{"id":"4dfe663e-af02-4f03-99f3-1565fc3e4ded","arxiv_id":"2505.05383","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Global existence of weak solutions is proved for a new diffuse interface Navier-Stokes/Cahn-Hilliard model with phase transition and for a quasi-stationary Stokes variant.","lead":"It proves that two new mathematical models for two-phase flows with phase transition have global weak solutions. A generalist should care because it gives rigorous foundations for simulations of boiling, evaporation, and other mass-exchange processes in fluid mixtures.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 2.4 is false as stated: the Neumann Laplacian has constants in its kernel, so the very weak pressure bound used for Theorem 1.2 is not justified without a mean-zero restriction.","rationale":"The reader identified the singular free energy assumption (A2) as the weakest point, but (A2) is explicitly stated and standard, and the proof depends on it in a controlled way. My concern is different and more concrete: the paper's preliminary Proposition 2.4 is internally inconsistent. The Neumann Laplace operator on W^{2,p'}_N(Ω) has the constants in its kernel and does not map onto all of L^{p'}(Ω); consequently its adjoint cannot be bijective from L^p(Ω) onto (W^{2,p'}_N(Ω))'. The proposition claims exactly this bijectivity and the estimate (2.10) for every right-hand side. This is not a matter of consensus or an unproven but standard estimate; it is a false statement. The proof of Theorem 1.2 invokes Proposition 2.4 precisely to control the mean-free pressure λ0^N. Without a correct version of that proposition, the pressure bound in Section 6.3 is unsupported. The error is repairable: for the actual application the right-hand side annihilates constants and the pressure is mean-free, so a corrected mean-zero Neumann-Laplace lemma would give the same estimate. Because the repair is straightforward but the manuscript as written contains a false load-bearing lemma, the appropriate verdict is conditional acceptance rather than outright acceptance or rejection. I do not see a comparably central flaw in Theorem 1.1; the velocity compactness argument is terse but follows standard discrete Aubin-Lions reasoning, and the remaining estimates appear consistent.","tokens_in":41383,"tokens_out":60994,"duration_ms":589804,"concrete_test":"Check Proposition 2.4 with f = 0: both u = 1 and u = 0 satisfy (u, Δφ) = 0 for all φ in W^{2,p'}_N(Ω), contradicting uniqueness in L^p(Ω). Then re-derive the pressure estimate in Section 6.3 with the corrected mean-zero statement: verify that the functional f_N satisfies ⟨f_N,1⟩ = 0, and replace (2.10) by the same estimate restricted to u in L^p_(0)(Ω) and f in the annihilator of constants. If the L^2(0,∞; L^r(Ω)) bound on λ0^N follows from this corrected lemma, Theorem 1.2 survives with a revision; if the bound fails, the existence proof for Model II is incomplete.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Proposition 2.4 in Section 2.3 is false as stated. It asserts that for every f in (W^{2,p'}_N(Ω))' there exists a unique u in L^p(Ω) with (u, Δφ) = ⟨f, φ⟩ for all φ in W^{2,p'}_N(Ω), together with estimate (2.10). This is the adjoint version of the preceding claim that Δ_N : W^{2,p'}_N(Ω) → L^{p'}(Ω) is bijective. The Neumann Laplacian is not bijective onto L^{p'}: constants lie in its kernel, so its range is the mean-zero subspace. For f = 0, both u = 1 and u = 0 satisfy (u, Δφ) = 0 for every φ, so uniqueness in L^p fails and (2.10) cannot hold for arbitrary f. This false proposition is load-bearing for Theorem 1.2: in Section 6.3 the pressure λ0^N is bounded in L^2(0,∞; L^r(Ω)) by applying Proposition 2.4 to the very weak equation for βλ0^N. As written, that bound is not justified. The gap is repairable: one should state the mean-zero version, valid for functionals f with ⟨f,1⟩ = 0, which holds for the functional f_N because testing its defining equation with φ = 1 gives ∇φ = 0. But the manuscript neither states nor proves the restricted version, and the bijectivity claim is a genuine mathematical error in the proof of the pressure estimate.