{"id":"81d4a436-f0a4-4e1a-94ca-4f015f27082e","arxiv_id":"2502.00778","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Lumped directed-area vectors and dual volumes for edge-based discretization can be computed directly from element face data, eliminating the need to construct dual control volumes.","lead":"This paper shows a simpler way to compute the grid geometric quantities used in a common CFD discretization method, without building the intermediate 'dual control volumes' that the usual method requires. The simplification speeds up the computation by a modest factor in 3D and is intended to make the method easier to extend to four-dimensional space-time grids.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Printed element-update formulas depend on an unstated vertex-orientation convention; on a standard CCW triangle they produce negative dual volumes, so the central implementation claim is not reproducible as written.","rationale":"The reader's weakest assumption concerns the closed conforming ring needed for cancellation of interior face terms. That is a relevant scope limitation, but the paper explicitly restricts itself to conforming triangular and tetrahedral grids, so it is not the most load-bearing issue. The more fundamental problem is that the implementation formulas as printed are signed arithmetic whose correctness depends on an unstated element-vertex ordering. A direct hand calculation on a single triangle using the paper's own definitions gives a negative dual volume through Eq. (4.6), and on a regular triangular lattice the printed edge vectors are orthogonal to the edges, which would make all dual volumes vanish. If this concern lands, the central claim that the algorithm can be implemented without forming dual volumes fails for standard mesh storage conventions; the lack of shown verification and shipped code leaves no way to determine the author's intended orientation. The abstract mathematical identity may be salvageable with explicit outward-face orientation and corrected boundary signs, but the manuscript as written is not reproducible, hence a REJECT verdict for the current version.","tokens_in":15584,"tokens_out":64539,"duration_ms":623477,"concrete_test":"Implement Eqs. (3.15)-(3.18) and (4.6) exactly as printed on the single triangle (0,0),(1,0),(0,1) and on a regular triangular-lattice patch with all elements stored counterclockwise. Compare each computed n_jk with the geometric directed-area sum of the median-dual segments for that edge, and compare the computed V_j with the polygon area of the median-dual cell. If V_j is negative, zero, or otherwise differs from the geometric value, the paper must state the required vertex ordering and provide corrected orientation-independent formulas or the shipped code.","verdict_should_be":"REJECT","load_bearing_attack":"The abstract identity (3.31) is plausible if n^E_j is the true outward directed-area vector, but the paper's concrete update rules never specify how to obtain that orientation. The signed expressions (3.15)-(3.16), (3.26)-(3.27), and the boundary corrections (3.17), (3.29) change sign with the cyclic order of element vertices, and no orientation convention is stated. This is not cosmetic. For the single triangle j=(0,0), k=(1,0), r=(0,1), following the printed formulas literally gives n^E_j=(-1,-1), initial contributions n_jk=n_jr=(-1/3,-1/3), boundary corrections n_jk=(-1/3,-1/2) and n_jr=(-1/2,-1/3), and then Eq. (4.6) yields V_j=-1/6 instead of the true median-dual area +1/6. On a regular triangular lattice, the same printed recipe gives each n_jk orthogonal to the corresponding edge, making every dot product in Eq. (4.6) vanish and all dual volumes zero. The paper states that verification tests were performed but results are not shown, so this sign/orientation ambiguity cannot be resolved from the manuscript.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes alternative element-loop algorithms for computing the lumped directed-area vectors n_jk and node-centered median dual volumes V_j required by edge-based finite-volume discretizations on triangular and tetrahedral grids, without explicitly constructing median dual control volumes. The central identities, Eqs. (3.14) and (3.25), express the lumped vector at an interior edge as a fixed multiple of the sum of element face vectors opposite the edge; boundary corrections are then added in a boundary loop. Two dual-volume algorithms are derived and compared: one based on an edge loop using Eq. (4.6), and one based on the element-volume fraction 1/(D+1) in Eq. (4.22). Complexity estimates and CPU timings are reported, showing modest speed-ups for the directed-area computation (1.12x in 2D, 2.16x in 3D), with the main motivation being simpler implementation and extension to four and higher dimensions.","tokens_in":15825,"tokens_out":20514,"duration_ms":195508,"significance":"If the implementation issues are resolved, the paper is a useful contribution to the practical side of edge-based discretization. The derivation is step-by-step, the formulas are parameter-free, and the paper provides a simple node-sum verification test and a divergence-theorem-based check for dual volumes. The reported speed-ups are modest, but the core value is the conceptual simplification of