{"id":"a05bcb64-cb72-4530-9a77-d512edf56f6a","arxiv_id":"2507.20874","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Strong convergence rates are proved for two-scale expansions of solutions to highly oscillatory diffusion and elasticity problems on thin plates, including the previously open bending case in two dimensions.","lead":"This paper proves that two-scale approximations of solutions to oscillatory diffusion and elasticity problems on thin plates converge in the energy norm as the plate thickness and heterogeneity scale vanish. It provides the first strong convergence result for the bending case of heterogeneous elastic plates, using a new proof strategy.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: main theorems are conditional on explicitly stated W^{1,∞} corrector/limit regularity, which is not automatic for L∞-elliptic coefficients but is disclosed.","rationale":"I read the main proofs, focusing on the novel bending-case machinery: Lemma 3.9 (the d=2 technical lemma), Lemma 3.15 (compactness of the rescaled stress), Lemma 3.17 (div-curl-type identification of Σ⋆), Lemma 3.19 (regularity of Σ⋆), Lemma 3.20 (the exact identity ∫(Σε−Σ⋆):e(v)=0), and the final assembly in Theorem 3.21. The arguments are internally coherent: the negative-Sobolev compactness is justified, the pointwise-in-xd zero-mean checks needed for Lemma 3.9 are present, and Lemma 3.20 follows from the established L2 regularity of Σ⋆ plus density. The restriction to d=2 is explicitly tied to Lemma 3.9. The only substantive soft spot is the corrector/solution W^{1,∞} regularity, which the reader already identified. It is genuinely load-bearing because the pointwise definition of the expansion and the boundary-layer estimates require L∞ control of correctors and their gradients, and this property does not follow from the natural L∞-ellipticity assumptions on A. However, the authors state these assumptions explicitly and do not claim the results for merely bounded coefficients. Thus the central claims hold as stated, and the verdict ACCEPT remains appropriate.","tokens_in":82906,"tokens_out":28224,"duration_ms":288879,"concrete_test":"Use a fine FEM or an exact laminate solution to solve the 2D cell problems (69) and (71) for a piecewise-constant, symmetric (80) periodic elastic tensor. Compute ess-sup|∇w_αβ| and ess-sup|∇W^{αβ}|; for discontinuous coefficients these norms grow without bound as mesh refines, confirming that the W^{1,∞} hypothesis fails for a standard class and that the theorems' rates do not apply there. For a C^{1,α} coefficient, verify the norms are finite, supporting the assumption's validity in the smooth setting.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The only load-bearing concern is the W^{1,∞} regularity assumed for the cell correctors (w_αβ, W^{αβ}) and the homogenized solution (e(u⋆+g), ∇²(u⋆_d+g_d)). This regularity is used essentially: the two-scale expansion is defined pointwise through corrector values, and the Step 1 boundary-cutoff estimates (e.g., (193)-(196) in Theorem 3.21, and the corresponding steps of Theorems 2.5 and 3.12) plus the applications of Lemma 3.9 require L∞ bounds on the correctors and their gradients. For periodic coefficients that are merely bounded and elliptic (e.g., piecewise-constant composites), correctors are only in H^1, so the assumption fails and the stated rates are not established in that physically common situation. This is a genuine limitation, but it is explicitly disclosed in the theorem statements and is standard in two-scale-expansion rate results; I did not find an internal gap in the proofs of Lemmas 3.9, 3.15, 3.17, 3.19, 3.20 or Theorems 3.12 and 3.21.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proves strong convergence results, in the scaled H^1_ε norm, between solutions of highly oscillatory problems on thin plates and their two-scale expansions. Section 2 treats scalar diffusion: under the shape regularity condition (20) and regularity assumptions on the homogenized solution and the cell correctors, Theorem 2.5 gives an explicit √ε bound with the dependence on |ω| tracked. Section 3 treats linear elasticity under the mirror symmetry assumption (80). The membrane case is handled in Theorem 3.12 by a careful adaptation of the diffusion argument, while the bending case, Theorem 3.21, requires a genuinely different strategy: one introduces rescaled stress fields Σε, proves weak compactness and regularity (Lemmas 3.15 and 3.19), identifies the in-plane stress limit through enriched correctors (Lemma 3.17), and then