{"id":"8bbef019-f787-4b14-8473-194688de54d3","arxiv_id":"2506.11414","paper_version":2,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"For the 2D free-boundary Euler equations with surface tension on a droplet, there exist arbitrarily small initial velocities such that the vorticity Hessian grows at least exponentially in time.","lead":"This paper constructs small smooth initial data for a 2D capillary liquid drop whose vorticity Hessian grows exponentially in time, showing that even tiny rotational motion can amplify small scales. The result provides a mathematical instability mechanism for free-boundary Euler droplets under surface tension.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 3.1's confinement proof uses a misquoted Bonnesen inequality (linear instead of quadratic in R−ρ), so the O(δ) boundary control and B√2⊂Dt are not proved as written; a weaker O(√δ) bound likely suffices.","rationale":"The reader's weakest_assumption identifies Lemma 3.1 as the hinge of the proof, and the stress-test agrees. The specific unjustified inequality is not merely cosmetic: the linear Bonnesen inequality (3.3) is false, so the claimed O(δ) boundary confinement is not proved. However, the role of Lemma 3.1 in the rest of the argument is only to keep a fixed disk of radius comfortably larger than √2 inside the droplet and to make the boundary correction perturbative with size √K(0). Replacing O(δ) by O(√δ) still gives 2−O(√δ)>√2 for sufficiently small δ, so the main growth mechanism in Section 5 can likely be preserved by shrinking ε0 and adjusting constants. Thus the paper is not fatally undermined, but it is not fully rigorous as printed; a conditional verdict remains appropriate. Other potential points—such as the delegation of the I11 computation to [31] and the elliptic interior estimates in Proposition 4.2—are less central and appear standard, though they should also be checked in a full revision.","tokens_in":13558,"tokens_out":26797,"duration_ms":275071,"concrete_test":"Check the stated Theorem 3.2 against its source, Osserman, 'Bonnesen-style isoperimetric inequalities', Theorem 4. If the correct inequality has (R−ρ)², recompute (3.6)–(3.9) with R−ρ ≤ π^{-1}√(δ(8π+δ)). Then verify whether the resulting weaker containment B_{2−C√δ}⊂Dt still implies B√2⊂Dt for δ below a universal threshold, and whether Proposition 4.2's error bound √K(0) remains sufficient; if so, amend Lemma 3.1 and choose ε0 accordingly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 3.1 is the sole mechanism ensuring B√2⊂Dt for all times, which is what makes the fixed-domain Biot–Savart decomposition in (4.3)–(4.5) and Proposition 4.1 legitimate. The proof invokes Theorem 3.2 in the form L²−4πA ≥ π²(R−ρ). This is not the standard Bonnesen inequality: Osserman [22, Thm 4] gives (R−ρ)² on the right. For a nearly circular ellipse with R−ρ<1, L²−4πA is of order (R−ρ)², which is smaller than π²(R−ρ); hence the printed inequality is false. Consequently (3.6) should read R−ρ ≤ π^{-1}√(δ(8π+δ)) = O(√δ), not O(δ), and the annulus confinement Γt⊂B_{2+c0δ}\\B_{2−c1δ} is not established. Later the paper only needs some fixed disk B_r with r>√2 to remain inside Dt; the O(√δ) version still gives 2−O(√δ)>√2 for δ small, so the main theorem is likely repairable by shrinking ε0. A second flaw in the same lemma: Observation 1 argues that if the circumcenter q lies in the first quadrant then r≥|q−p̃|>|p|, but |q−p̃|>|p| is false in general (e.g. p=(1,1), q=(0,0.1)); the origin-centeredness is true but the given proof is invalid.