{"id":"39615841-1bd9-4b22-9f18-2c2b69721f1e","arxiv_id":"2501.06988","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"MarineHydro.jl is a differentiable boundary element solver that computes hydrodynamic coefficients and their sensitivities via reverse-mode automatic differentiation, enabling gradient-based optimization of marine structures.","lead":"The authors built a new open-source solver, MarineHydro.jl, that computes wave forces on offshore structures together with their derivatives using automatic differentiation. A smart generalist might care because these gradients enable faster design optimization of wave energy converters and floating wind turbines, replacing slow finite-difference or gradient-free methods.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"FD-based mesh/hydrostatics Jacobians undercut the 'exact gradient' claim for geometry variables; the AD-vs-FD agreement in Table 3 is not an independent check.","rationale":"The most load-bearing premise is the exactness of gradients with respect to geometry variables, because the abstract and introduction promise 'exact gradients' and 'fully differentiable' for dimensions and separation, and both case studies rely on such gradients. The manuscript itself discloses that mesh and hydrostatics derivatives are computed by finite differences (Section 3.5, Section 4.5, Algorithm 1), so the only question is whether this still supports the central claim. It does not: for the design variables used in the case studies, the pipeline is not fully AD-differentiable, and the reported verification against finite differences is not an independent confirmation, since the same finite-difference approach is embedded in the custom rules. The reader's weakest_assumption identifies exactly this point, so I agree. The reader's CONDITIONAL verdict remains appropriate: the BEM-core differentiability and the verification with the Delhommeau Green's function are genuine contributions, but the exactness claim needs to be scoped to the BEM core or the mesh/hydrostatics path needs true AD. I recommend no change to the verdict, with the condition clearly requiring either restriction of the claim or implementation of AD through meshing/hydrostatics, plus documentation of the FD step and regularization constant.","tokens_in":25995,"tokens_out":2345,"duration_ms":25393,"concrete_test":"Replace the finite-difference rrule for f_mesh (Algorithm 1, Appendix D) with an analytic pullback for the sphere geometry used in Case Study I (spherical vertices parameterized explicitly by r and x), keeping the BEM core unchanged, and recompute the gradients in Table 3 and Fig. 12. In parallel, sweep the FD step h used in the custom rule over 10^-3 to 10^-7 and report the resulting gradient variation. If the analytic-pullback gradients differ from the published AD gradients by more than the reported 10^-9 tolerance, or if the gradients vary with h, then the geometry gradients are FD-limited and the 'exact gradients' claim for radius and separation distance does not hold as stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that MarineHydro.jl computes 'exact gradients' and is 'fully differentiable' with respect to inputs including body dimensions and separation distance. Section 3.5 admits that the mesh-generation function f_mesh is differentiated by finite differences ('The Jacobian of this function is then computed using finite differences'; Algorithm 1, Appendix D), and Section 4.5 confirms that geometry and hydrostatic gradients are 'estimated using finite differences'. Consequently, gradients with respect to radius r and separation x are not exact AD gradients; they inherit the truncation error and step-size sensitivity of the FD Jacobian, and 'fully differentiable' overreaches for design variables that pass through mesh generation or hydrostatics. The verification in Table 3 does not resolve this: both the 'finite difference gradient' and the 'AD gradient' use FD perturbations of the same non-differentiable external routines (Capytaine meshing and hydrostatics), so their agreement to 10^-9 may be circular rather than evidence of exactness. If the custom-rule FD step is poorly chosen, or if the mesh map is nonsmooth at the evaluated parameters, the reported geometry sensitivities are approximate, and the first case study's sensitivities with respect to r and x are not the exact gradients promised by the abstract.