Learning Koopman Representations for Stochastic Dynamics with Spectral Residual Constraints

Authors

  • Martina Romano Faculty of Engineering and Information Technology, University of Melbourne, Melbourne, Australia
  • Andrew Chuck Faculty of Engineering and Information Technology, University of Melbourne, Melbourne, Australia

Keywords:

Koopman Operator, Stochastic Dynamics, Deep Learning, Spectral Residual, Time Series Prediction

Abstract

The analysis and prediction of nonlinear dynamical systems are fundamental challenges across numerous scientific disciplines. The Koopman operator theory provides a powerful framework by transforming finite-dimensional nonlinear dynamics into an infinite-dimensional linear paradigm. However, extending this theory to stochastic systems introduces significant complexities, primarily due to the continuous spectrum and the difficulty of identifying a finite set of invariant observables. In this paper, we propose a novel deep learning architecture that learns finite-dimensional Koopman representations for stochastic dynamical systems by incorporating spectral residual constraints. Unlike traditional data-driven methods that rely solely on reconstruction and prediction errors, our approach enforces the spectral properties of the Koopman operator directly during the training phase. By minimizing the spectral residual, the neural network is guided to discover observable functions that closely approximate the true eigenfunctions of the stochastic system. We comprehensively evaluate the proposed framework on canonical stochastic systems and high-dimensional fluid flow simulations. The results demonstrate that the inclusion of spectral residual constraints significantly improves long-term prediction accuracy, enhances the physical interpretability of the learned representations, and provides robust eigenvalue estimation in the presence of noise. This research bridges the gap between deep learning and rigorous operator theory, offering a scalable solution for the data-driven analysis of complex stochastic processes.

References

1. Hao, Y., Xu, J., & Liao, S. (2026). RAPS-Net: A Risk-Aware CNN-LSTM Framework for Cross-Domain Risk Prediction and Dynamic Security Control in Cloud Payment Supply Chains. Economics and Management Innovation, 3(2), 18-27.

2. Xing, J., Liu, Y., Zhang, Z., Chen, L., Hasan, M. N., & Zhang, Z. B. (2025). An Intrinsically Knowledge-Transferring Developmental Spiking Neural Network for Tactile Classification. IEEE Transactions on Neural Networks and Learning Systems.

3. Xue, Xiao, et al. "Equivariant U-Shaped Neural Operators for the Cahn-Hilliard Phase-Field Model." arXiv preprint arXiv:2509.01293 (2025).

4. Xue, X., Wang, S., Yao, H. D., Davidson, L., & Coveney, P. V. (2024). Physics informed data-driven near-wall modelling for lattice Boltzmann simulation of high Reynolds number turbulent flows. Communications Physics, 7(1), 338.

5. Li, W. W., & Ma, T. (2025). Learn to Rank Risky Investors: A Case Study of Predicting Retail Traders’ Behaviour and Profitability. ACM Transactions on Information Systems, 44(1), 1-33.

6. Li, W. W., Kim, H., Cucuringu, M., & Ma, T. (2026, August). Can LLM-based financial investing strategies outperform the market in long run?. In Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 1 (pp. 2711-2722).

7. Kim, E. H., Wang, Z., Zong, H., Jiang, Z., Fu, Z., & Pedrycz, W. (2023). Design of tobacco leaves classifier through fuzzy clustering-based neural networks with multiple histogram analyses of images. IEEE Transactions on Industrial Informatics, 20(3), 4698-4709.

8. Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society: Series B, 58(1), 267–288.

9. Li, Z., Hao, Y., & Zeng, P. (2026). CLASNet: A Cognitive Load–Aware CNN-LSTM-Attention Framework for Supply Chain Demand Forecasting and Adaptive Human–Computer Interaction. Journal of Computer Science and Frontier Technologies, 3(2), 44-57.

10. Dong, J., Wang, Y., Lai, J. H., & Xie, X. (2022). Improving adversarially robust few-shot image classification with generalizable representations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 9025-9034).

