{"id":"512bdbb7-b64b-4b65-ac09-dbf989b00a71","arxiv_id":"2501.10331","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Small H^{1/2} initial data and small multiplicative noise give global-in-time stochastic Navier-Stokes solutions with probability arbitrarily close to 1 on the three-dimensional torus.","lead":"This mathematics paper shows that the stochastic Navier-Stokes equations on a three-dimensional torus have solutions that keep existing forever with probability as close to 1 as desired, provided the initial fluid motion and the random noise are small enough. It reaches a key regularity threshold for these fluid equations, the critical H^{1/2} space, that earlier stochastic results had missed.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.5's proof assumes unavailable H^{1/2+δ}/H^{3/2+δ} bounds on u^(k-1), invalidating the uniform energy estimate used to control stopping times.","rationale":"Analyzing the proof of Theorem 2.1, the most load-bearing point is the uniform H^{1/2+δ} energy control in Lemma 4.5, because it is used to keep the stopping times τ_k and ρ_k finite with small probability and to pass to the limit. The proof's estimate for the cross term I2 appears to require bounds on u^(k-1) that are not present in the hypothesis (4.29). This is an internal derivation gap, distinct from the imported Lemma 4.8 that the reader flagged, though both undermine the induction. The claim may be repairable—for example, by adding a higher-regularity induction hypothesis or by proving the needed control via the cutoffs—but the current text does not justify it. Since this is a proof gap rather than a demonstrated counterexample, the appropriate verdict remains conditional acceptance pending major revision.","tokens_in":24021,"tokens_out":14798,"duration_ms":117489,"concrete_test":"Verify the last inequality in (4.39) by constructing a counterexample: choose u(t) with sup_t‖u‖_{H^{1/2}} + ‖u‖_{L^2_t H^{3/2}} ≤ ε but ‖u‖_{L^2_t H^{3/2+δ}} arbitrarily large (e.g., high-frequency sinusoidal data nonzero only on a short time interval), and take v^(k) satisfying the cutoff constraints. If the integral of ‖u‖_{H^{3/2+δ}}^2 can be large while (4.29) holds, the step is false.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 4.5 claims a k-independent H^{1/2+δ} energy bound for v^(k) under hypothesis (4.29), which only controls u^(k-1) in sup_t H^{1/2} and L^2_t H^{3/2}. In the proof, the cross term I2 is estimated via (3.7) and then the last line of (4.39) bounds the expression by ε^2 (sup_t‖v^(k)‖_{H^{1/2+δ}}^2 + ∫‖v^(k)‖_{H^{3/2+δ}}^2). This requires ‖u^(k-1)‖_{H^{1/2+δ}} ≤ ε and ‖u^(k-1)‖_{L^2_t H^{3/2+δ}} ≤ ε, but these are not implied by (4.29): H^{3/2+δ} is strictly stronger than H^{3/2}, and H^{1/2+δ} cannot be bounded pointwise by the H^{1/2} sup plus the L^2 H^{3/2} norm. Moreover, u^(k-1) is the sum of v^(j) for j<k whose H^{1/2+δ} norms are allowed to grow with k (via M_j in (4.5)), so the missing smallness cannot be recovered. Lemma 4.5's estimate feeds into Lemma 4.9 to make the probabilities of τ_k and ρ_k summable; without it, the nearly global existence argument collapses.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims almost global existence for the 3D stochastic Navier-Stokes equations with multiplicative noise and small H^{1/2} initial data on the torus. The proof decomposes the initial datum into a series of smoother pieces, solves a hierarchy of truncated Navier-Stokes-like systems with time-dependent cutoffs, and shows that the limit solves the original equation up to a stopping time that is almost surely positive and finite with probability arbitrarily close to 1. A second theorem gives local existence on a deterministic time interval when the noise is not small. The main novelty is the use of H^{1/2+δ} regularity for each piece while maintaining H^{1/2} smallness of the sum.","tokens_in":24525,"tokens_out":10073,"duration_ms":89805,"significance":"If the argument were complete, the result would be a substantial advance: it extends almost global existence for the SNSE from the L3 setting of the authors' earlier work to the critical H^{1/2} Hilbert setting, and the second theorem provides a conditional