Bayesian Hidden Physics Models: Uncertainty Quantification for Discovery of Nonlinear Partial Differential Operators from Data

What do data tell us about physics-and what don't they tell us? There has been a surge of interest in using machine learning models to discover governing physical laws such as differential equations from data, but current methods lack uncertainty quantification to communicate their credibility. This work addresses this shortcoming from a Bayesian perspective. We introduce a novel model comprising "leaf" modules that learn to represent distinct experiments' spatiotemporal functional data as neural networks and a single "root" module that expresses a nonparametric distribution over their governing nonlinear differential operator as a Gaussian process. Automatic differentiation is used to compute the required partial derivatives from the leaf functions as inputs to the root. Our approach quantifies the reliability of the learned physics in terms of a posterior distribution over operators and propagates this uncertainty to solutions of novel initial-boundary value problem instances. Numerical experiments demonstrate the method on several nonlinear PDEs.

[1]  Carl E. Rasmussen,et al.  Bayesian Inference and Learning in Gaussian Process State-Space Models with Particle MCMC , 2013, NIPS.

[2]  A. O'Hagan,et al.  Bayesian inference for the uncertainty distribution of computer model outputs , 2002 .

[3]  S. Cox,et al.  Exponential Time Differencing for Stiff Systems , 2002 .

[4]  Carl E. Rasmussen,et al.  Closed-form Inference and Prediction in Gaussian Process State-Space Models , 2018, ArXiv.

[5]  Paris Perdikaris,et al.  Numerical Gaussian Processes for Time-Dependent and Nonlinear Partial Differential Equations , 2017, SIAM J. Sci. Comput..

[6]  Dominique Picard,et al.  Uncovering differential equations from data with hidden variables , 2020, ArXiv.

[7]  Geoffrey E. Hinton,et al.  On the importance of initialization and momentum in deep learning , 2013, ICML.

[8]  Steven L. Brunton,et al.  Data-driven discovery of partial differential equations , 2016, Science Advances.

[9]  George Em Karniadakis,et al.  NSFnets (Navier-Stokes flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations , 2020, J. Comput. Phys..

[10]  Paris Perdikaris,et al.  Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations , 2017, ArXiv.

[11]  James Hensman,et al.  Identification of Gaussian Process State Space Models , 2017, NIPS.

[12]  Yuan Yu,et al.  TensorFlow: A system for large-scale machine learning , 2016, OSDI.

[13]  David Duvenaud,et al.  Neural Ordinary Differential Equations , 2018, NeurIPS.

[14]  James O. Ramsay,et al.  Principal differential analysis : Data reduction by differential operators , 1996 .

[15]  Yoram Singer,et al.  Adaptive Subgradient Methods for Online Learning and Stochastic Optimization , 2011, J. Mach. Learn. Res..

[16]  Carl E. Rasmussen,et al.  Variational Gaussian Process State-Space Models , 2014, NIPS.

[17]  F. Takens Detecting strange attractors in turbulence , 1981 .

[18]  George Em Karniadakis,et al.  Adaptive activation functions accelerate convergence in deep and physics-informed neural networks , 2019, J. Comput. Phys..

[19]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[20]  Ilias Bilionis,et al.  Multi-output separable Gaussian process: Towards an efficient, fully Bayesian paradigm for uncertainty quantification , 2013, J. Comput. Phys..

[21]  Paris Perdikaris,et al.  Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations , 2019, J. Comput. Phys..

[22]  Carl E. Rasmussen,et al.  Non-Factorised Variational Inference in Dynamical Systems , 2018, ArXiv.

[23]  Natalia Gimelshein,et al.  PyTorch: An Imperative Style, High-Performance Deep Learning Library , 2019, NeurIPS.

[24]  Shirley Ho,et al.  Learning Symbolic Physics with Graph Networks , 2019, ArXiv.

[25]  Paris Perdikaris,et al.  Understanding and mitigating gradient pathologies in physics-informed neural networks , 2020, ArXiv.

[26]  Liping Wang,et al.  Data-driven discovery of free-form governing differential equations , 2019, ArXiv.

[27]  David J. Fleet,et al.  Gaussian Process Dynamical Models , 2005, NIPS.

[28]  Maziar Raissi,et al.  Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations , 2018, J. Mach. Learn. Res..

[29]  Neil D. Lawrence,et al.  Gaussian Processes for Big Data , 2013, UAI.

[30]  Hod Lipson,et al.  Automated reverse engineering of nonlinear dynamical systems , 2007, Proceedings of the National Academy of Sciences.

[31]  Liu Yang,et al.  B-PINNs: Bayesian Physics-Informed Neural Networks for Forward and Inverse PDE Problems with Noisy Data , 2020, J. Comput. Phys..

[32]  Paris Perdikaris,et al.  Physics Informed Deep Learning (Part II): Data-driven Discovery of Nonlinear Partial Differential Equations , 2017, ArXiv.

[33]  Michalis K. Titsias,et al.  Variational Learning of Inducing Variables in Sparse Gaussian Processes , 2009, AISTATS.

[34]  George Em Karniadakis,et al.  Physics-Informed Neural Network for Ultrasound Nondestructive Quantification of Surface Breaking Cracks , 2020, Journal of Nondestructive Evaluation.

[35]  Matthew D. Zeiler ADADELTA: An Adaptive Learning Rate Method , 2012, ArXiv.

[36]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.