{"id":"6176ab47-0174-4f47-9b6c-997f0cd5f5bf","arxiv_id":"2505.17535","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Equilibrium boundary conditions for monotone lattice Boltzmann schemes are shown to converge to the Bardos-Leroux-Nedelec entropy solution in the scalar multidimensional case.","lead":"Equilibrium-based boundary conditions for lattice Boltzmann schemes are proven to converge, in the scalar multidimensional case, to the correct weak entropy solution; the same treatment is tested on the Euler equations. The result gives numerical analysts a rigorous justification for a standard boundary trick used in fluid simulations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Bounded-domain convergence claim rests on an unproved 'analogous' extension from the quarter plane; corner and east/north wall estimates are missing.","rationale":"The reader's weakest assumption identifies the quarter-plane to bounded-domain extension, and my reading confirms that this is the most load-bearing issue. The internal estimates in the quarter-plane proof are largely written out, and the use of the companion preprint for monotonicity is a reasonable dependency rather than a gap in the argument itself. By contrast, the bounded-domain theorem as stated in the abstract and in Definition 1 simply does not follow from Theorem 3 without the missing Section 2.6 proof. The concern is serious but addressable: a full four-wall version with corner estimates would settle it. Hence the reader's CONDITIONAL verdict remains appropriate, and no change to the verdict is needed.","tokens_in":39381,"tokens_out":9226,"duration_ms":93229,"concrete_test":"Prove the bounded-domain analogue of Propositions 7-12 on (0,1)^2 with four walls. In particular, re-derive the total-variation recurrence (27) at the eastern and northern walls, check the signs of the incoming ghost equilibrium terms f6eq(ũ_E) and f5eq(ũ_N), and verify that the four corner contributions in the entropy inequality, including the north-east analogue of (35), are O(Δx) with constants depending only on u∞, the BV norms of the four boundary data, and T. If any corner term fails to vanish in the limit, Theorem 4 does not imply Definition 1; if all estimates go through, the gap is expositional and repairable.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorems 3 and 4 are proved on the quarter plane (R*_+)^2, but the central claim concerns Definition 1, which is a Bardos-Leroux-Nedelec entropy formulation on the bounded square (0,1)^2 with four boundary segments. Theorem 3 states L1 convergence on (R*_+)^2, and Theorem 4 says 'under the same assumptions' the limit satisfies Definition 1. The only bridge is Section 2.6's assertion that 'analogous properties and estimates hold for the numerical scheme on (0,1)^2,' with no proof supplied. The quarter-plane argument uses only the western and southern walls. Passing to the square requires: (i) L1, equicontinuity, and total-variation estimates with data on all four walls; (ii) trace inequalities on the eastern and northern walls, where the BLN boundary terms enter with opposite signs; and (iii) control of the four corner cells in the entropy proof, e.g., the north-east analogue of equation (35), where two boundary equilibria stream into the same cell. These are nontrivial estimates and none is written out. Because the advertised first convergence proof for lattice Boltzmann schemes with non-periodic boundaries is exactly the bounded-domain statement, this is a load-bearing gap.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes equilibrium-based boundary conditions for vectorial multi-dimensional lattice Boltzmann schemes and studies their behavior. In the scalar case, for monotone two-relaxation-times D2Q5 schemes satisfying condition (17), the authors prove on the quarter plane (R_+*)^2 that the numerical solution is L1-stable, equicontinuous in time, uniformly bounded in total variation, and within O(Delta x) of the equilibrium manifold. Combining these estimates in the Crandall-Majda framework, they obtain convergence up to subsequence in L∞_t L1_x to a limit u that is at equilibrium, and they claim in Theorem 4 that this limit satisfies the Bardos-Leroux-Nédelec weak entropy inequality stated in Definition 1 on the bounded square (0,1)^2. The paper also contains numerical experiments for scalar problems and for the Euler equations, including a double Mach-10 reflection.","tokens_in":39671,"tokens_out":7984,"duration_ms":65864,"significance":"If the proof were complete, the paper would provide the first convergence result for lattice Boltzmann schemes with non-periodic boundary conditions for weak solutions of hyperbolic conservation laws. The quarter-plane estimates are written out in detail, the '♭-procedure' is a clean device for bounding sums of equilibrium differences, and the numerical section demonstrates that the proposed boundary conditions work on demanding test cases. However, the advertised central claim is the bounded-domain statement of Theorem 4, and that statement is not actually proved: the paper proves its estimates only on the quarter plane with two boundaries, and the entropy proof applies Proposition 8 to equilibrium states that may lie outside the allowed box K. These