{"id":"e1d5bb8a-6419-4653-aca7-6f29d2286113","arxiv_id":"2607.26664","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A parameter-free stabilized Morley method for surface Stokes in stream-function form achieves first-order broken-H2 and second-order broken-H1 convergence under H3 regularity, via a new normal-separated geometric estimate.","lead":"This paper introduces a stabilized Morley finite element method for the surface Stokes equations written in stream-function form, and proves it converges at optimal rates on polyhedral approximations of smooth closed surfaces. The key analytical step is a new geometric estimate showing that normal-dependent consistency errors are second-order, despite the discrete normal being only first-order accurate.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Interpolation (4.9) appears to use the exact gradient rather than its Piola pullback; taken literally, Ih violates the V_h edge-moment condition, undermining Theorems 5.6/5.7.","rationale":"The reader's weakest assumption was the standard geometric approximation of Section 2.3, which is not the main risk. The paper's new geometric estimate Theorem 3.2 appears internally coherent, and the error analysis is detailed. The most serious issue is the definition of the interpolant Ih. If (4.9) is read literally, Ih does not belong to V_h for curved polyhedral surfaces, so the interpolation estimates and the central convergence theorems do not follow. The paper's own introduction of the Piola transform indicates the intended definition, but the displayed formulas and the proof of Lemma 4.5 are inconsistent with that intent. The concern is concrete and checkable; it does not attack the geometric-estimate machinery itself. Because the likely fix is notational/correctional rather than conceptual, the verdict should be CONDITIONAL rather than REJECT: the central claims can stand once the interpolation operator is correctly defined and Lemma 4.5 is re-verified with the Piola pullback.","tokens_in":30162,"tokens_out":39879,"duration_ms":355599,"concrete_test":"Take a nonplanar two-triangle patch on a sphere sharing an edge e, with discrete outward conormals n1,n2. For w(y)=y1, compute the literal right-hand side sum S = int_e (nabla_gamma w)^e·n1 ds + int_e (nabla_gamma w)^e·n2 ds. If S is nonzero, (4.9) cannot define a function in V_h. Then recompute using the Piola pullback \\breve{\\nabla_gamma w}, for which the corresponding sum should be zero. This single computation distinguishes the literal reading from the Piola-corrected reading and determines whether the interpolation estimates (4.10) and Theorems 5.6/5.7 are valid as proved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central convergence proofs rely on the interpolant Ih lying in the discrete space V_h. The definition of V_h (Section 4.1) requires, for every interior edge, the plus-convention normal-moment condition int_e (nabla_Gamma_h w|_{K1}·n1 + nabla_Gamma_h w|_{K2}·n2) ds = 0. Equation (4.9), as written, sets the two element contributions to int_e (nabla_gamma w)^e|_K · n_K ds. Their sum is int_e (nabla_gamma w)^e · (n1+n2) ds. On a genuinely curved polyhedral surface, n1+n2 is not zero; it is O(h), and for generic w this integral does not vanish. Hence the function defined by the displayed degrees of freedom is not an element of V_h, and the subsequent interpolation estimates (4.10a–c), Corollary 4.6 jump estimates, and the decompositions in Theorems 5.6 and 5.7 that use Ih phi as a V_h comparison function lose their foundation. The text immediately before (4.9) says the surface Piola transform is used precisely so that conormal components are preserved; the Piola pullback \\breve{\\nabla_gamma w} would satisfy the required flux-matching condition. This suggests a missing breve in (4.9), but the proof of Lemma 4.5 treats the term literally as \\nabla_gamma w (not \\breve{\\nabla_gamma w}), so the manuscript as written is internally inconsistent at a load-bearing point.