{"id":"7539b51f-4118-41fe-9e62-10cd0eb4fbe3","arxiv_id":"2607.13803","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"high","formal_verification":"none","parameter_count":1,"one_line_summary":"For sufficiently large Prandtl number, 3D non-diffusive Boussinesq flows with large initial data are globally well-posed and temperature patch boundaries preserve C^{1,γ}, W^{2,∞}, and C^{2,γ} regularity, with convergence to the Stokes-transport system as Pr→∞.","lead":"This paper proves that the 3D Boussinesq equations—a model of fluids where temperature differences drive motion—have smooth solutions for all time when the Prandtl number is large enough, even for large initial data. It also shows that sharp \"temperature patch\" surfaces keep their smooth boundaries forever, and that in the infinite-Prandtl limit the flow converges to a simpler Stokes-transport system.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Reader's embedding objection in Eq. (4.22) is a sign error: L∞ ↪ B^{-1/2}_{∞,1} is true. The genuine gap is that the C^{2,γ} proof relies on deferred estimates (2.9) and (2.19), whose proofs are in unpublished companion [44].","rationale":"The reader's central criticism targets Eq. (4.22), claiming that the embedding L∞(R³) ↪ B^{-1/2}_{∞,1}(R³) is false for functions with jump discontinuities. This is incorrect for the nonhomogeneous Besov spaces used in the paper: the norm sums 2^{-q/2}‖Δ_q f‖_{L∞} over q ≥ -1, and the dyadic blocks of an L∞ function are uniformly bounded, so the sum converges. The indicator function of a smooth domain therefore has finite B^{-1/2}_{∞,1} norm. The reader's concern is a sign/index confusion and does not land.\n\nStill, the manuscript is not fully self-contained. The C^{2,γ} boundary-regularity proof in Prop. 4.4 depends crucially on Lemma 2.6(2.9) and Lemma 2.8(2.19). The text explicitly defers the proofs of these estimates to the unpublished companion [44]. Since these estimates are not proved in the paper or in a publicly verifiable source, the uniformity-in-Pr persistence of C^{2,γ} regularity is not yet fully justified as written. This does not establish that the theorem is false; it means the supporting technical infrastructure is incomplete. The reader's verdict of CONDITIONAL is therefore still appropriate, but for a different reason than the one given. I weigh the missing proofs as a genuine but addressable gap rather than a demonstrated contradiction, hence UNCHANGED.","tokens_in":37162,"tokens_out":34774,"duration_ms":279448,"concrete_test":"Obtain the companion [44] (or ask the authors to include the proofs) and check that Lemma 2.8(2.19) follows from the stated commutator estimate (2.20) with no extra smallness assumption, and that Lemma 2.6(2.9) is valid as stated for all ε ∈ (0,1), (p,r) ∈ [1,∞]^2. Specifically, re-derive the two lines used in Prop. 4.4: (i) the bound ‖ΔW·∇Γ‖_{B^{-1}_{∞,∞}} ≤ C‖ΔW‖_{B^{-1}_{∞,∞}}‖∇Γ‖_{B^0_{∞,1}} from (2.9); (ii) the heat-smoothing inequality (2.19) for s = -1, r = 1. If either requires additional regularity of u or θ, or fails to be uniform in Pr, then Theorem 1.1(3) and Theorem 1.4(2) are unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakness flagged by the reader, the embedding used in (4.22), is not false. With the paper's nonhomogeneous Besov definition, every L∞ function satisfies ‖Δ_q f‖_{L∞} ≤ C‖f‖_{L∞}, so ∑_{q≥-1} 2^{-q/2}‖Δ_q f‖_{L∞} ≤ C‖f‖_{L∞}; an indicator function has finite B^{-1/2}_{∞,1} norm. Thus jump discontinuities do not invalidate (4.22).\n\nThe actual load-bearing gap is the incomplete proof of the technical estimates on which the C^{2,γ} persistence rests. Lemma 2.6(2.9) and Lemma 2.8(2.19) are stated as ready-made tools, but their proofs are explicitly deferred to the companion preprint [44]. Estimate (2.19) controls the decompositions F^{(1)}, F^{(2)} of ∂_W Γ in Proposition 4.4 (equations (4.42)–(4.45)) and is what produces the uniform-in-Pr C^γ control needed for (4.46)–(4.47); (2.9) is used just above to control ΔW·∇Γ in B^{-1}_{∞,∞}. Without a verifiable derivation of these two estimates, the central claim that C^{2,γ} patch regularity persists uniformly for all Pr ∈ [Pr*, ∞) is not fully supported by the manuscript. This is a verifiability gap, not a demonstrated contradiction.