avoiding dual-volume construction and the implied path toward higher-dimensional simplex grids. The mathematical core appears sound; however, the manuscript as written is not reproducible because the sign/orientation convention for simplex vertices is never specified, and the printed element-update formulas produce the wrong sign under standard conventions.","major_comments":[{"comment":"The element-update rules are not reproducible as written because the vertex-orientation convention for simplex elements is never specified. With the standard convention that triangles are stored counterclockwise and tetrahedra have positive orientation, the vector n^E_j defined by the printed coordinate formulas is inward, not outward. For the triangle j=(0,0), k=(1,0), r=(0,1), Eq. (3.15) with n^E_j=(y_k-y_r,x_r-x_k)=(-1,-1), followed by the boundary correction (3.17) with n_B=(y_j-y_k,x_k-x_j) and the volume formula (4.6), gives V_j=-1/6 instead of the true median-dual area +1/6. In three dimensions, Eq. (3.26) computes 2n^E_j=(x_r-x_k)×(x_ℓ-x_k), which for a positively oriented tetrahedron points into the tetrahedron rather than outward. Please state the vertex-ordering convention explicitly, or equivalently replace n^E_j by -n^E_j and adjust n_B and the boundary corrections, and add a simple single-triangle example as a verification.","section":"§3.2.1–3.2.2, Eqs. (3.15)–(3.16), (3.26)–(3.27), (3.17), (3.29)"},{"comment":"The paper states that the proposed implementations were verified by the node-sum test of §3.2.3 and the total-volume test of §4.3, but no verification results are shown; Table 3 contains only CPU times. Because the correctness of the formulas depends on the orientation convention (see previous comment) and a sign error would silently produce negative dual volumes, the absent verification data is load-bearing. Please include the actual verification output for the grids used in Table 3: the maximum node-sum residual after Eq. (3.34) and the sum of dual volumes compared with the exact domain volume.","section":"§5, first paragraph"},{"comment":"The cancellation of the intermediate face vectors n_r, n_ℓ, etc. requires every interior edge to be surrounded by a closed ring of conforming simplex elements. The paper mentions that an adjacent tetrahedron is needed for interior edges, but it does not explicitly state the conforming/manifold assumption in the introduction or in the statement of the general formula (3.31). For grids with hanging nodes, non-manifold edges, or mixed element types, formulas (3.14) and (3.25) are not valid as written and no correction or fallback is provided; please state this limitation clearly in the scope of the paper.","section":"§3.2.1–3.2.2, Eqs. (3.13), (3.25)"}],"minor_comments":[{"comment":"There is a typo in the sentence after Eq. (3.29): the vector nrℓ is described as belonging to the edge {k, ℓ}, but it should be the edge {r, ℓ}.","section":"Eq. (3.29)"},{"comment":"The tetrahedral-grid complexity count for the traditional algorithm does not add up: 33 additions per edge times 6 edges gives 198, plus 9 for the element centroid gives 207, not the 205 stated in the text and table. Please check the arithmetic and correct it.","section":"Table 1 and §3.3.2"},{"comment":"In the sentence after the edge-based dual-volume complexity count, the expression \"3Ne + Nv ≈ 2Ne + 0.2NE\" for three dimensions appears to be a typo; with Ne ≈ 7/6 NE and Nv ≈ NE/6, the correct approximation is about 3.7NE, which is what the final estimate in the same paragraph reports.","section":"§4.3"},{"comment":"Report the variance or per-execution times in addition to the 10-run averages; without any measure of variability it is difficult to judge whether the small differences, such as the 1.12x speed-up in 2D, are significant.","section":"§5"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a computational methods journal. The mathematical identities are sound, but the manuscript needs a clear statement of the element orientation convention and actual verification data before the implementation can be considered reproducible. I would not reject on mathematical grounds; the issues are local and fixable within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this one. The paper is a clean derivation of an implementation trick for computing edge-based discretization metrics without forming median dual volumes. The core formula (3.31) is a known identity, as the author concedes, so the novelty is the implementation and complexity analysis. The complexity numbers are believable and the reported speedups (1.12x in 2D, 2.16x in 3D) are honest—modest but real.\n\nThe problem is that the printed update formulas don't specify the element vertex orientation. The signs of n^E_j and the boundary correction n_B depend on the cyclic order of the triangle's vertices and the order of endpoints in the boundary edge loop, and the paper doesn't fix either. The consequence is not cosmetic. Take the single triangle with vertices (0,0), (1,0), (0,1) in CCW order. Following the formulas literally gives n^E_j = (-1,-1), and after the boundary corrections Eq (4.6) yields V_j = -1/6 instead of the true median dual area +1/6. I checked a regular right-triangle lattice with an interior node and got V_j = -1 instead of +1. So the algorithm as printed produces negative dual volumes on standard meshes. This is a load-bearing flaw in an implementation paper: a reader cannot reproduce the method without reverse-engineering the intended sign convention.