{"id":"arxiv/2602.10233","url":"https://arxiv.org/abs/2602.10233","tags":["cs.NE","cs.AI","math.CA","math.MG","math.OC"],"score":0,"title":"ImprovEvolve: Basin-Hopping Meets LLM-Guided Evolutionary Search","domain":"cs.NE","source":"arxiv","authors":["Alexey Kravatskiy","Valentin Khrulkov","Ivan Oseledets"],"reviews":0,"subject":"cs.NE","summary":"LLM-guided evolutionary computation, most notably AlphaEvolve, has been remarkably successful in discovering novel mathematical constructions by solving challenging optimization problems. The standard approach is to evolve a monolithic program that directly outputs a candidate solution. We present ImprovEvolve, an algorithmic alternative that drastically reduces cognitive load on the LLM. Instead of prompting the model for an end-to-end optimizer, we evolve a program with three specialized operators of initialization, local improvement, and perturbation. We then approach the optimum by…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-02-10","reactions":{}}