{"id":"9ac219ca-9844-45b2-ba68-96775b4c795a","arxiv_id":"2607.27385","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A calibrated kinetic surface-energy term lets the relaxed micromorphic model reproduce boundary-truncation-dependent scattering in finite metamaterials beyond scale separation.","lead":"This paper adds a surface-inertia term to a homogenized continuum model so it can tell apart finite metamaterial samples that have the same internal structure but different edge cuts. The extra parameter is fitted at one frequency and then checked at other frequencies and sizes, with mixed but mostly positive results.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Scalar surface density is not established as a frequency/size-independent physical constant; the claim rests on a single calibration point and an imperfect reference.","rationale":"Good-faith reading: the paper proposes a clean variational extension (kinetic surface energy → inertial boundary condition) and offers suggestive evidence of transfer across specimen sizes and higher frequencies. The concern is not that the model is internally inconsistent; it is that identification of ρ∂Ω is underdetermined by a single visual calibration on an imperfect reference, and no physical anchor is provided. The tables already show the correction is not uniformly beneficial, which is compatible with ρ∂Ω acting as an effective fitting parameter rather than a fixed interface property. The proposed refit is decisive because it tests the exact property—frequency independence—that makes a single scalar meaningful. The reader's CONDITIONAL verdict already reflects this fragility, so no change to the verdict is needed. Minor issue: figure captions alternate between +0.715 and −0.715 for the same parameter, which should be corrected.","tokens_in":19548,"tokens_out":6399,"duration_ms":59853,"concrete_test":"On the 5L×5L geometry, use the same L2 error metric as in Table 2 and perform a least-squares refit of ρ∂Ω at each of the ten tabulated frequencies (1.26–12.56 Mrad/s) instead of fixing the value from 12.56 Mrad/s. If the optimal ρ∂Ω varies by more than about 0.1 kg/m² across frequencies—or if refitting on a 10L×10L block at 12.56 Mrad/s gives a substantially different optimum—the load-bearing assumption of a single frequency/size-independent interface density fails and the transfer claim collapses.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central transfer claim requires that ρ∂Ω = 0.715 kg/m² is an intrinsic, frequency-independent property of the β-cut interface. The calibration in §3.3.2 determines it by direct visual comparison at a single frequency (12.56 Mrad/s) on one 5L×5L block, with the α-cut assigned ρ∂Ω = 0 as reference. That reference is not 'already reproduced' to high accuracy: Table 2 shows α-cut errors of 17–28% at high frequencies even with ρ∂Ω = 0. The fitted value may therefore be compensating bulk-model misfit rather than isolating interface inertia. The L2 tables also undercut the claimed systematic improvement: for N=5, β with ρ∂Ω = 0.715 is worse than β with ρ∂Ω = 0 at 1.26 and 2.51 Mrad/s (5.53 vs 3.96; 18.72 vs 12.14), with mixed results for N=10 and N=20. Thus the fixed scalar is an average compromise, not demonstrably a frequency-independent constant. No independent estimate of boundary-layer mass from the truncated microstructure is provided, and the surface kinetic energy in Eq. (6) is negative by construction, so the term is explicitly a relative correction. The γ-cut case is a second fit, not an unfitted prediction. The physical interpretation as the homogenized counterpart of interface inertia therefore remains unanchored.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes to augment the relaxed micromorphic model for finite-size metamaterials with a kinetic surface energy K∂Ω = −1/2 ρ∂Ω ⟨u̇,u̇⟩ (Eq. 6), which variationally yields the inertial boundary condition t = f + ρ∂Ω ü (Eq. 12). The interface density is calibrated for the β-cut against one fully resolved 5L×5L scattering simulation at 12.56 Mrad/s (§3.3.2), giving ρ∂Ω = 0.715 kg/m². The authors then report that this single scalar reproduces the β-cut response across 1.26–12.57 Mrad/s and for 5L×5L, 10L×10L, and 20L×20L specimens, while ρ∂Ω = 0 matches the α-cut. A second cut (γ) is calibrated separately and presented as validation. The claimed significance is that the homogenized model can capture truncation-dependent interface inertia in regimes where scale separation breaks down.","tokens_in":20037,"tokens_out":4899,"duration_ms":48011,"significance":"If the transfer claim held, this would be a meaningful extension of the relaxed micromorphic framework: a single fitted boundary parameter, with a clean variational derivation, would allow continuum simulations of finite metamaterial blocks at frequencies where standard homogenization fails. The derivation in §3.2 is compact and correct, and the high-frequency tables show substantial error reductions (e.g., Table 2: β-cut average error drops from 26.16% to 14.86% when ρ∂Ω = 0.715 is used). However, as presented the evidence is largely calibration-based: the γ-cut is a second fit rather than an unfitted prediction, and the low-frequency entries in the L2 tables contradict the claim of systematic