{"id":"cccf8fb1-53fe-47ba-b901-bc83f84c2cc1","arxiv_id":"2411.17879","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A stabilizer-free weak Galerkin method achieves optimal-order error estimates for linear elasticity on nonconvex polytopal meshes via bubble-function stability.","lead":"This paper introduces a weak Galerkin finite element method for linear elasticity that works on nonconvex polygonal and polyhedral meshes without needing stabilizer terms. The authors prove optimal convergence rates and demonstrate them on several meshes, simplifying implementation for complex geometries.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.5, the sole coercivity equivalence for the stabilizer-free method, rests on deferred bubble lemmas (4.3–4.4, (4.6)); on nonconvex polytopes a face-supporting hyperplane can pass through the interior, so the required uniform positivity of the bubble function and the inverse inequality…","rationale":"The paper's main theorems (6.3, 7.1) are conditional on the norm equivalence Lemma 4.5. The reader's weakest assumption correctly identifies this point. My stress-test concentrates on the least secure part of Lemma 4.5: (4.6) and Lemmas 4.3–4.4. The proof of Lemma 4.1 uses a 'domain inverse inequality' attributed to [52], but [52] does not establish such an inequality for a product of squared affine functions on nonconvex polytopes. On a nonconvex cell, the affine functions l_i vanish on the full hyperplane containing face e_i, and those hyperplanes may cross the cell interior, so the bubble can have interior zeros; the weighted norm-equivalence constant depends on how far one can get from all face-supporting hyperplanes, and no uniform lower bound is established. The paper itself defers the polynomial-extension details to preprints [36,37], and Remark 4.3 concedes that extra techniques are needed in difficult cases. Because the error analysis in (6.8), (6.9), and J6 in Section 7 all invoke ||v||_{1,h} ≤ C |||v|||, any nonuniformity in C destroys the stated convergence rates. The numerical tables show optimal orders on the tested nonconvex meshes, which is supportive but cannot certify the infinite family of nonconvex shape-regular cells covered by the theorem. Therefore the appropriate verdict is the same CONDITIONAL reached by the reader: accept once the deferred lemmas are supplied or verified. The proposed computational eigenvalue test would give direct evidence of whether the geometric degeneracy described here actually occurs.","tokens_in":25119,"tokens_out":22406,"duration_ms":207974,"concrete_test":"Verify Lemma 4.5 on a one-parameter family of star-shaped nonconvex elements that satisfy the paper's shape-regularity assumption but have a reentrant slit of width ε whose supporting lines pass arbitrarily close to the inscribed ball (e.g., a square with a V-notch of depth 1−ε). For k = 1, 2, compute the generalized eigenvalue λ_min of the pair (|||v|||^2, ||v||_{1,h}^2) over the local space V(k,T) restricted to one element, using sufficiently high-order quadrature for the bubble degrees. If λ_min → 0 as ε → 0, the uniform constant C1 in (4.9) does not exist and the central error estimates are unsupported; if λ_min stays bounded below on this family, the geometric concern is resolved for the tested configuration.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claims are the optimal-order error estimates in Theorems 6.3 and 7.1, and both rely on the norm equivalence C1||v||_{1,h} ≤ |||v||| ≤ C2||v||_{1,h} in Lemma 4.5, used in (6.8), (6.9), and in the estimate of J6. Lemma 4.5 is proved only by invoking three unproved technical ingredients: the domain inverse inequality (4.6) in Lemma 4.1, and the polynomial-extension/trace inequalities in Lemmas 4.3–4.4. Lemma 4.3 explicitly says 'For details on these extensions, see [36,37]', and Remark 4.3 states that for difficult nonconvex configurations additional techniques are needed but are also deferred to [36,37]. On a genuinely nonconvex polytope, the affine function l_i(x) associated with a face e_i vanishes on the whole supporting hyperplane of e_i, not just on e_i, and that hyperplane can pass through the interior of T even when e_i is a boundary face. Hence the bubble product Φ_B = product of l_i^2 can have interior zeros, and the inequality (4.6), (Φ_B p, p) ≥ C||p||^2, is not a standard consequence of the cited star-shape regularity. The finite-dimensional norm-equivalence argument only gives a constant depending on the distance from the chosen interior subdomain to all face-supporting hyperplanes; no bound on that distance is proved or cited for nonconvex cells. If the best constant degenerates for an admissible shape-regular family of nonconvex elements, then C1 in (4.9) is not uniform and the stated error estimates do not follow.