{"id":"af91c7d2-98d6-43c0-b263-36eab6808f86","arxiv_id":"1908.04556","paper_version":4,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A general optimization-based correction method enforces entropy conservation/dissipation and kinetic energy preservation for residual distribution and discontinuous Galerkin schemes, with explicit formulas.","lead":"This paper gives a general recipe for adding small correction terms to many numerical schemes so that they exactly preserve entropy, and optionally kinetic energy, even on arbitrary grids. It turns the correction into a simple optimization problem and tests it on gas dynamics problems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The semidiscrete correction is sound, but the fully discrete DeC entropy-conservation claim (Section 2.1, Remark 2.2) is not proven; without a discrete entropy identity the 'fully discrete' novelty is unsupported.","rationale":"The reader's weakest_assumption identifies positivity/admissibility as the key condition, and that is indeed a genuine limitation: when density or pressure leave the admissible set, the entropy variables and the denominator in (26) are undefined, and the paper's own Taylor-Green experiments (Section 4.2.2) show crashes. However, that limitation is explicitly acknowledged and can in principle be addressed by positivity-preserving limiters, as the paper does for other tests. The more load-bearing gap for the central claim as stated is the fully discrete entropy conservation via DeC: the abstract presents it as a novelty, the algorithmic description in Section 2.1 is only a sketch, and Remark 2.2 explicitly subordinates the fully discrete analysis. Since the semidiscrete Theorem 2.8 does not automatically imply a discrete-in-time entropy identity for the multi-stage DeC update, the fully discrete claim needs either a proof or a qualification. The reader's rationale does mention this issue, so agreement is partial rather than full. The concrete test above would settle whether the FEC claim is exact or only asymptotic in the time step, and would determine whether the verdict should remain CONDITIONAL or be strengthened. I do not see a reason to alter the reader's CONDITIONAL verdict: the core semidiscrete construction is correct under stated hypotheses, and the weaknesses are addressable.","tokens_in":26637,"tokens_out":8263,"duration_ms":92095,"concrete_test":"Test the FEC claim directly: for Burgers' equation with entropy eta(u)=u^2/2 on a periodic interval, implement DeC(2,2) with the correction applied to the full bracket (19) as described in Section 2.1. Over 100 steps with CFL=0.9, compute the discrete entropy residual R^n = eta(U^{n+1}) - eta(U^n) + Delta_t F_flux^n at each step. If max|R^n| is at the level of the time integrator's truncation error (O(Delta_t^2) or O(Delta_t^3)) rather than machine zero, then the claim of exact fully discrete entropy conservation is false as stated; if it is machine zero, the claim is supported. Either way the test settles whether the FEC assertion needs a proof or a qualification.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central semidiscrete correction r = lambda(F - mean) in Theorem 2.8 is mathematically sound: for admissible states with nonconstant F, the KKT conditions yield the stated unique minimizer. The load-bearing weakness is the fully discrete claim. The abstract and the algorithm in Section 2.1 (update (18), correction of the whole space-time residual (19)) assert that applying the correction to the DeC bracket yields fully discrete entropy conservative/dissipative schemes. Remark 2.2, however, states that the paper focuses on semidiscrete corrections, that mainly SEC/D schemes are investigated, and that relaxation combinations are deferred. No theorem derives a discrete entropy balance for the multi-stage DeC update; the correction is computed from stage values, but the final update U^{n+1} is obtained after p correction sweeps, so an entropy identity for the complete step does not follow from the semidiscrete proof. The numerical section reports fully discrete EC results without quantitative discrete entropy residuals. Positivity/admissibility is a real limitation (Section 4.2.2), but it is explicit and qualifiable; the FEC gap is a missing justification for a headline contribution.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a unified optimization-based interpretation of entropy correction terms for residual distribution (RD) and summation-by-parts/Simultaneous-Approximation-Term (SBP-SAT) semidiscretizations, including discontinuous Galerkin and finite difference methods. It characterizes the correction as the unique minimizer of a quadratic program subject to conservation and entropy constraints (Theorem 2.8), extends the setting to entropy inequalities (Theorem 3.1), multiple linear constraints (Theorem 3.2), and kinetic-energy preservation for the Euler equations (Prop. 3.5), and supplies explicit formulas in each case. The authors also describe a fully discrete variant obtained by correcting the whole space-time residual in a deferred-correction (DeC) framework, present numerical comparisons for CG, psi, and DG schemes, and give entropy-stable grid transfer operators in an appendix.","tokens_in":26883,"tokens_out":6340,"duration_ms":70939,"significance":"If the semidiscrete results are taken as the contribution, the paper is a valuable systematization: the explicit formulas remove the need for optimization solvers, the framework is independent of special quadrature rules, and the extension to multiple constraints and kinetic-energy preservation is a genuine generalization of Abgrall's earlier construction. The algebraic derivations in Theorems 2.8, 3.1, and 3.2 are