local-existence result without smallness of the noise. The paper is clearly written and the overall strategy is coherent. However, the current manuscript does not establish the stated theorems because two load-bearing lemmas are either proved with an unjustified step or deferred to an unpublished companion preprint.","major_comments":[{"comment":"The last inequality in (4.39) is unjustified. The chain estimates require, for the cross term I2, a bound of the form ||u^{(k-1)}||_{L^\\infty_t H^{1/2+\\delta}} \\lesssim \\varepsilon and ||u^{(k-1)}||_{L^2_t H^{3/2+\\delta}} \\lesssim \\varepsilon. Hypothesis (4.29) controls only the H^{1/2} sup norm and the L^2_t H^{3/2} norm of u^{(k-1)}; these are strictly weaker than the required H^{1/2+\\delta} and H^{3/2+\\delta} norms. Moreover, u^{(k-1)} is the sum of the preceding v^{(j)}'s, whose H^{1/2+\\delta} norms are allowed to grow via the constants M_j in (4.5) and are not small. Therefore the claimed k-independent estimate (4.37) is not established. Lemma 4.9 then uses (4.37) in (4.45) to make the probabilities P(ρ_k < ∞) summable; without a valid proof of Lemma 4.5, the proof of Theorem 2.1 collapses. This is a load-bearing gap that must be repaired by either a correct estimate or a substantially modified argument.","section":"§4.2, Lemma 4.5, Eq. (4.39)"},{"comment":"The pointwise H^{1/2} control (4.41) is asserted with the proof delegated to [KX2, Lemma 3.5] and a sentence saying that the argument follows upon replacing the L^3 norm by Q_{k,0}(t). This lemma is load-bearing: it is the mechanism that supplies hypothesis (4.29) for the inductive step and that ensures the infinite series \\sum_k v^{(k)} converges in the correct spaces. The cutoffs in the present paper, (4.8)–(4.9), differ from those of [KX2] by incorporating a time-integrated dissipation term, and the manuscript does not demonstrate that the 'energy decays once the threshold is exceeded' argument works for these new cutoffs. Since [KX2] is an unpublished preprint and the adaptation is nontrivial, the proof of Lemma 4.8 must be included in full or replaced by a verifiable argument.","section":"§4.2, Lemma 4.8, Eq. (4.41)"},{"comment":"The decomposition of the initial data into the series (4.1)–(4.4) is stated with the proof 'omitted' and a reference to [KX2, Lemma 3.1]. This decomposition is the foundation of the entire construction, including the geometric decay in (4.3) and the existence of the constants ~M_k in (4.5). The omission means the paper is not self-contained at a critical point. Even if the proof is genuinely analogous, the authors should either provide it or state precisely which result from [KX2] is being invoked and verify that all hypotheses, including the H^{1/2+\\delta} bounds, are satisfied.","section":"§4.1, Lemma 4.1"}],"minor_comments":[{"comment":"The proof of Lemma 4.3 is headed 'Proof of Lemma 4.2'; this is a typo and should be corrected.","section":"§4.1, proof of Lemma 4.3"},{"comment":"In the displayed sums after (4.45), the notation 'Q_{k,\\delta}(t)' should be 'Q_{j,\\delta}(t)' in the first sum, and correspondingly for the H^{1/2} term the index should be j; the current notation makes the sum over j unclear.","section":"§4.2, proof of Lemma 4.9"},{"comment":"The paragraph beginning 'Recalling (4.41), we obtain that u^{(k)} has a limit u ...' is very terse. The convergence in L^\\infty_t H^{1/2} \\cap L^2_t H^{3/2} and the passage to the limit in the stochastic integral in (5.3) should be spelled out, especially because the stopping time τ is common to all terms.","section":"§5, convergence passage"},{"comment":"The statement that the solution is probabilistically strong is clear, but the uniqueness claim in the final paragraph of Section 5 is only sketched; a precise uniqueness statement for solutions on [0,τ] with the given regularity would be helpful.","section":"§2, Theorem 2.1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript relies heavily on the companion preprint [KX2] for Lemma 4.1 and especially Lemma 4.8, while Lemma 4.5 contains a genuinely unjustified estimate. The editorial decision should take into account that the current version is not self-contained and that the main theorem is not proven as written. If the authors can supply a correct proof of Lemma 4.5 and a full proof of Lemma 4.8 in the present setting, the paper could be appropriate for publication, but substantial revision is required."