11. Lu, Yao, Kaiyi Hu, and Luyan Zhang. "S 3 G: Stock State Space Graph for Enhanced Stock Trend Prediction." ICASSP 2026-2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2026.

12. Wu, Yuxuan, et al. "Unsupervised Disentanglement of Content and Style via Variance-Invariance Constraints." arXiv preprint arXiv:2407.03824 (2024).

13. Xue, Xiao, et al. "Fast-Forward Lattice Boltzmann: Learning Kinetic Behaviour with Physics-Informed Neural Operators." arXiv preprint arXiv:2509.22411 (2025).

14. Guo, H., Lu, C., Bao, F., Pang, T., Yan, S., Du, C., & Li, C. (2023). Gaussian mixture solvers for diffusion models. Advances in Neural Information Processing Systems, 36, 25598-25626.

15. Huang, J. (2025, September). GRAPHTRUST: GRAPH NEURAL NETWORKS FOR RELIABLE SUPPLY CHAIN RISK REASONING. In The 5th International scientific and practical conference “Trends in the development of science by young scientists and students”(September 30-October 03, 2025) Warsaw, Poland. International Science Group. 2025. 122 p. (p. 18).

16. Qu, D., & Ma, Y. (2025). Magnet-bn: markov-guided Bayesian neural networks for calibrated long-horizon sequence forecasting and community tracking. Mathematics, 13(17), 2740.

17. Xing, Jiarui, Song Wang, and Jian Wang. "Divergence is Uncertainty: A Closed-Form Posterior Covariance for Flow Matching." arXiv preprint arXiv:2605.00941 (2026).

18. Yao, Zhengtao, et al. "Task-agnostic federated learning." arXiv preprint arXiv:2406.17235 (2024).

19. Dong, J., Moayedi, R. Z., Ong, Y. S., & Moosavi-Dezfooli, S. M. (2026). Allies teach better than enemies: Inverse adversaries for robust knowledge distillation. IEEE Transactions on Pattern Analysis and Machine Intelligence.

20. Qu, D., & Ma, Y. (2026). F2-CommNet: Fourier–Fractional neural networks with Lyapunov stability guarantees for hallucination-resistant community detection. Frontiers in Computational Neuroscience, 19, 1731452. (https://doi.org/10.3389/fncom.2025.1731452)

21. Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017). Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347.

22. Dong, J., Yang, L., Wang, Y., Xie, X., & Lai, J. (2023). Toward intrinsic adversarial robustness through probabilistic training. IEEE Transactions on Image Processing, 32, 3862-3872.

23. Qu, D., Ma, Y., & Zhang, S. (2025, November). OAMF: Optics-Accelerated Multimodal Learning with Markov Temporal Priors and Fourier Regularization. In 2025 4th International Conference on Image Processing, Computer Vision and Machine Learning (ICICML) (pp. 600-605). IEEE.

24. Lv, Qi, et al. "Spatial-temporal graph diffusion policy with kinematic modeling for bimanual robotic manipulation." Proceedings of the Computer Vision and Pattern Recognition Conference. 2025.

25. Zhao, S., Qian, K., & Wen, C. (2022). Research on Used Car Transaction Cycle Based on Soft Voting. Information Systems and Economics, 3(1), 10-15.

26. Wang, Z., Kim, E. H., Oh, S. K., Pedrycz, W., Fu, Z., & Yoon, J. H. (2024). Reinforced fuzzy-rule-based neural networks realized through streamlined feature selection strategy and fuzzy clustering with distance variation. IEEE Transactions on Fuzzy Systems, 32(10), 5674-5686.

27. Li, W. W., Ziser, Y., Xie, Y., Cohen, S. B., & Ma, T. (2025, July). Tsprank: Bridging pairwise and listwise methods with a bilinear travelling salesman model. In Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 1 (pp. 707-718).

28. Hu, L., & Shen, Y. (2026). A predictive analytics approach for forecasting global stock index returns using deep learning techniques. Decision Analytics Journal, 100685.

29. Chen, L. (2026). Beyond external constraints: The missing dimension of ai governance. Available at SSRN 6449738.

30. Che