{"id":"f58146b0-8971-4db8-aacf-d43f7c82f868","arxiv_id":"2509.06089","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A class of 3D boundary layer profiles with independent streamwise and spanwise shear supports Navier-Stokes eigenmodes growing at rate exp(C alpha^3 t/sqrt(nu)), an instability absent in two dimensions.","lead":"Boundary layer flows in three dimensions are shown to have an unstable mode growing like exp(C t/sqrt(nu)), far faster than any 2D instability. The result, if correct, means the standard boundary layer approximation fails in 3D for non-analytic data.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem is conditional on quantitative condition (2.2); abstract's 'generic' and Remark 1.5's 'any δ>0' are not established by the written proof.","rationale":"The reader's CONDITIONAL verdict is essentially right: the theorem is internally coherent, but its advertised scope is overstated. The reader identified the structural assumption (2.2) as the weakest point; I agree that this is the load-bearing condition. However, the reader's specific concern about a 'gap' in the exponential family between large and small amplitudes is not fully accurate: a direct computation of (2.2) for us=1-e^{-kuZ}, vs=v∞(1-e^{-kvZ}) indicates that the inequality may hold for all k>1 and all amplitudes, so the gap may be only in the written verification, not in the actual validity of the example. The more serious issue is the jump from an existence theorem for a restricted class to the abstract's 'generic boundary layer profiles' and Remark 1.5's 'any δ>0': the paper proves neither genericity nor a complete amplitude threshold. I therefore do not change the reader's verdict; the paper remains CONDITIONAL, pending clarification or proof of the advertised genericity, while the central construction itself appears internally consistent.","tokens_in":32395,"tokens_out":50900,"duration_ms":580147,"concrete_test":"Explicitly evaluate (2.2) for us=1-e^{-kuZ}, vs=δ(1-e^{-k kuZ}) as a function of (k,δ), with k>1, to determine the exact set where the inequality holds. If it holds for all k>1,δ>0, Lemma 2.2 can be repaired and Remark 1.5 is valid for this family. Separately, take a profile satisfying (2.1) but failing (2.2), e.g. a slowly varying spanwise profile with k<1 or a small-amplitude vs with moderate k, and check whether any unit vector (λ1,λ2) can satisfy U>0, U'(0)>0, U''(0)>0. If none exists, the instability mechanism is not generic.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 1.1 depends on Proposition 2.1, which requires the quantitative inequality (2.2) to build a direction (λ1,λ2) with U>0, U'(0)>0, and U''(0)>0. This is a genuine restriction: (2.2) is not implied by the linear-independence condition (2.1) that the abstract highlights, and no genericity argument is given. For the exponential family, Lemma 2.2 only verifies (2.2) under k|v∞| ≥ max{1, sqrt(2/(k-1))}; the paper does not treat smaller amplitudes, so the claim in Remark 1.5 that instability occurs for any δ>0 is unsupported by the text. A direct algebraic check suggests that for this family (2.2) may actually hold for all k>1,δ>0, so this particular gap may be repairable — but the repair is absent. What remains load-bearing is the gap between the theorem's actual hypotheses and the advertised 'generic boundary layer profiles': the result proves instability for a nonempty but restricted class, not for generic linearly independent profiles.