{"id":"dd39a580-462a-4ebc-932f-460918bc9aaa","arxiv_id":"2608.09327","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"High-order DG and finite volume schemes for relativistic Euler equations that are fully discrete entropy-stable for many entropy pairs and provably bound-preserving via one cellwise projection.","lead":"This paper builds numerical schemes for the special relativistic Euler equations that keep simulated gas states physically admissible while guaranteeing a discrete version of entropy decay for many entropy functions at once. The methods could make simulations of ultra-relativistic jets, shocks, and near-vacuum flows more robust, a practical problem in astrophysics and high-energy fluid dynamics.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The fully discrete multi-entropy-stability proof hinges on an unverified Riemann-solution hypothesis in Section 3.2; if the exact Riemann solution leaves G, the LF average H_n need not be admissible and inequality (7) can fail.","rationale":"I agree with the reader's weakest_assumption: the Section 3.2 Riemann-solution hypothesis is the single most load-bearing premise. It is the only place where the two-point average H_n acquires both admissibility and the per-pair entropy inequality (7), and those feed directly into the weak budgets (13) and the projection feasibility (17). Without it, the proof of the central claim has no foothold even for the canonical entropy pair. I did not find an internal inconsistency: the decomposition (10), the convexity/interval argument for the multi-entropy radius, and the SSP multistep carry-over are sound, and the numerical experiments are extensive. The secondary issues - scoping of \"arbitrary\" entropy families to the H_r-family and the optional COS module's time-level mixing - would need clarification, but they do not affect the core result as much as the Riemann assumption. Because the reader already identified this premise and recommended conditional acceptance, my stress-test does not change the verdict. The proposed test directly probes the two-point inequality and admissibility of H_n across the EOSs used; its outcome would tell us whether the Riemann assumption is necessary in practice or can be replaced by a direct algebraic proof. Until then, the advertised \"provably admissible\" guarantee should be read as conditional on the standard Riemann-solution hypothesis, exactly as the paper states.","tokens_in":28956,"tokens_out":17587,"duration_ms":174273,"concrete_test":"For the ideal, RC, IP, and TM EOSs, generate a dense structured/random sample of admissible left/right states (U_L,U_R) covering near-vacuum, high-Lorentz, and large-jump regimes. For each pair and for the entropy pairs used in Section 7 (canonical plus H_r with r in {0,0.2,0.39}), compute the LF average H_n from Section 3.2 and verify (i) D(H_n)>0 and q(H_n)>0, and (ii) the two-point inequality E^(l)(H_n) <= (E^(l)(U_L)+E^(l)(U_R))/2 - (Q_n^(l)(U_R)-Q_n^(l)(U_L))/2. Also, where an exact Riemann solver is available (ideal EOS), compare H_n with the light-cone average of the exact solution. Report any violation; one counterexample would show the proof's premise is not satisfied by the implemented flux, while millions of clean samples would support the conditional verdict.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.2 states: \"As is standard, we assume that the Riemann problem in the direction n has a self-similar solution that remains in G and satisfies the canonical entropy inequality.\" This single assumption carries the whole construction. It is used to conclude that H_n = (U_L+U_R)/2 - (F_n(U_R)-F_n(U_L))/2 lies in G (as the light-cone average of the Riemann solution, with G convex) and that the two-point entropy inequality (7) holds for every entropy pair in the family, via the canonical-to-family transfer using H'(S)>0. These two-point facts directly produce the weak budgets (13) and the projection feasibility (17), hence the final strong nodal inequalities (27) advertised in the abstract. For the TM, RC, and IP EOSs used in the experiments, existence of a self-similar Riemann solution that remains in G is not proved, and the light-cone bound of Section 3.1 bounds only wave speeds, not existence or invariant-domain properties. The paper explicitly flags the assumption and imports [68] only for projection geometry, not for this Riemann result. Thus the central \"provably admissible wherever an entropy is evaluated\" claim is conditional on an unproven property of the continuous PDE. This is an externally load-bearing gap rather than an internal inconsistency; the authors' algebra up to that point is coherent.