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a stabilizer-free weak Galerkin (WG) finite element method for linear elasticity on polytopal meshes, claiming validity on nonconvex meshes through the use of bubble functions. The method uses the standard weak strain and weak divergence operators, and the bilinear form contains no explicit stabilization terms. The authors state optimal-order error estimates in the discrete H1 norm and in the L2 norm (Theorems 6.3 and 7.1), under regularity assumptions, and support these claims with numerical experiments on convex and nonconvex polygonal/polyhedral meshes. The main novelty asserted is the removal of convexity constraints for stabilizer-free WG methods, achieved via polynomial bubble functions and associated norm equivalences.","tokens_in":25538,"tokens_out":14145,"duration_ms":112017,"significance":"If the central technical lemmas were valid, the method would be a genuinely useful contribution: it is symmetric, positive definite, and simpler to implement than existing WG methods, and the numerical results show optimal convergence rates on nonconvex polytopal meshes, including pressure-robust behavior. The paper also addresses an important gap in the stabilizer-free WG literature, where previous work was limited to convex meshes. However, the key theoretical foundation—the bubble-function-based norm equivalence for nonconvex elements—is not proved in the manuscript and is deferred to the authors' own preprints [36,37]. The main theorems therefore rest on unverified assumptions. The numerical experiments provide supporting evidence but cannot replace a rigorous uniformity argument for the constants in the norm equivalence. The paper is a reasonable candidate for publication if the missing technical lemmas are supplied and the nonconvex case is fully handled.","major_comments":[{"comment":"The proof of the norm equivalence Lemma 4.5, which is the sole coercivity estimate for the stabilizer-free bilinear form, rests on the domain inverse inequality (4.6), asserted in Lemma 4.1 for nonconvex polytopes. The construction of the bubble function Φ_B as a product of squared face linear functions does not guarantee the existence of a subdomain RT with Φ_B ≥ ρ0 > 0 for nonconvex elements: the supporting hyperplane of a face can pass through the interior of T, creating interior zeros of Φ_B. The finite-dimensional norm equivalence would then give a constant depending on the distance from RT to all such zeros, and no uniform (mesh-independent) bound is proved or cited. Since (4.9) is used at (6.8)-(6.9) and in the estimate of J6 in Theorem 7.1, Theorems 6.3 and 7.1 are not established for nonconvex polytopal meshes without additional arguments.","section":"Lemma 4.1, Eq. (4.6); Lemma 4.5, Eq. (4.9)"},{"comment":"Lemma 4.3 and Lemma 4.4, which supply the face-bubble control of the interface jumps used in the lower bound of Lemma 4.5, are not proved in this paper: the proof of Lemma 4.3 states 'For details on these extensions, see [36,37]', and Remark 4.3 concedes that 'additional techniques' are needed for certain nonconvex configurations and again defers to [36,37]. The manuscript is therefore not self-contained at the exact point that distinguishes it from the existing convex-domain theory. The authors should either include complete proofs of Lemmas 4.3-4.5 (including (4.6)) or make the paper explicitly conditional on the preprints [36,37]; in the latter case the main theorems cannot be assessed as standing alone.","section":"Lemmas 4.3-4.4; Remark 4.3"}],"minor_comments":[{"comment":"In the proof of Theorem 7.1, the text 'Substituting this equation into (8.1)' refers to the wrong equation; it should be (7.3).","section":"Theorem 7.1 proof"},{"comment":"The definition of l_i(x) in Lemma 4.1 is written only for points on the edge/face; it should be defined globally on R^d as the signed distance to the supporting hyperplane, so that the bubble function Φ_B is a polynomial on T.","section":"Lemma 4.1 definition"},{"comment":"Reference [37] has an incomplete arXiv identifier ('arXiv:.') and cannot be located; this should be completed or the reference removed.","section":"References"},{"comment":"Equation (4.10) states the boundary term equals ∫_{e_i} |v_b-v_0|^2 