Adaptive learning of effective dynamics for online modeling of complex systems
From MaRDI portal
Publication:6096436
DOI10.1016/J.CMA.2023.116204arXiv2304.01732MaRDI QIDQ6096436FDOQ6096436
Authors: Ivica Kičić, Pantelis R. Vlachas, Georgios Arampatzis, Michail Chatzimanolakis, Leonidas Guibas, Petros Koumoutsakos 
Publication date: 12 September 2023
Published in: Computer Methods in Applied Mechanics and Engineering (Search for Journal in Brave)
Abstract: Predictive simulations are essential for applications ranging from weather forecasting to material design. The veracity of these simulations hinges on their capacity to capture the effective system dynamics. Massively parallel simulations predict the systems dynamics by resolving all spatiotemporal scales, often at a cost that prevents experimentation. On the other hand, reduced order models are fast but often limited by the linearization of the system dynamics and the adopted heuristic closures. We propose a novel systematic framework that bridges large scale simulations and reduced order models to extract and forecast adaptively the effective dynamics (AdaLED) of multiscale systems. AdaLED employs an autoencoder to identify reduced-order representations of the system dynamics and an ensemble of probabilistic recurrent neural networks (RNNs) as the latent time-stepper. The framework alternates between the computational solver and the surrogate, accelerating learned dynamics while leaving yet-to-be-learned dynamics regimes to the original solver. AdaLED continuously adapts the surrogate to the new dynamics through online training. The transitions between the surrogate and the computational solver are determined by monitoring the prediction accuracy and uncertainty of the surrogate. The effectiveness of AdaLED is demonstrated on three different systems - a Van der Pol oscillator, a 2D reaction-diffusion equation, and a 2D Navier-Stokes flow past a cylinder for varying Reynolds numbers (400 up to 1200), showcasing its ability to learn effective dynamics online, detect unseen dynamics regimes, and provide net speed-ups. To the best of our knowledge, AdaLED is the first framework that couples a surrogate model with a computational solver to achieve online adaptive learning of effective dynamics. It constitutes a potent tool for applications requiring many expensive simulations.
Full work available at URL: https://arxiv.org/abs/2304.01732
Recommendations
machine learningNavier-Stokes equationscomputer simulationscontinuous learningadaptive reduced-order modelingonline real-time learning
Cites Work
- Online Dynamic Mode Decomposition for Time-Varying Systems
- Diffusion maps
- Equation-free, coarse-grained multiscale computation: enabling microscopic simulators to perform system-level analysis
- Nonintrusive and structure preserving multiscale integration of stiff ODEs, SDEs, and Hamiltonian systems with hidden slow dynamics via flow averaging
- Heterogeneous multiscale methods: a review
- Numerical Solution of the Navier-Stokes Equations
- The heterogeneous multiscale methods
- A penalization method to take into account obstacles in incompressible viscous flows
- Nonlinear model reduction for uncertainty quantification in large-scale inverse problems
- Title not available (Why is that?)
- Dynamic data-driven reduced-order models
- Data-driven operator inference for nonintrusive projection-based model reduction
- Online adaptive model reduction for nonlinear systems via low-rank updates
- Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks
- Model reduction and approximation. Theory and algorithms
- POD-DL-ROM: enhancing deep learning-based reduced order models for nonlinear parametrized PDEs by proper orthogonal decomposition
- Overcoming catastrophic forgetting in neural networks
- Data-driven reduced order model with temporal convolutional neural network
- Data-driven discovery of coordinates and governing equations
- Learning data-driven discretizations for partial differential equations
- Learning physics-based models from data: perspectives from inverse problems and model reduction
- Modeling the dynamics of PDE systems with physics-constrained deep auto-regressive networks
- LaSDI: parametric latent space dynamics identification
- gLaSDI: parametric physics-informed greedy latent space dynamics identification
- Unsteady flow prediction from sparse measurements by compressed sensing reduced order modeling
Cited In (1)
This page was built for publication: Adaptive learning of effective dynamics for online modeling of complex systems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6096436)
