{"id":"be0c2231-2903-4f25-a08f-92d770fc0ba0","arxiv_id":"2504.14063","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A four-equation compressible multiphase model with temperature-based characteristic reconstruction satisfies the interface equilibrium condition to near machine precision while preserving positivity and conservation.","lead":"A new simulation method keeps pressure, temperature, and velocity from spuriously oscillating at the boundaries between liquids and gases under strong shocks. It makes high-resolution simulations of shock-droplet interactions, underwater explosions, and similar multiphase flows more robust and cheaper.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"EOS closure Eqs. 21-22 is mathematically consistent only if e is specific internal energy, but the text defines e as per-unit-volume, making the printed closure ambiguous and the central IEC results non-reproducible without code.","rationale":"Reader's weakest assumption targets Eq. 22. Our re-derivation shows the algebra in Eqs. 21-22 is correct when e is the specific internal energy; the quadratic is homogeneous in (J kg^-1 K^-1) Pa^2, so a1=Cv does not create a unit mismatch. The real defect is the paper's definition of e as 'internal energy per unit volume' in Section 2.1, which would make E=(e-q)/(v-b) dimensionally inconsistent. This is a genuine reproducibility bug, though not a mathematical falsification of the IEC scheme. The IEC claim itself is supported by the Appendix C constant-characteristic-variable proof and by Table 6, and we found no error in the characteristic matrices or the root selection. The absence of code/data and the ambiguous EOS notation justify a conditional verdict pending a corrected closure and release of reproducible artifacts.","tokens_in":31317,"tokens_out":41638,"duration_ms":387311,"concrete_test":"Implement Eqs. 19-22 in a small script using e as specific internal energy and the Table 1 NASG parameters. Check (i) dimensional consistency of all terms, (ii) the pure-water limit P = [E(CP-Cv) - P_inf CP]/Cv and pure-air limit P = E(CP-Cv)/Cv, and (iii) that the Table 5 water-air droplet initial condition with rho=997 and 1.18 kg/m^3 at T=297 K, P=101325 Pa inverts to exactly these values to round-off. If all three pass, the concern is a notation typo; if any fails, the printed EOS closure is wrong and Table 6 is invalid.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The principal soft spot is the closed-form mixture EOS closure, Eqs. 21-22. Re-deriving from Eqs. 19-20 with e interpreted as specific internal energy shows the printed coefficients are dimensionally consistent: E=(e-q)/(v-b) is a pressure, and the quadratic a1 P^2 - a2 P - a3 has homogeneous units (J kg^-1 K^-1) Pa^2, so the root formula returns Pa. The pure-liquid and pure-gas limits also reduce correctly. However, Section 2.1 explicitly defines e as 'internal energy per unit volume,' which would give E units of Pa kg m^-3 and break Eq. 22. This is an internal inconsistency in the paper text, and since every simulation in Section 4 obtains P and T from this inversion, a reader cannot reproduce the machine-precision IEC results of Table 6 from the paper alone. Appendix C's constant-field proof for the W=[T,Y,u,P] reconstruction is sound under the standard interpretation; the issue is the printed closure, not the IEC argument itself. No code or data are provided, and the HTR solver extension is not released, compounding the reproducibility barrier.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a four-equation model for compressible multi-phase multi-component flows, combined with a primitive-variable characteristic ENO-type reconstruction that is designed to satisfy the interface equilibrium condition (IEC), positivity-preserving limiters, a multi-component extension of the conservative diffuse interface (CDI) regularization, and a phase-confined non-dilute species diffusion model. The central claim is that using W=[T,Y,u,v,w,P] as the reconstruction basis prevents pressure, temperature, and velocity oscillations across isothermal material interfaces to near machine precision without redundant equations. The numerical section includes one- and two-dimensional tests ranging from shock-bubble interaction and Richtmyer-Meshkov instability to droplet advection, shock-droplet interaction, a Mach 100 water column, and a two-layer ICF-like RMI implosion.","tokens_in":31576,"tokens_out":11210,"duration_ms":104659,"significance":"If