AlphaRLM: Aligning Language Models for Literature-Grounded Symbolic Factor Discovery | ArxivCSExplorer