Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Uniform-in-time weak propagation of chaos for consensus-based optimization

T0 review · 3 major / 4 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read Consensus-based optimization particles track their mean-field limit with error at most C/N, uniformly for all times.

desk verdict A serious and largely convincing uniform-in-time PoC result for cut-off CBO, but the central contraction estimate is imported without proof and Lemma 2.15 is sketched by analogy; the abstract also overclaims for the global minimizer. read the letter →

arxiv 2502.00582 v1 pith:G6PKQT4R submitted 2025-02-01 math.OC cs.LGmath.PR

Erhan Bayraktar , Ibrahim Ekren , Hongyi Zhou This is my paper · ORCID
classification math.OCcs.LGmath.PR MSC 35Q8937N4093D5082C3190C26
keywords consensus-basedoptimizationpropagationofchaosuniform-in-timemean-fieldlimitFokker-PlanckequationmasterSobolevspacesglobal
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

Consensus-based optimization (CBO) is a gradient-free algorithm in which $N$ particles drift toward a weighted consensus point to minimize a possibly non-convex objective. This paper proves that, on a bounded search domain cut off by a smooth function, the empirical distribution of the CBO particle system stays within weak error $O(N^{-1})$ of its mean-field limit uniformly over the infinite time horizon, with a constant independent of both time and the initial distribution. The number of particles can therefore be chosen without regard to how long the algorithm runs, eliminating the finite-horizon trade-off in earlier estimates. As a consequence, the empirical measure converges, simultaneously in $N$ and $t$, to the Dirac measure at the global minimizer in centered Wasserstein-type metrics.

What carries the argument

The machinery is the linearized Fokker–Planck equation (L-FPE) for fluctuations of the empirical measure around the mean-field flow, studied through its backward adjoint equation. The proof obtains estimates of the form $\|q_t - q_\infty \cdot \nabla \delta_{\tilde x_\mu}\|_{(n,\infty)'} \le C e^{-\kappa_0 t}$ for the L-FPE solutions, an ergodicity property that converts finite-horizon estimates into uniform-in-time ones. Because the CBO generator is not fully diffusive (its second-order coefficient vanishes at the consensus point), the backward equation does not decay directly; instead the paper uses the Feynman–Kac formula to show exponential decay of derivatives of the adjoint solution along geometric Brownian motions, at rate $e^{-\lambda t}$, and combines this with the exponential contraction of the mean-field CBO flow to a Dirac measure (Proposition 2.4).

What would settle it

Run the cut-off CBO particle system with a drift $\lambda$ below the threshold in Proposition 2.4 on a non-convex objective and compute $\sup_{t\ge 0} |\mathbb{E}[\Phi(\nu_t^N)] - \Phi(\bar\nu_t)|$ for the smooth translation-invariant centered Fourier–Wasserstein functional; if the supremum grows with $t$ for fixed $N$, the main theorem is false. Equivalently, check whether the mean-field consensus point $M(\mu_t)$ converges to a single point exponentially fast with a rate independent of the initial measure, since Lemma 3.3 uses exactly this to make the remainder $R^{1,i}_t$ decay.

Watch

Extended reading notes

Core claim

The central discovery is a uniform-in-time weak propagation of chaos for the cut-off consensus-based optimization system. For any smooth, translation-invariant functional $\Phi$ of the empirical measure, the paper establishes $\sup_{t\ge 0} |\mathbb{E}[\Phi(\nu_t^N)] - \Phi(\bar\nu_t)| \le C_{\mathrm{main}}/N$ for every $N \ge 2$, with $C_{\mathrm{main}}$ independent of the initial law. The proof identifies the second-order derivative of the solution operator $U(t,\mu)=\Phi(\mu_t)$ as the key object, decomposes it through the master equation into solutions of linearized Fokker–Planck equations, and shows that those solutions are uniformly bounded and decay exponentially in Sobolev dual norms to a term of the form $q_\infty \cdot \nabla \delta_{\tilde x_\mu}$. This exponential decay is what prevents errors from accumulating over long time intervals.

