A deep learning algorithm for computing mean field control problems via forward-backward score dynamics

We propose a deep learning approach to compute mean field control problems with individual noises. The problem consists of the Fokker-Planck (FP) equation and the Hamilton-Jacobi-Bellman (HJB) equation. Using the differential of the entropy, namely the score function, we first formulate the determin...

Full description

Saved in:
Bibliographic Details
Published inResearch in the mathematical sciences Vol. 12; no. 3; p. 42
Main Authors Zhou, Mo, Osher, Stanley, Li, Wuchen
Format Journal Article
LanguageEnglish
Published Cham Springer International Publishing 01.09.2025
Springer Nature B.V
Subjects
Online AccessGet full text
ISSN2522-0144
2197-9847
DOI10.1007/s40687-025-00531-9

Cover

Abstract We propose a deep learning approach to compute mean field control problems with individual noises. The problem consists of the Fokker-Planck (FP) equation and the Hamilton-Jacobi-Bellman (HJB) equation. Using the differential of the entropy, namely the score function, we first formulate the deterministic forward-backward characteristics for the mean field control system, which is different from the classical forward-backward stochastic differential equations (FBSDEs). We further apply the neural network approximation to fit the proposed deterministic characteristic lines. Numerical examples, including the control problem with entropy potential energy, the linear quadratic regulator, and the systemic risks, demonstrate the effectiveness of the proposed method.
AbstractList We propose a deep learning approach to compute mean field control problems with individual noises. The problem consists of the Fokker-Planck (FP) equation and the Hamilton-Jacobi-Bellman (HJB) equation. Using the differential of the entropy, namely the score function, we first formulate the deterministic forward-backward characteristics for the mean field control system, which is different from the classical forward-backward stochastic differential equations (FBSDEs). We further apply the neural network approximation to fit the proposed deterministic characteristic lines. Numerical examples, including the control problem with entropy potential energy, the linear quadratic regulator, and the systemic risks, demonstrate the effectiveness of the proposed method.
ArticleNumber 42
Author Osher, Stanley
Li, Wuchen
Zhou, Mo
Author_xml – sequence: 1
  givenname: Mo
  surname: Zhou
  fullname: Zhou, Mo
  organization: Department of Mathematics, University of California
– sequence: 2
  givenname: Stanley
  surname: Osher
  fullname: Osher, Stanley
  organization: Department of Mathematics, University of California
– sequence: 3
  givenname: Wuchen
  orcidid: 0000-0002-2218-5734
  surname: Li
  fullname: Li, Wuchen
  email: wuchen@mailbox.sc.edu
  organization: Department of Mathematics, University of South Carolina
BookMark eNp9kM1LAzEQxYMo-PkPeAp4jk42yWZzLOIXFLzoOaSb2bp1N6nJVul_b2oFb57mMfzem-GdksMQAxJyyeGaA-ibLKFuNINKMQAlODMH5KTiRjPTSH1YtKoqBlzKY3KR8woAuK6FFHBCljPqEdd0QJdCH5bUDcuY-ultpF1MtI3jejPt9iO6QLseB1-WYUpxoOsUFwOOmX72bkd_ueTZwrXvO0FzGxNSvw1u7Nt8To46N2S8-J1n5PX-7uX2kc2fH55uZ3PWVrqaWNNC47WWSsoGjZDaS9Sqqx2X2Elu1EI40QqDtak9SI5edLIDtTDeF9SLM3K1zy3PfWwwT3YVNymUk1ZUygholFKFqvZUm2LOCTu7Tv3o0tZysLtO7b5TWzq1P51aU0xib8oFDktMf9H_uL4B13p8YA
Cites_doi 10.1088/1748-3190/ab49a4
10.1007/s00332-018-9525-3
10.1214/21-AAP1715
10.1137/140985676
10.4310/CMS.2015.v13.n4.a4
10.1137/21M1402303
10.1137/19M1274377
10.1016/j.matpur.2021.10.010
10.1016/j.jcp.2022.111264
10.1214/14-AOP946
10.1007/s40687-024-00479-2
10.1073/pnas.2024713118
10.1007/978-1-4612-1466-3_7
10.1007/s00498-021-00310-1
10.1073/pnas.1922204117
10.1007/978-1-4612-1466-3
10.1017/9781009028943.021
10.1016/j.jcp.2020.109792
10.1007/978-3-319-56436-4
10.1007/978-1-4614-8508-7
10.1109/MIS.2020.2971597
10.23943/princeton/9780691190716.001.0001
10.1007/s11009-022-09946-1
10.1109/CDC.1997.652379
10.1016/j.jcp.2022.111409
10.1016/j.jcp.2023.112019
ContentType Journal Article
Copyright The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025 Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025.