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know about arXiv:2501.10331. It targets the critical H^{1/2} endpoint for 3D stochastic Navier-Stokes with multiplicative noise and claims almost global existence for small data and small noise, plus a local-in-time version without small noise. The strategy is the infinite decomposition of the data from KX2, adapted to Hilbert spaces with new cutoffs that incorporate the dissipation integral. If the proof holds, it is a real advance over Kim's subcritical s>1/2 and complements the L3 result from KX2. The paper is clearly written and honest about what it borrows.\n\nThe soft spots are central, not cosmetic. Lemma 4.5 claims a k-independent H^{1/2+δ} energy bound for v(k) assuming (4.29), which controls u(k-1) only in sup_t H^{1/2} and L^2_t H^{3/2}. In the proof of the cross term I2, after the Sobolev estimate, the expression is bounded using smallness of ||u(k-1)||_{H^{1/2+δ}} and ||u(k-1)||_{H^{3/2+δ}}. Those bounds are not consequences of (4.29), and interpolation does not make them small on the full time interval. Also u(k-1) is a sum of previous v(j), whose H^{1/2+δ} norms are allowed to grow via the M_j parameters. So the estimate in Lemma 4.5 is not justified as written. This matters: Lemma 4.5 feeds directly into Lemma 4.9, where the probabilities of τ_k and ρ_k are summed. Without a valid uniform energy bound, the almost global existence argument collapses.\n\nThe second issue is Lemma 4.8, the pointwise H^{1/2} control that keeps the sum of the pieces small. The proof is not reproduced; the text says it follows from [KX2, Lemma 3.5] upon replacing the L3 norm with Q_{k,0}(t). KX2 is an unreviewed preprint, and the cutoffs in this paper are structurally different, so this is a load-bearing omission, not a mere citation.\n\nThere are also minor typos (the proof of Lemma 4.3 is labeled 'Proof of Lemma 4.2'), but those are not substantive.\n\nThe central claim is plausible and the structure looks repairable. The paper is not confused; it is a serious attempt at a natural endpoint. The references are appropriate, and the dependence on KX2 is methodological rather than circular.\n\nI would send this to a serious referee, not desk reject it. Ask the authors to fix Lemma 4.5 and either prove Lemma 4.8 in detail or import the precise statement with a self-contained argument. I would not cite the paper in its present form; once repaired it would be worth citing. I might bring it to the reading group to discuss the decomposition method, but not for its conclusions.","headline":"A plausible attack on the critical H^{1/2} endpoint for stochastic Navier-Stokes, but Lemma 4.5 has a genuine gap and load-bearing material is imported from an unreviewed preprint; repairable but not established as written.","tokens_in":24926,"tokens_out":5746,"would_cite":false,"duration_ms":49040,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H15","35Q30","76D05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Small $H^{1/2}$ data plus small noise give almost-global stochastic Navier-Stokes solutions.","keywords":["stochastic Navier-Stokes equations","multiplicative noise","H^{1/2} initial data","almost global existence","infinite decomposition","stopping times","critical spaces","3D torus"],"falsifier":"One concrete check would be to investigate the truncated system (4.28) with the cutoffs (4.8)–(4.9) and determine whether $\\sup_{\\Omega \\times [0,\\infty)} Q_{k,0}(t)$ can exceed $\\bar{\\epsilon}/2^{k-1}$ for admissible data and noise; exhibiting such an example would disprove Lemma 4.8 and with it the probability bound $\\mathbb{P}(\\tau < \\infty) \\le p_0$. A positive check would supply the missing proof that the energy-decay argument transfers to the $H^{1/2}$ dissipation cutoffs.","tokens_in":23827,"feed_emoji":"🌊","tokens_out":13061,"duration_ms":100078,"temperature":0.7,"pith_summary":"The paper establishes that the stochastic Navier-Stokes equations with multiplicative noise