Pith. sign in

REVIEW 2 major objections 5 minor 2 cited by

A particle system approach towards the global well-posedness of master equations for potential mean field games of control

T0 review · 2 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves global classical well-posedness for the Hamilton-Jacobi-Bellman and master equations of generalized mean-field control and potential mean-field games of control, under an extended displacement-convexity condition.

desk verdict Substantial and original paper on global well-posedness for generalized MFC/MFGC master equations with joint position-momentum dependence, but the main a priori estimate rests on a sign condition that does not follow from the stated assumptions, and the Nash equilibrium rate claim overreaches. read the letter →

arxiv 2412.11742 v3 pith:E6T5TRBC submitted 2024-12-16 math.OC math.APmath.PR

Huafu Liao , Chenchen Mou This is my paper · ORCID
classification math.OCmath.APmath.PR MSC 49N8049L1235Q7060H3065C35
keywords meanfieldgamesofcontrolmasterequationsglobalwell-posednessdisplacementconvexityparticlesystemspropagationchaosbackwardstochasticdifferential
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that the Hamilton-Jacobi-Bellman equation (1.4) for generalized mean-field control and the master equation (1.5) for potential mean-field games of control have unique global classical solutions with bounded derivatives, and that the same remains true when the individual noise vanishes. That matters because global classical regularity is what turns these infinite-dimensional PDEs into usable objects: it gives well-defined optimal feedback functions and makes the $N$-particle approximation quantitative. The proof goes through the $N$-particle HJB equation, proving a uniform-in-$N$ bound on its Hessian, transferring the resulting displacement convexity and semi-concavity to the mean-field value function by propagation of chaos, and then closing the local-to-global step with a priori estimates on the Wasserstein derivatives. On the way, the analysis yields a Lipschitz approximator to the optimal feedback function with algebraic rate and an approximate Markovian Nash equilibrium with $O(N^{-1})$ error.

What carries the argument

The load-bearing object is the $N$-particle value function $V_N$ together with the uniform Hessian estimate (2.6): $0 \le \sum_{i,j=1}^N \xi_i\xi_j \partial^2_{x_j x_j} V_N(t,x) \le \frac{C}{N}\sum_{i=1}^N \xi_i^2$. The lower and upper bounds express displacement convexity and semi-concavity at the particle level, and they are proved by writing the differentiated HJB system for $V_N$ as a nonlinear Feynman-Kac representation with coefficients involving the Hamiltonian and its derivatives, using Riccati-type identities to control the quadratic terms. Propagation of chaos transfers this bound to the mean-field value function $V$, making $V$ displacement convex and semi-concave; those properties imply the Lipschitz bound on $\partial_\mu V$ and the a priori bounds on $\partial_x\partial_\mu V$ and $\partial^2_{\mu\mu} V$ that drive the global extension. Higher-order derivative bounds are obtained inductively through a Feynman-Kac representation for linear master equations, and the local solution generated by the mean-field FBSDE is extended by restarting it at successive short-time intervals whose length is controlled by the data-dependent constant (4.17).

What would settle it

Take a smooth displacement-convex Hamiltonian satisfying Assumption 2.1(2)--(3) and compute the mixed derivative $\partial_p \partial_{\tilde\mu} H^{(p)}$ on a two-point empirical measure; if it can be positive while (2.4) holds, then the Riccati step of Lemma 5.11 has a missing hypothesis. A direct numerical check would then solve the $N$-particle HJB equation (1.1) for $N=2,3$ and test whether the bound (2.6) continues to hold uniformly in $N$; a violation in the absence of that momentum-concavity condition would falsify the proof's mechanism.

Watch

Extended reading notes

Core claim

Under the extended displacement-convexity condition in Assumption 2.1 and the structural consistency condition in Assumption 2.5, the paper proves that the HJB equation (1.4) for generalized MFC and the master equation (1.5) for potential MFGC each admit a unique classical solution on the full time interval $[0,T]$ whose derivatives are bounded by constants depending only on the data. The same conclusion holds for the degenerate case $\sigma=0$, obtained by a limiting argument because all a priori bounds are uniform in $\sigma>0$. The solution is first generated locally as the decoupling field of a mean-field forward-backward SDE, and the global extension is made possible by the a priori estimates derived from the $N$-particle system, including higher-order estimates bootstrapped through a Feynman-Kac representation of linear master equations. Along the way, the paper establishes that the $N$-particle value function produces an $O(N^{-1/2}+N^{-1})$ Lipschitz approximator to the optimal feedback map $\partial_\mu V$, and that the feedback policy generated by $V_N$ forms an approximate Markovian Nash equilibrium for the $N$-player game with error $O(N^{-1})$.

