{"id":"fa65833e-022b-4898-b449-e80b2323b7ec","arxiv_id":"1908.04647","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Numerical experiments confirm that a spectral mixed DG method on anisotropic geometric meshes is inf-sup stable with mild degree dependence and converges exponentially for 3D elasticity with edge and corner singularities, including near the incompressible limit.","lead":"This paper tests a high-order discontinuous Galerkin method for 3D elasticity on meshes that zoom in on edges and corners, using computer experiments to check stability and exponential accuracy. It shows the method stays stable as materials become nearly incompressible and converges exponentially on test problems with edge and corner singularities.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; the inf-sup foundation (13) is a theorem from prior work and the numerical study supports the claimed stability and exponential convergence.","rationale":"The reader's weakest_assumption correctly identifies the discrete inf-sup condition (13) as the load-bearing premise of Theorem 3.2 and the exponential convergence theory. However, (13) is not merely an assumption: it was proved in [19] for the same type of geometric edge meshes with ρ=3/2, so the foundation is secure. The numerical experiments in Section 4.4.1 directly probe this condition and show robustness with respect to refinement, consistent with the theorem. The genuinely unverified ingredient is the analytic regularity of the manufactured solutions in Section 5; the paper explicitly assumes membership in the countably normed spaces A_{-1-β}×A_{-β}. The chosen singular functions (r^{1/2} edge, r^{1/6} corner) are canonical examples for which such regularity is expected with suitable weights, and the observed exponential convergence corroborates the assumption. I also checked the algebraic derivation of the inf-sup constants: Proposition 4.3 correctly identifies γ_B as the smallest positive singular value of the scaled matrix, and Proposition 4.4 correctly extends this to the saddle-point form ah despite its nonsymmetry. The augmented formulation (16) is consistent with the original problem; the apparent singularity concern in Proposition 4.2 resolves once the full integral in the second equation is distinguished from the mean integral in the third. Overall, the paper delivers a solid computational study that validates and sharpens the known theory, and I find no internal inconsistency or unsupported leap that would justify a downgrade.","tokens_in":17779,"tokens_out":40914,"duration_ms":363314,"concrete_test":"As a worthwhile verification step, recompute the γ_B values from Section 4.4.1 for k=12, 14, 16 on the Fichera corner mesh and check whether the decay remains far slower than the theoretical k^{-3/2} bound; if instead the values collapse toward the bound, the 'considerably more optimistic' conclusion would need qualification, though the central stability and convergence claims would remain intact.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After reviewing the SVD derivations (Propositions 4.3 and 4.4), the equivalence of the augmented system (16) with the original formulation, and the role of the discrete inf-sup condition (13), I find no load-bearing flaw in the paper's central argument. The assumption (13) is not an unverified conjecture: reference [19] proved it for geometric edge meshes with ρ=3/2, and the paper's numerical experiments in Section 4.4.1 are consistent with that result, while suggesting a milder k-dependence. The exponential convergence statement in Section 5 is explicitly conditional on the analytic regularity assumption (u,p)∈A_{-1-β}×A_{-β}; the manufactured solutions are canonical edge/corner singularities for which this assumption is expected to hold, and the observed nearly linear log-error graphs support it. The main weakness is the absence of shipped code or data, which is a reproducibility limitation rather than a correctness issue. I therefore do not see a concern that would change the reader's acceptance verdict.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a computational study of a spectral mixed discontinuous Galerkin method for the Lamé system of linear elasticity in mixed form on anisotropic geometric edge meshes in three-dimensional polyhedral domains. The authors recall the discrete inf-sup framework from their earlier work [24] and the assumption (13) on the form B_h, which was proved with rho = 3/2 in [19]. They then derive SVD-based characterizations (Propositions 4.3 and 4.4) for the discrete inf-sup constants of B_h and of the full mixed form a_h in the incompressible limit nu = 1/2, together with an augmented formulation (16) that avoids the zero-mean constraint on the pressure. Numerical experiments measure these constants on edge, corner, corner-edge, and Fichera-type geometric refinements, and show robustness with respect to mesh refinement and a k-dependence much milder than the theoretical bounds k^{-3/2} and k^{-3}. A second set of experiments with manufactured edge, corner, and corner-edge singularities and with smooth incompressible examples exhibits near-exponential convergence in the DG norm and robustness as nu approaches 1/2.","tokens_in":17943,"tokens_out":26788,"duration_ms":267492,"significance":"The paper's main contribution is a clean SVD-based