{"id":"fbd75bed-bcd2-4aff-aa71-253c2b208e6e","arxiv_id":"2507.18896","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A bubble-function-based weak Galerkin scheme without stabilizers is proven to converge at optimal rates for the Brinkman equations on convex and non-convex polytopal meshes.","lead":"This paper introduces a stabilizer-free weak Galerkin finite element method for the Brinkman equations on meshes with non-convex polygonal or polyhedral elements, with optimal-order error estimates. It is relevant to computational fluid dynamics in porous media, where flow can switch between Stokes and Darcy behavior.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The non-convex norm-equivalence Lemma 4.3 is asserted rather than proved; its crucial step extends v_b off the face by orthogonal projection and then treats it as a polynomial, and the bubble-product bound used is imported from companion papers without proof or experiment.","rationale":"The reader's weakest_assumption correctly identifies the companion Lemma 4.1 bubble bound on the subregion as load-bearing, but the more acute concern is the explicit construction in Lemma 4.3: its extension of v_b by orthogonal projection to the element interior is not shown to be polynomial, and the face bubble phi_{e_i} is used with the discrete weak-gradient identity requiring the test function to lie in [P_r(T)]^{dxd}. This is an internal gap in the present paper, not merely a missing citation. However, I do not change the verdict: the gap is plausibly fixable through a different construction (e.g., using the bubble phi_{e_i} and suitable polynomial extension on each face separately, or by citing a complete proof from [16]/[30]), and the numerical results show the method works in practice at least for the tested lower degree r. The missing inf-sup citation placeholder is a real but minor presentational defect. A CONDITIONAL acceptance remains the right stance; my concern strengthens one of the reader's conditions but does not overturn the verdict. I would still like the paper to include the proof of Lemma 4.3 for non-convex elements, referencing the extension polynomial space, and to add a numerical test with r = 2N+k-1 on a genuinely non-convex mesh.","tokens_in":23473,"tokens_out":1975,"duration_ms":18553,"concrete_test":"Verify Lemma 4.3's key step independently: on a fixed non-convex quadrilateral (e.g., L-shaped polygon with N=6 sides), check whether the orthogonal-projection extension of a face polynomial v_b, multiplied by phi_{e_i}, lies in [P_{2N+k-1}(T)]^{dxd}; if not, compute ||grad_w v|| for a nontrivial interpolant with v0=0 and face data and compare with the face-lift norm used in the lemma. If the equivalence constant C1 in (4.6) is not bounded independent of the face geometry, Lemma 4.3 fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is the optimal a priori estimates Theorems 6.5 and 7.1 on general polytopal meshes including non-convex elements. Every estimate downstream rests on the norm equivalence C1||v||_{1,h} <= |||v||| (Lemma 4.3), which in turn rests on Lemma 4.1 (||grad v0||_T <= C||grad_w v||_T, cited from companion [16]) and on the derivation around Eq. (4.7): the proof extends v_b - v0 from the face e_i to the whole element by orthogonal projection onto the hyperplane, and then uses phi = (v_b - v0)^T n phi_{e_i} as a test function in the discrete weak-gradient identity (2.2). This requires phi (or at least v_b - v0 times phi_{e_i}) to be a polynomial in [P_r(T)]^{dxd}; but the projected extension generally does not produce a polynomial on a non-convex polytope, and even on a convex polytope the projected values outside e_i may fall outside T. The paper does not prove that the extension is in the weak-gradient polynomial space; it only says 'vb(X) = vb(Proj_ei(X)), with a suitable extension when Proj_ei(X) is not in ei'. Lemma 4.2 is also imported from [30] and no proof is given. Since the companion results [16,30] are not verified here and the numerical experiments never run the degree r = 2N+k-1 demanded by Remark 4.1 (they use r = k+1, k+2, k+3 on meshes that are mostly triangles, and Tables 3-9 show N is not large but the gap is never tested), the non-convex analysis has an unverified load-bearing step. Moreover, Lemma 4.5 cites '(see, e.g., [?, ?, ?, ?, ?])' with a literal placeholder, so the inf-sup condition is not actually established with citations. The result may be true, but this paper does not supply a complete proof on non-convex elements.