pith. sign in

arxiv: 2501.11874 · v2 · submitted 2025-01-21 · 🧮 math.PR

Large Deviations for Slow-Fast Mean-Field Diffusions

Wei Hong , Wei Liu , Shiyuan Yang This is my paper

Pith reviewed 2026-05-23 05:45 UTC · model grok-4.3

classification 🧮 math.PR
keywords large deviationsslow-fast diffusionsmean-field interactionsrate functionfunctional occupation measuresviable pairsfeedback controlsLaplace principle
0
0 comments X

The pith

Slow-fast mean-field diffusions obey a large deviation principle with an explicit rate function.

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

This paper proves the large deviation principle for slow-fast mean-field diffusions in which the slow component depends on the time-marginal laws of the fast process. Standard weak-convergence arguments fail because of these law dependencies, so the authors characterize the limits of controlled sequences by means of functional occupation measures, viable pairs, and feedback controls. The characterization supplies both the upper and lower bounds of the Laplace principle. As a direct consequence an explicit formula for the rate function is obtained.

Core claim

The large deviation principle holds for the class of slow-fast mean-field diffusions that incorporate the laws of the fast process into the slow dynamics; the proof proceeds by identifying controlled-sequence limits through functional occupation measures, viable pairs and feedback controls, which in turn yields the explicit representation of the rate function.

What carries the argument

Functional occupation measures together with viable pairs and feedback controls, used to identify the limits of controlled sequences when direct weak-convergence methods are unavailable.

If this is right

  • The Laplace principle holds with the derived rate function.
  • Deviation probabilities can be computed from the explicit rate-function formula.
  • The result covers mean-field models in which the slow drift depends on the fast-process marginal law.
  • Upper and lower bounds are obtained separately through the occupation-measure characterization.

Where Pith is reading between the lines

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

  • The viable-pair technique may transfer to other slow-fast systems whose coefficients depend on empirical measures.
  • Approximate computation of the rate function could be attempted by discretizing the feedback-control representation.
  • The same occupation-measure framework might address moderate-deviation principles or central-limit corrections for the same class.

Load-bearing premise

The perturbations of the fast process and its time marginal law permit characterization of controlled-sequence limits via functional occupation measures, viable pairs, and feedback controls.

What would settle it

A concrete path measure whose large-deviation rate differs from the explicit formula constructed from the viable-pair limits would falsify the result.

read the original abstract

The aim of this paper is to investigate the large deviations for a class of slow-fast mean-field diffusions, which extends some existing results to the case where the laws of fast process are also involved in the slow component. Due to the perturbations of fast process and its time marginal law, one cannot prove the large deviations based on verifying the powerful weak convergence criterion directly. To overcome this problem, we employ the functional occupation measure, which combined with the notion of the viable pair and the controls of feedback form to characterize the limits of controlled sequences and justify the upper and lower bounds of Laplace principle. As a consequence, the explicit representation formula of the rate function for large deviations is also presented.

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

2 major / 2 minor

Summary. The manuscript establishes a large deviation principle for slow-fast mean-field diffusions in which the slow drift depends on the time-marginal law of the fast process. Because the standard weak-convergence criterion cannot be applied directly, the authors employ functional occupation measures together with the notions of viable pairs and feedback controls to characterize the limits of controlled sequences, prove the Laplace upper and lower bounds, and obtain an explicit variational representation of the rate function. The work is presented as an extension of existing results that did not incorporate the fast-process marginal into the slow component.

Significance. If the technical steps are correct, the result supplies a concrete rate-function formula for a class of multi-scale mean-field systems that appears in several applied contexts. The occupation-measure approach with viable pairs is a recognized route in the literature; carrying it through to an explicit representation strengthens the available tools for analyzing rare events in these models.

major comments (2)
  1. [§3.2] §3.2, Definition 3.3 and Proposition 3.5: the set of viable pairs is asserted to be compact in the appropriate topology once the fast marginal enters the slow drift, yet the proof sketch does not supply a uniform-integrability estimate on the family of controls that would guarantee tightness when the interaction term depends on the empirical measure of the fast process.
  2. [Theorem 2.4] Theorem 2.4 (Laplace principle): the lower bound is obtained by constructing a sequence of feedback controls whose cost converges to the variational problem, but the manuscript does not verify that the resulting controlled process remains a valid mean-field limit when the fast marginal is fed back into the slow equation; this step is load-bearing for the explicit rate-function representation.
minor comments (2)
  1. [§2] Notation for the occupation-measure space and the projection onto the slow component should be introduced once in §2 and used consistently thereafter.
  2. [Theorem 1.1] The statement of the main LDP (Theorem 1.1) refers to the topology of weak convergence on path space; a brief reminder of the precise metric would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments. The two major points concern the compactness argument for viable pairs and the verification step in the lower bound. We address each below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [§3.2] §3.2, Definition 3.3 and Proposition 3.5: the set of viable pairs is asserted to be compact in the appropriate topology once the fast marginal enters the slow drift, yet the proof sketch does not supply a uniform-integrability estimate on the family of controls that would guarantee tightness when the interaction term depends on the empirical measure of the fast process.