sound, and the paper is commendably explicit about the positivity limitation in Section 4.2.2. The numerical tests confirm that the corrections enforce the imposed constraints, as expected by construction rather than as empirical discovery. The main weakness is that the fully discrete entropy-conservation claim is asserted rather than proved, and the abstract presents this fully discrete extension as a headline contribution.","major_comments":[{"comment":"The fully discrete FEC/D claim is not supported by a proof. The paper shows that the correction is optimal in the semidiscrete setting (Theorem 2.8), but for the DeC update (18) the final value U^{n+1} is obtained after p correction sweeps over several substages, and no discrete entropy identity for the complete step is derived. The algorithm described after Eq. (19) is also underspecified: it gives the steps conceptually but no explicit correction formula for the full space-time residual analogous to (26). The numerical section reports 'fully discrete EC' results but does not provide a quantitative fully discrete entropy residual (for example, entropy change as a function of dt or number of correction sweeps). Without such an identity or error analysis, the abstract's statement that the approach yields 'for the first time a fully discrete entropy conservative/dissipative RD scheme' is a claim rather than a demonstrated result. I recommend either proving the discrete entropy balance for the DeC correction or reframing the contribution as semidiscrete, with the fully discrete variant presented as a numerical demonstration.","section":"2.1"},{"comment":"The 'universal tool' claim should be qualified in the abstract and introduction. The correction formulas (26) and (42) require the entropy variables F to be well-defined, which for the Euler equations requires positive density and pressure; the denominator in (26) also vanishes for constant states. The Taylor-Green experiments in Section 4.2.2 show that when negative density or pressure appears, the correction cannot be applied and the simulation crashes. Since the correction itself does not preserve admissibility, the paper should state clearly that the framework applies to any scheme only for admissible states, and that positivity preservation is an additional requirement outside the scope of the correction. This is not an internal inconsistency, but it limits the claimed universality and should be stated prominently.","section":"4.2.2"}],"minor_comments":[{"comment":"The denominator in (13) should use a norm squared, e.g., sum_i ||F_i - Fbar||^2, to avoid confusion about vector-valued entropy variables.","section":"Eq. (13)"},{"comment":"Remark 2.2 emphasizes that the paper focuses on semidiscrete corrections, yet the abstract presents the fully discrete extension as a major novelty. The abstract should reflect the actual emphasis of the paper.","section":"2.1"},{"comment":"There is a typo 'formulation formulation' in the sentence about motivating examples; it should read 'formulation'.","section":"4.2.2"},{"comment":"The observation that the entropy change decreases as dt^4 in Figure 10 is explained by the order of the time integrator, but the text should state this explicitly rather than leaving the reader to infer it.","section":"A.2"},{"comment":"The nondegeneracy condition for the multiple-constraint theorem is only expressed verbally as 'the constraints do not contradict each other'. It would help to state explicitly that the coefficient matrix in the theorem must be invertible, and to comment on when this can fail for the combined entropy and kinetic-energy constraints.","section":"3.2"}],"recommendation":"major_revision","confidential_remarks":"The semidiscrete contribution is sound and the paper is honest about its limitations. The main reason for major revision, rather than minor revision or acceptance, is the unsupported fully discrete FEC claim, which is part of the advertised novelty. The authors can likely fix this by either providing a discrete entropy identity for the DeC update or by visibly downscoping the FEC claim in the abstract and conclusions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a look. The semidiscrete correction formulas are solid, and the optimization reinterpretation is a genuinely useful way to think about Abgrall's construction. The paper delivers what it promises on the semidiscrete side: Theorem 2.8 and its extensions to inequality constraints, multiple constraints, and kinetic energy preservation are clearly derived, explicitly solvable, and do not need an optimizer. Section 5 also makes a fair case that this framework is attractive for non-calorically-perfect equations of state, where classical entropy-conservative flux algebra gets ugly. Credit where due: the algebra is correct, the explicit formulas are reproducible, and the paper is unusually candid about its own limitations.\n\nNow the soft spots, in proportion. The abstract and Section 2.1 float a fully discrete entropy-conservative/dissipative (FEC/D) claim via DeC, but Remark 2.2 admits that the paper focuses on semidiscrete corrections and that relaxation combinations are deferred. No theorem proves a discrete entropy balance for the full multi-stage DeC update, and the numerical section reports fully discrete results without quantitative entropy residuals. That is a real gap between headline and content, though not a fatal one: the semidiscrete core stands, and the FEC claim could likely be repaired by proving a discrete identity or by rescoping the abstract. The positivity issue is also real but honestly acknowledged. The correction formulas require admissible states with nonconstant entropy variables; the Taylor-Green experiments show crashes when density or pressure goes negative. This makes the \"universal tool\" framing too strong, but the paper says \"no free lunch\" itself and points to limiters as future work. Minor, but note: no code or data is provided, which slows reproduction.