{"id":"e6601c69-0cc6-44d7-92d2-416b1e5b49a5","arxiv_id":"1908.03375","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A stochastic Keller-Segel equation with common environmental noise is shown to be well-posed and to arise as the mean-field limit of an interacting particle system.","lead":"This paper introduces a stochastic version of the chemotaxis (Keller-Segel) equation where bacteria experience both individual random motion and a shared environmental noise. It proves the new equation has unique solutions under small-density conditions, and that it emerges as the mean-field limit of a system of many interacting particles.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: Assumption 1(iii) is the load-bearing premise, but it is explicit, disclosed, and the proof uses it exactly as stated.","rationale":"The reader's verdict ACCEPT with moderate confidence is appropriate. The paper's most fragile-looking premise is the divergence-free condition on σ, because it is exactly the identity that removes the L4 stochastic integral and makes the small-L4 fixed-point argument work. But this is an explicit standing assumption, not a hidden or contradictory step, and the authors state in Remark 3.1 that dropping it would require a different stopping-time argument. A stress-test should not penalize a theorem for being conditional on a clearly stated technical condition. I checked the main chains of estimates in Sections 3-5: the contraction map is well-defined on B, the contraction constant is made <1 by the stated smallness on χ‖ρ0‖_4, the higher-regularity result in Theorem 3.4 uses only bounds already available from Theorem 3.2, the backward-SPDE duality identifies ρ as the conditional density of the McKean-Vlasov SDE, and the mean-field bound in Theorem 5.1 handles the singular ∇Gε with the ε^{-d} cost controlled by choosing ε^{-d}=δ ln N. No circularity, sign error, or missing integrability was found. The one point that a referee would want expanded is the application of Krylov's Lp theory to the linear SPDE; the paper says 'standard computations' rather than displaying them. Since the equation is in divergence form with bounded lower-order coefficients, this is a verification task, not a demonstrated failure. Accordingly the verdict should remain unchanged.","tokens_in":25581,"tokens_out":29621,"duration_ms":295524,"concrete_test":"Independently verify the hypotheses of [36, Theorems 5.1 and 7.1, n=-1] for the linear SPDE (3.5) in divergence form: check that b_t(x)=D_j a^{ij}(t,x)-χ(∇G*ξ_t(x))^i is uniformly bounded for all ξ in the fixed-point ball B and that Assumption 1(i)-(iii) imply the required super-parabolicity for the stochastic term; also recompute the divergence-free cancellation immediately after (3.8) and in (3.14) to confirm the L4 stochastic integral vanishes identically.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No significant objection identified. The central claim is conditional on Assumption 1, and the proof of Theorem 3.2 really does require Assumption 1(iii): the divergence-free identity after (3.8) makes the L4 stochastic integral vanish, which yields the S∞_FW(L4) bound and permits the contraction in B. This is precisely the point the paper flags in Remarks 2.1 and 3.1, so the limitation is disclosed rather than hidden. I traced the L4 estimate (3.8)-(3.12), the contraction computation (3.14)-(3.16), the regularity bootstrap in Theorem 3.4, the density-identification duality in Theorem 4.1, and the ε and N error estimates (5.5)-(5.21); the inequalities close, and the final rate with ε^{-d}=δ ln N and Cδ<1 vanishes. The only residual verification risk is the undisplayed check that the linear SPDE (3.5) satisfies the divergence-form hypotheses of Krylov's Lp theory, but in that form the lower-order coefficient b_t=-χ∇G*ξ_t is bounded by (3.2), so I see no visible gap.