    Authors: We agree that the current proof sketch of Proposition 3.5 is too concise on this point. The uniform integrability follows from the linear growth and boundedness assumptions on the coefficients together with the moment bounds on the occupation measures, but these estimates are not written out explicitly. We will expand the proof in the revised version to include the required uniform-integrability argument. revision: yes

  2. Referee: [Theorem 2.4] Theorem 2.4 (Laplace principle): the lower bound is obtained by constructing a sequence of feedback controls whose cost converges to the variational problem, but the manuscript does not verify that the resulting controlled process remains a valid mean-field limit when the fast marginal is fed back into the slow equation; this step is load-bearing for the explicit rate-function representation.

    Authors: The verification is contained in the definition of viable pairs (Definition 3.3), which requires that any admissible pair satisfies the controlled integral equation with the fast marginal fed back into the slow drift. The feedback-control construction in the proof of Theorem 2.4 is built precisely so that the resulting pair is viable. We nevertheless accept that an explicit remark or short lemma confirming this property would improve clarity and will add it in the revision. revision: yes

Circularity Check

0 steps flagged

No significant circularity; standard occupation-measure method applied to extension

full rationale

The derivation relies on functional occupation measures, viable pairs, and feedback controls to obtain Laplace bounds and an explicit rate function. These are established tools from the large-deviations literature for slow-fast systems and are invoked to handle the additional dependence on fast-process marginals. The abstract explicitly frames the work as an extension of prior results rather than a self-contained re-derivation. No equations or steps reduce by construction to fitted parameters, self-defined quantities, or load-bearing self-citations. The central claim therefore remains independent of its own inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; all technical machinery is referenced at the level of named tools (occupation measures, viable pairs) without further breakdown.

pith-pipeline@v0.9.0 · 5638 in / 976 out tokens · 36433 ms · 2026-05-23T05:45:13.271755+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages

  1. [1]

    Ansari, Mean first passage time solution of the Smoluchowski equatio n: Application to re- laxation dynamics in myoglobin, Journal of Chemical Physics

    A. Ansari, Mean first passage time solution of the Smoluchowski equatio n: Application to re- laxation dynamics in myoglobin, Journal of Chemical Physics. 112(2000), 2516-2522

    work page 2000

  2. [2]

    Billingsley, Convergence of probability measures, John Wiley and Sons

    P. Billingsley, Convergence of probability measures, John Wiley and Sons. New York, 2013

    work page 2013

  3. [3]

    Bou´ e, P

    M. Bou´ e, P. Dupuis, A variational representation for certain functionals of Br ownian motion, Annals of Probability. 26(1998), 1641-1659

    work page 1998

  4. [4]

    Budhiraja, P

    A. Budhiraja, P. Dupuis, A variational representation for positive functionals of i nfinite di- mensional Brownian motion, Probab. Math. Satist. 20(2000), 39-61

    work page 2000

  5. [5]

    Budhiraja, P

    A. Budhiraja, P. Dupuis, V. Maroulas, Large deviations for infinite dimensional stochastic dynamical systems, Ann. Probab. 36(2008), 1390-1420

    work page 2008

  6. [6]

    Budhiraja, P

    A. Budhiraja, P. Dupuis, Analysis and approximation of rare events: Representation s and Weak Convergence Methods, Springer. 2019

    work page 2019

  7. [7]

    Bezemek, K

    Z. Bezemek, K. Spiliopoulos, Large deviations for interacting multiscale particle syst ems, Sto- chastic Processes. Appl. 155(2023), 27-108

    work page 2023

  8. [8]

    Bezemek, K

    Z. Bezemek, K. Spiliopoulos, Moderate deviations for fully coupled multiscale weakly in teracting particle systems, Stoch PDE: Anal Comp. 12(2024), 1265-1373

    work page 2024

  9. [9]