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the 3D non-diffusive Boussinesq system (B) and proves global existence and uniqueness of strong solutions when the Prandtl number Pr is larger than a threshold depending only on scale-invariant norms of the initial data (Theorem 1.1, Section 3). For non-constant temperature patch initial data, the authors further prove global persistence of C^{1,γ}, W^{2,∞}, and C^{2,γ} regularity of the patch boundary uniformly in Pr ∈ [Pr*,∞): Theorem 1.1(1)-(3), proved in Sections 4.1-4.3. Finally, Theorem 1.4 and Proposition 5.1 justify the infinite-Prandtl limit to the 3D Stokes-transport system and show that the limit patch solution preserves the same boundary regularity. The central technical tool is the 'good unknown' Γ introduced in (1.11), which satisfies the transport-diffusion equation (4.5) and allows uniform-in-Pr estimates.","tokens_in":37621,"tokens_out":9561,"duration_ms":92078,"significance":"If the proof is complete, the result is a significant advance: it gives the first global strong well-posedness for the 3D non-diffusive Boussinesq system with large data in a physically relevant parameter regime, and it provides the 3D analogue of Grayer II's 2D Stokes-transport patch result. The uniform-in-Pr boundary-regularity estimates and the rigorous infinite-Prandtl limit are valuable additions. The paper is generally well structured, and the main geometric framework (admissible conormal vectors, striated estimates, the Γ equation) is appropriate. The main weakness is that several load-bearing commutator and smoothing estimates — Lemma 2.6(2.9), (2.11), and Lemma 2.8(2.19)-(2.20) — are asserted with proofs deferred to an unreviewed companion preprint [44] by the same research group, including co-author J. Yang. Because these estimates are used in the core proofs of Propositions 4.3 and 4.4 and hence in the main persistence theorems, the paper is not self-contained at key points. This is a verifiability gap that must be closed before the results can be considered fully supported.","major_comments":[{"comment":"The product estimate (2.9), the commutator estimates (2.11)-(2.12), and the smoothing estimates (2.19)-(2.20) are all asserted with proofs omitted and referred to the companion preprint [44] ('see e.g. [44]', 'one can see [44] for more details', 'one can see [44] for the details'). These are not cosmetic: (2.9) and (2.19) are used directly in the proof of Proposition 4.4 (equations (4.42)-(4.47)), and (2.11) is used in Proposition 4.3 to control a commutator in the W^{2,∞} persistence proof. Thus the C^{2,γ} and W^{2,∞} boundary-regularity claims of Theorems 1.1 and 1.4 rest on estimates whose derivations are not available in the manuscript. A referee cannot verify such load-bearing steps from a preprint that is not provided and is not peer-reviewed. Please include complete proofs of (2.9), (2.11), (2.19), and (2.20), either in the paper or in a clearly identified, freely accessible comp","section":"§2, Lemmas 2.6 and 2.8"},{"comment":"The proof of the C^{2,γ} persistence relies on the smoothing estimate (2.19) for the low-regularity source in F^{(1)}. This estimate is stated in Lemma 2.8(2) with a proof deferred to [44]; the supporting commutator estimate (2.20) is also deferred. Since Proposition 4.4 is the only place where the central C^{2,γ} uniformity in Pr is established, this is a load-bearing gap. Even if (2.19) is true, the manuscript as submitted does not provide the tools to verify it, and the proof of Theorem 1.1(3) is therefore incomplete.","section":"§4.3, Proposition 4.4"},{"comment":"The global existence and uniqueness part of Theorem 1.1 is only sketched: after the a priori estimates (3.7)-(3.8), the text says that uniqueness follows by adapting [42, Lemma 3.2] and that existence follows by 'a standard approximation process (e.g. see [42])'. This is acceptable if the adaptation is indeed routine, but the paper does not spell out how the H^{1/2} initial data and the L^1∩L^s temperature are handled in the approximation. Given that the main