uses the exact identity ∫(Σε−Σ⋆):e(v)=0 for all v∈V (Lemma 3.20) to close the error estimate. Both elasticity theorems are restricted to d=2 because the auxiliary Lemma 3.9 is established only in that dimension. Appendices A–D collect the rescaling identities, H^div trace results, Korn inequalities, and self-contained proofs of the weak homogenization theorems.","tokens_in":83145,"tokens_out":14687,"duration_ms":191612,"significance":"If the results are correct, this is a valuable contribution: strong convergence rates for two-scale expansions in heterogeneous elastic plates were previously available in very few settings, and the bending-case argument is genuinely new. In particular, the exact stress identity of Lemma 3.20, and the use of enriched bending correctors to identify the in-plane stress limit in Lemma 3.17, are substantial technical advances. The explicit dependence of the constants on |ω| is important for the companion MsFEM error analysis. The proofs are detailed and the main theorems are conditional in a transparent way: they assume W^{1,∞} regularity of the cell correctors and W^{2,∞}-type regularity of the homogenized quantities, which is standard in two-scale expansion rate results but does not follow from mere L∞ ellipticity; the d=2 restriction is explicitly traced to Lemma 3.9. I found no internal derivation gap, and the stress-limit identification is not circular, as Remark 3.18 correctly notes.","major_comments":[],"minor_comments":[{"comment":"The symbol B is first used for J^i_{·,j}·∇_εφ and then for the symmetrized matrix eB+eB^T; this makes the manipulations in (95) and the following paragraph hard to follow. Rename the second object, for example B^{sym}, or use a different font.","section":"§3.5.1, proof of Lemma 3.9, around (88)–(95)"},{"comment":"The displayed exponents for |ω| in the bounds on E_0 and E_1 appear scrambled (e.g. |ω|^{d−1 d−2}); they should read |ω|^{(d−2)/(d−1)}. Since the theorem is stated for d=2 the issue is harmless, but the typography should be corrected for readability.","section":"Theorem 3.21, Step 1, bounds for E_0 and E_1"},{"comment":"The proof of Theorem 3.21 constructs Σ⋆ along a weakly convergent subsequence, and the statement of the theorem concerns the whole sequence. Uniqueness of the full matrix Σ⋆ follows from Lemma 3.17 for the in-plane block and from the formulas in Lemma 3.19 for the remaining components; it would help to state this explicitly so that the reader sees immediately that the estimate holds without a final subsequence argument.","section":"§3.5, Theorem 3.21 proof and Lemmas 3.15–3.20"},{"comment":"When Lemma 3.17 is invoked to establish the bending homogenized equation before (84) has been proved, it is not immediately obvious that the argument is non-circular. A sentence referring the reader to Remark 3.18, and noting that u⋆∈V^B_KL follows from the symmetry of uε before the homogenized equation is identified, would remove potential confusion.","section":"Appendix D.3"}],"recommendation":"accept","confidential_remarks":"This is a strong paper with detailed, coherent proofs and a new strategy for the bending case. The unconditional limitations — d=2 for elasticity, and the W^{1,∞} corrector regularity assumptions — are explicitly disclosed and are standard for this type of rate result. I have no concerns requiring further review; the minor comments are purely typographical or expository."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper delivers what its title promises: H^1_epsilon error estimates for two-scale expansions in heterogeneous plates, in the diffusion, membrane, and bending cases. The genuinely new part is the bending case (Theorem 3.21), where the authors replace the standard test-function route with an enriched family of correctors (166) and a stress-identification lemma (Lemma 3.17). That strategy is not in the prior strong-convergence literature, which only handled homogeneous plates [8,10,11] or weak convergence [4,5]. The proofs are detailed and, as far as I can tell, coherent. The stress-tensor identification (170) is a real div-curl-type argument, and Lemma 3.20 gives the exact identity (no limit) needed to close the estimate. The paper also tracks the dependence on |ω| and on the loads, which is the right thing for the MsFEM application in their companion paper.