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript derives two diffuse interface models for binary fluid flows with unmatched densities and phase transitions, and establishes existence of global weak solutions for both: a quasi-incompressible Navier–Stokes/Cahn–Hilliard system with volume-averaged velocity (Theorem 1.1) and a quasi-stationary Stokes/Cahn–Hilliard system with mass-averaged velocity (Theorem 1.2). The analysis is based on implicit time discretization, a Leray–Schauder fixed-point argument, discrete energy estimates, and compactness arguments, with singular free energies of logarithmic type admitted. Section 3 derives the first model from balance laws and a dissipation inequality, while Section 5 reformulates the Aki–Dreyer–Giesselmann–Kraus model with mass-averaged velocity before the analytical treatment.","tokens_in":41685,"tokens_out":10826,"duration_ms":98207,"significance":"The paper contributes a thermodynamically consistent derivation of a phase-transition extension of the Abels–Garcke–Grün model and provides existence theories for two systems that have not been analyzed in this generality. The proofs are detailed, follow established strategies, and are largely self-contained up to standard subgradient results. The main obstruction to acceptance is the incorrect statement of Proposition 2.4 and its use in the pressure estimate for Theorem 1.2; this is a real but localized flaw that can be repaired by a mean-zero reformulation of the Neumann–Laplace very weak solution theory.","major_comments":[{"comment":"Proposition 2.4 is false as stated. The Neumann Laplacian Δ_N : W^{2,p'}_N(Ω) → L^{p'}(Ω) is not bijective: every constant function lies in its kernel (so the map is not injective), and its range is the mean-zero subspace L^{p'}_{(0)}(Ω), not all of L^{p'}(Ω). Consequently, the adjoint Δ'_N : L^p(Ω) → (W^{2,p'}_N(Ω))' is not injective: for f = 0 every constant function u satisfies (u, Δφ) = 0 for all φ, so the asserted uniqueness in L^p fails and estimate (2.10) is false for arbitrary f. This error is load-bearing: in Section 6.3 the bound ∥λ_0^N∥_{L^2(0,∞;L^r(Ω))} ≤ C∥f_N∥_{L^2(0,∞;(W^{2,r'}_N(Ω))')} is obtained by invoking (2.10), so the pressure estimate as written is unjustified. The repair is simple: reformulate Proposition 2.4 for u ∈ L^p_{(0)}(Ω) and f ∈ (W^{2,p'}_N(Ω))' satisfying ⟨f,1⟩=0 (or, equivalently, work with Δ_N on W^{2,p'}_N ∩ L^{p'}_{(0)}), and verify that the particular functional f_N in (6.8a) satisfies this compatibility condition because Δ1=0. The authors should correct this proposition and the preceding incorrect claim that Δ_N : W^{2,p}_N(Ω) → L^p(Ω) is bijective, and then re-derive the pressure estimate accordingly.","section":"Section 2.3, Proposition 2.4; used in Section 6.3"}],"minor_comments":[{"comment":"The phrase “Naver-slip boundary condition” should read “Navier-slip boundary condition”.","section":"Section 1, paragraph on Model II"},{"comment":"The product space X defined in the Leray–Schauder setup is not a Banach space because D(∂E) and D(∂E_m) are not linear spaces; this does not affect the argument, but the notation could be adjusted to avoid confusion.","section":"Lemma 4.8 and Lemma 6.7"}],"recommendation":"major_revision","confidential_remarks":"The flaw in Proposition 2.4 is genuine but localized; I see no other load-bearing error in the existence proofs. The authors should be asked to correct the proposition and the related bijectivity statement, and to check explicitly that f_N satisfies the mean-zero compatibility condition. After these changes, the paper should be publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nTwo things to know. The paper delivers the first global-in-time weak solutions for two diffuse-interface two-phase-flow models with phase transition and unmatched densities, and both existence proofs follow the established implicit-time-discretization strategy of the Abels school. The second thing is less happy: the very weak pressure bound that Theorem 1.2 depends on rests on a Proposition 2.4 that is false as stated.\n\nWhat is actually new: Model (1.1), a thermodynamically consistent extension of Abels–Garcke–Grün with phase-transition source terms, is derived cleanly from balance laws and an energy dissipation inequality in Section 3. The existence proof for that system (Theorem 1.1) looks solid to me: energy estimates, subgradient machinery for the singular potential, Aubin–Lions compactness, and strong convergence of velocity and phase field are all in place. The quasi-stationary Stokes/Cahn–Hilliard variant from Aki et al. is also analyzed, and the modified chemical potential and Navier-slip setup are handled carefully.