Load-bearing premise

Everything rests on the assumption that the Hamiltonian is concave enough in the momentum variable to make its second-derivative matrix in momentum non-positive along the particle dynamics; without that, the uniform curvature bound on the $N$-particle value function, and the global well-posedness built on it, does not follow.

Editorial extensions

If this is right

  • The HJB and master equations for generalized MFC and potential MFGC are globally well-posed, so the optimal feedback function exists for the whole time horizon rather than only near the terminal time.
  • The estimates are uniform in the individual-noise parameter $\sigma>0$, so the well-posedness survives the degenerate limit $\sigma=0$.
  • The $N$-particle value function gives a Lipschitz approximation to the optimal feedback map with algebraic rate $O(N^{-1/2}+N^{-1})$ in Wasserstein distance.
  • The feedback strategies obtained from $V_N$ form an approximate Markovian Nash equilibrium with error $O(N^{-1})$ when the initial data are i.i.d., with explicit dependence on the spread of initial positions.
  • The results cover Hamiltonians that depend on the joint distribution of position and momentum, which includes extended mean-field control and centralized-control particle systems that do not fit the additive form (1.6)--(1.7).

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: the restriction to $P_2(\mathbb{R})$ is made for notation, so the same particle-system route should extend the global well-posedness theorem to $P_2(\mathbb{R}^d)$; a direct test is to rerun the Hessian estimate (2.6) with the $\mathbb{R}^d$-valued empirical measure.
  • Beyond the paper: the proof isolates the momentum-concavity of the Hamiltonian as the quantity controlling global regularity; if a milder damping mechanism could replace the non-positivity of the $p$-Hessian, the class of displacement-convex models covered by the theorem would widen.
  • Beyond the paper: the $O(N^{-1})$ approximate-equilibrium result suggests a numerical recipe -- solve the finite-$N$ HJB equation and use its feedback map as a surrogate equilibrium -- whose error is comparable to the classical Nash-system limit, a comparison the paper leaves to Remark 7.5.
  • Beyond the paper: because all a priori constants are independent of $\sigma$, the degenerate solution could be approached by simulating small-noise particle systems, giving a stochastic numerical scheme for deterministic mean-field games of control.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper develops a global well-posedness theory for the HJB equation (1.4) of generalized mean field control and the master equation (1.5) of potential mean field games of control, under a displacement convexity-type assumption on the generalized Hamiltonian H(μ̃) defined on joint position-momentum distributions. The strategy is to first solve a mean field FBSDE locally in time (Lemma 4.6), then derive uniform-in-N Hessian estimates for the N-particle HJB equation (1.1), use propagation of chaos to transfer these to displacement convexity/semi-concavity of V, obtain a priori bounds on ∂_μV and ∂^2_{μμ}V, and finally bootstrap the short-time solution to a global one. Consequences include global classical solutions in the nondegenerate and degenerate cases, rates for a Lipschitz approximation of ∂_μV, and an approximate Markovian Nash equilibrium. The central claim is that, under Assumptions 2.1 and 2.5, the relevant equations have unique global classical solutions with bounds uniform in σ, and that these results extend to σ = 0 (Theorems 2.3, 2.4, 2.7, 2.8, 2.9).