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a stochastic Keller-Segel equation with common environmental noise, proves well-posedness for small L4 initial data under a divergence-free condition on the common-noise coefficient, establishes the well-posedness of a McKean-Vlasov SDE whose conditional density solves the SPDE, and proves a quantitative mean-field limit for a regularized interacting particle system. The main results are Theorems 3.2, 3.4, 4.1, 5.1 and Corollary 5.2. The proofs combine Krylov's Lp theory of SPDEs, a contraction mapping in S∞_{FW}(L4), a forward-backward SPDE duality argument, and Gronwall-type estimates with explicit rates in N and ε.","tokens_in":25745,"tokens_out":16890,"duration_ms":162841,"significance":"If the results are correct, this is the first rigorous well-posedness and mean-field limit result for a Keller-Segel type SPDE with a genuine common noise and a singular (Bessel) interaction. The paper is transparent about the load-bearing divergence-free condition and the smallness assumption on the L4 norm of the initial data. Its strengths include the explicit smallness thresholds, the duality-based identification of the conditional density, and quantitative convergence rates in the mean-field limit. The technical arguments are laid out in detail, and I found no circularity or internal inconsistency; the main external dependencies are clearly cited.","major_comments":[],"minor_comments":[{"comment":"The sentence 'This section is devoted to the global existence and uniqueness of the solution to nonlinear SPDE (1.7)' should refer to SPDE (1.3); the displayed equation numbers in the text sometimes lag the actual numbering.","section":"Section 3, first paragraph"},{"comment":"The phrase 'Unlike the classical deterministic KS system, which only allows for idiosyncratic noises' is inaccurate because the deterministic KS system has no noise; it should refer to classical stochastic KS models with only idiosyncratic noise.","section":"Abstract and Introduction"},{"comment":"The verification of the hypotheses of Krylov's Lp theory for the linear SPDE is compressed into 'standard computations.' Since this theory is used repeatedly (Theorems 3.2, 3.4, and Lemma 4.1), please add a short paragraph listing the checked conditions, particularly the boundedness of the first-order coefficient b_t = χ∇G*ξ_t and the treatment of the lower-order terms.","section":"Equation (3.6) and surrounding text"},{"comment":"The bound ||(∇Gε−∇G)*ρ^ε_s||∞ ≤ C ε ||ρ^ε_s||_{W^{1,4}} is asserted without proof; a sentence explaining the Sobolev embedding and the representation of ∇^2G*ρ as a Riesz potential would help the reader.","section":"Equation (5.5)"},{"comment":"There are numerous typos and spacing errors throughout the text (e.g., 'DERIV A TION', 'EQUA TION', 'Keller-Se gel' in the header); a careful proofreading is needed.","section":"General"}],"recommendation":"minor_revision","confidential_remarks":"This is a solid paper in scope for the journal. The central claims are internally consistent, and the limitations are disclosed rather than hidden. The only requests are for clarifications and proofreading, which do not affect the validity of the results."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a genuinely new stochastic Keller-Segel model: common noise plus Bessel potential, neither in the literature as far as I can tell. The paper proves unique well-posedness in an L4-small-data regime, identifies the conditional density of a McKean-Vlasov SDE as the solution, and establishes a mean-field limit from a regularized particle system. The package is coherent and the main estimates are detailed.\n\nThe key trick is clean: under the divergence-free assumption on the common-noise coefficient, the stochastic integral in the L4 estimate vanishes, giving a deterministic Gronwall bound and a contraction. That assumption is load-bearing, as the paper admits. If the environmental noise is incompressible, fine; if not, the proof collapses. That is the biggest soft spot, but it is disclosed in Remark 3.1 and used exactly as stated. The smallness condition on the initial L4 norm is also strong and rules out the interesting near-critical chemotaxis regime, but that is a scope limitation, not a flaw in the proof.\n\nThe duality argument connecting the forward SPDE with a backward SPDE to show that the conditional density solves the equation is well done. The mean-field estimates are carefully laid out; I traced the L4 estimates, the contraction, the duality, and the epsilon and N error bounds, and they close. The final rate with epsilon^{-d} = delta ln N and C*delta < 1 is fine. I did not independently verify every hypothesis of Krylov's Lp theory invoked for the linear SPDE, but the lower-order terms look controlled and I saw no visible gap.