pith. sign in

arxiv: 1907.09096 · v1 · pith:T6XSRMIOnew · submitted 2019-07-22 · 🧮 math.PR

Rate of propagation of chaos for diffusive stochastic particle systems via Girsanov transformation

Jean-Francois Jabir This is my paper

Pith reviewed 2026-05-24 18:23 UTC · model grok-4.3

classification 🧮 math.PR
keywords propagation of chaosMcKean-VlasovGirsanov transformationtotal variation distancemean-field limitexponential martingalestochastic particle systems
0
0 comments X

The pith

Girsanov transformation produces optimal total variation rate for propagation of chaos in diffusive particle systems.

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

This paper develops a quantitative bound on the rate at which a finite system of interacting diffusive particles converges to its McKean-Vlasov mean-field limit. The method uses a Girsanov change of measure to control the exponential martingale cost arising from the difference between the interacting drift and the mean-field drift. Under a sub-Gaussian moment assumption on the path-dependent nonlinearity, the bound is uniform and yields an optimal convergence rate measured in total variation distance. A reader would care because such rates determine how many particles are needed for reliable approximation in models of large-scale interacting systems.

Core claim

For Brownian-driven McKean-Vlasov dynamics whose nonlinearity is concentrated in a path-dependent drift obeying sub-Gaussian moment control, the Girsanov transformation supplies a uniform estimate on the exponential martingale between each particle trajectory and the corresponding mean-field process, which directly implies an optimal rate of propagation of chaos in total variation distance.

What carries the argument

Girsanov transformation producing a uniform bound on the exponential martingale cost between the interacting particle system and its McKean-Vlasov limit.

If this is right

  • The technique supplies an explicit constant in the convergence rate that depends only on the sub-Gaussian parameter.
  • The result strengthens the qualitative propagation of chaos statements of Lacker (2018) by adding a quantitative rate.
  • The argument supplies a stochastic interpretation of the entropy control method introduced in Jabin and Wang (2016).
  • The rate applies to finite-time horizons for systems driven by Brownian motion with the stated drift structure.

Where Pith is reading between the lines

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

  • The Girsanov estimate could be adapted to obtain rates in other distances such as Wasserstein by combining it with standard coupling inequalities.
  • Similar moment controls might permit quantitative chaos results for systems with jumps or state-dependent diffusion coefficients.
  • The explicit rate could be used to calibrate the number of particles required for a prescribed approximation error in numerical mean-field simulations.

Load-bearing premise

The McKean nonlinearity is confined to a path-dependent drift component that satisfies a particular sub-Gaussian moment control.

What would settle it

A concrete particle system satisfying the sub-Gaussian condition for which the total variation distance to the mean-field limit decays slower than the rate derived from the exponential martingale bound.

read the original abstract

This paper focus on investigating the explicit rate of convergence for the propagation of chaos, in a pathwise sense a family of interacting stochastic particle related to some Brownian driven McKean-Vlasov dynamics. Precisely the McKean form of nonlinearity is concentrated on a path dependent drift component and satisfies a particular sub-gaussian moment control. Such control enables to derive a uniform estimate of the cost in terms of exponential martingale between the particle and its McKean/mean-field limit system which in turn provide an optimal rate of propagation of chaos in terms of the total variation distance. As a by-product, we deepen some recent propagation of chaos results due to Lacker 2018 and provides a partial stochastic interpretation of the entropy control technique introduced in Jabin and Wang 2016.

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 / 2 minor

Summary. The manuscript develops a Girsanov-transform argument to obtain an explicit rate of propagation of chaos in total variation distance for N-particle systems whose mean-field limit is a Brownian-driven McKean-Vlasov equation whose drift is path-dependent and obeys a sub-Gaussian moment bound. The central step is a uniform-in-N bound on the exponential martingale that quantifies the Radon-Nikodym cost between the empirical particle law and the McKean-Vlasov law; this bound is then converted into a TV-distance estimate that is claimed to be optimal. The same estimate is used to recover and strengthen results of Lacker (2018) and to supply a probabilistic interpretation of the entropy-control technique of Jabin-Wang (2016).