\n\nWho is this for? Numerical analysts working on entropy-stable schemes for hyperbolic conservation laws, especially those tired of equation-specific entropy flux derivations. They will get a clean, useful toolkit for semidiscrete corrections and a fair comparison against flux-differencing approaches. I would bring it to a reading group and would cite it.\n\nRecommendation: send it to peer review. It deserves a serious referee. The referee should push on the fully discrete claim and ask for either a proof, a counterexample, or a title/abstract that sticks to semidiscrete results. The core contribution is solid enough that this is fixable rather than fatal.","headline":"Sound semidiscrete core, honest limitations, but the fully discrete claim is over-sold and needs either proof or careful rescoping.","tokens_in":27352,"tokens_out":1202,"would_cite":true,"duration_ms":14349,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M12","65M60","65M70","65M06"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper shows that entropy conservation for residual distribution and discontinuous Galerkin schemes can be enforced by adding one explicit, optimally small correction term to each degree of freedom.","keywords":["entropy stability","kinetic energy preservation","conservation laws","residual distribution schemes","discontinuous Galerkin schemes","Euler equations","deferred correction","summation-by-parts"],"falsifier":"Compute the discrete entropy budget $F^T M \\,\\partial_t U$ plus the numerical entropy flux boundary term on one element of a corrected nodal DG scheme for a smooth Euler vortex, using the paper's own quadrature; Theorem 2.8 predicts the budget is zero up to rounding whenever $F$ is not constant, so any significantly nonzero value would contradict the claimed identity.","tokens_in":26448,"feed_emoji":"🧮","tokens_out":11214,"duration_ms":98395,"temperature":0.7,"pith_summary":"This paper tries to establish that any numerical scheme in the residual-distribution class—which includes many continuous/discontinuous Galerkin and flux-reconstruction methods—can be made entropy-conservative or entropy-dissipative by adding one explicitly computable correction term to each degree of freedom. The correction is the unique minimizer of a quadratic program that enforces local conservation and a discrete entropy condition, and the paper gives closed-form formulas for it in both the residual-distribution and the SBP/SAT/DG settings. It then extends the same idea to entropy inequalities, to multiple simultaneous constraints such as entropy plus kinetic energy for the Euler equations, and to fully discrete schemes via deferred correction time integration. If correct, this provides a systematic universal tool that removes the need to derive special entropy-conservative numerical fluxes case by case.","feed_headline":"One explicit term can make many numerical schemes entropy-conservative","feed_subtitle":"The correction is the unique minimal fix: any baseline scheme gains entropy stability without new flux design.","key_machinery":"The load-bearing object is the explicit correction term $r_i = \\lambda (F_i - \\bar F)$ (with $\\bar F$ the mass-matrix-weighted element average of the entropy variables), chosen as the unique minimizer of the quadratic program $\\min \\frac{1}{2}\\|r\\|_M^2$ subject to local conservation and the discrete entropy equality. The coefficient $\\lambda$ is given in closed form as $\\mathcal{E}/\\|F - \\bar F\\|_M^2$, so no optimization solver is needed; the same quadratic program with linear constraints is the template for every extension in the paper, including the inequality, multi-constraint, kinetic-energy, and fully discrete variants.","core_discovery":"The central claim is that entropy conservation can be imposed on any semidiscretization of the form (25) by adding $r_i = \\lambda\\bigl(F_i - \\frac{1^T M F}{1^T M 1}\\mathbf{1}\\bigr)$, where $F$ are the entropy variables, $M$ is the mass matrix, and $\\lambda = \\mathcal{E}/\\|F - \\bar F\\|_M^2$; Theorem 2.8 shows this is the unique solution of $\\min \\frac{1}{2}\\|r\\|_M^2$ subject to the conservation constraint $1^T M r_i = 0$ and the entropy constraint $F^T M r = \\mathcal{E}$. The paper also claims the same quadratic-program characterization yields closed-form corrections for entropy inequalities (Theorem 3.1), for two simultaneous constraints such as entropy and kinetic energy (Theorem 3.2), and for a fully discrete DeC update, so the framework is a unified recipe rather than a single formula for a single scheme.","pith_inferences":["The optimization formulation is constraint-agnostic: any additional linear conservation law (e.g., angular momentum) should be enforceable by adding its generalized variables to the constraint set, a route the paper does not explore.","Because the correction is the minimal-norm repair, one can view the method as a stabilization-by-projection that preserves the baseline scheme's behavior where it is already good; this suggests using it as a general filter for any unstable discretization, not only for entropy.","The vanishing-denominator case limits the approach at strong shocks: an element whose entropy variables are nearly constant acts like a finite-volume element, so a robust implementation would need to switch to the flux-difference correction there; the paper leaves this hybrid as future work.","A testable extension is to use the same quadratic-program correction for grid