Tag
This paper introduces an autoresearch paradigm using LLM agents (coding and theory agents) to discover convex relaxations for sharp-constant inequalities, improving certified lower bounds on two optimization constants.