\n\nSelf-citations are used as tools, not as the target result, so no circularity. The paper is honest about its limitations. This is a serious contribution to stochastic aggregation equations and common-noise propagation of chaos, though not a breakthrough outside that subfield. I would send it to a serious referee without desk rejection, and I expect it to be accepted after the usual checks. Worth a close reading if you work on SPDEs or mean-field limits.","headline":"Genuinely new stochastic Keller-Segel SPDE with common noise, proved well-posed under a disclosed divergence-free condition and connected to a mean-field limit; a solid, honest paper worth a serious referee.","tokens_in":26309,"tokens_out":3206,"would_cite":true,"duration_ms":35395,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H15","35Q92","35K55","60K35"],"pacs":[],"model":"deepseek-v4-flash","headline":"Under a smallness condition on the initial density and a divergence-free common noise, the stochastic Keller–Segel equation admits a unique nonnegative solution, and the empirical measures of the regularized particle system converge to it…","keywords":["chemotaxis","stochastic Keller-Segel equation","common noise","mean-field limit","propagation of chaos","Bessel potential","stochastic partial differential equation","well-posedness"],"falsifier":"Take a common-noise coefficient $\\sigma$ whose divergence is not zero and an initial density satisfying the theorem's smallness condition, then test whether the Itô formula for $\\|\\rho_t\\|_4^4$ retains a nonzero martingale term; if that term is large enough to violate the Grönwall bound used in the contraction argument, or if simulations of the compressible-noise particle system show the empirical measure drifting away from (1.3), then the stated hypotheses are not sufficient.","tokens_in":121,"feed_emoji":"🦠","tokens_out":8959,"duration_ms":154113,"temperature":0.7,"pith_summary":"The paper proposes a stochastic Keller–Segel equation in dimensions $d=2,3$ that describes chemotaxis under a common environmental noise, and it tries to show that this equation is well-posed and is the true mean-field limit of a microscopic particle system. In the classical deterministic Keller–Segel equation, only independent idiosyncratic noises appear, and they average out; here the particles also share a common Wiener noise, so the limiting density remains random and satisfies an SPDE. The main results state that when the $L^4$-norm of the initial density is small and the common-noise coefficient is divergence-free, the SPDE has a unique nonnegative solution, and the empirical measure of the mollified interacting particle system converges weakly to that solution as the number of particles $N$ tends to infinity and the regularization parameter $\\varepsilon$ tends to zero. A sympathetic reader would care because real bacterial colonies experience shared fluctuating environments, and this gives a rigorous macroscopic equation for that setting rather than treating noise as merely particle-by-particle.","feed_headline":"Shared noise gives chemotaxis a well-posed random density","feed_subtitle":"From many interacting particles with a common environmental noise, a unique macroscopic bacterial density emerges in the large-N limit.","key_machinery":"The load-bearing object is the $L^4$ a priori estimate for the SPDE, combined with the Bessel potential's smoothing property $\\|\\nabla G*\\rho\\|_\\infty \\le S_d\\|\\rho\\|_4$. The divergence-free condition on $\\sigma$ (Assumption 1(iii)) makes the stochastic integral in Itô's formula for $\\|\\rho_t\\|_4^4$ vanish identically, reducing the estimate to a deterministic differential inequality; Grönwall's inequality then keeps $\\rho$ in $S^\\infty_{\\mathcal F^W}([0,T];L^4)$. This boundedness lets the Keller–Segel nonlinearity be absorbed into a small parameter $\\kappa$, and the solution map on the ball $B$ is shown to be a contraction. For the mean-field step, a duality analysis with a backward