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents MarineHydro.jl, a Julia boundary element solver for linear potential-flow wave-structure interactions that supports direct and indirect boundary integral formulations, two free-surface Green's function options, and reverse-mode automatic differentiation for computing sensitivities. The main claimed novelty is the ability to compute hydrodynamic coefficients together with 'exact gradients' with respect to inputs such as wave frequency, body radius, and inter-body separation, using one extra linear solve via a discrete adjoint. The paper verifies the hydrodynamic coefficients against Hulme's analytical hemisphere results and Capytaine, and verifies gradients against analytical and finite-difference results. Two applications are presented: a sensitivity study of two identical floating spheres and a gradient-based optimization of the mechanical power of a pair of wave energy converters. The central claim, however, is qualified by the fact that mesh generation and hydrostatics are differentiated through custom finite-difference Jacobians, which conflicts with the abstract's 'fully differentiable / exact gradient' language for geometry variables.","tokens_in":26230,"tokens_out":7307,"duration_ms":78734,"significance":"The contribution is potentially useful: an open-source BEM solver with AD through the Green's function evaluation, influence-matrix assembly, and implicit linear solve would enable adjoint-based design optimization for wave-energy and offshore-structure problems. The manuscript includes concrete verification cases, including the exact analytical added-mass gradient for the Rankine-only sphere (Fig. 6), AD-vs-finitedifference agreement at the 1e-7 level for heave coefficients, and two worked case studies. These are genuine strengths. The significance is reduced, however, by the finite-difference treatment of mesh generation and hydrostatics: the exactness claim is only established for the hydrodynamic core, not for geometry design variables that pass through external meshing routines. The core idea remains valuable, and the issue is fixable by re-scoping claims or by implementing differentiable mesh/hydrostatics routines.","major_comments":[{"comment":"The central claim of 'exact gradients' and 'fully differentiable' is not supported for geometry design variables. The custom reverse-mode rule for mesh generation computes its vector-Jacobian product using a finite-difference Jacobian ('The Jacobian of this function is then computed using finite differences'), and Section 4.5 states that hydrostatic gradients are also 'estimated using finite differences'. Consequently, gradients with respect to radius r and separation x in the case studies are not exact derivatives of the discretized BEM problem; they inherit truncation error and step-size sensitivity from those finite-difference Jacobians. The authors themselves acknowledge the limitation in Section 3.5 and in the conclusion, so the abstract and Section 5 should either qualify the exactness claim or the missing differentiability should be supplied.","section":"Section 3.5 and Algorithm 1"},{"comment":"The AD-versus-finite-difference agreement reported for radius and separation is not an independent verification of geometry gradients. Both the 'AD gradient' (via the custom finite-difference mesh/hydrostatics rules) and the 'finite difference gradient' perturb the same external Capytaine routines, so the agreement to about 1e-9 in Table 3 primarily confirms consistency of the custom rule with a finite-difference perturbation, not exactness of the underlying mesh Jacobian. Please add an independent validation of the mesh and hydrostatics Jacobians, for example complex-step differentiation, analytic differentiation of the meshing map, or a Richardson extrapolation over finite-difference step sizes, and report the chosen step and its effect on the reported gradients.","section":"Table 3 and Figs. 7-8"},{"comment":"The scaling demonstration using 'dummy' design variables does not establish that reverse-mode AD keeps gradient cost constant for geometric design variables. The dummy variables apparently do not enter mesh generation or hydrostatics, while the actual mesh and hydrostatics Jacobians are computed by finite differences and therefore cost additional perturbations per geometric input. The constant-time result in Fig. 16 is expected for dummy variables that are not used in the computation; it does not demonstrate scalability with respect to body dimensions, shape, or layout variables. Please demonstrate scaling on actual geometric design variables or restrict the scalability claim to hydrodynamic and frequency variables.","section":"Section 4.6 and Fig. 16"},{"comment":"The adjoint derivation itself is sound for the BEM core, but the paper should be explicit that Eq. (14) only gives exact gradients if all partial derivatives of S, D, and b are exact. Since the mesh and hydrostatics contributions use finite differences, the 'one extra linear solve yields exact gradients' statement in the introduction and abstract should be carefully scoped to the hydrodynamic part of the pipeline. The current wording overstates the result for geometry-coupled optimization.","section":"Section 3.1 and Eq. (14)"}],"minor_comments":[{"comment":"The terms 