{"id":"540b8ebc-8833-4f72-ae8c-f4a251e2fe02","arxiv_id":"1908.02216","paper_version":1,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Global well-posedness for the 2D inhomogeneous incompressible Navier-Stokes equations is proved in critical Besov spaces with large initial data, almost the energy space.","lead":"This mathematics paper proves that a fluid model with varying density, the 2D inhomogeneous incompressible Navier-Stokes equations, has a global unique solution starting from large, rough initial data in critical Besov spaces. It removes a longstanding extra regularity assumption and lands at the natural scaling-invariant threshold, nearly the energy space.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The short-time bootstrap in Proposition 3.2 is not closed: making the coefficient in (3.21) small requires assuming smallness of the very quantity being estimated.","rationale":"The reader's verdict identifies the same weakest point: the closure of the short-time bootstrap in Proposition 3.2. My reading of the manuscript confirms that the transition from (3.21) to (3.7) is the essential junction of the proof. The coefficient on the principal unknown contains an exponential of ||grad u||_{L1_t(L^infty)}, which is not a priori small and is controlled only through the quantity being estimated. A continuity/bootstrap argument is needed to justify 'taking t sufficiently small', and this argument must be uniform in the mollification index n for the compactness step in Section 4 to work. The gap is real but probably repairable by a specialist: the heat-kernel term is uniformly small by dominated convergence and the convergence of the mollified data, and the remaining nonlinear terms can be absorbed by a bootstrap if the threshold is chosen appropriately. I do not see an internal inconsistency that would force rejection; the theorem and the overall strategy are plausible. The absence of the bootstrap argument, however, justifies the reader's CONDITIONAL verdict and low confidence. I also checked the apparent mismatch in uniqueness where Proposition 2.2 requires a in L^infty_t(B^2_{2,1}); this is not a genuine flaw because Proposition 2.2 is applied to the smoothed coefficient S_k a_2 in (4.5), which does belong to B^2_{2,1} with a k-dependent bound. The more serious unaddressed point remains the short-time bootstrap, so no change to the reader's verdict is needed.","tokens_in":25245,"tokens_out":20013,"duration_ms":197574,"concrete_test":"Write out the missing bootstrap for Proposition 3.2. Concretely, for the mollified initial data u_{0,n}, define X_n(t)=||u_n||_{L1_t(dot B^2_{2,1})}+||grad u_n||_{L2_t(L^2)}. Establish that there exist delta>0, eta>0, k, and T1>0, all independent of n, such that whenever X_n(t)= T1 for a common T1 does not follow. This is load-bearing for the existence part of Theorem 1.1.","section":"Section 3, end of proof of Proposition 3.2, around (3.21)"},{"comment":"The passage from the mollified initial data to a fixed existence interval is not established. The authors infer (4.2) from Proposition 3.2 and then conclude that the lifespans T*_n of the mollified solutions satisfy T*_n >= T1 for some T1 independent of n. This conclusion requires the uniform-in-n bootstrap described above; otherwise T1 = T1(n) could shrink to 0, and the compactness argument on [0,T1] cannot be performed. A standard fix would be to prove a quantitative version of Proposition 3.2 in which T1 and the constants depend only on the norms of the initial data, or to use a different mechanism (for example, a diagonal compactness argument combined with uniform global bounds from [14,20]) to obtain a solution on a fixed time interval.","section":"Section 4, Step 1, around (4.1)-(4.2)"},{"comment":"The uniqueness proof uses Proposition 2.1 at the endpoint alpha = 1, but Proposition 2.1 is stated only for alpha in [0,1), and the proof of the underlying commutator estimate (Lemma 2.1) restricts to alpha < 1. Specifically, equation (4.3) applies (2.7) to \\dot B^1_{2,1}, and (4.9) requires a tail estimate for a2 - S_j a2 in B^1_{2,1} on a small time interval. If the intended estimate for alpha = 1 is true, it needs to be stated and proved or cited from a source that covers this endpoint; alternatively, one can use the fact that u lies in L1_t(\\dot