gaps are load-bearing and must be fixed before the main result can be accepted.","major_comments":[{"comment":"The convergence and entropy theorems are proved on the quarter plane (R_+*)^2 with only western and southern boundary conditions, while the paper's central claim is convergence to the Bardos-Leroux-Nédelec weak entropy solution of Definition 1 on the bounded square (0,1)^2 with four boundary segments. Section 2.6 merely asserts that 'analogous properties and estimates hold for the numerical scheme on (0,1)^2' without any proof. Passing to the square requires additional L1, equicontinuity, and total-variation estimates with data on all four walls, trace inequalities on the eastern and northern walls where the entropy boundary terms carry opposite signs, and control of the four corner cells in the entropy proof (for instance, the north-east analogue of Eq. (35), where two boundary equilibria stream into the same cell). None of these estimates are written out, so Theorem 4 as stated is unsupported.","section":"Section 2.6 and Theorems 3-4"},{"comment":"The entropy inequality is claimed for all κ ∈ R, but the proof applies Proposition 8 to the tuple f_eq(κ). Proposition 8 is stated only for arguments in K = ∏_p [f_eq_p(-u∞), f_eq_p(u∞)], and by the monotonicity of the equilibria (Proposition 6), f_eq(κ) is not in K when |κ| > u∞. The proof does not justify the ℓ1-contractivity inequality for such κ, nor does it explain how the full BLN inequality for all κ follows from the range κ ∈ [-u∞,u∞]. The entropy proof is therefore incomplete even on the quarter plane.","section":"Section 2.11, proof of Theorem 4"}],"minor_comments":[{"comment":"In the display after Eq. (28), the term |f^{n,*}_{6,1,j_y} - f^{eq}_i(tilde u^n_{,j_y})| should read |f^{n,*}_{6,1,j_y} - f^{eq}_6(tilde u^n_{,j_y})|; the subscript i is undefined in that term.","section":"Eq. (28)"},{"comment":"Definition 1 requires the limit u to belong to BV((0,T) × (0,1)^2), but Theorem 3 only states L∞_t L1_x convergence and pointwise boundedness. The paper should explicitly state how the uniform BV_x bounds and L1-time continuity imply the space-time BV regularity needed to apply Definition 1.","section":"Theorem 3 and Definition 1"},{"comment":"The phrase 'analogous properties and estimates hold' is too cursory for a statement that underpins the main theorem. Even if a full proof is deferred, the authors should outline the necessary changes for the four-wall problem, particularly the signs of the boundary terms and the handling of corners.","section":"Section 2.6"}],"recommendation":"major_revision","confidential_remarks":"The main result is potentially significant, but the gap between the quarter-plane proofs and the bounded-square claim is central, and the entropy proof's use of Proposition 8 for arbitrary κ is a real technical issue. Both are fixable in principle, but they require substantial additional work rather than minor edits."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dan,\n\nRead this one carefully because it's a serious analysis paper with a genuine new result and one load-bearing gap. The new thing is the scalar multidimensional convergence of monotone lattice Boltzmann schemes with equilibrium boundary conditions to the BLN entropy solution — the first convergence proof for LB with non-periodic boundaries in a weak-solution setting. The BV, equicontinuity, and equilibrium-closeness estimates are mostly written out and look right, and the boundary-layer analysis for the linear D1Q2 scheme is a nice extra: explicit formulas, GKS-style, actually useful.\n\nThe problem is the bounded-domain theorem. Section 2.6 says the proof is done on the quarter plane for simplicity and that analogous properties hold on (0,1)^2. Theorems 3 and 4 are stated for the quarter plane, but Theorem 4 says the limit satisfies Definition 1, which lives on the bounded square. That's not a harmless notational shift. Getting the east and north walls into the trace estimates and the entropy inequality requires handling flipped signs in the BLN boundary terms and corners where two boundary equilibria stream into one cell. The stress-test note is right: this is the part that makes the advertised result the first convergence proof with boundaries, and it's missing.\n\nThere's also a smaller, fixable gap in the entropy proof. Proposition 8 (ℓ1-contractivity) needs both arguments in the invariant box K. The proof applies it to f_eq(κ) for arbitrary κ ∈ R, but for |κ| > u∞ that state is outside K. The standard truncation argument would fix it, but it isn't there.\n\nThe heavy reliance on arXiv:2501.07934 for the monotonicity and contractivity engine is legitimate, not circular — those results are cited, not re-derived — but it does mean the paper isn't self-contained until that preprint is vetted.