the claims hold, the main contribution is practically important: a concise, conservative four-equation scheme that satisfies the IEC for general equations of state and mixing rules, avoiding the redundant equations or non-conservative corrections used by many competing methods. The paper provides a mathematical argument in Appendix C, a direct numerical IEC test in Table 6, and validation against experiments and prior computational studies. The CDI and phase-confined diffusion extensions are useful, although they build on prior work from the same group. The main barrier to accepting the central claim is not the IEC construction itself but the presentation of the equation-of-state closure, which is currently internally inconsistent and not reproducible from the text.","major_comments":[{"comment":"The mixture EOS inversion is the load-bearing closure for every simulation in Sec. 4, but it is not reproducible from the text as printed. Section 2.1 defines e as \"internal energy per unit volume,\" whereas Eqs. (9), (18), (19), (22), and the parameters in Table 1 are only dimensionally consistent if e and v are specific (per-unit-mass) quantities. With e per unit volume, r=(e-q)/(v-b) in Eq. (22) is dimensionally meaningless, since q has units J/kg and b has units m^3/kg, and the quadratic in Eq. (21) mixes pressure and temperature dimensions. Re-deriving Eqs. (19)-(20) with e and v specific yields a1=Cv, a2=r(CP-Cv)-Pinf*Cv-Pinf*Y1(CP1-Cv1), and a3=r*Pinf*[CP-Cv-Y1(CP1-Cv1)], so the printed coefficients appear to be the intended ones. Nevertheless, the definitional error and the absence of any derivation leave the reader unable to verify that the implemented inversion matches the paper. Please correct the definition of e and E, state the units of a1, a2, and a3, and include the derivation of Eq. (21)-(22) or a precise reference for it.","section":"Secs. 2.1 and 2.3, Eqs. (19)-(22) and Eq. (39)"},{"comment":"The paper states that the positivity-preserving limiter guarantees admissible states, but the described algorithm only checks whether the high-order HLLC flux is admissible and, if not, replaces it with the first-order flux F_HLLC(U_i,U_{i+1}). No argument is given that this first-order flux is itself positivity-preserving for the four-equation NASG system under the stated CFL condition (Eq. (44)). Without such an argument or a citation to a proof that covers this system, the guarantee is unsubstantiated. The numerical evidence in Sec. 4.2.3 suggests the limiter works in practice, but the claimed guarantee should be proven or explicitly qualified.","section":"Sec. 3.3.2, flux limiter"}],"minor_comments":[{"comment":"The phrase \"near machine precision\" is not uniformly supported: the TENO6 normalized temperature and velocity errors are 3.99e-8 and 1.12e-8, respectively, which are several orders of magnitude above double-precision epsilon. The text partially acknowledges this, but the wording should be qualified to avoid overstating the headline result.","section":"Sec. 4.2.2, Table 6"},{"comment":"The proof that the characteristic variables w5 and w6 are constant relies on evaluating the coefficients such as a*rho at a single face-averaged state and using that same coefficient for every point in the stencil. This is implied by the Godunov algorithm in Sec. 3.1.3 but should be stated explicitly, since the overline notation can be misread as a per-cell average.","section":"Appendix C, Eq. (C.1)"},{"comment":"The units of the co-volume b are listed as kg^{-1}; they should be m^3 kg^{-1}.","section":"Table 1"},{"comment":"The blending coefficients theta_P and theta_T are used to define the positivity-preserving interpolation but are never defined; please state how they are computed.","section":"Sec. 3.3.1, Eqs. (41)-(42)"},{"comment":"The claim that the phase-confined diffusion model prevents leakage of air/helium into SF6 is demonstrated only qualitatively. A quantitative measure, such as the total mass of the confined species across the interface or the maximum SF6 mass fraction inside the bubble, would make the test more convincing.","section":"Sec. 4.3.2, Fig. 15"},{"comment":"\"Nobel-Abel\" should be \"Noble-Abel\" (e.g., in Sec. 2.3 and the references).","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"I believe the unit inconsistency in the EOS closure is likely a typo rather than an error in the method, and the IEC construction