Pith. sign in

REVIEW 2 major objections 4 minor 26 references

Relaxed Lagrangian Approach to First-Order Non-Convex Mean Field Type Control Problem

T0 review · 2 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read Relaxed controls give non-convex mean field control a fixed-point equilibrium

desk verdict A genuinely interesting relaxed-MFC existence paper with a real gap in the final fixed-point argument, though the negative-L counterexample in the reader's report does not stand up. read the letter →

arxiv 2412.03308 v2 pith:EIQJH2NN submitted 2024-12-04 math.OC

classification math.OC MSC 49N8049N9068T0793C10
keywords MeanFieldControlRelaxedLagrangianApproachcontrolsResidualNeuralNetworksNon-convexoptimalWassersteinspaceExistenceofequilibriaKakutanifixedpoint
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 aims to prove that first-order mean field type control (MFC) problems with non-convex running costs still admit equilibria, provided controls are allowed to be randomized. The proposed 'relaxed Lagrangian approach' replaces pointwise controls by probability measures over the control space, turning the non-convex minimization into a linear-in-measure problem on the Wasserstein space. Under smoothness and controlled-growth assumptions on the Lagrangian and drift, the authors claim there always exists a relaxed MFC equilibrium — a probability law over state trajectories and control measures that is optimal for its own induced population distribution. They also show that when the data satisfy a classical convexity condition, the relaxed equilibrium can be written back as a classical feedback control, recovering the standard MFC existence result. A sympathetic reader would care because the result would extend mean field control theory to settings where convexity fails, such as mean-field formulations of neural network training.

What carries the argument

The central object is the set $P^R_U$ of relaxed controls: probability measures $\mu = dt\otimes \mu_t$ on $[0,T]\times\mathbb{R}^n$ with $\int |u|^q \mu(dt,du) \le R$. The state equation becomes $\dot\gamma(t)=\int f(\gamma(t),u,m(t))\,\mu_t(du)$, and the cost $J^m(\gamma,\mu)=\int L(\gamma(t),u,m(t))\,\mu(dt,du)$ is linear in $\mu$, so convexity in the control is no longer needed. The proof is carried by the set-valued map $E(\eta)=\{\pi_{1\sharp}P : P\in R^*(m)\}$ with $m(t)=e_{t\sharp}\eta$; the load-bearing results are that $E$ has non-empty, convex, compact values and a closed graph, obtained via the lower semicontinuity of $J^m$ in $\mu$, and then Kakutani's fixed-point theorem yields $\bar\eta\in E(\bar\eta)$, hence the equilibrium $P$.

What would settle it

Take $L(x,u,\nu) = -1/(1+|u|^q)$ (which satisfies (L1)–(L4)) with a drift $f$ satisfying (F1)–(F3), and compute whether the inequality $\liminf_i J^m(\gamma,\mu_i) \ge J^m(\gamma,\mu)$ in Lemma 2.4(2) still holds; if for some $m\in M_r$ the argmin set $R^*(m)$ is empty, then the existence theorem as stated is not valid.

Watch

Extended reading notes

Core claim

The paper's central claim is that under assumptions (L1)–(L4) and (F1)–(F3) — smoothness, controlled growth, and Lipschitz dependence on the measure — there exists at least one relaxed MFC equilibrium for the first-order mean field type control problem, even when the running cost $L(x,u,m)$ is non-convex in the control $u$. The relaxation consists in taking controls to be probability measures $\mu\in P^R_U$ on $[0,T]\times\mathbb{R}^n$ with bounded $q$-th moment, so the state equation and the cost become linear in the control variable. An equilibrium is a joint law $P$ of state trajectories $\gamma$ and relaxed controls $\mu$ that minimizes the total cost $J(m,P)$ for its own induced distribution $m(t)=e_{t\sharp}\pi_{1\sharp}P$. The proof models this as a fixed point of the set-valued map $E(\eta)=\{\pi_{1\sharp}P : P\in R^*(m)\}$ with $m(t)=e_{t\sharp}\eta$, shows $E$ is non-empty, convex, compact-valued and has closed graph, and applies Kakutani's theorem. Under an additional convexity condition on the epigraph set $\mathcal{L}(t,x)$, the same machinery produces a strict relaxed equilibrium and recovers the classical MFC existence theorem.

