{"id":"98314b4c-4b6c-4c90-8433-e0d61d1ee23a","arxiv_id":"2607.23234","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For bounded Lipschitz domains whose boundary charts have sufficiently small Sobolev multiplier norm, the Navier–Stokes equations with nonhomogeneous Dirichlet data admit a unique local-in-time very weak solution, and a global one under small data.","lead":"This paper proves that the Navier–Stokes equations have unique 'very weak' solutions on certain rough three-dimensional domains, extending results previously restricted to smoother C^{2,1} boundaries. The proof uses the Stokes semigroup and a duality-based formulation to avoid trace integration on irregular boundaries, then solves an equivalent integral equation by contraction.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1 rests on Proposition 2.3, whose Stokes estimates are only cited from an unpublished preprint; if that toolkit is wrong or not uniform in p, the fixed-point proof collapses.","rationale":"The reader's weakest assumption identifies exactly the point I find most load-bearing: Proposition 2.3 is the black box on which all of Sections 3–6 depend, and the paper's Appendix A is only a sketch that defers to an unpublished preprint. The rest of the proof appears internally consistent: the index relations check out (e.g., γ=1/2+3/(2q)=1−1/s for the HLS argument in Lemma 5.1), the duality manipulations use the stated identities, and the fixed-point contraction and interval-partition uniqueness argument are standard. Thus the appropriate disposition is not rejection but conditionality: the central claim is plausible and formally coherent conditional on Proposition 2.3. I also note a secondary issue: the abstract's advertised special case 'bounded Lipschitz domains with sufficiently small Lipschitz constants' is not proved, and the multiplier norm in Definition 2.1 is not automatically controlled by a small Lipschitz constant alone. This does not change the verdict because the main theorem is stated for the multiplier class itself, but it is another reason to keep the paper conditional until Proposition 2.3 and the domain-class inclusions are substantiated.","tokens_in":18375,"tokens_out":19217,"duration_ms":211123,"concrete_test":"Obtain the Breit–Gaudin preprint and verify, with full proofs, the following for domains of class M_W^{1+α,ρ}(ε): (i) D(A_p^{1/2}) = W^{1,p}_{0,σ} with norm equivalence for all p∈{q,q′,r,r′}, including p=r′ close to 1; (ii) the maximal regularity estimate (2.8); and (iii) the gradient estimate (2.10) with the stated exponent, with constants independent of p and valid for one common ε0 independent of α,ρ. If [5] requires additional regularity, or permits ε to depend on p, α, or ρ, then Theorem 1.1 does not follow as stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Every subsequent step in the paper requires Proposition 2.3: the Helmholtz decomposition used in the Hodge decomposition (3.8), the auxiliary Stokes solves in Propositions 3.1–3.2, the linear theory in Proposition 4.2, the definition of R_q in (5.1)–(5.2), and the K-estimate in Lemma 5.1. In particular, the square-root identification D(A_p^{1/2}) = W^{1,p}_{0,σ}, the maximal regularity (2.8), and the gradient estimate (2.10) are used essentially. The paper does not prove any of these; Appendix A only says they follow from Breit–Gaudin [arXiv:2511.19091], with ε chosen 'sufficiently small'. Since [5] is an unpublished preprint, the central claim is conditional on the truth and uniformity of an external theorem. Concretely, (2.10) is what makes u⊗u pair with ∇Φ and gives the contraction in Section 6; if the gradient estimate fails for p=q' with the stated exponent γ=1/2+3/(2q), Lemma 5.1 and the fixed-point argument collapse. Moreover, Theorem 1.1 states a universal ε0 for all α∈(0,1), ρ∈[1,∞], while Proposition 2.3 only promises, for each p, that 'there exists ε' small enough; the required uniformity in p∈{q,q′,r,r′} and in α,ρ is not demonstrated. The internal argument after Proposition 2.3 appears coherent, but the load-bearing input is an unverified black box.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a well-posedness theory for very weak solutions of the Navier–Stokes equations on bounded Lipschitz domains whose boundary charts belong to a Sobolev multiplier class M_W^{1+α,ρ}(ε) with sufficiently small multiplier norm. The authors define very weak solutions via duality with the Stokes–Dirichlet operator, construct stationary auxiliary fields for the force and boundary data, establish the linear theory, and then reformulate the nonlinear problem as a fixed-point equation in L^s(0,T;L^q(Ω)). Local well-posedness for arbitrary data and global well-posedness under a smallness condition are claimed. The proof builds on a package of Stokes estimates — Helmholtz decomposition, square-root domain identification, maximal regularity, semigroup decay, and gradient estimates — collected in Proposition 2.3, which is only sketched in Appendix A and is said to follow from the unpublished preprint [5] by Breit and Gaudin.","tokens_in":18778,"tokens_out":6671,"duration_ms":58862,"significance":"If the