English

Twitch: Learning Abstractions for Equational Theorem Proving

Logic in Computer Science 2026-03-10 v1 Artificial Intelligence

Abstract

Several successful strategies in automated reasoning rely on human-supplied guidance about which term or clause shapes are interesting. In this paper we aim to discover interesting term shapes automatically. Specifically, we discover abstractions : term patterns that occur over and over again in relevant proofs. We present our tool Twitch which discovers abstractions with the help of Stitch, a tool originally developed for discovering reusable library functions in program synthesis tasks. Twitch can produce abstractions in two ways: (1) from a partial, failed proof of a conjecture; (2) from successful proofs of other theorems in the same domain. We have also extended Twee, an equational theorem prover, to use these abstractions. We evaluate Twitch on a set of unit equality (UEQ) problems from TPTP, and show that it can prove 12 rating-1 problems as well as yielding significant speed-ups on many other problems.

Keywords

Cite

@article{arxiv.2603.06849,
  title  = {Twitch: Learning Abstractions for Equational Theorem Proving},
  author = {Guy Axelrod and Moa Johansson and Nicholas Smallbone},
  journal= {arXiv preprint arXiv:2603.06849},
  year   = {2026}
}

Comments

20 pages, submitted to IJCAR 2026

R2 v1 2026-07-01T11:07:56.878Z