novelty is the large-Prandtl criterion, the authors should either include the approximation/uniqueness argument or give a precise statement of which part of [42] is being used and why it applies verbatim.","section":"§3, global well-posedness"}],"minor_comments":[{"comment":"The reader's concern about L∞↪B^{-1/2}_{∞,1} is not valid. With the nonhomogeneous Besov normalization in Definition 2.5, one has ∑_{q≥-1} 2^{-q/2}‖Δ_q f‖_{L∞} ≤ C‖f‖_{L∞}, since the sum over q≥-1 converges. Thus (4.22) is not invalidated by the discontinuity of the patch indicator. The estimate is legitimate.","section":"Eq. (4.22)"},{"comment":"In the paragraph following (4.44), when p>3 the displayed estimate for the initial-data part uses F^{(1)} in two places, but the equation (4.43) for F^{(2)} is the one with nonzero initial data. This appears to be a typo: the terms with kΔ_{W0}Γ0k_{B^{-1}_{∞,∞}} should involve F^{(2)}. Please correct.","section":"Eq. (4.44) and surrounding paragraph"},{"comment":"The remark about persistence of higher C^{k,γ} and W^{k,∞} regularity refers again to [44] for the proof. Since [44] is unpublished and not provided, this remark should be clearly labelled as conditional on the companion preprint.","section":"Remark 1.2"},{"comment":"The notation 'Ceexp{CT}' is nonstandard and ambiguous: it could mean C exp(C T) or C exp(exp(C T)). Since the paper repeatedly uses Grönwall inequalities in which the right-hand side contains e^{∫‖∇u‖_{L∞}}, please clarify the convention so the reader can track whether constants are single- or double-exponential.","section":"Throughout"},{"comment":"Reference [44] is to a preprint by the same authors (including co-author J. Yang). The manuscript should state whether [44] has been posted to arXiv and, if so, provide the arXiv identifier so that referees and readers can access it.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The core mathematical strategy is plausible and the results are potentially important, but the paper's dependence on the unreviewed companion preprint [44] for several load-bearing estimates is a serious verifiability problem. I would recommend requiring the authors to either include full proofs of (2.9), (2.11), (2.19), (2.20) or to resubmit with these as explicitly stated assumptions whose proofs are supplied. The reader's original embedding objection appears to be a false alarm; the real gap is the companion dependence. I also note the self-citation pattern: [44] is by the same group and includes the second author, so the referee should verify that the deferred estimates are not circular. The manuscript otherwise reads as a well-motivated and technically sophisticated contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. First, the paper is a genuine advance: it gives the first global large-data well-posedness for 3D non-diffusive Boussinesq at large Prandtl number, and the patch-boundary persistence plus the Pr→∞ limit to Stokes–transport is a real 3D analogue of Grayer II. Second, the reader's strongest objection is wrong: the embedding L∞↪B^{-1/2}_{∞,1} in (4.22) is true with the nonhomogeneous Besov definition used here, because Δ_q f is controlled uniformly in L∞ for f∈L∞ and the weight 2^{-q/2} sums. So the indicator data is not a problem at that step.\n\nWhat the paper does well: the Γ-good-unknown decomposition is coherent, the bootstrap for the large-Prandtl global existence is standard but carefully set up, and the geometric persistence arguments in §4 are structured sensibly around the conormal frame. If the quoted auxiliary estimates are correct, Theorem 1.4 looks like the right way to connect Boussinesq patches to Stokes-transport patches.\n\nThe real soft spot is verifiability, not a false embedding. Lemma 2.6 (2.9), (2.11)–(2.12) and Lemma 2.8 (2.19) are stated as tools but their proofs are explicitly deferred to [44], an unpublished preprint sharing a co-author. These are not incidental. (2.9) is what handles ΔW·∇Γ in Proposition 4.4; (2.19) is what gives the uniform-in-Pr smoothing bounds for F^{(1)} and F^{(2)}; (2.11)–(2.12) are used in the curvature persistence. Without seeing [44], a referee cannot certify the C^{2,γ} persistence or the upper C^{1,γ} range. The estimates look plausible — they are in the standard Littlewood-Paley toolbox — but 'plausible' isn't 'verified'. There are also minor typos, e.g. F^{(1)} appears where F^{(2)} is meant near (4.45).