{"id":"3baaa3fc-5734-44c7-a26b-414665094f94","arxiv_id":"2411.12958","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A rigorous computer-assisted proof establishes the existence of non-convex analytic uniformly rotating vortex patches with 6-fold symmetry for the 2D Euler equation.","lead":"This paper proves that non-convex vortex patches with smooth analytic boundaries can rotate uniformly in two-dimensional fluid flow, a configuration previously only seen in numerics. The proof combines classical analysis with rigorous computer-assisted estimates around an explicit approximate solution.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The analyticity half of Theorem 1.1 rests on Proposition 5.5, whose C^8 bootstrap is not a closed proof as written.","rationale":"The reader's weakest_assumption focuses on the computer-assisted bounds, and those are indeed load-bearing: the fixed point, the invertibility of L, and the non-convexity all depend on rigorous interval enclosures, and the code is not pinned or independently checked. However, I find a more internal and equally load-bearing soft spot in Section 5.1. Theorem 1.1 claims analyticity, and the route to analyticity is Proposition 5.5 (C^8) followed by Proposition 5.6 (free-boundary elliptic regularity). The C^8 bootstrap is presented with '≲' estimates and a generic 'take δ small enough' absorption argument. A bootstrap of this kind can be made rigorous, but it must start from the actual regularity available after the fixed point, namely R ∈ H^1. The displayed derivative estimates appear to quantify B^{k+2}R before establishing its existence: for example, the bound for d^j/dx^j g uses a mean-value comparison of B^{j-l+1}R at nearby points, which requires differentiability of B^{k+1}R, and for j=k that is precisely the assertion being proved. Similarly, bounding ∂_x log A through ∂_x A/A ∼ 1/z and then multiplying by h, which is strictly positive at z=0, suggests the differentiated integral is not classically integrable unless a cancellation mechanism is exhibited. If this step fails, the analyticity conclusion of Theorem 1.1 is not obtained, even though the H^1 non-convex V-state may still exist. I am not claiming the theorem is false; rather, the proof as written has a genuine gap at a central point. A rigorous revision of Proposition 5.5 with explicit constants and a weak-form derivation would settle the issue and would likely allow the result to stand. Hence I keep the reader's CONDITIONAL verdict unchanged, with the emphasis shifted from the numerical code to the regularity bootstrap.","tokens_in":57552,"tokens_out":14037,"duration_ms":193334,"concrete_test":"Re-derive Proposition 5.5 rigorously at the first rung: starting only from the H^1 solution of Proposition 3.9 and the L^∞ bound on R' from the first step, write (1.4) in weak form and prove that R'' exists as a weak derivative, with all '≲' constants made explicit. In particular, check whether the differentiated RHS is absolutely integrable when R' is merely L^∞: if ∫ cos(z)(B_x A/A) h dz requires principal-value interpretation because h does not vanish at z=0, then the displayed estimate (5.3) is not justified and Proposition 5.5 needs a substantive repair.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The fixed-point argument yields R in H^1 (Proposition 3.9), and the first step of Proposition 5.5 upgrades only R' to L^∞. Equation (5.1) is then obtained by differentiating (1.4), but existence of R'' is exactly what is being proved; no weak formulation or mollification argument is supplied. The estimates (5.2)-(5.5) are stated with non-explicit '≲' constants and, more seriously, bound d^j/dx^j g and d^j/dx^j(B_x A/A) using ||B^{k+2}R||_{L^∞} before that quantity is known to exist or be finite. For k=0 this means ∂_x log A is treated as a function: A ~ z^2 and ∂_x A ~ z would give a 1/z singularity multiplied by h(x,x-z) ≈ R(x)^2 + R'(x)^2 > 0, which is not absolutely integrable unless additional cancellation from higher regularity is justified. The absorption step ('taking δ small enough') is therefore circular as written. Since Proposition 5.6 invokes [73, Theorem 3.1'] only after establishing C^2 regularity, Theorem 1.1's analyticity claim is not fully established by the text. This concern is independent of the computer-assisted constants and would remain even if every Arb bound is correct.