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies linearized 3D Navier–Stokes equations around a boundary-layer shear flow u_s(z/√ν), v_s(z/√ν), 0. It seeks normal modes e^{-iαcν^{-1/2}t} e^{iν^{-1/2}(σx+βy)} (ũ,ṽ,w̃)(z/√ν). By decoupling the vertical velocity, the authors reduce the problem to an Orr–Sommerfeld equation for w̃, with an induced base flow U = (σ/α)u_s + (β/α)v_s. Under structural conditions on u_s,v_s guaranteeing U'(0)>0, U''(0)>0, U>0, they construct an inviscid Rayleigh unstable mode via a detailed asymptotic expansion of φ_Ray(0;c), obtaining c_Ray ≈ α U_∞^2/U'(0) + i α^2 U''(0)U_∞^4 π/U'(0)^4. They then use a Rayleigh–Airy iteration to show this instability persists for the viscous Orr–Sommerfeld equation, and recover the tangential velocity components from the vertical one. The main theorem asserts a spectral instability with growth rate e^{t/√ν} for a class of spanwise profiles satisfying the quantitative condition (2.2).","tokens_in":32728,"tokens_out":15697,"duration_ms":172363,"significance":"If the proof is correct, this is a substantial contribution: it gives the first rigorous construction of a 3D boundary-layer instability that is stronger than the classical 2D Tollmien–Schlichting instability, and it indicates that 3D Prandtl expansions fail in general without analytic regularity. The argument is genuinely parameter-free at the level of the dispersion relation: the eigenvalue c_Ray is solved from φ_Ray(0;c)=0, and the asymptotic expansions in Lemmas 3.4 and 3.7 are explicit. The use of Rouché's theorem in Theorem 4.5 is appropriate given the claimed analyticity of φ_Ray(0;c). The paper also gives a self-contained recovery of the tangential velocity components in H^1. The main caveat is that the advertised 'generic' and 'any δ>0' statements go beyond what the written hypotheses and Lemma 2.2 establish.","major_comments":[{"comment":"The abstract says the instability occurs for 'generic boundary layer profiles', and Remark 1.5 says it occurs for any spanwise amplitude δ>0. The theorem actually requires the quantitative structural inequality (2.2), which is not shown to be generic and is not implied by the linear-independence condition (2.1). Lemma 2.2 verifies (2.2) for the exponential family only under k>1 and k|v∞| ≥ max{1, sqrt(2/(k-1))}; the small-amplitude regime is not treated in the paper. Thus 'generic' and 'any δ>0' are unsupported as written. Either prove (2.2) for an open dense set (or for all linearly independent profiles), or revise the abstract and Remark 1.5 to state the actual conditional class.","section":"Abstract; Remark 1.5; Proposition 2.1; Lemma 2.2"},{"comment":"The parameter set H2 is stated as 'α, c_r, c_i ∼ O(1)', but the actual regime is α≪1 and c_i ∼ α^2. This is internally inconsistent: c_i∼O(1) and c_i≪α≪1 cannot both hold. The convergence proof of the Rayleigh–Airy iteration relies on 'Since α, c_i ∼ O(1)' to make the factor |ε|^{1/4} c_i^{-3/2}|log c_i| small. With c_i∼α^2 the factor is ν^{1/8} α^{-13/4}|log α|, so the written proof only gives convergence for ν sufficiently small depending on α. Theorem 4.5 states existence of ν0 for each fixed α, so the argument can be repaired by tracking the α-dependence explicitly, but the current text does not do so. This needs to be fixed in a revision.","section":"§4, definition of H2 and estimates (4.10)–(4.12)"}],"minor_comments":[{"comment":"There are several typos: 'centain' in §1.1, 'suffcient' in Lemma 2.2, 'vetical' in the proof of Theorem 1.1, and inconsistent spelling of 'Rouché'. These should be corrected.","section":"Throughout"},{"comment":"The phrase 'α, c ∼ O(1)' appears in the roadmap and in H2; it should be replaced by 'α≪1, c_i≪c_r∼α' to match the actual scaling used in Sections 3 and 4.","section":"§1.3 and §4"},{"comment":"The proof of Lemma 2.2 compares with the expression 2(k^3v∞^2 -1)/(k|v∞|(1+k)). This expression is correct only after assuming v∞ is dimensionless and the profiles are as in (2.3); the presentation would be clearer if the nondimensionalization were stated explicitly. In particular, the notation k^3v∞^2 could be misread as a dimensional quantity.","section":"Lemma 2.2"},{"comment":"In the Rouché argument, the bound |φ_Ray(0;c)| ≥ ν^δ/(2U_∞^2) uses |∂_c φ_Ray| = O(α|log c_i| + |c|). Since |c|∼α, the condition α|log α|≪1 is needed; this is true for α≪1 but should be stated explicitly.","section":"Theorem 4.5 proof"}],"recommendation":"major_revision","confidential_remarks":"The core analytic construction appears sound and is a significant advance if the parameter-regime and genericity issues are resolved. The main concern is not circularity or a fundamental