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a stabilizer-free weak Galerkin (WG) method for the Brinkman equations on polytopal meshes, including non-convex polytopal elements. The velocity trial and test space is piecewise P_k in element interiors and on faces, the pressure space is piecewise P_{k-1}, and the discrete weak gradient is computed in a space P_r with r=2N+k-1 for non-convex elements and r=N+k-1 for convex elements. Existence and uniqueness of the discrete solution are established through a bubble-function-based norm equivalence and an inf-sup condition. Optimal a priori error estimates are claimed in the discrete H^1 norm (Theorem 6.5) and the L^2 norm (Theorem 7.1), with rates h^k and h^{k+1}, respectively. Numerical experiments on triangular, non-convex polygonal, tetrahedral, and non-convex polyhedral meshes report optimal convergence rates for r=k+1, k+2, and k+3.","tokens_in":23871,"tokens_out":10723,"duration_ms":104526,"significance":"The proposed scheme is potentially significant because it removes explicit stabilizers, has a simple implementation, and aims to cover non-convex polytopal meshes, a setting that is uncommon in the stabilizer-free WG literature. The analytical framework is standard once the norm equivalence Lemma 4.3 is granted, and the paper does provide a full error-equation analysis and a conventional duality argument. The manuscript would be strengthened by making the proof of the non-convex norm equivalence self-contained and by running experiments at the polynomial degrees required by the theory; without those additions the central non-convex claim remains conditional.","major_comments":[{"comment":"The proof of the norm equivalence is load-bearing for every subsequent result, but as written it is incomplete for non-convex elements. The displayed test function φ=(v_b-v0)^T n φ_{e_i} is scalar-valued, whereas the test functions in (2.2) are d×d matrix polynomials; the intended object is presumably the rank-one matrix (v_b-v0)n^T φ_{e_i}. More importantly, the extension v_b(X)=v_b(Proj_{e_i}(X)) is asserted with only 'a suitable extension' when Proj_{e_i}(X)∉e_i, and it is not demonstrated that the resulting matrix-valued test function lies in [P_r(T)]^{d×d} or that the bound ∥φ∥_T^2≤C h_T∫_{e_i}|v_b-v0|^2 ds holds; that bound is imported as Lemma 4.2 from [30] without proof. Since Lemma 4.3 supplies the lower bound in the norm equivalence, which feeds the inf-sup condition (4.12) and hence Theorems 6.5 and 7.1, this gap must be closed in the present manuscript.","section":"§4, Lemma 4.3 and Eq. (4.7)"},{"comment":"The numerical experiments never use the polynomial degrees required by the theory. Remark 4.1 and Lemma 4.1 require r=2N+k-1 for non-convex elements and r=N+k-1 for convex elements, but the computations use r=k+1, k+2, and k+3 throughout. For the non-convex polygonal meshes of Figures 2–4 with N≥4, this is far below the required degree (e.g., r=2N+k-1≥k+7 for k=1 and N=4), and even Table 1 on triangles uses r=k+1, below the convex requirement k+2. The tables therefore do not validate the central non-convex error estimates. Please either add experiments at the theoretically required degrees or prove the error estimates for the smaller degrees actually used.","section":"§8, Tables 1–9"},{"comment":"The main a priori estimates depend on Lemma 4.1, which is cited from the companion paper [16] and is not proved or tested here. This lemma asserts a uniform bound ∥∇v0∥_T≤C∥∇w v∥_T for the specific choice r=2N+k-1 (non-convex) or r=N+k-1 (convex). Because the assertion is parameter-free and non-obvious for non-convex polytopes, the manuscript should either include a proof of this lemma, or state the precise hypotheses under which it holds and explain why the companion paper establishes it in the form needed here.","section":"§4, Lemma 4.1 and Remark 4.1"}],"minor_comments":[{"comment":"The proof contains the placeholder citation '(see, e.g., [ ?, ?, ?, ?, ?])'; please replace it with concrete references for the surjectivity of the divergence operator from H^1_0(Ω) to L^2_0(Ω).","section":"§4, Lemma 4.5"},{"comment":"In the estimates of I2 and I3, the displayed bounds use ∥p∥_{k+1} while the theorem statement and the final result use ∥p∥_k; this appears to be a typo and should be corrected.","section":"§6, Theorem 6.5 proof"},{"comment":"The notation φ is reused for the test function and for the