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims the first rigorous construction of a non-convex, 6-fold symmetric V-state with analytic boundary for the 2D Euler equation, far from the known circle and ellipse families. The proof is computer-assisted: fixing an explicit 30-mode approximation R0 and angular velocity Ω = 1537/3750, the authors reformulate the boundary equation (1.4) as the fixed-point equation Lu = NL[u] + δ on an L² space, invert the linear operator L using a certified finite-rank approximation (Lemmas 3.4–3.7), prove contraction estimates (Section 4 and Lemma A.1), and thereby obtain a unique H¹ solution R in a ball of radius ε = 2×10⁻⁵ (Theorem 3.1, Proposition 3.9). Section 5 then attempts to bootstrap regularity from H¹ to C⁸ (Proposition 5.5), upgrade to analyticity via a free-boundary elliptic argument (Proposition 5.6), and prove non-convexity by enclosing R between R0 ± ε√(p_m) (Section 5.3). The paper includes its Arb-based C++ code and explicit numerical constants.","tokens_in":57842,"tokens_out":11645,"duration_ms":131165,"significance":"If the proof is completed, this is a substantial result: it would be the first existence proof of a non-convex V-state with analytic boundary outside a perturbative neighborhood of known solutions, and it illustrates a general computer-assisted strategy for rotating vortex patches. The fixed-point part is carefully designed: the linear inversion is reduced to a certified matrix bound (C2 = 8.8) and a certified operator approximation error (C3 = 0.085), the nonlinear estimates are stated with explicit constants, and the machine-checked bounds are backed by attached code. The non-convexity argument is short and robust. However, the analyticity claim depends on the regularity bootstrap in Proposition 5.5, and that bootstrap is not a closed proof as written; the theorem is therefore not fully established by the manuscript.","major_comments":[{"comment":"The step from the H¹ solution to C¹/C² regularity is not justified. Equation (5.1) is asserted by differentiating (1.4), but at that point R'' has not been shown to exist. The first step of Proposition 5.5 only upgrades R' to L^∞; this does not imply that the right-hand side of (1.4) is differentiable, and the text provides no weak formulation, mollification, or difference-quotient argument. This is load-bearing because Proposition 5.6 invokes [73, Theorem 3.1'] only after establishing that the boundary is C².","section":"§5.1, Proposition 5.5, Eq. (5.1)"},{"comment":"The bootstrap estimates are circular as written. The displayed inequalities use non-explicit '≲' constants and, more seriously, bound d^j/dx^j g and d^j/dx^j(B_x A / A) using ||B^{k+2}R||_{L^∞} and ||B^{j+1}R||_{L^∞}^{2j+1} before those derivatives are known to exist. For k = 0, g(x,x−z) = R(x)R'(x−z) − R'(x)R(x−z) expands as z[(R'(x))² − R(x)R''(x)] + o(z), so the local contribution to the right-hand side of (5.1) contains R'' itself. The absorption step 'taking δ small enough' therefore cannot close unless it is applied to a uniformly regularized family with a priori control on the second derivative; no such family is supplied.","section":"§5.1, estimates (5.2)–(5.5)"},{"comment":"Because the C² bootstrap in Proposition 5.5 is not closed, the hypothesis needed to apply [73, Theorem 3.1'] is not verified, so the analyticity conclusion of Theorem 1.1 does not follow from the manuscript as written. This concern is independent of the computer-assisted constants and would remain even if every Arb bound in the paper is correct.","section":"§5.2, Proposition 5.6"}],"minor_comments":[{"comment":"In the first displayed computation of ∇^⊥φ·t, the second integral is written as '∫ C R'(y) + S R'(y) dy'; from the preceding identity (5.8) the second term should be S R(y), not S R'(y). Please correct this typo.","section":"§5.2, Eq. (5.9)"},{"comment":"The notation mR0, MR0, MR1_0, MR2_0 in Appendix A is inconsistent with the notation m_R, M_R, M_{R1_0}, M_{R2_0} used in the main text; unify