    Cerrai, M

    S. Cerrai, M. Freidlin, Averaging principle for stochastic reaction-diffusion equ ations, Probab. Theory Related Fields. 144 (2009), 137-177

    work page 2009

  10. [10]

    Dupuis, R

    P. Dupuis, R. Ellis, A weak convergence approach to the theory of large deviation s,Wiley, New York. 1997

    work page 1997

  11. [11]

    Delgadino, R.S

    M.G. Delgadino, R.S. Gvalani, G.A. Pavliotis, On the diffusive-mean field limit for weakly interacting diffusions exhibiting phase transitions, Arch. Ration. Mech. Anal. 241 (2021), 91- 148

    work page 2021

  12. [12]

    Dupuis, K

    P. Dupuis, K. Spiliopoulos, Large deviations for multiscale problems via weak converge nce methods, Stochastic Process. Appl. 122 (2012), 1947-1987

    work page 2012

  13. [13]

    Dupuis, K

    P. Dupuis, K. Spiliopoulos, H. Wang, Rare event simulation for rough energy landscapes, Proceedings of the 2011 Winter Simulation Conference. IEEE. 2011 , 504–515

    work page 2011

  14. [14]

    Dos Reis, W

    G. Dos Reis, W. Salkeld, J. Tugaut, Freidlin-Wentzell LDP in path space for McKean-Vlasov equations and the functional iterated logarithm law, Ann. Appl. Probab. 29 (2019), 1487-1540

    work page 2019

  15. [15]

    Dembo, O

    A. Dembo, O. Zeitouni, Large deviations techniques and applications, Springer, New York. 2009

    work page 2009

  16. [16]

    W. E and B. Engquis, Multiscale modeling and computations, Notice of AMS, 50 (2003), 1062-1070

    work page 2003

  17. [17]

    J. Feng, M. Forde, J.-P. Fouque, Short-maturity asymptotics for a fast mean-reverting Hest on stochastic volatility model, SIAM J. Financ. Math. 1 (2010) 126-141

    work page 2010

  18. [18]

    Feng, J.-P

    J. Feng, J.-P. Fouque, R. Kumar, Small-time asymptotics for fast mean-reverting stochasti c volatility models, Ann. Appl. Probab. 22 (2012) 1541-1575

    work page 2012

  19. [19]

    Freidlin, A

    M. Freidlin, A. Wentzell, Random Perturbations of Dynamical Systems, Springer-Verlag, 2012

    work page 2012

  20. [20]

    J. Gao, W. Hong, W. Liu, Small noise asymptotics of multi-scale McKean-Vlasov stoc hastic dynamical systems, J. Differential Equations. 364 (2023), 521-575

    work page 2023

  21. [21]

    Gomes, G.A

    S.N. Gomes, G.A. Pavliotis, Mean field limits for interacting diffusions in a two-scale po tential, J. Nonlinear Sci. 28(3) (2018), 905-941. LDP FOR SLOW-F AST MEAN-FIELD DIFFUSIONS 33

    work page 2018

  22. [22]

    Gasteratos, M

    I. Gasteratos, M. Salins, K. Spiliopoulos, Moderate deviations for systems of slow-fast sto- chastic reaction-diffusion equations, Stoch PDE: Anal Comp. 11(2023) 503-598

    work page 2023

  23. [23]

    W. Hong, S. Li, W. Liu, Strong convergence rates in averaging principle for slow-f ast McKean- Vlasov SPDEs, J. Differential Equations 316(2022), 94-135

    work page 2022

  24. [24]

    W. Hong, G. Li, S. Li, Multi-scale McKean-Vlasov SDEs: moderate deviation princ iple in different regimes, to appear in Ann. Inst. Henri Poincar´ e Probab. Stat

    work page

  25. [25]

    W. Hong, S. Li, W. Liu, X. Sun, Central limit type theorem and large deviation principle fo r multi-scale McKean-Vlasov SDEs, Probab. Theory Related Fields. 187 (2023), 133-201

    work page 2023

  26. [26]

    W. Hu, M. Salins, K. Spiliopoulos, Large deviations and averaging for systems of slow-fast stochastic reaction-diffusion equations, Stoch PDE: Anal Comp. 7 (2019), 808-874

    work page 2019

  27. [27]

    Hammersley, D

    W. Hammersley, D. Siska, L. Szpruch, Weak existence and uniqueness for McKean-Vlasov SDEs with common noise, Ann. Probab. 49 (2021) 527-555

    work page 2021

  28. [28]