bubble function φ_{e_i}; please use distinct symbols for clarity.","section":"§4, Eq. (4.7)"},{"comment":"The range 1≤n≤2N+k-1 should be stated separately for convex elements, where the projection degree is N+k-1, or restricted to n=k, which is the value actually used in the paper.","section":"§6, Lemma 6.1"},{"comment":"Several figure captions call the meshes 'triangular' even though Figures 2–4, 6–9 display non-convex polygonal or polyhedral meshes; please correct the captions.","section":"§8, figure captions"}],"recommendation":"major_revision","confidential_remarks":"The central non-convex claim hinges on Lemma 4.3 and the imported bubble-function estimates, and the numerical section's degree mismatch is a real weakness. I would prioritize a self-contained proof of the non-convex norm equivalence (or a precise statement of the assumptions) and experiments at the theoretically required degrees before considering the paper for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The stress-test note goes too far on one detail but is right on the main structural issues. The specific worry about the affine extension of vb is not valid: orthogonal projection onto the face hyperplane is an affine map, so composing with the face polynomial gives a polynomial on the whole element, and the test function φ = (vb−v0)·n φ_ei lands in P_{2N+k−2}, which is inside the discrete weak-gradient space P_{2N+k−1}. That step in Lemma 4.3 is fine.\n\nWhat is actually new: this is the first stabilizer-free WG scheme for the Brinkman equations on general polytopal meshes, including non-convex elements. The κ^{-1} term is absorbed into the energy norm without introducing new difficulties, and the a priori error analysis is a clean, standard extension of the authors' Stokes framework. The numerical tables show optimal convergence rates on convex and non-convex meshes for k=1..4, which is credible evidence that the method works in practice.\n\nThe soft spots are real, and they are exactly where the reader put them. First, the norm equivalence (Lemma 4.3) rests entirely on Lemma 4.1 and Lemma 4.2, which are imported from the authors' own companion papers [16] and [30] without proofs or even precise statements. For a standalone paper, that is a heavy dependency, especially since the non-convex case is the advertised novelty. Second, the inf-sup proof in Lemma 4.5 contains a literal placeholder citation \"[?, ?, ?, ?, ?]\"; the argument itself is plausible, but the reference gap is sloppy and needs fixing. Third, and most importantly, the numerical experiments use r = k+1, k+2, k+3 while the theory on non-convex elements demands r = 2N+k−1. On the tested meshes that means the experiments never actually exercise the regime the analysis covers. Moreover, every experiment uses κ = 1, so the abstract's claim of robustness across Stokes- and Darcy-dominated flows is never demonstrated; there are no variable-permeability tests and no Darcy-limit or Stokes-limit checks.\n\nThis is not a desk-reject case. The paper is a legitimate extension of a working method, the analysis is coherent conditional on the companion lemmas, and the numerical results are useful. The right outcome is peer review with referees who will demand either proofs or full statements of the bubble-function lemmas, a proper inf-sup citation, and at least one experiment with spatially varying κ or with κ driving the problem into the Darcy regime. For researchers in weak Galerkin methods or polytopal finite elements, the paper is worth reading and worth citing as an extension, but the non-convex theory should be treated as inherited rather than established here.","headline":"A clean extension of stabilizer-free WG to Brinkman, but the non-convex analysis leans on unproved companion lemmas and the experiments never test the theory or the claimed Darcy/Stokes robustness.","tokens_in":24467,"tokens_out":4481,"would_cite":true,"duration_ms":49271,"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":"The paper proves that a stabilizer-free weak Galerkin finite element method achieves optimal-order convergence for the Brinkman equations on general polytopal meshes, including non-convex elements.","keywords":["weak Galerkin","Stabilizer-Free","weak gradient","weak divergence","bubble functions","non-convex","polytopal meshes","Brinkman equations"],"falsifier":"Take a non-convex polygonal cell with one re-entrant angle approaching $360^\\circ$, keep its diameter