\n\nThe soft spot, and it is a real one: Proposition 2.4 claims that the Neumann Laplacian Δ_N maps W^{2,p'}_N bijectively onto L^{p'}. It does not; constants are in the kernel, and the range is the mean-zero subspace. Consequently the very weak solution to the Neumann–Laplace equation is not unique in L^p, and estimate (2.10) cannot hold for arbitrary f. This proposition is exactly what carries the L^2(L^r) bound on λ_0^N in the proof of Theorem 1.2 (Section 6.3). As written, that bound is not justified.\n\nThe fix is straightforward and local: state the mean-zero version (f has ⟨f,1⟩=0, and u is chosen with mean zero), and note that the functional f_N in Section 6.3 does satisfy the compatibility condition because testing the defining equation with φ=1 gives zero on both sides. But the manuscript as it stands contains a genuine mathematical error at a load-bearing point, so the proof of Theorem 1.2 needs revision, not just typo-fixing. The overall strategy is sound, and I would bet the corrected statement goes through unchanged.\n\nWho is this for: anyone working on Navier–Stokes/Cahn–Hilliard with non-matched densities, and especially people who want a usable thermodynamically consistent model with phase transition. It deserves a serious referee; the error is repairable, but the referee should require the fix before publication.","headline":"New models with first existence theorems, and a false Proposition 2.4 that makes the pressure estimate for Theorem 1.2 unjustified as written—repairable, but a real flaw.","tokens_in":42233,"tokens_out":3666,"would_cite":true,"duration_ms":33811,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q30","35Q35","35D30","35G61","76D05","76D03","76T06"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves global existence of weak solutions for two diffuse-interface models of two-phase flow that include phase transitions and unmatched densities, and it derives the first model from continuum thermodynamics.","keywords":["two-phase flow","Navier–Stokes equations","Cahn–Hilliard equation","diffuse interface model","phase transition","weak solutions","singular free energy","quasi-incompressible"],"falsifier":"A concrete test is to implement the implicit time-discrete scheme of Lemma 4.8 for the logarithmic Flory–Huggins potential and check whether the discrete energy inequality (4.5) and the a priori bounds of Lemma 4.6 hold for a range of admissible initial data; if any single admissible datum produces approximate solutions for which the discrete energy grows or for which $\\|F'_0(\\varphi_N)\\|_{L^2}$ becomes unbounded, the compactness construction behind Theorem 1.1 would be invalid.","tokens_in":41175,"feed_emoji":"🌊","tokens_out":10699,"duration_ms":98667,"temperature":0.7,"pith_summary":"The paper proves that two diffuse-interface descriptions of two-phase flow remain solvable when the two fluids have different densities and can exchange mass through phase transitions. The first description, a new quasi-incompressible Navier–Stokes/Cahn–Hilliard system with volume-averaged velocity, is derived from mass and momentum balance together with an energy-dissipation inequality; it generalizes the Abels–Garcke–Grün model by adding phase-transition source terms. The second description is the quasi-stationary Stokes version of the Aki–Dreyer–Giesselmann–Kraus model with mass-averaged velocity, relevant at small Reynolds numbers. The main theorems state that both systems have global weak solutions for all admissible initial data, provided the free energy is singular at the pure-phase values. This matters because phase change is precisely where unmatched densities and mass transfer occur, and few existence results cover such coupled systems.","feed_headline":"Weak solutions proven for phase-changing two-phase flows","feed_subtitle":"Two diffuse-interface models with unmatched densities and singular free energies are shown to have global weak solutions.","key_machinery":"The load-bearing object is the subgradient of the singular free energy $E(\\varphi)=\\int_\\Omega F_0(\\varphi)+\\frac12|\\nabla\\varphi|^2\\,dx$, where $F_0$ is the convex part of the double-well potential $F$ fixed in (A2) (for the mass-averaged model, the same functional is