\n\nSoft spots, in proportion:\n\n1. The W^{1,∞} regularity assumptions on the correctors and on the homogenized solution are load-bearing. For merely L^∞-elliptic period coefficients (e.g., piecewise-constant composites), correctors are only H^1, so the rates in Theorems 2.5, 3.12, and 3.21 are not established in that common physical setting. This is a genuine scope limitation, but it is stated plainly in each theorem, and it is standard for two-scale-expansion rate results. I don't see it as a hidden flaw.\n\n2. The d=2 restriction for elasticity is explicit: Lemma 3.9 is proved only for d=2, as the authors disclose at length. That is a real restriction for plate applications (d=3), but not an internal gap.\n\n3. The regularity assumptions on f, g, h are somewhat strong (e.g., gd ∈ H^4(ω) in the bending case), but they are compatible with elliptic regularity on u⋆ and are not hidden.\n\nI checked the stress-test concern: it correctly identifies that the W^{1,∞} corrector regularity is not automatic and is needed for the pointwise cutoff estimates. That is a fair concern, but it is already in the paper as a stated assumption, not an unacknowledged gap. The stress-test note itself makes clear the proofs appear internally gap-free. I agree with that.\n\nThere is no circularity or fitted constants. The estimates are derived from the original PDE and corrector problems; the stress limit is characterized, not calibrated. The citation pattern is appropriate; the companion-paper reference [13] is in preparation, but the present results stand alone.\n\nWho should read it: anyone working on rigorous numerical analysis of MsFEM for plates, or on strong-convergence homogenization in thin domains. It deserves a serious referee; the main theorems are tightly stated and proven with unusual care about the local dependence on ω. My own verdict is accept, with the standard caveat that the W^{1,∞} regularity assumption should be highlighted in the abstract or introduction so users see it immediately.\n\nRecommendation: engage with it. Send to a referee who knows homogenization and plate theory, and expect the referee to verify Lemma 3.17 and the Step 1 boundary estimates in Theorem 3.21, but not to find a fatal gap.","headline":"A rigorous, well-disclosed proof of new H^1_epsilon strong-convergence estimates for two-scale expansions of heterogeneous plates; the bending case is genuinely new and the main limitation (W^{1,∞} corrector regularity, d=2) is openly stated.","tokens_in":83655,"tokens_out":842,"would_cite":true,"duration_ms":13376,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B27","74K20","74Q05","74B05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes order-$\\sqrt{\\varepsilon}$ strong convergence of two-scale expansions for oscillatory plate problems, including the previously resistant bending case.","keywords":["two-scale expansion","homogenization","thin plates","linear elasticity","membrane problem","bending problem","strong convergence","periodic coefficients"],"falsifier":"Numerically compute the two-scale error $\\|u^\\varepsilon-u^{\\varepsilon,1}\\|_{H^1_\\varepsilon(\\Omega)}$ for a two-dimensional periodic bending problem whose cell corrector $W^{11}$ is only in $H^1$ (for instance a coefficient field with strong periodic discontinuities or corner-like oscillations), and check whether the rate is still $O(\\sqrt{\\varepsilon})$; if the rate worsens, the $W^{1,\\infty}$ regularity assumptions are load-bearing rather than technical.","tokens_in":82736,"feed_emoji":"📐","tokens_out":7599,"duration_ms":90978,"temperature":0.7,"pith_summary":"This paper proves that, for heterogeneous plates whose in-plane microstructure varies on the same scale $\\varepsilon$ as the thickness, the true solution is approximated by an explicit two-scale expansion with an error of order $\\sqrt{\\varepsilon}$ in the natural scaled energy norm. The first main result covers linear diffusion in arbitrary dimension; the constant is tracked explicitly in the size of the plate, which is what a numerical error analysis needs. For linear elasticity, under a classical mid-plane symmetry of the elasticity tensor, the problem decouples into a membrane and a bending problem, and the paper proves the membrane estimate in dimension two by adapting classical arguments. The bending case is the real obstacle: the classical proof scheme fails because no admissible $H^2_0(\\omega)$ test function can be built from an arbitrary $H^1$ test field, and the authors replace it by a new strategy that identifies the weak limit of the stress tensor and then uses an exact identity that holds before passing to the limit.","feed_headline":"Two-scale expansions match oscillatory plates to order √ε","feed_subtitle":"A new proof covers bending, where classical homogenization