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces an intrinsic stabilized Morley finite element for the stream-function formulation of the surface Stokes problem on a polyhedral approximation Γ_h of a smooth closed surface γ. The method uses a parameter-free value-jump stabilization, is defined entirely through edge lengths and vertex values, and does not use discrete curvature information. The main analytical contribution is a normal-separated geometric estimate (Theorem 3.2) that converts first-order normal errors into second-order consistency for trace-free Hessians, Stokes-type tensor Green identities, and conormal fluxes. Under C^4 surface and H^3 solution regularity, the authors prove first-order broken-H^2 convergence, second-order broken-H^1 convergence, and a second-order L^2 estimate for the recovered velocity. Numerical experiments on a sphere, a torus, and a general implicit surface support the predicted rates.","tokens_in":30556,"tokens_out":61986,"duration_ms":597256,"significance":"If the results stand, this is a substantial advance. The normal-separated estimate is a clean and apparently powerful tool: it recovers the classical P_hν estimate, yields superconvergence for first-order tangential forms, and supplies the second-order interior and skeleton consistency needed for optimal broken-H^1 convergence under only H^3 regularity. The method itself is attractive: intrinsic to Γ_h, no tunable penalty parameter, no Gaussian-curvature approximation. The proofs are detailed and largely self-contained, with external results (notably a discrete Korn inequality for H(div)-conforming surface BDM spaces) clearly cited. The numerical experiments cover a nontrivial coexact case on a torus and confirm the predicted orders.","major_comments":[{"comment":"As printed, equation (4.9) defines the edge normal-derivative degrees of freedom using ∇_γ w rather than the Piola pullback \\breve{∇_γ w}. Taken literally, the two contributions on an interior edge e=∂K_1∩∂K_2 sum to ∫_e (∇_γ w)^e·(n_1+n_2) ds_h, which is not zero on a curved polyhedral surface. Hence \\tilde I_h w would not satisfy the V_h edge-moment condition, and the interpolation estimates (4.10), Corollary 4.6, and the comparison arguments in Theorems 5.6 and 5.7 would lose their foundation. The surrounding text and the use of the Piola transformation property (4.6) in the proof of Lemma 4.5 make clear that \\breve{∇_γ w} is intended. Please restore the missing breve in (4.9) and in the corresponding line of Lemma 4.5, and add one sentence explicitly verifying that the Piola pullback makes the two edge contributions cancel.","section":null}],"minor_comments":[{"comment":"The decomposition of a_h(w_I,ψ_I)-a(w,ψ) lists five terms but labels them I_1,...,I_4, and then reuses I_1 for the first two terms. This is confusing; please relabel the terms consistently.","section":null},{"comment":"In the bound for [t_h^T(H_γ w)^e n_h], the text refers to P J n_h K, but the natural object is P(n_1-n_2); the estimate is O(h^2) after using H_γ w's tangentiality, but the displayed identity should be corrected or clarified.","section":null},{"comment":"The identity ∂_{t_1}[∇_{Γ_h}w·n_h] = [t_h^T H_{Γ_h} w t_h] is asserted without proof and is not immediate. Since it underpins the crucial bound (4.22), please add a short derivation in the planar edge frame.","section":null},{"comment":"The proof is only sketched ('Sketch of proof'). In a paper whose central claim depends on this consistency bound, please expand the pointwise comparison or give a precise reference to the exact statement in [15].","section":null},{"comment":"The proof is omitted and referred to [23, Lemma 3.4]. This is acceptable, but a one-line indication of how (4.6), norm equivalence, and interpolation combine would improve readability.","section":null},{"comment":"The notation \\breve{u} is used before it is defined. Please state explicitly that \\breve{u} := \\breve{\\mathrm{curl}_γ ϕ} is the Piola pullback of the exact velocity to Γ_h.","section":null},{"comment":"The torus H^2 orders at the coarser levels (0.54, 0.77) are far from the asymptotic rate 1; only the final refinement approaches it. Please comment on this pre-asymptotic behavior or add one more refinement.","section":null}],"recommendation":"minor_revision","confidential_remarks":"The paper is strong and within scope. The only genuinely load-bearing defect is the missing Piola accent in (4.9) and in the proof of Lemma 4.5; I am confident this is a typographical error. If the authors confirm the intended \\breve{∇_γ w} and fix the other presentation issues, I would support publication. No concerns about citation practices or novelty disclosure."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Strong paper. The main analytic contribution, Theorem 3.2, is genuinely new: a normal-separated geometric estimate that contains the classical P_hν estimate as a special case and, more importantly, yields second-order consistency for the trace-free Hessian, the curl–divergence pairing, and the conormal fluxes. That is what lets the authors get optimal broken-H1 and velocity L2 rates under the natural H3 regularity, with a parameter-free stabilization and no curvature input. The proofs are detailed and the key lemmas actually compute the q_F terms; this is not a sketch. The discrete Korn inequality via the H(div)-conforming relative is a nice piece of work. The numerical experiments line up with the theory, and the destabilization test in Table 6.4 gives good evidence that the stabilization is doing something real.