Load-bearing premise

The whole argument rests on the exponential contraction of the mean-field CBO flow toward a single Dirac measure, imported under a large-drift condition on $\lambda$; if that contraction rate were zero, negative, or time-dependent, the uniform-in-time $O(N^{-1})$ bound would collapse to a finite-horizon estimate.

Editorial extensions

If this is right

  • For any prescribed error tolerance, the number of particles $N$ can be fixed independently of the running time $t$, eliminating the finite-horizon trade-off in which $N$ had to grow exponentially with $t$.
  • The empirical measure of the CBO system converges jointly in $N$ and $t$: $\mathbb{E}\|\nu_t^N \circ \tau_{\langle \mathrm{Id}, \nu_t^N\rangle} - \delta_0\|^2_{-s,2} \le C_{FW}(N^{-1} + e^{-2\kappa t})$ and $\mathbb{E} W_2^2(\nu_t^N \circ \tau_{\langle \mathrm{Id}, \nu_t^N\rangle}, \delta_0) \le C_W(N^{-1} + e^{-\kappa t})$.
  • The uniform-in-time estimate holds for any initial distribution supported in the search ball, so the same $N$ works for any restart or initialization.
  • The proof transfers a recently developed uniform-in-time weak propagation methodology from torus settings to compactly supported, degenerate-diffusion CBO dynamics.

Reading between the lines

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

  • A natural testable extension is to remove the cutoff function $\varphi$: if the objective grows fast enough at infinity to control the exponential moments appearing in the estimates, the same $O(N^{-1})$ uniform bound might hold on the whole space.
  • The translation-invariance condition (2.5) suggests that functionals sensitive to the mean, such as uncentered Wasserstein distances, would not satisfy the theorem; one could test whether their weak error grows with $t$.
  • The exponential-decay estimates on L-FPE solutions could be reused to prove uniform-in-time propagation of chaos for related consensus algorithms (minimax CBO, constrained CBO), as long as their mean-field dynamics contract exponentially to a Dirac measure.
  • If the contraction rate $\kappa$ in Proposition 2.4 can be bounded below explicitly in terms of $\lambda, \sigma, \alpha$ and the objective's parameters, the constant $C_{\mathrm{main}}$ becomes computable and the result turns into a practical resource-allocation rule for CBO users.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper studies the cut-off consensus-based optimization particle system (2.1) on a bounded domain and establishes a uniform-in-time weak propagation of chaos result: for test functionals Φ satisfying (2.4)-(2.5), it claims sup_{t≥0} |E[Φ(ν_t^N)] - Φ(ν̄_t)| ≤ C_main/N for all N ≥ 2, with C_main independent of the initial distribution. The proof follows the Delarue-Tse methodology: decompose the weak error via the master equation, express second-order derivatives of U through linearized Fokker-Planck equations, establish exponential decay of their Sobolev dual norms using the ergodicity of the mean-field CBO flow, and conclude by Grönwall arguments. Corollaries give convergence of the centered empirical measure to δ_0 in Fourier-Wasserstein and Wasserstein distances with rates in N and t separately.

Significance. If the technical gaps are filled, this is a valuable contribution: it provides one of the first uniform-in-time propagation of chaos results for CBO on a bounded domain, with an explicit O(N^{-1}) rate and independent choices of N and t. The paper is substantial and well-structured, adapting the DT25 machinery with detailed GBM estimates in Appendix B, and it contains no fitted parameters. The explicit nature of the estimates and the careful master-equation decomposition are strengths. However, the central uniform-in-time claim rests on an imported contraction result for the cut-off dynamics and on a partially sketched proof of Lemma 2.15; both are load-bearing and need to be addressed before the result can be considered established.

