{"id":"2402f747-e29e-478a-b295-7b2f015bae2f","arxiv_id":"2411.18336","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For a 2D chemotaxis-Navier-Stokes system with singular sensitivity up to c^{-5/6}, global bounded weak solutions exist under mild diffusion growth and bounded initial oxygen; if gamma > 1/2, initial bacterial mass must also be bounded.","lead":"This paper proves that a two-dimensional model of bacteria swimming in fluid, with oxygen as attractant and with diffusion that grows only mildly at high density, has global solutions that stay bounded whenever the initial oxygen level is bounded. For stronger singular sensitivities the proof additionally needs a bound on the initial bacterial mass; with strictly positive diffusion the solution is classical.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Case split in Lemmas 5.2–7.4 omits γ=1/2, so the written proof does not cover Theorem 1.1 at this parameter value; the boundary-estimate concern raised by the reader is not valid.","rationale":"The reader's conditional verdict is based mainly on a supposed boundary-estimate failure on non-convex domains, but that estimate is a standard consequence of bounded geometry and does not require convexity; it therefore does not threaten the argument. However, the proof has a genuine, if small, omission: the pivotal lemmas after Section 5 are stated only for γ∈[0,1/2) and γ∈(1/2,5/6], leaving γ=1/2 uncovered, even though Theorem 1.1 explicitly includes γ=1/2 and the L¹ condition on n0 is not assumed there. The gap is easily fixed by extending the first case to γ∈[0,1/2] and using Lemma 3.6, so the conditional verdict remains appropriate. My read does not change the reader's verdict, but it shifts the reason: the concern is a repairable missing case, not an invalid boundary lemma.","tokens_in":45602,"tokens_out":26476,"duration_ms":214726,"concrete_test":"Re-run the proof of Lemma 5.2 with γ=1/2 assigned to the first case, taking L(M) from Lemma 3.6 instead of Lemma 3.1, and check that Lemma 4.2 and the L¹ condition on n0 are never used for γ=1/2. If the energy functional Fε of Lemma 3.6 yields exactly the same differential inequality (3.20) with L=L(M) finite, and Lemmas 5.4–7.4 then follow verbatim with the first-case hypotheses, the gap closes and Theorem 1.1 holds at γ=1/2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakest point in the paper is not the boundary estimate highlighted by the reader. The estimate ∂|∇w|²/∂ν ≤ C|∇w|² for w∈C²(Ω̄) with w_ν=0 is valid on any smooth bounded domain: in an adapted frame it reads ∂_ν|∇w|² = -2 II(∇_τ w,∇_τ w), and |II(τ,τ)|≤C|τ|² by boundedness of the second fundamental form; no convexity is required. It is used correctly in Lemmas 3.2 and 3.4.\n\nThe actual gap is the case split on γ. Theorem 1.1 allows γ∈[0,5/6], including γ=1/2. Yet Lemmas 5.2, 5.4, 5.5, Corollary 5.6, Lemmas 5.7, 5.8, 6.1, 7.1, 7.3 and 7.4 are all stated with first case γ∈[0,1/2) and second case γ∈(1/2,5/6]. The first case uses L(M) from Lemma 3.1 (clearly a typo for Lemma 3.6) and does not require ∥n0∥_{L¹}≤M; the second case requires ∥n0∥_{L¹}≤M. Thus γ=1/2 falls into neither branch as written, and the proof does not establish Theorem 1.1 for γ=1/2. This is repairable—Lemma 3.6 and Lemma 4.2 both cover γ=1/2—but it is a genuine omitted case in the chain of estimates.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a global boundedness and existence result for a two-dimensional chemotaxis-Navier-Stokes system with degenerate nonlinear diffusion D(n) and singular, possibly tensor-valued sensitivity satisfying |S(x,n,c)| ≤ S0(c)/c^γ for γ∈[0,5/6]. The main theorem (Theorem 1.1) states that for each M>0 there is a threshold L(M) such that whenever D has liminf at infinity greater than L and D(n)/n remains positive near zero, and the initial data satisfy the stated regularity with ||c0||_∞ ≤ M (and additionally ||n0||_1 ≤ M when γ>1/2), the approximate problems admit a limit which is a global bounded weak solution; if D(0)>0, the solution is classical. The proof follows and extends the strategy of Winkler [52]: it derives spatio-temporal estimates via a