Latent-space time evolution of non-intrusive reduced-order models using Gaussian process emulation
Abstract
Non-intrusive reduced-order models (ROMs) have recently generated considerable interest for constructing computationally efficient counterparts of nonlinear dynamical systems emerging from various domain sciences. They provide a low-dimensional emulation framework for systems that may be intrinsically high-dimensional. This is accomplished by utilizing a construction algorithm that is purely data-driven. It is no surprise, therefore, that the algorithmic advances of machine learning have led to non-intrusive ROMs with greater accuracy and computational gains. However, in bypassing the utilization of an equation-based evolution, it is often seen that the interpretability of the ROM framework suffers. This becomes more problematic when black-box deep learning methods are used which are notorious for lacking robustness outside the physical regime of the observed data. In this article, we propose the use of a novel latent-space interpolation algorithm based on Gaussian process regression. Notably, this reduced-order evolution of the system is parameterized by control parameters to allow for interpolation in space. The use of this procedure also allows for a continuous interpretation of time which allows for temporal interpolation. The latter aspect provides information, with quantified uncertainty, about full-state evolution at a finer resolution than that utilized for training the ROMs. This research assesses the viability of thismore »
- Authors:
-
- Argonne National Lab. (ANL), Argonne, IL (United States)
- The Alan Turing Institute, London (United Kingdom)
- Imperial College, London (United Kingdom)
- Publication Date:
- Research Org.:
- Argonne National Laboratory (ANL), Argonne, IL (United States)
- Sponsoring Org.:
- USDOE Office of Science (SC), Basic Energy Sciences (BES). Scientific User Facilities Division; Engineering and Physical Sciences Research Council (EPSRC); Imperial College, London (United Kingdom)
- OSTI Identifier:
- 1797914
- Grant/Contract Number:
- AC02-06CH11357
- Resource Type:
- Accepted Manuscript
- Journal Name:
- Physica. D, Nonlinear Phenomena
- Additional Journal Information:
- Journal Volume: 416; Journal ID: ISSN 0167-2789
- Publisher:
- Elsevier
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; Deep learning; Gaussian process regression; Reduced-order models
Citation Formats
Maulik, Romit, Botsas, Themistoklis, Ramachandra, Nesar, Mason, Lachlan Robert, and Pan, Indranil. Latent-space time evolution of non-intrusive reduced-order models using Gaussian process emulation. United States: N. p., 2021.
Web. doi:10.1016/j.physd.2020.132797.
Maulik, Romit, Botsas, Themistoklis, Ramachandra, Nesar, Mason, Lachlan Robert, & Pan, Indranil. Latent-space time evolution of non-intrusive reduced-order models using Gaussian process emulation. United States. https://doi.org/10.1016/j.physd.2020.132797
Maulik, Romit, Botsas, Themistoklis, Ramachandra, Nesar, Mason, Lachlan Robert, and Pan, Indranil. Mon .
"Latent-space time evolution of non-intrusive reduced-order models using Gaussian process emulation". United States. https://doi.org/10.1016/j.physd.2020.132797. https://www.osti.gov/servlets/purl/1797914.
@article{osti_1797914,
title = {Latent-space time evolution of non-intrusive reduced-order models using Gaussian process emulation},
author = {Maulik, Romit and Botsas, Themistoklis and Ramachandra, Nesar and Mason, Lachlan Robert and Pan, Indranil},
abstractNote = {Non-intrusive reduced-order models (ROMs) have recently generated considerable interest for constructing computationally efficient counterparts of nonlinear dynamical systems emerging from various domain sciences. They provide a low-dimensional emulation framework for systems that may be intrinsically high-dimensional. This is accomplished by utilizing a construction algorithm that is purely data-driven. It is no surprise, therefore, that the algorithmic advances of machine learning have led to non-intrusive ROMs with greater accuracy and computational gains. However, in bypassing the utilization of an equation-based evolution, it is often seen that the interpretability of the ROM framework suffers. This becomes more problematic when black-box deep learning methods are used which are notorious for lacking robustness outside the physical regime of the observed data. In this article, we propose the use of a novel latent-space interpolation algorithm based on Gaussian process regression. Notably, this reduced-order evolution of the system is parameterized by control parameters to allow for interpolation in space. The use of this procedure also allows for a continuous interpretation of time which allows for temporal interpolation. The latter aspect provides information, with quantified uncertainty, about full-state evolution at a finer resolution than that utilized for training the ROMs. This research assesses the viability of this algorithm for an advection-dominated system given by the inviscid shallow water equations.},
doi = {10.1016/j.physd.2020.132797},
journal = {Physica. D, Nonlinear Phenomena},
number = ,
volume = 416,
place = {United States},
year = {Mon Feb 01 00:00:00 EST 2021},
month = {Mon Feb 01 00:00:00 EST 2021}
}
Works referenced in this record:
Predicting clustered weather patterns: A test case for applications of convolutional neural networks to spatio-temporal climate data
journal, January 2020
- Chattopadhyay, Ashesh; Hassanzadeh, Pedram; Pasha, Saba
- Scientific Reports, Vol. 10, Issue 1
Data-driven discovery of coordinates and governing equations
journal, October 2019
- Champion, Kathleen; Lusch, Bethany; Kutz, J. Nathan
- Proceedings of the National Academy of Sciences, Vol. 116, Issue 45
Layered reduced-order models for nonlinear aerodynamics and aeroelasticity
journal, January 2017
- Kou, Jiaqing; Zhang, Weiwei
- Journal of Fluids and Structures, Vol. 68
Time-series learning of latent-space dynamics for reduced-order model closure
journal, April 2020
- Maulik, Romit; Mohan, Arvind; Lusch, Bethany
- Physica D: Nonlinear Phenomena, Vol. 405
Aerodynamic Data Reconstruction and Inverse Design Using Proper Orthogonal Decomposition
journal, August 2004
- Bui-Thanh, T.; Damodaran, M.; Willcox, K.
