pith. sign in

arxiv: 2606.06193 · v1 · pith:TJJCGFXXnew · submitted 2026-06-04 · 🧮 math.PR · math.OC

Peng's Maximum Principle for McKean-Vlasov Stochastic Differential Equations with Common Noise

Johan Benedikt Spille , Wilhelm Stannat This is my paper

Pith reviewed 2026-06-28 00:06 UTC · model grok-4.3

classification 🧮 math.PR math.OC
keywords McKean-Vlasov SDEscommon noisestochastic maximum principlePeng's principleadjoint processesconditional backward SDEsmean-field optimal control
0
0 comments X

The pith

A third adjoint state from a conditional McKean-Vlasov backward SDE is required for the maximum principle in McKean-Vlasov SDEs with common noise.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper derives a stochastic maximum principle of Peng type for optimal control of McKean-Vlasov SDEs whose dynamics depend on the conditional law of the state process. Unlike the standard McKean-Vlasov setting, the common noise case demands a third adjoint process to account for all second-order Lions derivatives in the Taylor expansion of the cost functional. This adjoint is defined via a conditional McKean-Vlasov backward SDE, and the three adjoints together achieve a full linearization that includes cross terms between conditionally independent copies. The work also establishes well-posedness of these conditional backward equations. Readers interested in mean-field stochastic control with shared noise factors would see this as a necessary extension of existing necessary conditions for optimality.

Core claim

The central claim is that the maximum principle for the common noise case contains a third adjoint state needed to dualize all second-order Lions derivatives in the Taylor expansion of the cost functional, with the additional adjoint state given by a conditional McKean-Vlasov backward SDE; all three adjoint states allow for a complete linearization of all contributions in the second-order expansion, including interactions between conditionally independent copies of the first variational process.

What carries the argument

The third adjoint state defined by a conditional McKean-Vlasov backward SDE that dualizes second-order terms involving the conditional law and interactions between copies.

If this is right

  • The maximum principle holds without convexity assumptions on the control domain.
  • The three adjoint processes together linearize all second-order contributions including cross-interactions.
  • Well-posedness is established for the conditional McKean-Vlasov backward SDEs used in the derivation.
  • The result applies specifically to dynamics that depend on the conditional law given the common noise.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The necessity of the third adjoint may indicate that common noise creates additional quadratic variation terms not captured by standard adjoints.
  • Similar conditional backward equations could appear in other control problems with partial information or filtering.
  • One testable extension is to derive explicit solutions for linear-quadratic examples to verify the form of the third adjoint.
  • The framework might connect to mean-field games where agents share a common noise source.

Load-bearing premise

The conditional McKean-Vlasov backward SDEs supplying the third adjoint process are well-posed under suitable coefficient conditions.

What would settle it

A concrete counterexample consisting of Lipschitz coefficients for which the conditional McKean-Vlasov backward SDE fails to have a unique solution would falsify the applicability of the derived maximum principle.

read the original abstract

We study a stochastic optimal control problem for McKean-Vlasov stochastic differential equations (SDEs) with common noise, where the dynamics depend on the conditional law of the state. We derive a stochastic maximum principle of Peng type without imposing convexity assumptions on the control domain. In comparison to the standard McKean-Vlasov case, the maximum principle for the common noise case contains a third adjoint state, which is needed to dualize all second-order Lions derivatives in the Taylor expansion of the cost functional. The additional adjoint state is given by a conditional McKean-Vlasov backward SDE. All three adjoint states together allow for a complete linearization of all contributions in the second-order expansion, including interactions between conditionally independent copies of the first variational process. As part of our analysis, we also prove a general well-posedness result for conditional McKean-Vlasov backward SDEs.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The paper derives a Peng-type stochastic maximum principle for optimal control of McKean-Vlasov SDEs driven by both idiosyncratic and common noise, without convexity assumptions on the control set. The key novelty is a third adjoint process, defined via a conditional McKean-Vlasov backward SDE, that dualizes the second-order Lions derivatives arising in the Taylor expansion of the cost; together with the usual first- and second-order adjoints this yields a complete first-order condition. As a supporting result the authors establish well-posedness for a class of conditional McKean-Vlasov BSDEs under standard Lipschitz and growth hypotheses.