Load-bearing premise

The existence proof needs the running cost $L$ to be nonnegative at the step where the factor $1+\varepsilon|u|^q$ is dropped from a lower-semicontinuity estimate; the stated assumptions only bound $|L|$, not its sign, so if $L$ can go negative the argument breaks.

Editorial extensions

If this is right

  • Every first-order MFC problem with $C^2$, controlled-growth, possibly non-convex data admits a relaxed MFC equilibrium in the sense of Definition 1.2.
  • When the epigraph set $\mathcal{L}(t,x)$ is convex, the relaxed equilibrium can be refined to a strict relaxed equilibrium, and the relaxed problem reduces to the classical MFC optimal control problem, so the result generalizes the existing convex theory.
  • The relaxed equilibrium carries all the information needed to design an optimal neural-network architecture in the mean-field training formulation: the parameters are read off from the second marginal of the equilibrium.
  • The fixed-point structure gives an algorithmic route: any numerical scheme that approximates the set-valued map $E$ and computes a fixed point would produce an approximate relaxed equilibrium.

Reading between the lines

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

  • If the existence theorem is correct, the relaxed MFC equilibrium can be interpreted as a mixed-strategy Nash equilibrium of a mean field game, suggesting a connection between the relaxed Lagrangian approach and randomized strategies in multi-agent reinforcement learning.
  • The nonnegativity used in the lower-semicontinuity step suggests the theorem likely needs an explicit lower bound on $L$, and the relaxation method may still be applicable under such a sign condition even where pointwise convexity fails.
  • The same relaxation on the Wasserstein space could be extended to second-order or stochastic MFC problems, replacing the pathwise Lagrangian by a relaxed control over the control space in a McKean-Vlasov dynamics.
  • A numerical test could check whether the relaxed equilibrium coincides with the classical equilibrium in the convex case and whether the gap between them measures the cost of non-convexity.
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 / 4 minor

Summary. The paper proposes a relaxed Lagrangian formulation for first-order mean-field-type control problems with non-convex Lagrangians. For a distribution m, the authors define P_R(m), the set of probability measures on curve–relaxed-control pairs whose support satisfies the state equation driven by m, and R*(m), the set of minimizers of the total cost J(m,P) over P_R(m). A relaxed MFC equilibrium is a measure P satisfying P∈R*((e_t#π1#P)_t). The main result, Theorem 1, asserts the existence of such an equilibrium under assumptions (L1)–(L4) and (F1)–(F3), proved by applying Kakutani's fixed-point theorem to the set-valued map E(η)=π1#R*((e_t#η)_t). Theorem 2 states that under a pointwise convexity condition there exists a strict relaxed equilibrium, recovering in particular the classical MFC existence result. The paper also motivates the framework through residual neural network training.

Significance. If the arguments are correct, Theorem 1 would be a useful extension of mean-field-type control existence theory to non-convex Lagrangians, a setting for which the authors note no direct first-order results are available. The relaxed-control construction on the Wasserstein space is natural, and the compactness apparatus for P_R_U and Γ_R^T is mostly well chosen. The paper is self-contained, uses standard measure-theoretic tools, and does not appear to rely on circular reasoning or fitted parameters. However, the proof as written contains a load-bearing gap in Proposition 2.4: the comparison that establishes the closed graph of E is only valid for competitors feasible for m_i, while the proof applies it to competitors feasible only for the limit m. This gap is local and, in my judgment, repairable as described in the major comments. The consequence is that the manuscript is not acceptable in its present form, but the central claim is defensible.