improvement. The conceptual advance is therefore plausible but not yet established quantitatively.","major_comments":[{"comment":"The text states that the calibrated ρ∂Ω 'systematically improves' and 'always moves' the β-cut response closer to the fully resolved solution. Tables 2–4 contradict this at several low frequencies. For example, Table 2 (β reference) shows the L2 error increasing from 3.96% to 5.53% at 1.26 Mrad/s and from 12.14% to 18.72% at 2.51 Mrad/s when ρ∂Ω is changed from 0 to 0.715. Table 3 also worsens at 1.26 and 2.51 Mrad/s, and Table 4 worsens at 1.26 Mrad/s. Thus the scalar value is at best an average compromise over frequency, not a demonstrated frequency-independent constant. The claim must be restricted to the high-frequency regime, or a mechanism for the low-frequency degradation must be provided, before the central transfer claim can be accepted.","section":"§4.1, Tables 2–4"},{"comment":"The γ-cut is presented as an 'additional and independent validation case', but its surface density is obtained by 'the same calibration strategy adopted for the other interfaces' (§4.2). This is a second fit, not an unfitted prediction. The improvement in Table 5 (e.g., average error 19.51%→11.79% for the 5.75L specimen) only demonstrates that the parametrization can absorb another cut's response after calibration. An independent test would require predicting ρ∂Ω|γ from the boundary-layer mass distribution of the γ-truncated lattice, or calibrating ρ∂Ω|γ on one frequency/size and testing it on all other frequencies and sizes for the γ-cut without re-fitting.","section":"§4.2"},{"comment":"The α-cut is assigned ρ∂Ω = 0 as the reference because the standard model 'already reproduces the response of the α-truncated specimen'. Yet Table 2 shows that for the α reference the standard model has 17–28% relative L2 errors in the high-frequency range (average 21.54%), and the figures in the Appendix show visible discrepancies. The fitted ρ∂Ω for the β-cut may therefore be compensating bulk-model misfit rather than isolating interface inertia. The paper should report the bulk-model error as baseline uncertainty and show that the calibrated ρ∂Ω is insensitive to it—for example by re-calibrating against a corrected bulk model or by directly estimating the boundary-layer mass from the truncated microstructure.","section":"§3.3.2 and Table 2"},{"comment":"The calibration of ρ∂Ω = 0.715 kg/m² is determined 'by direct inspection' of Fig. 10 at a single frequency and specimen size, with no objective function, no uncertainty estimate, and no mesh-convergence data for the fully resolved simulations. Since the central claim is that this single scalar is a frequency- and size-independent physical property, the calibration needs to be quantitative and accompanied by sensitivity and convergence analysis. In addition, the surface energy in Eq. (6) is negative by construction; the paper should state explicitly whether this is an effective correction or a physical inertia, because the physical interpretation as 'interface inertia' rests on that distinction.","section":"§3.3.2"}],"minor_comments":[{"comment":"Several captions list ρ∂Ω = −0.715 or −0.3575 while the text calibrates positive values. The sign convention should be made consistent, or the captions corrected.","section":"Figures 12–20"},{"comment":"The captions refer to 'δ-cut' and 'δ-type boundaries' while the text and tables refer to the γ-cut. Please unify the nomenclature.","section":"§4.2, Figures 18–20"},{"comment":"The caption says 'five types of implementation' but lists four columns; check and correct.","section":"§4.1, Figure 11"},{"comment":"The header repeats µ*m; the micro-inertia tensors Jm and Te in Eq. (5) appear identical. If this is intentional, state it; otherwise correct the typographical duplication.","section":"Table 1"},{"comment":"The scale-separation breakdown threshold λ = 10.33L is inferred visually from only two specimen sizes. The paper should state the limited precision of this threshold and avoid using it as a sharp delimiter without further analysis.","section":"§2.1 and Appendix"},{"comment":"No mesh-convergence or discretization details are provided for the fully resolved microstructured simulations. Please add or cite such details so the reader can assess whether the reported discrepancies are converged numerical results.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The variational formulation and the high-frequency results are promising, but the current evidence overstates the predictive power: the γ-cut is a second calibration, the low-frequency L2 tables contradict the 'always improves' claim, and the α-cut reference is not as accurate as stated. I would ask for a reanalysis of the tables, a quantitative calibration procedure with uncertainty, and an unfitted test before considering publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The variational extension is real: adding the kinetic surface energy K∂Ω = −1/2 ρ∂Ω⟨u̇,u̇⟩ to the relaxed micromorphic action yields the inertial boundary condition t = f + ρ∂Ωu¨ cleanly, and that is a legitimate way to let one bulk model distinguish different truncations of the same lattice. That part is done carefully. The second thing: the calibration evidence is thinner than the abstract suggests, and the γ-cut story is not an independent validation.