Pith. sign in

REVIEW 2 major objections 4 minor 2 cited by

Propagation of Chaos for 2D Log Gas on the Whole Space

T0 review · 2 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read For $N$ Brownian particles with 2D Coulomb repulsion on the whole plane, relative entropy to the mean-field law stays within $C e^{\frac{C}{\varepsilon}t^\varepsilon}(E_0+1/N)$, for every $\beta>0$.

desk verdict The first quantitative entropic propagation of chaos for the 2D log gas on R^2 is plausible and the log-growth estimates are genuinely new, but the whole-space proof of Lemma 4.1 is delegated, not done, so Theorem 1.1 currently rests on an unverified Gronwall input. read the letter →

arxiv 2411.14777 v1 pith:XSFRPC74 submitted 2024-11-22 math.AP math.PR

Shuzhe Cai , Xuanrui Feng , Yun Gong , Zhenfu Wang This is my paper · ORCID
classification math.APmath.PR MSC 60K3582C2235K5535B40
keywords propagationofchaosmeanfieldlimit2DloggasCoulombrelativeentropymodulatedfreeenergyPoisson-Nernst-Planckequationmaximumprinciple
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

The paper proves that the $N$-particle stochastic system with 2D Coulomb repulsion on the whole plane converges, as $N\to\infty$, to the mean-field Poisson-Nernst-Planck density, in normalized relative entropy, with a quantitative rate. The rate combines a $1/N$ statistical error with a prefactor that grows at most like $e^{\frac{C}{\varepsilon}t^\varepsilon}$ in time, for any $\varepsilon>0$, and it holds for every inverse temperature $\beta>0$. This is the first quantitative propagation of chaos for the 2D log gas on the whole space; earlier whole-space results were qualitative, and quantitative results on the torus did not cover the unbounded domain. The proof works by transplanting the modulated free energy method to $\mathbb{R}^2$, using new logarithmic growth estimates for the mean-field density.

What carries the argument

The central object is the modulated free energy $E_N(\rho_N|\bar\rho_N)=H_N(\rho_N|\bar\rho_N)+\frac{1}{\beta}F_N(\mu_N,\bar\rho)$, which combines the normalized relative entropy with the modulated Coulomb energy between the empirical measure and the mean-field density. Its time derivative cancels the singular $\delta_0$ term coming from $\operatorname{div}K=-\Delta g$, leaving a commutator term that is controlled by the new logarithmic growth estimates $|\nabla\log\bar\rho|^2\lesssim \frac{1}{1+t}(1+\log(1+t)+\frac{|x|^2}{1+t})$ and a corresponding Hessian bound, obtained through a parabolic maximum principle and a Gaussian lower bound for $\bar\rho$. These estimates replace the positive lower bound that is available on the torus but not on the whole space.

What would settle it

A direct computation of the commutator integral in formula (4.2) for a two-particle Gaussian test law on $\mathbb{R}^2$ at small $t$ should show all boundary terms at infinity vanish and the integral is finite; any divergence would falsify Lemma 4.1. Alternatively, a Monte Carlo simulation of the $N$-particle system initialized as a product Gaussian should show normalized relative entropy staying below the stated $O(1/N)$ rate up to the time prefactor, and a clearly faster growth would contradict Theorem 1.1.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.1: if $\rho_N$ is an entropy solution to the $N$-particle Liouville equation and $\bar\rho$ solves the mean-field PNP equation with initial data satisfying the logarithmic growth conditions and Gaussian upper bound, then for any $\varepsilon>0$ and any $t\ge 0$, $$H_N(\rho_N|\bar\$rho^{{\otimes N}}$)(t)\le C\,$e^{{\frac{C}}${\varepsilon}t^\varepsilon}\left(E_N(\rho_N|\bar\$rho^{{\otimes N}}$)(0)+\frac{1}{N}\right).$$ The constant $C$ depends only on $\beta$, the initial mean-field density, and the Gaussian bound. This gives the first quantitative propagation of chaos for the 2D Coulomb gas on $\mathbb{R}^2$, upgrading the earlier qualitative compactness argument to a rate and extending the torus result to the whole space for every $\beta>0$.

Load-bearing premise

The load-bearing premise is that the whole-space version of the modulated-free-energy time evolution (Lemma 4.1) is valid with all boundary terms at infinity vanishing, even though its proof is delegated to the torus case; if that adaptation fails, the Gronwall bound collapses.

