{"id":"arxiv/2608.07407","url":"https://arxiv.org/abs/2608.07407","tags":["math.OC"],"score":0,"title":"A Domain-Specific Harness for End-to-End Automation of Optimization Research","domain":"math.OC","source":"arxiv","authors":["Heechang Kim","Ernest K. Ryu","Shuvomoy Das Gupta"],"reviews":0,"subject":"math.OC","summary":"We present AutoOPT, a domain-specific harness for end-to-end automation of optimization research. AutoOPT organizes the discovery of optimal first-order methods into four stages: numerical design through the BnB-PEP methodology; symbolic discovery of the analytic description and a convergence proof through frontier large language models (LLMs); formal verification in the Lean 4 proof assistant; and human interpretation and write-up. We demonstrate the framework on two case studies, each of independent interest. The first, lemniscate acceleration, is a new accelerated gradient method for…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-08-07","reactions":{}}