arguments fail; rates are explicit in plate size.","key_machinery":"The load-bearing object is the two-scale expansion itself, built as $u^\\star(x')+\\varepsilon\\,w_\\alpha(x'/\\varepsilon,x_d)\\,\\partial_\\alpha(u^\\star+g)$ for diffusion, $u^\\star+\\varepsilon\\,w_{\\alpha\\beta}\\,e_{\\alpha\\beta}(u^\\star+g)$, with the transverse component scaled by $\\varepsilon^2$, for the membrane case, and $u^\\star+\\varepsilon\\,W^{\\alpha\\beta}\\,\\partial_{\\alpha\\beta}(u^\\star_d+g_d)$ for the bending case. The rate is carried by three mechanisms: the scaled differential operators $\\nabla_\\varepsilon$ and $e_\\varepsilon$, which encode the plate scaling in the $H^1_\\varepsilon$ norm; the oscillation lemmas (Lemma 2.9 and its matrix counterpart Lemma 3.9) that kill the resonant in-plane oscillation of the corrector residual $Z(x'/\\varepsilon,x_d)$ by representing it as the divergence of a periodic skew-symmetric potential and exploiting $\\varepsilon\\nabla_\\varepsilon$, and which, for $d=2$, use a two-scale Taylor expansion of the test displacement; and, in the bending case, the enriched correctors $W^{\\alpha\\beta,\\xi}$ together with the exact identity $(\\Sigma^\\varepsilon-\\Sigma^\\star):e(v)=0$, which replaces the unavailable homogenized test function by a stress-limit characterization.","core_discovery":"The paper's central claim is that the two-scale ansatz—homogenized displacement plus cell corrector applied to the in-plane strain in the membrane case, or to the Hessian of the transverse displacement in the bending case—is accurate at order $\\sqrt{\\varepsilon}$ in the $H^1_\\varepsilon$ norm for solutions of elliptic problems on plates of thickness $\\varepsilon$ with $\\varepsilon$-periodic in-plane heterogeneities. In the diffusion case (Theorem 2.5) this holds in any dimension with a constant explicit in $|\\omega|$. In elasticity, under the symmetry assumption (80) and in dimension two, the membrane case (Theorem 3.12) and, by a different and new argument, the bending case (Theorem 3.21) both deliver the same $\\sqrt{\\varepsilon}$ rate, with the bending bound containing the data norms $N^{\\mathrm{bend}}$ and $N^{\\mathrm{bend}}_\\Omega$. The genuinely new ingredient is the bending proof: instead of constructing a homogenized test function from the error, it passes to the limit in the stress through enriched bending correctors $W^{\\alpha\\beta,\\xi}$, identifies the in-plane stress limit $\\Sigma^\\star_{\\alpha\\beta}=S^\\star_{\\alpha\\beta\\gamma\\delta}\\partial_{\\gamma\\delta}(u^\\star_d+g_d)$, upgrades the regularity of $\\Sigma^\\star$, and then uses the exact identity $\\int_\\Omega (\\Sigma^\\varepsilon-\\Sigma^\\star):e(v)=0$, valid for every $v$ without taking $\\varepsilon\\to0$.","pith_inferences":["As an extension the authors leave implicit: because the bending proof produces an explicit tensor $S^\\star_{\\alpha\\beta\\gamma\\delta}$ and an exact identity $(\\Sigma^\\varepsilon-\\Sigma^\\star):e(v)=0$ for every admissible $v$, the same machinery should yield a computable a posteriori error indicator proportional to $\\|\\Sigma^\\varepsilon-\\Sigma^\\star\\|_{L^2}$ for plate multiscale methods; this is our","A testable extension is whether the rate sharpens to $O(\\varepsilon)$ when the data are smoother: the displayed bound already contains an $\\varepsilon^{3/2}N^{\\mathrm{bend}}$ term, but removing the leading $\\sqrt{\\varepsilon}\\,|\\omega|^{(d-2)/(2(d-1))}$ term would require a boundary-layer correction of higher order than the paper constructs.","For random stationary coefficients the periodic divergence-potential argument used in Lemmas 2.9 and 3.9 has no direct analogue, so an open and natural test is whether the same $\\sqrt{\\varepsilon}$ two-scale rate survives in expectation for random microstructures."],"forward_implications":["The estimates make two-scale expansions a legitimate tool for numerical analysis of plate multiscale methods: local elements of size $|\\omega|^{1/(d-1)}$ inherit explicit error bounds, the application the paper identifies as pivotal for the companion MsFEM analysis.","The diffusion theorem upgrades the known weak homogenization limit for thin domains to strong convergence in the scaled $H^1$ norm in arbitrary dimension, with no restriction to $d=2$.","Under the symmetry assumption (80), membrane and bending problems are each approximated at the same $\\sqrt{\\varepsilon}$ order in dimension two, so the decoupled plate model is justified beyond weak convergence.","The bending theorem supplies a strong-convergence