restricted to a fixed spatial mean). Here 'subgradient' is the set-valued analogue of a derivative that remains meaningful when $F'$ blows up at $\\pm1$. Known estimates, stated as (2.4)–(2.5) and (2.8)–(2.9), convert bounds on the subgradient into $H^2$ or $W^{2,r}$ regularity of $\\varphi$ and $L^r$ bounds on $F'_0(\\varphi)$; these are exactly the compactness and limit-passage tools used in the proofs. Around this, the construction uses an implicit time discretization, a Leray–Schauder fixed-point argument for the discrete systems, and a discrete energy-dissipation inequality that yields uniform a priori estimates. For the quasi-stationary model, a damping term $h\\lambda_0$ plus a very weak formulation of the Neumann–Laplace problem controls the pressure, and a measure-theoretic argument identifies the weak limit of $F'(\\varphi_N)$.","core_discovery":"The central claim, stated as Theorems 1.1 and 1.2, is that both models are globally well-posed in the weak sense. For a bounded domain with $C^2$ boundary in dimension two or three, under assumptions (A1)–(A3), the quasi-incompressible Navier–Stokes/Cahn–Hilliard system (1.1) admits a weak solution $(v,\\lambda,\\mu,\\varphi)$ for any initial velocity $v_0\\in L^2(\\Omega)^d$ and phase field $\\varphi_0\\in H^1(\\Omega)$ with $|\\varphi_0|\\le 1$ a.e. and $\\langle\\varphi_0\\rangle\\in(-1,1)$; under (A1)–(A4) the same holds for the quasi-stationary Stokes/Cahn–Hilliard system (1.3) for any such $\\varphi_0$. The solutions satisfy the respective energy inequalities, so the total energy is non-increasing along the flow. In addition, the paper derives model (1.1) from continuum thermodynamics, starting from partial mass balances and momentum conservation and fixing the constitutive laws by requiring a local energy-dissipation inequality to hold. The phase field is governed by a Cahn–Hilliard equation with a source term modeling the phase transition, and the chemical potential is defined through the singular free energy.","pith_inferences":["Because the discrete energy inequality is uniform in the time step, the proof is a natural starting point for designing and analyzing energy-stable finite-element schemes for phase-change flows; the paper does not take that numerical step.","The pressure-control strategy used for model II—damping term plus very weak Neumann–Laplace solution—looks transferable to other quasi-incompressible diffuse-interface systems in which the pressure enters the chemical potential equation.","One route beyond the paper is to study the singular limit in which the quasi-stationary model (1.3) emerges from the instationary model (1.1) as the Reynolds number tends to zero; the paper develops both models but does not connect them by such a limit.","The assumption that the mobilities are bounded away from zero (A3) excludes degenerate mobilities that vanish in the pure phases; extending the two theorems to degenerate mobility would need new estimates near $\\varphi=\\pm1$."],"forward_implications":["Every admissible initial datum with $\\langle\\varphi_0\\rangle\\in(-1,1)$ starts a global weak solution of the volume-averaged system (1.1), and likewise of the quasi-stationary system (1.3) under the slightly stronger assumptions.","Because the energy inequality holds for the weak solutions, the total energy is non-increasing and the dissipation rate has the explicit form $\\int_\\Omega S(\\varphi,Dv):Dv + m_j(\\varphi)|\\nabla\\mu|^2 + m_r(\\varphi)(\\mu+\\alpha\\lambda)^2\\,dx$ for model I, and the analogous expression with $c_+^2$ and the boundary friction term for model II.","The singular free energy is not merely an admissible choice: it enforces $|\\varphi|\\le1$ throughout the evolution, which is physically the statement that the phase field stays within the pure-phase interval.","Stationary solutions of both systems are characterized by Cahn–Hilliard equilibria: in model I they solve $F'(\\varphi_*)-\\Delta\\varphi_*=\\mu_*$ with constant $\\mu_*,\\lambda_*$, while in model II they solve the same equation with $-\\alpha\\lambda_*$ on the right-hand side.","The generalization of the Abels–Garcke–Grün model derived in Section 3 reduces to the original model when the reaction rates vanish, so the new result contains the earlier existence theory as a special case."],"supporting_citations":[{"why":"The Abels–Garcke–Grün model that