major comments (3)
  1. [Section 2.2, Proposition 2.4] Proposition 2.4 asserts for the cut-off mean-field SDE (2.2) exponential convergence of the mean and consensus to a point x̃_μ with rate κ = 2(λ - dσ²e^{9αc_E r_cut² - αE}) and a constant C uniform in μ0, but no proof is given; Remark 2.5 calls it an adaptation of Theorem 4.1 of CCTT18. That theorem is stated for the standard CBO dynamics without the compactly supported cut-off φ in the diffusion coefficient, and the cut-off changes the noise degeneracy exactly on the boundary of the confinement. Since the exponential decay (4.1) is used throughout Lemmas 3.1, 3.3, and 3.4 and ultimately in Theorem 2.6 to obtain integrability uniformly in t, this is a load-bearing point. The authors should either prove Proposition 2.4 for (2.2) with the stated uniformity, or identify a published result that applies verbatim to the cut-off dynamics; a short citation is not sufficient.
  2. [Section 3.3 and Appendix A, Lemma 2.15] The proof of Lemma 2.15 is incomplete. Section 3.3 provides only a sketch, and Appendix A contains a 'Substitute proof of Lemma 2.15' that says 'By replacing all m(1)(t; μ, δz) with d(1)_j(t; μ, z) in the above proof, we get exactly (3.7) for Lemma 2.15.' This does not verify that the remainder expansion, the cancellation steps, and the associated Sobolev-norm bounds of the Lemma 2.14 proof carry over to d(2), particularly the treatment of the initial condition q(1)_{j,∞}·∇∂_{x_j}δ_{z_1} and the separate bounds (3.7) and (3.8). Lemma 2.15 is used in the proof of Theorem 2.6 to obtain the exponential decay of ∂_{z1_j}∂_{z2_j}δ²U/δm², which is essential for the uniform-in-time bound in (2.9). The authors should provide a complete, self-contained proof.
  3. [Section 4, Lemma 3.1, Step 3] In the treatment of the case q0 = ∂²_{x_j}δ_z, the displayed formula for γ^rem_j(T0; z) writes the first integral with upper limit T instead of T0, while the subsequent Cauchy-sequence argument treats T1,T2 → ∞. If the integral is genuinely truncated at T, the convergence argument does not apply; if it is a typo and should read T0, the correction should be made. In addition, the bound for the second-derivative terms is asserted to follow from Lemma B.1, but the display records only the expectation identities and not the uniform-in-x estimates needed for the dual-norm conclusion; this step should be expanded explicitly.
minor comments (4)
  1. [Section 2.4, Proof of Theorem 2.6] The proof relies on Lemma 4.11 of DT25 for the identities (2.11), but it does not explicitly check that the functionals satisfying (2.4)-(2.5) lie in the domain of the required measure derivatives of the flow map P_t. The authors should add a short justification or cite the precise result ensuring the needed regularity of U.
  2. [Corollary 2.8] The terminology 'centered Wasserstein distance' is used for W2(ν_t^N ∘ τ_{⟨Id,ν_t^N⟩}, δ0)^2, which is the variance functional; the name may confuse readers, since it is not a metric between the centered law and δ0 in the usual Wasserstein sense. A brief clarification would help.
  3. [Lemma 2.12] The statement of Lemma 2.12 includes 'for every j ∈ [d]' although the quantity m(1) involves no index j; this is a copy-paste artifact and should be removed.
  4. [Throughout] The manuscript contains numerous typographical errors and OCR artifacts, e.g., 'Prop agation' in the title/abstract, 'W e' and 't he' in the abstract, and 'Lemmata' vs 'lemmas'. A careful proofreading pass is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the uniform-in-time O(N^{-1}) weak propagation of chaos bound is derived from master-equation and linearized Fokker–Planck estimates, with external prior-work contraction results used only as inputs.