\n\nWho should read it: researchers in Boussinesq and patch regularity. I'd send this to peer review rather than desk reject; the potential significance is high and the structure is coherent enough that the gaps are assessable if the referee has [44] in hand. Ask the authors for the companion preprint.","headline":"Genuinely new large-Prandtl 3D Boussinesq patch result, but the written proof depends on deferred estimates in an unpublished companion, so referees need that preprint in hand.","tokens_in":38118,"tokens_out":6880,"would_cite":true,"duration_ms":64587,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q35","76D03","35Q86"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the 3D non-diffusive Boussinesq system has unique global strong solutions whenever the Prandtl number is large enough, and that temperature patch boundaries keep their Hölder or W2,∞ regularity through the infinite-Pr","keywords":["Boussinesq system","temperature patches","Prandtl number","global well-posedness","boundary regularity","Stokes-transport system","Besov spaces","infinite Prandtl limit"],"falsifier":"Compute the B^{-1/2}_{∞,1}(R^3) norm of the characteristic function 1_{D0} of a smooth bounded domain. The high-frequency Littlewood–Paley blocks are of order one, so the sum defining this norm diverges; this shows the embedding invoked to derive estimate (4.22) cannot hold for patch data. A decisive check is whether some alternative estimate supplies the same L^p control of the commutator [R^{-1}, u·∇]θ; without such a replacement, the global boundary-regularity results lack support.","tokens_in":37012,"feed_emoji":"🌡️","tokens_out":9764,"duration_ms":83625,"temperature":0.7,"pith_summary":"This paper aims to show that a large Prandtl number can substitute for small initial data in the 3D non-diffusive Boussinesq system, the standard convection model in which temperature is advected without diffusion. Its main theorem states that whenever Pr exceeds a threshold set by a scale-invariant norm of the initial data—the sum of the L3,∞ norm of the velocity and the L1 norm of the temperature—a unique global strong solution exists. For non-constant temperature patch data, where the initial temperature is a smooth density times the indicator of a bounded domain, the patch boundary keeps its C1,γ, W2,∞, or C2,γ regularity for all time, uniformly in the large-Prandtl range. A second theorem justifies the limit Pr→∞, showing that the Boussinesq patch solutions converge to the unique patch solution of the 3D Stokes-transport system with the same boundary regularity. This matters because large Prandtl numbers occur in real fluids and because the result offers a path around the smallness assumptions that have blocked global regularity in three dimensions.","feed_headline":"Large Prandtl number gives global 3D Boussinesq solutions","feed_subtitle":"Temperature patch boundaries keep their C1,γ, W2,∞, and C2,γ regularity for all time.","key_machinery":"The load-bearing object is the good unknown Γ := Ω − (R^{-1,2}θ, −R^{-1,1}θ, 0)^t, the difference between vorticity and a Riesz-potential expression of temperature; it satisfies (1/Pr)(∂tΓ + u·∇Γ) − ΔΓ = (1/Pr)Ω·∇u + (1/Pr)[R^{-1}, u·∇]θ. All regularity bounds reduce to estimates on Γ and its tangential derivatives through the Biot–Savart identity, which splits ∇u into a part controlled by Γ and parts controlled directly by θ. For boundary regularity, an admissible system of five divergence-free conormal vector fields W tangent to the initial boundary is transported by the flow; a lemma converts striated regularity of W into C^{k,γ} or W^{k,∞} regularity of the patch boundary. Large Pr enter","core_discovery":"The central claim is Theorem 1.1: for divergence-free u0∈H^{1/2}(R^3) and θ0∈L^1∩L^s(R^3), s>3, the condition Pr ≥ (‖u0‖_{L^{3,∞}} + ‖θ0‖_{L^1})/c* with a universal small constant c* guarantees a unique global strong solution of the Boussinesq