    Huang, F.-Y

    X. Huang, F.-Y. Wang, Distribution dependent SDEs with singular coefficients, Stochastic Process. Appl. 129 (2019), 4747-4770

    work page 2019

  29. [29]

    Janke, Rugged Free-Energy Landscapes,Lecture Notes in Physics, vol

    W. Janke, Rugged Free-Energy Landscapes,Lecture Notes in Physics, vol. 736, Springer Science and Business Media, 2008

    work page 2008

  30. [30]

    Kac, Foundations of kinetic theory, in: Proceedings of the Third Berkeley Symposium on Mathematical Statistics and Probability, University of California Press

    M. Kac, Foundations of kinetic theory, in: Proceedings of the Third Berkeley Symposium on Mathematical Statistics and Probability, University of California Press. Berkeley and Los Angeles. 3(600)(1956), 171-197

    work page 1956

  31. [31]

    Khasminskii, Stochastic stability of differential equations, Springer Science and Business Me- dia, 2011

    R. Khasminskii, Stochastic stability of differential equations, Springer Science and Business Me- dia, 2011

    work page 2011

  32. [32]

    Liang, M

    M. Liang, M. Majka, J. Wang, Exponential ergodicity for SDEs and McKean-Vlasov process es with L ´evy noise, Ann. Inst. Henri Poincar´e Probab. Stat. 57 (3) (2021) 1665-1701

    work page 2021

  33. [33]

    W. Liu, M. R¨ ockner, Stochastic Partial Differential Equations: An Introductio n, Universitext, Springer, 2015

    work page 2015

  34. [34]

    W. Liu, M. R¨ ockner, X. Sun, Y. Xie, Strong averaging principle for slow-fast stochastic parti al differential equations with locally monotone coefficients, Appl. Math. Optim. 87 (2023), Paper no. 39

    work page 2023

  35. [35]

    McKean, Propagation of chaos for a class of nonlinear parabolic equa tions,Lecture Series in Differential Equations, 7 (1967), 41-57

    H.P. McKean, Propagation of chaos for a class of nonlinear parabolic equa tions,Lecture Series in Differential Equations, 7 (1967), 41-57

    work page 1967

  36. [36]

    Ren, F.-Y

    P. Ren, F.-Y. Wang, Bismut formula for Lions derivative of distribution depend ent SDEs and applications, J. Differ. Equ. 267 (2019) 4745-4777

    work page 2019

  37. [37]

    R¨ ockner, X

    M. R¨ ockner, X. Zhang, Well-posedness of distribution dependent SDEs with singul ar drifts, Bernoulli. 27 (2021) 1131-1158

    work page 2021

  38. [38]

    D. W. Stroock, An introduction to the theory of large deviations, Springer. New York. 2012

    work page 2012

  39. [39]

    Spiliopoulos, Large deviations and importance sampling for systems of slo w-fast motion, Appl

    K. Spiliopoulos, Large deviations and importance sampling for systems of slo w-fast motion, Appl. Math. Optim. 67(2013), 123-161

    work page 2013

  40. [40]

    G. Shen, H. Zhou, J.-L. Wu, Large deviation principle for multi-scale distribution de pendent stochastic differential equations driven by fractional Bro wnian motions, Journal of Evolution Equations. 24(2024), 35

    work page 2024

  41. [41]

    S. R. S. Varadhan, Large deviations and applications, CBMS-NSF Regional Conf. Ser. in Appl. Math. 46, SIAM, Philadelphia, 1984

    work page 1984

  42. [42]

    Wang, Distribution dependent SDEs for Landau type equations, Stochastic Process

    F.-Y. Wang, Distribution dependent SDEs for Landau type equations, Stochastic Process. Appl. (2018), 595-621

    work page 2018

  43. [43]

    H. Wu, J. Hu, C. Yuan, Large deviation for slow-fast McKean-Vlasov stochastic di fferential equations driven by fractional Brownian motions and Browni an motions, Stoch. Dyn. (2023), Paper No. 2450044. 34 LDP FOR SLOW-F AST MEAN-FIELD DIFFUSIONS

    work page 2023

  44. [44]

    Zwanzig, Diffusion in a rough potential, Proc

    R. Zwanzig, Diffusion in a rough potential, Proc. Natl. Acad. Sci. USA 85 (1988) 2029-2030

    work page 1988