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the two-dimensional incompressible Euler equations with surface tension for a capillary liquid drop, with initial domain a disk of radius 2. The main result, Theorem 1.1, constructs arbitrarily small smooth rotational initial velocities u0 = εv0 such that, for a threshold ε0, either the solution loses regularity before a time T1(ε) or the L∞ norm of the vorticity Hessian satisfies sup_{t≤T} ||∇²ω(t)||_{L∞(D_t)} ≥ ε e^{εT} for every T > T1(ε). The proof strategy is to use the conserved energy K(t)+σL(t) to confine the free boundary in a thin annulus, then to decompose the velocity near the origin into a fixed-disk Dirichlet part U and a small error e via an approximate Biot-Savart law, and finally to run a Zlatoš-type hyperbolic-flow argument that amplifies a initially tiny vorticity gradient into exponential growth of the Hessian. The argument relies heavily on symmetry, on two prior works [31] and [16], and on a geometric confinement lemma (Lemma 3.1).","tokens_in":13948,"tokens_out":11017,"duration_ms":125748,"significance":"If the proof is correct, the result is significant: it shows that arbitrarily small smooth rotational data on a circular droplet can lead to exponential growth of the vorticity Hessian, extending to a free-boundary problem the small-scale-creation mechanisms previously developed for fixed domains. The paper's construction is quantitative and does not fit parameters to the claimed growth rate; the rate εe^{εT} is explicit and the initial data are explicit up to a smooth odd function. The use of the conserved energy to control the free boundary and the separation of the growth mechanism from the boundary are conceptually appealing. However, the soundness of the paper hinges on the geometric confinement lemma and on several estimates imported from prior work, so the result is plausible but not yet fully established as written.","major_comments":[{"comment":"The Bonnesen inequality is misquoted. Theorem 3.2 states L² − 4πA ≥ π²(R−ρ), but the cited result [22, Theorem 4] has the quadratic form L² − 4πA ≥ π²(R−ρ)². Therefore the derivation of (3.6) gives R−ρ ≤ C√δ, not O(δ). The linear-width conclusion Γ_t ⊂ B_{2+c0δ}\\B_{2−c1δ} is thus not proved, and for near-circular domains a linear bound in the isoperimetric deficit is generally false. The later argument only requires that a fixed disk B_r with r > √2 remain inside D_t, so the theorem appears repairable by choosing ε0 smaller, but Lemma 3.1 and every subsequent use of its quantitative form must be rewritten.","section":"Section 3, Theorem 3.2 and Eq. (3.6)"},{"comment":"The inequality |q − p̃| > |p| used to prove that the circumcenter q must be the origin is false in general. For example, with p = (1,1) and q = (0,0.1), one has p̃ = (−1,1), |q − p̃| = sqrt(1.81) < sqrt(2) = |p|. The intended conclusion that the circumscribed circle is centered at the origin should instead be deduced from the uniqueness of the minimal enclosing circle combined with the two-fold reflective symmetry of D_t; this step needs to be supplied.","section":"Section 3, Observation 1 (proof of Lemma 3.1)"},{"comment":"The key pointwise estimate for the main term U1 is not derived in the paper but is imported as \"an identical argument to [31, Lemma 2.1]\". Since [31] treats the torus and the present setting involves a fixed-disk Green's function with additional boundary correction terms, the reduction to the torus calculation should be stated explicitly, or the precise form of the imported lemma should be quoted with all constants. This estimate is the quantitative engine of Proposition 4.1, so the argument is not fully checkable without this detail.","section":"Section 4.2, Eqs. (4.11)–(4.12)"}],"minor_comments":[{"comment":"The phrase \"D0 = D0\" appears to be a typo; it should state that the domain is invariant under the reflection.","section":"Section 2, proof of Lemma 2.2"},{"comment":"The description \"e1 is odd (even) in x1 (x2)\" is unclear and should read \"e1 is odd in x1 and even in x2\", with the analogous statement for e2.","section":"Section 4.1, Proposition 4.2, item 2"},{"comment":"The assertion that u2(t,0,z) > 0 for 0 < z < η/2 is used in the characteristic argument but is not justified; it should be derived by passing to the limit x1 → 0 in (5.8) or by a separate parity argument.","section":"Section 5, around Eq. (5.14)"},{"comment":"The notation p̃ and ¯p for the two reflections is easy to confuse; using distinct symbols throughout, including in Section 