- AIAA Journal, Vol. 42, Issue 8
Nonlinear aeroelastic reduced order modeling by recurrent neural networks
journal, July 2014
- Mannarino, Andrea; Mantegazza, Paolo
- Journal of Fluids and Structures, Vol. 48
Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks
journal, May 2018
- Vlachas, Pantelis R.; Byeon, Wonmin; Wan, Zhong Y.
- Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, Vol. 474, Issue 2213
Non-autoregressive time-series methods for stable parametric reduced-order models
journal, August 2020
- Maulik, Romit; Lusch, Bethany; Balaprakash, Prasanna
- Physics of Fluids, Vol. 32, Issue 8
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
journal, February 2019
- Raissi, M.; Perdikaris, P.; Karniadakis, G. E.
- Journal of Computational Physics, Vol. 378
Physically constrained data‐driven correction for reduced‐order modeling of fluid flows
journal, October 2018
- Mohebujjaman, M.; Rebholz, L. G.; Iliescu, T.
- International Journal for Numerical Methods in Fluids, Vol. 89, Issue 3
Principal interval decomposition framework for POD reduced-order modeling of convective Boussinesq flows: PRINCIPAL INTERVAL DECOMPOSITION MODEL REDUCTION FRAMEWORK
journal, February 2015
- San, O.; Borggaard, J.
- International Journal for Numerical Methods in Fluids, Vol. 78, Issue 1
On the stability and convergence of a Galerkin reduced order model (ROM) of compressible flow with solid wall and far-field boundary treatment
journal, August 2010
- Kalashnikova, I.; Barone, M. F.
- International Journal for Numerical Methods in Engineering, Vol. 83, Issue 10
PDE-Net 2.0: Learning PDEs from data with a numeric-symbolic hybrid deep network
journal, December 2019
- Long, Zichao; Lu, Yiping; Dong, Bin
- Journal of Computational Physics, Vol. 399
Koopman-Based Approach to Nonintrusive Projection-Based Reduced-Order Modeling with Black-Box High-Fidelity Models
journal, October 2018
- Renganathan, S. Ashwin; Liu, Yingjie; Mavris, Dimitri N.
- AIAA Journal, Vol. 56, Issue 10
Nonlinear Aerodynamic Reduced-Order Model for Limit-Cycle Oscillation and Flutter
journal, October 2016
- Zhang, Weiwei; Kou, Jiaqing; Wang, Ziyi
- AIAA Journal, Vol. 54, Issue 10
Applying Neural Network Models to Prediction and Data Analysis in Meteorology and Oceanography
journal, September 1998
- Hsieh, William W.; Tang, Benyang
- Bulletin of the American Meteorological Society, Vol. 79, Issue 9
Non-linear Petrov–Galerkin methods for reduced order hyperbolic equations and discontinuous finite element methods
journal, February 2013
- Fang, F.; Pain, C. C.; Navon, I. M.
- Journal of Computational Physics, Vol. 234
Efficient non-linear model reduction via a least-squares Petrov-Galerkin projection and compressive tensor approximations
journal, October 2010
- Carlberg, Kevin; Bou-Mosleh, Charbel; Farhat, Charbel
- International Journal for Numerical Methods in Engineering, Vol. 86, Issue 2
Model reduction for compressible flows using POD and Galerkin projection
journal, February 2004
- Rowley, Clarence W.; Colonius, Tim; Murray, Richard M.