Trudinger-Moser inequality, constructs energy-like functionals involving the second primitive of D and powers of ∇c, treats the regimes γ≤1/2 and γ>1/2 separately in Sections 3 and 4, then performs an iterative bootstrap to obtain uniform L∞ bounds, establishes time-derivative bounds, and passes to the limit via an Aubin-Lions argument. Corollaries cover porous medium type diffusion D(n)=n^{m-1}, m∈(1,2].","tokens_in":15,"tokens_out":9521,"duration_ms":141255,"significance":"If correct, this is a meaningful advance: it extends the three-dimensional Stokes result of [52] to the two-dimensional Navier-Stokes setting and enlarges the admissible singular exponent from the prototypical γ=1/2 to γ=5/6 under very mild diffusion enhancement. The proof is largely self-contained, with detailed estimates and explicit constants, and the energy structure is elegantly adapted to avoid convexity restrictions on the domain. In particular, the boundary estimate used in Lemmas 3.2 and 3.4 is valid on arbitrary smooth bounded domains, so the apparent non-convexity concern raised in the evaluation of this paper does not actually land. The main defect is a repeated case-split omission of γ=1/2 in the statements of several lemmas; this is local and repairable, but it does leave the written proof of Theorem 1.1 incomplete at that parameter value.","major_comments":[{"comment":"Theorem 1.1 asserts the result for all γ∈[0,5/6], including γ=1/2. However, all these lemmas are stated with a first case γ∈[0,1/2) and a second case γ∈(1/2,5/6], so that γ=1/2 falls under neither branch. The proof of Lemma 5.2 actually begins with 'If γ∈[0,1/2]' and invokes Lemma 3.6, which does cover γ=1/2, so the estimates themselves are available; nevertheless, as written, the proof of Theorem 1.1 does not cover γ=1/2 through the cited lemmas. Please correct the case split to γ∈[0,1/2] in the first branch throughout, and in the first branch cite L(M) from Lemma 3.6 rather than Lemma 3.1, which does not define L(M).","section":"Lemmas 5.2, 5.4, 5.5, 5.7, 5.8, 6.1, 7.1, 7.3, 7.4 and Corollary 5.6"}],"minor_comments":[{"comment":"The first case in these lemmas refers to 'L(M)>0 provided by Lemma 3.1'; Lemma 3.1 only furnishes a constant C(M) in a differential inequality, while L(M) is introduced in Lemma 3.6. This citation should be corrected.","section":"Lemmas 5.2 onward"},{"comment":"The proof of Lemma 5.2 writes 'If γ∈[0,1/2]' in the first branch, while the statement says γ∈[0,1/2); please harmonize the statement and proof so that the intended coverage of γ=1/2 is explicit.","section":"Lemma 5.2, proof"},{"comment":"The boundary estimate ∂|∇w|²/∂ν ≤ C|∇w|² for w∈C²(Ω̄) with ∂w/∂ν=0 is valid on any smooth bounded domain, not only convex ones; the citations to [27, Lemma 4.2] and [17, Lemma 4.2] are appropriate, so no revision is needed there, but it would help to state this explicitly once to avoid confusion.","section":"Section 3, Lemmas 3.2 and 3.4"}],"recommendation":"major_revision","confidential_remarks":"The paper is technically sound and the missing γ=1/2 case in the case splits is an oversight in the statements, not a fatal mathematical flaw: the proofs, especially Lemma 5.2, already indicate the correct coverage. The boundary-estimate concern raised in the review is unfounded. I recommend major revision to correct the case splits and the L(M) citation; no new mathematical ideas are required."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main take: the theorem is likely correct and deserves a serious referee. The paper extends Winkler's [52] from 3D Stokes with gamma=1/2 to 2D Navier-Stokes with gamma up to 5/6, removes the convexity assumption, and adds a genuinely new second energy for gamma>1/2. The estimates are detailed and coherent; the Section 4 energy is not a routine tweak.\n\nThe soft spot is a set of statement-level typos: Lemmas 5.2, 5.4, 5.5, Corollary 5.6, Lemmas 5.7, 5.8, 6.1, 7.1, 7.3, and 7.4 split as gamma in [0,1/2) vs gamma in (1/2,5/6], omitting gamma=1/2. The proof of Lemma 5.2 actually says \"If gamma in [0,1/2]\", so the mathematics covers the omitted case; the statements just need to be corrected. This is a nuisance, not a gap in the argument.