Parameterized neural ordinary differential equations: applications to computational physics problems

被引:35
作者
Lee, Kookjin [1 ,2 ]
Parish, Eric J. [2 ]
机构
[1] Arizona State Univ, Sch Comp Informat & Decis Syst Engn, 699 South Mill Ave, Tempe, AZ 85281 USA
[2] Sandia Natl Labs, Extreme Scale Data Sci & Analyt Dept, 7011 East Ave,MS 9159, Livermore, MS 94550 USA
来源
PROCEEDINGS OF THE ROYAL SOCIETY A-MATHEMATICAL PHYSICAL AND ENGINEERING SCIENCES | 2021年 / 477卷 / 2253期
关键词
deep learning; autoencoders; nonlinear manifolds; model reduction; neural ordinary differential equations; latent-dynamics learning; PETROV-GALERKIN PROJECTION; FLUID-DYNAMICS; NETWORKS; FRAMEWORK;
D O I
10.1098/rspa.2021.0162
中图分类号
O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];
学科分类号
07 ; 0710 ; 09 ;
摘要
This work proposes an extension of neural ordinary differential equations (NODEs) by introducing an additional set of ODE input parameters to NODEs. This extension allows NODEs to learn multiple dynamics specified by the input parameter instances. Our extension is inspired by the concept of parameterized ODEs, which are widely investigated in computational science and engineering contexts, where characteristics of the governing equations vary over the input parameters. We apply the proposed parameterized NODEs (PNODEs) for learning latent dynamics of complex dynamical processes that arise in computational physics, which is an essential component for enabling rapid numerical simulations for time-critical physics applications. For this, we propose an encoder-decoder-type framework, which models latent dynamics as PNODEs. We demonstrate the effectiveness of PNODEs on benchmark problems from computational physics.
引用
收藏
页数:25
相关论文
共 82 条
[1]  
Acemoglu D, 2020, 27102 NAT BUR EC RES
[2]  
[Anonymous], 2017, DEEP LEARNING UNIVER
[3]  
[Anonymous], 2019, PROC ADV NEURAL INFO
[4]  
[Anonymous], 2017, Computational Science & Engineering, DOI 10.1137/1.9781611974829
[5]  
[Anonymous], 2021, PROC 35 AAAI C ARTIF
[6]  
[Anonymous], 2019, Physics-informed autoencoders for lyapunov-stable fluid flow prediction
[7]  
Ayed I., 2019, Learning dynamical systems from partial observations
[8]   A Survey of Projection-Based Model Reduction Methods for Parametric Dynamical Systems [J].
Benner, Peter ;
Gugercin, Serkan ;
Willcox, Karen .
SIAM REVIEW, 2015, 57 (04) :483-531
[9]  
Buffoni M., 2010, 40 FLUID DYN C EXH C, DOI 10.2514/6.2010-5008
[10]   Galerkin v. least-squares Petrov-Galerkin projection in nonlinear model reduction [J].
Carlberg, Kevin ;
Barone, Matthew ;
Antil, Harbir .
JOURNAL OF COMPUTATIONAL PHYSICS, 2017, 330 :693-734