itself is plausible and well supported by the numerical IEC test. However, the current text does not allow a reader to reproduce the pressure and temperature inversion, which is used in every simulation, and the positivity-preserving guarantee is asserted rather than proved. A revision that fixes the definition of e, adds the derivation of Eq. (22), and clarifies or proves the first-order flux guarantee would make this a strong JCP paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a real contribution to the four-equation compressible multiphase literature, and the referee time is worth spending. The new piece is the choice of primitive variables W=[T,Y,u,v,w,P] as the characteristic basis. It is simple, it is not in the cited prior work, and the numerical evidence in Table 6 backs it up: for the isothermal droplet advection, pressure/temperature/velocity errors sit around 1e-11 to 1e-8, many orders below the density-based basis in Table 7. I checked Appendix C and the proof is coherent under the stated procedure, where the left/right eigenvectors are frozen at the single face-average state. Under that convention, constant T,P,u indeed produce constant characteristic variables, so the ENO interpolation keeps them constant.\n\nThe CDI extension to multi-component mixtures and the phase-confined diffusion model are also sensible, and the test list is appropriate: shock-bubble vs Haas-Sturtevant, RMI vs Terashima and Adams, shock-droplet centroid vs Colonius and Wong, and the Mach-100 water column as a stress test. The positivity-preserving limiter is described in enough detail to reproduce.\n\nThe soft spot is the EOS closure, and it is exactly where the reader's report points. Section 2.1 defines e as internal energy per unit volume, but Eqs. 19-20 only make dimensional sense if e is specific internal energy (J/kg). As printed, Eq. 22 mixes units; a reader cannot re-derive the mixture pressure and therefore cannot reproduce Table 6 from the text. Table 1 also lists b with units kg^-1, which should be m^3/kg. I think this is a typesetting/definition slip rather than a flaw in the implemented method—the pure-liquid and pure-gas limits reduce correctly under the specific-energy reading—but it is a genuine reproducibility barrier. The paper does not ship code or data, and the HTR extensions are not released, which makes the barrier concrete.\n\nOne minor point: the TENO6 temperature error in Table 6 is 4e-8, which is small but not machine precision; the text says 'near machine precision' for all schemes, then qualifies it. That is fine, but the abstract's phrasing should track the table.\n\nBottom line: the central IEC claim holds up to my reading; the EOS closure needs a clean correction and, ideally, a code release. The paper deserves peer review. If I were the editor, I would send it out and ask the referee to verify the implemented EOS against the corrected text.","headline":"A genuinely useful idea—temperature-based characteristic reconstruction for the four-equation model—with a reproducible-text problem in the EOS closure that should be fixed before publication.","tokens_in":32079,"tokens_out":3740,"would_cite":true,"duration_ms":35786,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A primitive-variable ENO reconstruction satisfies the interface equilibrium condition to near machine precision in the four-equation model, preventing pressure, velocity, and temperature oscillations across material interfaces.","keywords":["four-equation model","interface equilibrium condition","ENO-type schemes","compressible multiphase flow","conservative diffuse interface","positivity preservation","Noble-Abel stiffened gas","multicomponent mixing"],"falsifier":"Compare the pressure and temperature predicted by Eqs. 21-22 with an independent iterative solve of the equilibrium conditions for the water/air parameters of Table 1; any mismatch would show that the printed closure is not the implemented one.","tokens_in":1851,"feed_emoji":"💧","tokens_out":2071,"duration_ms":91320,"temperature":0.7,"pith_summary":"This paper claims that the four-equation multi-phase model, which assumes one shared temperature, pressure, and velocity in each cell, can be discretized so that these quantities remain constant across an isothermal material interface to near machine precision. The key move is to build the high-order ENO-type reconstructions from the primitive variables $W=[T,Y^p_c,u,v,w,P]^T$ rather than