ml:automatic_theorem_proving
Table of Contents
Automatic Theorem Proving
Overviews
- Formalizing mathematics
- Autoformalization
Papers
Neural Theorem Proving
- Overviews
- Bibliographies
- Step-by-step Proof Generation
- GPT-f: Polu & Sutskever 2020 - Generative Language Modeling for Automated Theorem Proving Maintains a proof tree with a queue of open goals, and tries to expand the goal with the highest LLM logprob to reach it. Metamath formalization language. Introduces WebMath pre-training.
- Han et al 2021 - Proof Artifact Co-training for Theorem Proving with Language Models Generates the next tactic using an LLM. Uses a best-first search algorithm: maintains a priority queue of search nodes, of tactic state (partial proof) and metadata, sorted by LLM heuristic score. For Lean. Introduces the LeanStep dataset to to predict the next tactic give a goal state, built from mathlib. Uses an interesting PACT pretraining dataset (sec 3.2).
- Jiang et al 2022 - Thor: Wielding Hammers to Integrate Language Models and Automated Theorem Provers Uses automated theorem prover premise selection (hammers) to retrieve the relevant premises, and then uses LLMs as in GPT-f (Polu & Sutskever 2020). Increases success rate on PISA from 39.0% to 57.0%.
- Construction All-In-One Go
Formal Automated Theorem Proving
Systems
- Vampire
- Kovacs & Voronkov 2013 - First-Order Theorem Proving and VAMPIRE Intro to VAMPIRE
- For Mizar dataset
- Kaliszyk et al 2013 - MizAR 40 for Mizar 40 Uses Vampire, Epar, and Z3 solvers.
- For Lean
- HOL Light
Datasets
- Mizar
- MPTP
- MizarTPTP
- Dataset: website
Formalizing Mathematics
- Overviews
- Mizar
- Byliński et al 2021 - Syntactic-Semantic Form of Mizar Articles Gives an overview and a good historical introduction in the intro.
Autoformalization
- Overviews
- Avigad et al 2023 - Mathematics and the Formal Turn Great overview of the reasons to autoformalize mathematics
Informal Math to Formal Math Parsing
- For Mizar dataset
- Wang et al 2018 - First Experiments with Neural Translation of Informal to Formal Mathematics Automatic translation of informal mathematical statements in latex to formal Mizar statements. Uses a synthetically constructed dataset.
- Wang et al 2020 - Exploration of Neural Machine Translation in Autoformalization of Mathematics in Mizar Automatic translation of informal mathematical statements in latex to formal Mizar statements. Uses a synthetically constructed dataset.
- For Isabelle
- Li et al 2021 - IsarStep: a Benchmark for High-level Mathematical Reasoning github Fill in a missing intermediate proposition given surrounding proofs.
Automatic Premise Selection
Software
- Mizar proof checker Open-source reimplementation of the Mizar proof checker
- Tutorial paper: Kovacs & Voronkov 2013 - First-Order Theorem Proving and VAMPIRE
- CMU slides: Vampire
- Lecture: slides
- Mizar (miscellaneous software)
- tptp4mizar: Generating Mizar texts from TPTP problems and solutions
- Generate Latex files from Mizar articles:
Courses and Tutorials
- Automated Theorem Proving (in general)
- Mizar
- Tutorial Good tutorial, use along with Mizar in a Nutshell and Mizar: An Impression (see also recommended reading at the end of the tutorial)
- Lean
- Natural Number Game Intro to Lean (great). Youtube video
- Learning Lean
- Learning Lean 4 - List of places to learn
- A glimpse of Lean - Fast paced, for mathematicians
- Theorem Proving in Lean 4 Lean 3 version - Jeff can't stand this book
Workshops and Conferences
People
- Christian Szegedy (at Google)
- Yuhuai Wu (at Google)
Related Pages
ml/automatic_theorem_proving.txt · Last modified: 2025/10/07 14:31 by jmflanig