Copyright_xml – notice: The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025 Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
– notice: The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025.
DBID AAYXX
CITATION
DOI 10.1007/s40687-025-00531-9
DatabaseName CrossRef
DatabaseTitle CrossRef
DatabaseTitleList

DeliveryMethod fulltext_linktorsrc
Discipline Mathematics
EISSN 2197-9847
ExternalDocumentID 10_1007_s40687_025_00531_9
GroupedDBID 0R~
406
5VS
AACDK
AAHNG
AAJBT
AASML
AATNV
AATVU
AAUYE
AAYZH
ABAKF
ABBRH
ABDBE
ABDZT
ABECU
ABFSG
ABFTV
ABJNI
ABJOX
ABKCH
ABMQK
ABRTQ
ABSXP
ABTEG
ABTKH
ABTMW
ABXPI
ACAOD
ACDTI
ACGFS
ACHSB
ACMLO
ACOKC
ACPIV
ACSTC
ACZOJ
ADHHG
ADKNI
ADKPE
ADTPH
ADURQ
ADYFF
ADZKW
AEFQL
AEJRE
AEMSY
AESKC
AEZWR
AFBBN
AFDZB
AFHIU
AFOHR
AFQWF
AGDGC
AGJBK
AGMZJ
AGQEE
AGRTI
AHBYD
AHPBZ
AHWEU
AHYZX
AIAKS
AIGIU
AILAN
AITGF
AIXLP
AJZVZ
ALMA_UNASSIGNED_HOLDINGS
AMKLP
AMXSW
AMYLF
AMYQR
ASPBG
ATHPR
AXYYD
AYFIA
BGNMA
CSCUP
DPUIP
EBLON
EBS
FIGPU
FNLPD
GGCAI
IKXTQ
IWAJR
J-C
JZLTJ
KOV
LLZTM
M4Y
NPVJJ
NQJWS
NU0
O9J
PT4
ROL
RSV
SISQX
SJYHP
SNE
SNPRN
SOHCF
SOJ
SPISZ
SRMVM
SSLCW
STPWE
TSG
UOJIU
UTJUX
UZXMN
VFIZW
ZMTXR
AAAVM
AAYXX
AHSBF
AMVHM
CITATION
EJD
FINBP
FSGXE
OK1
ID FETCH-LOGICAL-c272t-8c08d7745448e9347d4e75f6a14ef4195b3a3c39e696d041ed3f4f05b9ddd4ed3
ISSN 2522-0144
IngestDate Sat Oct 11 06:55:43 EDT 2025
Wed Oct 01 05:31:18 EDT 2025
Sat Sep 06 10:14:39 EDT 2025
IsPeerReviewed true
IsScholarly true
Issue 3
Keywords Deep learning algorithms
Forward-backward score dynamics
Mean field control
Language English
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c272t-8c08d7745448e9347d4e75f6a14ef4195b3a3c39e696d041ed3f4f05b9ddd4ed3
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ORCID 0000-0002-2218-5734
PQID 3259308555
PQPubID 2044439
ParticipantIDs proquest_journals_3259308555
crossref_primary_10_1007_s40687_025_00531_9
springer_journals_10_1007_s40687_025_00531_9
ProviderPackageCode CITATION
AAYXX
PublicationCentury 2000
PublicationDate 2025-09-01
PublicationDateYYYYMMDD 2025-09-01
PublicationDate_xml – month: 09
  year: 2025
  text: 2025-09-01
  day: 01
PublicationDecade 2020
PublicationPlace Cham
PublicationPlace_xml – name: Cham
– name: Heidelberg
PublicationTitle Research in the mathematical sciences
PublicationTitleAbbrev Res Math Sci
PublicationYear 2025
Publisher Springer International Publishing
Springer Nature B.V
Publisher_xml – name: Springer International Publishing
– name: Springer Nature B.V
References W Li (531_CR14) 2023; 480
M Zhou (531_CR19) 2021; 43
D Bauso (531_CR2) 2016; 54
531_CR12
W Li (531_CR15) 2022; 466
531_CR29
531_CR28
W Li (531_CR13) 2022; 463
531_CR27
R Carmona (531_CR20) 2021; 59
L Ruthotto (531_CR18) 2020; 117
P-EC Raynal (531_CR10) 2021; 156
RA Carmona (531_CR31) 2015; 13
M Germain (531_CR11) 2022; 24
K Elamvazhuthi (531_CR3) 2019; 15
531_CR1
531_CR4
R Carmona (531_CR21) 2022; 32
S Ji (531_CR22) 2020; 35
531_CR5
R Carmona (531_CR9) 2015; 43
531_CR6
531_CR7
531_CR26
A Angiuli (531_CR16) 2022; 34
531_CR25
531_CR24
531_CR23
R Flamary (531_CR32) 2021; 22
531_CR17
NM Boffi (531_CR8) 2023; 4
J Han (531_CR30) 2020; 423
References_xml – volume: 15
  issue: 1
  year: 2019
  ident: 531_CR3
  publication-title: Bioinspiration Biomimetics
  doi: 10.1088/1748-3190/ab49a4
– ident: 531_CR17
  doi: 10.1007/s00332-018-9525-3