B^2_{2,1}) subset L1_t(\\dot B^1_{\\infty,1}), so the flow is Lipschitz and standard transport-regularity arguments give a in C([0,T]; B^1_{2,1}) and the needed tail smallness. As written, the reference to Proposition 2.1 does not cover the case used.","section":"Section 4, Step 2, equations (4.3) and (4.9)"}],"minor_comments":[{"comment":"In the sentence 'Therefore, thanks to (3.2) and (3.18), we obtain from (3.19) that ...' the reference to (3.19) appears to be a typo; the inequality being substituted into is (3.17), since (3.19) is the displayed estimate that follows.","section":"Section 3, after (3.17)"},{"comment":"The text refers to 'Lemma 3.2' but the statement in Section 3 is Proposition 3.2; the numbering should be made consistent.","section":"Section 4, Step 1"},{"comment":"After deriving (4.7), the proof says 'Then applying Proposition 3.3 to (4.7) leads to ...' but the displayed estimate with the factor (1 + 2^j ||a2||^2_{...}) and the norms in B^{-2}_{2,\\infty} is precisely Proposition 2.3, not Proposition 3.3. Please correct the citation.","section":"Section 4, Step 2, around (4.7)"},{"comment":"There are malformed double norms in several displays, e.g. 'sup_{q>= -1} 2^{-q} || [v, \\Delta_q P] \\cdot \\nabla u ||_{L^1_T(L^2)}' has an extra bar; these should be cleaned up for readability.","section":"Section 2, proof of Proposition 2.2, (2.12)-(2.13)"},{"comment":"The proof repeatedly invokes 'Lemma 1 of [22]' for commutator estimates involving the Leray projector P. Since this is a key technical tool, the lemma should be stated explicitly in the paper or its exact form should be quoted, rather than referenced in passing.","section":"Section 3 and Section 4, commutator estimates"},{"comment":"The inequality 'ln(e + alpha x^{-1}) <= ln(e + alpha)(1 - ln x)' is stated for alpha >= 0 and x in (0,1]; the argument also needs the resulting factor (1 - ln x) to be positive, so the smallness of ||\\delta u||_{...} should be stated explicitly when applying Osgood's lemma.","section":"Section 4, Step 2, after (4.13)"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is significant and the overall strategy is plausible, but the missing bootstrap in Proposition 3.2 is a genuine gap in the proof of existence, and the endpoint-alpha issue in the uniqueness argument needs to be repaired. Both appear to be fixable within the manuscript's scope, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the paper genuinely extends prior global well-posedness results for the 2D inhomogeneous incompressible Navier-Stokes system: it handles large initial data with velocity in \\dot B^0_{2,1} and density perturbation in \\dot B^\\epsilon_{2/\\epsilon,1}, both critical, without the extra H^s or \\dot B^1 regularity that earlier results required. That is a real advance. Second, the proof is mostly careful, but there is one genuine skipped step in the bootstrap that closes Proposition 3.2.\n\nWhat the paper does well: the main theorem is credible and the Littlewood-Paley machinery is applied seriously. The authors are honest about the main limitation—uniqueness in Theorem 1.1 needs the density in B^1_{2,1}, even though the abstract's phrase 'global unique solvability' omits that caveat. The citation practice is fine; the self-citations to [2,4] are to technical tools for critical-regularity estimates, not to the theorem being proved. The paper is not circular and no constants are fitted.\n\nThe real soft spot is the end of Proposition 3.2, around (3.21). To close the estimate for X(t) = ||u||_{L^1_t(\\dot B^2_{2,1})} + ||\\nabla u||_{L^2_t(L^2)}, the coefficient on the right contains exp(C ||\\nabla u||_{L^1_t(L^\\infty)}), and through the embedding that exponential is controlled by the same X(t). The text says 'taking \\eta small, k large, and t sufficiently small' makes the coefficient absorbable, but that is not immediate: small t only helps if one already has a uniform small bound on X(t) for the family of mollified solutions. The needed continuity/bootstrap argument—X_n(t) tends to 0 as t tends to 0, uniformly in n, and stays below a threshold on a common interval—is not written. This is an omission, not a demonstrated contradiction, and a specialist can probably fill it: the heat-kernel term tends to 0 as t→0 uniformly in n, and the material to make the bootstrap work is present. But as written, the lifespan T1 could depend on the mollification index, which would break the compactness argument in Section 4. A referee should ask for the bootstrap to be written out.