Significance. If the results are correct, the paper would provide the first global classical well-posedness results for HJB/master equations of generalized MFC and potential MFGC that involve the joint distribution of position and momentum. The approach is attractive: it leverages classical finite-dimensional PDE estimates for the N-particle system and propagates them to the mean field limit, yielding not only existence but explicit Lipschitz approximators and quantified Nash equilibrium errors. The paper is ambitious and technically rich, and the applications in Section 7—Lipschitz approximation of the optimal feedback function with rate O(N^{-1/2}+N^{-1}) and an approximate Markovian Nash equilibrium with error O(N^{-1})—are valuable if the underlying well-posedness is established. However, the proof currently has a load-bearing sign mismatch in the key Hamiltonian conditions, and one of the central local well-posedness lemmas is stated without proof; both points must be addressed before the main claims can be considered verified.

major comments (2)
  1. [Assumption 2.1(3) and Lemmas 5.4, 5.11] Equation (2.4) in Assumption 2.1(3) is written as a condition on ∂_x∂_μ̃ H^{(p)} in its local term, but the proofs of Lemma 5.4 and Lemma 5.11 require the sign of ∂_p∂_μ̃ H^{(p)}. Specifically, the matrix H^pp_N defined in (5.15) contains the term δ_{kl} N ∂_p∂_μ̃ H^{(p)}, and the assertion in (5.16) that H^pp_N ≤ 0 “according to Assumption 2.1” does not follow from (2.4) as written. Likewise, Lemma 5.11 states “In view of Lemma 2.2, ∂p∂˜µ H(p)(...) ≤ 0”, but Lemma 2.2 only yields −∂_x∂_μ̃ H^{(p)} ≥ 0 and the cross-derivative identity (2.5); it says nothing about ∂_p∂_μ̃ H^{(p)}. The upper bound in Lemma 5.4 drops the term Y_N H^pp_N Y_N in (5.17), which is only valid if H^pp_N ≤ 0, and the Riccati equation in Lemma 5.11 needs the same sign for boundedness of ˆP^s_x. Since the uniform estimate (2.6) and the subsequent global well-posedness theorems build directly on these lemmas, the central derivation is not valid as written. If (2.4) is intended to encode displacement convexity in the p-marginal, the local term should presumably be ∂_p∂_μ̃ H^{(p)} instead of ∂_x∂_μ̃ H^{(p)}; otherwise, an additional argument is needed to justify the required signs.
  2. [Lemma 4.6 (Section 4)] Lemma 4.6 is the sole source of local well-posedness and C^4 regularity for the mean field FBSDE system (4.16), and it is used as the starting point of the bootstrap in Theorem 2.4 and Theorem 2.7. Its proof is omitted with the sentence “Since the proof is basically the same as those in [19, 22, 25, 59], we omit it here.” This is not a routine extension: the present FBSDE involves the joint law of (X_s,Y_s), the conditional law given common noise, and the additional scalar component V^µ_s whose recovery via (4.1) and Lemma 4.2 is essential for the identification ∂_μV = Y. It is not immediate that the standard contraction arguments apply directly to this coupled system without further assumptions. Given that the global results depend critically on this lemma, the authors should either include a complete proof or provide a precise reference with a statement whose hypotheses are verified here.
minor comments (5)
  1. [Throughout] There are numerous typographical errors: “well-posedne ss” in the abstract, “+ +Z” in the FBSDE on page 13, “X^{ε,µns}” near (4.6), and “Proposition 5.1” in the proof of Lemma 5.5 where Lemma 5.1 is meant.
  2. [Lemma 2.2] The proof of Lemma 2.2 uses φ(x) = δ_{x_1}(x), but test functions are required to be smooth; this is a standard approximation argument, but it would be clearer to spell it out.
  3. [Section 5.4.1] Theorem 5.15 is a “modification of Theorem 7.2 of [11]” but the proof is only sketched. The modification is nontrivial because the coefficients in (5.36) depend on the conditional law given common noise and the master equation (5.38) involves additional cross terms. A more detailed verification of the hypotheses, or a statement of the exact modification, would increase confidence.
  4. [Proposition 7.4 proof] The notation for the k-th player’s cost in the proof of Proposition 7.4 is inconsistent: J^k_N is defined in (7.9), but later the text writes J̄^k_N (e.g., line after (7.10) and in (7.13)) without introducing the bar notation. Using one symbol consistently would avoid confusion.
  5. [Theorem 2.8 proof] The proof of Theorem 2.8 refers to “Proof Theorem 2.8” and uses the Arzelà–Ascoli argument with subsequences; the uniqueness of the limit is shown, but the argument would be easier to follow if the extraction of the convergent subsequence were written more explicitly.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the local-to-global bootstrap is self-contained; self-citations ([54], [58], [59]) serve as technique/completeness references alongside external anchors ([19], [22], [25], [11], [46]), and the flagged Assumption 2.1(3)/(2.4) mismatch is a correctness risk, not circularity.