refinement and coarsening with explicit closed-form weights, since the paper only solves the inequality-constrained version with a generic optimizer."],"forward_implications":["Any locally conservative semidiscretization in the RD/DG/SBP class can be made discretely entropy-conservative without changing quadrature rules or grid structure, as demonstrated for continuous Galerkin, psi-scheme, and nodal DG discretizations.","The inequality variant yields entropy-dissipative schemes that reduce to the baseline scheme when it is already dissipative, so no spurious dissipation is added to a good method.","Multiple linear constraints can be imposed simultaneously, e.g., entropy conservation and kinetic energy preservation for the Euler equations, with a closed-form two-by-two solve.","Applying the correction to the full space-time residual in the deferred correction framework gives fully discrete entropy-conservative schemes, removing the need for special time integrators.","The framework sidesteps the case-by-case derivation of entropy-conservative numerical fluxes, which the paper argues works cleanly for calorically perfect gases but becomes unclear for general or tabulated equations of state."],"supporting_citations":[{"why":"Grounds the original residual-correction construction that this paper reinterprets as an optimization problem.","marker":"[1]"},{"why":"Provides the mass-matrix-weighted variant of the correction that is adopted in formulas (26) and (42).","marker":"[43]"},{"why":"Supplies the discrete entropy-stability theory and the numerical entropy flux framework used to state conditions (10) and (24).","marker":"[57]"},{"why":"Defines the entropy-conservative flux condition and flux potential used in Remark 2.5 and in the motivation of Section 5.","marker":"[58]"},{"why":"Introduces the deferred correction time discretization that the paper uses to obtain fully discrete entropy-conservative schemes.","marker":"[3]"},{"why":"Gives the split-form SBP-DG discretizations used in numerical comparisons and as the flux-differencing baseline.","marker":"[24]"},{"why":"Provides the kinetic-energy-preserving numerical flux and the discrete kinetic energy identity (41) used in Section 3.3 and the experiments.","marker":"[42]"},{"why":"Supplies the comparison of entropy-conservative fluxes that motivates the claim that flux construction is equation-specific.","marker":"[40]"},{"why":"Defines kinetic energy preserving schemes for the Euler equations, the target property extended here to DG via correction terms.","marker":"[29]"}],"fun_headline_variants":["One minimal term makes many schemes entropy-stable","Closed-form entropy fix: the unique optimal choice","Optimization-based entropy correction for many schemes","From RD to DG: one entropy fix to unify them","Entropy stability by construction: the minimal fix"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction assumes the entropy variables $F(u)$ can be evaluated from the numerical solution and are not constant inside an element, so the denominator in formulas (26) and (42) is nonzero; the paper's own experiments in Section 4.2.2 show that once density or pressure turns negative this assumption fails and the corrected scheme blows up.","fun_headline_variants_meta":{"raw":{"variants":["One minimal term makes many schemes entropy-stable","Closed-form entropy fix: the unique optimal choice","Optimization-based entropy correction for many schemes","From RD to DG: one entropy fix to unify them","Entropy stability by construction: the minimal fix"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00081,"raw_usage":{"total_tokens":3564,"prompt_tokens":965,"completion_tokens":2599,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":581,"completion_tokens_details":{"reasoning_tokens":2526}},"tokens_in":581,"tokens_out":2599,"duration_ms":22376,"temperature":1.0,"reasoning_tokens":2526,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:39:41.176579+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the discrete entropy budget $F^T M \\,\\partial_t U$ plus the numerical entropy flux boundary term on one element of a corrected nodal DG scheme for a smooth Euler vortex, using the paper's own quadrature; Theorem 2.8 predicts the budget is zero up to rounding whenever $F$ is not constant, so any significantly nonzero value would contradict the claimed identity.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Grounds the original residual-correction construction that this paper reinterprets as an optimization problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the mass-matrix-weighted variant of the correction that is adopted in formulas (26) and (42)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the discrete entropy-stability theory and the numerical entropy flux framework used to state conditions (10) and (24)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the entropy-conservative flux condition and flux potential used in Remark 2.5 and in the motivation of Section 5."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the deferred correction time discretization that the paper uses to obtain fully discrete entropy-conservative schemes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the split-form SBP-DG discretizations used in numerical comparisons and as the flux-differencing baseline."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the kinetic-energy-preserving numerical flux and the discrete kinetic energy identity (41) used in Section 3.3 and the experiments."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines kinetic energy preserving schemes for the Euler equations, the target property extended here to DG via correction terms."}],"review_version":1}