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper constructs fully discrete, bound-preserving, multi-entropy-stable discontinuous Galerkin and finite volume schemes for the special relativistic Euler equations with a general causal equation of state. The central claims are: for any prescribed finite family of convex entropy pairs, the scheme satisfies local and global discrete entropy inequalities for every member of the family; every quadrature state that enters an entropy evaluation or a primitive recovery is provably admissible; and conservation and high-order accuracy are retained with a single cellwise radial projection per time step. The construction combines a light-cone speed bound that fixes the Lax-Friedrichs viscosity at alpha = 1 (Section 3.1), a two-point average that under a Riemann-solution hypothesis inherits one entropy inequality per pair (Section 3.2), Gauss-Lobatto decompositions that convert two-point inequalities into weak cell-average budgets (Section 4), a radial projection that enforces positivity before entropy (Section 5), and SSP multistep integration (Section 6). Numerical experiments with four equations of state, including near-vacuum flows and jets with Lorentz factor above 70, report third-order convergence, absence of inadmissible states, and monotone decay of every enforced discrete total entropy.","tokens_in":29201,"tokens_out":29012,"duration_ms":299391,"significance":"If the advertised properties hold as stated, this is a substantial advance: the existing entropy-stable relativistic hydrodynamics literature is semi-discrete, single-entropy, and positivity is imposed by a separate limiter, whereas this paper couples admissibility and entropy control in one conservative projection for an arbitrary finite entropy family, in a fully discrete setting. The structural argument is attractive, since only convexity of the admissible set and the universal speed bound are used, which suggests transferability to RMHD and GRHD. The paper also contains genuinely parameter-free elements (alpha = 1 from causality), a clear weak-budget mechanism, and extensive, demanding numerical tests with four equations of state; the reported monotone entropy histories and absence of inadmissible states are crisp falsifiable diagnostics. The main theorems, however, are conditional on an unverified Riemann-solution hypothesis, the global decay claim for the multistep realization is stronger than the estimates prove, and a load-bearing part of the projection machinery is imported from an unpublished preprint by the same author.","major_comments":[{"comment":"The assumption that the directional Riemann problem has a self-similar solution that remains in G and satisfies the canonical entropy inequality is load-bearing: it is the only argument given for the admissibility of the two-point average H_n and for the per-pair inequality (7), which in turn produce the weak budgets (13), the projection input (17), and the strong nodal inequalities (27) advertised in the abstract. For the TM, RC, and IP equations of state used in the experiments, existence of such a solution for arbitrary admissible data is not established in the paper, and the light-cone bound of Section 3.1 controls only wave speeds. Since the abstract states \"provably admissible wherever an entropy is evaluated\" unconditionally, the authors should either state the main theorems as conditional on this hypothesis or replace the assumption by a direct proof: the admissibility of the LF average may be provable by the kind of algebraic invariant-domain lemma used in the positivity-preserving literature on relativistic hydrodynamics, and (7) by the standard entropy-stability theory for LF-type fluxes; neither route is currently taken, and both would make the advertised guarantees unconditional.","section":"Section 3.2, Eqs. (7), (10), (13), (17), (27)"},{"comment":"The global monotone decay statement is proven in Section 5.5 only at the forward-Euler level, where summing (13) gives E^{n+1}_{corr} <= E^n_{corr}. For the third-order SSP multistep scheme of Example 6.1, summing the local strong inequalities (27) with the multistep budget of Section 6 yields at best a convex-combination bound of the form S_{n+1} <= (16/27) S_n + (11/27) S_{n-3} in the periodic case, possibly with weighted boundary terms; this does not imply stepwise monotone decay. The abstract and the Conclusions (\"Every reported entropy history decreases monotonically, in agreement with the theoretical estimates\") therefore overstate what the estimates prove for the fully discrete multistep scheme. The authors should either prove a stronger estimate for the multistep realization or state the exact convex-combination inequality as the theoretical claim and present the monotone histories as numerical observations.","section":"Section 6 and Section 5.5"},{"comment":"A load-bearing part of the construction is imported from the unpublished preprint [68] by the same author: the radial projection geometry, the SSP convex-combination argument for the limiting step, and the Cartesian-mesh extension are stated to be taken from [68], and Section 6 explicitly refers to [68, Section 7] for the SSP limiting argument. While Section 5 does prove the relativistic-specific radii (positivity radius via concavity of q, entropy radius via monotonicity of the profile), the general machinery that guarantees mean preservation, nodal admissibility, and budget inheritance under the SSP limiting is not restated here. For a paper claiming a first construction of fully discrete multi-entropy-stable schemes, the main theorems should be self-contained modulo clearly identified statements; I recommend restating the needed results of [68] as numbered assumptions or lemmas, with at least proof sketches.","section":"Sections 1.2, 5, 6; reference [68]"}],"minor_comments":[{"comment":"The symbol H denotes both the entropy-generating function H(S) in (3) and the two-point average H_n(U_L,U_R) introduced in Section 3.2; this clash is confusing and should be resolved, for example by renaming the two-point average.","section":"Section 