full rationale

The paper's derivation chain is a standard local-to-global bootstrap and contains no fitted parameters, no empirical predictions, and no quantity defined in terms of the target result. Local well-posedness (Lemma 4.6) is generated independently by contraction/FBSDE methods; although the proof is omitted, it is anchored to the external references [19], [22], [25] (Carmona–Delarue; Chassagneux–Crisan–Delarue), with the authors' own [59] cited only alongside them, so it does not reduce to an unverified self-citation. The uniform N-particle estimates (Theorem 2.3, via Lemmas 5.4–5.5) are proven in-paper from the stated Assumption 2.1 using the Feynman–Kac representation (5.14) and standard PDE regularity ([46]); the appeal to "a contraction method similar to that in Proposition 3.10 of [54]" calls on the authors' prior work as a proof technique, a completeness burden rather than a load-bearing premise, since the regularized system (5.20) and the bootstrap are written out. The transfer of convexity/semi-concavity from V_N to V (Lemmas 5.6–5.7) uses the in-paper propagation-of-chaos comparison (Proposition 3.1), which assumes only local classical solutions as granted by Lemma 4.6, so the continuation argument is not circular. Second- and higher-order estimates (Theorem 5.12, Proposition 5.17) are in-paper consequences of the previous steps, and the linear master-equation Feynman–Kac ingredient (Theorem 5.15) is a modification of the external [11] (Buckdahn–Li–Peng–Rainer). The approximation claims of Section 7 (Lipschitz approximator ∂μVN with the O(N^{-1/2}+N^{-1}) rate, and the O(N^{-1}) approximate Markovian Nash equilibrium) are derived quantitatively from the established estimates, not assumed. The same-authors citations [2], [37], [38], [57], [58] are contextual or set structural assumptions; no uniqueness theorem is imported from the authors' prior work to forbid alternatives. A genuine concern flagged by the paper's own Lemmas 5.4 and 5.11 is that Assumption 2.1(3), inequality (2.4), controls ∂x∂μH^(p) (with Lemma 2.2 giving the sign of −∂x∂μH^(p)), whereas those lemmas need ∂p∂μH^(p) ≤ 0; this is a potential assumption-mismatch and a correctness risk, but it is not circularity. Hence the central claim is self-contained against external benchmarks; no step reduces to its inputs by construction.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

The paper introduces no new physical entities or fitted parameters. It relies on smoothness and convexity-type assumptions on U and H, on a structural constraint linking the control and game data, and on several substantial analytic results (local FBSDE well-posedness and a Feynman-Kac representation for linear master equations) that are imported with proof omitted or sketched. The most fragile item is the printed form of Assumption 2.1(3), which appears mismatched with the sign conditions used in the main estimates.

assumptions (8)
  • domain assumption Smoothness and boundedness of U in C^6(P2(R)) and H in C^6(P2(R^2)) with Lipschitz bounded derivatives (Assumption 2.1(1)).
    Main regularity source for all estimates; invoked throughout Sections 4 to 6.
  • domain assumption Displacement convexity of U and H with respect to the x-marginal, (2.2) and (2.3).
    Used to obtain U_N(T) >= 0 and H^xx_N >= 0 in Lemma 5.4, and later to infer displacement convexity of V in Lemma 5.7.
  • domain assumption Displacement concavity of -H with respect to the p-marginal, (2.4).
    Intended to give H^pp_N <= 0 in Lemma 5.4 and the Riccati sign in Lemma 5.11. As written it controls d_x d_mu H^(p), not the d_p d_mu H^(p) that the proofs appear to use. This is the weakest load-bearing assumption.
  • domain assumption Structural constraints relating H, H-bar, G and U (Assumption 2.5, equations (2.8) and (2.9)).
    Needed to pass from the MFC HJB equation to the potential MFGC master equation in Section 6.2.
  • domain assumption Smoothness Property 5.13 and Assumption 5.14 for the Feynman-Kac linear master equation.
    Used in Theorem 5.15 to propagate higher-order derivative estimates in Section 5.4.
  • ad hoc to paper Lemma 4.6: local well-posedness of the mean field FBSDE by a contraction method.
    Stated without proof, with reference to [19,22,25,59]; it is the starting point for every extension step in the bootstrap.
  • ad hoc to paper Theorem 5.15: modification of Theorem 7.2 of [11] giving a Feynman-Kac representation for linear master equations.
    Only a proof sketch is provided, yet the higher-order a priori estimates in Section 5.4 rest on it.
  • standard math Fenchel biconjugation for H_N and the Lagrangian L_N (Section 3).
    Converts the Hamiltonian into a Lagrangian and justifies the variational representations used in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A particle system approach towards the global well-posedness of master equations for potential mean field games of control." pith.science (2026). https://pith.science/paper/E6T5TRBC