fixed, and compute $\\sup_v \\|\\nabla v_0\\|_T/\\|\\nabla_w v\\|_T$ over the discrete space for $k=1$; if this supremum grows without bound as the angle tends to $360^\\circ$, the central recovery lemma fails and the optimal error estimates cannot hold uniformly.","tokens_in":23243,"feed_emoji":"🌊","tokens_out":9917,"duration_ms":90096,"temperature":0.7,"pith_summary":"The paper's central claim is that the Brinkman equations—the model that blends Stokes flow in free regions with Darcy flow in porous media—can be solved accurately by a weak Galerkin finite element scheme that contains no explicit stabilization term. The scheme is shown to work on general polytopal meshes whose cells may be non-convex polygons or polyhedra, a case earlier stabilizer-free WG methods left out. If the proof is right, simulations of flow in fractured or vuggy porous media get a simpler discretization with the same optimal accuracy as stabilized schemes, since the stabilizer is replaced by a higher-degree polynomial weak gradient. The central result is that the discrete velocity and pressure errors are $O(h^k)$ in the discrete $H^1$ norm, and the velocity error is $O(h^{k+1})$ in $L^2$, where $k$ is the polynomial degree of the velocity approximation.","feed_headline":"No-stabilizer weak Galerkin scheme hits optimal convergence","feed_subtitle":"Proves optimal h^k error bounds for Brinkman flow on general polygonal and polyhedral meshes, convex or not.","key_machinery":"The load-bearing construction is the element bubble function. For a polytopal cell $T$ with $N$ edges or faces $e_1,\\dots,e_N$, each with a linear function $l_i$ vanishing on $e_i$, the non-convex case uses $\\Phi_B = l_1^2 l_2^2 \\cdots l_N^2 \\in P_{2N}(T)$, scaled to be one at the barycenter and uniformly positive on a subregion $\\hat T$; the discrete weak gradient is then taken in $[P_r(T)]^{d\\times d}$ with $r=2N+k-1$ (with the linear-product variant $r=N+k-1$ in the convex case). The bubble product enables the recovery bound $\\|\\nabla v_0\\|_T \\le C \\|\\nabla_w v\\|_T$ on non-convex elements, which yields the norm equivalence between $\\|\\cdot\\|_{1,h}$ and $\\||\\cdot\\||$, and that equivalence feeds every subsequent step: the inf-sup condition, the error equation, and the $O(h^k)$ and $O(h^{k+1})$ bounds.","core_discovery":"The paper's central claim is that Algorithm 3.1, which computes the weak gradient in $[P_r(T)]^{d\\times d}$ with $r=2N+k-1$ on non-convex cells (or $r=N+k-1$ on convex cells) and uses no stabilizer, produces a unique discrete solution of the Brinkman problem and satisfies the error estimate $\\|u-u_h\\|_{\\text{energy}} + \\|p-p_h\\| \\le C h^k (\\|u\\|_{k+1} + \\|p\\|_k)$, together with the $L^2$ velocity bound $\\|u-u_0\\| \\le C h^{k+1}(\\|u\\|_{k+1}+\\|p\\|_k)$. The proof constructs bubble functions on each polytopal element to establish a norm equivalence between the discrete $H^1$-type norm and the weak-gradient norm, derives the inf-sup condition for the mixed formulation, obtains the error equation with consistency terms, and bounds those terms by projection approximation estimates.","pith_inferences":["The paper's theory requires $r=2N+k-1$ on non-convex cells, but the experiments use only $r=k+1$, $k+2$, or $k+3$ and still show optimal rates; a plausible reading is that the theoretical degree is a very safe overestimate and the true requirement is much weaker.","The unproved recovery lemma from the companion paper is the single step that ties the whole proof together; proving it directly, or testing it on the actual meshes used in the experiments, would either close the gap or reveal a shape-dependent constant.","Because only the weak-gradient/weak-divergence structure and the inf-sup condition are used, the construction should carry over to variable permeability tensors and to other coupled Stokes–Darcy systems, although the paper itself assumes constant $\\kappa$.","A concrete prediction of the analysis is that on a fixed non-convex mesh the convergence order should not change once $r$ is large enough; the error constant may decrease, but the rate is governed by $k$."],"forward_implications":["The scheme eliminates the explicit stabilizer term, so the global stiffness matrix keeps the size and sparsity of a standard WG discretization while implementation simplifies.","Meshes with non-convex polygons or polyhedra can be used as-is, without convexity preprocessing, so arbitrary polytopal partitions from