Significance. If the moment hypothesis indeed closes the Girsanov estimate uniformly, the paper supplies a concrete, optimal-rate PoC result in the strong total-variation metric for a nontrivial class of path-dependent mean-field interactions. This is a technically stronger conclusion than the Wasserstein or weak-convergence statements that dominate the literature, and the Girsanov route offers a probabilistic alternative to the PDE/entropy methods used in the cited works.

minor comments (2)
  1. The abstract contains several grammatical and stylistic infelicities (e.g., 'This paper focus on', 'Precisely the McKean form of nonlinearity is concentrated on a path dependent drift component') that should be corrected before publication.
  2. Notation for the particle system, the McKean-Vlasov limit, and the Girsanov density should be introduced with explicit equation numbers in the main text so that the uniform martingale estimate can be traced directly to the sub-Gaussian hypothesis.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive summary and recommendation of minor revision. No major comments were listed in the report, so we have no specific points to address point-by-point at this stage. We will make any minor changes suggested during the revision process.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The derivation applies standard Girsanov change-of-measure estimates to a given sub-Gaussian moment bound on the path-dependent McKean drift, producing uniform control on the exponential martingale and thence a TV-distance rate for propagation of chaos. The argument cites external works (Lacker 2018, Jabin-Wang 2016) whose authors are distinct from the present paper; no self-citation is load-bearing, no parameter is fitted to a subset and then relabeled a prediction, and no quantity is defined in terms of the target rate. The central estimate therefore remains independent of the claimed conclusion and rests on classical martingale inequalities outside the paper's own inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central argument rests on the sub-Gaussian moment control stated in the abstract; no free parameters or invented entities are mentioned.

axioms (1)
  • domain assumption Sub-Gaussian moment control on the path-dependent drift component
    Abstract states that this control enables the uniform exponential-martingale estimate.

pith-pipeline@v0.9.0 · 5657 in / 1002 out tokens · 18310 ms · 2026-05-24T18:23:51.263560+00:00 · methodology

discussion (0)

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

Forward citations

Cited by 2 Pith papers

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

  1. Quantitative propagation of chaos for non-exchangeable diffusions via first-passage percolation

    math.PR 2024-09 unverdicted novelty 7.0

    The paper establishes sharp relative entropy estimates for marginals of non-exchangeable interacting particle systems by linking a BBGKY hierarchy to first-passage percolation.

  2. Regularization of a mean-field SDE by an additive common noise: The conditional expectation case

    math.PR 2026-05 unverdicted novelty 6.0

    Existence and uniqueness of weak solutions hold for McKean-Vlasov SDEs with common noise and conditional expectation interactions under bounded measurable drifts (with individual noise) or additional Lipschitz conditi...

Reference graph

Works this paper leans on

23 extracted references · 23 canonical work pages · cited by 2 Pith papers

  1. [1]

    Bossy, M., Some stochastic particle methods for nonlinear parabolic P DEs, Proceedings of 2005 GRIP Summer School, volume 15 of ESAIM Proc.: 18-57, 2003

    work page 2005

  2. [2]

    Bossy, M., and Jabir, J.-F., and Talay, D., On conditional McKean Lagrangian stochastic models, Probability Theory and Related Fields, 151:319-351, 2011

    work page 2011

  3. [3]

    Bougeron, S., and Lugosi, G., and Massart, P ., Concentration inequalities: A Nonasymptotic Theory of Independence, Oxford, 2016

    work page 2016

  4. [4]

    Carlen, E., and Krée, P ., Lp estimates on iterated stochastic integrals, Annals of Probability, 19(1): 354-368, 1991

    work page 1991

  5. [5]

    Carmona, R., and Delarue, F., Probabilistic Theory of Mean Field Games I: Mean Field FBSDE s, Control and Games, Springer, 2018

    work page 2018

  6. [6]

    Carmona, R., and Delarue, F., Probabilistic Theory of Mean Field Games II: Mean Field Game s with Common Noise and Master Equations , Springer, 2018

    work page 2018

  7. [7]

    and Mischler, S., On Kac’s chaos and related pr oblems, Journal of Functional Analysis, 266(10): 6055-6157, 2014

