{"id":"7c3808df-c52f-4f2e-a41c-d09c718185aa","arxiv_id":"2504.13549","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A 1D adaptive integer lattice gas with velocity-dependent collision fractions reproduces lattice Boltzmann equilibrium and cosine-wave dynamics, and can be encoded in log(N)+3 qubits with a linear collision operator.","lead":"This paper designs a new grid-based fluid simulation method that adjusts its collision rules locally to mimic a standard technique called lattice Boltzmann, and also gives a quantum circuit version. It matters because it is a concrete attempt to make fluid dynamics algorithms more quantum-friendly, though the current quantum version still has to read out the whole grid after every step.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Adaptive ALGA's hydrodynamic limit is never derived: λc varies in space and time, so the LBM-equivalence claim rests only on equilibrium matching and one cosine benchmark.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing gap: the adaptive model's hydrodynamic limit is not derived. The paper has real independent support: Eq. 17 follows from the equilibrium condition with D1Q3 weights w±=1/6, the quantum encoding in Eq. 18 is a nontrivial construction, and the cosine-wave comparison provides one empirical point. But the central claim of LBM equivalence requires more than matching equilibria; it requires that the full space-time-dependent relaxation operator, with λc varying through Eq. 17, produces the same low-Mach Navier-Stokes behavior as LBM. That is not shown. The constant-parameter Chapman-Enskog result in Appendix A does not apply, and the paper's own numerical section explicitly leaves the τ–λs relation unproven. These issues justify the reader's CONDITIONAL verdict, but do not move it: the underlying construction is coherent and testable, and the missing derivation could in principle be supplied. Therefore no change to the reader's verdict is recommended.","tokens_in":15566,"tokens_out":12120,"duration_ms":113528,"concrete_test":"Perform a second-order Chapman-Enskog expansion for the adaptive collision term (Eq. 10) with λc given by Eq. 17 as a function of local conserved variables, linearizing around the LBM equilibrium geq. Compare the resulting viscous stress and dispersion relation to LBM BGK with τ=1. Complement this analytically by running the adaptive ALGA on a small-amplitude monochromatic sine wave on a 512-site periodic lattice with λs=0.2, measuring the amplitude decay rate for several wavenumbers k, and inferring ν(k). If ν is not k-independent and equal to the LBM value ν=1/6, or if the decay is not exponential, the agreement in Fig. 6 is not evidence of general equivalence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim that ALGA achieves the same simulation results as LBM is made to rest on the statement in Sec. II B that reproducing LBM equilibrium distributions should be sufficient for reproducing LBM macroscopic dynamics. That inference is not backed by a derivation for the adaptive case. The only Chapman-Enskog analysis in Appendix A treats λc and λs as constants and yields a first-order momentum equation containing ∂|ρu|/∂x (Eq. A13), not the LBM convective term. In the adaptive algorithm, λc(x,t) from Eq. 17 depends on the local momentum u(x,t), so the collision operator's relaxation rate varies in space and time. The effective viscosity and all higher-order Knudsen terms are therefore uncontrolled, and matching the equilibrium distribution alone does not fix them. The paper itself concedes in Sec. V D that no analytical τ–λs relation was obtained and that the comparison with τ=1.5 'did not prove any confirmation.' The single cosine-wave benchmark, without error quantification or variation of Reynolds number and wavelength, is the only evidence that the unmatched terms are negligible. This is a missing derivation rather than a disagreement with consensus, and it directly affects the main claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces an adaptive integer lattice gas algorithm (ALGA) for the one-dimensional D1Q3 model. The collision step applies local fractions λ_s and λ_c of the maximum possible splittings and crunches; λ_c is adjusted at each cell via Eq. (17) so that the equilibrium populations match the LBM equilibrium (Eq. 9) in the low-velocity limit. The authors claim that ALGA 'achieves the same simulation results of LBM' in this limit, and support this with equilibrium-distribution measurements and a cosine-wave shock-formation comparison. They then present a quantum encoding (Eq. 18) in which the collision operator is represented as a fixed linear operator (Eq. 19), implemented via SVD and LCU, with streaming by quantum shifts. The per-step circuit cost is stated as O(log² N) using log(N)+3 qubits, but the full evolution still requires measurement and reinitialization, giving total cost Ω(NT). The paper explicitly leaves several issues open, including an analytical τ–λ_s relation and the possibility of time-step concatenation.","tokens_in":15800,"tokens_out":8678,"duration_ms":72942,"significance":"If the central claim is established, ALGA would be a deterministic, quantum-friendly lattice gas that avoids the random sampling of MCLGA while reproducing low-Mach LBM behavior, with a linear collision operator suitable for quantum circuits. The paper is honest about its current limitations: the quantum implementation does not achieve an end-to-end advantage because of tomography and reinitialization, and the authors concede that no analytical viscosity relation is obtained. The main value lies in proposing a concrete construction that bridges ILGA and LBM and in demonstrating numerically that the local-adaptation idea can match LBM for a benchmark. However, the claim of 'same simulation results as LBM' is not yet backed by a hydrodynamic derivation for the adaptive case, and the only external validation is a single cosine-wave test. The paper would be a useful contribution if the missing analysis is supplied.","major_comments":[{"comment":"The Chapman-Enskog expansion in Appendix A is performed only for constant λ_c and λ_s. It yields Eq. A13, which contains a term λ_c/(λ_c+λ_s) ∂|ρu|/∂x rather than the LBM convective term. In the adaptive algorithm, λ_c(x,t) from Eq. 17 depends on the local momentum, so the relaxation