{"id":"6bc23ed9-8ab2-4996-9bff-52669d398a0e","arxiv_id":"2412.13002","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A high-order entropy stable discontinuous Galerkin method is constructed and verified on cut meshes using skew-hybridized summation-by-parts operators and Carathéodory-pruned quadrature.","lead":"This paper develops a high-order numerical method for fluid-flow equations on computational grids cut by embedded shapes, and shows the method stays entropy stable without artificial tuning. It is the first such entropy stable scheme on cut meshes, combining an existing discontinuous Galerkin framework with a quadrature-pruning step.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper does not prove that Carathéodory pruning always yields a positive/non-negative exact quadrature rule with enough strictly positive weights on arbitrary cut elements, and the appendix's α formula appears to use max where min is required.","rationale":"The most load-bearing condition for the central claim is not the state redistribution observation (explicitly unproven) nor the novelty assertion, but the quadrature construction that makes the skew-hybridized SBP entropy stability proof operative on cut elements. The paper's own appendix concedes that the required positivity condition is only 'typically easily met,' and the described pruning formula contains an apparent typo (max vs. min for the positive-entry bound). The numerical experiments demonstrate the pipeline works for the specific geometries tested, which is credit where due, but they do not establish the general claim. The manufactured-solution convergence study is also compromised by the missing source term, but that is a verification defect rather than a defect in the method's construction; it would not by itself falsify the theoretical accuracy argument. The reader's conditional verdict already captures this, so no verdict change is needed.","tokens_in":21684,"tokens_out":10741,"duration_ms":98703,"concrete_test":"Implement the pruning algorithm exactly as described in Appendix A, including the printed α+ = max_{Δw_i>0} w_i/Δw_i, and run it on a suite of degenerate cut elements: sliver triangles, cuts with very small measure, near-tangent boundary intersections, and high-curvature boundaries, for polynomial degrees N = 2 through 6. Verify termination with at most dim P_{2N-1} + 1 points, non-negative weights, exactness for degree 2N-1 volume and 2N surface rules, and positive-definiteness of M = V_q^T W V_q. Then correct α+ to min_{Δw_i>0} w_i/Δw_i and repeat; if the printed version fails on any case, or the corrected version still produces an indefinite mass matrix for a valid cut element, the general robustness claim is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on the skew-hybridized SBP formulation of Section 2.1, whose entropy stability and high-order accuracy require, per Section 2.1.4, volume quadrature exact for degree 2N-1, surface quadrature exact for degree 2N, non-negative weights, and a positive-definite mass matrix M = V_q^T W V_q. Section 2.2 and the Appendix construct these rules by subtriangulation and Carathéodory pruning, but provide no proof that pruning terminates for arbitrary degenerate cut elements with the required properties; the Appendix only states that enough strictly positive weights remain is 'typically easily met.' Additionally, the printed pruning step appears incorrect: α+ is defined as max_{Δw_i>0} w_i/Δw_i, whereas maintaining non-negativity w_i - α Δw_i ≥ 0 for positive Δw_i requires α ≤ min_{Δw_i>0} w_i/Δw_i. With the printed formula, α = min{α-, α+} with α- ≤ 0 ≤ α+ selects the negative bound and may not reduce the point count, while choosing the positive bound can drive weights negative. If the pruning algorithm is not robust for every cut element, the entropy stability proof collapses for that element, undermining the claim of an entropy stable method on cut meshes generally. The §3.3 convergence study also omits the source term required by the manufactured solution, further weakening the numerical verification of high-order accuracy.