\n\nThe uniqueness step also contains a few 'take m large' absorptions that are sketched, but those look routine by comparison. Overall, the central theorem is plausible and important. This paper deserves a serious referee; with the bootstrap made explicit, I would expect it to be accepted.","headline":"First large-data critical-space global well-posedness for 2D inhomogeneous Navier-Stokes; proof is mostly solid but the short-time bootstrap in Proposition 3.2 is sketched, not fully closed.","tokens_in":26244,"tokens_out":6172,"would_cite":true,"duration_ms":62574,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q30","76D05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The 2-D inhomogeneous Navier-Stokes equations have global solutions for large initial data in critical Besov spaces that are 'almost the energy space'.","keywords":["Inhomogeneous Navier-Stokes equations","global well-posedness","critical Besov spaces","large initial data","Littlewood-Paley theory","density-dependent incompressible fluids","uniqueness","Osgood lemma"],"falsifier":"A reader could try to reproduce the closing of (3.21) with explicit constants: for a fixed initial pair $(a_0, u_0)$ with large critical norms, determine whether there exist $\\eta > 0$, $k \\in \\mathbb{Z}$, $t > 0$ such that the coefficient of $\\|u\\|_{L^1_t(\\dot B^2_{2,1})}$ on the right-hand side of (3.21) is at most $1/2$ while the remaining terms are finite; if for some admissible data the absorption condition fails for every choice, then (3.7) is false. Since the paper does not prove the required continuity of $\\|\\nabla u\\|_{L^1_t(L^\\infty)} \\to 0$ as $t \\to 0$ for the approximate solutions, that calculation is the natural place to look.","tokens_in":25005,"feed_emoji":"🌊","tokens_out":17896,"duration_ms":156785,"temperature":0.7,"pith_summary":"The paper tries to establish that the 2-D incompressible inhomogeneous Navier-Stokes system — the standard model for two miscible incompressible fluids with different densities — admits a global-in-time solution for initial data that are large, as long as the data lie in critical Besov spaces whose scaling matches the energy space. Previous global well-posedness results in two dimensions required extra regularity on the density or the velocity; here that extra regularity is dropped, except that uniqueness still asks for a bit more density regularity. The central object is the reformulated unknown $a = 1/\\rho - 1$ transported by the flow, and the proof splits the density-dependent viscous coefficient into a smooth low-frequency part plus a small high-frequency remainder, then closes a short-time bootstrap on the velocity's $L^1_t(\\dot B^2_{2,1})$ norm. A smart reader would care because the result moves well-posedness theory for this system from small-data or extra-regularity regimes to the scaling-optimal critical spaces, the same threshold where the classical energy-space weak solutions live.","feed_headline":"2-D fluid mixing flows solved globally, no smallness needed","feed_subtitle":"New proof reaches critical regularity almost equal to the energy space, where earlier results stalled.","key_machinery":"The mechanism that carries the argument is the decomposition of the density-dependent diffusion coefficient by a dyadic partial sum, a frequency-by-frequency splitting inherited from Littlewood-Paley theory. In the linearized momentum equation $\\partial_t u - (1+a)(\\Delta u - \\nabla\\Pi) = f$, one rewrites $1+a$ as $1 + \\dot S_m a + (a - \\dot S_m a)$, so that the high-frequency remainder $a - \\dot S_m a$ is small in the critical norm (the tail of $a_0$'s Besov norm vanishes as $m \\to \\infty$), while the low-frequency part $\\dot S_m a$ is smooth enough for basic energy estimates. This split feeds a short-time bootstrap (Proposition 3.2) that proves $\\|u\\|_{L^1_t(\\dot B^2_{2,1})} + \\|\\nabla u\\|_{L^2_t(L^2)} \\lesssim \\sum_{j \\in \\mathbb{Z}} (1 - e^{-ct2^{2j}}) \\|\\dot\\Delta_j u_0\\|_{L^2} + \\sqrt{t}$ on a small interval $[0,T_1]$, using Littlewood-Paley commutator estimates, the basic energy inequality, and