system. The new input is a global a priori bound u∈L∞_T(H^{1/2})∩L^2_T(H^{3/2}) obtained from an energy argument and an L∞_T(L^{3,∞}) estimate of u; once that is in hand, uniqueness follows by adapting the existing 3D framework. For temperature patch data θ0 = θ̄0 1_{D0}, the paper proves that the transported boundary ∂D(t) = X_t(∂D0) remains in C^{1,γ}, W^{2,∞}, and C^{2,γ} with bounds uniform in Pr∈[Pr*,∞). Theorem 1.4 passes to the limit Pr→∞ and id","pith_inferences":["If the mechanism is as robust as the proof suggests, the same large-Prandtl idea should transfer to bounded domains or periodic boxes, where the authors note the argument naturally extends.","A concrete next step is to upgrade the boundary-regularity persistence to C^{k,γ} for all k≥3 with uniform-in-Pr estimates, which the authors sketch in a remark but do not fully carry out here.","The convergence proof requires well-prepared initial data; removing this well-preparedness would make the infinite-Prandtl limit a more robust selection principle for 3D convective flows.","The proof's dependence on the L3,∞ norm of the velocity suggests that a mild-solution formulation using the heat semigroup might work in the large-Prandtl regime, potentially opening a route to data that are only locally integrable."],"forward_implications":["Global well-posedness holds for the 3D non-diffusive Boussinesq system with large initial data, provided the Prandtl number is above the stated scale-invariant threshold; no smallness of data is required.","Temperature patch boundaries of class C1,γ, W2,∞, or C2,γ remain in that class globally in time, uniformly over all Pr in [Pr*, ∞).","The infinite-Prandtl limit is rigorously justified: Boussinesq patch solutions converge to the unique patch solution of the 3D Stokes-transport system, with the same boundary regularity persistence.","The threshold depends only on a scale-invariant norm of the initial data, so the result identifies a physically meaningful large-parameter regime for convection models rather than a small-data regime.","The theorem provides a genuinely 3D analogue of the earlier 2D Stokes-transport patch regularity result, extending the known theory to whole-space R^3."],"fun_headline_variants":["Large Prandtl number yields global 3D Boussinesq solutions","3D Boussinesq global strong solutions for large Prandtl number","Temperature patches stay smooth globally in 3D Boussinesq","Global 3D Boussinesq solutions from large Prandtl number"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The argument hinges on a frequency-space commutator bound that assumes the temperature is smooth enough to lie in a certain Besov space; for a temperature patch, whose temperature is a step function across the boundary, that assumption is invalid, and the regularity-persistence estimates collapse unless the bound is replaced by a correct estimate.","fun_headline_variants_meta":{"raw":{"variants":["Large Prandtl number yields global 3D Boussinesq solutions","3D Boussinesq global strong solutions for large Prandtl number","Temperature patches stay smooth globally in 3D Boussinesq","Global 3D Boussinesq solutions from large Prandtl number"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000528,"raw_usage":{"total_tokens":2498,"prompt_tokens":977,"completion_tokens":1521,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":721,"completion_tokens_details":{"reasoning_tokens":1449}},"tokens_in":721,"tokens_out":1521,"duration_ms":46571,"temperature":1.0,"reasoning_tokens":1449,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T03:41:06.484795+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the B^{-1/2}_{∞,1}(R^3) norm of the characteristic function 1_{D0} of a smooth bounded domain. The high-frequency Littlewood–Paley blocks are of order one, so the sum defining this norm diverges; this shows the embedding invoked to derive estimate (4.22) cannot hold for patch data. A decisive check is whether some alternative estimate supplies the same L^p control of the commutator [R^{-1}, u·∇]θ; without such a replacement, the global boundary-regularity results lack support.","supporting_citations":[],"review_version":1}