Learning nonlinear projections for reduced-order modeling of dynamical systems using constrained autoencoders

Recently developed reduced-order modeling techniques aim to approximate nonlinear dynamical systems on low-dimensional manifolds learned from data. This is an effective approach for modeling dynamics in a post-transient regime where the effects of initial conditions and other disturbances have decay...

Full description

Saved in:
Bibliographic Details
Published in:Chaos (Woodbury, N.Y.) Vol. 33; no. 11
Main Authors: Otto, Samuel E, Macchio, Gregory R, Rowley, Clarence W
Format: Journal Article
Language:English
Published: 01.11.2023
ISSN:1089-7682, 1089-7682
Online Access:Get more information
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract Recently developed reduced-order modeling techniques aim to approximate nonlinear dynamical systems on low-dimensional manifolds learned from data. This is an effective approach for modeling dynamics in a post-transient regime where the effects of initial conditions and other disturbances have decayed. However, modeling transient dynamics near an underlying manifold, as needed for real-time control and forecasting applications, is complicated by the effects of fast dynamics and nonnormal sensitivity mechanisms. To begin to address these issues, we introduce a parametric class of nonlinear projections described by constrained autoencoder neural networks in which both the manifold and the projection fibers are learned from data. Our architecture uses invertible activation functions and biorthogonal weight matrices to ensure that the encoder is a left inverse of the decoder. We also introduce new dynamics-aware cost functions that promote learning of oblique projection fibers that account for fast dynamics and nonnormality. To demonstrate these methods and the specific challenges they address, we provide a detailed case study of a three-state model of vortex shedding in the wake of a bluff body immersed in a fluid, which has a two-dimensional slow manifold that can be computed analytically. In anticipation of future applications to high-dimensional systems, we also propose several techniques for constructing computationally efficient reduced-order models using our proposed nonlinear projection framework. This includes a novel sparsity-promoting penalty for the encoder that avoids detrimental weight matrix shrinkage via computation on the Grassmann manifold.Recently developed reduced-order modeling techniques aim to approximate nonlinear dynamical systems on low-dimensional manifolds learned from data. This is an effective approach for modeling dynamics in a post-transient regime where the effects of initial conditions and other disturbances have decayed. However, modeling transient dynamics near an underlying manifold, as needed for real-time control and forecasting applications, is complicated by the effects of fast dynamics and nonnormal sensitivity mechanisms. To begin to address these issues, we introduce a parametric class of nonlinear projections described by constrained autoencoder neural networks in which both the manifold and the projection fibers are learned from data. Our architecture uses invertible activation functions and biorthogonal weight matrices to ensure that the encoder is a left inverse of the decoder. We also introduce new dynamics-aware cost functions that promote learning of oblique projection fibers that account for fast dynamics and nonnormality. To demonstrate these methods and the specific challenges they address, we provide a detailed case study of a three-state model of vortex shedding in the wake of a bluff body immersed in a fluid, which has a two-dimensional slow manifold that can be computed analytically. In anticipation of future applications to high-dimensional systems, we also propose several techniques for constructing computationally efficient reduced-order models using our proposed nonlinear projection framework. This includes a novel sparsity-promoting penalty for the encoder that avoids detrimental weight matrix shrinkage via computation on the Grassmann manifold.