underlying Stokes estimates hold, the paper makes a substantial advance by extending the very weak solution theory from C^{2,1} domains to a class of irregular Lipschitz domains that is strictly between Lipschitz and C^{1,α}. The internal argument is largely coherent: the duality-based very weak formulation, the construction of the stationary auxiliary fields, and the reduction to the integral equation are carefully presented. The paper does not fit parameters or rely on circular assumptions. However, the entire edifice rests on Proposition 2.3, whose proof is not included and whose uniformity in the relevant exponents is not demonstrated. The contribution is therefore conditional on the external toolkit in [5].","major_comments":[{"comment":"Proposition 2.3 is load-bearing: every subsequent result uses it, including the Helmholtz decomposition in (3.8), the auxiliary Stokes solves in Propositions 3.1–3.2, the linear theory in Proposition 4.2, the definition of R_q in (5.1)–(5.2), and the K-estimate in Lemma 5.1. Appendix A does not prove these estimates; it only explains how they follow from [5, Theorems 4.38, 6.5; Propositions 4.40, 6.19; Meta-Theorem 6.20], and [5] is an unpublished preprint. As submitted, Theorem 1.1 is contingent on the correctness and completeness of [5]. The authors should either supply a full proof of Proposition 2.3 or state precisely the external theorem and verify all of its hypotheses, including the exact smallness condition on ε.","section":"Section 2, Proposition 2.3 and Appendix A"},{"comment":"Theorem 1.1 asserts a universal ε0 > 0 that works for all α∈(0,1), ρ∈[1,∞], and for the whole set of exponents p∈{q,q′,r,r′}. Proposition 2.3, however, only says that for each p there exists ε>0 sufficiently small. The proof requires one smallness threshold that simultaneously gives (2.8), (2.10), and the square-root identification for all four exponents. The uniformity of the smallness condition in p, α, and ρ is not shown in Appendix A. If the threshold in [5] is not uniform, then ε0 and the contraction threshold μ0 may depend on p in a way that invalidates the fixed-point argument in Section 6.","section":"Theorem 1.1 and Proposition 2.3 (uniformity of ε0)"},{"comment":"The boundedness of R_q and K is the only mechanism that makes the nonlinear term meaningful. It depends on the gradient estimate (2.10), i.e., on the bound ||∇e^{-tA_{q'}}ψ||_{L^{(q/2)'}} ≤ C t^{-1/2-3/(2q)} ||ψ||_{L^{q'}}. The exponent computation is correct for q>3, but (2.10) is cited from [5] and is not proved. If the constant in this estimate depends on the target exponent (q/2)' = q/(q-2) in an uncontrolled way, then (5.3) and (5.6) would fail and the bilinear estimate (5.7) would break. The authors should verify that (2.10) holds with a constant depending only on q and the domain class, uniformly in α and ρ, for the stated range.","section":"Section 5, Eqs. (5.1)–(5.5) and Lemma 5.1"}],"minor_comments":[{"comment":"The norm in the estimate is written as ∥u∥_{L^s(0,T;L^q(Ω))}, but the theorem has already introduced T*∈(0,T]; it should be L^s(0,T*;L^q(Ω)).","section":"Theorem 1.1, display (1.4)"},{"comment":"The phrase 'ε>0 sufficiently small' appears before the proposition that quantifies ε. This is ambiguous; state explicitly that the smallness condition is the one appearing in Proposition 2.3.","section":"Definition 2.2"},{"comment":"The claimed adaptation to Besov spaces is stated without proof. Since it is not used in the main theorem, it should be either removed or clearly labeled as a conjecture, or the necessary Besov analogues should be stated.","section":"Remark 1.2"},{"comment":"There are numerous formatting issues: missing spaces in 'C 1 c', nonstandard 'L s t Lq x', and inconsistent use of L^s(0,T;L^q) versus L^s(0,T*;L^q). These should be corrected.","section":"Throughout"},{"comment":"Reference [5] is an arXiv preprint with DOI. If accepted for publication, the authors should update the reference to the published version and, more importantly, include the precise theorem statements they rely on.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is mathematically coherent from Section 3 onward, but Proposition 2.3 is the foundation and it is outsourced to an unpublished preprint. This is not an accusation of circularity; rather, it is a question of verifiability. If the editors are willing to accept a paper whose main theorem is conditional on an external preprint, then this merits publication after demonstration of uniformity. In the present form, the central claim is defensible only if the reader trusts [5], which is not available in a peer-reviewed venue. I recommend requesting a complete proof or at least a self-contained statement and verification of all hypotheses of Proposition 2.3 before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline: the paper gives a coherent extension of very weak Navier–Stokes well-posedness to bounded Lipschitz domains whose boundary charts have sufficiently small Sobolev multiplier norm, and the internal argument from the auxiliary fields through the fixed point is clean. But Theorem 1.1 is not self-contained: every load-bearing Stokes estimate is imported from the unpublished Breit–Gaudin preprint [5]. If that toolkit holds, the theorem plausibly holds; if it does not, the construction collapses.