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a high-order discontinuous Galerkin method for hyperbolic conservation laws on Cartesian cut meshes and claims it is entropy stable. The method combines skew-hybridized summation-by-parts operators from Chan [1] with flux-differencing and entropy conservative/stable two-point fluxes. Quadrature rules on arbitrarily shaped cut elements are constructed by subtriangulation followed by Carathéodory pruning. The authors verify the method numerically with the shallow water and compressible Euler equations, including entropy residual tests, an h-convergence study with manufactured solutions, and several benchmark problems with state redistribution.","tokens_in":21957,"tokens_out":9646,"duration_ms":99649,"significance":"If the central claim is fully supported, this is a meaningful advance: it would be the first entropy stable high-order DG scheme on cut meshes, combining nonlinear stability at arbitrary order with the geometric flexibility of cut-cell meshes. The paper has several concrete strengths: it builds on a parameter-free skew-hybridized SBP framework with explicit assumptions, it provides open-source Julia packages and a reproducibility repository, and the numerical experiments include round-off-level entropy conservation for the entropy conservative scheme, non-positive entropy residuals for the entropy stable scheme, and convergence checks for N=2,3,4. However, two load-bearing points need attention before the claim can be accepted: the quadrature-pruning argument does not prove the conditions needed for positive-definiteness of the mass matrix on arbitrary cut elements, and the manufactured-solution convergence study as written omits the source term required by the proposed solution.","major_comments":[{"comment":"The entropy stability proof requires a positive-definite mass matrix M in Eq. (29), which is used to define Pq in Eq. (32). Section 2.1.4 states that this holds if quadrature weights are non-negative and enough weights are strictly positive, and the Appendix concludes that this condition is 'typically easily met'. This is not a proof, and it is load-bearing: the final Carathéodory-pruned rule is only guaranteed by Theorem 1.4 to have at most dim(P_N)+1 non-negative weights, not that the corresponding points are unisolvent for P_N. The iterative update in Eqs. (68)-(73) preserves exactness and non-negativity under one step, but the paper does not prove that repeated pruning terminates with at least dim(P_N) strictly positive weights whose points form a unisolvent set, nor does it state geometric assumptions under which this is guaranteed. Since the central claim is stated for cut meshes generally, the proof should either establish this property under explicit, verifiable assumptions or the claim should be qualified accordingly.","section":"§2.1.4 and Appendix"},{"comment":"The manufactured solution does not satisfy the homogeneous shallow water equations stated in Eq. (51). Substituting h = sin(2πx)sin(2πy)cos(πt)+3, u1=u2=1 into the continuity equation gives a nonzero residual: ∂h/∂t + ∂(hu1)/∂x + ∂(hu2)/∂y = -π sin(2πx)sin(2πy)sin(πt) + 2π cos(2πx)sin(2πy)cos(πt) + 2π sin(2πx)cos(2πy)cos(πt). The paper never introduces or specifies a source term for this manufactured solution. As written, therefore, the h-convergence study in Figure 8 does not verify the scheme for the stated equations. The authors should state the source term explicitly, explain how it is discretized, and if any code used a source term, this must be documented.","section":"§3.3, Eqs. (51) and (58)"}],"minor_comments":[{"comment":"The definition α+ = max_{Δwi>0} w_i/Δwi is not the bound needed if one wants to prune a positive-Δw weight; the correct bound is min_{Δwi>0} w_i/Δwi. As printed, α = min{α-, α+} always equals α- because α- ≤ 0 ≤ α+, so the α+ formula is inert and potentially misleading. Please clarify whether the algorithm intentionally always prunes a negative-Δw weight or whether the positive-side formula is a typo.","section":"Appendix, Eqs. (68)-(70)"},{"comment":"The construction of the reference-triangle quadrature rules used after the geometric mapping is not described. Table 1 lists integrand degrees up to 134 for N=8, but the paper does not state which quadrature rules or libraries are used to achieve exactness at those degrees; this information is needed for reproducibility and for assessing the practical cost of the method.","section":"§2.2"},{"comment":"The convergence study reports only a log-log plot. Adding a table with L2 errors, orders of convergence, and the corresponding mesh parameters would make the verification more transparent and easier to reproduce.","section":"§3.3"},{"comment":"The displayed equation for the entropy wave reduction is typeset in a confusing way; the scalar advection equation ∂ρ/∂t + u1∂ρ/∂x + u2∂ρ/∂y = 0 should be displayed separately from the vector-valued momentum/energy relations.","section":"Eq. (59)"},{"comment":"The statement that the method 'can be adapted to any hyperbolic conservation law taking the form in Equation (1)' should be qualified: the construction requires an entropy-entropy flux pair, an entropy conservative/stable two-point flux, and the quadrature conditions discussed in Section 2.1.4.","section":"Conclusions"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is fundamentally in-scope for a numerical-analysis journal and the central idea is attractive. My main concern is that the quadrature-pruning argument is an existence argument combined with an empirical assertion, while the paper's headline claim is unconditional. The manufactured-solution issue is likely fixable and may just be an omission in the text, but it should be corrected before publication. I did not find evidence of a circular derivation; the entropy stability argument is inherited from the cited SBP framework, and the paper's own numerical tests do target the claimed properties."