\n\nNet: worth refereeing, and the referee should condition acceptance on either proving the square extension or honestly restating the theorem for the quarter plane, plus fixing the κ-range issue. The core ideas are solid and worth engaging with.","headline":"Real first convergence proof for LB with boundaries, but the bounded-domain version is asserted, not proved, and the entropy proof has a κ-range gap.","tokens_in":40189,"tokens_out":4367,"would_cite":true,"duration_ms":46587,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M12","76M28"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that filling ghost cells with equilibrium states makes monotone lattice Boltzmann schemes converge to the weak entropy solution of scalar hyperbolic conservation laws on bounded multi-dimensional domains.","keywords":["lattice Boltzmann schemes","equilibrium boundary conditions","hyperbolic conservation laws","weak entropy solution","monotone schemes","two-relaxation-times","convergence analysis","boundary layers"],"falsifier":"Run the monotone D2Q5 scheme on (0,1)^2 with zero initial data and incompatible boundary data on the west and south walls, such as a constant value 1 on the west and a constant value -1 on the south, and check whether the total variation of the reconstructed distribution functions at a fixed time remains bounded independently of the grid size as Δx → 0; if it grows without bound, or if the limiting solution violates the boundary entropy inequality near the corner, the bounded-domain form of the theorem fails.","tokens_in":39188,"feed_emoji":"🌊","tokens_out":5324,"duration_ms":66518,"temperature":0.7,"pith_summary":"This paper establishes that lattice Boltzmann schemes can be given reliable boundary conditions by filling ghost cells with equilibrium states computed from the prescribed macroscopic boundary data. In the scalar multi-dimensional case, under a monotonicity condition, the numerical solution is shown to converge, up to a subsequence, to the weak entropy solution of the hyperbolic conservation law with boundary conditions. This matters because it is the first convergence proof for lattice Boltzmann schemes with non-periodic boundaries for weak solutions of hyperbolic problems. The proof works because the discrete solution stays uniformly bounded in L∞ and in total variation, is equicontinuous in time, and remains within O(Δx) of the equilibrium manifold, so compactness arguments apply and the limit satisfies the correct entropy inequality.","feed_headline":"How lattice Boltzmann schemes can handle walls and still converge","feed_subtitle":"Equilibrium ghost cells give the first convergence proof for lattice Boltzmann methods with non-periodic boundaries in hyperbolic problems.","key_machinery":"The machinery is the equilibrium bridge f_i^eq(u) between macroscopic data and mesoscopic distribution functions, together with the two-relaxation-times collision operator and the monotonicity condition (17). The consistency of the equilibria with the fluxes, expressed in (9), is what allows boundary data to be injected through f_i^eq without breaking conservation. The proof combines a maximum principle, ℓ1-contractivity of relaxation, L1 and total variation estimates, equicontinuity, and a geometric relaxation estimate showing the solution stays within O(Δx) of equilibrium; the entropy argument then uses Krushkov-type entropies defined as |f_i - f_i^eq(κ)| and a boundary trace inequality that follows from the monotonicity and consistency of the equilibria.","core_discovery":"The central claim is that equilibrium boundary conditions, in which each incoming distribution just outside the domain is set to the equilibrium evaluated at the prescribed boundary trace, make monotone two-relaxation-times D2Q5 lattice Boltzmann schemes convergent for scalar hyperbolic conservation laws on multi-dimensional domains. Under condition (17), the scheme satisfies a maximum principle, is ℓ1-contractive in relaxation, admits L1 and total variation bounds controlled by the data, is equicontinuous in time, and its distribution functions stay within O(Δx) of equilibrium. These ingredients yield a subsequence converging in L∞-in-time and L1-in-space to a limit u that is at equilibrium and satisfies the weak entropy inequality with boundary trace terms. The convergence holds regardless of whether the boundary is an inflow or an outflow, and simpler stencils and relaxation models are included as limiting cases.","pith_inferences":["If the asserted extension to the finite square with corners is completed, the same compactness framework should yield convergence with all four walls active, provided a corner-aware trace inequality controls the two incoming directions simultaneously.","The linear boundary-layer analysis for wrong outflow traces suggests that, in smooth-data regimes, the Lp boundary-layer error scales as O(Δx^{1/p}), which would dominate for p > 1; this sharp rate could be tested experimentally for nonlinear scalar problems.","The numerical success for the Euler system hints that equilibrium boundary conditions may also be convergent for systems under analogous spectral or entropy-stability conditions, but the proof in the vectorial case remains open.","Because the equilibrium bridge depends only on macroscopic data, the same ghost-cell closure could be applied to other kinetic or relaxation schemes, with monotonicity-like constraints playing the role of the sub-characteristic condition."],"forward_implications":["Lattice Boltzmann schemes with equilibrium ghost-cell boundary conditions are convergent solvers for scalar hyperbolic conservation laws on bounded domains, so this classical 'wet-node' approach