full rationale

The paper's central claim, Theorem 2.6, is a derived bound: the weak error is decomposed via the master equation (2.6)–(2.9), expressed in terms of solutions to linearized Fokker–Planck equations (Lemmas 2.10–2.11), and then bounded using the decay estimates in Lemmas 2.12–2.15. Those lemmas are proved in Sections 3–4 and Appendices A–B by direct estimates of backward Cauchy problems and geometric Brownian motions (Lemmas B.1–B.2), not by assuming the theorem's conclusion. The only imported nonlinear ingredient is the exponential contraction of the mean-field CBO flow, Proposition 2.4, which is explicitly taken from Carrillo–Choi–Totzeck–Tse (CCTT18) and Fornasier–Klock–Riedl (FKR24). These are independent works by different authors, not self-citations, and their result is not identical to the paper's target: the target is a uniform-in-time weak propagation-of-chaos estimate for the N-particle system, which is not an input of those prior theorems. The path from Proposition 2.4 through Lemmas 3.1, 3.3, 3.4 to Theorem 2.6 is a genuine proof chain, not a renaming or a fitted-input-called-prediction. No fitted parameters, data subsets, or definitions of the predicted quantity in terms of itself appear. If Proposition 2.4's adaptation to the cut-off dynamics were invalid or only finite-horizon, that would be a correctness risk, but it is not circularity under the stated evidentiary standard.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new entities, particles, or forces. It relies on standard stochastic analysis, a domain cutoff, and imported prior results. The large-lambda condition and finite-alpha bias are assumptions, not fitted parameters.

assumptions (4)
  • domain assumption Condition 2.1: E has a unique global minimizer x*, quadratic growth away from x*, and bounded derivatives up to order 4.
    Used to define the CBO dynamics and to control the consensus operator and its derivatives throughout the estimates.
  • domain assumption Assumption 2.2: the global minimizer is known to lie in B(c0, r_cut), and a smooth cutoff phi restricts dynamics to B(c0, 2r_cut).
    The cutoff is required to keep empirical measures compactly supported and to make exponential moment bounds finite; it modifies the original CBO algorithm.
  • domain assumption Proposition 2.4 (from CCTT18/FKR24): under large lambda, the mean-field flow has a unique weak solution and converges exponentially to delta_{tilde x_mu} with rate kappa.
    This imported exponential contraction is used at every stage of the proof, especially in Lemma 3.3 and the proofs of Lemmas 2.12-2.15.
  • standard math Master-equation and linearization calculus from DT25, Tse21, CST22, including Lemma 4.11 of DT25 and Theorem 2.14 of CST22.
    The decompositions (2.6)-(2.9) and the identities in (2.11) are imported from these references.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Uniform-in-time weak propagation of chaos for consensus-based optimization." pith.science (2026). https://pith.science/paper/G6PKQT4R

@misc{pith2026250200582,
  author       = {Pith},
  title        = {Pith review of: Uniform-in-time weak propagation of chaos for consensus-based optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G6PKQT4R}},
  note         = {Machine review of arXiv:2502.00582}
}
abstract

We study the uniform-in-time weak propagation of chaos for the consensus-based optimization (CBO) method on a bounded searching domain. We apply the methodology for studying long-time behaviors of interacting particle systems developed in the work of Delarue and Tse (ArXiv:2104.14973). Our work shows that the weak error has order $O(N^{-1})$ uniformly in time, where $N$ denotes the number of particles. The main strategy behind the proofs are the decomposition of the weak errors using the linearized Fokker-Planck equations and the exponential decay of their Sobolev norms. Consequently, our result leads to the joint convergence of the empirical distribution of the CBO particle system to the Dirac-delta distribution at the global minimizer in population size and running time in Wasserstein-type metrics.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Well-posedness and mean-field limit estimate of a consensus-based algorithm for multiplayer games

    math.OC 2025-05 conditional novelty 5.0 of 10

    This paper establishes existence, uniqueness, and a finite-particle mean-field error rate of order N^{-γ} for a multi-species consensus-based algorithm for multiplayer Nash games.

Reference graph

Works this paper leans on

31 extracted references · 20 canonical work pages · cited by 1 Pith paper

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTIO...