SPDE — whose solution has a probabilistic representation as a conditional expectation — identifies the SPDE solution as the conditional density of the nonlinear SDE, and mollifier estimates control the difference between $\\nabla G_\\varepsilon$ and $\\nabla G$.","core_discovery":"The central claim is that the stochastic aggregation-diffusion equation of Keller–Segel type, with chemical concentration $c_t=(I-\\Delta)^{-1}\\rho_t$ where $G$ is the Bessel potential, is globally well-posed in the spaces $M$ (Theorem 3.2) or $M_1$ (Theorem 3.4) provided Assumption 1 holds and $\\|\\rho_0\\|_4$ is small enough. The same solution is shown to be the conditional density of the McKean-Vlasov SDE (1.7), so the stochastic particle system is consistent with the SPDE at the level of law. Finally, the empirical measure of the regularized interacting particle system (1.5) converges weakly to that solution with an explicit rate in $N$ and $\\varepsilon$ (Theorem 5.1 and Corollary 5.2). In plain terms: the paper claims that chemoattracting particles buffeted by a shared environmental noise have a unique macroscopic density, and that this density is exactly what the particle system produces in the mean-field limit.","pith_inferences":["A natural test of the paper's scope is whether the divergence-free condition can be relaxed: any coefficient structure that makes the stochastic integral in the $L^4$ estimate vanish or remain controllable would likely allow the same contraction argument to run, and the paper's Remark 3.1 suggests the authors view this as the main technical gap rather than a physical necessity.","The smallness condition is phrased in $L^4$, not in the mass $m_0\\chi$ familiar from deterministic Keller–Segel; one might conjecture a stochastic critical-mass threshold, but the paper does not address it.","The explicit convergence rate suggests a practical random-particle blob method for stochastic chemotaxis; a numerical check of whether the $\\delta\\ln N$ window is sharp could be done with standard particle simulations.","The forward–backward SPDE duality used here is not tied to the Bessel kernel specifically; it may extend to other singular nonlocal SPDEs with common noise, such as stochastic aggregation equations with interaction kernels satisfying similar Sobolev bounds."],"forward_implications":["If the theorem is correct, chemotaxis models with common environmental noise have a unique macroscopic density, so questions about aggregation and blow-up can be posed for the SPDE rather than only for the particle system.","The conditional-density result gives a well-posed McKean-Vlasov SDE whose law under the common noise is the SPDE solution, closing the circle between microscopic and macroscopic descriptions.","For a test function $\\varphi$, the empirical measure error is bounded by $C\\big((\\delta\\ln N)^{(2d-2)/d}N^{-(1-C\\delta)} + N^{-1} + (\\delta\\ln N)^{-2/d}\\big)$, so the convergence is quantified and the regularization $\\varepsilon$ must vanish logarithmically slowly relative to $N$.","The non-mollified singular particle system is known to have strictly positive collision probability, so the regularization in (1.5) is not an artifact: it is necessary for a global strong solution, and the limit $\\varepsilon\\to 0$ is taken after $N\\to\\infty$.","Under the same smallness condition, the solution is nonnegative and conserves mass ($\\|\\rho_t\\|_1=1$), matching the probabilistic interpretation of a density."],"supporting_citations":[{"why":"Introduces the Keller–Segel chemotaxis model that the paper generalizes to a stochastic equation.","marker":"[35]"},{"why":"Supplies the deterministic existence and blow-up threshold for the classical equation, the baseline the stochastic smallness condition is compared with.","marker":"[32]"},{"why":"Provides the $L^p$-theory of SPDEs used to solve the linearized equation (3.5) and to obtain the required solution regularity.","marker":"[36]"},{"why":"Gives the Itô formula for $L^p$-norms of $W^1_p$-valued processes used in the central $L^4$ estimate.","marker":"[37]"},{"why":"Shows the non-mollified singular particle system has positive collision probability, motivating the regularization in (1.5).","marker":"[22]"},{"why":"Supplies