@misc{pith2026241211742,
  author       = {Pith},
  title        = {Pith review of: A particle system approach towards the global well-posedness of master equations for potential mean field games of control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E6T5TRBC}},
  note         = {Machine review of arXiv:2412.11742}
}
abstract

This paper studies the $N$-particle systems as well as the HJB/master equations for a class of generalized mean field control (MFC) problems and the corresponding potential mean field games of control (MFGC). A local in time classical solution for the HJB equation is generated via a probabilistic approach based on the mean field maximum principle. Given an extension of the so called displacement convexity condition, we obtain the uniform estimates on the HJB equation for the $N$-particle system. Such estimates imply the displacement convexity/semi-concavity and thus the prior estimates on the solution to the HJB equation for generalized MFC problems. The global well-posedness of HJB/master equation for generalized MFC/potential MFGC is then proved thanks to the local well-posedness and the prior estimates. In view of the nature of the displacement convexity condition, such well-posedness is also true for the degenerated case. Our analysis on the $N$-particle system also induces an Lipschitz approximator to the optimal feedback function in generalized MFC/potential MFGC where an algebraic convergence rate is obtained. Furthermore, an alternative approximate Nash equilibrium is proposed based on the $N$-particle system, where the approximation error is quantified thanks to the aforementioned uniform estimates.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Quantitative convergence for displacement monotone Mean Field Games of control

    math.PR 2025-07 conditional novelty 8.0 of 10

    Open-loop and closed-loop Nash equilibria of N-player Mean Field Games of Controls converge to the mean field equilibrium at explicit rates under displacement monotonicity, without assuming separable Hamiltonians.

  2. Robust mean field control: an application to optimal execution under composite uncertainty

    math.OC 2026-07 conditional novelty 6.0 of 10

    Proves well-posedness of HJBI equations for mean-field optimal execution with composite (stochastic + deterministic) uncertainty, and derives an explicit multi-dimensional LQ solution with strict liquidation constraint.

Reference graph

Works this paper leans on

63 extracted references · 58 canonical work pages · cited by 2 Pith papers

  1. [58]

    Mou and J

    C. Mou and J. Zhang. Mean field games of controls: Propaga tion of monotonicities. Probab. Uncertain. Quant. Risk , 7(3):247–274, 2022

    work page 2022

  2. [37]

    Gangbo and A.R

    W. Gangbo and A.R. Mészáros. Global well-posedness of m aster equations for deterministic displace- ment convex potential mean field games. Comm. Pure Appl. Math. , 75:2685–2801, 2022

    work page 2022

  3. [1]

    Acciaio, J

    B. Acciaio, J. Backhoff-Veraguas, and R. Carmona. Extend ed mean field control problems: stochastic maximum principle and transport perspective. SIAM J. Control Optim. , 57(6):3666–3693, 2019

    work page 2019

  4. [2]

    Bansil, A.R

    M. Bansil, A.R. Mészáros, and C. Mou. Global well-posedn ess of displacement monotone degenerate mean field games master equations. SIAM J. Control Optim. , to appear, 2025+

    work page 2025

  5. [3]

    Bayraktar, I

    E. Bayraktar, I. Ekren, and X. Zhang. Convergence rate of particle system for second-order PDEs on Wasserstein space. arXiv:2408.06013, 2024

    arXiv 2024

  6. [4]

    Bensoussan, P.J

    A. Bensoussan, P.J. Graber, and S.C.P. Yam. Stochastic c ontrol on space of random variables. arXiv:1903.1260

    work page arXiv 1903

  7. [5]