3, would improve readability.","section":"Section 4.2, Eq. (4.8)"}],"recommendation":"major_revision","confidential_remarks":"The central claim is important and the overall strategy is credible, but the current version contains a misquoted isoperimetric inequality in a load-bearing lemma and relies on several unstated imports from prior papers. Since the main theorem appears repairable by switching to an O(√δ) confinement bound and by choosing ε0 sufficiently small, I recommend major revision rather than rejection. The author should also be asked to reduce the number of \"identical argument\" and \"similar argument\" delegations in the proof of Proposition 4.1."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the take: this paper proves something genuinely new—exponential growth of the vorticity Hessian for a 2D capillary droplet with small, smooth, rotational initial data. That is the first small-scale creation result for a free-boundary droplet, and it extends the fixed-domain constructions of Kiselev–Šverák and Zlatoš to a moving boundary. The strategy is sensible: use the conserved energy K+σL to keep the boundary far from the hyperbolic mechanism near the origin, then adapt Zlatoš's torus construction with an approximate Biot–Savart law. The main theorem is derived, not assumed; no parameter is fitted to force the growth rate. Credit is due for that.\n\nThe soft spots are real but not fatal. Lemma 3.1, the confinement lemma, misquotes the Bonnesen inequality: Theorem 3.2 is stated as L²−4πA ≥ π²(R−ρ), but the correct Osserman inequality has (R−ρ)² on the right. Consequently the O(δ) annulus width is not proved; you only get O(√δ). That still suffices, because the proof only needs some fixed disk r>√2 to stay in Dt, and 2−O(√δ)>√2 for δ small. So the main theorem survives, but only after shrinking ε0 and rewriting the estimates. A second, smaller issue in the same lemma: Observation 1 asserts |q−p̃| > |p|, which is false in general; the origin-centeredness of the circumcircle is probably true by symmetry, but the given proof is invalid. These are fixable, but a referee will want to see them cleaned up.\n\nOther soft spots are more stylistic. Several key estimates are imported from [31] and [16] with phrases like 'identical argument' rather than full derivations. For a paper of this technical density, that is acceptable if the references are precise, but it makes verification harder. The self-citation to [16] is legitimate here, because that is the framework being extended.\n\nWho is this for? People working on free-boundary Euler, small-scale creation, or hydrodynamic instability. The paper deserves a serious referee: the result is meaningful, the proof is serious, and the gaps are local and repairable. I would engage with it, but the author needs to fix Lemma 3.1 before I'd accept it as written.","headline":"Real new result for droplet small-scale creation, but the confinement lemma has a misquoted Bonnesen inequality that is repairable.","tokens_in":14428,"tokens_out":2154,"would_cite":true,"duration_ms":23996,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q31","35R35","76B45"],"pacs":[],"model":"deepseek-v4-flash","headline":"Arbitrarily small smooth swirls on a circular capillary droplet force exponential growth of the vorticity Hessian (or finite-time loss of regularity).","keywords":["free-boundary Euler equations","capillary liquid drop","surface tension","vorticity Hessian","exponential growth","small scale creation","hyperbolic flow","instability"],"falsifier":"Numerically solve the free-boundary Euler equations with surface tension for the constructed data ($D_0=B_2$, $u_0=\\varepsilon\\nabla^\\perp\\psi$ with $f$ odd-odd, $f\\equiv 1$ on most of the first quadrant, $f=\\sin^3(x_1)\\sin(x_2)$ near $0$). If for some $\\varepsilon<\\varepsilon_0$ and some $T>T_1(\\varepsilon)$ the solution remains regular while $\\|\\nabla^2\\omega(T)\\|_{L^\\infty(D_T)}<\\varepsilon e^{\\varepsilon T}$, the theorem is false. A cheaper test is to check the two load-bearing