the notation for readability.","section":"Appendix A, Lemmas A.7–A.9"},{"comment":"In the final displayed integral of the proof of (A.2), there is a stray 'dz' at the end of the line and the variable z is used both as the integration variable and as an endpoint in the preceding line; please clean up the display.","section":"Lemma A.2, proof"},{"comment":"The corollary asserts that the same R0 remains an approximate solution for Ω̃ close to Ω, but the defect bound CE0 is proved only at the fixed value Ω. A short justification that the defect and all relevant bounds vary continuously with Ω would make this corollary fully rigorous.","section":"Corollary 1.2"}],"recommendation":"major_revision","confidential_remarks":"The computer-assisted fixed-point argument appears carefully executed and I did not find a flaw in the L² contraction machinery; my reservations are purely analytic and localized in Section 5. If the authors can supply a rigorous regularity bootstrap—for example via difference quotients, a C¹ base step, or an alternative regularity theorem—the result would be a strong and publishable contribution. The paper should not be rejected on the computer-assisted part; it should be revised to close the analyticity gap."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about Castro-López and Gómez-Serrano's arXiv:2411.12958. First, the core fixed-point result is real: they construct an H^1 solution to (1.4) near an explicit 6-fold approximate solution, with verified computer-assisted bounds and attached Arb code. If the computer-assisted lemmas are correct, the existence of a non-convex 6-fold symmetric V-state is established, and that would be the first rigorous construction far from the perturbative regime. Second, the advertised upgrade to analyticity is not proven as written. The regularity bootstrap in Proposition 5.5 has a genuine gap.\n\nThe fixed-point part is handled carefully: finite-rank approximation of the kernel, Neumann series, explicit constants, and the contraction argument closes with epsilon = 2e-5. The nonlinear estimates in Section 4 are lengthy but appear systematic. The non-convexity argument (Section 5.3) is clean and only needs the L^∞ closeness to R0.\n\nThe problem is the bootstrap. After obtaining R in H^1, they first show R' is L^∞. That step is okay. Then they differentiate (1.4) to write equation (5.1) for R''. But R'' has not been shown to exist. No weak formulation, mollification, or difference quotient argument is given. The estimates (5.2)–(5.5) compound this: to bound derivatives of g and B_xA/A near z=0, they use MVT differences that require ||B^{k+2}R||_{L^∞} — exactly the quantity the induction step is trying to bound. For k=0 this is circular. The integrand B_xA/A ~ 1/z, and without the extra cancellation from the |z| factor in g, the principal-value integrals do not converge absolutely. The sentence 'taking delta small enough' does not resolve this because the problematic term is present at every scale. The stress-test note is correct on this point, and this is independent of the computer-assisted constants.\n\nSo: the existence statement for a non-convex V-state is likely correct, but the analyticity claim is not supported by the text as it stands. The paper deserves a serious referee — the main idea is valuable and the gap seems repairable, perhaps via elliptic regularity on the free boundary problem or a more careful bootstrap using the equation's structure. I would send it to peer review, asking the authors to supply a fully rigorous regularity argument before the analyticity claim is accepted.","headline":"Solid fixed-point construction of a non-convex V-state, but the analyticity upgrade rests on a regularity bootstrap in Proposition 5.5 that is not a closed proof as written.","tokens_in":58340,"tokens_out":4828,"would_cite":false,"duration_ms":52101,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q31","35R35","76B47","35B32"],"pacs":[],"model":"deepseek-v4-flash","headline":"There exists an analytic, non-convex, six-fold symmetric uniformly rotating vortex patch, proved by a computer-assisted fixed-point argument.","keywords":["V-states","vortex patches","2D Euler equations","analytic