Significance. The result supplies the first rigorous maximum principle that fully accounts for the interaction between conditionally independent copies and the common-noise filtration in the mean-field setting. The accompanying well-posedness theorem for conditional McKean-Vlasov BSDEs is of independent interest and removes a technical obstacle that had previously limited the scope of such control problems. If the derivation is free of hidden regularity gaps, the work materially extends the Peng framework to a practically relevant class of models.

minor comments (3)
  1. §2.2, Definition 2.4: the notion of “conditional McKean-Vlasov BSDE” is introduced via a fixed-point argument; a short remark clarifying why the conditional expectation is taken with respect to the common-noise filtration (rather than the full filtration) would help readers unfamiliar with the common-noise literature.
  2. Theorem 3.1 (maximum principle): the statement lists the three adjoint processes but does not explicitly record the terminal conditions for the third adjoint; adding one line would make the theorem self-contained.
  3. §4.3, proof of well-posedness: the a-priori estimate (4.12) is derived under a uniform Lipschitz constant; it would be useful to note whether the same constant works uniformly in the conditional law or whether an additional smallness condition on the time horizon is tacitly used.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the careful reading of the manuscript and the positive recommendation to accept. The summary accurately captures the role of the third adjoint process in handling the second-order Lions derivatives under common noise, as well as the supporting well-posedness result for conditional McKean-Vlasov BSDEs.

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The derivation introduces a third adjoint process via a conditional McKean-Vlasov BSDE and explicitly proves the required well-posedness result internally as part of the analysis. No load-bearing step reduces by construction to a fitted input, self-citation chain, or renamed ansatz; the central claim remains independent of the paper's own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The result rests on standard well-posedness assumptions for SDEs and BSDEs that are not enumerated in the abstract; no free parameters or new entities are introduced.

axioms (2)
  • domain assumption Existence and uniqueness for the underlying McKean-Vlasov SDE and its variational processes under suitable Lipschitz and growth conditions.
    Implicit prerequisite for the Taylor expansion and adjoint construction in stochastic control.
  • domain assumption Well-posedness of the conditional McKean-Vlasov backward SDE that defines the third adjoint.
    Stated as proved in the paper but required for the maximum principle to be usable.

pith-pipeline@v0.9.1-grok · 5688 in / 1256 out tokens · 19983 ms · 2026-06-28T00:06:21.392168+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references · 23 canonical work pages

  1. [1]

    A maximum principle for SDEs of mean-field type.Appl

    Daniel Andersson and Boualem Djehiche. A maximum principle for SDEs of mean-field type.Appl. Math. Optim., 63(3):341–356, 2011. ISSN 0095-4616,1432-0606. doi: 10.1007/s00245-010-9123-8. URL https://doi.org/10.1007/s00245-010-9123-8

    work page doi:10.1007/s00245-010-9123-8 2011

  2. [2]

    An introductory approach to duality in optimal stochastic control.SIAM Rev., 20(1):62–78, 1978

    Jean-Michel Bismut. An introductory approach to duality in optimal stochastic control.SIAM Rev., 20(1):62–78, 1978. ISSN 1095-7200. doi: 10.1137/1020004. URLhttps://doi.org/10.1137/1020004

    work page doi:10.1137/1020004 1978

  3. [3]

    Centralized systemic risk control in the interbank system: weak formulation and Gamma-convergence.Stochastic Process

    Lijun Bo, Tongqing Li, and Xiang Yu. Centralized systemic risk control in the interbank system: weak formulation and Gamma-convergence.Stochastic Process. Appl., 150:622–654, 2022. ISSN 0304- 4149,1879-209X. doi: 10.1016/j.spa.2022.05.005. URLhttps://doi.org/10.1016/j.spa.2022.05. 005. 28

    work page doi:10.1016/j.spa.2022.05.005 2022

  4. [4]

    Extended mean-field control problems with pois- sonian common noise: Stochastic maximum principle and hamiltonian-jacobi-bellman equation, 2026

    Lijun Bo, Jingfei Wang, Xiaoli Wei, and Xiang Yu. Extended mean-field control problems with pois- sonian common noise: Stochastic maximum principle and hamiltonian-jacobi-bellman equation, 2026. URLhttps://arxiv.org/abs/2407.05356

    Pith/arXiv arXiv 2026

  5. [5]