major comments (2)
  1. [Section 2.2, Proposition 2.4] The final displayed chain in the proof of Proposition 2.4 uses the inequality J(m_i,P_i) ≤ J(m_i,\hat P) for an arbitrary \hat P ∈ P_R(m). Since P_i is only known to minimize over P_R(m_i), this comparison requires \hat P ∈ P_R(m_i). The consistency condition (1.7) is formulated with the distribution m, and \hat P ∈ P_R(m) does not imply \hat P ∈ P_R(m_i) when f depends on the measure argument; convergence m_i → m does not make the two feasible sets coincide. This step is load-bearing because it is exactly what yields the closed graph of E, which is needed for the Kakutani fixed-point argument in Theorem 1. The gap is repairable: for a fixed \hat P ∈ P_R(m), define T_i(γ,μ)=(γ_i,μ), where γ_i solves the state equation (1.4) with m_i and γ_i(0)=γ(0). Then \hat P_i = T_i#\hat P ∈ P_R(m_i), \hat P_i → \hat P, and J(m_i,\hat P_i) → J(m,\hat P), so the argument can be completed with \hat P replaced by \hat P_i.
  2. [Section 2.2, Proposition 2.3] The measure h#η̃ is not well-defined as written. The map h is introduced only on the set {γ̃_x : x∈T^d} by γ̃_x ↦ (γ̃_x, μ_x), but no measurable selection of the associated optimal controls μ_x is specified. Without such a selection, h need not be a Borel map and h#η̃ may not exist as a push-forward. Since Γ*_m(x) is compact-valued and has closed graph, a measurable selection of optimal controls can be obtained by standard arguments, so this appears to be a technical gap rather than a fatal flaw.
minor comments (4)
  1. [Lemma 2.4(2)] The proof's displayed inequality L = L(1+ε|u|^q)/(1+ε|u|^q) ≥ L/(1+ε|u|^q) requires L ≥ 0, which is not among assumptions (L1)–(L4). The claimed lower semicontinuity is nevertheless true: L is continuous, |L| ≤ C(1+|u|^q), and measures in P_R_U have uniformly bounded q-th moments, so Proposition B.2(3) gives ∫ L dμ_i → ∫ L dμ. The proof should be corrected; in particular, the negative-L example sometimes cited against this lemma does not make R*(m) empty.
  2. [Sections 2 and 3] Proposition 2.1 has only parts (1)–(3), but Proposition 2.3 and the proof of Theorem 2 refer to Proposition 2.1(4). These references should be to Proposition 2.1(3).
  3. [Theorem 2(1)] The proof establishes only m0-a.e. optimality of the family {u_x^*} from the aggregate inequality ∫[...]m0(dx), not pointwise optimality for every x. If the statement 'for any x ∈ T^d' is intended to mean individual optimality at every x, an additional argument is needed; as written, the proof does not support that stronger reading.
  4. [Proposition 2.2(1)] The non-emptiness of Γ*_m(x) is dismissed with 'easily proved by convexity'; since J^m is linear in μ, the existence of a minimizer follows from compactness of P_R_U and continuity of μ ↦ J^m(γ(·;x,μ,m),μ). The argument should be stated explicitly rather than deferred.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the relaxed fixed-point existence proof is self-contained and does not reduce to its inputs.

full rationale

The paper proves existence of a relaxed MFC equilibrium by a Kakutani fixed-point argument on the set-valued map E(eta) = {pi1#P : P in R*((et#eta)_t)} (Section 2.2). The equilibrium definition in Definition 1.2 is self-referential by design, but the proof does not assume the desired conclusion; it constructs E from the minimization problem R*(m) and then shows non-emptiness, convexity, compactness, and closed graph using compactness of Gamma^R_T and P^R_U, continuity of L and f, and standard measure-theoretic results. No parameter is fitted to data and then renamed a prediction. The authors cite their own earlier work only as background for the classical Lagrangian approach to Mean Field Control, not as the load-bearing justification for the new existence theorem. The cited fixed-point, selection, and disintegration theorems are standard external results. Potential mathematical defects in the proof, such as the sign-sensitive inequality in Lemma 2.4(2) and the reference to Proposition 2.1(4) where only (1)-(3) are stated, are correctness questions, not instances of circularity: even if those steps fail, the failure is not that a conclusion is presupposed by its own definition or by a self-citation. The central claim therefore has independent mathematical content, and no circularity score is warranted.