\n\nThe stress-test concern about the boundary estimate d|grad w|^2/dnu <= C|grad w|^2 is a red herring; it holds on any smooth bounded domain via the second fundamental form, no convexity needed. The reader's worry is unfounded.\n\nOne minor thing: Corollary 1.3 depends on an unpublished preprint [2]. That is acceptable for a preprint, but the referee should check whether it is actually needed.\n\nOverall: the paper is careful, the main result is an advance for the subfield, and the flaws are cosmetic. Send it to peer review.","headline":"Solid extension of Winkler's 3D Stokes result to 2D Navier-Stokes with gamma up to 5/6; the gamma=1/2 case-split typo in several lemma statements is cosmetic, and the boundary-estimate worry is a red herring.","tokens_in":46542,"tokens_out":5262,"would_cite":true,"duration_ms":42691,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35D30","35A01","35K65","35Q35","35Q92","92C17"],"pacs":[],"model":"deepseek-v4-flash","headline":"For any prescribed bound on the initial oxygen concentration, a two-dimensional chemotaxis-Navier–Stokes system has a global bounded weak solution when diffusion grows, however slowly, to a sufficiently high limit and vanishes no faster…","keywords":["chemotaxis-Navier-Stokes system","degenerate diffusion","singular chemotactic sensitivity","global weak solution","boundedness","porous medium diffusion","two-dimensional chemotaxis-fluid"],"falsifier":"Take a smooth bounded domain in the plane with a concave boundary patch and a sequence of functions $w_j\\in C^2(\\overline{\\Omega})$ satisfying $\\partial w_j/\\partial\\nu=0$ on $\\partial\\Omega$; compute the ratio $\\partial|\\nabla w_j|^2/\\partial\\nu$ divided by $|\\nabla w_j|^2$ on the boundary. If this ratio is unbounded, the boundary estimate quoted from [27, Lemma 4.2] fails on non-convex domains, and the boundary integrals used to close the energy estimates cannot be absorbed, so the theorem's claim about arbitrary smoothly bounded domains would not follow.","tokens_in":45363,"feed_emoji":"🧫","tokens_out":7464,"duration_ms":64285,"temperature":0.7,"pith_summary":"This paper proves that a chemotaxis-Navier–Stokes system in a two-dimensional bounded domain, describing oxygen-consuming bacteria in a fluid, admits global bounded weak solutions under two structural assumptions on bacterial movement. The chemotactic sensitivity may be singular, blowing up like a negative power $c^{-\\gamma}$ of the oxygen concentration, with $\\gamma$ up to $5/6$; this goes beyond the exponent $\\gamma=1/2$ previously treated in a three-dimensional Stokes setting. In return, the diffusion coefficient only needs to be enhanced in a very mild sense: at large cell densities it must eventually exceed a threshold $L(M)$ that depends only on the initial oxygen bound, and near zero it must vanish no faster than linearly. No smallness condition on the initial data is needed, and if diffusion stays positive at zero density the constructed solution is classical.","feed_headline":"Mild diffusion growth yields bounded chemotaxis-fluid solutions","feed_subtitle":"New proof allows singular sensitivity up to exponent 5/6 in a 2D Navier–Stokes coupling.","key_machinery":"The proof revolves around a family of energy-like functionals built from the double primitive $D_{2,\\varepsilon}(n)=\\int_0^n\\int_0^s D_\\varepsilon(\\sigma)\\,d\\sigma\\,ds$ of the diffusion coefficient, weighted gradient integrals of the oxygen concentration, and a truncated primitive $\\Psi$ of the reciprocal diffusion that captures behavior at low densities. For $\\gamma\\le\\tfrac12$ the functional $F_\\varepsilon=\\int D_{2,\\varepsilon}(n_\\varepsilon)+b_1\\int n_\\varepsilon|\\nabla c_\\varepsilon|^2/c_\\varepsilon+b_2\\int|\\nabla c_\\varepsilon|^4/c_\\varepsilon^3+b_3\\int\\Psi(n_\\varepsilon)$ closes the estimates, while for $\\gamma>\\tfrac12$ a slimmer functional without the mixed $n|\\nabla c|^2/c$ term is closed using a Young-splitting and the additional $L^1$ bound on $n_0$. A Trudinger–Moser inequality converts mass and gradient information into time-space control of $n\\ln(n+1)$, an ODE comparison lemma transfers those bounds to the fluid component, and a Moser-type iteration upgrades the bounds to $L^\\infty$; boundary integrals are controlled without convexity via a cited Neumann boundary estimate.","core_discovery":"Theorem 1.1 asserts the following: for every $M>0$ and $\\gamma\\in[0,\\tfrac56]$, if the tensor-valued sensitivity satisfies $|S(x,n,c)|\\le S_0(c)/c^\\gamma$ with non-decreasing $S_0$, and if $D$ is smooth and positive on $(0,\\infty)$ with $\\liminf_{n\\to\\infty}D(n)>L(M)$ and $\\liminf_{n\\to0}D(n)/n>0$, then suitably regular initial data with $\\|c_0\\|_{L^\\infty}\\le M$ (and additionally $\\|n_0\\|_{L^1}\\le M$ when $\\gamma>\\tfrac12$) lead to a global weak solution satisfying uniform $L^\\infty$ bounds on $n$, $\\nabla c$, and $\\nabla u$. If $D(0)>0$, the solution is classical. The theorem covers all smoothly bounded planar domains, not only convex ones, and as a corollary it applies to porous-medium diffusion $D(n)=n^{m-1}$ for $m\\in(1,2]$, yielding continuous solutions.","pith_inferences":["The exponent $\\tfrac56$ appears to be a technical cutoff imposed by the Young-splitting in the proof, so a natural test is whether the same result extends to every $\\gamma<1$ with a sharper interpolation.","The double-primitive energy combined with the small-density correction term may transfer to three-dimensional Navier–Stokes, but the diffusion threshold would likely depend on additional norms of the initial data.","If the quoted boundary estimate turns out to require convexity, the theorem would shrink to convex or weakly convex domains, so verifying or refuting that estimate on concave boundary patches is a concrete way to test the paper's scope."],"forward_implications":["For any fixed upper bound $M$ on the initial oxygen level, choosing diffusion that eventually exceeds $L(M)$ at high density and remains linear near zero guarantees global bounded weak solutions, regardless of the size of other data components (apart from an $L^1$ bound on cell mass when $\\gamma>\\tfrac12$).","The allowed singularity exponent $\\gamma=\\tfrac56$ improves the previous $\\gamma=\\tfrac12$ in the three-dimensional Stokes analogue and applies directly to the two-dimensional Navier–Stokes coupling.","Porous-medium diffusion $D(n)=n^{m-1}$ with $m\\in(1,2]$ falls under the theorem, and the corresponding weak solutions are continuous rather than merely bounded.","If the diffusion coefficient does not degenerate at zero ($D(0)>0$), the obtained solution is classical, with $C^{2,1}$ regularity in space-time and a pressure.","The result holds on arbitrary smoothly bounded planar domains; convexity of the domain is not required."],"supporting_citations":[{"why":"Supplies the energy-like functional built from the double primitive of $D$, the truncated reciprocal diffusion, and the baseline result with $\\gamma=1/2$ in a 3D Stokes system that this paper extends to 2D Navier–Stokes and $\\gamma$ up to $5/6$.","marker":"[52]"},{"why":"Provides the boundary estimate $\\partial|\\nabla w|^2/\\partial\\nu\\le C|\\nabla w|^2$ for Neumann functions $w$, used in Lemmas 3.2 and 3.4 to control boundary integrals and to avoid convexity assumptions.","marker":"[27]"},{"why":"Establishes small-data global classical solutions for the linear-diffusion case with $\\gamma<1$; Theorem 1.1 partially recovers this and removes the smallness restriction under mild diffusion enhancement.","marker":"[26]"},{"why":"Gives global bounded weak solutions for porous-medium diffusion with sufficiently large exponent and singular sensitivity; the paper's corollary extends the same conclusion to $m\\in(1,2]$.","marker":"[41]"},{"why":"Supplies the Poincaré-type inequality for measurable subsets used in Lemma 3.8 to replace convex-domain Poincaré arguments, contributing to the removal of the convexity requirement.","marker":"[23]"},{"why":"Provides the ODE-comparison lemma (Lemma 2.8) that turns spatio-temporal bounds into uniform bounds for the energy functional and the solution components.","marker":"[49]"},{"why":"Contains boundary arguments for non-convex domains that inspire the treatment