boundary","non-convex","computer-assisted proof","6-fold symmetry","rigorous numerics"],"falsifier":"Re-implement the certified computations in a different validated-arithmetic system and verify the key inequalities: the defect bound $\\|E[0]\\|_{L^2} \\le 3 \\times 10^{-8}$, the matrix bound $\\|(I+A)^{-1}\\|_2 \\le 8.8$, the operator error $\\|L-L_F\\|_2 \\le 0.085$, and the positivity of the constant $C_{K_1}$ in Lemma 5.3. The first violated certified bound would invalidate the proof; if all hold, Theorem 1.1 stands as argued.","tokens_in":57342,"feed_emoji":"🌀","tokens_out":9956,"duration_ms":97382,"temperature":0.7,"pith_summary":"Uniformly rotating vortex patches of the 2D incompressible Euler equation, called V-states, have only two explicit families: circles and Kirchhoff ellipses. This paper proves that a third kind exists: a non-convex, six-fold symmetric vortex patch whose boundary is an analytic curve and which sits outside the perturbative regime where earlier bifurcation arguments apply. The proof starts from an explicit cosine-polynomial approximate solution, rewrites the boundary equation as a fixed-point problem for a small correction, and closes the argument with certified computer-assisted bounds. If the proof is correct, this is the first rigorous construction of a non-convex V-state with analytic boundary, and it validates numerically predicted shapes that had no existence proof.","feed_headline":"First analytic non-convex V-state shown to exist","feed_subtitle":"Only circles and ellipses were known; this proof adds a six-fold symmetric patch.","key_machinery":"The load-bearing object is the linearized operator $L = I + K$ acting on $X_m = L^2([0,\\pi/m])$, with $m=6$, defined by $Lu(x) = u(x) + \\int_0^{\\pi/m} K(x,y)u(y)\\,dy$. Because $\\|K\\|_2 > 1.3$, a simple Neumann bound does not work; instead $K$ is approximated by a finite-rank operator $K_F$ built from the first 201 Fourier modes, and the paper certifies $\\|(I+A)^{-1}\\|_2 \\le 8.8$ and $\\|L - L_F\\|_2 \\le 0.085$ using rigorous interval arithmetic. A Neumann series then yields $\\|L^{-1}\\|_2 \\le 35$, and the final fixed-point theorem is a Banach contraction on the ball of radius $\\varepsilon$. Analyticity of the boundary is obtained by recasting the rotating patch as a free-boundary elliptic problem and applying the regularity theorem, using that the rotating-frame velocity is nonzero along the boundary.","core_discovery":"The central claim is Theorem 1.1: there exists an analytic solution $R(x)$ of the boundary equation $RR' = F[R]$ that parametrizes a vortex patch $D \\subset \\mathbb{R}^2$ which is non-convex and has 6-fold symmetry. The solution is built as $R = R_0 + v$ with $v = \\int_0^x \\tilde{u}$, where $R_0$ is an explicit 30-term cosine polynomial. The perturbation $u$ solves $Lu = N_L[u] + \\delta$ on $L^2([0,\\pi/6])$; the paper proves $L$ is invertible, with norm bound $\\|L^{-1}\\| \\le 35$, and that the nonlinear map is a contraction on a ball of radius $\\varepsilon = 2 \\times 10^{-5}$. Regularity is then bootstrapped from $H^1$ to $C^8$, upgraded to analyticity through the free-boundary elliptic formulation, and non-convexity is certified by enclosing $R$ between explicit functions $R_0(x) \\pm \\varepsilon \\sqrt{p_6(x)}$.","pith_inferences":["Editorial inference: the same scheme should work for other $m \\ge 4$ symmetry branches; the $m = 6$ choice specifically avoids $m = 2,3$, where branches appear to remain convex, so the bottleneck is computing an accurate approximate solution with small certified defect rather than any structural obstruction.","Editorial inference: because all the certified constants are explicit, the fixed-point argument could be rerun with $\\Omega$ as an interval parameter to extract a concrete interval of angular velocities, not merely the open neighborhood asserted by Corollary 1.2.","Editorial inference: the quantitative non-convexity certificate suggests a testable numerical prediction—continuing from this solution should immediately produce nearby non-convex analytic V-states, which could serve as an independent check of the certified bounds."],"forward_implications":["The constructed V-state is