procedure for evaluating discrete inf-sup constants of mixed DG forms, together with the first systematic numerical evidence that the inf-sup constants of the spectral mixed DG method are robust on anisotropic geometric edge meshes with respect to refinement and Poisson ratio, and that exponential convergence is achieved for solutions with edge and corner singularities. The derivations in Section 4 are mathematically sound given the inf-sup hypothesis (13), and the experimental design tests the theory rather than fitting constants. The absence of shipped code or data is a reproducibility limitation, but the algorithmic description is sufficiently detailed to be reimplemented in deal.II.","major_comments":[],"minor_comments":[{"comment":"The text states that the singular examples use nu in {1/8, 1/2, 3/8}, but the figures show nu = 1/8, 1/4, 3/8 and the formula p = -div u/(1-2 nu) is undefined at nu = 1/2; the set should be corrected.","section":"Section 5.1"},{"comment":"The notation 'exp(-b 5 sqrt(N))' in (28) is ambiguous; it should be typeset as exp(-b N^{1/5}) or exp(-b times the fifth root of N), and the relationship between this bound and the N^{1/4} axes used for the edge and corner cases in Figure 7 should be stated explicitly.","section":"Equation (28) and Figures 7-8"},{"comment":"The manufactured solutions do not satisfy the homogeneous Dirichlet condition u = 0 on the whole boundary; the paper explains that only u dot n = 0 is enforced and that the remaining boundary data are imposed through flux terms, but the precise modification of the right-hand side in (10) is not given, which would improve reproducibility.","section":"Section 5.1"},{"comment":"The reference solution is computed only at k = l = 5; since the reported errors approach the level of this reference, a comparison against a finer reference or an estimate of the reference discretization error would make the claimed exponential convergence more conclusive.","section":"Section 5.2, Example 5.2"},{"comment":"Equation (19) is not readable as typeset; the relation between M and N should be written clearly, for example M = 3((k+1)/k)^3 N.","section":"Section 4.2, Eq. (19)"},{"comment":"No code or data files are provided; for a purely computational paper, making the deal.II scripts and mesh-generation routines available would substantially aid reproducibility.","section":"General"}],"recommendation":"minor_revision","confidential_remarks":"The paper is essentially a numerical companion to the authors' earlier M3AS paper [24], with the theoretical framework taken from [24] and [19]; the genuinely new content is the SVD-based computation of discrete inf-sup constants and the numerical verification. If the journal expects a substantial amount of new analysis, this is a validation study, but it is a careful and useful one. The minor issues listed are local and should be corrected in a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a numerical companion paper to the authors' earlier theory paper [24], and it does what it says. The new contributions are the SVD-based procedure for computing discrete inf-sup constants (Propositions 4.3 and 4.4) and the quantitative experiments on edge, corner, corner-edge, and Fichera meshes. The derivations are clean and correct — the reduction of γ_B to the smallest positive singular value of the scaled matrix B is standard but nicely adapted to the DG setting, and the reformulation with the Lagrange multiplier for the pressure mean (Section 4.1) is rigorously shown equivalent to the original system. The observed stabilization of γ_B and γ_a under refinement, and the much milder k-dependence than the theoretical k^{-3/2} and k^{-3}, are credible and clearly labeled as numerical evidence, not proofs.\n\nThe paper is honest about its scope. The stability theory comes from [24] and [19]; the abstract says \"computational study\", and the conclusions repeat that. The exponential convergence plots for the edge/corner/corner-edge manufactured solutions are consistent with the analytic regularity assumption quoted from [6].\n\nSoft spots, in proportion:\n\n1. No code or data is shipped, and the experiments are not described at the level of reproducibility (exact meshes, tolerances, and the reference solution for Example 5.2 are described, but the SVD implementation details and the deal.II configurations are not). For a purely numerical paper, that is a real limitation, though not a correctness flaw.\n\n2. The manufactured singular solutions are not the true corner/edge singularities of the homogeneous Dirichlet problem. They have u·n=0 on ∂Ω, not u=0, so they test approximation with inhomogeneous data, which is fine but not exactly the setting of the theory. Also, the singular tests are only for ν = 1/8, 1/4, 3/8 (the text in Section 5.1 lists ν = 1/2, but the figures show 1/4; presumably a typo). The incompressible robustness for nonsmooth solutions is therefore inferred from the inf-sup computations plus two smooth/divergence-free experiments, not directly demonstrated.\n\n3. The choice γ=10 and σ=1/2 is not varied. Since the coercivity and inf-sup estimates depend on γ, some sensitivity study would strengthen the claim.