3.2"},{"comment":"There is a typo: \"Define the prodile\" should read \"Define the profile\".","section":"Section 5.4.2"},{"comment":"The same symbol E^{(ℓ)}_j is used for the entropy function evaluated at a single state, such as E^{(ℓ)}(\\bar U^⋆_j), and for the quadrature entropy of a nodal array, such as E^{(ℓ)}_j(U^{n-r}_j); this makes the budgets (13) and (27) harder to parse, and a separate notation for the discrete quadrature operator would improve readability.","section":"Sections 2.4 and 5.3"},{"comment":"The title and Section 1.3 claim high-order DG and finite volume schemes, but Section 7 presents only P2 DG computations; either add a finite volume test or state explicitly that the FV variant is constructed but not separately benchmarked.","section":"Section 7"},{"comment":"The entropy inequalities (27) are enforced only for time steps after the SSP-RK startup, since the startup steps apply only the bound-preserving module; the paper should state explicitly that the fully discrete entropy guarantee holds for n >= 3, because the conclusions' monotone-histories phrasing covers t = 0 onward.","section":"Section 6"},{"comment":"The displayed factorizations that establish the light-cone bound are unnumbered; numbering them, or at least the final bound, would help readers verify the key estimate that motivates alpha = 1.","section":"Section 3.1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's headline claim rests substantially on the author's own unpublished preprint [68] (the EPO radial projection and SSP limiting geometry) and on in-press or same-group work [66] and [70] for the entropy family and the COS operator. For a \"first\" claim of this scope, I would ask the editor to ensure that [68] and [66] are publicly available and carefully vetted, since the present paper does not re-derive the core geometry that makes the one-projection construction work. The paper is otherwise a good fit for a numerical analysis journal, and the conditional-theorem concern raised in my major comments is, in my view, fixable without changing the scheme."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a strong construction, and the main proof chain—two-point decomposition, weak budgets, one-ray multi-entropy projection—is coherent and I spot-checked the algebra. It delivers the first fully discrete, multi-entropy-stable, bound-preserving schemes for the special relativistic Euler equations. The α=1 light-cone bound, the transfer from the thermodynamic entropy pair to the whole H-generated family via H′(S)>0, and the positivity-before-entropy ordering forced by the implicit primitive recovery are all real contributions. The numerical work is also serious: four EOSs, accuracy tests within 10^-5 of vacuum, ultra-relativistic shocks, jets with Lorentz factor above 70, and monotone entropy diagnostics throughout.\n\nThe main caveat is load-bearing: the entire two-point building block rests on the Section 3.2 assumption that the directional Riemann problem has a self-similar solution in the admissible set G and satisfies the canonical entropy inequality. That hypothesis is what makes the LF average admissible and what turns Riemann entropy admissibility into inequality (7) for every member of the family. It is standard in the literature, but for the TM, RC, and IP EOSs used in the numerics it is not proved, and the light-cone speed bound alone does not give existence or invariant-domain properties of the Riemann solution. The authors flag it honestly, but as it stands the central claim that computed states are 'provably admissible wherever an entropy is evaluated' is conditional on a property of the continuous PDE. A referee should press for a proof or a precise citation for this EOS class, or for an explicit conditional statement.\n\nTwo smaller issues. The projection geometry and SSP convexity argument are imported from the same author's unpublished preprint [68], so the paper is not self-contained at that point; that is common, but it should be resolved. And the entropy guarantee starts only after the SSP-RK startup, when the full EPO operator kicks in; this is disclosed in Section 6, but the abstract's unqualified 'fully discrete' claim and 'arbitrary family' (really any H-generated family, not every convex pair) overreach slightly. Also, code and data are only 'available upon reasonable request,' which is a missed chance for a paper selling provable properties.\n\nSend it to peer review. This deserves a serious referee—it is a substantial step for numerical RHD, and the open points are addressable rather than fatal. I'd bring it to reading group.","headline":"Strong construction, first fully discrete multi-entropy-stable bound-preserving schemes for relativistic Euler, but the headline guarantee is conditional on an unproved Riemann-solution hypothesis.","tokens_in":29769,"tokens_out":4064,"would_cite":true,"duration_ms":41224,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M60","65M12","35L65","76Y05"],"pacs":[],"model":"deepseek-v4-flash","headline":"One cellwise projection per step makes high-order relativistic Euler schemes fully discrete entropy stable for a whole family of entropy pairs and keeps every quadrature state admissible.","keywords":["relativistic Euler equations","fully discrete entropy stability","multi-entropy stability","bound-preserving schemes","discontinuous Galerkin