– volume: 32
  start-page: 4065
  issue: 6
  year: 2022
  ident: 531_CR21
  publication-title: The Annals of Applied Probability
  doi: 10.1214/21-AAP1715
– volume: 54
  start-page: 3225
  issue: 6
  year: 2016
  ident: 531_CR2
  publication-title: SIAM Journal on Control and Optimization
  doi: 10.1137/140985676
– ident: 531_CR24
– volume: 13
  start-page: 911
  issue: 4
  year: 2015
  ident: 531_CR31
  publication-title: Communications in Mathematical Sciences
  doi: 10.4310/CMS.2015.v13.n4.a4
– volume: 43
  start-page: 4043
  issue: 6
  year: 2021
  ident: 531_CR19
  publication-title: SIAM Journal on Scientific Computing
  doi: 10.1137/21M1402303
– volume: 59
  start-page: 1455
  issue: 3
  year: 2021
  ident: 531_CR20
  publication-title: SIAM Journal on Numerical Analysis
  doi: 10.1137/19M1274377
– volume: 156
  start-page: 1
  year: 2021
  ident: 531_CR10
  publication-title: Journal de Mathématiques Pures et Appliquées
  doi: 10.1016/j.matpur.2021.10.010
– volume: 463
  year: 2022
  ident: 531_CR13
  publication-title: Journal of Computational Physics
  doi: 10.1016/j.jcp.2022.111264
– volume: 43
  start-page: 2647
  issue: 5
  year: 2015
  ident: 531_CR9
  publication-title: The Annals of Probability
  doi: 10.1214/14-AOP946
– ident: 531_CR29
  doi: 10.1007/s40687-024-00479-2
– ident: 531_CR4
  doi: 10.1073/pnas.2024713118
– ident: 531_CR12
– ident: 531_CR28
– volume: 22
  start-page: 1
  issue: 78
  year: 2021
  ident: 531_CR32
  publication-title: Journal of Machine Learning Research
– ident: 531_CR23
  doi: 10.1007/978-1-4612-1466-3_7
– volume: 34
  start-page: 217
  issue: 2
  year: 2022
  ident: 531_CR16
  publication-title: Mathematics of Control, Signals, and Systems
  doi: 10.1007/s00498-021-00310-1
– volume: 117
  start-page: 9183
  issue: 17
  year: 2020
  ident: 531_CR18
  publication-title: Proceedings of the National Academy of Sciences
  doi: 10.1073/pnas.1922204117
– ident: 531_CR6
  doi: 10.1007/978-1-4612-1466-3
– ident: 531_CR1
  doi: 10.1017/9781009028943.021
– volume: 423
  year: 2020
  ident: 531_CR30
  publication-title: Journal of Computational Physics
  doi: 10.1016/j.jcp.2020.109792
– volume: 4
  issue: 3
  year: 2023
  ident: 531_CR8
  publication-title: Machine Learning: Science and Technology
– ident: 531_CR5
  doi: 10.1007/978-3-319-56436-4
– ident: 531_CR26
  doi: 10.1007/978-1-4614-8508-7
– volume: 35
  start-page: 71
  issue: 3
  year: 2020
  ident: 531_CR22
  publication-title: IEEE Intelligent Systems
  doi: 10.1109/MIS.2020.2971597
– ident: 531_CR27
  doi: 10.23943/princeton/9780691190716.001.0001
– volume: 24
  start-page: 2557
  issue: 4
  year: 2022
  ident: 531_CR11
  publication-title: Methodology and Computing in Applied Probability
  doi: 10.1007/s11009-022-09946-1
– ident: 531_CR7
  doi: 10.1109/CDC.1997.652379
– volume: 466
  year: 2022
  ident: 531_CR15
  publication-title: Journal of Computational Physics
  doi: 10.1016/j.jcp.2022.111409
– volume: 480
  year: 2023
  ident: 531_CR14
  publication-title: Journal of Computational Physics
  doi: 10.1016/j.jcp.2023.112019
– ident: 531_CR25
SSID ssj0001763430
ssib053846527
Score 2.3047657
Snippet We propose a deep learning approach to compute mean field control problems with individual noises. The problem consists of the Fokker-Planck (FP) equation and...