\n\nThe stress-test note claimed that on a regular lattice the dot products all vanish and dual volumes become zero. That specific mechanism didn't hold in my check—I got nonzero dots with negative sign—but the conclusion (implementation not reproducible as written) is right.\n\nWhat's good: the derivations are step-by-step, the complexity analysis is detailed and matches the timings, and the author is upfront about prior art and about the speedup being small relative to total CFD runtime. The dual-volume formulas in Section 4 are correctly derived and useful. The paper is also honest about verification tests being performed but not shown, which is a minor gap.\n\nThis paper is for CFD practitioners implementing edge-based solvers. The 4D motivation is plausible but not realized here. I'd send it to peer review—the math is sound and the fix is likely a matter of stating the orientation convention and showing the verification results. It should not be accepted without that clarification.","headline":"Clean write-up of a known identity, but the printed implementation formulas have an unstated sign convention that makes them fail on standard meshes; worth reviewing with a mandated revision.","tokens_in":16352,"tokens_out":41173,"would_cite":false,"duration_ms":328743,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M50","65M08","65N08"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that lumped directed-area vectors and dual volumes for edge-based discretization can be computed directly from element-face data, without ever forming median dual control volumes.","keywords":["edge-based discretization","lumped directed-area vectors","median dual control volume","unstructured grids","triangular grids","tetrahedral grids","grid metric computation","space-time extension"],"falsifier":"Construct a grid with one interior edge whose surrounding element fan does not close, for example a hanging node placed on that edge, and compute $n_{jk}$ both by the proposed formula (3.25) and by the traditional sum of explicit median-dual face directed areas. The two values will differ by uncancelled terms of the form $\\frac{1}{12}(n^E_\\ell + n^E_r)$, showing that the identity depends on the closed conforming fan.","tokens_in":15377,"feed_emoji":"⚡","tokens_out":14192,"duration_ms":129671,"temperature":0.7,"pith_summary":"The paper claims that the two grid metrics needed for node-centered edge-based discretization—the lumped directed-area vector at each edge and the median dual volume at each node—can be computed on triangular and tetrahedral grids without constructing median dual control volumes at all. The directed-area vector follows from a scaled element-loop sum of the face vectors opposite each node, plus a boundary correction, and the dual volume follows either from an edge sum over those same vectors or from each element contributing one part in $D+1$ of its volume. If the claim is right, edge-based solver code becomes simpler, metric setup is faster, and the main geometric obstacle to carrying the discretization into four-dimensional space-time grids is removed. Timed tests show about 1.12x faster directed-area computation on triangular grids and 2.16x faster on tetrahedral grids.","feed_headline":"Grid metric setup skips dual volumes, runs 2.16x faster in 3D","feed_subtitle":"Element-loop formulas for edge vectors and dual volumes simplify edge-based solvers and open a route to four-dimensional space-time grids.","key_machinery":"The load-bearing identity is the cancellation of side-face vectors: inside each element, the dual-face contribution to an edge is written as a multiple of the element face opposite one endpoint plus terms involving the faces opposite the other nodes, and those extra terms cancel when contributions are summed over all elements sharing the edge. This reduces the metric computation to formulas (3.31) and (3.32), with only the outward face vectors of each element and each boundary element needed. The companion device is the divergence theorem applied to $x - x_j$ over the median dual volume, which turns the volume into an exact edge sum over the lumped directed-area vectors, yielding formulas (4.6) and (4.22).","core_discovery":"For a simplex grid of dimension $D=2$ or $D=3$, every edge's lumped directed-area vector can be written as $n_{jk} = \\frac{2}{D(D+1)} \\sum_{E \\in \\{E_{jk}\\}} n^E_j$, where $n^E_j$ is the outward directed-area vector of the face of element $E$ opposite node $j$, plus the boundary correction $n_{jk} \\leftarrow n_{jk} + \\frac{1}{D(D+1)} n_B$ for each boundary element. The median dual volume satisfies $V_j = \\frac{1}{2D} \\sum_{k\\in\\{k_j\\}} (x_k - x_j)\\cdot n_{jk}$, and equivalently every element contributes exactly $V_E/(D+1)$ to each of its nodes. These identities follow from canceling side-face terms in the closed ring of elements around an interior edge and from applying the divergence theorem with an exact edge-based quadrature. The paper's central claim is that these formulas, implemented in element and edge loops, reproduce the traditional dual-face