\n\nThe stress-test concern about (4.9) does not hold up. The displayed formula is missing a breve: the RHS should be the Piola pullback \\breve{\\nabla_γ w}, not the plain extension. The sentence before the equation says the normal-preserving property of the Piola transform is what makes the degrees of freedom well-defined, and the proof of Lemma 4.5 invokes the Piola estimate (4.6) to bound the difference between \\nabla_{Γ_h} w^e and \\breve{\\nabla_γ w}. So the intended definition preserves conormal moments and the interpolant does lie in V_h. It is a typo, and the authors should fix it, but it is not a load-bearing flaw.\n\nOther soft spots are minor. The paper leans on the authors' own [23] for the Piola estimates and on [15] for the discrete Korn inequality; both are published results and the usage is appropriate. The geometric assumptions (C4 surface, O(h^2) position/area, O(h) normals) are standard for a piecewise-linear fit. The torus experiment is fine: the solution is coexact, which sidesteps the simple-connectivity requirement, and the authors say so. No code is shipped, but the experiments are simple enough to reproduce from the text.\n\nWho should read this: anyone working on surface fourth-order problems, nonconforming methods on surfaces, or the stream-function formulation of surface Stokes. It deserves a serious referee. My recommendation: accept after minor revision — fix the missing breve and add a line in (4.9) or its caption clarifying that the RHS is the Piola pullback.","headline":"Strong paper: the new normal-separated geometric estimate is real and delivers the advertised optimal rates; the flagged interpolation bug is a typo in the displayed formula, not a load-bearing flaw.","tokens_in":30979,"tokens_out":6355,"would_cite":true,"duration_ms":59396,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65N12","65N15","65N30"],"pacs":[],"model":"deepseek-v4-flash","headline":"A stabilized Morley finite element method for the surface Stokes stream-function formulation reaches optimal convergence by upgrading first-order normal errors to second order through an integral cancellation.","keywords":["surface Stokes","stream-function formulation","Morley element","nonconforming finite element","geometric consistency","normal-separated estimate","broken H^1 error","discrete Korn inequality"],"falsifier":"Run the stabilized scheme on a manufactured C^4 surface problem with a sequence of meshes that violate the geometric approximation: vertices deliberately offset from the surface by O(h) (or strongly graded triangles), and record the broken-H^1 error order. If second order persists despite the first-order position error, the claimed dependence on the geometric approximation is wrong; if the rate drops below 2, the cancellation argument is doing the work.","tokens_in":30081,"feed_emoji":"🌊","tokens_out":6622,"duration_ms":63964,"temperature":0.7,"pith_summary":"This paper claims that a stabilized Morley finite element method for the stream-function form of surface Stokes equations converges at optimal rates on plain polyhedral surface meshes: first order in the broken H^2 norm and second order in the broken H^1 norm, with second-order L^2 recovery of the velocity, under only H^3 regularity of the stream function. The obstruction to such rates was the first-order pointwise accuracy of the polyhedral normal. The paper's main step is a new normal-separated geometric estimate showing that normal-dependent consistency errors gain a second order through an integral cancellation in the first normal variation. A sympathetic reader would care because this removes the need for higher-order surface approximation, extra solution regularity, or a tuning parameter to obtain optimal accuracy for a fourth-order surface problem.","feed_headline":"Morley surface Stokes method reaches second-order accuracy","feed_subtitle":"A geometric cancellation turns first-order normal errors into optimal convergence, with no tuned penalty.","key_machinery":"The load-bearing object is the normal-separated geometric estimate (Theorem 3.2). It treats a surface integrand F(y, ν(y)) as a function of the normal variable, Taylor-expands in that variable, and separates the derivative into tangential and normal parts. The tangential part is a tangential vector field q_F; Lemma 3.1 (normal-to-divergence estimate) shows its integral against the discrete normal is O(h^2) with