result for heterogeneous plates in the bending regime, and its proof route—identify the stress limit through enriched correctors, then freeze the identity before $\\varepsilon\\to0$—is new to this setting.","If the two-dimensional technical Lemma 3.9 is ever proven in dimension $d\\ge3$, the same statements follow in all dimensions, as the authors explicitly note."],"supporting_citations":[{"why":"Supplies the weak homogenization limit for diffusion in thin domains, which Theorem 2.5 upgrades to strong two-scale convergence.","marker":"[4]"},{"why":"Supplies the homogenized plate model and Theorem 3.1, the weak convergence result that the membrane and bending theorems refine.","marker":"[5]"},{"why":"Provides the homogeneous-plate comparison strategy that the bending-case proof adapts to heterogeneous plates.","marker":"[11]"},{"why":"Provides the classical strong two-scale convergence arguments and the periodic divergence-potential lemma used in Lemma 2.9.","marker":"[20]"},{"why":"Gives the arbitrary-order asymptotics for homogeneous clamped plates that set the context and benchmark for strong plate convergence.","marker":"[10]"},{"why":"Supplies the Korn inequalities used for coercivity of the elasticity problem and for the weighted Poincaré estimate in Lemma 3.4.","marker":"[6]"},{"why":"Provides the H-convergence and div-curl background behind the stress-limit identification in Lemma 3.17.","marker":"[28]"}],"fun_headline_variants":["Strong convergence proven for two-scale plate expansions","New proof: two-scale expansion converges for bending plates","Bending plates: two-scale expansion proven at order √ε","Two-scale expansions validated for elastic plates at √ε"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proofs require the cell correctors and the homogenized displacement to be Lipschitz, and in the bending case the transverse displacement to be smooth up to fourth order, regularity that is not automatic for general bounded elliptic coefficients.","fun_headline_variants_meta":{"raw":{"variants":["Strong convergence proven for two-scale plate expansions","New proof: two-scale expansion converges for bending plates","Bending plates: two-scale expansion proven at order √ε","Two-scale expansions validated for elastic plates at √ε"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000183,"raw_usage":{"total_tokens":1360,"prompt_tokens":1033,"completion_tokens":327,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":649,"completion_tokens_details":{"reasoning_tokens":264}},"tokens_in":649,"tokens_out":327,"duration_ms":4587,"temperature":1.0,"reasoning_tokens":264,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T13:11:02.175156+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically compute the two-scale error $\\|u^\\varepsilon-u^{\\varepsilon,1}\\|_{H^1_\\varepsilon(\\Omega)}$ for a two-dimensional periodic bending problem whose cell corrector $W^{11}$ is only in $H^1$ (for instance a coefficient field with strong periodic discontinuities or corner-like oscillations), and check whether the rate is still $O(\\sqrt{\\varepsilon})$; if the rate worsens, the $W^{1,\\infty}$ regularity assumptions are load-bearing rather than technical.","supporting_citations":[{"cited_title":"Caillerie","cited_arxiv_id":null,"evidence_quote":"Supplies the weak homogenization limit for diffusion in thin domains, which Theorem 2.5 upgrades to strong two-scale convergence."},{"cited_title":"Caillerie","cited_arxiv_id":null,"evidence_quote":"Supplies the homogenized plate model and Theorem 3.1, the weak convergence result that the membrane and bending theorems refine."},{"cited_title":"Destuynder","cited_arxiv_id":null,"evidence_quote":"Provides the homogeneous-plate comparison strategy that the bending-case proof adapts to heterogeneous plates."},{"cited_title":"Jikov, S.M","cited_arxiv_id":null,"evidence_quote":"Provides the classical strong two-scale convergence arguments and the periodic divergence-potential lemma used in Lemma 2.9."},{"cited_title":"Dauge and I","cited_arxiv_id":null,"evidence_quote":"Gives the arbitrary-order asymptotics for homogeneous clamped plates that set the context and benchmark for strong plate convergence."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Korn inequalities used for coercivity of the elasticity problem and for the weighted Poincaré estimate in Lemma 3.4."},{"cited_title":"Murat and L","cited_arxiv_id":null,"evidence_quote":"Provides the H-convergence and div-curl background behind the stress-limit identification in Lemma 3.17."}],"review_version":1}