domain decomposition or mesh generators are admissible.","For velocity polynomials of degree $k$ and pressure polynomials of degree $k-1$, the method delivers $O(h^k)$ errors in the discrete energy norm and in pressure, and $O(h^{k+1})$ for velocity in $L^2$, whenever the exact solution has the stated Sobolev regularity.","The same mixed framework covers both Stokes-dominated and Darcy-dominated regimes, so a single code can handle a Brinkman problem as the permeability $\\kappa$ varies, at least in the constant-$\\kappa$ case analyzed.","The duality argument shows that the $L^2$ velocity error converges one order faster under the standard elliptic-regularity assumption."],"supporting_citations":[{"why":"Supplies the unproved recovery bound $\\|\\nabla v_0\\|_T\\le C\\|\\nabla_w v\\|_T$ for non-convex polytopal elements, the load-bearing input of the norm equivalence.","marker":"[16]"},{"why":"Defines the discrete weak gradient and weak divergence and supplies the identities and inf-sup scaffolding reused throughout the paper.","marker":"[30]"},{"why":"Provides the shape-regularity condition and the trace inequalities used in every error estimate.","marker":"[27]"},{"why":"Documents the failure of standard Stokes-stable elements in Darcy-dominated regimes, the baseline the new method must improve on.","marker":"[3]"},{"why":"The earlier weak Galerkin discretization of the Brinkman equations that this stabilizer-free scheme simplifies and extends.","marker":"[37]"},{"why":"Supplies the $L^2$ projection approximation estimates used to bound consistency errors.","marker":"[21]"}],"fun_headline_variants":["Stabilizer-free WG method nails Brinkman on any polytopal mesh","No-stabilizer weak Galerkin: optimal Brinkman error bounds","Robust weak Galerkin for Stokes-Darcy without stabilizers","Optimal convergence for Brinkman, even on non-convex meshes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole proof rests on a lemma borrowed from a companion paper that is not proved or tested here: on every non-convex polytopal element the bubble product $l_1^2\\cdots l_N^2$ is uniformly positive on some subregion with a constant independent of the element shape, so the recovery bound $\\|\\nabla v_0\\|_T\\le C\\|\\nabla_w v\\|_T$ holds with a fixed $C$.","fun_headline_variants_meta":{"raw":{"variants":["Stabilizer-free WG method nails Brinkman on any polytopal mesh","No-stabilizer weak Galerkin: optimal Brinkman error bounds","Robust weak Galerkin for Stokes-Darcy without stabilizers","Optimal convergence for Brinkman, even on non-convex meshes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000224,"raw_usage":{"total_tokens":1489,"prompt_tokens":1002,"completion_tokens":487,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":618,"completion_tokens_details":{"reasoning_tokens":407}},"tokens_in":618,"tokens_out":487,"duration_ms":4632,"temperature":1.0,"reasoning_tokens":407,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:06:48.909345+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a non-convex polygonal cell with one re-entrant angle approaching $360^\\circ$, keep its diameter fixed, and compute $\\sup_v \\|\\nabla v_0\\|_T/\\|\\nabla_w v\\|_T$ over the discrete space for $k=1$; if this supremum grows without bound as the angle tends to $360^\\circ$, the central recovery lemma fails and the optimal error estimates cannot hold uniformly.","supporting_citations":[{"cited_title":"W ang and S","cited_arxiv_id":null,"evidence_quote":"Defines the discrete weak gradient and weak divergence and supplies the identities and inf-sup scaffolding reused throughout the paper."},{"cited_title":"W ang, and X","cited_arxiv_id":null,"evidence_quote":"Provides the shape-regularity condition and the trace inequalities used in every error estimate."},{"cited_title":"Mardal, X","cited_arxiv_id":null,"evidence_quote":"Documents the failure of standard Stokes-stable elements in Darcy-dominated regimes, the baseline the new method must improve on."},{"cited_title":"A Stable Numerical Algorithm for the Brinkman Equations by Weak Galerkin Finite Element Methods","cited_arxiv_id":"1312.2256","evidence_quote":"The earlier weak Galerkin discretization of the Brinkman equations that this stabilizer-free scheme simplifies and extends."},{"cited_title":"W ang and J","cited_arxiv_id":null,"evidence_quote":"Supplies the $L^2$ projection approximation estimates used to bound consistency errors."}],"review_version":1}