'fully differentiable' and 'exact gradients' are used without the qualifications stated in Section 3.5; the abstract and conclusion should be aligned with the actual implementation, which is exact for the hydrodynamic core but finite-difference-based for mesh generation and hydrostatics.","section":"Abstract and Section 5"},{"comment":"There are typos in the caption: 'hydrodyamic' and 'with and analytical results' should read 'hydrodynamic' and 'with analytical results'.","section":"Fig. 4 caption"},{"comment":"The regularization constant epsilon in the '(d + epsilon)' distance is a free parameter, but its value and its influence on gradient accuracy are not reported; please state the value used and include a brief sensitivity check.","section":"Section 3.2"},{"comment":"The analytical added-mass expression in Eq. (16) corresponds to a full sphere in unbounded fluid, while the surrounding text and Fig. 6 mention a 'hemisphere sphere'; please clarify which geometry is being benchmarked and make the notation consistent.","section":"Eq. (16) and Fig. 6"},{"comment":"The statement that the sensitivities of added mass and damping have a 'pi/2 phase difference' appears to be a general claim drawn from a specific numerical case; please add a supporting derivation or citation, or phrase it as an observation for the computed configuration.","section":"Section 4.2"}],"recommendation":"major_revision","confidential_remarks":"The reviewer's stress-test concern is valid and is partially acknowledged by the authors in Section 3.5: the geometry and hydrostatics stage of the pipeline is differentiated by finite differences, so the 'fully differentiable / exact gradient' label is overstated. I do not see a fatal flaw in the BEM-core adjoint derivation or in the numerical verification of the hydrodynamic gradients; the issue is scoping and verification of the FD-based custom rules. A major revision that either makes mesh generation and hydrostatics genuinely AD-differentiable or carefully re-scopes all exactness and scalability claims would bring the manuscript in line with its actual contribution. The 'first application of exact gradients' novelty claim should also be benchmarked against the analytical-gradient work of Rohrer and Bachynski-Polić (2024), which the authors cite."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHere is the short version: MarineHydro.jl is a real contribution — an open-source Julia BEM solver that differentiates through the free-surface Green's function, influence matrix assembly, and linear solve via reverse-mode AD. That is new in marine hydrodynamics; Capytaine, HAMS, and NEMOH don't offer AD. The verification is convincing where it matters: the Rankine-only added mass gradient matches the analytical expression to the eye (Fig. 6), and AD-vs-FD agreement for frequency derivatives is at 1e-7 or better. The code being on GitHub helps a lot.\n\nThe soft spot is the one flagged in the stress test. The abstract and intro promise 'exact gradients' and 'fully differentiable', but geometry gradients (radius, separation) pass through mesh generation and hydrostatics, and Section 3.5 admits those Jacobians are computed by finite differences. So the sensitivities in Case Study I and the optimization gradients in Case Study II are not exact AD gradients; they carry FD error. Table 3's 1e-9 agreement between 'AD' and 'FD' doesn't solve this, because both share the same FD-perturbed mesh routine — the agreement shows consistency, not exactness. To the paper's credit, this is disclosed clearly, but the language overreaches.\n\nMinor points: the complex adjoint is written with a transpose rather than a conjugate transpose (likely fine since real and imaginary parts are split, but the text should say so); irregular frequencies are unaddressed (acknowledged); and the two-variable WEC optimization is too simple to showcase the method, as the authors admit. The dummy-variable scaling study is a nice touch.\n\nThe paper deserves a serious referee. The right revision is to either implement AD through meshing and hydrostatics or restrict the exactness claim to the BEM core, and to report the FD step size and the regularization constant. I'd accept with minor revisions.","headline":"A genuinely useful differentiable BEM solver, but the 'exact gradient' headline overreaches where geometry passes through finite-differenced mesh Jacobians.","tokens_in":26818,"tokens_out":5403,"would_cite":true,"duration_ms":48072,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"MarineHydro.jl is a boundary element solver whose wave-force answers come with exact, automatically differentiated gradients.","keywords":["automatic differentiation","boundary element method","wave-structure interaction","hydrodynamic coefficients","adjoint method","sensitivity analysis","wave energy converter","differentiable