error in the estimates, but a mismatch between the paper's advertised claims ('generic', 'any δ>0') and the hypotheses actually used. This is fixable with a careful revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing to know: this paper constructs, for the linearized 3D Navier-Stokes equations around a class of boundary layer shear flows, an unstable mode with growth rate e^{C t / sqrt(nu)}. That is genuinely new. It beats the 2D Tollmien-Schlichting rate and comes from a different mechanism: the spanwise profile plus the 3D perturbation yields an effective base flow U with U''(0)>0, which is convex at the wall and inviscidly unstable. The structure is clever, and the Rayleigh-Airy construction is carried out in detail. I can't find an obvious error in the core estimates (Lemmas 3.2-3.7, 4.1-4.4, Theorems 3.9, 4.5). The use of Rouche's theorem appears legitimate. The recovery of tangential velocities checks out. So the main theorem, conditional on (2.1)-(2.2), appears sound.\n\nThe soft spot is the gap between the theorem and the advertising. The abstract says 'generic' profiles, and Remark 1.5 says instability occurs for any spanwise amplitude δ>0. Neither is established. Proposition 2.1 requires the quantitative inequality (2.2), which is not a consequence of linear independence. For the exponential family, Lemma 2.2 only verifies (2.2) under k|v_infty| >= max{1, sqrt(2/(k-1))}. The paper leaves a hole in the middle range. The stress-test note suggests (2.2) may actually hold for all k>1, δ>0 for that family, and if so the gap is repairable—but it is not repaired in this version. The reader's verdict of CONDITIONAL is fair.\n\nOne more concern: the abstract and remarks imply the result forbids Prandtl validity in Sobolev/Gevrey classes. That is a leap. The paper proves linear spectral instability, not nonlinear instability. The discussion in Remark 1.4 is a reasonable speculation but is presented as a conclusion.\n\nI would send this to a serious referee. The core construction deserves scrutiny, and the overclaims will need to be fixed before acceptance. The math is coherent enough that a referee can do useful work. If the genericity gap is closed, this is an important paper.\n\nFor the reading group: yes, it's worth the time.","headline":"A genuinely new 3D boundary-layer instability mechanism with a clean core proof, but the advertised 'generic' profiles and 'any δ>0' claim outrun what the stated hypotheses actually support.","tokens_in":33245,"tokens_out":1993,"would_cite":true,"duration_ms":21419,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q30","76D10","76E05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs unstable modes for the linearized Navier-Stokes equations around three-dimensional boundary-layer shear flows in the small-viscosity limit, with growth rates of order exp(t/sqrt(nu)).","keywords":["3D boundary layer","Navier-Stokes linear stability","Tollmien-Schlichting waves","Rayleigh equation","Orr-Sommerfeld equation","zero-viscosity limit","Prandtl expansion","spectral instability"],"falsifier":"Solve the linearized eigenvalue problem (1.4) numerically for the exponential profiles u_s=1-e^{-Z}, v_s=v_infty(1-e^{-kZ}) with k>1 and k|v_infty| large, at nu=10^{-6}. Theorem 1.1 predicts an eigenvalue with c_i ~ alpha^2 U''(0)U_inf^4 pi/U'(0)^4 and growth rate ~nu^{-1/2}; if no unstable eigenvalue appears, or if the growth rate scales differently in nu, the central claim fails. A second check: test profiles violating (2.2); the theorem makes no prediction there, so finding instability would not falsify it but would show the hypotheses are not sharp.","tokens_in":32282,"feed_emoji":"🌊","tokens_out":5504,"duration_ms":59224,"temperature":0.7,"pith_summary":"When the streamwise and spanwise velocity profiles are linearly independent near the wall, the authors prove there is a growing mode with rate exp(t/sqrt(nu)), faster than the classical two-dimensional Tollmien-Schlichting instability. The instability arises because the three-dimensional perturbation tilts the base