AbstractList Recently developed reduced-order modeling techniques aim to approximate nonlinear dynamical systems on low-dimensional manifolds learned from data. This is an effective approach for modeling dynamics in a post-transient regime where the effects of initial conditions and other disturbances have decayed. However, modeling transient dynamics near an underlying manifold, as needed for real-time control and forecasting applications, is complicated by the effects of fast dynamics and nonnormal sensitivity mechanisms. To begin to address these issues, we introduce a parametric class of nonlinear projections described by constrained autoencoder neural networks in which both the manifold and the projection fibers are learned from data. Our architecture uses invertible activation functions and biorthogonal weight matrices to ensure that the encoder is a left inverse of the decoder. We also introduce new dynamics-aware cost functions that promote learning of oblique projection fibers that account for fast dynamics and nonnormality. To demonstrate these methods and the specific challenges they address, we provide a detailed case study of a three-state model of vortex shedding in the wake of a bluff body immersed in a fluid, which has a two-dimensional slow manifold that can be computed analytically. In anticipation of future applications to high-dimensional systems, we also propose several techniques for constructing computationally efficient reduced-order models using our proposed nonlinear projection framework. This includes a novel sparsity-promoting penalty for the encoder that avoids detrimental weight matrix shrinkage via computation on the Grassmann manifold.Recently developed reduced-order modeling techniques aim to approximate nonlinear dynamical systems on low-dimensional manifolds learned from data. This is an effective approach for modeling dynamics in a post-transient regime where the effects of initial conditions and other disturbances have decayed. However, modeling transient dynamics near an underlying manifold, as needed for real-time control and forecasting applications, is complicated by the effects of fast dynamics and nonnormal sensitivity mechanisms. To begin to address these issues, we introduce a parametric class of nonlinear projections described by constrained autoencoder neural networks in which both the manifold and the projection fibers are learned from data. Our architecture uses invertible activation functions and biorthogonal weight matrices to ensure that the encoder is a left inverse of the decoder. We also introduce new dynamics-aware cost functions that promote learning of oblique projection fibers that account for fast dynamics and nonnormality. To demonstrate these methods and the specific challenges they address, we provide a detailed case study of a three-state model of vortex shedding in the wake of a bluff body immersed in a fluid, which has a two-dimensional slow manifold that can be computed analytically. In anticipation of future applications to high-dimensional systems, we also propose several techniques for constructing computationally efficient reduced-order models using our proposed nonlinear projection framework. This includes a novel sparsity-promoting penalty for the encoder that avoids detrimental weight matrix shrinkage via computation on the Grassmann manifold.
Author Rowley, Clarence W
Otto, Samuel E
Macchio, Gregory R
Author_xml – sequence: 1
  givenname: Samuel E
  surname: Otto
  fullname: Otto, Samuel E
– sequence: 2
  givenname: Gregory R
  surname: Macchio
  fullname: Macchio, Gregory R
– sequence: 3
  givenname: Clarence W
  surname: Rowley
  fullname: Rowley, Clarence W
BookMark eNpNkLtOAzEQRS0UJJJAwR-4pNng99olinhJK9FAHfkxizbatcHeLfL3OIKCauZqdI50Z4NWMUVA6JaSHSWK38sdocoorS_QmhJtmlZptvq3X6FNKUdCCGVcrtHUgc1xiJ-4isYh1oS_cjqCn4cUC-5TxhnC4iE0KQfIeEoBxjOQehxO0U6DtyMupzLDVPBSzidf0TnbqgvYLnOC6CuVyzW67O1Y4OZvbtHH0-P7_qXp3p5f9w9d47kicxM4Z6TVwbWtE0y4tg9WQrCOe8uI7IUwRnGnnQxUUSKFsb3nlEqteH0CsC26-_XWKt8LlPkwDcXDONoIaSkHpo1omeKGsB9TEl9X
CitedBy_id crossref_primary_10_1016_j_physd_2024_134299
crossref_primary_10_1007_s11071_024_10712_w
crossref_primary_10_1016_j_jcp_2024_113371
crossref_primary_10_1016_j_jcp_2025_113817
crossref_primary_10_1063_5_0237494
crossref_primary_10_1063_5_0243849
ContentType Journal Article
Copyright 2023 Author(s). Published under an exclusive license by AIP Publishing.
Copyright_xml – notice: 2023 Author(s). Published under an exclusive license by AIP Publishing.