of the boundary integral in the energy estimates.","marker":"[17]"}],"fun_headline_variants":["2D chemotaxis-fluid: bounded states from mild diffusion","Bounded weak solutions for 2D chemotaxis-Navier-Stokes","Singular sensitivity up to 5/6 tamed in 2D fluid chemotaxis","Mild diffusion and singular sensitivity yield global bounds","Mild diffusion enhancement gives bounded 2D solutions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof's load-bearing premise is a boundary estimate valid for all smoothly bounded domains without convexity: for any $C^2$ function $w$ with zero normal derivative on the boundary, the normal derivative of $|\\nabla w|^2$ is bounded by a constant times $|\\nabla w|^2$ there. That estimate is quoted from a cited lemma; if it fails for some smooth non-convex domain, the boundary integrals in Lemmas 3.2 and 3.4 cannot be absorbed and the energy inequalities do not close.","fun_headline_variants_meta":{"raw":{"variants":["2D chemotaxis-fluid: bounded states from mild diffusion","Bounded weak solutions for 2D chemotaxis-Navier-Stokes","Singular sensitivity up to 5/6 tamed in 2D fluid chemotaxis","Mild diffusion and singular sensitivity yield global bounds","Mild diffusion enhancement gives bounded 2D solutions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001564,"raw_usage":{"total_tokens":6432,"prompt_tokens":1316,"completion_tokens":5116,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":932,"completion_tokens_details":{"reasoning_tokens":5025}},"tokens_in":932,"tokens_out":5116,"duration_ms":31661,"temperature":1.0,"reasoning_tokens":5025,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:17:47.903335+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a smooth bounded domain in the plane with a concave boundary patch and a sequence of functions $w_j\\in C^2(\\overline{\\Omega})$ satisfying $\\partial w_j/\\partial\\nu=0$ on $\\partial\\Omega$; compute the ratio $\\partial|\\nabla w_j|^2/\\partial\\nu$ divided by $|\\nabla w_j|^2$ on the boundary. If this ratio is unbounded, the boundary estimate quoted from [27, Lemma 4.2] fails on non-convex domains, and the boundary integrals used to close the energy estimates cannot be absorbed, so the theorem's claim about arbitrary smoothly bounded domains would not follow.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the energy-like functional built from the double primitive of $D$, the truncated reciprocal diffusion, and the baseline result with $\\gamma=1/2$ in a 3D Stokes system that this paper extends to 2D Navier–Stokes and $\\gamma$ up to $5/6$."},{"cited_title":"Mizoguchi and P","cited_arxiv_id":null,"evidence_quote":"Provides the boundary estimate $\\partial|\\nabla w|^2/\\partial\\nu\\le C|\\nabla w|^2$ for Neumann functions $w$, used in Lemmas 3.2 and 3.4 to control boundary integrals and to avoid convexity assumptions."},{"cited_title":"Globalclassicalsolvabilityandstabilizationinatwo-dimensionalchemotaxis–fluidsystemwithsub-logarithmic sensitivity","cited_arxiv_id":null,"evidence_quote":"Establishes small-data global classical solutions for the linear-diffusion case with $\\gamma<1$; Theorem 1.1 partially recovers this and removes the smallness restriction under mild diffusion enhancement."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives global bounded weak solutions for porous-medium diffusion with sufficiently large exponent and singular sensitivity; the paper's corollary extends the same conclusion to $m\\in(1,2]$."},{"cited_title":"Lankeit and M","cited_arxiv_id":null,"evidence_quote":"Supplies the Poincaré-type inequality for measurable subsets used in Lemma 3.8 to replace convex-domain Poincaré arguments, contributing to the removal of the convexity requirement."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the ODE-comparison lemma (Lemma 2.8) that turns spatio-temporal bounds into uniform bounds for the energy functional and the solution components."},{"cited_title":"Ishida, K","cited_arxiv_id":null,"evidence_quote":"Contains boundary arguments for non-convex domains that inspire the treatment of the boundary integral in the energy estimates."}],"review_version":1}