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: this is a genuine first — high-order entropy stable DG on cut meshes — and the core construction is believable. The authors combine Chan's skew-hybridized SBP with Carathéodory pruning to build the quadrature rules that the framework needs on arbitrary cut elements, and the numerics show entropy conservation to round-off, nonpositive entropy residual for the ES version, and stable runs on airfoil and dam-break problems. Code is provided. That part is solid.\n\nThe soft spots are real but mostly fixable. The biggest one is Section 3.3. The 'manufactured solution' (58) does not satisfy the shallow water equations with u1=u2=1; there is a nonzero residual, and the paper never mentions adding a source term. That means the reported h^{N+1} convergence rates are not actually measuring convergence to the manufactured solution. This needs to be redone, either by adding the correct source term or by using a genuine solution.\n\nSecond, the appendix's pruning step as printed looks wrong. To keep weights nonnegative when Δw_i > 0 you need α ≤ w_i/Δw_i, so α+ should be the minimum of those ratios, not the maximum. As written, α = min{α-, α+} can pick a negative α- and drive weights negative. The numerical results suggest the implementation does something sane, so this may be a typo, but the exposition needs fixing. Relatedly, there is no proof that pruning always leaves enough strictly positive weights for positive-definiteness of the mass matrix on arbitrary degenerate cuts — the paper says 'typically easily met,' which is not a guarantee. That weakens the generality claim, though not the observed behavior on the test cases.\n\nThird, the state redistribution entropy-stability observation is explicitly unproven, and the authors say so. Fine, but it should not be read as more than a promising numerical hint.\n\nThe central stability proof is inherited and correctly applied; the novelty is in the quadrature pipeline, and that pipeline's robustness is the main open question. I'd send this to peer review. The method is important enough and the framework is sound enough to merit referee time, but I'd require the convergence study to be redone and the pruning formula corrected before acceptance. If fixed, this will be a useful reference for anyone doing embedded-boundary high-order CFD.","headline":"Genuine first for entropy stable DG on cut meshes, but the convergence study and pruning formula need correction before the claims are fully supported.","tokens_in":22487,"tokens_out":3091,"would_cite":true,"duration_ms":28505,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M60","65M12","65D32"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper presents the first high-order, entropy stable discontinuous Galerkin method that works on cut meshes with arbitrarily shaped elements.","keywords":["entropy stability","discontinuous Galerkin","cut meshes","summation-by-parts operators","Carathéodory pruning","quadrature","hyperbolic conservation laws","embedded boundaries"],"falsifier":"Construct a family of degenerate cut elements, such as near-zero-area slivers, cuts that pass almost through a mesh vertex, and cuts that intersect a background cell in several disjoint pieces, run the subtriangulation and Carathéodory pruning for increasing polynomial degree $N$, and check exactness of degree $2N-1$ integrals, non-negativity of all weights, and positive-definiteness of the mass matrix. A single geometry in which the pruned rule has fewer than $\\mathrm{dim}(P_N)$ strictly positive weights, or in which a mass-matrix eigenvalue becomes zero or negative, would disprove the general claim.","tokens_in":21469,"feed_emoji":"📐","tokens_out":7364,"duration_ms":68036,"temperature":0.7,"pith_summary":"The paper aims to establish the first high-order accurate scheme that is entropy stable on cut meshes: grids containing arbitrarily shaped elements produced when embedded boundaries slice a Cartesian background mesh. Entropy stability