Assumptions & free parameters 1 free parameters · 5 assumptions · 2 invented entities

The central claim rests on regularity assumptions on L and f, on fixed point and selection theorems, and on an unstated lower boundedness of L. No data fitting parameters are involved. The invented entities are new equilibrium concepts, not empirical objects.

free parameters (1)
  • R = arbitrary positive constant
    Defines the compact control moment budget P^R_U; the statement is meant for every R, so it is an input rather than a fitted number.
assumptions (5)
  • ad hoc to paper L is nonnegative (implicitly assumed)
    Lemma 2.4(2) multiplies the integrand by (1+epsilon|u|^q) and then drops the factor via an inequality that is only valid when L is nonnegative; assumptions only give |L| <= C1(1+|u|^q).
  • standard math Kakutani fixed point theorem applies to the correspondence E
    Used in the proof of Theorem 1 to obtain eta in E(eta); requires upper hemicontinuity and nonempty convex compact values.
  • standard math Skorokhod representation theorem
    Invoked as Theorem 3 in Appendix B to derive lower-semicontinuity consequences in Corollary B.1.
  • standard math Disintegration theorem
    Invoked as Theorem 4 to construct the measurable kernel q_{t,x} in Proposition 3.1.
  • standard math Measurable selection theorem of Haussmann and Lepeltier, Theorem A.9 of [21]
    Used in Theorem 2(1) to extract the measurable feedback alpha from the convexified set L(t,x).
invented entities (2)
  • Relaxed MFC equilibrium
    purpose: Solution concept for non-convex first-order mean field type control; a probability P over state-measure pairs that is a fixed point of the relaxed argmin correspondence.
    Definition 1.2. It is a mathematical solution concept with no falsifiable handle outside the paper; the existence theorem for it fails under the stated assumptions.
  • Strict relaxed MFC equilibrium
    purpose: A relaxed equilibrium whose relaxed control is a Dirac on a feedback control, used to connect back to classical MFC equilibria.
    Definition 1.3. Same status as the relaxed equilibrium; its existence is derived from Theorem 2, which depends on the flawed Theorem 1 proof.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Relaxed Lagrangian Approach to First-Order Non-Convex Mean Field Type Control Problem." pith.science (2026). https://pith.science/paper/EIQJH2NN

@misc{pith2026241203308,
  author       = {Pith},
  title        = {Pith review of: Relaxed Lagrangian Approach to First-Order Non-Convex Mean Field Type Control Problem},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EIQJH2NN}},
  note         = {Machine review of arXiv:2412.03308}
}
read the original abstract

This paper addresses the existence of equilibria for Mean Field type Control problems of first-order with non-convex action functional. Introducing a relaxed Lagrangian approach on the Wasserstein space to handle the lack of convexity. we prove the existence of new relaxed Nash equilibria and we show that our existence result encompasses the classical Mean Field Control problem's existence result under convex data conditions.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 24 canonical work pages

  1. [1]

    Ambrosio, N

    L. Ambrosio, N. Gigli, and G. Savaré.Gradient flows in metric spaces and in the space of probability measures. Lectures in Mathematics ETH Zürich. Birkhäuser Verlag, Basel, second edition, 2008

  2. [2]

    Bensoussan, J

    A. Bensoussan, J. Frehse, and P. Yam.Mean field games and mean field type control theory. SpringerBriefs in Mathematics. Springer, New York, 2013

  3. [3]

    Billingsley.Convergence of probability measures

    P. Billingsley.Convergence of probability measures. Wiley Series in Probability and Statistics: Probability and Statistics. John Wiley & Sons, Inc., New York, second edition, 1999

  4. [4]

    Buttazzo

    G. Buttazzo. Some relaxation problems in optimal control theory. J. Math. Anal. Appl., 125(1):272–287, 1987

  5. [5]

    Semicontinuity, relaxation and integral representation in the calculus of variations, volume 207 ofPitman Research Notes in Mathematics Series

    G.Buttazzo. Semicontinuity, relaxation and integral representation in the calculus of variations, volume 207 ofPitman Research Notes in Mathematics Series. Longman Scientific & Technical, Harlow; copublished in the United States with John Wiley & Sons, Inc., New York, 1989

  6. [6]

    Buttazzo

    G. Buttazzo. Relaxed optimal control problems and applications to shape optimization. In Nonlinear analysis, differential equations and control (Montreal, QC, 1998), volume 528 of NATO Sci. Ser. C Math. Phys. Sci., pages 159–206. Kluwer Acad. Publ., Dordrecht, 1999