on the three-dimensional torus possess a unique strong solution that exists for all time with probability arbitrarily close to one, provided the $H^{1/2}$ norm of the divergence-free initial datum and the noise coefficient's Lipschitz constant are sufficiently small. This is the stochastic counterpart of the classical small-data global well-posedness result in the critical space $H^{1/2}$, and it reaches lower regularity than earlier stochastic results that required $H^s$ with $s>1/2$ or $H^1$ data. The proof decomposes the initial datum into an infinite series of smooth pieces, solves a truncated difference equation for each piece, and keeps every piece small in $H^{1/2}$ with an explicit geometric decay, so the infinite sum converges to a global solution. A second theorem removes the small-noise assumption and still yields a strong solution on any fixed finite time interval with probability arbitrarily close to one.","feed_headline":"Small data plus small noise: almost-global stochastic Navier-Stokes","feed_subtitle":"Splitting the initial data into tiny smooth pieces keeps the solution global with probability arbitrarily close to 1.","key_machinery":"The load-bearing construction is an infinite decomposition $u_0 = v_0^{(0)} + v_0^{(1)} + \\cdots$ in $H^{1/2}$, where each $v_0^{(k)}$ is smooth, divergence-free, of size at most $\\epsilon_0/4^k$ in $H^{1/2}$, and of controlled size $M_k$ in $H^{1/2+\\delta}$. Each piece $v^{(k)}$ solves a truncated difference equation whose nonlinearity and noise are multiplied by cutoffs $\\psi_k$ and $\\varphi_k$; the cutoffs vanish once the $H^{1/2+\\delta}$ size exceeds $M_k$ or once the quantity $Q_{k,0}(t) = \\|v^{(k)}(t)\\|_{H^{1/2}} + (\\int_0^t \\|v^{(k)}(s)\\|^2_{H^{3/2}}\\,ds)^{1/2}$ exceeds $\\bar{\\epsilon}/2^k$. The central estimate, Lemma 4.8, asserts the pointwise bound $Q_{k,0}(t) \\le \\bar{\\epsilon}/2^{k-1}$ uniformly in time and realization, which keeps the assembled sum $u = \\sum_k v^{(k)}$ small in $H^{1/2}$. Markov's inequality then shows the event that any piece reaches its cutoff occurs with probability at most $p_0/2^{2k+2}$, so the common stopping time $\\tau = \\inf_k \\tau_k$ is finite with probability at most $p_0$, while positivity of $\\tau$ follows from the same energy estimates on short time intervals.","core_discovery":"On the torus, if the divergence-free, zero-average initial datum satisfies $\\sup_\\Omega \\|u_0\\|_{H^{1/2}} \\le \\epsilon_0$ and the multiplicative noise coefficient satisfies a small Lipschitz bound in $H^{1/2}$ and $H^{1/2+\\delta}$, then for every $p_0 \\in (0,1]$ there is a stopping time $\\tau$ and a unique probabilistically strong solution $(u,\\tau)$ of the stochastic Navier-Stokes equations with $\\mathbb{E}[\\sup_{0\\le t\\le \\tau} \\|u(t)\\|^2_{H^{1/2}} + \\int_0^\\tau \\|u(t)\\|^2_{H^{3/2}}\\,dt] \\le C \\epsilon_0^2$ and $\\mathbb{P}(\\tau < \\infty) \\le p_0$. In plain terms, the solution never blows up before time infinity with probability at least $1-p_0$. If the noise is not assumed small, the same machinery produces a solution on any prescribed deterministic interval $[0,T]$ with probability at least $1-p_0$, with the energy bound now depending on $T$.","pith_inferences":["Because the decomposition and cutoffs are the $H^{1/2}$ Hilbert-space version of an earlier $L^3$-based scheme, the same proof strategy plausibly transfers the almost-global small-data conclusion to critical Besov or $L^3$ data, where the analogous pointwise control would be the missing ingredient.","The constants in the stopping-time estimate are explicit through Markov's inequality, so the result could be made quantitative: from $p_0$ one can read off how small $\\epsilon_0$ must be relative to $\\bar{\\epsilon}$ and how large $M_k$ must be relative to the $H^{1/2+\\delta}$ sizes of the pieces.","The second theorem suggests that smallness of the noise is needed only for the global-in-time statement, not for well-posedness on a finite horizon; this may matter for applications that fix an observation window, such as filtering or control, where arbitrarily large noise can be accommodated.","The unverified pointwise control is a natural target for future work: supplying a self-contained proof for the new dissipation cutoffs would remove the paper's main imported assumption, while a counterexample would expose a gap in the induction."],"forward_implications":["For every $p_0 \\in (0,1]$, there is a smallness threshold $\\epsilon_0$ such that all divergence-free zero-average $H^{1/2}$ data below the threshold and all sufficiently small multiplicative noise coefficients admit a unique strong solution whose stopping time is finite with probability at most $p_0$.","The solution satisfies an explicit energy bound: the expected value of $\\sup_{0\\le t\\le \\tau} \\|u(t)\\|^2_{H^{1/2}}$ plus the $H^{3/2}$ dissipation integral up to $\\tau$ is at most $C\\epsilon_0^2$.","When the noise coefficient is not small, the same construction gives a unique strong solution on any fixed interval $[0,T]$ with probability arbitrarily close to one, provided only the $H^{1/2}$ data are sufficiently small.","The almost-global statement extends to initial data in $L^1(\\Omega; H^{1/2})$ by a Markov-truncation argument, so integrable random data are covered as well."],"supporting_citations":[{"why":"Supplies the infinite-decomposition method and, by citation, the pointwise $H^{1/2}$ control (Lemma 4.8) on which the induction rests.","marker":"[KX2]"},{"why":"Establishes the deterministic local and small-data global well-posedness in $H^{1/2}$ that the paper extends to multiplicative noise.","marker":"[FK]"},{"why":"Provides the stochastic heat equation estimates used as Lemma 3.1 to solve each truncated difference equation.","marker":"[KXZ]"},{"why":"Gives earlier local existence for $H^s$ data with $s>1/2$ and a small-data large-probability global result that the present paper improves to the critical $H^{1/2}$ regularity.","marker":"[Ki]"},{"why":"Constructs maximal strong solutions for stochastic Navier-Stokes in bounded domains with $H^1$ data, the baseline strong-solution framework used here.","marker":"[GZ]"},{"why":"Supplies the linear stochastic evolution theory used in the mollification argument within Lemma 3.1.","marker":"[LR]"}],"fun_headline_variants":["Stochastic Navier-Stokes: global solutions with near certainty","High-probability global existence for small stochastic data","Small initial data and noise: almost-global stochastic Navier-Stokes","Near-certain global existence for stochastic Navier-Stokes","Tiny data and noise make stochastic Navier-Stokes global"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rests on Lemma 4.8, which asserts that every truncated piece $v^{(k)}$ obeys the pointwise $H^{1/2}$ control $Q_{k,0}(t) \\le \\bar{\\epsilon}/2^{k-1}$ for all times and all realizations; the paper cites this as following from earlier work and does not reproduce the proof for the new cutoffs, so the induction and limit construction collapse if that control fails.","fun_headline_variants_meta":{"raw":{"variants":["Stochastic Navier-Stokes: global solutions with near certainty","High-probability global existence for small stochastic data","Small initial data and noise: almost-global stochastic Navier-Stokes","Near-certain global existence for stochastic Navier-Stokes","Tiny data and noise make stochastic Navier-Stokes global"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000259,"raw_usage":{"total_tokens":1549,"prompt_tokens":875,"completion_tokens":674,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":491,"completion_tokens_details":{"reasoning_tokens":589}},"tokens_in":491,"tokens_out":674,"duration_ms":6891,"temperature":1.0,"reasoning_tokens":589,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T19:12:39.448194+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete check would be to investigate the truncated system (4.28) with the cutoffs (4.8)–(4.9) and determine whether $\\sup_{\\Omega \\times [0,\\infty)} Q_{k,0}(t)$ can exceed $\\bar{\\epsilon}/2^{k-1}$ for admissible data and noise; exhibiting such an example would disprove Lemma 4.8 and with it the probability bound $\\mathbb{P}(\\tau < \\infty) \\le p_0$. A positive check would supply the missing proof that the energy-decay argument transfers to the $H^{1/2}$ dissipation cutoffs.","supporting_citations":[],"review_version":1}