the derivative bounds for the regularized Bessel/Newtonian potential $\\nabla G_\\varepsilon$ used in the particle approximation error estimates.","marker":"[29]"},{"why":"Establishes well-posedness of the backward SPDE whose probabilistic representation is used in the duality argument.","marker":"[19]"},{"why":"Provides the forward–backward SPDE duality analysis that identifies the SPDE solution as the conditional density of the nonlinear SDE.","marker":"[41]"}],"fun_headline_variants":["Shared noise gives chemotaxis a unique density","Common noise makes stochastic chemotaxis well-posed","Particles with common noise: unique chemotaxis density","Mean-field limit with shared noise: unique bacterial density","Stochastic Keller-Segel: common noise, unique solution"],"cache_read_input_tokens":28416,"weakest_assumption_plain":"Everything rests on the common-noise coefficient $\\sigma$ being divergence-free, meaning the environmental fluctuations are incompressible; without that, the central $L^4$ estimate acquires a stochastic term the proof cannot control, as the paper's Remark 3.1 concedes.","fun_headline_variants_meta":{"raw":{"variants":["Shared noise gives chemotaxis a unique density","Common noise makes stochastic chemotaxis well-posed","Particles with common noise: unique chemotaxis density","Mean-field limit with shared noise: unique bacterial density","Stochastic Keller-Segel: common noise, unique solution"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000245,"raw_usage":{"total_tokens":1482,"prompt_tokens":837,"completion_tokens":645,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":453,"completion_tokens_details":{"reasoning_tokens":567}},"tokens_in":453,"tokens_out":645,"duration_ms":6434,"temperature":1.0,"reasoning_tokens":567,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:15:22.110959+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a common-noise coefficient $\\sigma$ whose divergence is not zero and an initial density satisfying the theorem's smallness condition, then test whether the Itô formula for $\\|\\rho_t\\|_4^4$ retains a nonzero martingale term; if that term is large enough to violate the Grönwall bound used in the contraction argument, or if simulations of the compressible-noise particle system show the empirical measure drifting away from (1.3), then the stated hypotheses are not sufficient.","supporting_citations":[{"cited_title":"Keller and Lee A","cited_arxiv_id":null,"evidence_quote":"Introduces the Keller–Segel chemotaxis model that the paper generalizes to a stochastic equation."},{"cited_title":"2, 819–824","cited_arxiv_id":null,"evidence_quote":"Supplies the deterministic existence and blow-up threshold for the classical equation, the baseline the stochastic smallness condition is compared with."},{"cited_title":"Krylov, An analytic approach to SPDEs , Vol","cited_arxiv_id":null,"evidence_quote":"Provides the $L^p$-theory of SPDEs used to solve the linearized equation (3.5) and to obtain the required solution regularity."},{"cited_title":"Krylov, Itˆ o’s formula for theLp-norm of stochastic W 1 p -valued processes, Probability theory and related fields 147 (2010), no","cited_arxiv_id":null,"evidence_quote":"Gives the Itô formula for $L^p$-norms of $W^1_p$-valued processes used in the central $L^4$ estimate."},{"cited_title":"5, 2807–2861","cited_arxiv_id":null,"evidence_quote":"Shows the non-mollified singular particle system has positive collision probability, motivating the regularization in (1.5)."},{"cited_title":"308, 2719–2744","cited_arxiv_id":null,"evidence_quote":"Supplies the derivative bounds for the regularized Bessel/Newtonian potential $\\nabla G_\\varepsilon$ used in the particle approximation error estimates."},{"cited_title":"2, 175–219","cited_arxiv_id":null,"evidence_quote":"Establishes well-posedness of the backward SPDE whose probabilistic representation is used in the duality argument."},{"cited_title":"Department of Mathematics, Technische Universit ¨at M ¨unchen, Boltzmannstr","cited_arxiv_id":null,"evidence_quote":"Provides the forward–backward SPDE duality analysis that identifies the SPDE solution as the conditional density of the nonlinear SDE."}],"review_version":1}