flow into a modified profile U=(sigma/alpha)u_s+(beta/alpha)v_s that is positive, increasing, and convex at the boundary, which destabilizes the Rayleigh equation. The same mode persists in the viscous Orr-Sommerfeld equation through a Rayleigh-Airy iteration, because the viscous boundary layer only shifts the eigenvalue by a small amount. If correct, this means generic high-Reynolds-number three-dimensional boundary layers are analytically unstable, and Prandtl expansion validity in three dimensions cannot be expected without analytic regularity.","feed_headline":"3D boundary layers explode at e^{t/√ν} speed","feed_subtitle":"New unstable mode grows faster than Tollmien-Schlichting waves and survives at small viscosity.","key_machinery":"The central object is the modified shear flow U(Z)=lambda_1 u_s(Z)+lambda_2 v_s(Z), with (lambda_1,lambda_2) chosen as in Proposition 2.1 so that, under the independence condition (2.1) and the quantitative bound (2.2), U is positive for Z>0, increasing and convex at the wall. This U acts as the effective advection speed in the Rayleigh and Orr-Sommerfeld equations for the vertical velocity. The proof machinery is the second-order asymptotic expansion of the Rayleigh solution phi_Ray(0;c) (Lemmas 3.4 and 3.7), which yields the unstable inviscid eigenvalue c_Ray approx alpha U_inf^2/U'(0) + i alpha^2 U''(0)U_inf^4 pi/U'(0)^4, followed by a Rayleigh-Airy iteration that constructs the Orr-Somme","core_discovery":"Theorem 1.1 states: for any small viscosity 00, c_i>0, and growth |(u,v,w)| ~ e^{C nu^{-1/2} t}. The eigenvalue is found first for the inviscid Rayleigh equation, where U''(0)>0 supplies a positive imaginary part, and then shown to persist for the Orr-Sommerfeld equation: the viscous boundary layer only shifts t","pith_inferences":["A direct numerical check for intermediate spanwise amplitudes, where Lemma 2.2 is silent, would test whether the instability extends beyond the proven parameter range; the theorem itself makes no claim there.","If the instability persists at arbitrarily small spanwise amplitude, the bifurcation picture in Remark 1.5 suggests that physically realized 3D boundary layers with weak spanwise drift should show rapid linearized transient growth even when the secondary flow is infinitesimal.","The mechanism inverts the usual role of convexity: U''(0)>0 destabilizes rather than stabilizes, which could guide spanwise-forcing design or numerical experiments aimed at suppressing boundary-layer transition.","The paper stops at the linear level; an analogous nonlinear instability statement for the 3D Navier-Stokes system would be a natural next step, following the pattern of the 2D Tollmien-Schlichting nonlinear theory."],"forward_implications":["The 3D linearized Navier-Stokes boundary-layer problem has unstable modes with growth rate of order nu^{-1/2}, faster than the two-dimensional Tollmien-Schlichting growth rate.","The instability appears already at the Euler level (inviscid Rayleigh equation) and persists for small viscosity, because the viscous boundary layer only shifts the eigenvalue by O(nu^{1/4-}).","For any nonzero spanwise amplitude delta, no matter how small, instability occurs; at delta=0 the flow is stable in Gevrey-3/2 for concave profiles, so delta=0 is a bifurcation point.","The growing mode has frequency of order nu^{-1/2}, so it lives on the boundary-layer scale and is genuinely three-dimensional: it does not occur when u_s and v_s are linearly dependent.","Validity of the 3D Prandtl expansion cannot be expected without analytic regularity, in contrast to the 2D case where Gevrey-3/2 suffices."],"supporting_citations":[{"why":"Supplies the 2D Tollmien-Schlichting spectral-instability baseline and the Rayleigh-Airy iteration method that the paper compares against and adapts.","marker":"[12]"},{"why":"Supplies the Rayleigh-Airy iteration framework and the benchmark that 2D Prandtl expansion validity holds in Gevrey-3/2, against which the 3D analytic-instability contrast is drawn.","marker":"[9]"},{"why":"Source of the observation