    A general stochastic maximum principle for SDEs of mean-field type.Appl

    Rainer Buckdahn, Boualem Djehiche, and Juan Li. A general stochastic maximum principle for SDEs of mean-field type.Appl. Math. Optim., 64(2):197–216, 2011. ISSN 0095-4616,1432-0606. doi: 10.1007/ s00245-011-9136-y. URLhttps://doi.org/10.1007/s00245-011-9136-y

    work page doi:10.1007/s00245-011-9136-y 2011

  6. [6]

    A stochastic maximum principle for general mean-field systems

    Rainer Buckdahn, Juan Li, and Jin Ma. A stochastic maximum principle for general mean-field systems. Appl. Math. Optim., 74(3):507–534, 2016. ISSN 0095-4616,1432-0606. doi: 10.1007/s00245-016-9394-9. URLhttps://doi.org/10.1007/s00245-016-9394-9

    work page doi:10.1007/s00245-016-9394-9 2016

  7. [7]

    A mean-field stochastic control problem with partial ob- servations.Ann

    Rainer Buckdahn, Juan Li, and Jin Ma. A mean-field stochastic control problem with partial ob- servations.Ann. Appl. Probab., 27(5):3201–3245, 2017. ISSN 1050-5164,2168-8737. doi: 10.1214/ 17-AAP1280. URLhttps://doi.org/10.1214/17-AAP1280

    work page doi:10.1214/17-aap1280 2017

  8. [8]

    Mean-field stochastic differential equa- tions and associated PDEs.Ann

    Rainer Buckdahn, Juan Li, Shige Peng, and Catherine Rainer. Mean-field stochastic differential equa- tions and associated PDEs.Ann. Probab., 45(2):824–878, 2017. ISSN 0091-1798,2168-894X. doi: 10.1214/15-AOP1076. URLhttps://doi.org/10.1214/15-AOP1076

    work page doi:10.1214/15-aop1076 2017

  9. [9]

    Forward-backward stochastic differential equations and con- trolled McKean-Vlasov dynamics.Ann

    Ren´ e Carmona and Fran¸cois Delarue. Forward-backward stochastic differential equations and con- trolled McKean-Vlasov dynamics.Ann. Probab., 43(5):2647–2700, 2015. ISSN 0091-1798,2168-894X. doi: 10.1214/14-AOP946. URLhttps://doi.org/10.1214/14-AOP946

    work page doi:10.1214/14-aop946 2015

  10. [10]

    I, volume 83 ofProbability Theory and Stochastic Modelling

    Ren´ e Carmona and Fran¸cois Delarue.Probabilistic theory of mean field games with applications. I, volume 83 ofProbability Theory and Stochastic Modelling. Springer, Cham, 2018. ISBN 978-3-319- 56437-1; 978-3-319-58920-6. Mean field FBSDEs, control, and games

    2018

  11. [11]

    II, volume 84 ofProbability Theory and Stochastic Modelling

    Ren´ e Carmona and Fran¸cois Delarue.Probabilistic theory of mean field games with applications. II, volume 84 ofProbability Theory and Stochastic Modelling. Springer, Cham, 2018. ISBN 978-3-319- 56435-7; 978-3-319-56436-4. Mean field games with common noise and master equations

    2018

  12. [12]

    A probabilistic approach to mean field games with major and minor players.Ann

    Ren´ e Carmona and Xiuneng Zhu. A probabilistic approach to mean field games with major and minor players.Ann. Appl. Probab., 26(3):1535–1580, 2016. ISSN 1050-5164,2168-8737. doi: 10.1214/ 15-AAP1125. URLhttps://doi.org/10.1214/15-AAP1125

    work page doi:10.1214/15-aap1125 2016

  13. [13]

    Control of McKean-Vlasov dynamics versus mean field games.Math

    Ren´ e Carmona, Fran¸cois Delarue, and Aim´ e Lachapelle. Control of McKean-Vlasov dynamics versus mean field games.Math. Financ. Econ., 7(2):131–166, 2013. ISSN 1862-9679,1862-9660. doi: 10.1007/ s11579-012-0089-y. URLhttps://doi.org/10.1007/s11579-012-0089-y

    work page doi:10.1007/s11579-012-0089-y 2013

  14. [14]

    Mean field games with common noise.Ann

    Ren´ e Carmona, Fran¸cois Delarue, and Daniel Lacker. Mean field games with common noise.Ann. Probab., 44(6):3740–3803, 2016. ISSN 0091-1798,2168-894X. doi: 10.1214/15-AOP1060. URLhttps: //doi.org/10.1214/15-AOP1060

    work page doi:10.1214/15-aop1060 2016

  15. [15]