φ_{e_i} ds, but for the matrix-valued test function φ=(v_b-v_0)n^T φ_{e_i}, the boundary term is actually 1/2∫(|v_b-v_0|^2 + ((v_b-v_0)·n)^2) φ_{e_i} ds; the displayed equality is therefore not literally correct, although the subsequent inequality can still be obtained.","section":"Equation (4.10)"},{"comment":"There are numerous typographical issues, e.g., 'ploytopal' for 'polytopal', 'impressible' for 'incompressible' (in Section 8 before Table 4), and inconsistent use of O(hr) in tables (some rows show '0.0' order on the first grid, which is expected but not explained).","section":"Section 8, general"}],"recommendation":"major_revision","confidential_remarks":"The paper's central claim of handling nonconvex polytopal meshes hinges entirely on technical lemmas (4.1-4.5) whose proofs are deferred to the same author's preprints [36,37], one of which [37] has an incomplete arXiv identifier. This makes the manuscript difficult to evaluate in isolation. If the missing lemmas are not included, the paper should at minimum be reframed as relying on established results from the preprints, with clear statements of what is assumed. The editor may also wish to verify the novelty with respect to [36], which appears to develop the same auto-stabilized bubble-function technique for other equations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about this paper. First, it does something genuinely useful: it extends stabilizer-free weak Galerkin methods to linear elasticity on nonconvex polytopal meshes, using bubble functions to handle both the strain and divergence terms. The method is symmetric, positive definite, and needs no stabilizers. Second, the central stability argument is not self-contained. Lemma 4.5, the norm equivalence that carries the whole error analysis, rests on a domain inverse inequality and two polynomial-extension lemmas whose proofs are deferred to the authors' preprints [36,37]. One of those preprints does not even have an arXiv number. So the paper's main claim—free from convexity constraints—is only as solid as those unpublished results.\n\nWhat the paper does well: the overall structure is standard WG machinery, and the error estimates in Theorems 6.3 and 7.1 are the right target. The numerical experiments on nonconvex polygonal and polyhedral meshes show clean optimal convergence rates and pressure robustness, which is encouraging evidence that the method actually works. The authors also honestly flag, in Remark 4.3, that some nonconvex configurations need extra techniques, and they point to the preprints.\n\nThe soft spots are real but not fatal if the deferred lemmas hold. The stress-test worry about Lemma 4.5 is legitimate: on a nonconvex polytope, a face-supporting hyperplane can pass through the element interior, so the squared bubble product may vanish inside the element, and the uniform positivity needed for (4.6) is not a consequence of standard shape regularity. The paper does not address this. The finite-dimensional argument gives a constant that depends on the distance from an interior subdomain to the supporting hyperplanes, and no uniform control is proved or cited for nonconvex cells. A referee will need to see those proofs.\n\nThere are also two numerical table issues. In Table 2(b), grid 6 shows an error of 0.135E-03 with order 0.0, which is inconsistent with the surrounding rows and should be explained. Table 5 is labeled P3/P4 WG, but the convergence rates are 1 and 2, so the labels are almost certainly wrong. These are minor, but they should be fixed.\n\nWho is this for? Researchers working on WG or polytopal finite element methods. The idea is coherent and the method is likely correct, but the paper does not fully stand alone. I would send it to peer review, with the requirement that the authors either include the deferred lemmas or make the preprints publicly available and convincing, and that they address the nonconvex bubble geometry explicitly. Not a desk reject, but not a quick accept either.","headline":"Plausible extension of stabilizer-free WG to elasticity on nonconvex polytopes, but the central stability proof is deferred to unpublished preprints and the geometry of the bubble functions is not fully settled; still deserves refereeing.","tokens_in":26022,"tokens_out":2002,"would_cite":false,"duration_ms":19723,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65N30","65N15","65N12","65N20"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that a bubble-function-enriched weak Galerkin method solves linear