gains a rigorous theoretical foundation.","The boundary treatment does not need to detect inflow versus outflow: stability and convergence hold for both, matching the fact that the macroscopic PDE only requires data along incoming characteristics.","The monotonicity conditions derived here give explicit parameter constraints on relaxation rates and lattice coefficients that guarantee safe boundary behavior; violating them is linked to boundary layers or oscillations.","The proof covers D1Q2, D1Q3, D2Q4, and BGK variants as limiting cases, so the result transfers to a family of practical schemes by reducing the stencil or setting relaxation parameters equal.","The O(Δx) closeness to equilibrium supports using equilibria for initialization as well as for boundary data, and the entropy dissipation near boundaries is inherited from the bulk relaxation property."],"supporting_citations":[{"why":"Defines the weak entropy solution with boundary trace terms that the numerical scheme is proved to converge to.","marker":"[Bardos et al., 1979]"},{"why":"Supplies the monotone-approximation compactness path that the proof follows to extract a convergent subsequence from L1 ∩ BV bounds.","marker":"[Crandall and Majda, 1980]"},{"why":"Provides the prior relaxation-scheme convergence and the boundary lemma that the entropy argument adapts to two dimensions.","marker":"[Aregba-Driollet and Milišić, 2004]"},{"why":"Establishes the whole-space monotonicity, ℓ1-contractivity, and closeness-to-equilibrium estimates that are extended here to domains with boundaries.","marker":"[Aregba-Driollet and Bellotti, 2025]"},{"why":"Motivates the stability-style analysis of Dirichlet boundary conditions for hyperbolic finite-difference schemes, which the authors adapt to the nonlinear lattice Boltzmann setting.","marker":"[Goldberg and Tadmor, 1981]"},{"why":"Provides the numerical boundary-layer framework used in the linear D1Q2 analysis of wrong outflow traces.","marker":"[Boutin and Coulombel, 2017]"}],"fun_headline_variants":["First convergence proof for lattice Boltzmann with non-periodic boundaries","Equilibrium boundary conditions prove convergent for lattice Boltzmann schemes","Lattice Boltzmann gets a rigorous convergence guarantee at boundaries","Set ghost cells to equilibrium for a new lattice Boltzmann convergence proof"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The theorem is proved on an unbounded quarter-plane with only two boundary walls, while the finite square with four walls and corners is handled by an unproved assertion in Section 2.6 that the same estimates carry over.","fun_headline_variants_meta":{"raw":{"variants":["First convergence proof for lattice Boltzmann with non-periodic boundaries","Equilibrium boundary conditions prove convergent for lattice Boltzmann schemes","Lattice Boltzmann gets a rigorous convergence guarantee at boundaries","Set ghost cells to equilibrium for a new lattice Boltzmann convergence proof"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000773,"raw_usage":{"total_tokens":3368,"prompt_tokens":841,"completion_tokens":2527,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":457,"completion_tokens_details":{"reasoning_tokens":2461}},"tokens_in":457,"tokens_out":2527,"duration_ms":12166,"temperature":1.0,"reasoning_tokens":2461,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:45:08.417182+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the monotone D2Q5 scheme on (0,1)^2 with zero initial data and incompatible boundary data on the west and south walls, such as a constant value 1 on the west and a constant value -1 on the south, and check whether the total variation of the reconstructed distribution functions at a fixed time remains bounded independently of the grid size as Δx → 0; if it grows without bound, or if the limiting solution violates the boundary entropy inequality near the corner, the bounded-domain form of the theorem fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the weak entropy solution with boundary trace terms that the numerical scheme is proved to converge to."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the monotone-approximation compactness path that the proof follows to extract a convergent subsequence from L1 ∩ BV bounds."},{"cited_title":"and Mili s i \\'c , V","cited_arxiv_id":null,"evidence_quote":"Provides the prior relaxation-scheme convergence and the boundary lemma that the entropy argument adapts to two dimensions."},{"cited_title":"Monotonicity and convergence of two-relaxation-times lattice Boltzmann schemes for a non-linear conservation law","cited_arxiv_id":"2501.07934","evidence_quote":"Establishes the whole-space monotonicity, ℓ1-contractivity, and closeness-to-equilibrium estimates that are extended here to domains with boundaries."},{"cited_title":"and Tadmor, E","cited_arxiv_id":null,"evidence_quote":"Motivates the stability-style analysis of Dirichlet boundary conditions for hyperbolic finite-difference schemes, which the authors adapt to the nonlinear lattice Boltzmann setting."},{"cited_title":"and Coulombel, J.-F","cited_arxiv_id":null,"evidence_quote":"Provides the numerical boundary-layer framework used in the linear D1Q2 analysis of wrong outflow traces."}],"review_version":1}