Editorial extensions

If this is right

  • For fixed time $t$, the normalized relative entropy is $O(1/N)$ up to a constant depending on $t$, giving the optimal statistical rate of mean-field convergence.
  • Quantitative propagation of chaos holds on $\mathbb{R}^2$ for every inverse temperature $\beta>0$, removing the earlier restriction to qualitative, rate-free convergence.
  • The same maximum-principle estimates yield a quantitative propagation-of-chaos result for the 2D viscous vortex model on the whole space, refining the previous time-growth estimate.
  • The equilibrium Gibbs measure of the $N$-particle system concentrates around the mean-field equilibrium in relative entropy with rate $(\log N)/N$ in two dimensions, as stated in Theorem 1.2.
  • The sufficient hypotheses on the initial data are the Gaussian upper bound and logarithmic growth of $\nabla\log\bar\rho_0$ and $\nabla^2\log\bar\rho_0$; no positive lower bound on $\bar\rho$ is required.

Reading between the lines

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

  • Editorial extension: the $e^{\frac{C}{\varepsilon}t^\varepsilon}$ prefactor is likely non-optimal. The paper itself notes that the bottleneck is the suboptimal $\frac{\log(1+t)}{1+t}$ decay in the maximum-principle estimates, so sharper maximum-principle bounds would probably yield a polynomial time growth instead.
  • Editorial extension: the same machinery may extend to the attractive 2D log gas or the Patlak-Keller-Segel system on the whole space, provided matching Gaussian lower and upper bounds hold; the paper states the logarithmic estimates are valid for such models on finite time intervals.
  • Editorial extension: a numerical test with moderate $N$ comparing the $N$-particle Coulomb system to the PNP solution could check whether the entropy difference actually follows the predicted $1/N$ rate up to a mild time factor; a rate worse than $1/N$ or a prefactor growing faster than $e^{\frac{C}{\varepsilon}t^\varepsilon}$ would indicate a missing whole-space boundary effect.
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 claims the first quantitative entropic propagation of chaos for the 2D log gas (weakly interacting particle system with 2D Coulomb interaction) on the whole space. The authors adapt the modulated free energy method of Bresch-Jabin-Wang to R^2 and prove logarithmic gradient and Hessian estimates for the mean-field PNP equation via a parabolic maximum principle. The main theorem states that, under initial bounds (1.11)-(1.12), the normalized relative entropy H_N(rho_N | rho_bar_N)(t) is bounded by C exp(C/eps t^eps)(E_N(rho_N|rho_bar_N)(0)+1/N) for any eps>0 and any beta>0. The proof combines a time evolution lemma for the modulated free energy, a large deviation estimate, and a Gronwall argument.

Significance. If the proof is completed, the result would be a substantial advance: it would close the gap between the torus result of BJW19a and the qualitative whole-space result of Liu-Yang, and it would provide one of the first quantitative propagation-of-chaos estimates for singular Coulomb interaction on R^2. The logarithmic growth estimates in Section 3, especially the Gaussian lower bound with the sharp exp(-|x|^2/t) scaling, are a meaningful technical contribution and improve on the earlier work by Feng-Wang for the vortex model. The paper also uses the large deviation theorem with test functions of quadratic growth, which is an interesting extension of the original JW18 framework. However, the whole-space adaptation of the modulated free energy evolution is not carried out in the manuscript, and there is a regularity mismatch between the assumptions of Theorem 1.1 and the lemmas used in the proof.

