{"id":"f439f1b7-3199-4660-9097-29ee86865ab1","arxiv_id":"2607.06045","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Sparse invariant-domain-preserving mortar fluxes based on LGL subcell characteristic functions enable convex limiting of LGL-DGSEM on nonconforming Cartesian AMR meshes.","lead":"Researchers built sparse low-order interface fluxes that keep high-order discontinuous Galerkin methods stable and physically admissible on meshes with hanging nodes. This closes the gap that previously blocked combining high-order accuracy, positivity/shock limiting, and adaptive mesh refinement for gas-dynamics simulations.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The central algebraic construction is self-contained and the IDP proof is the standard bar-state argument once the fluxes are shown to fit form (11). The reader correctly identifies the Cartesian/one-level/transfer limitations as the weakest assumptions; those limitations are already stated by the authors and do not affect the validity of the claim inside its declared scope. No hidden inconsistency or unstated hypothesis that would invalidate (31)–(32) or the CFL argument was found. Consequently the ACCEPT verdict stands without modification.","tokens_in":29271,"tokens_out":470,"duration_ms":5022,"concrete_test":"Independently recompute the local weights for N=3 from the characteristic-function definition (26) and verify that they reproduce Table 5 of the appendix; then confirm that the resulting sparse stencil still satisfies the discrete metric identity (39) and that the bar-state form (42) continues to hold. If either identity fails, the IDP argument collapses.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's strongest claim is correct as stated. The sparsified mortar fluxes (31)–(32) are derived from non-negative LGL-subcell characteristic weights (26) that satisfy the partition-of-unity identities (28). Conservation follows immediately from the antisymmetry of the normal vectors and the weight identities. On matching interfaces the characteristic functions coincide, so the weights collapse to Kronecker deltas times the face quadrature weights (33) and the flux reduces exactly to the standard conforming LLF flux. Remark 2 rewrites the fluxes in the graph-viscosity form (11) with discrete metric identities (38)–(39) that continue to hold for the nonconforming sparse stencil; the bar-state rewriting (42) and the CFL restriction (44) then give the standard convex-combination argument for IDP under forward Euler (and hence under SSP-RK). All of these steps are algebraic and do not rely on unstated assumptions beyond the Cartesian, equal-degree, one-level setting already declared by the authors. The only genuine limitations (Cartesian meshes, 2-to-1 interfaces, non-IDP transfer operators) are explicitly scoped in the paper and do not undermine the claim that is actually made.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper constructs sparse, conservative, invariant-domain-preserving (IDP) mortar fluxes for nonconforming interfaces in LGL-DGSEM on Cartesian meshes with hanging nodes (equal polynomial degree, one-level 2-to-1 refinement). Starting from an all-to-all central mortar integral, the authors replace the dense L2-projection weights by non-negative overlap integrals of LGL-subcell characteristic functions, add a local Lax–Friedrichs graph-viscosity term, and obtain interface fluxes (31)–(32) that reduce exactly to the standard conforming LLF flux on matching faces, satisfy discrete metric identities, fit the graph-viscosity low-order form (11), and therefore admit the standard bar-state convex-combination argument for IDP under the CFL restriction (44). The fluxes are then blended with high-order L2-mortar fluxes via a posteriori FCT-type limiting; a Zhang–Shu scaling limiter is applied after AMR transfers. Numerical tests (density-wave and isentropic-flow convergence, Kelvin–Helmholtz, Sedov, Double Mach, high-Mach jet) confirm conservation, expected orders under pure positivity limiting, and robustness on adaptively refined meshes.","tokens_in":29560,"tokens_out":826,"duration_ms":9588,"significance":"The construction supplies the missing low-order interface operator that allows existing convex-limiting / graph-viscosity DGSEM frameworks to be used with hanging-node AMR. The algebraic derivation is self-contained once the standard LGL-DGSEM and graph-viscosity machinery are granted; conservation, reduction to the conforming case, and the IDP property under the stated CFL follow by direct verification without free parameters. Implementations are provided in the open-source Trixi.jl framework, and the numerical suite includes both order verification and genuinely challenging Euler problems that require positivity and shock capturing. Within the declared