a pressure bound from Proposition 3.1; this bootstrap is the step that makes the estimate independent of the size of the data. Uniqueness is then closed by an Osgood-type lemma applied to the difference of two solutions, where a logarithmic factor yields the divergent integral $\\int_0^1 \\frac{dx}{x(1-\\ln x)}$ that forces the difference to vanish on a short interval, then inductively for all times.","core_discovery":"The central claim, in the authors' own terms, is Theorem 1.1: for any $\\varepsilon \\in (0,1)$, if $u_0$ is divergence-free in $\\dot B^0_{2,1}(\\mathbb{R}^2)$ and $1/\\rho_0 - 1$ belongs to $\\dot B^\\varepsilon_{2/\\varepsilon,1}(\\mathbb{R}^2)$ with $m \\leq \\rho_0 \\leq M$ for some positive constants $m$ and $M$, then the system has a global solution $(\\rho, u, \\nabla\\Pi)$ with $1/\\rho - 1 \\in C(\\mathbb{R}_+; \\dot B^\\varepsilon_{2/\\varepsilon,1})$, $u \\in C(\\mathbb{R}_+; \\dot B^0_{2,1}) \\cap L^1_{\\rm loc}(\\mathbb{R}_+; \\dot B^2_{2,1})$, and $\\partial_t u, \\nabla\\Pi \\in L^1_{\\rm loc}(\\mathbb{R}_+; \\dot B^0_{2,1})$. If in addition $1/\\rho_0 - 1 \\in B^1_{2,1}$, the solution is unique. The force of the theorem is the 'without smallness' clause: the critical norms of the two initial quantities may be arbitrarily large, and in terms of the scaling transformation (1.4) the space for the density deviation has the same scaling as $L^\\infty$ while the space for the velocity has the same scaling as $L^2$, which is exactly the energy space of the classical weak-solution theory.","pith_inferences":["The low/high-frequency splitting used here should adapt to the variable-viscosity system $\\mu(\\rho)$, since the same coefficient appears in the diffusion term; checking that the commutator and pressure estimates survive the extra factor would test the method's reach.","A natural open question is whether uniqueness can be relaxed to exactly the critical density space $\\dot B^\\epsilon_{2/\\epsilon,1}$; the Osgood argument here needs $B^1_{2,1}$ regularity for the density difference, so the logarithmic loss appears to be the obstacle rather than the existence proof.","One could test the bootstrap quantitatively by computing the constants in (3.21) for simple families of large data; the theorem does not quantify how the lifespan $T_1$ shrinks as the critical norms grow, so a numerical exploration would show whether the absorption step is robust in practice."],"forward_implications":["The 2-D inhomogeneous Navier-Stokes system is globally solvable for arbitrarily large initial velocity in $\\dot B^0_{2,1}$ and density deviation in $\\dot B^\\epsilon_{2/\\epsilon,1}$, with no smallness condition on the data.","Uniqueness holds in the slightly more regular class with $1/\\rho_0 - 1 \\in B^1_{2,1}$, so the large-data solutions are determined by the initial data, not merely existing.","The regularity reached is scaling-optimal: under the scaling transformation (1.4) the spaces for the density deviation and the velocity match the energy spaces $L^\\infty$ and $L^2$, closing the gap between energy-space weak solutions and critical-space strong solutions in two dimensions.","Earlier 2-D well-posedness theorems required extra regularity on the density or the velocity; this result operates at the critical regularity itself, the same threshold where the classical weak-solution theory lives."],"supporting_citations":[{"why":"Supplies the reformulation with the partial-sum split and the 3-D critical-space well-posedness result without size restriction on the density that the paper adapts to 2-D.","marker":"[2]"},{"why":"Provides the Littlewood-Paley machinery, commutator estimates, and transport-diffusion estimates that carry the a priori bounds.","marker":"[6]"},{"why":"Gives the local critical-space theory and the transport-diffusion estimates used for the linearized momentum equations (1.7)-(1.8).","marker":"[11]"},{"why":"Yields the local well-posedness for the mollified initial data used at the start of the existence proof.","marker":"[12]"},{"why":"Used together with [20] to extend the local solution to all times once the velocity reaches H^1 at some positive time.","marker":"[14]"},{"why":"Supplies Osgood's lemma, the divergence-of-the-logarithmic-integral tool that closes the uniqueness proof.","marker":"[17]"},{"why":"Provides