model I generalizes; supplies the original energy structure and the framework the derivation extends.","marker":"[9]"},{"why":"The Aki–Dreyer–Giesselmann–Kraus model whose quasi-stationary version is model II; supplies the mass-averaged velocity formulation and base system.","marker":"[12]"},{"why":"Supplies the subgradient estimates (2.4)–(2.5) for singular free energies with prescribed mean, used in Lemma 6.5 for model II.","marker":"[11]"},{"why":"Supplies the subgradient estimates (2.8)–(2.9) without prescribed mean used in Lemma 4.6, plus the Korn-type inequalities used in both proofs.","marker":"[1]"},{"why":"Provides the implicit time discretization and Leray–Schauder fixed-point strategy that the present proofs follow.","marker":"[6]"},{"why":"Supplies the damping-term device and the very weak Neumann–Laplace argument used to control the pressure in model II.","marker":"[2]"},{"why":"Supplies the measure-theoretic argument (Egorov/Chebyshev) used to identify the weak limit of $F'(\\varphi_N)$ in Theorem 1.2.","marker":"[5]"}],"fun_headline_variants":["Existence of weak solutions for phase-change two-fluid systems","Diffuse interface models with phase transition: weak solvability","Global weak solutions for two-phase flows with mass transfer","Phase transition in two-phase flow: mathematical existence proof","Weak well-posedness for two phase-transition diffuse models"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire proof leans on the free energy being singular at the pure-phase values $\\pm1$, so that its derivative blows up there and the subgradient estimates yield enough regularity of $\\varphi$ and $F'(\\varphi)$; with a smooth double-well potential the arguments as written would break down.","fun_headline_variants_meta":{"raw":{"variants":["Existence of weak solutions for phase-change two-fluid systems","Diffuse interface models with phase transition: weak solvability","Global weak solutions for two-phase flows with mass transfer","Phase transition in two-phase flow: mathematical existence proof","Weak well-posedness for two phase-transition diffuse models"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000317,"raw_usage":{"total_tokens":1817,"prompt_tokens":992,"completion_tokens":825,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":744}},"tokens_in":608,"tokens_out":825,"duration_ms":8449,"temperature":1.0,"reasoning_tokens":744,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:05:19.664359+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete test is to implement the implicit time-discrete scheme of Lemma 4.8 for the logarithmic Flory–Huggins potential and check whether the discrete energy inequality (4.5) and the a priori bounds of Lemma 4.6 hold for a range of admissible initial data; if any single admissible datum produces approximate solutions for which the discrete energy grows or for which $\\|F'_0(\\varphi_N)\\|_{L^2}$ becomes unbounded, the compactness construction behind Theorem 1.1 would be invalid.","supporting_citations":[{"cited_title":"Abels, H","cited_arxiv_id":null,"evidence_quote":"The Abels–Garcke–Grün model that model I generalizes; supplies the original energy structure and the framework the derivation extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The Aki–Dreyer–Giesselmann–Kraus model whose quasi-stationary version is model II; supplies the mass-averaged velocity formulation and base system."},{"cited_title":"Abels and M","cited_arxiv_id":null,"evidence_quote":"Supplies the subgradient estimates (2.4)–(2.5) for singular free energies with prescribed mean, used in Lemma 6.5 for model II."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the subgradient estimates (2.8)–(2.9) without prescribed mean used in Lemma 4.6, plus the Korn-type inequalities used in both proofs."},{"cited_title":"Abels, D","cited_arxiv_id":null,"evidence_quote":"Provides the implicit time discretization and Leray–Schauder fixed-point strategy that the present proofs follow."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the damping-term device and the very weak Neumann–Laplace argument used to control the pressure in model II."},{"cited_title":"Abels, S","cited_arxiv_id":null,"evidence_quote":"Supplies the measure-theoretic argument (Egorov/Chebyshev) used to identify the weak limit of $F'(\\varphi_N)$ in Theorem 1.2."}],"review_version":1}