ReasFlow: Assisting Reasoning-Centric Scientific Discovery in Applied Mathematics via a Knowledge-Based Multi-Agent System

arXiv cs.AI · 6 hours ago

ReasFlow is an autonomous AI agent system designed to assist with reasoning-centric scientific discovery in applied mathematics by combining large language models with verification loops and knowledge retrieval mechanisms. The system generated five complete research papers with theoretical and empirical content from minimal prompts and achieved the highest evaluation scores compared to open-access baselines. This enables a collaborative model where human experts guide research direction while AI agents handle rigorous derivations, theorem proving, and manuscript preparation.