major comments (2)
  1. [Section 4.1, Lemma 4.1 (Eq. (4.2))] The time evolution of the modulated free energy is stated on the whole space, but its proof is delegated to [BJW19a, Proposition 2.1], which is a torus computation. This adaptation is not automatic. On T^2 the vector field w(x)=nabla log rho_bar(x)+beta nabla g*rho_bar(x) is bounded and every integration by parts is boundary-free; on R^2, w grows linearly in x and the commutator integrand (w(x)-w(y))·nabla g(x-y) is singular at x=y and only conditionally integrable as a function of the signed measure (d mu_N - d rho_bar)^(x)2. The entropy-solution hypothesis (1.10) gives a uniform entropy dissipation but no weighted spatial decay of rho_N, so the boundary terms at infinity in the integration by parts over R^{2N} are not shown to vanish, and no estimate is provided for the absolute value of the commutator integral before taking expectations. Since the Gronwall argument in (4.8)-(4.11) and the final bound in Theorem 1.1 rest entirely on (4.2), the central estimate is unsupported as written. A self-contained proof of Lemma 4.1 on R^2, or a localization/truncation argument with explicit control of the truncation error, is required.
  2. [Section 2.2 / Theorem 1.1 / Lemma 4.2] Lemma 2.2 assumes Schwartz initial data rho_bar_0 in S(R^2) and uses weighted L^2 moment estimates and Gagliardo-Nirenberg inequalities to obtain the decay rates (2.8). Theorem 1.1, however, assumes only rho_bar_0 in W^{2,1} cap W^{2,infty}(R^2), and the statement of Theorem 1.1 does not even include the hypothesis rho_bar in C([0,T], W^{2,1} cap W^{2,infty}) that Lemma 4.1 explicitly requires. The proofs of Theorem 3.1, Theorem 3.2, and Lemma 4.2 cite Lemma 2.2 for the decay of nabla rho_bar, nabla^2 rho_bar, and nabla^2 g*rho_bar. Consequently, the proof as written does not cover the stated regularity class. Either Theorem 1.1 must be strengthened to Schwartz (or sufficiently decaying C^2) initial data, or Lemma 2.2 and its consequences must be proved under the W^{2,1} cap W^{2,infty} assumptions with constants depending only on those norms.
minor comments (4)
  1. [Section 4.3, after Eq. (4.11)] The Gronwall argument yields an exponential factor of the form exp(C eps^{-2} ((1+t)^eps -1)), not exp(C eps^{-1} t^eps) as stated in Theorem 1.1; the statement should be adjusted, or the constant should be allowed to depend on eps in the stated way.
  2. [Sections 3.1 and 3.4] The repeated assertion that f = g*rho_bar is bounded is false on R^2, since f(x) behaves like -(1/(2 pi)) log|x| at infinity. The growth condition (3.6) can still be verified using the Gaussian decay of the auxiliary functions, but the justification given in the proofs of Theorem 3.1 and Theorem 3.2 should be corrected.
  3. [Lemma 4.2, Eq. (4.10)] The bound for w_2(x) uses the factor sup_z |z| rho_bar(z) without an explicit proof; it follows from the Gaussian upper bound Lemma 2.3, but this should be stated for the reader.
  4. [Throughout] There are numerous typos and minor notation issues: 'logathimic' in the proof of Theorem 3.1, 'convinence' in Section 2.1, 'Risze' and 'Rosezweig' in the references, and the notation '1s=0' in Theorem 1.2 should be clarified as an indicator function.

Circularity Check

1 steps flagged · score 4.0 of 10

No definitional or fitted-parameter circularity; the main entropy bound is not assumed. However, the central Gronwall input (Lemma 4.1) is a whole-space claim delegated to the authors' own torus paper, a load-bearing self-citation with the R^2 boundary issues unproved.

  1. self citation load bearing [Section 4.1, Lemma 4.1, Eq. (4.2)]
    "In this subsection we present the time derivative of the modulated free energy on the whole space, which is almost the same as the torus case for Coulomb potentials. Therefore, we adapt the proof of [BJW19a, Proposition 2.1] and claim the following lemma, and refer the proof to [BJW19a]."

    Lemma 4.1 is the only derivation of the time evolution of the modulated free energy and feeds the Gronwall estimate (4.8)-(4.11) that proves Theorem 1.1. Instead of proving it, the paper reduces it to [BJW19a, Proposition 2.1], a torus computation by the same authors. On T^2 the density is bounded below and integrations by parts have no boundary terms; on R^2, w(x)=∇log ρbar(x)+β∇g*ρbar(x) grows linearly and the boundary terms and commutator integral in (4.2) need separate justification. Thus the central premise rests on a load-bearing self-citation whose whole-space version is asserted, not established.