ingredients: the confinement $\\Gamma_t\\subset B_{2+c_0\\delta}\\setminus B_{2-c_1\\delta}$ (Lemma 3.1) and the pointwise velocity formula (4.1) with the stated error bound near the origin.","tokens_in":13386,"feed_emoji":"💧","tokens_out":12997,"duration_ms":120491,"temperature":0.7,"pith_summary":"This paper proves that a perfectly circular two-dimensional droplet of an ideal fluid, held together only by surface tension, is strongly unstable: even if the initial swirling motion is arbitrarily small and perfectly smooth, the curvature of the vorticity field (its Hessian) must grow at least like $\\varepsilon e^{\\varepsilon T}$ by time $T$, unless the solution loses smoothness in finite time first. The author constructs explicit initial data on the disk of radius $2$ with velocity of size $\\varepsilon$ for arbitrarily small $\\varepsilon$, so the statement is about genuinely small perturbations of a stationary round drop. The mechanism is an interior hyperbolic flow near the drop's center, which stretches vorticity contours into exponentially thin filaments without any help from the free boundary. A conserved energy—kinetic energy plus surface tension times boundary length—keeps the boundary far from the growth region, so the free interface acts only as a small perturbative error. If true, this gives the first quantitative instability result for rotational capillary droplets and shows that small-scale creation is not an artifact of fixed-wall boundaries.","feed_headline":"Vorticity Hessian grows exponentially in tiny swirling droplets","feed_subtitle":"Even arbitrarily weak initial swirls force exponentially sharp vorticity curvature on a free-boundary droplet.","key_machinery":"The load-bearing object is the approximate Biot-Savart law (Proposition 4.1). It states that for initial data with the symmetry of Lemma 2.2 and sufficiently small initial kinetic energy $K(0)\\le \\delta_0\\sigma$, every $x\\in B_{1/2}$ in the first quadrant satisfies $u_j(t,x)=(-1)^j \\frac{4}{\\pi}\\left(\\int_{Q(2x)} \\frac{y_1 y_2}{|y|^4}\\omega(t,y)\\,dy + B_j(t,x)\\right)x_j$, with $|B_j|$ bounded by $C_0\\big(\\|\\omega_0\\|_{L^\\infty}(1+\\min\\{\\log(1+x_{3-j}/x_j),\\, x_{3-j}\\|\\nabla\\omega\\|_{L^\\infty([0,2x_{3-j}]^2)}/\\|\\omega_0\\|_{L^\\infty}\\})+\\sqrt{K(0)}\\big)$. The integral term is the hyperbolic-flow kernel introduced for the torus in the paper's reference [31]; because $\\omega$ is odd in each variable, this kernel makes particle paths near the origin behave like hyperbolas, approaching the coordinate axes and compressing the support of vorticity. The error term $B_j$ is controlled through the decomposition $u=U+e$, where $U$ is the Euler flow on the fixed disk $B_{\\sqrt{2}}$ with vanishing stream function and $e$ is shown to satisfy $|e_j|\\le C\\sqrt{K(0)}|x_j|$ using harmonicity, a Calder\\'on-Zygmund estimate, and the symmetry; the crucial input that $B_{\\sqrt{2}}\\subset D_t$ for all times comes from the confinement Lemma 3.1, proved via the conserved quantity $K(t)+\\sigma L(t)$ and a Bonnesen-style isoperimetric inequality.","core_discovery":"The central theorem (Theorem 1.1) states: for $\\sigma=1$ and initial domain $D_0=B_2$, there exist a smooth divergence-free vector field $v_0$ and a threshold $\\varepsilon_0>0$ such that for every $\\varepsilon\\in(0,\\varepsilon_0)$ the regular solution with initial velocity $u_0=\\varepsilon v_0$ either loses regularity by some time $T_1(\\varepsilon)$, or for every $T>T_1(\\varepsilon)$ the spatial supremum of $|\\nabla^2\\omega(t)|$ on the moving droplet $D_t$ is bounded below by $\\varepsilon e^{\\varepsilon T}$. In other words, arbitrarily small smooth rotational data on a circular capillary drop are nonlinearly unstable at the level of the vorticity Hessian: the second derivatives of vorticity grow exponentially in time along the regular evolution. The result is independent of the surface tension coefficient, since the $\\sigma=1$ case transfers to general $\\sigma>0$ by the scaling $(\\sigma^{1/2}u(\\sigma^{1/2}t,x), D_{\\sigma^{1/2}t})$. The