    Hauray, M. and Mischler, S., On Kac’s chaos and related pr oblems, Journal of Functional Analysis, 266(10): 6055-6157, 2014

    work page 2014

  8. [8]

    Jabin, P .-E., and Wang, Z., Mean field limit and propagation of chaos for Vlasov systems w ith bounded forces, J. Funct. Anal., 271(12): 3588-3627, 2016

    work page 2016

  9. [9]

    V olume 1, Theory, Models, Applications, Birkhauser-Springer, pp

    Jabin, P .-E., and Wang, Z., Mean field limit for stochastic particle systems , Active Particles. V olume 1, Theory, Models, Applications, Birkhauser-Springer, pp. 3 79-402, 2017

    work page 2017

  10. [10]

    and Wang, Z., Quantitative estimates of propagation of chaos for stochas tic systems with W − 1, ∞ kernels, Invent

    Jabin, P .- E. and Wang, Z., Quantitative estimates of propagation of chaos for stochas tic systems with W − 1, ∞ kernels, Invent. math., 214: 523-591, 2018. 16

    work page 2018

  11. [11]

    Jabir, J.-F., and Menozzi, S., Density estimates for some Langevin-McKean-Vlasov models and applications2 In preparation, 2019

    work page 2019

  12. [12]

    Jourdain, B., Diffusions with a nonlinear irregular drift coefficient and probabilistic interpretation of gener- alized burgers equations ESAIM Probability and Statistics, 1: 339-355, 1997

    work page 1997

  13. [13]

    Third Berkeley Sympos

    Kac, M., F oundation of kinetic theory, Proc. Third Berkeley Sympos. on Math. Statist, and Probab. Univ. Calif. Press., 3: 171-197, 1956

    work page 1956

  14. [14]

    N., Nonlinear Markov Processes and Kinetic Equations , Cambridge University Press, 2010

    Kolokoltsov, V . N., Nonlinear Markov Processes and Kinetic Equations , Cambridge University Press, 2010

    work page 2010

  15. [15]

    Lacker, D., On a strong form of propagation of chaos for McKean-Vlasov eq uations, Electron. Commun. Probab., V olume 23, paper no. 45, 11, 2018

    work page 2018

  16. [16]

    N., Statistics of Random Processes I , Springer-V erlag Berlin, 2001

    Liptser, R., and Shiryaev, A. N., Statistics of Random Processes I , Springer-V erlag Berlin, 2001

    work page 2001

  17. [17]

    R., A class of Markov processes associated with nonlinear parab olic equations, Proc

    McKean, J. R., A class of Markov processes associated with nonlinear parab olic equations, Proc. National Academy Sci. U.S.A., 56: 1907-1911, 1966

    work page 1907

  18. [18]

    R., Propagation of chaos for a class of non-linear parabolic equ ations, Lecture Series in Differential Equations

    McKean, J. R., Propagation of chaos for a class of non-linear parabolic equ ations, Lecture Series in Differential Equations. Catholic Univ. Washington D.C.: 4 1-57, 1967

    work page 1967

  19. [19]

    S., and V eretennikov, A

    Mishura, Y . S., and V eretennikov, A. Y .,Existence and uniqueness theorems for solutions of McKean- Vlasov stochastic equations, Preprint, 2016-2018

    work page 2016

  20. [20]

    T., Probability Metrics and the Stability of Stochastic Models , John Wiley, 1991

    Rachev, S. T., Probability Metrics and the Stability of Stochastic Models , John Wiley, 1991

    work page 1991

  21. [21]

    Simon, Schrödinger semigroups, Bull

    B. Simon, Schrödinger semigroups, Bull. Amer. Math. Soc. (N.S.), 7(3): 447â ˘A ¸ S526, 1982

    work page 1982

  22. [22]

    Sznitman, A.-S., T opics in propagation of chaos, Ecole d’été de Saint-Flour, 1989

    work page 1989

  23. [23]

    2Title subject to change

    V eretennikov, A., On mean-field {GI/GI/ 1} queueing model: existence, uniqueness, convergence, Preprint, April 2018. 2Title subject to change. 17

    work page 2018