Lean 4 Programming Language 90%
- used by CatalyzeX 90%
- developed by Stacks Project 90%
- used by rust-lang 90%
- used by Stacks Project 90%
- used by Axiom 90%
25 day(s) with sentiment data
-
AI agents refactor Lean proofs under $3 cap, using tiered models
A research paper details a novel approach to refactoring Lean proofs using enterprise agent practices, achieving a score of 86.31 on a closed-track competition. The method employed a cost-effective strategy, utilizing d…
-
Anthropic's Fable 5.2 Aids Quantum Computing Breakthrough
PSI, a research systems startup, has utilized an early version of Anthropic's Fable 5.2 model to solve the 18-year-old Aaronson–Ambainis conjecture in quantum computing. The solution, converted into a verifiable format,…
-
AI solves 18-year-old quantum computing problem, signaling shift in scientific discovery
Physical Superintelligence (PSI) is leveraging AI to tackle complex scientific problems, with a recent breakthrough in quantum computing. A PSI researcher used Anthropic's Fable 5.2 model to solve the Aaronson-Ambainis …
-
Google DeepMind uses AI agents to formally verify Signal's cryptographic protocol
Google DeepMind has introduced AlphaProof Code, an AI-powered framework that formally verifies complex software. Researchers used this system to mathematically prove key security properties of Signal's cryptographic pro…
-
Anthropic releases Claude Fable 5.2 with 1M context, price cut
Anthropic has released Claude Fable 5.2 and Claude Mythos 5.2, with Fable 5.2 now available to all developers via API and cloud platforms. Both models feature a 1M-token context window and improved speed, but Mythos 5.2…
-
1,000 Anthropic Claude Agents Prove Fermat's Last Theorem Autonomously
Anthropic's Claude Agents, an early version of Claude Fable 5.2, successfully proved Fermat's Last Theorem by developing their own organizational structure. A team of 1,000 agents completed the complex mathematical proo…
-
New research challenges transformer limits with novel layer designs
A new research paper explores the limitations of standard transformer models in learning complex graph structures and games like SET. The authors propose a novel approach using higher-order layers and specific channel d…
-
AI Labs Challenge Theory, Expose RSA Flaws, and Spark Microsoft Reorg · 1 source tracked
Leading AI labs are making significant theoretical and security advancements. OpenAI and Anthropic, using AI, have published Lean-verified proofs that challenge a 1998 complexity theorem, suggesting faster solutions for…
-
AI proves "enormous theorem" with 43.1M lines of code; new GDPx benchmark launched
An AI has been used to prove "the enormous theorem," also known as the Classification of Finite Simple Groups. This proof was formalized using 43.1 million lines of code in Lean, a proof assistant designed for rigorous …
-
Quantum Conjecture Proven, Google Buys Wyoming Data Center, Microsoft Exits Oracle Deal
A research system developed by PSI has successfully proven the Aaronson-Ambainis Conjecture, a 2008 quantum computing question, in just three days. This proof, which utilized non-public Anthropic model instances and was…
-
Anthropic AI model proves 18-year-old quantum conjecture, verified in Lean · 4 sources tracked
An unreleased Anthropic AI model has reportedly proven the Aaronson-Ambainis conjecture, a quantum computing problem that has been open since 2008. Physical Superintelligence (PSI) stated that their research system util…
-
PSI finds quantum computing proof using private Anthropic model
PSI has announced a breakthrough in quantum computing complexity, finding a proof for a conjecture posed in 2008. The research system, which utilized a private Anthropic model, generated its first proof within three day…
-
AI discovers and rigorously certifies new water wave solutions
Researchers have developed a novel computational framework that merges machine learning with rigorous computer-assisted numerical certification and formal verification. This approach uses physics-informed neural network…
-
AI agents formalize mathematical proofs in Lean library
Researchers have developed AutoIP, a library within the Lean proof assistant that uses AI agents to formalize classical linear regularization theory. This library aims to make mathematical proofs more accessible by allo…
-
New LeanContext benchmark disentangles math resources for AI theorem proving
Researchers have developed LeanContext, a new benchmark comprising over 129,000 Lean whole-proof generation problems. This benchmark aims to disentangle the impact of mathematical resources (Mathematical Context) from l…
-
Axiom formalizes 7,500-page Stacks Project in Lean, creating largest math library
Axiom has released a formalization of the entire Stacks Project, a foundational reference in algebraic geometry, into the Lean proof assistant. This extensive library, comprising 16 million lines of code, represents the…
-
Axiom Math uses AI agents to formally verify vast math library
Axiom Math, a startup valued at $1.6 billion, is developing AI agents to formally verify mathematical proofs and prevent the influx of unverified AI-generated content into the field. The company has successfully verifie…
-
AI advancements drive formal verification across software and hardware
Several companies are advancing the field of formal verification, applying it to both AI systems and the software/hardware they run on. Harmonic is using its Aristotle agent for scalable verification, while SF-Tensor is…
-
Whiel platform enables certified relational program verification
Researchers have developed Whiel, a novel neurosymbolic platform designed for certified relational program verification. This system integrates with Lean and can encode complex problems related to Datalog and graph quer…
-
OpenAI launches Images 2.6, Ripple-65B sets new coding benchmark · 1 source tracked
OpenAI has released Images 2.6, a new model capable of generating GIFs and images 50% faster than its predecessor, with API access for advanced versions. Separately, Ripple-65B Preview, an open-weight model, has demonst…