methods","positivity preservation","SSP time stepping","relativistic hydrodynamics"],"falsifier":"Run the proposed P2 DG scheme with a non-ideal causal EOS on a one-dimensional Riemann problem whose pressure ratio spans twelve orders of magnitude, and at every time step check at every Gauss-Lobatto node that $D>0$ and $E - \\sqrt{D^2 + |m|^2} > 0$ and that each prescribed entropy budget $B_j^{(\\ell)}$ is satisfied; a single violation would refute the universal claim, as would a direct computation of $H_n(U_L,U_R)$ for two admissible states whose exact Riemann solution leaves the admissible set.","tokens_in":1849,"feed_emoji":"🚀","tokens_out":5568,"duration_ms":146785,"temperature":0.7,"pith_summary":"The paper claims that for the special relativistic Euler equations one can build high-order discontinuous Galerkin and finite volume schemes that are entropy stable in the fully discrete sense for any prescribed finite family of convex entropy pairs, and that every quadrature state is provably physically admissible wherever an entropy is evaluated. This matters because in relativistic hydrodynamics the two properties cannot simply be added together: an inadmissible state has no defined entropy, and the conservative-to-primitive recovery is well posed only on the admissible set. The construction uses a single cellwise projection per time step, so conservation and the formal order of accuracy are retained, and the numerical viscosity is fixed once for all states and all equations of state by relativistic causality. If correct, the result supplies fully discrete, multi-entropy-stable, bound-preserving high-order schemes for this system.","feed_headline":"One projection yields fully discrete multi-entropy-stable schemes","feed_subtitle":"A single cellwise limiter enforces many entropy inequalities and physical bounds at once for relativistic Euler","key_machinery":"The load-bearing object is the two-point local Lax-Friedrichs average with unit viscosity, $H_n(U_L,U_R) = (U_L+U_R)/2 - (F_n(U_R)-F_n(U_L))/2$, combined with the light-cone factorization showing that every directional characteristic speed of the relativistic Euler system lies in $[-1,1]$ for any causal equation of state. Because $H_n$ is the average of the self-similar Riemann solution over the light cone, it is admissible and satisfies the two-point entropy inequality (7) for every prescribed entropy pair, with only the canonical thermodynamic pair needing an entropy admissibility assumption. The Gauss-Lobatto endpoint decomposition (10) then expresses a forward Euler cell average as a convex combination of admissible states and two such averages, giving one weak budget per pair; the radial multi-entropy projection with radius $\\theta^{\\mathrm{EPO}}_j = \\min\\{\\theta^P_j \\theta^{ME}_j, \\theta^O_j\\}$ converts those weak budgets into strong nodal inequalities, with the convex nondecreasing feasibility profile $\\Phi_j(\\vartheta)$ ensuring that the intersection of per-pair feasible intervals is itself a single interval.","core_discovery":"The central discovery is that relativistic causality turns a potential obstacle into the engine of the method: every directional characteristic speed of the system lies in $[-1,1]$ for any causal equation of state, so the Lax-Friedrichs viscosity can be set to $\\alpha = 1$ uniformly, with no wave-speed estimate and no dependence on the state or equation of state. The resulting two-point average $H_n(U_L,U_R)$ is the average of the self-similar Riemann solution over the light cone; assuming that solution stays admissible and satisfies the canonical entropy inequality, $H_n$ is itself admissible and inherits a two-point entropy inequality for every member of any finite family of convex entropy pairs. A Gauss-Lobatto decomposition writes one forward Euler step of the high-order scheme as a convex combination of nodal states and two such averages, producing one weak entropy budget per pair for the updated cell average. The high-order candidate is then contracted along a single ray anchored at that average: a positivity radius is computed first, then the common feasible interval of all entropy budgets is intersected on the shortened admissible ray, together with an optional oscillation-suppressing interval. Because SSP multistep methods are convex combinations of forward Euler steps, the fully discrete update satisfies local and global discrete entropy inequalities for the entire family while preserving the cell average and the design order.","pith_inferences":["One extension the paper leaves implicit is that the enforced entropy histories could serve as a runtime health check in production codes: because a rise would signal a violated assumption or an implementation error, not merely discretization error.","The authors show that three entropy pairs remove a post-contact overshoot that one pair leaves; a natural next step is to make the number or weighting of entropy pairs adaptive, using the gap between feasible radii as an error indicator.","The uniform alpha=1 flux is more dissipative than a local Rusanov speed in smooth regions, and the paper notes the restrictive CFL condition from the Gauss-Lobatto endpoint weight; sharper compatible fluxes or optimal cell-average decompositions could reduce cost without touching the entropy guarantee."],"forward_implications":["The same conservative update and one projection enforce