is the nonlinear analogue of energy stability for hyperbolic conservation laws, and it has been available for simplicial and tensor-product elements but not for cut elements, whose arbitrary shapes defeat standard quadrature assumptions. The authors show that a skew-hybridized summation-by-parts formulation only needs polynomially exact, non-negative-weight quadrature rules on each cut element, and they construct such rules by subtriangulating each cut element and then pruning the composite rule with a Carathéodory-based algorithm. If the construction holds for every possible cut, then robustness, high-order accuracy, and efficiency on embedded-boundary domains can be combined without artificial stabilization. The claim is verified numerically for the shallow water and compressible Euler equations, including shock waves and sharp airfoil geometry.","feed_headline":"First high-order entropy stable method for cut meshes","feed_subtitle":"Skew-hybridized summation-by-parts plus pruned quadrature keep high-order schemes stable on embedded-boundary grids.","key_machinery":"The load-bearing object is the skew-hybridized summation-by-parts operator, a block operator built from a non-symmetric differentiation matrix $Q_d$, an extrapolation operator $E$ mapping volume quadrature values to surface points, and a boundary integration matrix $B_d$, arranged so that the operator plus its transpose equals a boundary-only matrix. It carries the argument because it restores summation-by-parts on arbitrary element shapes without requiring the diagonal mass matrix of classical SBP theory. The second mechanism is Carathéodory pruning: starting from an exact positive-weight composite rule obtained by subtriangulation, repeatedly subtract a null vector of the quadrature Vandermonde matrix to zero out weights while preserving exactness and non-negativity, stopping at the Carathéodory bound. This produces the small, positive-weight volume and surface rules that the skew-hybridized operator needs on every cut element.","core_discovery":"On cut meshes, classical diagonal-norm summation-by-parts operators are impractical because every cut element has a different shape and cannot be mapped to one reference element. The paper's central claim is that the skew-hybridized summation-by-parts operator satisfies the hybridized SBP property and needs only two things: volume quadrature exact for polynomials of degree $2N-1$ and surface quadrature exact for polynomials of degree $2N$, with non-negative weights and enough positive weights to keep the mass matrix positive definite. Because no special orthogonality is imposed on the mass matrix, these rules can be built individually for each cut element by splitting it into curved triangles, composing known reference-triangle rules, and pruning the many-point rule using null vectors of the quadrature Vandermonde matrix until the Carathéodory bound of at most $N^*+1$ points is reached. The scheme then uses entropy conservative or entropy stable two-point fluxes on the pruned quadrature points and delivers semi-discrete entropy conservation or stability at high order. In the authors' words, this constitutes the first instance of a high-order accurate entropy stable method on cut meshes.","pith_inferences":["Beyond the paper: the same subtriangulation-plus-pruning construction should port to three-dimensional cut meshes by subtetrahedralization, provided the geometric-mapping degree and quadrature-exactness conditions scale; this extension is not demonstrated in the paper.","Beyond the paper: the observed compatibility between state redistribution and entropy stability suggests a target theorem, namely that the projection and averaging steps of state redistribution are entropy-dissipative or entropy-bounded for entropy stable fluxes; if proven, the small-cell restriction would be removable while retaining the stability guarantee.","Beyond the paper: the method inherits the explicit boundary parameterization used to build the cut mesh; switching to a polynomial level-set representation would allow a different quadrature construction, but the surface-rule exactness conditions and the entropy argument would have to be re-verified for that representation."],"forward_implications":["High-order entropy stability is no longer restricted to fitted simplicial or tensor-product meshes; embedded-boundary domains can use coarse Cartesian cut meshes without artificial viscosity or tuned limiters.","The same formulation applies to any hyperbolic conservation law of the form in Equation (1), so the shallow-water and compressible-Euler demonstrations should transfer to other systems that admit an entropy-entropy flux pair.","Because