that the vertical velocity decouples from the other variables, used to derive the Orr-Sommerfeld equation.","marker":"[3]"},{"why":"Provides the 2D Prandtl-equation instability for non-monotone profiles, one of the background instability mechanisms the paper distinguishes itself from.","marker":"[8]"},{"why":"Motivates the question whether three-dimensional effects can destabilize flows that are stable in two dimensions.","marker":"[6]"},{"why":"Establishes a growing mode for the 3D Prandtl equation in Gevrey-2 space, the preceding 3D instability result that this paper strengthens to Navier-Stokes with faster growth.","marker":"[21]"}],"fun_headline_variants":["3D boundary layer instability grows at e^{t/√ν}","Spanwise flow triggers unstable mode in Navier-Stokes","New instability in 3D boundary layers: e^{t/√ν} growth","Viscous 3D shear flows show exponential blow-up","3D shear flow unstable: growth e^{t/√ν}"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The construction needs the quantitative structural inequality (2.2), which forces the modified profile U=lambda_1 u_s+lambda_2 v_s to be positive, increasing, and convex at the wall; this is not shown for generic profiles, and for the explicit exponential family it is only verified under a large-spanwise-amplitude condition, so the word 'generic' in the abstract overreaches the hypotheses.","fun_headline_variants_meta":{"raw":{"variants":["3D boundary layer instability grows at e^{t/√ν}","Spanwise flow triggers unstable mode in Navier-Stokes","New instability in 3D boundary layers: e^{t/√ν} growth","Viscous 3D shear flows show exponential blow-up","3D shear flow unstable: growth e^{t/√ν}"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000776,"raw_usage":{"total_tokens":3246,"prompt_tokens":695,"completion_tokens":2551,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":439,"completion_tokens_details":{"reasoning_tokens":2458}},"tokens_in":439,"tokens_out":2551,"duration_ms":19927,"temperature":1.0,"reasoning_tokens":2458,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T04:29:33.133593+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Solve the linearized eigenvalue problem (1.4) numerically for the exponential profiles u_s=1-e^{-Z}, v_s=v_infty(1-e^{-kZ}) with k>1 and k|v_infty| large, at nu=10^{-6}. Theorem 1.1 predicts an eigenvalue with c_i ~ alpha^2 U''(0)U_inf^4 pi/U'(0)^4 and growth rate ~nu^{-1/2}; if no unstable eigenvalue appears, or if the growth rate scales differently in nu, the central claim fails. A second check: test profiles violating (2.2); the theorem makes no prediction there, so finding instability would not falsify it but would show the hypotheses are not sharp.","supporting_citations":[{"cited_title":"Grenier, Y","cited_arxiv_id":null,"evidence_quote":"Supplies the 2D Tollmien-Schlichting spectral-instability baseline and the Rayleigh-Airy iteration method that the paper compares against and adapts."},{"cited_title":"G´ erard-Varet, Y","cited_arxiv_id":null,"evidence_quote":"Supplies the Rayleigh-Airy iteration framework and the benchmark that 2D Prandtl expansion validity holds in Gevrey-3/2, against which the 3D analytic-instability contrast is drawn."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Source of the observation that the vertical velocity decouples from the other variables, used to derive the Orr-Sommerfeld equation."},{"cited_title":"G´ erard-Varet and E","cited_arxiv_id":null,"evidence_quote":"Provides the 2D Prandtl-equation instability for non-monotone profiles, one of the background instability mechanisms the paper distinguishes itself from."},{"cited_title":"Gallaire, D","cited_arxiv_id":null,"evidence_quote":"Motivates the question whether three-dimensional effects can destabilize flows that are stable in two dimensions."},{"cited_title":"Liu , Y.-G","cited_arxiv_id":null,"evidence_quote":"Establishes a growing mode for the 3D Prandtl equation in Gevrey-2 space, the preceding 3D instability result that this paper strengthens to Navier-Stokes with faster growth."}],"review_version":1}