local and global discrete entropy inequalities for every member of a user-chosen finite entropy family, not just one pair.","The schemes keep the design order (verified for P2 DG at third order) even when smooth solutions approach vacuum within $10^{-5}$, and conservation is exact because the projection anchors at the cell average.","Because the viscosity coefficient $\\alpha=1$ is independent of state and equation of state, the entropy mechanism needs no wave-speed estimate and no EOS-specific tuning beyond primitive recovery.","SSP multistep time integration makes the entropy bound fully discrete, so the boundary-corrected discrete total entropy decreases at every step for each enforced pair.","The authors argue the same construction transfers to any related system with a convex admissible set and a uniform speed bound, such as relativistic magnetohydrodynamics and general relativistic hydrodynamics."],"supporting_citations":[{"why":"Supplies the radial projection geometry, the SSP convex-combination argument, and the Cartesian extension on which the single-projection limiter is built.","marker":"[68]"},{"why":"Constructs the infinite family of convex entropy pairs for the causal EOS whose members the scheme enforces simultaneously.","marker":"[66]"},{"why":"Gives the acoustic-speed formula and causality condition used to prove the uniform light-cone bound and the admissible invariant domain.","marker":"[9]"},{"why":"Characterizes the admissible set in conservative variables and its convexity, which underlies the positivity radius and the feasibility of convex averages.","marker":"[28]"},{"why":"Establishes that SSP multistep methods are convex combinations of forward Euler steps, carrying the entropy budgets to fully discrete time integration.","marker":"[69]"},{"why":"Provides the convex oscillation-suppressing operator whose feasible interval is intersected with the entropy radii.","marker":"[70]"},{"why":"Supplies the safeguarded primitive-recovery iteration whose well-posedness on the admissible set makes entropy evaluation possible only after admissibility is enforced.","marker":"[11]"}],"fun_headline_variants":["Single projection yields discrete multi-entropy stability","One limiter enforces many entropies and physical bounds","Relativistic Euler: fully discrete multi-entropy stability","Causality enables a simple projection for entropy stability"],"cache_read_input_tokens":31872,"weakest_assumption_plain":"The whole construction rests on the standard assumption that the directional Riemann problem has a self-similar solution staying inside the physically admissible set and obeying the canonical entropy inequality, together with a radial-projection geometry taken from a companion preprint rather than proved here.","fun_headline_variants_meta":{"raw":{"variants":["Single projection yields discrete multi-entropy stability","One limiter enforces many entropies and physical bounds","Relativistic Euler: fully discrete multi-entropy stability","Causality enables a simple projection for entropy stability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000216,"raw_usage":{"total_tokens":1529,"prompt_tokens":1139,"completion_tokens":390,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":755,"completion_tokens_details":{"reasoning_tokens":326}},"tokens_in":755,"tokens_out":390,"duration_ms":4344,"temperature":1.0,"reasoning_tokens":326,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:20:17.914222+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the proposed P2 DG scheme with a non-ideal causal EOS on a one-dimensional Riemann problem whose pressure ratio spans twelve orders of magnitude, and at every time step check at every Gauss-Lobatto node that $D>0$ and $E - \\sqrt{D^2 + |m|^2} > 0$ and that each prescribed entropy budget $B_j^{(\\ell)}$ is satisfied; a single violation would refute the universal claim, as would a direct computation of $H_n(U_L,U_R)$ for two admissible states whose exact Riemann solution leaves the admissible set.","supporting_citations":[{"cited_title":"Wu, EPO: A unified framework for entropy stability, positivity, and oscillation sup- pression, arXiv preprint arXiv:2604.00301v1 (2026)","cited_arxiv_id":null,"evidence_quote":"Supplies the radial projection geometry, the SSP convex-combination argument, and the Cartesian extension on which the single-projection limiter is built."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the acoustic-speed formula and causality condition used to prove the uniform light-cone bound and the admissible invariant domain."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Characterizes the admissible set in conservative variables and its convexity, which underlies the positivity radius and the feasibility of convex averages."},{"cited_title":"Gottlieb, D","cited_arxiv_id":null,"evidence_quote":"Establishes that SSP multistep methods are convex combinations of forward Euler steps, carrying the entropy budgets to fully discrete time integration."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the convex oscillation-suppressing operator whose feasible interval is intersected with the entropy radii."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the safeguarded primitive-recovery iteration whose well-posedness on the admissible set makes entropy evaluation possible only after admissibility is enforced."}],"review_version":1}