solver"],"falsifier":"Run the same radius-gradient calculation with several different finite-difference step sizes in the mesh-generation Jacobian, and through a meshing map with a kink such as a panel collocation point crossing a seam as the radius varies; if the resulting AD gradient changes with step size or fails to match a complex-step or analytic gradient, the fully differentiable claim holds only for the hydrodynamic core, not for geometry design variables. A second test is to evaluate gradients at an irregular frequency for a surface-piercing body and check whether the reported $10^{-7}$ agreement with finite differences degrades.","tokens_in":25757,"feed_emoji":"🌊","tokens_out":7113,"duration_ms":62232,"temperature":0.7,"pith_summary":"This paper introduces MarineHydro.jl, a boundary element solver for linear potential-flow wave-structure problems that is designed to be differentiable end to end. The central claim is that the solver can compute hydrodynamic coefficients, such as added mass, radiation damping, and excitation forces, together with their derivatives with respect to wave frequency, body radius, and inter-body spacing, using one forward BEM solve plus one adjoint linear solve driven by reverse-mode automatic differentiation. The authors verify these gradients against finite-difference results to $10^{-7}$ or better and against analytical results for a hemisphere, then use the gradients to optimize the mechanical power of a pair of wave energy converters. If the solver is right, it removes the finite-difference bottleneck that has limited gradient-based optimization and sensitivity analysis in marine hydrodynamics.","feed_headline":"Exact wave-force gradients from a differentiable BEM solver","feed_subtitle":"One adjoint solve per gradient evaluation opens up gradient-based WEC layout and design optimization.","key_machinery":"The load-bearing mechanism is the free-surface Green's function rendered AD-differentiable. MarineHydro.jl implements both a global polynomial approximation and an exact expression for the Green's function, and because both are expressed through ordinary polynomial, Bessel, Struve, and elementary operations, reverse-mode automatic differentiation propagates derivatives through their evaluation and through the panel integration that assembles the influence matrices. Around the linear solve, the paper uses implicit differentiation rather than unrolling the iterative solver, so the adjoint equation reuses the same factorization and only needs the transpose matrix. A custom reverse-mode chain rule, whose Jacobian for mesh generation and hydrostatics is computed by finite differences, connects external geometry routines into the otherwise AD pipeline.","core_discovery":"The core discovery is that a BEM solver for wave-structure interactions can be built so that differentiation is a core feature rather than a post-processing add-on. MarineHydro.jl assembles the influence matrices $S$, $D$, and $K$ through AD-differentiable evaluation and integration of the free-surface Green's function, then obtains the sensitivity of any scalar output by solving a discrete adjoint system $\\lambda^T D = \\partial\\mathcal{J}/\\partial\\phi$ in addition to the forward system $D\\phi - Sb = 0$. The total gradient $d\\mathcal{J}/d\\theta$ is assembled from $\\lambda$, the solution $\\phi$, and AD-computed partials $\\partial S/\\partial\\theta$, $\\partial D/\\partial\\theta$, and $\\partial b/\\partial\\theta$, costing about one extra linear solve regardless of the number of design variables. The paper demonstrates this for two interacting heaving spheres and for wave-energy-converter power optimization, reporting absolute agreement with finite differences at or below $10^{-7}$ for added mass and damping gradients and below $10^{-9}$ for the power objective. A notable feature is that the gradients are exact with respect to the discretized numerical problem, which the authors describe as a deliberate property of their discretize-then-optimize approach.","pith_inferences":["Extension beyond the paper: if the finite-difference Jacobian in the mesh chain rule were replaced by a natively AD mesh generator, the exact-differentiation claim would extend to arbitrary CAD parameters rather than just radius and separation.","Extension beyond the paper: the same adjoint engine could serve gradient-based uncertainty quantification or gradient-enhanced surrogate training, neither of which the paper demonstrates.","Extension beyond the paper: the paper's own caution about irregular frequencies points to a differentiable irregular-frequency removal scheme as the natural next stress test for the architecture."],"forward_implications":["For a fixed BEM discretization, gradients of hydrodynamic coefficients with respect to any differentiable input can be obtained at the cost of one extra linear solve, independent