quantitatively controlled: its boundary lies between explicit envelopes and has exactly six-fold symmetry.","Corollary 1.2 gives an open interval of angular velocities around $\\Omega = 1537/3750$ for which such non-convex analytic patches also exist.","This is the first existence result for V-states that supplies quantitative information outside the small neighborhoods of the circle and the ellipses used in local bifurcation theory.","The computer-assisted fixed-point scheme is designed to be applicable to other branches and other active scalar equations whenever an approximate solution with sufficiently small certified defect can be computed.","The proof also rules out loss of $C^2$ regularity at the constructed boundary, since analyticity follows once the free-boundary problem has $C^2$ data."],"supporting_citations":[{"why":"Supplies the numerical continuation and Newton-Raphson method that produces the approximate solution $R_0$ about which the whole perturbation argument is built.","marker":"[28]"},{"why":"Gives the local bifurcation branches of m-fold patches from the disk that the m=6 branch is a numerical continuation of.","marker":"[9]"},{"why":"Provides the rigorous eigenvalue-enclosure lemma used to certify invertibility of the finite-rank matrix $I+A$ and the bound $\\|(I+A)^{-1}\\| \\le 8.8$.","marker":"[49]"},{"why":"Is the arbitrary-precision interval-arithmetic library used to compute every certified bound in the computer-assisted part.","marker":"[71]"},{"why":"Is the free-boundary elliptic regularity theorem that upgrades the $C^2$ boundary to analyticity once the velocity in the rotating frame is shown nonvanishing.","marker":"[73]"}],"fun_headline_variants":["Non-convex V-states exist with analytic boundary","Computer-assisted proof shows non-convex V-states","First analytic proof of non-convex vortex patches","Six-fold symmetric non-convex V-state proven to exist","Analytic non-convex V-states beyond known examples"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof stands on the correctness of the computer-assisted bounds carried out with interval arithmetic—the defect estimate, the matrix and operator norm bounds, and the constant checks—so a bug in any of those certified computations would collapse the fixed-point argument.","fun_headline_variants_meta":{"raw":{"variants":["Non-convex V-states exist with analytic boundary","Computer-assisted proof shows non-convex V-states","First analytic proof of non-convex vortex patches","Six-fold symmetric non-convex V-state proven to exist","Analytic non-convex V-states beyond known examples"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000742,"raw_usage":{"total_tokens":3251,"prompt_tokens":826,"completion_tokens":2425,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":442,"completion_tokens_details":{"reasoning_tokens":2351}},"tokens_in":442,"tokens_out":2425,"duration_ms":18036,"temperature":1.0,"reasoning_tokens":2351,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:00:09.587855+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-implement the certified computations in a different validated-arithmetic system and verify the key inequalities: the defect bound $\\|E[0]\\|_{L^2} \\le 3 \\times 10^{-8}$, the matrix bound $\\|(I+A)^{-1}\\|_2 \\le 8.8$, the operator error $\\|L-L_F\\|_2 \\le 0.085$, and the positivity of the constant $C_{K_1}$ in Lemma 5.3. The first violated certified bound would invalidate the proof; if all hold, Theorem 1.1 stands as argued.","supporting_citations":[{"cited_title":"G ´omez-Serrano and G","cited_arxiv_id":null,"evidence_quote":"Provides the rigorous eigenvalue-enclosure lemma used to certify invertibility of the finite-rank matrix $I+A$ and the bound $\\|(I+A)^{-1}\\| \\le 8.8$."},{"cited_title":"Johansson","cited_arxiv_id":null,"evidence_quote":"Is the arbitrary-precision interval-arithmetic library used to compute every certified bound in the computer-assisted part."},{"cited_title":"Kinderlehrer, L","cited_arxiv_id":null,"evidence_quote":"Is the free-boundary elliptic regularity theorem that upgrades the $C^2$ boundary to analyticity once the velocity in the rotating frame is shown nonvanishing."}],"review_version":1}