Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
158 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

PINF: Continuous Normalizing Flows for Physics-Constrained Deep Learning (2309.15139v1)

Published 26 Sep 2023 in cs.LG and cs.AI

Abstract: The normalization constraint on probability density poses a significant challenge for solving the Fokker-Planck equation. Normalizing Flow, an invertible generative model leverages the change of variables formula to ensure probability density conservation and enable the learning of complex data distributions. In this paper, we introduce Physics-Informed Normalizing Flows (PINF), a novel extension of continuous normalizing flows, incorporating diffusion through the method of characteristics. Our method, which is mesh-free and causality-free, can efficiently solve high dimensional time-dependent and steady-state Fokker-Planck equations.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (35)
  1. Flow-based generative models for markov chain monte carlo in lattice field theory. Physical Review D, 100(3):034515, 2019.
  2. Neural ordinary differential equations. Advances in neural information processing systems, 31, 2018.
  3. Weihua Deng. Finite element method for the space and time fractional fokker–planck equation. SIAM journal on numerical analysis, 47(1):204–226, 2009.
  4. Nice: non-linear independent components estimation. In International Conference on Learning Representations Workshop, 2015.
  5. Density estimation using real nvp. In International Conference on Learning Representations, 2016.
  6. Solving time dependent fokker-planck equations via temporal normalizing flow. Communications in Computational Physics, 32(2):401–423, 2022.
  7. How to train your neural ode: the world of jacobian and kinetic regularization. In International conference on machine learning, pp. 3154–3164. PMLR, 2020.
  8. Deep learning. MIT press, 2016.
  9. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp.  770–778, 2016.
  10. Flow++: Improving flow-based generative models with variational dequantization and architecture design. In International Conference on Machine Learning, pp. 2722–2730. PMLR, 2019.
  11. Algorithms of data generation for deep learning and feedback design: A survey. Physica D: Nonlinear Phenomena, 425:132955, 2021.
  12. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014.
  13. Glow: Generative flow with invertible 1x1 convolutions. Advances in neural information processing systems, 31, 2018.
  14. Pankaj Kumar and S Narayanan. Solution of fokker-planck equation by finite element and finite difference methods for nonlinear systems. Sadhana, 31:445–461, 2006.
  15. Artificial neural networks for solving ordinary and partial differential equations. IEEE transactions on neural networks, 9(5):987–1000, 1998.
  16. A neural network approach for stochastic optimal control. arXiv preprint arXiv:2209.13104, 2022.
  17. A causality-free neural network method for high-dimensional hamilton-jacobi-bellman equations. In 2020 American Control Conference (ACC), pp.  787–793. IEEE, 2020.
  18. Bernt Oksendal. Stochastic differential equations: an introduction with applications. Springer Science & Business Media, 2013.
  19. Ot-flow: Fast and accurate continuous normalizing flows via optimal transport. Proceedings of the AAAI Conference on Artificial Intelligence, 35(10):9223–9232, 2021a.
  20. A neural network approach applied to multi-agent optimal control. In 2021 European Control Conference (ECC), pp.  1036–1041. IEEE, 2021b.
  21. A neural network approach for high-dimensional optimal control applied to multi-agent path finding. IEEE Transactions on Control Systems Technology, 31(1):235–251, 2022.
  22. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378:686–707, 2019.
  23. Variational inference with normalizing flows. In International conference on machine learning, pp. 1530–1538. PMLR, 2015.
  24. Hannes Risken. Fokker-planck equation. Springer, 1996.
  25. Carl Runge. Über die numerische auflösung von differentialgleichungen. Mathematische Annalen, 46(2):167–178, 1895.
  26. A machine learning framework for solving high-dimensional mean field game and mean field control problems. Proceedings of the National Academy of Sciences, 117(17):9183–9193, 2020.
  27. Dgm: A deep learning algorithm for solving partial differential equations. Journal of computational physics, 375:1339–1364, 2018.
  28. Adaptive deep density approximation for fokker-planck equations. Journal of Computational Physics, 457:111080, 2022.
  29. Cédric Villani. Topics in Optimal Transportation. Number 58. American Mathematical Soc., 2003.
  30. Physics-guided deep learning for dynamical systems: A survey. arXiv preprint arXiv:2107.01272, 2021.
  31. Solving ordinary differential equations II, volume 375. Springer Berlin Heidelberg New York, 1996.
  32. M Fo Wehner and WG Wolfer. Numerical evaluation of path-integral solutions to fokker-planck equations. Physical Review A, 27(5):2663, 1983.
  33. E Weinan and Bing Yu. The deep ritz method: A deep learning-based numerical algorithm for solving variational problems. Communications in Mathematics and Statistics, 6(1):1–12, 2018.
  34. Potential flow generator with l2subscript𝑙2l_{2}italic_l start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT optimal transport regularity for generative models. IEEE Transactions on Neural Networks and Learning Systems, 33(2):528–538, 2020.
  35. Monge-ampère flow for generative modeling. arXiv e-prints, pp.  arXiv–1809, 2018.
Citations (1)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now