\n\nNone of these undermines the central conclusion: the method is stable and exponentially convergent on these geometric meshes, and the numerical evidence supports the theory. The paper should be refereed. I would recommend acceptance with minor revisions, asking for code/data or a reproducibility appendix.\n\nWho it's for: numerical analysts working on DG/hp methods for elasticity and Stokes; anyone needing a practical recipe for computing discrete inf-sup constants. Cite it if you use the SVD tool.","headline":"A sound, honest computational validation of the authors' own theory; the SVD inf-sup tool is genuinely useful, and the soft spots are reproducibility and a narrow set of manufactured tests.","tokens_in":18499,"tokens_out":5075,"would_cite":true,"duration_ms":48710,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65N30","65N35","65N12"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper uses numerical experiments to claim that a spectral mixed discontinuous Galerkin method for 3D linear elasticity on anisotropic geometric meshes is stable up to the incompressible limit and converges exponentially in a natural…","keywords":["mixed discontinuous Galerkin method","linear elasticity","polyhedral domains","geometric edge meshes","inf-sup stability","exponential convergence","incompressible limit","spectral methods"],"falsifier":"Recompute the discrete inf-sup constants for a geometric mesh family with a different refinement ratio or with polynomial degrees beyond $k=10$, or test the corner-edge example on a Fichera domain with traction boundary conditions; if the constants decay to zero with degree or the exponential error decay flattens to algebraic as $\\nu \\to 1/2$, the central claim is refuted.","tokens_in":17550,"feed_emoji":"📐","tokens_out":13977,"duration_ms":118448,"temperature":0.7,"pith_summary":"This paper makes the case, through extensive numerical experiments, that a spectral mixed discontinuous Galerkin method can solve the three-dimensional Lamé system of linear elasticity in polyhedral domains — shapes with edges and corners where solutions develop singularities — with no loss of stability. It focuses on anisotropic geometric meshes that refine toward edges, corners, and corner-edges, and on the behavior of discrete inf-sup constants as the Poisson ratio $\\nu$ approaches $1/2$, the incompressible Stokes limit. The authors claim two things: the discrete stability constants remain robust under mesh refinement and in the incompressible limit, with a polynomial-degree dependence far milder than the proved $k^{-3/2}$ bound, and the method converges exponentially in a natural DG norm for solutions with edge, corner, and corner-edge singularities under analytic regularity assumptions. If these claims hold, the method is a practical high-order solver for 3D elasticity and Stokes problems, including nearly incompressible materials.","feed_headline":"DG solver for 3D elasticity converges exponentially on refined meshes","feed_subtitle":"Spectral mixed DG stays stable as materials become incompressible, with exponential accuracy on edge-refined meshes.","key_machinery":"The load-bearing mechanism is the discrete inf-sup condition (13): the assumption that the inf-sup constant of $B_h$ over the discrete spaces $V_h \\times Q_h$ is bounded below by $\\kappa k^{-\\rho}$ with $\\kappa$ and $\\rho$ independent of the polynomial degree $k$, the number of refinement layers $\\ell$, and the aspect ratio of anisotropic elements. This condition upgrades coercivity — which degenerates as $(1-2\\nu) \\to 0$ — into full inf-sup stability of the mixed form $a_h$ even at $\\nu=1/2$, and it underlies the exponential convergence estimate (28). Computationally, the paper evaluates these constants as the smallest positive singular values of scaled system matrices, and builds the meshes from canonical geometric patches toward edges, corners, and corner-edges.","core_discovery":"The central discovery is computational but with theoretical teeth: on anisotropic geometric edge meshes, the discrete inf-sup constant $\\gamma_B$ of the bilinear form $B_h$ stabilizes after a few geometric refinement layers and shows a dependence on the polynomial degree $k$ that is considerably more optimistic than the theoretical $k^{-3/2}$ bound; likewise, the inf-sup constant $\\gamma_a$ of the mixed form $a_h$ remains bounded away from zero uniformly as $\\nu \\to 1/2$, confirming that the DG scheme does not degenerate in the incompressible limit. Moreover, for manufactured solutions with edge, corner, and corner-edge singularities, and with the polynomial degree grown proportionally to the refinement level, the DG error in the norm (12) decays as $\\exp(-b N^{1/5})$ or $\\exp(-b N^{1/4})$, exhibiting the exponential convergence predicted by the theory. These experiments together support the conclusion that the spectral mixed DG method (8) is stable and exponentially convergent on geometric edge meshes for 3D linear elasticity.","pith_inferences":["If the observed mild $k$-dependence of the inf-sup constants is proved rather than merely observed, the exponential convergence estimates for hp-DG methods on geometric meshes could be sharpened, yielding better cost predictions for high-accuracy 3D elasticity.","The same SVD-based constant computation could serve as a numerical pre-screen for stability of hp-adaptive DG variants (variable or anisotropic polynomial degrees), which the paper mentions as a future extension but does not analyze.","Since