of the number of design variables.","The same pipeline works for both direct and indirect boundary integral formulations and for both the approximate and exact Green's functions, with approximate-Green's-function gradients agreeing with exact-Green's-function gradients to within about 2%.","Sensitivity maps of coupled added mass and damping for two heaving spheres show that off-diagonal interaction terms remain strongly oscillatory with separation, so plane-wave-approximation cutoffs should be used with care for mutual interaction terms.","Gradient-based optimization of WEC mechanical power becomes feasible with exact BEM gradients; the demonstration converged in three L-BFGS iterations to a design favoring minimal radius and maximal separation.","Because gradients scale with outputs rather than inputs, the method becomes increasingly advantageous relative to finite differences as the number of design variables grows, as shown by the dummy-variable scaling study."],"supporting_citations":[{"why":"Supplies the global approximate free-surface Green's function whose elementary-function form makes AD through matrix assembly practical.","marker":"Wu et al. (2017)"},{"why":"Provides the exact Green's function expression used as a verification benchmark and for exact-derivative checks.","marker":"Delhommeau (1987)"},{"why":"Capytaine is the comparison solver for excitation forces and supplies the external mesh and hydrostatics routines that the custom chain rules differentiate.","marker":"Ancellin and Dias (2019)"},{"why":"Gives the analytical added mass and damping coefficients for a surface-piercing hemisphere used to benchmark solver accuracy.","marker":"Hulme (1982)"},{"why":"Provides the Rankine panel integration algorithm used to handle the singular $1/R$ terms inside the differentiable pipeline.","marker":"Newman (1986)"},{"why":"Defines the plane-wave-approximation cutoff and interaction analysis that the two-sphere sensitivity case study probes.","marker":"Singh and Babarit (2013)"},{"why":"Provides the implicit-differentiation machinery used to differentiate through the linear solve without unrolling the solver.","marker":"Ning and McDonnell (2023)"},{"why":"Represents the prior adjoint-BEM effort whose finite-difference influence-matrix Jacobians and memory limits motivate the fully differentiable design.","marker":"Rohrer and Bachynski-Polić (2024)"},{"why":"Supplies the reverse-mode AD engine used to propagate gradients through the solver's computational graph.","marker":"Innes (2018)"}],"fun_headline_variants":["Adjoint BEM yields exact wave-structure gradients in one extra solve","Exact hydrodynamic gradients from a fully differentiable BEM solver","One adjoint solve gives exact power gradients for wave-energy converters","Differentiable BEM with exact adjoint gradients for marine design","Exact wave-force gradients via adjoint BEM, one extra solve"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the finite-difference Jacobian used inside the custom reverse rule for mesh generation and hydrostatics is accurate enough that the reported gradients with respect to body radius and separation can be called exact; if the finite-difference step is poorly chosen or the meshing map is nonsmooth, those geometry gradients are no longer exact AD gradients even though the BEM core itself is.","fun_headline_variants_meta":{"raw":{"variants":["Adjoint BEM yields exact wave-structure gradients in one extra solve","Exact hydrodynamic gradients from a fully differentiable BEM solver","One adjoint solve gives exact power gradients for wave-energy converters","Differentiable BEM with exact adjoint gradients for marine design","Exact wave-force gradients via adjoint BEM, one extra solve"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000763,"raw_usage":{"total_tokens":3350,"prompt_tokens":876,"completion_tokens":2474,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":492,"completion_tokens_details":{"reasoning_tokens":2385}},"tokens_in":492,"tokens_out":2474,"duration_ms":16240,"temperature":1.0,"reasoning_tokens":2385,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:50:33.164441+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same radius-gradient calculation with several different finite-difference step sizes in the mesh-generation Jacobian, and through a meshing map with a kink such as a panel collocation point crossing a seam as the radius varies; if the resulting AD gradient changes with step size or fails to match a complex-step or analytic gradient, the fully differentiable claim holds only for the hydrodynamic core, not for geometry design variables. A second test is to evaluate gradients at an irregular frequency for a surface-piercing body and check whether the reported $10^{-7}$ agreement with finite differences degrades.","supporting_citations":[],"review_version":1}