    Mean Field Type Control Problems, Some Hilbert-space-valued FBSDEs, and Related Equations

    A. Bensoussan, H.M. Tai, and S.C.P. Yam. Mean field type co ntrol problems, some Hilbert-space-valued FBSDEs and related equations. arXiv:2305.04019v1, 2023

    work page Pith review arXiv 2023

  8. [6]

    Bensoussan and S.C.P

    A. Bensoussan and S.C.P. Yam. Control problem on space of random variables and master equation. ESAIM Control Optim. Calc. Var. , 25(Paper No. 10), 2019

    work page 2019

Show all 63 references
  1. [7]

    L. Bo, A. Capponi, and H. Liao. Large sample mean–field sto chastic optimization. SIAM J. Control Optim., 60(4):2538–2573, 2022

    2022

  2. [8]

    L. Bo, T. Li, and X. Yu. Centralized systemic risk control in the interbank system: Weak formulation and Gamma–convergence. Stochastic Process Appl., 150:622–654, 2022. 45

    2022

  3. [9]

    L. Bo, J. Wang, X. Wei, and X. Yu. Extended mean-field contr ol problems with Poissonian common noise: Stochastic maximum principle and Hamiltonian-Jaco bi-Bellman equation. arXiv:2407.05356, 2024

    2024 arXiv

  4. [10]

    Briani and P

    A. Briani and P. Cardaliaguet. Stable solutions in pote ntial mean field game systems. Nonlinear Differ. Equ. Appl. , 25:1–26, 2018

    2018

  5. [11]

    Buckdahn, J

    R. Buckdahn, J. Li, S. Peng, and C. Rainer. Mean-field sto chastic differential equations and associated pdes. Ann. Probab., 45(2):824–878, 2017

    2017

  6. [12]

    Cardaliaguet

    P. Cardaliaguet. Mean field games systems of first order. ESAIM: Contr. Optim. Cal. Variat. , 21(3):690– 722, 2015

    2015

  7. [13]

    Cardaliaguet

    P. Cardaliaguet. The convergence problem in mean field g ames with local coupling. Appl. Math. Optim. , 76:177–215, 2017

    2017

  8. [14]

    Cardaliaguet, S

    P. Cardaliaguet, S. Daudin, J. Jackson, and P. Souganid is. An algebraic convergence rate for the optimal control of Mckean–Vlasov dynamics. SIAM J. Control Optim. , 61(6):3341–3369, 2023

    2023

  9. [15]

    Cardaliaguet, F

    P. Cardaliaguet, F. Delarue, J.-M. Lasry, and P.-L. Lio ns. The Master Equation and the Convergence Problem in Mean Field Games , volume 201 of Ann. of Math. Stud. Princeton University Press, Princeton, NJ, 2019

    2019

  10. [16]

    Cardaliaguet, P.J

    P. Cardaliaguet, P.J. Graber, A. Porretta, and D. Tonon . Second order mean field games with degenerate diffusion and local coupling. Nonlinear Differ. Equ. Appl. , 22:1287–1317, 2015

    2015

  11. [17]

    Cardaliaguet and S

    P. Cardaliaguet and S. Hadikhanloo. Learning in mean fie ld games: the fictitious play. ESAIM: Contr. Optim. Cal. Variat. , 23(2):569–591, 2017

    2017

  12. [18]

    Cardaliaguet, J

    P. Cardaliaguet, J. Jackson, N. Mimikos-Stamatopoulo s, and P.E. Souganidis. Sharp convergence rates for mean field control in the region of strong regularity. arXiv:2312.11373, 2023

    2023 arXiv

  13. [19]

    Carmona and F

    R. Carmona and F. Delarue. Probabilistic Theory of Mean Field Games with Applications I , volume 83 of Probability Theory and Stochastic Modelling . Springer Cham, 2018

    2018

  14. [20]

    Carmona and F

    R. Carmona and F. Delarue. Mean field forward-backward s tochastic differential equations. Electron. Commun. Probab., (68):1–15, 2013

    2013

  15. [21]

    Carmona and F

    R. Carmona and F. Delarue. Forward–backward stochasti c differential equations and controlled Mck- ean–Vlasov dynamics. Ann. Probab., 43(5):2647–2700, 2015

    2015

  16. [22]