    Conditional McKean–Vlasov control, 2025

    Ren´ e Carmona, Ludovic Tangpi, and Kaiwen Zhang. Conditional McKean–Vlasov control, 2025. URL https://arxiv.org/abs/2510.06543

    arXiv 2025

  16. [16]

    Transposition approach to optimal control of McKean–Vlasov spdes, 2026

    Liangying Chen and Wilhelm Stannat. Transposition approach to optimal control of McKean–Vlasov spdes, 2026. URLhttps://arxiv.org/abs/2603.06245

    arXiv 2026

  17. [17]

    McKean-Vlasov optimal control: the dynamic programming principle.Ann

    Mao Fabrice Djete, Dylan Possama¨ ı, and Xiaolu Tan. McKean-Vlasov optimal control: the dynamic programming principle.Ann. Probab., 50(2):791–833, 2022. ISSN 0091-1798,2168-894X. doi: 10.1214/ 21-aop1548. URLhttps://doi.org/10.1214/21-aop1548

    work page doi:10.1214/21-aop1548 2022

  18. [18]

    McKean-Vlasov optimal control: limit theory and equivalence between different formulations.Math

    Mao Fabrice Djete, Dylan Possama¨ ı, and Xiaolu Tan. McKean-Vlasov optimal control: limit theory and equivalence between different formulations.Math. Oper. Res., 47(4):2891–2930, 2022. ISSN 0364- 765X,1526-5471. doi: 10.1287/moor.2021.1232. URLhttps://doi.org/10.1287/moor.2021.1232. 29

    work page doi:10.1287/moor.2021.1232 2022

  19. [19]

    Peng’s maximum principle for stochastic delay differential equations of mean-field type, 2025

    Giuseppina Guatteri, Federica Masiero, and Lukas Wessels. Peng’s maximum principle for stochastic delay differential equations of mean-field type, 2025. URLhttps://arxiv.org/abs/2512.00934

    arXiv 2025

  20. [20]

    The Connected-Component Labeling Problem: A Review of State-of-the-Art Algorithms

    Ben Hambly and Philipp Jettkant. Optimal control of the nonlinear stochastic Fokker-Planck equation. Stochastic Process. Appl., 191:Paper No. 104774, 35, 2026. ISSN 0304-4149,1879-209X. doi: 10.1016/j. spa.2025.104774. URLhttps://doi.org/10.1016/j.spa.2025.104774

    work page doi:10.1016/j 2026

  21. [21]

    Optimal control of mean field equations with mono- tone coefficients and applications in neuroscience.Appl

    Antoine Hocquet and Alexander Vogler. Optimal control of mean field equations with mono- tone coefficients and applications in neuroscience.Appl. Math. Optim., 84:S1925–S1968, 2021. ISSN 0095-4616,1432-0606. doi: 10.1007/s00245-021-09816-1. URLhttps://doi.org/10.1007/ s00245-021-09816-1

    work page doi:10.1007/s00245-021-09816-1 2021

  22. [22]

    Huang, R

    Minyi Huang, Roland P. Malham´ e, and Peter E. Caines. Large population stochastic dynamic games: closed-loop McKean-Vlasov systems and the Nash certainty equivalence principle.Commun. Inf. Syst., 6(3):221–251, 2006. ISSN 1526-7555,2163-4548. doi: 10.4310/cis.2006.v6.n3.a5. URLhttps://doi. org/10.4310/cis.2006.v6.n3.a5

    work page doi:10.4310/cis.2006.v6.n3.a5 2006

  23. [23]

    M. Kac. Foundations of kinetic theory. InProceedings of the Third Berkeley Symposium on Mathematical Statistics and Probability, 1954–1955, vol. III, pages 171–197. Univ. California Press, Berkeley-Los Angeles, Calif., 1956

    1954

  24. [24]

    Mean field games.Jpn

    Jean-Michel Lasry and Pierre-Louis Lions. Mean field games.Jpn. J. Math., 2(1):229–260, 2007. ISSN 0289-2316,1861-3624. doi: 10.1007/s11537-007-0657-8. URLhttps://doi.org/10.1007/ s11537-007-0657-8

    work page doi:10.1007/s11537-007-0657-8 2007

  25. [25]

    H. P. McKean, Jr. Propagation of chaos for a class of non-linear parabolic equations. InStochastic Differential Equations (Lecture Series in Differential Equations, Session 7, Catholic Univ., 1967), vol- ume Session 7 ofLecture Series in Differential Equations, pages 41–57. Air Force Office of Scientific Research, Office of Aerospace Research, United Stat...