\n\nThe new piece here is the specific application to the relaxed micromorphic model, with per-cut scalar densities calibrated and then swept over frequency and specimen size. That is a real extension, not just a copy of classical surface mechanics. What the paper does well is test its calibrated density over a broad high-frequency range on 5L×5L, 10L×10L, and 20L×20L blocks. At high frequencies the improvement is visible and quantitative: for the β-cut, the average L2 error drops from 26% to 15% at N=5 and from 24% to 13% at N=10, with similar gains at N=20. That is suggestive evidence that the boundary inertia term captures something real. The authors also state in §5 that scalar surface inertia is a simplification and mention surface elasticity and tensorial inertia as needed extensions, which is honest.\n\nNow the soft spots, in proportion.\n\nThe γ-cut case is presented as \"validation,\" but §4.2 says its density ρ∂Ω|γ = 0.3575 kg/m² is found using the same calibration strategy. That is a second fit, not an out-of-sample prediction. The paper should say so explicitly.\n\nThe β-cut calibration is a direct visual fit at one frequency on one 5×5 block, and the α-cut reference is not clean: Table 2 shows α-cut errors of 17–28% even with ρ∂Ω = 0. So the fitted value may be partly compensating bulk-model misfit rather than isolating interface inertia.\n\nThe claim that the correction “systematically improves” agreement is contradicted by the paper’s own tables. For N=5, β with ρ∂Ω = 0.715 is worse at 1.26 and 2.51 Mrad/s (5.53 vs 3.96 and 18.72 vs 12.14), and the intermediate-frequency regime is mixed. The fixed scalar looks like an average compromise, not a demonstrated frequency-independent constant.\n\nNo error bars, mesh-convergence data, or code are provided, and the λ = 10.33L scale-separation threshold is inferred from two specimen sizes. The Introduction’s “fundamentally new concept” language is also over-stated; surface kinetic energy is classical, and the authors cite the relevant Gurtin–Murdoch literature.\n\nWho this is for: people working on enriched continua for finite metamaterials, especially RMM users. It deserves a serious referee because the variational framework is clean and the high-frequency transfer is plausible. My recommendation: send to peer review, expect major revision, and require an unfitted validation, a quantitative calibration protocol, and weaker interpretation of the scalar density.","headline":"A clean variational extension of the relaxed micromorphic model with boundary inertia, but the scalar calibration and the γ-cut validation are weaker than the framing claims.","tokens_in":20424,"tokens_out":3119,"would_cite":true,"duration_ms":32329,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["74Q05","74J20","74E30"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes that the inertial contribution of the boundary itself — modeled as a kinetic surface energy in the relaxed micromorphic variational principle — is what lets a homogenized continuum reproduce truncation-dependent scatte","keywords":["interface inertia","relaxed micromorphic model","homogenization","mechanical metamaterials","boundary effects","scale separation","surface kinetic energy","finite-size metamaterials"],"falsifier":"Take the calibrated value ρ∂Ω = 0.715 kg/m² and apply it to a specimen or excitation outside the fitted range — e.g., a 3×3 or 40×40 β-cut block, shear-wave incidence, or oblique angles — in fully-resolved simulation; if the enhanced model no longer tracks the β-cut scattering while ρ∂Ω = 0 tracks the α-cut, the scalar-interface-inertia hypothesis is falsified. A more direct check is to compute the mass per area of the boundary layer of partial cells exposed by each truncation plane and see whether it matches the calibrated ρ∂Ω values; if the fitted densities contradict the geometric boundary","tokens_in":19445,"feed_emoji":"〰️","tokens_out":6266,"duration_ms":53585,"temperature":0.7,"pith_summary":"This paper argues that when a periodic metamaterial is truncated at different planes, the boundary layer itself carries different inertia, and this interface inertia — not just bulk elasticity — is what makes finite specimens behave differently at high frequencies. To capture this, the authors add a kinetic surface energy to the relaxed micromorphic model, producing a boundary condition in which the interface carries an effective mass; the bulk model is unchanged. A single surface density calibrated on one 5×5 β-cut specimen at one frequency reproduces the fully-resolved β-cut scattering response across frequencies from 6.28 to 12.56 Mrad/s and for specimens up to 20×20 unit cells, while the standard zero-density model tracks the α-cut. The payoff is that homogenized models can remain usable in frequency regimes where the classical separation-of-scales assumption fails, as long as the cut geometry is accounted for by a surface inertia parameter.","feed_headline":"One scalar boundary density recreates full metamaterial response","feed_subtitle":"A single fitted surface inertia lets one homogenized model tell two lattice cuts apart across sizes and frequencies.","key_machinery":"The kinetic surface energy