elasticity on nonconvex polygonal and polyhedral meshes at optimal convergence rates, with no stabilizer terms needed.","keywords":["weak Galerkin finite element method","linear elasticity","stabilizer-free","bubble functions","nonconvex polytopal meshes","error estimates","discrete weak strain tensor","discrete weak divergence"],"falsifier":"On a single nonconvex polygonal cell, compute the quotient $\\inf_{v_0} \\|\\epsilon_w(v)\\| / \\|\\epsilon(v_0)\\|$ over nonzero interior polynomials with $v_b = 0$; Lemma 4.1 predicts a positive lower bound uniform over a mesh sequence, so any sequence of cells whose reentrant angle approaches $360^\\circ$ and drives this quotient to zero would falsify the stability claim.","tokens_in":24889,"feed_emoji":"🧮","tokens_out":7003,"duration_ms":60664,"temperature":0.7,"pith_summary":"The paper aims to show that the weak Galerkin finite element method for linear elasticity can be stripped of its usual stabilization terms and still be stable and optimally accurate on meshes whose cells need not be convex. The authors construct discrete weak strain and discrete weak divergence operators using higher-degree polynomial spaces, then use bubble functions to prove the resulting stabilizer-free bilinear form is coercive in a discrete $H^{1}$ norm. For an exact solution in $H^{{k+1}}$, they prove errors of order h^k in the discrete energy norm, and order $h^{{k+α}}$ in $L^{2}$ when the dual problem has $H^{{1+α}}$ regularity. The payoff is a simpler, symmetric, positive-definite linear system that works on general polygonal and polyhedral meshes, including nonconvex ones. This matters because existing stabilizer-free WG treatments required convex cells, while standard WG methods carry an extra programming and theoretical burden for stabilizers.","feed_headline":"Bubble functions strip stabilizers from elasticity WG method","feed_subtitle":"A stabilizer-free weak Galerkin scheme reaches optimal convergence on nonconvex polygonal and polyhedral meshes.","key_machinery":"The central object is the element bubble function $\\Phi_B = \\ell_1^2 \\ell_2^2 \\cdots \\ell_N^2$, built from squared affine functions that vanish on each face of an N-faced polytopal element, together with its face-based sibling $\\phi_{e_i} = \\prod_{k \\neq i} \\ell_k^2$. These functions vanish on the element boundary, which removes the boundary terms in the integration-by-parts identities (2.2) and (2.4), and each is bounded below on a subdomain, which lets the domain inverse inequality convert weak-strain control into true-strain control. The face-based bubble function isolates one face at a time, so the boundary discrepancy $v_0 - v_b$ on every face is controlled; combined with reverse trace inequalities, this produces the two-sided norm equivalence of Lemma 4.5, the load-bearing stability estimate of the whole scheme.","core_discovery":"On the paper's own terms, the central discovery is that bubble functions, meaning polynomials that vanish on the boundary of each polytopal cell, make the stabilizer-free WG method for linear elasticity coercive without any convexity condition on the mesh. Choosing test functions of the form $\\Phi_B \\epsilon(v_0)$ in the discrete weak strain identity and $\\Phi_B (\\nabla \\cdot v_0)$ in the discrete weak divergence identity forces the boundary terms to vanish, so the true strain and divergence of the interior component $v_0$ are bounded by the weak strain and weak divergence of the pair $\\{v_0, v_b\\}$. An edge-based bubble function then transfers the boundary mismatch $v_0 - v_b$ into the same bound, yielding the norm equivalence $C_1\\|v\\|_{1,h} \\le |\\!|\\!|v|\\!|\\!| \\le C_2\\|v\\|_{1,h}$ of Lemma 4.5. From that norm equivalence the uniqueness proof, the error equation, and the optimal-order estimates follow: Theorem 6.3 gives $|\\!|\\!|u - u_h|\\!|\\!