    Carmona and F

    R. Carmona and F. Delarue. Probabilistic Theory of Mean Field Games with Applications II , volume 84 of Probability Theory and Stochastic Modelling . Springer Cham, 2018

    2018

  17. [23]

    Carmona and M

    R. Carmona and M. Laurère. Convergence analysis of mach ine learning algorithms for the numerical solution of mean field control and games: II – the finite horizo n case. Ann. Appl. Probab., 32(6):4065–4105, 2022

    2022

  18. [24]

    Cecchin and F

    A. Cecchin and F. Delarue. Weak solutions to the master e quation of potential mean field games. arXiv:2204.04315, 2022

    2022 arXiv

  19. [25]

    Chassagneux, D

    J.-F. Chassagneux, D. Crisan, and F. Delarue. A probabi listic approach to classical solutions of the master equation for large population equilibria. Mem. Amer. Math. Soc. , 280(1379):v+123, 2022

    2022

  20. [26]

    Cosso and H

    A. Cosso and H. Pham. Zero-sum stochastic differential g ames of generalized mckean-vlasov type. J. Math. Pures Appl. , 129(9):180–212, 2019

    2019

  21. [27]

    Delarue, D

    F. Delarue, D. Lacker, and K. Ramanan. From the master eq uation to mean field game limit theory: a central limit theorem. Electron. J. Probab., 24:1–54, 2018

    2018

  22. [28]

    Delarue, D

    F. Delarue, D. Lacker, and K. Ramanan. From the master eq uation to mean field game limit theory: large deviations and concentration of measure. Ann. Probab., 48(1):211–263, 2020. 46

    2020

  23. [29]

    Daudin, F

    S. Daudin, F. Delarue, and J. Jackson. On the optimal rat e for the convergence problem in mean field control. J. Funct. Anal. , 287(12):Paper No. 110660, 94, 2024

    2024

  24. [30]

    M.F. Djete. Extended mean field control problem: a propa gation of chaos result. Electron. J. Probab., 27:1–53, 2022

    2022

  25. [31]

    Djete, D

    M.F. Djete, D. Possamaï, and X. Tan. Mckean-Vlasov opti mal control: the dynamic programming principle. Ann. Prob., 50(2):791–833, 2022

    2022

  26. [32]

    W. E. A proposal on machine learning via dynamical syste ms. Commun. Math. Stat. , 5(1):1–11, 2017

    2017

  27. [33]

    W. E, J. Han, and Q. Li. A mean–field optimal control formu lation of deep leanring. Res. Math. Sci. , 6(10), 2019

    2019

  28. [34]

    Fleming and H.M

    W.H. Fleming and H.M. Soner. Controlled Markov Processes and Viscosity Solutions . Springer New York, NY, 2006

    2006

  29. [35]

    Fournier and A

    N. Fournier and A. Guillin. On the rate of convergence in Wasserstein distance of the empirical measure. Probab. Theory Relat. Fields , 162(3):707–738, 2015

    2015

  30. [36]

    Gangbo, S

    W. Gangbo, S. Mayorga, and A. Święch. Finite dimensiona l approximation of Hamilton–Jacobi–Bellman equations in spaces of probability measures. SIAM J. Math. Anal. , 53(2):1320–1356, 2021

    2021

  31. [38]

    Gangbo, A.R

    W. Gangbo, A.R. Mészáros, C. Mou, and J. Zhang. Mean field games master equations with nonsepa- rable Hamiltonians and displacement monotonicity. Ann. Probab., 50(6):2178–2217, 2022

    2022

  32. [39]

    Gangbo and A

    W. Gangbo and A. Święch. Existence of a solution to an equ ation arising from the theory of mean field games. J. Differential Equations , 259(11):6573–6643, 2015

    2015

  33. [40]

    Graber and A

    P.J. Graber and A. Mullenix and L. Pfeiffer. Weak solutio ns for potential mean field games of controls Nonlinear Differ. Equ. Appl. , 28: article number 50, 2021

    2021

  34. [41]

    Guo and Y

    X. Guo and Y. Zhang. Towards an analytical framework for potential games. arXiv:2310.02259, 2023

    2023 arXiv

  35. [42]

    Höfer and H.M

    F. Höfer and H.M. Soner. Potential mean-field games and g radient flows. arXiv:2408.00733v1, 2024

    2024 arXiv

  36. [43]