  2. [2]

    Aarts and J

    E. Aarts and J. Korst. Simulated Annealing and Boltzmann Machines: A Stochastic Approach to Combinatorial Optimization and Neural Computing . Wiley Series in Discrete Mathematics & Optimization. John Wiley and Sons Inc., 1991

    work page 1991

  3. [3]

    Baeck, D

    T. Baeck, D. B. Fogel, and Z. Michalewicz, editors. Handbook of Evolutionary Computation . CRC Press, 1st edition, 1997. doi:10.1201/9780367802486

    work page doi:10.1201/9780367802486 1997

  4. [4]

    Buckdahn, J

    R. Buckdahn, J. Li, S. Peng, and C. Rainer. Mean-field stochastic differential equations and associated pdes. The Annals of Probability , 45(2):824--878, 2017

    work page 2017

  5. [5]

    J. A. Carrillo, Y.-P. Choi, C. Totzeck, and O. Tse. An analytical framework for consensus-based global optimization method. Mathematical Models and Methods in Applied Sciences , 28(06):1037--1066, 2018, https://doi.org/10.1142/S0218202518500276 http://arxiv.org/abs/https://doi.org/10.1142/S0218202518500276 . doi:10.1142/S0218202518500276

    doi:10.1142/s0218202518500276 2018

  6. [6]

    Carmona and F

    R. Carmona and F. Delarue. Probabilistic Theory of Mean Field Games with Applications I. Mean Field FBSDEs, Control, and Games . Probability Theory and Stochastic Modelling. Springer Cham, New York, NY, 2018. doi:10.1007/978-3-319-58920-6

    doi:10.1007/978-3-319-58920-6 2018

  7. [7]

    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 . Probability Theory and Stochastic Modelling. Springer Cham, New York, NY, 2018. doi:10.1007/978-3-319-56436-4

    doi:10.1007/978-3-319-56436-4 2018

  8. [8]

    Cardaliaguet, F

    P. Cardaliaguet, F. Delarue, J.-M. Lasry, and P.-L. Lions. The Master Equation and the Convergence Problem in Mean Field Games . Annals of Mathematics Studies. Princeton University Press, Princeton, NJ, 2019

    work page 2019

Show all 31 references
  1. [9]

    J. A. Carrillo, S. Jin, H. Zhang, and Y. Zhu. An interacting particle consensus method for constrained global optimization, 2024. doi:10.48550/arXiv.2405.00891

    2024 doi

  2. [10]

    Q. Cormier. On the stability of the invariant probability measures of mckean-vlasov equations, 2024. doi:10.48550/arXiv.2201.11612

    2024 doi arXiv

  3. [11]

    Chassagneux, L

    J.-F. Chassagneux, L. Szpruch, and A. Tse. Weak quantitative propagation of chaos via differential calculus on the space of measures . The Annals of Applied Probability , 32(3):1929 -- 1969, 2022. doi:10.1214/21-AAP1725

    1929 doi

  4. [12]

    Delarue and A

    F. Delarue and A. Tse. Uniform in time weak propagation of chaos on the torus. To appear in Annales de l’Institut H enri P oincar\'e , 2025+

    2025

  5. [13]

    Fornasier, T

    M. Fornasier, T. Klock, and K. Riedl. Consensus-based optimization methods converge globally. SIAM Journal on Optimization , 34(3):2973--3004, 2024. doi:10.1137/22M1527805

    2024 doi

  6. [14]

    D. B. Fogel. Evolutionary Computation: Toward a New Philosophy of Machine Intelligence, 3rd Edition . IEEE Press Series on Computational Intelligence. Wiley-IEEE Press, 2006

    2006

  7. [15]

    N. J. Gerber, F. Hoffmann, and U. Vaes. Mean-field limits for consensus-based optimization and sampling, 2023. doi:10.48550/arXiv.2312.07373

    2023 doi

  8. [16]

    Guillin, P

    A. Guillin, P. Le Bris, and P. Monmarché. On systems of particles in singular repulsive interaction in dimension one: log and Riesz gas. Journal de l'École polytechnique Mathématiques , 10:867--916, 2023. doi:10.5802/jep.235

    2023 doi

  9. [17]

    J. H. Holland. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence . Complex Adaptive Systems. MIT Press, Cambridge, MA, 1992. doi:10.7551/mitpress/1090.001.0001

    1992 doi

  10. [18]

    Huang and J

    H. Huang and J. Qiu. On the mean-field limit for the consensus-based optimization. Mathematical Methods in the Applied Sciences , 45(12):7814--7831, 2022, https://onlinelibrary.wiley.com/doi/pdf/10.1002/mma.8279 http://arxiv.org/abs/https://onlinelibrary.wiley.com/doi/pdf/10.1...