from density-based variables. Because temperature, pressure, and velocity are interpolated directly, a uniform state stays uniform across the interface, so the interface equilibrium condition holds to about $10^{-11}$. Around this core, the paper extends conservative diffuse-interface regularization to multi-phase multi-component mixtures, confines species diffusion within each phase, and adds a locally applied positivity-preserving limiter. Validation spans shock-bubble interaction, gas-liquid Riemann problems, shock-droplet interaction, a Mach 100 water column, and an ICF-like two-layer Richtmyer-Meshkov implosion.","feed_headline":"Four-equation model hits machine-precision interface equilibrium","feed_subtitle":"Reconstructing temperature instead of density in ENO interpolation keeps pressure, velocity, and temperature uniform across interfaces.","key_machinery":"The load-bearing object is the primitive reconstruction basis $W=[T,Y^p_c,u,v,w,P]^T$. ENO-type schemes, a family of high-order interpolation and shock-capturing schemes, are applied to characteristic variables derived from this basis and then projected back to physical space; density is recovered afterward from the equation of state, and fluxes are closed with an HLLC Riemann solver. Interpolating temperature rather than density is what enforces the interface equilibrium condition, because a constant $T$, $P$, $u$ field remains constant through the nonlinear ENO interpolation. Supporting mechanisms are the closed-form Noble-Abel stiffened-gas mixture pressure and temperature inversion, the kinetic-energy and entropy-preserving (KEEP) split-form central discretization of the diffuse-interface regularization flux, and the local positivity-preserving limiter that independently limits pressure, temperature, and mass fractions.","core_discovery":"The central claim is that the four-equation model can satisfy the interface equilibrium condition without adding redundant equations. The authors achieve this by using the primitive vector $W=[T,Y^p_c,u,v,w,P]^T$, with one redundant mass fraction omitted, as the basis for characteristic projection and ENO-type interpolation, instead of the density-based basis $[\\rho Y^p_c,u,v,w,P]$ used in five-equation settings. Since a well-formulated scheme preserves constant fields after interpolation, reconstructing $T$, $P$, and $u$ directly enforces the interface equilibrium condition, and the characteristic decomposition in this basis is shown analytically to keep $T$, $P$, and $u$ constant across the interface. Around this construction the paper adds a conservative positivity-preserving limiter applied locally for minimal dissipation, a multi-phase multi-component extension of the conservative diffuse interface model, and a phase-confined species diffusion model. Validation spans shock-bubble interaction, gas-liquid Riemann problems, droplet advection with error tables near machine precision, shock-droplet interaction, a Mach 100 water column, and a two-layer air-helium-SF6 Richtmyer-Meshkov implosion.","pith_inferences":["A natural extension the paper does not pursue is that the same equilibrium-variable reconstruction basis should transfer to other equations of state, such as temperature-dependent specific heats or non-ideal mixing, where the mixture state is still inverted iteratively; the remaining challenge would be an efficient equilibrium solver, not a new oscillation cure.","The phase-confined diffusion construction suggests a clean route to phase change: add interfacial mass-transfer terms that respect the same confinement, rather than letting Fickian fluxes cross the regularized interface.","A testable consequence of the finite-thickness interface regularization is that breakup statistics in high-Weber-number simulations depend on the regularization parameters $\\epsilon$ and $\\Gamma$; results at under-resolved scales should report these parameters to be interpretable.","The Table 6 error statistics provide a simple quantitative benchmark: other four-equation schemes could be compared on the same inviscid droplet-advection test by reporting max normalized $P$, $T$, and $u$ errors after one flow-through time."],"forward_implications":["Four-equation models can satisfy the interface equilibrium condition with normalized pressure, temperature, and velocity errors around $10^{-11}$ for WENO-type schemes, making the minimal four-equation model viable for high-density-ratio compressible