constant governed by div_γ q_F, because the O(h) pointwise normal error cancels under integration. This single estimate yields the P_hν-type estimate, a superconvergence identity for first-order tangential forms, second-order consistency of the trace-free Hessian, the Stokes-type tens","core_discovery":"The central claim is Theorem 5.7: for a C^4 surface, if φ is the exact stream function and φ_h is the stabilized Morley solution, then ||φ^e − φ_h||_{H_h^1(Γ_h)} ≤ C h^2 (||φ||_{H^3(γ)} + ||f||_{L^2(γ)}), with the companion broken-H^2 bound O(h) from Theorem 5.6 and the recovered velocity u_h = curl_{Γ_h} φ_h satisfying ||u − u_h||_{L^2} ≤ C h^2 ||f||_{L^2(γ)}. These rates are optimal for piecewise quadratic elements and hold under the minimal H^3 regularity that the equation itself provides. What makes them attainable is the geometric estimate of Theorem 3.2: although the discrete normal differs from the exact normal pointwise by O(h), the integral of a tangential pairing with that normal d","pith_inferences":["The same Taylor-in-normal mechanism may extend to higher-order geometric approximations, potentially yielding higher-order nonconforming surface methods; the paper mentions this direction as work in preparation but does not prove it.","Because the argument is not specific to the Stokes stream-function equation, it may sharpen error analyses of other fourth-order surface problems (for example surface biharmonic or vector-Laplacian formulations) that currently settle for first-order geometric consistency.","The error constants depend on surface curvature through div_γ q_F; this dependence is not made explicit, so surfaces with large curvature gradients could be the first place the practical rates degrade.","A natural test is whether the cancellation survives mild violations of the geometric approximation — for example vertices offset by O(h^2) rather than lying on the surface — which would weaken the mesh-generation requirement."],"forward_implications":["On a C^4 surface, the stabilized Morley scheme converges at O(h) in the broken H^2 norm and O(h^2) in the broken H^1 norm under the natural H^3 regularity, so no super-regularity assumption is needed.","The recovered tangential velocity u_h = curl_{Γ_h} φ_h converges in L^2 at O(h^2), giving a directly usable fluid velocity from the scalar stream function.","The scheme is intrinsic to the polyhedral mesh: it uses only edge lengths and vertex/edge degrees of freedom, no discrete Gaussian curvature and no user-tuned penalty parameter.","The geometric estimate is a general tool: it recovers classical geometric consistency estimates and proves second-order consistency for several tensorial forms, so it can be reused in other surface finite element analyses."],"fun_headline_variants":["Geometric cancellation unlocks second-order surface Stokes","Morley FEM hits optimal rates via normal variation cancellation","Surface Stokes: second-order via integral geometric estimate","Stabilized Morley method: second-order on surfaces, no tuning"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the polyhedral mesh is a genuine second-order geometric approximation of a C^4 surface — vertices on the surface, O(h^2) position and area errors, O(h) normal errors, with shape-regular quasi-uniform triangles — because the cancellation that upgrades all key estimates relies on those specific error scales.","fun_headline_variants_meta":{"raw":{"variants":["Geometric cancellation unlocks second-order surface Stokes","Morley FEM hits optimal rates via normal variation cancellation","Surface Stokes: second-order via integral geometric estimate","Stabilized Morley method: second-order on surfaces, no tuning"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000168,"raw_usage":{"total_tokens":1133,"prompt_tokens":814,"completion_tokens":319,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":558,"completion_tokens_details":{"reasoning_tokens":255}},"tokens_in":558,"tokens_out":319,"duration_ms":4034,"temperature":1.0,"reasoning_tokens":255,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T11:13:04.730472+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the stabilized scheme on a manufactured C^4 surface problem with a sequence of meshes that violate the geometric approximation: vertices deliberately offset from the surface by O(h) (or strongly graded triangles), and record the broken-H^1 error order. If second order persists despite the first-order position error, the claimed dependence on the geometric approximation is wrong; if the rate drops below 2, the cancellation argument is doing the work.","supporting_citations":[],"review_version":1}