    Huang and M

    J. Huang and M. Huang. Robust mean field linear-quadrati c-gaussian games with unknown L2- disturbance. SIAM J. Control Optim. , 55(5):2811–2840, 2017

    2017

  37. [44]

    Huang and S.L

    M. Huang and S.L. Nguyen. Mean field games for stochastic growth with relative utility. Appl. Math. Optim., 74(3):643–668, 2016

    2016

  38. [45]

    Ismail and H

    A. Ismail and H. Pham. Robust Markowitz mean-variance p ortfolio selection under ambiguous covari- ance matrix. Math. Financ. , 29(1):174–207, 2019

    2019

  39. [46]

    N.V. Krylov. Controlled Diffusion Processes . Springer Berlin, Heidelberg, 1980

    1980

  40. [47]

    D. Lacker. A general characterization of the mean field l imit for stochastic differential games. Probab. Th. Rel. Fields , 165:581–648, 2016

    2016

  41. [48]

    D. Lacker. Limit theory for controlled McKean–Vlasov d ynamics. SIAM J. Control Optim. , 55(3):1641– 1672, 2017

    2017

  42. [49]

    D. Lacker. On a strong form of propagation of chaos for Mc Kean–Vlasov equations. Electron. Commun. Probab., 23(45):1–11, 2018

    2018

  43. [50]

    D. Lacker. On the convergence of closed-loop Nash equil ibria to the mean field game limit. Ann. Appl. Probab., 30(4):1693–1761, 2020. 47

    2020

  44. [51]

    Lasry and P.-L

    J.-M. Lasry and P.-L. Lions. Mean field games. Japan. J. Math. , 2(1):229–260, 2007

    2007

  45. [52]

    Laurière and L

    M. Laurière and L. Tangpi. Convergence of large populat ion games to mean field games with interaction through the controls. SIAM J. Math. Anal. , 54(3):3535–3574, 2022

    2022

  46. [53]

    Li and H

    J. Li and H. Min. Weak solutions of mean-field stochastic differential equations and application to zero-sum stochastic differential games. SIAM J. Control Optim. , 54(3):1826–1858, 2016

    2016

  47. [54]

    Liao, A.R

    H. Liao, A.R. Mészáros, C. Mou, and C. Chao. Convergence analysis of controlled particle systems arising in deep learning: from finite to infinite sample size. arXiv:2404.05185, 2024

    2024 arXiv

  48. [55]

    S. Mayorga. Short time solution to the master equation o f a first order mean field game system. J. Differential Equations , 268(10):6251–6318, 2020

    2020

  49. [56]

    Mayorga and A

    S. Mayorga and A. Świ¸ ech. Finite dimensional approxim ations of Hamilton–Jacobi–Bellman equations for stochastic particle systems with common noise. SIAM J. Control Optim. , 61(2):820–851, 2023

    2023

  50. [57]

    Mou and J

    C. Mou and J. Zhang. Mean field game master equations with anti-monotonicity conditions. J. Eur. Math. Soc. , published online first

  51. [59]

    Mou and J

    C. Mou and J. Zhang. Wellposedness of second order maste r equations for mean field games with nonsmooth data. Mem. Amer. Math. Soc. , 302, 2024

    2024

  52. [60]

    Pham and X

    H. Pham and X. Wei. Dynamic programming for optimal cont rol of stochastic Mckean–Vlasov dynamics. SIAM J. Control Optim. , 55(2):1069–1101, 2017

    2017

  53. [61]

    H. Pham, X. Wei, and C. Zhou. Portfolio diversification a nd model uncertainty: A robust dynamic mean-variance approach. Math. Financ. , 32(1):349–404, 2022

    2022

  54. [62]

    Ruthotto, S.J

    L. Ruthotto, S.J. Osher, W. Li, L. Nurbekyan, and S.W. Fu ng. A machine learning framework for solving high-dimensional mean field game and mean field contr ol problems. PNAS, 117(17):9183–9193, 2020

    2020

  55. [63]

    Święch and L

    A. Święch and L. Wessels. Finite dimensional projectio ns of HJB equations in the Wasserstein space. arXiv:2408.07688v1, 2024. 48

    2024 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.