Cross-Domain Reasoning Transfer via Reinforcement Learning Alignment of Large Language Models. ijaies 2026, 3 (2).