DBID 7X8
DOI 10.1063/5.0169688
DatabaseName MEDLINE - Academic
DatabaseTitle MEDLINE - Academic
DatabaseTitleList MEDLINE - Academic
Database_xml – sequence: 1
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod no_fulltext_linktorsrc
Discipline Sciences (General)
Physics
EISSN 1089-7682
GroupedDBID ---
-~X
.DC
1UP
2-P
29B
4.4
53G
5VS
7X8
AAAAW
AABDS
AAGWI
AAPUP
AAYIH
ABJGX
ABJNI
ACBRY
ACGFS
ACLYJ
ACNCT
ACZLF
ADCTM
AEJMO
AENEX
AFATG
AFHCQ
AGKCL
AGLKD
AGMXG
AGTJO
AHSDT
AJJCW
AJQPL
ALEPV
ALMA_UNASSIGNED_HOLDINGS
AMVHM
AQWKA
ATXIE
AWQPM
BDMKI
BPZLN
DU5
EBS
F5P
FDOHQ
FFFMQ
HAM
M6X
M71
M73
N9A
NPSNA
O-B
P2P
RIP
RNS
RQS
TAE
WH7
ID FETCH-LOGICAL-c360t-d332078db77b424b7fda5edab3ca205f449963b8b5d1610549afc3115863063e2
IEDL.DBID 7X8
ISICitedReferencesCount 11
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001115002100003&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1089-7682
IngestDate Wed Oct 01 17:12:47 EDT 2025
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 11
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c360t-d332078db77b424b7fda5edab3ca205f449963b8b5d1610549afc3115863063e2
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
OpenAccessLink https://pubs.aip.org/aip/cha/article-pdf/doi/10.1063/5.0169688/18226985/113130_1_5.0169688.pdf
PQID 2894726390
PQPubID 23479
ParticipantIDs proquest_miscellaneous_2894726390
PublicationCentury 2000
PublicationDate 20231101
PublicationDateYYYYMMDD 2023-11-01
PublicationDate_xml – month: 11
  year: 2023
  text: 20231101
  day: 01
PublicationDecade 2020
PublicationTitle Chaos (Woodbury, N.Y.)
PublicationYear 2023
SSID ssj0001235
Score 2.5114918
Snippet Recently developed reduced-order modeling techniques aim to approximate nonlinear dynamical systems on low-dimensional manifolds learned from data. This is an...
SourceID proquest
SourceType Aggregation Database
Title Learning nonlinear projections for reduced-order modeling of dynamical systems using constrained autoencoders
URI https://www.proquest.com/docview/2894726390
Volume 33
WOSCitedRecordID wos001115002100003&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1LSwMxEA5qFbyorYpvInjQQ-g2yW6Sk4hYPBUPCr2VPIsHu9pt_f1msik9eBG8b9gQJjPfPPJ9CN0MYliXhjrCOWeEi6CIkYITCT0fJZh1TiaxCTEayfFYveSCW5PHKlc-MTlqV1uokfdjYsAFjfG0uP_8IqAaBd3VLKGxiTosQhkY6RLjNVs4vANtB-wVibCarpiFKtYHsk7ghZG_fHAKLMP9_27pAO1lSIkfWhvoog0_66GdNNppmx7q5uvb4NvMMX13iD4yr-oUz1quDD3HuSgDdogjlMVzYHX1jiR2Tpwkc2BBHbBrZezjT1si6AbD-PwUWwCboDnhHdbLRQ0kmTAofYTehk-vj88kKy8Qy6piQRxjNGIHZ4QwnHIjgtOld9owq2lRBh7zpIoZaUoXEWNEfUoHC7w9soopCPP0GG3F7fsThBmzpRqwsmBa8SIoLTwNRmhvTFkZxk_R9epwJ9GyoV2hZ75eNpP18Z794ZtztAtK8O0zwQvUCfH2-ku0bb8X7838KhnGD6rnw74
linkProvider ProQuest
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Learning+nonlinear+projections+for+reduced-order+modeling+of+dynamical+systems+using+constrained+autoencoders&rft.jtitle=Chaos+%28Woodbury%2C+N.Y.%29&rft.au=Otto%2C+Samuel+E&rft.au=Macchio%2C+Gregory+R&rft.au=Rowley%2C+Clarence+W&rft.date=2023-11-01&rft.issn=1089-7682&rft.eissn=1089-7682&rft.volume=33&rft.issue=11&rft_id=info:doi/10.1063%2F5.0169688&rft.externalDBID=NO_FULL_TEXT
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1089-7682&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1089-7682&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1089-7682&client=summon