proof constructs data with odd-odd symmetric vorticity, tracks a single fluid particle near the origin, and uses the approximate Biot-Savart law to show the particle is swept toward the axes while the vorticity it carries is stretched, producing a large $\\partial_1\\omega$ difference across a small interval; the Hessian bound then follows either from that difference (Case 2) or, if $\\nabla\\omega$ is already large, from the mean value theorem applied to $\\nabla\\omega$ vanishing at the origin (Case 1).","pith_inferences":["If the theorem is correct, the same interior-hyperbolic mechanism should produce exponential growth of the vorticity Hessian in other free-boundary or fixed-domain settings where an odd-odd symmetric hyperbolic point can be planted away from all boundaries; the torus construction is one instance, and the droplet construction shows the boundary only needs to be kept at a distance.","The confinement lemma suggests a general principle for capillary droplets: small initial kinetic energy prevents the free boundary from approaching any interior compact set, and the quantitative gap is governed by the isoperimetric deficit of the boundary. One could test whether the threshold $\\delta<2\\pi/27$ is sharp or merely an artifact of the particular isoperimetric inequality used.","A natural numerical test is to simulate the free-boundary Euler equations with the constructed symmetry data and measure $\\|\\nabla^2\\omega\\|_{L^\\infty(D_t)}$; a successful simulation should observe the exponential rate $\\varepsilon e^{\\varepsilon T}$ until either numerical blow-up or loss of regularity, and should verify the boundary stays outside $B_{\\sqrt{2}}$.","The paper leaves open whether the dichotomy resolves as finite-time singularity or eternal exponential growth; if a global continuation criterion were available, the same data set could decide which branch occurs."],"forward_implications":["For any arbitrarily small smooth rotational perturbation of a circular capillary droplet, a regular solution cannot have uniformly bounded vorticity Hessian: $\\sup_{t\\le T}\\|\\nabla^2\\omega(t)\\|_{L^\\infty(D_t)}\\ge \\varepsilon e^{\\varepsilon T}$ for all large $T$, so some derivative of vorticity must grow without bound on the maximal time interval.","The growth mechanism operates strictly in the fluid bulk, independent of the free boundary: the free interface contributes only through a perturbative error of size $\\sqrt{K(0)}$, so the construction shows small-scale creation can be driven by interior hyperbolic flows rather than boundary effects.","The result extends to any surface tension $\\sigma>0$ via the scaling symmetry, so the instability is not specific to the normalized coefficient $\\sigma=1$.","As the paper notes, the dichotomy in Theorem 1.1 is compatible with finite-time loss of regularity: by the continuation criterion discussed in Remark 1.2, the constructed solutions may cease to be regular in finite time, and the growth bound applies to the regular part of the evolution."],"supporting_citations":[{"why":"Supplies the hyperbolic-flow growth mechanism and the exact integral identity (the $I_{11}$ term coincides with Equation (2.5) of that paper) used in the approximate Biot-Savart law.","marker":"[31]"},{"why":"Provides the free-boundary small-scale-creation framework: the conserved quantity $K(t)+\\sigma L(t)$, symmetry propagation, and the velocity decomposition $u=U+e$ that treats the boundary as a perturbation.","marker":"[16]"},{"why":"Introduces hyperbolic flows for small-scale creation in 2D Euler, here adapted to an interior hyperbolic point rather than a fixed boundary.","marker":"[19]"},{"why":"Supplies the quantitative Bonnesen-style isoperimetric inequality used in the proof of the confinement Lemma 3.1.","marker":"[22]"},{"why":"Establishes local well-posedness for the free-boundary Euler equations with surface tension, so the constructed $H^\\infty$ initial data produce regular solutions.","marker":"[10]"},{"why":"Supplies the