Cross-Domain Reasoning Transfer via Reinforcement Learning Alignment of Large Language Models. ijaies 2026, 1 (2).