Cartesian, equal-degree, one-level setting the result is a clean and useful building block for high-order adaptive simulations of nonlinear hyperbolic systems.","major_comments":[],"minor_comments":[{"comment":"Section 2.2, after (33): a short explicit statement that the sparsified weights remain non-negative and form a partition of unity on both sides of a 2-to-1 interface would make the subsequent bar-state argument completely self-contained without reference to the appendix.","section":null},{"comment":"Remark 6 and Section 3.4: the practical use of the less restrictive low-order CFL together with low-order-solution bounds is well motivated, but a one-sentence clarification that the resulting scheme is no longer provably IDP (only positivity-preserving in practice) would avoid any ambiguity for readers who skip the remark.","section":null},{"comment":"Section 2.4: the Zhang–Shu transfer limiter is correctly described as positivity-stabilizing rather than fully IDP; a brief remark that a fully IDP transfer operator remains open would be helpful.","section":null},{"comment":"Figures 9b, 12a, 13a, 16, 18a: the colour scales for limiting factors and indicator variables are not always labelled; adding a colour bar or a short legend would improve readability.","section":null},{"comment":"A few typographical slips remain (e.g., “adaptivemeshrefinement”, missing spaces after commas in the abstract and introduction, “theso-called”). A careful proof-reading pass would remove them.","section":null},{"comment":"Table 5 and the accompanying text in Appendix A give a useful concrete example of the sparse weights; a one-line reference to this table already in Section 2.2 would help readers who want an immediate illustration of the stencil.","section":null}],"recommendation":"accept","confidential_remarks":"The manuscript is a natural and carefully executed extension of the authors’ previous conforming-mesh work. The Cartesian / one-level restriction is clearly stated and does not undermine the central claim. I see no reason to request major changes; the paper is ready for publication after the minor polishing items listed above."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This paper supplies the missing low-order mortar operator that lets existing graph-viscosity / convex-limiting DGSEM frameworks work on nonconforming Cartesian meshes. That is the whole contribution, and it is real.\n\nWhat is new is the sparsified interface flux built from integrals of LGL-subcell characteristic functions (their eqs. 26, 31–32). The all-to-all mortar is first written in central + graph-viscosity form, then sparsified so that only overlapping subcells couple. The resulting weights stay non-negative, sum to the correct face measures, collapse exactly to the usual conforming LLF flux when nodes match, and fit the standard low-order residual (11) so the usual bar-state / convex-combination IDP argument goes through under the stated CFL. Conservation and discrete metric identities are checked algebraically. That chain is clean and does not hide free parameters.\n\nThey also show how to do FCT-style limiting at the mortar level (one factor per mortar, budget split among incident faces) and add a simple Zhang–Shu scaling step for the refinement/coarsening transfer. Numerics cover the expected ground: high-order convergence when limiting is inactive, reduced order when local bounds are active, and a battery of hard Euler tests (isentropic rarefaction, KHI, Sedov, double Mach, Mach-2000 jet) with AMR. The code base is Trixi.jl; configs are available on request.\n\nSoft spots are exactly the ones they declare: Cartesian only, equal polynomial degree, at most one-level (2-to-1) hanging nodes, and the transfer operators are only positivity-stabilized, not fully IDP. The IDP CFL is more restrictive than the usual low-order CFL, so they sometimes fall back to low-order bounds; that is practical but weakens the guarantee. None of these undercut the claim that is actually made.\n\nThis is for people already using or building IDP/subcell DGSEM who want AMR without abandoning the invariant-domain machinery. It is not a conceptual breakthrough, but it removes a concrete roadblock. I would send it to referees without hesitation; the math and the evidence are solid enough to deserve that time.","headline":"Clean, correctly scoped fix that finally lets IDP/convex-limiting LGL-DGSEM run on Cartesian AMR with hanging nodes.","tokens_in":30138,"tokens_out":553,"would_cite":true,"duration_ms":6611,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M60","65M70","76M10","35L65"],"pacs":[],"model":"grok-4.5","headline":"Sparse low-order mortar fluxes let high-order DG methods keep invariant domains on hanging-node adaptive meshes.","keywords":["DGSEM","invariant domain preservation","convex limiting","mortar