    2022 doi

  11. [19]

    Huang, J

    H. Huang, J. Qiu, and K. Riedl. Consensus-based optimization for saddle point problems. SIAM Journal on Control and Optimization , 62(2):1093--1121, 2024, https://doi.org/10.1137/22M1543367 http://arxiv.org/abs/https://doi.org/10.1137/22M1543367 . doi:10.1137/22M1543367

    2024 doi

  12. [20]

    Kennedy and R

    J. Kennedy and R. Eberhart. Particle swarm optimization. Proceedings of ICNN'95 - International Conference on Neural Networks , 4:1942--1948, 1995. doi:10.1109/ICNN.1995.488968

    1942

  13. [21]

    D. Lacker. Hierarchies, entropy, and quantitative propagation of chaos for mean field diffusions. Probab. Math. Phys. , 4, 05 2021. doi:10.2140/pmp.2023.4.377

    2021 doi

  14. [22]

    Lacker and L

    D. Lacker and L. Le Flem. Sharp uniform-in-time propagation of chaos. Probab. Theory Related Fields , 187(1):443--480, 2023. doi:10.1007/s00440-023-01192-x

    2023 doi

  15. [23]

    F. Malrieu. Logarithmic sobolev inequalities for some nonlinear pde's. Stochastic Processes and their Applications , 95(1):109--132, 2001. doi:https://doi.org/10.1016/S0304-4149(01)00095-3

    2001 doi

  16. [24]

    F. Malrieu. Convergence to equilibrium for granular media equations and their Euler schemes . The Annals of Applied Probability , 13(2):540 -- 560, 2003. doi:10.1214/aoap/1050689593

    2003

  17. [25]

    Mischler, C

    S. Mischler, C. Mouhot, and B. Wennberg. A new approach to quantitative propagation of chaos for drift, diffusion and jump processes. Probability Theory and Related Fields , 161(1):1--59, 2015. doi:10.1007/s00440-013-0542-8

    2015 doi

  18. [26]

    Pinnau, C

    R. Pinnau, C. Totzeck, O. Tse, and S. Martin. A consensus-based model for global optimization and its mean-field limit. Mathematical Models and Methods in Applied Sciences , 27(01):183--204, 2017. doi:10.1142/S0218202517400061

    2017 doi

  19. [27]

    C. R. Reeves and J. E. Rowe. Genetic Algorithms: Principles and Perspectives . Operations Research/Computer Science Interfaces Series. Springer New York, NY, 2002. doi:10.1007/b101880

    2002 doi

  20. [28]

    Rosenzweig and S

    M. Rosenzweig and S. Serfaty. Global-in-time mean-field convergence for singular Riesz-type diffusive flows . Ann.Appl.Probab. , 33(2):954 -- 998, 2023. doi:10.1214/22-AAP1833

    2023 doi

  21. [29]

    Shi and R

    Y. Shi and R. Eberhart. A modified particle swarm optimizer. 1998 IEEE International Conference on Evolutionary Computation Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98TH8360) , pages 69--73, 1998. doi:10.1109/ICEC.1998.699146

    1998

  22. [30]

    Sznitman

    A.-S. Sznitman. Topics in propagation of chaos . Lecture Notes in Mathematics. Springer-Verlag, New York, 1991

    1991

  23. [31]

    A. Tse. Higher order regularity of nonlinear fokker-planck pdes with respect to the measure component. Journal de Mathématiques Pures et Appliquées , 150:134--180, 2021. doi:https://doi.org/10.1016/j.matpur.2021.04.005

    2021 doi

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.