  7. [7]

    Cannarsa and R

    P. Cannarsa and R. Capuani. Existence and uniqueness for mean field games with state constraints. In PDE models for multi-agent phenomena, volume 28 ofSpringer INdAM Ser., pages 49–71. Springer, Cham, 2018

  8. [8]

    Cannarsa and T

    P. Cannarsa and T. D’Aprile.Introduction to measure theory and functional analysis, volume 89 of Unitext. Springer, Cham, 2015

Show all 26 references
  1. [9]

    Cannarsa and C

    P. Cannarsa and C. Mendico. Mild and weak solutions of mean field game problems for linear control systems.Minimax Theory Appl., 5(2):221–250, 2020

  2. [10]

    Cannarsa and C

    P. Cannarsa and C. Sinestrari.Semiconcave functions, Hamilton-Jacobi equations, and opti- mal control, volume 58 ofProgress in Nonlinear Differential Equations and their Applications. Birkhäuser Boston, Inc., Boston, MA, 2004

  3. [11]

    Cardaliaguet.Notes on mean field game

    P. Cardaliaguet.Notes on mean field game. unpublished manuscript, 2013

  4. [12]

    Carmona and F

    R. Carmona and F. Delarue. Probabilistic analysis of mean-field games. SIAM J. Control Optim., 51(4):2705–2734, 2013. 24

  5. [13]

    Carmona and F

    R. Carmona and F. Delarue.Probabilistic theory of mean field games with applications I. Mean field FBSDEs, control, and games, volume 83 ofProbability Theory and Stochastic Modelling. Springer, Cham, 2018

  6. [14]

    Carmona and F

    R. Carmona and F. Delarue. Probabilistic theory of mean field games with applications II. Mean field games with common noise and master equations, volume 84 ofProbability Theory and Stochastic Modelling. Springer, Cham, 2018

  7. [15]

    Carmona, F

    R. Carmona, F. Delarue, and A. Lachapelle. Control of McKean-Vlasov dynamics versus mean field games. Math. Financ. Econ., 7(2):131–166, 2013

  8. [16]

    F. H. Clarke, Y. S. Ledyaev, R. J. Stern, and P. R. Wolenski.Nonsmooth analysis and control theory, volume 178 ofGraduate Texts in Mathematics. Springer-Verlag, New York, 1998

  9. [17]

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

  10. [18]

    W. E, J. Han, and Q. Li. A mean-field optimal control formulation of deep learning.Res. Math. Sci., 6(1):Paper No. 10, 41, 2019

  11. [19]

    Fischer and F

    M. Fischer and F. J. Silva. On the asymptotic nature of first order mean field games.Appl. Math. Optim., 84(2):2327–2357, 2021

  12. [20]

    D. A. Gomes and J. Saúde. A mean-field game approach to price formation.Dyn. Games Appl., 11(1):29–53, 2021

  13. [21]

    U. G. Haussmann and J. P. Lepeltier. On the existence of optimal controls.SIAM J. Control Optim., 28(4):851–902, 1990

  14. [22]

    W. Lee, S. Liu, H. Tembine, W. Li, and S. Osher. Controlling propagation of epidemics via mean-field control. SIAM J. Appl. Math., 81(1):190–207, 2021

  15. [23]

    Maass and J

    J. Maass and J. Fontbona. Symmetries in overparametrized neural networks: A mean-field view, https://arxiv.org/abs/2405.19995, 2024

  16. [24]

    Mazanti and F

    G. Mazanti and F. Santambrogio. Minimal-time mean field games. Math. Models Methods Appl. Sci., 29(8):1413–1464, 2019

  17. [25]

    C. Villani. Topics in optimal transportation, volume 58 ofGraduate Studies in Mathematics. American Mathematical Society, Providence, RI, 2003

  18. [26]

    M. A. Zaman, E. Miehling, and T. Başar. Reinforcement learning for non-stationary discrete- time linear-quadratic mean-field games in multiple populations.Dyn. Games Appl., 13(1):118– 164, 2023

Pith tools

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