the method tolerates $\\nu \\to 1/2$ without deterioration, coupling it with the iterative Newton-DG approach mentioned in the conclusions could yield robust solvers for nonlinear hyperelasticity near the incompressible limit.","The canonical-patch experiments use manufactured solutions; a natural next test, not performed here, would be on a Fichera domain with traction boundary conditions, where the combination of corner and edge singularities is genuinely three-dimensional and the analytic regularity assumptions are harder to satisfy."],"forward_implications":["The spectral mixed DG method (8) can be used as a reliable solver for 3D linear elasticity in polyhedral domains, including near-incompressible and exactly incompressible materials, without special locking treatment.","Exponential convergence in the DG norm is achievable for solutions with edge, corner, and corner-edge singularities when geometric refinement is paired with a polynomial degree that grows with the number of layers.","The measured $k$-dependence of the inf-sup constants indicates that the theoretical worst-case decay $k^{-3/2}$ (and $k^{-3}$ for the mixed form) is not sharp, so sharper stability bounds should be attainable.","The augmented Lagrange-multiplier formulation (16) offers a practical way to enforce the zero-mean pressure constraint and to compute inf-sup constants using standard local basis functions."],"supporting_citations":[{"why":"Supplies the theoretical inf-sup stability and exponential convergence results for 3D mixed DG that the present study validates computationally.","marker":"[24]"},{"why":"Proves the discrete inf-sup condition (13) with $\\rho = 3/2$ on geometric edge meshes, the rigorous basis for the stability hypothesis.","marker":"[19]"},{"why":"Establishes the analytic regularity of solutions in weighted Sobolev spaces that underlies the assumed exponential convergence.","marker":"[6]"},{"why":"Provides the singular value decomposition technique used to evaluate the discrete inf-sup constants as smallest positive singular values.","marker":"[5]"},{"why":"Develops the hp-DG framework on geometric meshes in polyhedra, including the mesh construction and exponential convergence techniques adapted here.","marker":"[20–22]"},{"why":"Introduces the mixed DG formulation for nearly incompressible linear elasticity from which the spectral scheme is derived.","marker":"[12]"},{"why":"The finite element library used to perform all numerical experiments reported in the paper.","marker":"[3, 4]"}],"fun_headline_variants":["Stable DG for near-incompressible 3D elasticity, exponential accuracy","Exponential DG convergence for 3D elasticity on refined meshes","Spectral mixed DG stable in incompressible limit, exponential convergence","Robust spectral DG method: exponential accuracy in 3D elasticity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire argument rests on an assumed stability bound, the discrete inf-sup condition (13), which says the mixed system cannot become singular as the mesh is refined or the polynomial degree grows; the paper verifies this bound numerically but does not prove it beyond the exponent $3/2$ already established in the literature.","fun_headline_variants_meta":{"raw":{"variants":["Stable DG for near-incompressible 3D elasticity, exponential accuracy","Exponential DG convergence for 3D elasticity on refined meshes","Spectral mixed DG stable in incompressible limit, exponential convergence","Robust spectral DG method: exponential accuracy in 3D elasticity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00084,"raw_usage":{"total_tokens":3634,"prompt_tokens":892,"completion_tokens":2742,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":508,"completion_tokens_details":{"reasoning_tokens":2665}},"tokens_in":508,"tokens_out":2742,"duration_ms":18562,"temperature":1.0,"reasoning_tokens":2665,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:36:26.507118+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the discrete inf-sup constants for a geometric mesh family with a different refinement ratio or with polynomial degrees beyond $k=10$, or test the corner-edge example on a Fichera domain with traction boundary conditions; if the constants decay to zero with degree or the exponential error decay flattens to algebraic as $\\nu \\to 1/2$, the central claim is refuted.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the theoretical inf-sup stability and exponential convergence results for 3D mixed DG that the present study validates computationally."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the discrete inf-sup condition (13) with $\\rho = 3/2$ on geometric edge meshes, the rigorous basis for the stability hypothesis."},{"cited_title":"Dauge, M","cited_arxiv_id":null,"evidence_quote":"Establishes the analytic regularity of solutions in weighted Sobolev spaces that underlies the assumed exponential convergence."},{"cited_title":"Brezzi and M","cited_arxiv_id":null,"evidence_quote":"Provides the singular value decomposition technique used to evaluate the discrete inf-sup constants as smallest positive singular values."},{"cited_title":"Houston, D","cited_arxiv_id":null,"evidence_quote":"Introduces the mixed DG formulation for nearly incompressible linear elasticity from which the spectral scheme is derived."}],"review_version":1}