| \\le C h^k \\|u\\|_{k+1}$ in the discrete $H^{1}$ norm, and Theorem 7.1 gives $\\|u_0 - u_{0,h}\\| \\le C h^{k+\\alpha} \\|u\\|_{k+1}$ in the $L^{2}$ norm under the stated dual regularity.","pith_inferences":["The same bubble-function machinery is likely transferable to other stabilizer-free weak Galerkin formulations, such as Stokes, Maxwell, or fourth-order problems on nonconvex polytopal meshes, because the norm-equivalence mechanism is geometric rather than specific to elasticity.","Since the only mesh-dependent construction is the face-distance bubble function, the scheme may combine naturally with adaptive or hp-refinement, where reentrant cells are common, although the cost of the higher-degree weak-derivative spaces would need to be weighed in practice.","A formal analysis of how the constants in Theorems 6.3 and 7.1 depend on $\\lambda$ would settle whether the observed pressure independence holds uniformly as $\\lambda \\to \\infty$, which the present statements do not claim."],"forward_implications":["On any shape-regular polytopal mesh, including meshes with nonconvex polygonal and polyhedral cells, Algorithm 3.1 is well-posed: it has a unique solution for every $\\lambda \\geq 0$ and produces a symmetric positive-definite linear system.","For an exact solution in $[H^{k+1}(\\Omega)]^d$, the $P_k$ version converges at order $k+1$ in $L^2$ under dual regularity and at order $k$ in the discrete energy norm, matching the optimal rates for such elements.","Because no stabilizer parameter enters the formulation, the method avoids parameter tuning and retains the sparsity pattern of standard weak Galerkin discretizations.","The numerical experiments reported in the paper show the predicted orders of convergence on nonconvex polygonal and polyhedral meshes, with errors essentially independent of the Lamé parameter $\\lambda$ in the tested cases."],"supporting_citations":[{"why":"Supplies the shape-regular mesh assumptions, trace inequalities, and domain inverse inequality used in Lemmas 4.1 and 4.5.","marker":"[52]"},{"why":"Source of the discrete weak strain tensor and discrete weak divergence definitions on which the scheme is built.","marker":"[56]"},{"why":"Earlier weak Galerkin elasticity formulation whose weak strain and discrete weak divergence constructions are reused here.","marker":"[47]"},{"why":"Companion preprint that supplies the polynomial-extension and bubble-function lemmas whose proofs are deferred in this paper, including Lemmas 4.3 and 4.4.","marker":"[36]"},{"why":"Companion preprint continuing the deferred technical lemmas needed for nonconvex polytopal elements.","marker":"[37]"},{"why":"The prior stabilizer-free weak Galerkin method that this paper extends from convex to nonconvex polytopal meshes.","marker":"[59]"}],"fun_headline_variants":["Bubble functions let WG method drop stabilizers and convexity","Stabilizer-free WG method for elasticity works on nonconvex meshes","Bubble functions remove stabilizers in elasticity solver","Bubble functions enable stabilizer-free WG for elasticity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the bubble-function norm equivalence of Lemma 4.5 holds with constants independent of the element, uniformly for nonconvex polytopal cells, and the proofs of the underlying polynomial-extension lemmas are deferred to companion preprints.","fun_headline_variants_meta":{"raw":{"variants":["Bubble functions let WG method drop stabilizers and convexity","Stabilizer-free WG method for elasticity works on nonconvex meshes","Bubble functions remove stabilizers in elasticity solver","Bubble functions enable stabilizer-free WG for elasticity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000691,"raw_usage":{"total_tokens":3142,"prompt_tokens":969,"completion_tokens":2173,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":585,"completion_tokens_details":{"reasoning_tokens":2104}},"tokens_in":585,"tokens_out":2173,"duration_ms":14608,"temperature":1.0,"reasoning_tokens":2104,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:43:54.058291+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a single nonconvex polygonal cell, compute the quotient $\\inf_{v_0} \\|\\epsilon_w(v)\\| / \\|\\epsilon(v_0)\\|$ over nonzero interior polynomials with $v_b = 0$; Lemma 4.1 predicts a positive lower bound uniform over a mesh sequence, so any sequence of cells whose reentrant angle approaches $360^\\circ$ and drives this quotient to zero would falsify the stability claim.","supporting_citations":[{"cited_title":"W ang and S","cited_arxiv_id":null,"evidence_quote":"Source of the discrete weak strain tensor and discrete weak divergence definitions on which the scheme is built."},{"cited_title":"W ang, Auto-Stabilized Weak Galerkin Finite Element Methods for B iharmonic Equations on Polytopal Meshes without Convexity Assumptions , arXiv:","cited_arxiv_id":null,"evidence_quote":"Companion preprint continuing the deferred technical lemmas needed for nonconvex polytopal elements."},{"cited_title":"Xiu and S","cited_arxiv_id":null,"evidence_quote":"The prior stabilizer-free weak Galerkin method that this paper extends from convex to nonconvex polytopal meshes."}],"review_version":1}