    1967

  26. [26]

    Etienne Pardoux and Aurel R˘ a¸ scanu.Backward Stochastic Differential Equations, pages 353–

  27. [27]

    ISBN 978-3-319-05714-9

    Springer International Publishing, Cham, 2014. ISBN 978-3-319-05714-9. doi: 10.1007/ 978-3-319-05714-9 5

    2014

  28. [28]

    A general stochastic maximum principle for optimal control problems.SIAM J

    Shi Ge Peng. A general stochastic maximum principle for optimal control problems.SIAM J. Control Optim., 28(4):966–979, 1990. ISSN 0363-0129. doi: 10.1137/0328054. URLhttps://doi.org/10. 1137/0328054

    work page doi:10.1137/0328054 1990

  29. [29]

    Linear quadratic optimal control of conditional McKean-Vlasov equation with ran- dom coefficients and applications.Probab

    Huyˆ en Pham. Linear quadratic optimal control of conditional McKean-Vlasov equation with ran- dom coefficients and applications.Probab. Uncertain. Quant. Risk, 1:Paper No. 7, 26, 2016. ISSN 2095-9672,2367-0126. doi: 10.1186/s41546-016-0008-x. URLhttps://doi.org/10.1186/ s41546-016-0008-x

    work page doi:10.1186/s41546-016-0008-x 2016

  30. [30]

    Dynamic programming for optimal control of stochastic McKean-Vlasov dynamics.SIAM J

    Huyˆ en Pham and Xiaoli Wei. Dynamic programming for optimal control of stochastic McKean-Vlasov dynamics.SIAM J. Control Optim., 55(2):1069–1101, 2017. ISSN 0363-0129,1095-7138. doi: 10.1137/ 16M1071390. URLhttps://doi.org/10.1137/16M1071390

    work page doi:10.1137/16m1071390 2017

  31. [31]

    L. S. Pontryagin, V. G. Boltyanskii, R. V. Gamkrelidze, and E. F. Mishchenko.The mathematical theory of optimal processes. Interscience Publishers John Wiley & Sons, Inc., New York-London, 1962

    1962

  32. [32]

    A novel approach to Peng’s maximum principle for McKean–Vlasov stochastic differential equations, 2026

    Johan Benedikt Spille and Wilhelm Stannat. A novel approach to Peng’s maximum principle for McKean–Vlasov stochastic differential equations, 2026. URLhttps://arxiv.org/abs/2602.12006

    arXiv 2026

  33. [33]

    Sznitman,Topics in propagation of chaos, in École d’Été de Probabilités de Saint-Flour XIX—1989, vol

    Alain-Sol Sznitman. Topics in propagation of chaos. In ´Ecole d’ ´Et´ e de Probabilit´ es de Saint-Flour XIX—1989, volume 1464 ofLecture Notes in Math., pages 165–251. Springer, Berlin, 1991. ISBN 3-540-53841-0. doi: 10.1007/BFb0085169. URLhttps://doi.org/10.1007/BFb0085169

    work page doi:10.1007/bfb0085169 1989

  34. [34]

    Springer Berlin Heidelberg, Berlin, Heidelberg,

    C´ edric Villani.The Wasserstein distances, pages 93–111. Springer Berlin Heidelberg, Berlin, Heidelberg,

  35. [35]

    Springer, Berlin, Heidelberg, 2009

    ISBN 978-3-540-71050-9. doi: 10.1007/978-3-540-71050-9 6. 30

    work page doi:10.1007/978-3-540-71050-9

  36. [36]

    Springer New York, New York, NY, 1999

    Jiongmin Yong and Xun Yu Zhou.The Relationship Between the Maximum Principle and Dynamic Programming, pages 217–280. Springer New York, New York, NY, 1999. ISBN 978-1-4612-1466-3. doi: 10.1007/978-1-4612-1466-3 5. URLhttps://doi.org/10.1007/978-1-4612-1466-3_5. 31

    work page doi:10.1007/978-1-4612-1466-3 1999