K∂Ω(˙u) = −½ρ∂Ω⟨˙u,˙u⟩, where ρ∂Ω is a scalar surface density assigned to the truncation plane, is the central new object. Placing it in the variational principle turns a pure bulk continuum into one whose boundary carries inertia: the first variation produces the non-coherent inertial boundary condition t = f + ρ∂Ω ü, with the sign chosen so the interface force acts like an applied traction. The relaxed micromorphic model — a continuum with macroscopic displacement u and micro-distortion P, with bulk kinetic and strain energies — supplies the bulk response; the surface term is the only new ingredient, and it preserves the variational structure of the bulk model.","core_discovery":"The central claim is that different truncations of the same periodic lattice behave like boundaries with different effective mass, and a kinetic surface energy K∂Ω = −½ρ∂Ω⟨u̇,u̇⟩ added to the relaxed micromorphic action captures this. Variation of the augmented action yields the modified boundary condition t = f + ρ∂Ω ü, in which interface inertia appears on the same footing as external tractions. Calibrating ρ∂Ω = 0.715 kg/m² for the β-cut from a single 5L×5L scattering simulation at 12.56 Mrad/s reproduces the β-cut scattering response over frequencies from 6.28 to 12.56 Mrad/s and for 5L×5L, 10L×10L, and 20L×20L specimens, while ρ∂Ω = 0 keeps the model on the α-cut. The paper is explicit","pith_inferences":["The near factor-of-two ratio between the calibrated β- and γ-cut densities hints that ρ∂Ω may equal a fully computable geometric quantity — the mass of the partial unit cells left by the truncation plane — which would turn calibration into prediction.","If the scalar density is truly a boundary-layer mass, then the same parameter should also control other boundary-driven effects, such as reflection coefficients or mode conversion at the cut face; this can be tested in direct scattering simulations.","The authors' own list of residual discrepancies in the intermediate regime suggests a tensorial or elastic surface term will be needed; a natural test is oblique incidence or shear-wave excitation, where normal and tangential interface inertia should separate.","A physical experiment with two finite blocks cut differently from the same lattice should show the scattering difference predicted here; demonstrating it outside numerics would settle that the effect is inertial rather than an artifact of parameter fitting."],"forward_implications":["Different truncations of the same metamaterial become distinguishable at the continuum level without resolving the microstructure.","One calibrated surface density carries over to other frequencies and specimen sizes, so interface inertia behaves like a property of the cut rather than a curve fit.","The variational formulation is preserved, so the surface term can be used alongside other boundary conditions and in time-domain simulations.","The γ-cut validation with a different fitted density (0.3575 kg/m²) shows the framework generalizes to other truncation planes.","Finite metamaterial building blocks can be assembled in multiscale models using homogenized descriptions at scales where fully resolved simulations are too expensive."],"fun_headline_variants":["Single boundary density recreates different material edge responses","Interface inertia lets a homogenized model sense lattice cuts","One fitted surface mass distinguishes metamaterial boundaries","Kinetic surface energy breaks scale separation in homogenization","Metamaterial model with interface inertia matches finite-size responses"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The argument rests on the premise that a single number — the extra inertia of the cut boundary — captures all the dynamic difference between truncations, independent of frequency and specimen size, with the α-cut taken as the zero reference.","fun_headline_variants_meta":{"raw":{"variants":["Single boundary density recreates different material edge responses","Interface inertia lets a homogenized model sense lattice cuts","One fitted surface mass distinguishes metamaterial boundaries","Kinetic surface energy breaks scale separation in homogenization","Metamaterial model with interface inertia matches finite-size responses"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0008,"raw_usage":{"total_tokens":3405,"prompt_tokens":846,"completion_tokens":2559,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":2485}},"tokens_in":590,"tokens_out":2559,"duration_ms":20928,"temperature":1.0,"reasoning_tokens":2485,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T08:18:45.731016+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the calibrated value ρ∂Ω = 0.715 kg/m² and apply it to a specimen or excitation outside the fitted range — e.g., a 3×3 or 40×40 β-cut block, shear-wave incidence, or oblique angles — in fully-resolved simulation; if the enhanced model no longer tracks the β-cut scattering while ρ∂Ω = 0 tracks the α-cut, the scalar-interface-inertia hypothesis is falsified. A more direct check is to compute the mass per area of the boundary layer of partial cells exposed by each truncation plane and see whether it matches the calibrated ρ∂Ω values; if the fitted densities contradict the geometric boundary","supporting_citations":[],"review_version":1}