• Combines Koopman operator theory with linear RNNs to yield interpretable dynamics for long-horizon forecasting.
• Achieves state-of-the-art accuracy on traffic and energy datasets while improving stability over standard RNNs.