SourceID proquest
crossref
springer
SourceType Aggregation Database
Index Database
Publisher
StartPage 42
SubjectTerms Algorithms
Applications of Mathematics
Brownian motion
Computational Mathematics and Numerical Analysis
Deep learning
Differential equations
Entropy
Lagrange multiplier
Linear quadratic regulator
Machine learning
Mathematics
Mathematics and Statistics
Neural networks
Potential energy
Shadow prices
Velocity
Viscosity
Title A deep learning algorithm for computing mean field control problems via forward-backward score dynamics
URI https://link.springer.com/article/10.1007/s40687-025-00531-9
https://www.proquest.com/docview/3259308555
Volume 12
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVLSH
  databaseName: SpringerLink Journals
  customDbUrl:
  mediaType: online
  eissn: 2197-9847
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0001763430
  issn: 2522-0144
  databaseCode: AFBBN
  dateStart: 20140617
  isFulltext: true
  providerName: Library Specific Holdings
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV1Lb9NAEF6F9AIHxFMEWrQHxMUY2d5d23tMq0YVasMlEREXy_auUyTyaJ1w4MY_Z2btXbstIOBiRRvLSnY-z87jmxlC3iiuwU0QkV9xTDOCTeLnSia-TkWMeSw407A4-WIan835h4VYDAY_-tUlu-J9-f2XdSX_I1VYA7lilew_SNY9FBbgM8gXriBhuP6VjMee0nprJz8svfzrcgPO_uXKkAdLM7AB11cYbjdcNUdNbwfJ1N63LznebcizBQbzMK5QY3NLTzXT6uu-AWuZepYfuXJtX211ZdmxEj9fbvYm4rpxsVw7aB2nF9ucMvKBDKng0x4gtO4HIiLhmFY3A5HIssbchyuUQV0WgZnno_PWHDvNWgiYkGnTcNMp46gHOtbTrE0PrjsKv-F41GCWgLI0Pwq1ii-7482m9Kcfs8n8_DybnS5mb7dXPg4ewwR9O4XlHjmI4GAIhuRgPDk-nlqlBAcCj0XbKtUE7kAjczPDxv2pthjLlGTe-SU3DZ7Oi7mVeDf2zOwRedg6InTcoOoxGej1E_LgwomzfkqWY4r4ohZf1OGLAmKowxdFfFGDL9rii1p8UcAXvY0vavBFLb6ekfnkdHZy5reDOfwySqKdn5ZBqsBvEODba8l4Am98Iqo4D7mueChFwXJWMqljGauAh1qxileBKKRScKtiz8lwvVnrF4SqlBVJBT6LBjsx1hH4G-BCwJkblqoIZD4int29bNv0X8lcp22z1xnsdWb2OpMjcmg3OGvf0zpj4OEzpGOKEXlnN737-vdPe_nnp70i97vX4JAMd9d7fQQW6q543WLoJxLykHM
linkProvider Library Specific Holdings
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=A+deep+learning+algorithm+for+computing+mean+field+control+problems+via+forward-backward+score+dynamics&rft.jtitle=Research+in+the+mathematical+sciences&rft.au=Zhou%2C+Mo&rft.au=Osher%2C+Stanley&rft.au=Li%2C+Wuchen&rft.date=2025-09-01&rft.pub=Springer+Nature+B.V&rft.issn=2522-0144&rft.eissn=2197-9847&rft.volume=12&rft.issue=3&rft.spage=42&rft_id=info:doi/10.1007%2Fs40687-025-00531-9&rft.externalDBID=NO_FULL_TEXT
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=2522-0144&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=2522-0144&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=2522-0144&client=summon