methods","adaptive mesh refinement","graph viscosity","Euler equations","positivity preservation"],"falsifier":"On a 2-to-1 Cartesian mortar with equal polynomial degree, replace the characteristic-function weights by the dense L2-projection mortar weights and check whether the resulting low-order scheme still keeps density and pressure non-negative for every CFL-stable time step on a near-vacuum isentropic vortex or Sedov blast; if it does not, the sparsification claim is essential.","tokens_in":30208,"feed_emoji":"📐","tokens_out":650,"duration_ms":5774,"temperature":0.7,"pith_summary":"High-order discontinuous Galerkin spectral-element schemes are accurate and efficient, yet they can produce non-physical states near shocks or rarefactions. Invariant-domain-preserving limiters fix that on conforming meshes by blending a carefully designed low-order graph-viscosity scheme with the high-order residual. Adaptive mesh refinement, however, introduces hanging nodes whose standard mortar couplings destroy the low-order structure and the proof of domain preservation. This paper constructs a new set of interface fluxes for exactly those nonconforming faces: they remain conservative, collapse to ordinary conforming fluxes when nodes match, and use LGL-subcell characteristic functions so that each node couples only to a few nearby nodes on the opposite side. The resulting sparse low-order operator fits the existing convex-limiting framework, so positivity of density and pressure (and optional local bounds) can be enforced under a CFL restriction even while the mesh is refined and coarsened. Numerical tests confirm high-order accuracy in smooth regions and robust shock-capturing on classic Euler problems that demand both adaptivity and limiting.","feed_headline":"Sparse mortar fluxes keep high-order DG schemes physical on hanging nodes","feed_subtitle":"Adaptive refinement and invariant-domain limiting now work together for compressible Euler flows","key_machinery":"The sparsified low-order mortar fluxes (31)–(32) whose interface weights are integrals of LGL-subcell characteristic functions; these weights produce a compact stencil that still satisfies the discrete metric identities and the bar-state convexity argument.","core_discovery":"On Cartesian meshes with at most one-level hanging-node interfaces, the sparsified mortar fluxes built from LGL-subcell characteristic-function weights are conservative, reduce to the standard local-Lax–Friedrichs flux on conforming faces, fit the graph-viscosity low-order form, and therefore yield an invariant-domain-preserving semi-discrete scheme under an explicit CFL restriction.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Sparsified mortar fluxes enable IDP limiting for LGL-DGSEM hanging nodes","Compact LGL subcell weights keep AMR DG schemes invariant-domain preserving","Graph-viscosity form extends to nonconforming interfaces via sparse mortars","Hanging-node LGL-DGSEM stays conservative and IDP with sparsified fluxes","AMR and convex limiting unite through sparse LGL mortar couplings"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"The construction and the IDP proof hold only for Cartesian meshes, equal polynomial degree on both sides of a mortar, and at most a single level of refinement difference; the operators that transfer the solution during mesh refinement or coarsening are only positivity-stabilized, not proved fully invariant-domain-preserving.","fun_headline_variants_meta":{"raw":{"variants":["Sparsified mortar fluxes enable IDP limiting for LGL-DGSEM hanging nodes","Compact LGL subcell weights keep AMR DG schemes invariant-domain preserving","Graph-viscosity form extends to nonconforming interfaces via sparse mortars","Hanging-node LGL-DGSEM stays conservative and IDP with sparsified fluxes","AMR and convex limiting unite through sparse LGL mortar couplings"]},"model":"grok-4.5","effort":"low","cost_usd":0.006558,"raw_usage":{"total_tokens":1673,"prompt_tokens":774,"num_sources_used":0,"completion_tokens":103,"cost_in_usd_ticks":65580000,"prompt_tokens_details":{"text_tokens":774,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":796,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":774,"tokens_out":103,"duration_ms":8436,"temperature":1.0,"reasoning_tokens":796,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T16:07:10.864516+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"On a 2-to-1 Cartesian mortar with equal polynomial degree, replace the characteristic-function weights by the dense L2-projection mortar weights and check whether the resulting low-order scheme still keeps density and pressure non-negative for every CFL-stable time step on a near-vacuum isentropic vortex or Sedov blast; if it does not, the sparsification claim is essential.","supporting_citations":[],"review_version":2}