flows.","The same numerical framework covers single-phase, multi-component, multi-phase, and multi-phase multi-component regimes without changing equilibrium assumptions or adding redundant PDEs.","High-Mach shock-interface interactions, including a Mach 100 water column, remain stable because the positivity-preserving limiter activates on a tiny fraction of faces and does not visibly degrade the baseline scheme.","Generalized conservative diffuse-interface regularization keeps immiscible interfaces sharp while the phase-confined diffusion model allows intraphase mixing without leakage across phases, enabling multi-layer implosion simulations.","The framework runs high-resolution simulations on GPUs as well as CPUs, so the method scales to production-scale problems."],"supporting_citations":[{"why":"Baseline fully conservative pressure-equilibrium-preserving flux-splitting scheme, restricted to perfect-gas single-phase multi-component flows without shocks; the approach this work generalizes.","marker":"[69]"},{"why":"Recent extension of the pressure-equilibrium-preserving approach to real-gas equations of state with approximate IEC at limited order; the state of the art this method improves on.","marker":"[70]"},{"why":"Density-based primitive reconstruction for five-equation multicomponent WENO; the basis the paper replaces with the temperature-based one because it fails the four-equation IEC.","marker":"[34]"},{"why":"Positivity-preserving flux-limiter framework that this work extends to the four-equation model with independent per-field limiting.","marker":"[36]"},{"why":"KEEP split-form central scheme used for the diffuse-interface regularization flux so that it also satisfies the IEC.","marker":"[79]"},{"why":"Noble-Abel stiffened-gas equation of state that supplies the closed-form mixture pressure-temperature closure in Eqs. 18-22.","marker":"[77]"},{"why":"Iterative equilibrium-state procedure for multicomponent mixtures; the context for the closed-form inversion and the fallback for more complex mixtures.","marker":"[53]"},{"why":"Compressible conservative diffuse-interface method whose boundedness and consistency properties the extended regularization builds on.","marker":"[47]"}],"fun_headline_variants":["Primitive-variable ENO hits exact interface equilibrium","ENO with T-P projection enforces interface constraints","No extra equations for interface equilibrium in four-equation ENO","Positivity-preserving ENO for multiphase flows with exact interfaces","Reconstruct T and P, not density, for balanced phase interfaces"],"cache_read_input_tokens":34304,"weakest_assumption_plain":"The load-bearing premise is that the closed-form mixture pressure-temperature inversion printed in Eqs. 21-22 is exactly what the simulation code evaluates; every pressure and temperature value is reconstructed from conserved variables through this formula.","fun_headline_variants_meta":{"raw":{"variants":["Primitive-variable ENO hits exact interface equilibrium","ENO with T-P projection enforces interface constraints","No extra equations for interface equilibrium in four-equation ENO","Positivity-preserving ENO for multiphase flows with exact interfaces","Reconstruct T and P, not density, for balanced phase interfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001725,"raw_usage":{"total_tokens":6855,"prompt_tokens":1014,"completion_tokens":5841,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":630,"completion_tokens_details":{"reasoning_tokens":5756}},"tokens_in":630,"tokens_out":5841,"duration_ms":35055,"temperature":1.0,"reasoning_tokens":5756,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T11:57:40.135919+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the pressure and temperature predicted by Eqs. 21-22 with an independent iterative solve of the equilibrium conditions for the water/air parameters of Table 1; any mismatch would show that the printed closure is not the implemented one.","supporting_citations":[{"cited_title":"Coralic, T","cited_arxiv_id":null,"evidence_quote":"Density-based primitive reconstruction for five-equation multicomponent WENO; the basis the paper replaces with the temperature-based one because it fails the four-equation IEC."},{"cited_title":"Le M´ etayer, R","cited_arxiv_id":null,"evidence_quote":"Noble-Abel stiffened-gas equation of state that supplies the closed-form mixture pressure-temperature closure in Eqs. 18-22."}],"review_version":1}