- Physica D: Nonlinear Phenomena, Vol. 189, Issue 1-2
DGM: A deep learning algorithm for solving partial differential equations
journal, December 2018
- Sirignano, Justin; Spiliopoulos, Konstantinos
- Journal of Computational Physics, Vol. 375
The Proper Orthogonal Decomposition in the Analysis of Turbulent Flows
journal, January 1993
- Berkooz, G.; Holmes, P.; Lumley, J. L.
- Annual Review of Fluid Mechanics, Vol. 25, Issue 1
An intrinsic stabilization scheme for proper orthogonal decomposition based low-dimensional models
journal, May 2007
- Kalb, Virginia L.; Deane, Anil E.
- Physics of Fluids, Vol. 19, Issue 5
Deep Learning Methods for Reynolds-Averaged Navier–Stokes Simulations of Airfoil Flows
journal, January 2020
- Thuerey, Nils; Weißenow, Konstantin; Prantl, Lukas
- AIAA Journal, Vol. 58, Issue 1
A long short-term memory embedding for hybrid uplifted reduced order models
journal, August 2020
- Ahmed, Shady E.; San, Omer; Rasheed, Adil
- Physica D: Nonlinear Phenomena, Vol. 409
Multi-level convolutional autoencoder networks for parametric prediction of spatio-temporal dynamics
journal, December 2020
- Xu, Jiayang; Duraisamy, Karthik
- Computer Methods in Applied Mechanics and Engineering, Vol. 372
Non-intrusive reduced order modeling of nonlinear problems using neural networks
journal, June 2018
- Hesthaven, J. S.; Ubbiali, S.
- Journal of Computational Physics, Vol. 363
Non-linear Petrov–Galerkin methods for reduced order modelling of the Navier–Stokes equations using a mixed finite element pair
journal, March 2013
- Xiao, D.; Fang, F.; Du, J.
- Computer Methods in Applied Mechanics and Engineering, Vol. 255
Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders
journal, November 2019
- Lee, Kookjin; Carlberg, Kevin T.
- Journal of Computational Physics
Proper orthogonal decomposition closure models for turbulent flows: A numerical comparison
journal, September 2012
- Wang, Zhu; Akhtar, Imran; Borggaard, Jeff
- Computer Methods in Applied Mechanics and Engineering, Vol. 237-240
Neural network closures for nonlinear model order reduction
journal, January 2018
- San, Omer; Maulik, Romit
- Advances in Computational Mathematics, Vol. 44, Issue 6
Recurrent neural network closure of parametric POD-Galerkin reduced-order models based on the Mori-Zwanzig formalism
journal, June 2020
- Wang, Qian; Ripamonti, Nicolò; Hesthaven, Jan S.
- Journal of Computational Physics, Vol. 410
On the stability and convergence of a Galerkin reduced order model (ROM) of compressible flow with solid wall and far-field boundary treatment
journal, August 2010
- Kalashnikova, I.; Barone, M. F.
- International Journal for Numerical Methods in Engineering, Vol. 83, Issue 10
Nonlinear aeroelastic reduced order modeling by recurrent neural networks
journal, July 2014
- Mannarino, Andrea; Mantegazza, Paolo
- Journal of Fluids and Structures, Vol. 48
Non-autoregressive time-series methods for stable parametric reduced-order models
journal, August 2020
- Maulik, Romit; Lusch, Bethany; Balaprakash, Prasanna
- Physics of Fluids, Vol. 32, Issue 8
Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks
journal, May 2018
- Vlachas, Pantelis R.; Byeon, Wonmin; Wan, Zhong Y.
- Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, Vol. 474, Issue 2213
Deep convolutional recurrent autoencoders for learning low-dimensional feature dynamics of fluid systems
preprint, January 2018
- Gonzalez, Francisco J.; Balajewicz, Maciej
- arXiv
Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders
preprint, January 2018
- Lee, Kookjin; Carlberg, Kevin
- arXiv
Data-driven discovery of coordinates and governing equations
text, January 2019
- Champion, Kathleen; Lusch, Bethany; Kutz, J. Nathan
- arXiv
Analog forecasting of extreme-causing weather patterns using deep learning
text, January 2019
- Chattopadhyay, Ashesh; Nabizadeh, Ebrahim; Hassanzadeh, Pedram
- arXiv
Recurrent Neural Network Architecture Search for Geophysical Emulation
preprint, January 2020
- Maulik, Romit; Egele, Romain; Lusch, Bethany
- arXiv
Works referencing / citing this record:
An AI-based Domain-Decomposition Non-Intrusive Reduced-Order Model for Extended Domains applied to Multiphase Flow in Pipes
text, January 2022
- Heaney, Claire E.; Wolffs, Zef; Tómasson, Jón Atli
- arXiv