metric data without forming edge midpoints, face centroids, or element centroids.","pith_inferences":["Extrapolating the general form to $D=4$ would give the coefficient $2/(4\\cdot 5)=1/10$ for directed-area vectors and a per-element dual-volume share of $1/5$, a pattern consistent with the paper's stated 4D goal but not derived here.","Because the boundary correction depends only on outward boundary-element normals, it could in principle be applied on boundary representations not derived from median dual faces, such as cut-cell or mixed-boundary treatments, as long as each boundary facet has an unambiguous outward normal.","An implementation that computes the lumped vectors first could reuse them in formula (4.6) within the same edge loop, producing dual volumes with nearly no extra geometric work; the paper times the two stages separately and does not combine them.","The non-cancellation mechanism for hanging-node grids is visible in equation (3.24): missing elements in the fan leave the side-face terms $n^E_\\ell + n^E_r$ uneliminated, so a robust hanging-node version would need correction terms proportional to the imbalance rather than a modified coefficient."],"forward_implications":["Edge-based solvers on triangles and tetrahedra can be implemented without any routine that connects edge midpoints, face centroids, and element centroids to build dual faces.","Metric precomputation is faster: about 1.12x for triangular-grid directed-area vectors and 2.16x for tetrahedral-grid directed-area vectors in the paper's timed tests, with tetrahedral dual volumes about 1.89x faster.","In time-accurate simulations with deforming grids, where metrics are recomputed at every time step, the savings compound; on the paper's large tetrahedral grid, 1,000 recomputations take about 3.75 hours instead of 8 hours.","Because no dual-volume geometry is formed, the formulas remove the main geometric obstacle to extending edge-based discretization to four-dimensional space-time simplex grids, the stated motivation of the paper.","The edge-based dual-volume formula (4.6) is the faster choice on tetrahedral grids, while the element-based formula (4.22) remains competitive on triangular grids, so the preferred implementation is dimension-dependent."],"supporting_citations":[{"why":"It defines the standard median-dual edge-based discretization and its conventional dual-face metric construction, which the proposed formulas replace.","marker":"[1]"},{"why":"It supplies the edge-based quadrature formula used in Section 4 to derive the dual-volume contribution $V_E/(D+1)$ from the divergence theorem.","marker":"[31]"},{"why":"It provides the accuracy-preserving boundary flux quadrature used for closure and verification of edge-based residuals at boundary nodes.","marker":"[32]"}],"fun_headline_variants":["Edge-based grid metrics without forming dual control volumes","2.16x faster edge metrics by skipping dual volume construction","Simpler edge-based discretization: no dual volumes needed","New algorithm computes edge vectors without dual control volumes","Efficient edge-based setup: dual volumes bypassed via element loops"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Every interior edge must be surrounded by a closed ring of conforming triangles or tetrahedra, so that the side-face normal vectors cancel exactly; grids with hanging nodes, mixed element types, or non-manifold edges lie outside the method's stated scope.","fun_headline_variants_meta":{"raw":{"variants":["Edge-based grid metrics without forming dual control volumes","2.16x faster edge metrics by skipping dual volume construction","Simpler edge-based discretization: no dual volumes needed","New algorithm computes edge vectors without dual control volumes","Efficient edge-based setup: dual volumes bypassed via element loops"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000547,"raw_usage":{"total_tokens":2628,"prompt_tokens":975,"completion_tokens":1653,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":1573}},"tokens_in":591,"tokens_out":1653,"duration_ms":12344,"temperature":1.0,"reasoning_tokens":1573,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T17:46:11.776750+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a grid with one interior edge whose surrounding element fan does not close, for example a hanging node placed on that edge, and compute $n_{jk}$ both by the proposed formula (3.25) and by the traditional sum of explicit median-dual face directed areas. The two values will differ by uncancelled terms of the form $\\frac{1}{12}(n^E_\\ell + n^E_r)$, showing that the identity depends on the closed conforming fan.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It defines the standard median-dual edge-based discretization and its conventional dual-face metric construction, which the proposed formulas replace."},{"cited_title":"Nishikawa","cited_arxiv_id":null,"evidence_quote":"It supplies the edge-based quadrature formula used in Section 4 to derive the dual-volume contribution $V_E/(D+1)$ from the divergence theorem."},{"cited_title":"Nishikawa","cited_arxiv_id":null,"evidence_quote":"It provides the accuracy-preserving boundary flux quadrature used for closure and verification of edge-based residuals at boundary nodes."}],"review_version":1}