full rationale

The target bound HN <= C e^{C/eps t^eps}(EN(0)+1/N) is not assumed anywhere; the paper derives the logarithmic gradient and Hessian estimates in Theorems 3.1-3.2 from the stated initial hypotheses (1.11)-(1.12) via the maximum principle, and those estimates are the genuinely new content. There are no fitted parameters or data-dependent predictions. The only circularity-adjacent step is Lemma 4.1, whose proof the paper delegates to the authors' own [BJW19a] while merely asserting that the whole-space case is 'almost the same as the torus case'; this leaves the boundary/commutator terms in (4.2) unproved and makes the Gronwall input depend on a self-citation rather than on a derivation in this paper. A further non-circular correctness gap is that Lemma 2.2 assumes Schwartz initial data while Theorem 1.1 assumes only W^{2,1} ∩ W^{2,∞}; this affects the proof of the decay used inside Lemma 4.2 but is not a circular reduction. Because the central new estimates are independently proved and no equation in the paper is equivalent to its input by construction, the score reflects a load-bearing self-citation (4) rather than full circularity (6+).

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The central claim rests on standard external theorems (Grigor'yan maximum principle, Gross LSI, Jabin-Wang large deviation) and on well-posedness and entropy-solution existence results from prior literature. The new work contributes the logarithmic growth estimates and their application, with no fitted parameters and no new postulated physical entities.

assumptions (7)
  • standard math Grigor'yan maximum principle (Theorem 3.3): if F satisfies (3.5) and a growth condition (3.6)-(3.7), then F <= 0.
    Used in Section 3 to prove the Gaussian lower bound and the logarithmic gradient and Hessian estimates. Proven in [Gri09] and extended here; the proof in the paper is a sketch.
  • standard math Gross's logarithmic Sobolev inequality (Proposition 2.1)
    Used in Section 2.2 to obtain optimal smoothing and decay estimates for the limit equation.
  • standard math Jabin-Wang Large Deviation theorem (Theorem 4.1)
    Used in Section 4.2 to bound the exponential integral of the commutator test function; stated without proof, referenced to [JW18].
  • domain assumption Existence of entropy solutions to the Liouville equation (Proposition 1.1)
    Needed to apply the modulated free energy evolution to rho_N; proof sketched and referenced to [LY16, Theorem 2.1].
  • domain assumption Global well-posedness and regularity of the limit equation in C([0,T], L^1 cap L^infinity) with W^{2,1} cap W^{2,infinity} spatial regularity
    Invoked throughout Sections 2-4; existence and uniqueness cited to [dCRS23b, Proposition 3.1, 3.2] with a contraction argument sketched in Section 2.1.
  • domain assumption Initial data conditions (1.11)-(1.12): |grad log rho_0| lesssim 1+|x|, |grad^2 log rho_0| lesssim 1+|x|^2, and rho_0 Gaussian upper bound
    These are hypotheses of Theorem 1.1 and drive the logarithmic growth estimates in Section 3.
  • domain assumption Entropy solution hypothesis for rho_N in Theorem 1.1
    The main theorem applies only to entropy solutions of (1.5) as defined in Definition 1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Propagation of Chaos for 2D Log Gas on the Whole Space." pith.science (2026). https://pith.science/paper/XSFRPC74

@misc{pith2026241114777,
  author       = {Pith},
  title        = {Pith review of: Propagation of Chaos for 2D Log Gas on the Whole Space},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XSFRPC74}},
  note         = {Machine review of arXiv:2411.14777}
}
read the original abstract

We derive the quantitative propagation of chaos in the sense of relative entropy for the first time for the 2D Log gas or the weakly interacting particle systems with 2D Coulomb interactions on the whole space. We resolve this problem by adapting the modulated free energy method in [BJW23] to the whole space setting and establishing the crucial logarithmic growth estimates for the mean-field Poisson-Nernst-Planck (PNP) equation of single component via the parabolic maximum principle.

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. Sharp mean-field estimates for the repulsive log gas in any dimension

    math.PR 2025-06 conditional novelty 8.0 of 10

    The log-gas partition function is uniformly bounded in N for arbitrary bounded base measures, giving a mean-field convergence rate of order 1/N instead of (log N)/N for repulsive logarithmic interactions.

  2. Kac's Program for the Landau Equation

    math.AP 2025-06 conditional novelty 8.0 of 10

    The k-particle velocity marginals of Kac's particle system converge to the factorized law of the Landau solution for all power-law potentials, including Coulomb collisions, in weak, Wasserstein, entropic, and strong L...