cut meshes have fewer faces than equivalent quad-tri meshes, the scheme reduces the dominant cost of flux-differencing, which is the number of numerical flux evaluations.","State redistribution, the standard fix for the small-cell CFL problem, can be applied to the entropy stable version of this method and is observed numerically not to destroy entropy stability.","The h-convergence study recovers order $N+1$ in $L^2$ for $N=2,3,4$, matching the expected high-order DG accuracy on cut meshes."],"supporting_citations":[{"why":"Supplies the skew-hybridized SBP operator and the quadrature exactness conditions that the entire cut-mesh formulation depends on.","marker":"[1]"},{"why":"Provides the discrete entropy conservation/stability proof for the flux-differencing DG formulation and the condition $Q_{H,d}\\mathbf{1}=0$.","marker":"[26]"},{"why":"Supplies the QR-factorization-based pruning algorithm used to reduce composite quadrature rules while preserving non-negative weights and polynomial exactness.","marker":"[51]"},{"why":"Supplies the explicit-parameterization cut-mesh generation and approximate Fekete points used to define the solution spaces on cut elements.","marker":"[45]"},{"why":"Provides Carathéodory's theorem, which gives the upper bound of $N^*+1$ points used to stop the pruning process.","marker":"[53]"},{"why":"Defines entropy conservative and entropy stable two-point fluxes and entropy potentials that the scheme's volume and boundary flux evaluations rely on.","marker":"[10]"}],"fun_headline_variants":["Entropy stable high-order DG works on cut meshes","Cut meshes conquered by entropy stable high-order DG","First high-order entropy stable scheme for cut grids","Skew-hybridized SBP unlocks entropy stability on cuts","Pruned quadrature enables entropy stable cut-mesh DG"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire entropy-stability guarantee rests on the assumption that, for every cut element, the pruning algorithm returns a quadrature rule that is exact to the required degree, keeps all weights non-negative, and leaves enough strictly positive weights for the mass matrix to remain positive definite; the paper constructs such rules numerically but does not prove that pruning always terminates successfully on degenerate cut shapes.","fun_headline_variants_meta":{"raw":{"variants":["Entropy stable high-order DG works on cut meshes","Cut meshes conquered by entropy stable high-order DG","First high-order entropy stable scheme for cut grids","Skew-hybridized SBP unlocks entropy stability on cuts","Pruned quadrature enables entropy stable cut-mesh DG"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000204,"raw_usage":{"total_tokens":1401,"prompt_tokens":968,"completion_tokens":433,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":584,"completion_tokens_details":{"reasoning_tokens":353}},"tokens_in":584,"tokens_out":433,"duration_ms":4063,"temperature":1.0,"reasoning_tokens":353,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:31:18.077612+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a family of degenerate cut elements, such as near-zero-area slivers, cuts that pass almost through a mesh vertex, and cuts that intersect a background cell in several disjoint pieces, run the subtriangulation and Carathéodory pruning for increasing polynomial degree $N$, and check exactness of degree $2N-1$ integrals, non-negativity of all weights, and positive-definiteness of the mass matrix. A single geometry in which the pruned rule has fewer than $\\mathrm{dim}(P_N)$ strictly positive weights, or in which a mass-matrix eigenvalue becomes zero or negative, would disprove the general claim.","supporting_citations":[{"cited_title":"Chan, Skew-symmetric entropy stable modal discontinuous Galerkin formulations, Journal of Scientific Computing 81 (1) (2019) 459–485","cited_arxiv_id":null,"evidence_quote":"Supplies the skew-hybridized SBP operator and the quadrature exactness conditions that the entire cut-mesh formulation depends on."},{"cited_title":"Chan, On discretely entropy conservative and entropy stable discontinuous Galerkin meth- ods, Journal of Computational Physics 362 (2018) 346–374","cited_arxiv_id":null,"evidence_quote":"Provides the discrete entropy conservation/stability proof for the flux-differencing DG formulation and the condition $Q_{H,d}\\mathbf{1}=0$."},{"cited_title":"Tadmor, The numerical viscosity of entropy stable schemes for systems of conservation laws","cited_arxiv_id":null,"evidence_quote":"Defines entropy conservative and entropy stable two-point fluxes and entropy potentials that the scheme's volume and boundary flux evaluations rely on."}],"review_version":1}