\n\nWhat is genuinely new is the duality-based formulation (Definition 3.3) that bypasses the W^{2,q} trace theory used on C^{2,1} domains, and the use of the R_q operator to interpret A_q^{-1} P_q div(u⊗u). The proof of the equivalence between this very weak notion and the integral equation (Prop 5.3) is careful, and the index relations in Lemma 5.1 check out (γ = 1/2 + 3/(2q) < 1 since q > 3). The contraction argument is standard but executed correctly, including the partition-of-unity uniqueness.\n\nThe soft spot is exactly where the stress-test points. Proposition 2.3 supplies the Helmholtz decomposition, square-root identification, maximal regularity, and the gradient estimate (2.10). All of it is cited from [5]. Appendix A shows how [5]'s results would be combined, but it does not prove them or even state the precise smallness assumptions in the multiplier norm under which they hold uniformly in p ∈ {q, q', r, r'} and in α, ρ. That uniformity matters, since Theorem 1.1 advertises a universal ε0. Without it, the statement is ambiguous. The paper is honest about the dependence — the authors flag it in the introduction and appendix — but honesty does not close the gap.\n\nOne more small thing: the abstract promises that the multiplier class contains bounded Lipschitz domains with small Lipschitz constants. I don't see that implication proved or even stated as a proposition in the text. It is plausible from Sobolev multiplier theory, but it should be shown.\n\nWho should read this: people working on very weak solutions or Stokes theory on rough domains will want to know this framework, assuming [5] appears. It deserves a serious referee: the internal logic is sound, and the conditional result is a step beyond Farwig–Galdi–Sohr. My recommendation is to send it out, but instruct the referee to treat Proposition 2.3 as the crux and require the authors to either prove the needed estimates or explicitly rephrase Theorem 1.1 as a conditional statement with precise hypotheses on ε and the external a priori bounds.","headline":"The paper's internal argument is clean, but Theorem 1.1 is a deduction from an unverified external preprint; it deserves review, not blind acceptance.","tokens_in":19295,"tokens_out":3069,"would_cite":false,"duration_ms":29336,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q30","76D05","76D03"],"pacs":[],"model":"deepseek-v4-flash","headline":"On bounded Lipschitz domains whose boundary graphs have small Sobolev-multiplier norm, the three-dimensional Navier-Stokes equations with nonhomogeneous Dirichlet data admit a unique very weak solution in L^s_t L^q_x, locally in time, and g","keywords":["very weak solutions","Navier-Stokes equations","Sobolev multipliers","Lipschitz domains","nonhomogeneous Dirichlet boundary data","Stokes semigroup","well-posedness","maximal regularity"],"falsifier":"One concrete test: take a bounded Lipschitz domain in the multiplier class M_W^{1+α,ρ}(ε) with ε arbitrarily small, and solve the Stokes problem with a forcing term in L^p; if the Stokes semigroup fails any of the estimates of Proposition 2.3 (for instance, if the square-root domain D(A_p^{1/2}) ≠ W^{1,p}_{0,σ} for some p in the exponent set, or if maximal L^p regularity fails), then the theorem is false. Conversely, verifying these estimates for a nontrivial family of such domains would confirm the theory.","tokens_in":18267,"feed_emoji":"🌊","tokens_out":14918,"duration_ms":107670,"temperature":0.7,"pith_summary":"This paper establishes a well-posedness theory for very weak solutions of the three-dimensional Navier-Stokes equations on a class of rough bounded domains: bounded Lipschitz domains whose boundary charts belong to a Sobolev-multiplier class with sufficiently small norm. The main theorem states that, under natural scaling assumptions on the forcing, boundary data, and initial data, a unique very weak solution exists in L^s(0,T*;L^q(Ω)); the solution exists on the whole half-line if the data are small. This matters because the standard theory, which relies on W^{2,q} elliptic regularity, is limited to C^{2,1} domains, and the authors' substitution — semigroup estimates and a duality-based weak formulation that avoids boundary integration by parts — extends existence and uniqueness to a strictly larger class of irregular domains.","feed_headline":"Navier-Stokes admits unique very weak solutions on rough domains","feed_subtitle":"Extends existence and uniqueness to rough domains where smooth-boundary theory fails, with global results for small data","key_machinery":"The load-bearing mechanism is the Sobolev-multiplier boundary class M_W^{1+α,ρ}(ε), a class of bounded Lipschitz domains strictly between Lipschitz and C^{1,α}; on this class the Stokes-Dirichlet operator A_p is assumed to have the full set of semigroup properties: bounded Helmholtz projection, square-root domain identification D(A_p^{1/2}) = W^{1,p}_{0,σ}, maximal L^p regularity, exponential decay, and a gradient estimate. These