Calder\\'on-Zygmund estimate used to control the $C^1$ norm of the error term $e$ on $B_{1/2}$.","marker":"[14]"}],"fun_headline_variants":["Tiny swirls, exponential vorticity Hessian growth","Weak swirls cause exponential vorticity growth","Arbitrary small swirls, exponential Hessian growth","Capillary drop: small vorticity, exponential Hessian","Exponential vorticity Hessian from tiny initial swirls"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof relies on Lemma 3.1, which says that if the initial kinetic energy is small enough, the free boundary stays inside a thin annulus around the original circle for all time, so the fixed disk $B_{\\sqrt{2}}$ is always contained in the droplet; if the boundary ever got close to the origin, the approximate Biot-Savart law would fail and the hyperbolic growth mechanism would break down.","fun_headline_variants_meta":{"raw":{"variants":["Tiny swirls, exponential vorticity Hessian growth","Weak swirls cause exponential vorticity growth","Arbitrary small swirls, exponential Hessian growth","Capillary drop: small vorticity, exponential Hessian","Exponential vorticity Hessian from tiny initial swirls"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000965,"raw_usage":{"total_tokens":4100,"prompt_tokens":928,"completion_tokens":3172,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":544,"completion_tokens_details":{"reasoning_tokens":3108}},"tokens_in":544,"tokens_out":3172,"duration_ms":20996,"temperature":1.0,"reasoning_tokens":3108,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:10:28.231622+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically solve the free-boundary Euler equations with surface tension for the constructed data ($D_0=B_2$, $u_0=\\varepsilon\\nabla^\\perp\\psi$ with $f$ odd-odd, $f\\equiv 1$ on most of the first quadrant, $f=\\sin^3(x_1)\\sin(x_2)$ near $0$). If for some $\\varepsilon<\\varepsilon_0$ and some $T>T_1(\\varepsilon)$ the solution remains regular while $\\|\\nabla^2\\omega(T)\\|_{L^\\infty(D_T)}<\\varepsilon e^{\\varepsilon T}$, the theorem is false. A cheaper test is to check the two load-bearing ingredients: the confinement $\\Gamma_t\\subset B_{2+c_0\\delta}\\setminus B_{2-c_1\\delta}$ (Lemma 3.1) and the pointwise velocity formula (4.1) with the stated error bound near the origin.","supporting_citations":[{"cited_title":"Exponential growth of the vorticity gradient for the euler equation on the torus","cited_arxiv_id":null,"evidence_quote":"Supplies the hyperbolic-flow growth mechanism and the exact integral identity (the $I_{11}$ term coincides with Equation (2.5) of that paper) used in the approximate Biot-Savart law."},{"cited_title":"Small scale creation for 2d free boundary euler equations with surface tension","cited_arxiv_id":null,"evidence_quote":"Provides the free-boundary small-scale-creation framework: the conserved quantity $K(t)+\\sigma L(t)$, symmetry propagation, and the velocity decomposition $u=U+e$ that treats the boundary as a perturbation."},{"cited_title":"Small scale creation for solutions of the incompressible two- dimensional euler equation","cited_arxiv_id":null,"evidence_quote":"Introduces hyperbolic flows for small-scale creation in 2D Euler, here adapted to an interior hyperbolic point rather than a fixed boundary."},{"cited_title":"Bonnesen-style isoperimetric inequalities","cited_arxiv_id":null,"evidence_quote":"Supplies the quantitative Bonnesen-style isoperimetric inequality used in the proof of the confinement Lemma 3.1."},{"cited_title":"Well-posedness of the free-surface incompressible euler equations with or without surface tension","cited_arxiv_id":null,"evidence_quote":"Establishes local well-posedness for the free-boundary Euler equations with surface tension, so the constructed $H^\\infty$ initial data produce regular solutions."},{"cited_title":"Regularity Theory for Elliptic PDE","cited_arxiv_id":null,"evidence_quote":"Supplies the Calder\\'on-Zygmund estimate used to control the $C^1$ norm of the error term $e$ on $B_{1/2}$."}],"review_version":1}