the bounded-density global well-posedness result invoked alongside [14] for the global extension step.","marker":"[20]"},{"why":"Supplies the commutator estimate used repeatedly to control the nonlinear terms in the a priori estimates.","marker":"[22]"}],"fun_headline_variants":["2D Navier-Stokes: global well-posedness without smallness","Global solutions for 2D inhomogeneous Navier-Stokes with large data","No smallness requirement for global 2D fluid mixing solutions","Arbitrarily large initial data still yield global 2D solutions","2D fluid mixing: global existence beyond the smallness regime"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the unproven continuity step that closes the short-time bootstrap in Proposition 3.2 — namely that for the mollified approximate solutions the norms $\\|\\nabla u\\|_{L^1_t(L^\\infty)}$ and $\\|\\nabla u\\|^2_{L^2_t(L^2)}$ tend to $0$ as $t \\to 0$ uniformly enough to make the right-hand side of (3.21) absorbable into the left-hand side.","fun_headline_variants_meta":{"raw":{"variants":["2D Navier-Stokes: global well-posedness without smallness","Global solutions for 2D inhomogeneous Navier-Stokes with large data","No smallness requirement for global 2D fluid mixing solutions","Arbitrarily large initial data still yield global 2D solutions","2D fluid mixing: global existence beyond the smallness regime"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00036,"raw_usage":{"total_tokens":1941,"prompt_tokens":932,"completion_tokens":1009,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":548,"completion_tokens_details":{"reasoning_tokens":915}},"tokens_in":548,"tokens_out":1009,"duration_ms":10535,"temperature":1.0,"reasoning_tokens":915,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:51:53.449205+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A reader could try to reproduce the closing of (3.21) with explicit constants: for a fixed initial pair $(a_0, u_0)$ with large critical norms, determine whether there exist $\\eta > 0$, $k \\in \\mathbb{Z}$, $t > 0$ such that the coefficient of $\\|u\\|_{L^1_t(\\dot B^2_{2,1})}$ on the right-hand side of (3.21) is at most $1/2$ while the remaining terms are finite; if for some admissible data the absorption condition fails for every choice, then (3.7) is false. Since the paper does not prove the required continuity of $\\|\\nabla u\\|_{L^1_t(L^\\infty)} \\to 0$ as $t \\to 0$ for the approximate solutions, that calculation is the natural place to look.","supporting_citations":[{"cited_title":"Abidi, G","cited_arxiv_id":null,"evidence_quote":"Supplies the reformulation with the partial-sum split and the 3-D critical-space well-posedness result without size restriction on the density that the paper adapts to 2-D."},{"cited_title":"Bahouri, J","cited_arxiv_id":null,"evidence_quote":"Provides the Littlewood-Paley machinery, commutator estimates, and transport-diffusion estimates that carry the a priori bounds."},{"cited_title":"Danchin, Local theory in critical spaces for compres sible viscous and heat-conductive gases, Comm","cited_arxiv_id":null,"evidence_quote":"Gives the local critical-space theory and the transport-diffusion estimates used for the linearized momentum equations (1.7)-(1.8)."},{"cited_title":"Danchin, Local and global well-posedness resultats for flows of inhomogenenous viscous fluids, Adv","cited_arxiv_id":null,"evidence_quote":"Yields the local well-posedness for the mollified initial data used at the start of the existence proof."},{"cited_title":"Danchin and P","cited_arxiv_id":null,"evidence_quote":"Used together with [20] to extend the local solution to all times once the velocity reaches H^1 at some positive time."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Osgood's lemma, the divergence-of-the-logarithmic-integral tool that closes the uniqueness proof."},{"cited_title":"Paicu, P","cited_arxiv_id":null,"evidence_quote":"Provides the bounded-density global well-posedness result invoked alongside [14] for the global extension step."},{"cited_title":"Planchon, An extension of the Beale-Kato-Majda crit erion for the Euler equations, Comm","cited_arxiv_id":null,"evidence_quote":"Supplies the commutator estimate used repeatedly to control the nonlinear terms in the a priori estimates."}],"review_version":1}