imported estimates allow the construction of stationary fields absorbing the force and boundary data, and define the very weak formulation via duality with D(A_{q'}) test functions. The key operator is the bilinear map Q(w,v) = K(w⊗v), built from the Stokes semigrou","core_discovery":"The paper's central claim is that the very weak formulation of the Navier-Stokes system with nonhomogeneous Dirichlet boundary data is well-posed on domains of Sobolev-multiplier class M_W^{1+α,ρ}(ε) for sufficiently small ε, with a unique solution in L^s_t L^q_x. The proof reduces the problem to an equivalent integral equation u = U - Q(u,u), built from the Stokes semigroup, and defines the inverse Stokes operator on the rough nonlinear term by duality; the argument closes by a contraction. Local-in-time existence follows from the decay of the semigroup term as t→0, and global-in-time existence from an explicit smallness condition on the data.","pith_inferences":["If the same Stokes estimates hold on wider classes of rough domains than the Sobolev-multiplier class, the argument would transfer directly, since the construction only uses semigroup properties and duality.","The partition argument used for uniqueness gives a quantitative relation between the L^s_t L^q_x norm of the solution and the length of the existence interval; turning it around could yield a blow-up criterion for very weak solutions on irregular domains.","The duality-based treatment of the nonlinear term is not tied to the specific form of the quadratic term; analogues for other incompressible models (e.g., MHD or Boussinesq) would follow if their semigroups satisfy the same set of estimates.","A concrete check of the theorem's hypothesis would be to numerically or analytically verify the Stokes semigroup estimates (particularly the square-root domain property) on an explicit family of multiplier-class domains; non-uniformity across the exponent set would narrow the admissible q-range."],"forward_implications":["The class of domains on which very weak solutions to the Navier-Stokes equations are known to be well-posed is enlarged from C^{2,1} domains to a class of bounded Lipschitz domains with small Sobolev-multiplier boundary norm.","For initial data, forcing, and boundary data satisfying the natural scaling regularity, there is a unique very weak solution on a time interval (0,T*) whose length depends on the data; the estimate (1.4) holds with a constant independent of T.","If, in addition, the data are small in the sense of (1.5), the unique very weak solution exists for all positive times.","The solution satisfies the divergence-free condition and the boundary condition in a weak sense, and a pressure distribution is recovered from the equation.","The same proof, with the semigroup estimates replaced by Besov-space analogues, yields well-posedness for data in Besov spaces B^s_{p,q}(Ω) in the indicated range."],"fun_headline_variants":["Unique very weak solutions for Navier-Stokes on rough domains","Well-posedness on rough domains for Navier-Stokes very weak solutions","Navier-Stokes well-posedness extends to small-multiplier Lipschitz domains","Very weak Navier-Stokes solutions: existence and uniqueness on rough domains","Rough domains admit unique Navier-Stokes very weak solutions"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole proof assumes that the Stokes-Dirichlet operator on Sobolev-multiplier domains—with the specific smallness of the multiplier norm—satisfies the full set of semigroup estimates (bounded Helmholtz projection, square-root domain identification, maximal regularity, exponential decay); if those imported estimates fail for any exponent in the set used, the construction of the solution collapses.","fun_headline_variants_meta":{"raw":{"variants":["Unique very weak solutions for Navier-Stokes on rough domains","Well-posedness on rough domains for Navier-Stokes very weak solutions","Navier-Stokes well-posedness extends to small-multiplier Lipschitz domains","Very weak Navier-Stokes solutions: existence and uniqueness on rough domains","Rough domains admit unique Navier-Stokes very weak solutions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000199,"raw_usage":{"total_tokens":1193,"prompt_tokens":712,"completion_tokens":481,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":456,"completion_tokens_details":{"reasoning_tokens":393}},"tokens_in":456,"tokens_out":481,"duration_ms":3807,"temperature":1.0,"reasoning_tokens":393,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T00:00:14.806594+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete test: take a bounded Lipschitz domain in the multiplier class M_W^{1+α,ρ}(ε) with ε arbitrarily small, and solve the Stokes problem with a forcing term in L^p; if the Stokes semigroup fails any of the estimates of Proposition 2.3 (for instance, if the square-root domain D(A_p^{1/2}) ≠ W^{1,p}_{0,σ} for some p in the exponent set, or if maximal L^p regularity fails), then the theorem is false. Conversely, verifying these estimates for a nontrivial family of such domains would confirm the theory.","supporting_citations":[],"review_version":1}