Reference graph

Works this paper leans on

14 extracted references · 4 canonical work pages · cited by 2 Pith papers

  1. [1]

    Duality method for mean-field limits with singular interactions

    [BDJ24] Didier Bresch, Mitia Duerinckx, and Pierre-Emmanuel Jabin. Duality method for mean-field limits with singular interactions. arXiv preprint arXiv:2402.04695 ,

    arXiv

  2. [7]

    Quantitative pr opagation of chaos in a bimolecu- lar chemical reaction-diffusion model

    [LLN20] Tau Shean Lim, Yulong Lu, and James H Nolen. Quantitative pr opagation of chaos in a bimolecu- lar chemical reaction-diffusion model. SIAM Journal on Mathematical Analysis , 52(2):2098–2133,

    work page

  3. [8]

    Modulated logarith mic sobolev inequalities and gener- ation of chaos

    [RS23b] Matthew Rosenzweig and Sylvia Serfaty. Modulated logarith mic sobolev inequalities and gener- ation of chaos. arXiv preprint arXiv:2307.07587 ,

    arXiv

  4. [9]

    Relative entropy a nd modulated free energy without confinement via self-similar transformation

    [RS24a] Matthew Rosenzweig and Sylvia Serfaty. Relative entropy a nd modulated free energy without confinement via self-similar transformation. arXiv preprint arXiv:2402.13977 ,

    arXiv

  5. [10]

    Sharp commutato r estimates of all order for coulomb and riesz modulated energies

    [RS24b] Matthew Rosenzweig and Sylvia Serfaty. Sharp commutato r estimates of all order for coulomb and riesz modulated energies. arXiv preprint arXiv:2407.15650 ,

    work page arXiv

  6. [12]

    Quantitative particle approxim ations of stochastic 2d navier- stokes equation

    [SZ24] Yufei Shao and Xianliang Zhao. Quantitative particle approxim ations of stochastic 2d navier- stokes equation. arXiv preprint arXiv:2402.02336 ,

    arXiv

  7. [14]

    Quantitative Propagation of Chaos for the Mixed-Sign Viscous Vortex Model on the Torus

    [Wyn21] Dominic Wynter. Quantitative propagation of chaos for the mixed-sign viscous vortex model on the torus. arXiv preprint arXiv:2106.05255 ,

    work page Pith review arXiv

  8. [1991]

    Sharp local propagation of chaos for mean fi eld particles with w−1, ∞ kernels

    [Wan24] Songbo Wang. Sharp local propagation of chaos for mean fi eld particles with w−1, ∞ kernels. arXiv preprint arXiv:2403.13161 ,

    arXiv

Show all 14 references
  1. [1996]

    The attractive log gas: stability, uniqueness, and propagation of chaos

    [dCRS23a] Antonin Chodron de Courcel, Matthew Rosenzweig, and S ylvia Serfaty. The attractive log gas: stability, uniqueness, and propagation of chaos. arXiv preprint arXiv:2311.14560 ,

    arXiv

  2. [2014]

    Quantitative propagation o f chaos for 2d viscous vortex model on the whole space

    [FW23] Xuanrui Feng and Zhenfu Wang. Quantitative propagation o f chaos for 2d viscous vortex model on the whole space. arXiv preprint arXiv:2310.05156 ,

  3. [2016]

    Heat kernels on weighted manifolds and applications

    [Gri06] Alexander Grigor’yan. Heat kernels on weighted manifolds and applications. Cont. Math , 398(2006):93–191,

    2006

  4. [2020]

    Lectures on coulomb and riesz gases

    [Ser24] Sylvia Serfaty. Lectures on coulomb and riesz gases. arXiv preprint arXiv:2407.21194 ,

    arXiv

  5. [2023]

    Sharp uniform-in-time mean-field convergence for singular periodic riesz flows

    [dCRS23b] Antonin Chodron de Courcel, Matthew Rosenzweig, and S ylvia Serfaty. Sharp uniform-in-time mean-field convergence for singular periodic riesz flows. arXiv preprint arXiv:2304.05315 ,

    arXiv

  6. [2024]

    A ne w approach to the mean-field limit of vlasov-fokker-planck equations

    [BJS22] Didier Bresch, Pierre-Emmanuel Jabin, and Juan Soler. A ne w approach to the mean-field limit of vlasov-fokker-planck equations. arXiv preprint arXiv:2203.15747 ,

    arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.