rate varies in space and time and the effective viscosity and higher-order Knudsen terms are uncontrolled. The paper's central claim that ALGA 'achieves the same simulation results of LBM' therefore rests on the assertion in Sec. II B that matching equilibrium distributions suffices, but no derivation or quantitative error bound is given for the adaptive case. This missing derivation is load-bearing and directly affects the abstract's main claim.","section":"Appendix A, Eq. A13"},{"comment":"The equilibrium-distribution agreement in Sec. V C is partially a verification of the construction rather than an independent prediction: λ_c is obtained by substituting the LBM equilibrium g_i^eq into the equilibrium condition Eq. 16, so agreement with Eq. 9 in the |u|≈0 region is enforced by design. The only genuinely external test is the cosine-wave dynamics in Sec. V D. The paper should state this circularity explicitly and distinguish the constructed equilibrium match from an independent validation of the hydrodynamic behavior.","section":"Sec. V C, Eq. 17"},{"comment":"The cosine-wave comparison is the sole evidence for the adaptive case, but it is not quantified: no error metric, convergence study, or variation of Reynolds number and wavelength is reported, and the simulation is performed for a single initial condition and a single relaxation time (τ=1). The paper itself states that the numerical analysis of the spatially averaged absolute difference 'did not prove any confirmation' for the τ=1.5 hypothesis and that no analytical τ–λ_s relation was obtained. As a result, the claim that ALGA reproduces LBM behavior in general is not established beyond the single benchmark shown; this is a limitation of the central claim, not a minor presentation issue.","section":"Sec. V D"},{"comment":"The D1Q3 weights and the equilibrium expansion in Eq. 9 are not the standard LBM equilibrium. The text states w_- = w_+ = 1/3, whereas with lattice velocities ±1 the standard D1Q3 weights are w_± = 1/6, and Eq. 9 as written lacks the velocity-dependent quadratic term and the −ρu²/(2c_s²) correction of the standard low-Mach expansion. Since Eq. 17 is derived from Eq. 9, the authors should specify the velocity rescaling and weight convention used, and verify that the resulting λ_c formula and the numerical fits are consistent with the LBM implementation they compare against.","section":"Sec. II C, Eq. 9"}],"minor_comments":[{"comment":"The notation n1 and n2 is used for the populations appearing in the quantum encoding, but the paper elsewhere denotes the populations as n_-, n0, n_+. Please define n1 and n2 explicitly to avoid ambiguity.","section":"Sec. IV, Eq. 18"},{"comment":"The condition '|u(x,t)| < 3' appears to be a typo; the bound for λ_c to remain physical is stated in Sec. III A as |u| < √(2/3) (up to the λ_s dependence noted below). Please correct the threshold.","section":"Sec. V D"},{"comment":"The statement that 0 < λ_c < 1 is 'in general' ensured if |u| < √(2/3) is not a complete condition, because the upper bound λ_c < 1 also depends on λ_s through Eq. 17. Please give the precise inequality for λ_c in the allowed range.","section":"Sec. III A"},{"comment":"The notation 'v_i u' in Eq. 9 is undefined; v_i is not introduced, and it is unclear whether it denotes the lattice velocity c_i, a rescaled velocity, or something else.","section":"Sec. II C, Eq. 9"},{"comment":"There is a typo 'algorihtm' in the paragraph after Eq. 19; the word should be 'algorithm'.","section":"Sec. IV"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses an interesting and timely topic, and the authors are transparent about the limitations of the current quantum implementation. The main obstacle to publication is the missing hydrodynamic derivation for the adaptive case: the central claim 'same simulation results as LBM' is supported only by an equilibrium match that is partly built into the construction and by a single cosine-wave benchmark. This is fixable with additional analysis or a more thorough numerical study, so I recommend major revision rather than rejection. I also suggest that the authors clarify the nonstandard equilibrium expansion in Eq. 9, as it affects the derivation of Eq. 17."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a real step forward in the MCLGA-to-LBM program, with a genuinely new deterministic collision adaptation and a clean quantum encoding, but the paper's central claim—that ALGA \"achieves the same simulation results of LBM\"—is only tested, not derived, for the adaptive case. I'd send it to peer review, but the authors need to do a proper asymptotic analysis of the space-dependent λc and give us code and at least a second benchmark.\n\nWhat's new: Eq. 17 for λc(u), locally adapted to force LBM equilibrium, replacing the random sampling of MCLGA with a deterministic fraction; and the amplitude encoding of the nonlinear min term in the |11> state, which turns the collision into a linear operator (Eq. 19). The per-step O(log^2 N) streaming and the honest statement that measurement/reinitialization brings total cost to Ω(NT) are communicated clearly. The Qiskit statevector experiments are a real check of the circuit construction.\n\nWhere it's soft: The equilibrium verification in Sec. V.C is partly circular, since λc was derived by imposing LBM equilibrium. The cosine-wave match is the only external evidence, and it is one low-Mach case with no error quantification and no variation of Reynolds number or wavelength. More importantly, the Chapman-Enskog analysis in Appendix A treats λc and λs as constants; the adaptive λc varies in space and time, so the effective viscosity and higher-order terms are uncontrolled. The paper itself concedes no analytical τ–λs relation. Reproducing equilibrium distributions is plausibly necessary but not obviously sufficient for matching macroscopic dynamics.\n\nThere are also some notation problems that need fixing: Sec. II.C states w−=w+=1/3, but the standard D1Q3 weights are 1/6, and Eq. 9 as written is not the LBM equilibrium; Eq. 17 appears to rely on the 1/6 weights. These look like typos, but in a paper whose derivation lives in these equations they obscure the result. No code or data are included.\n\nNone of this kills the contribution. The construction is coherent, the quantum encoding is new, and the authors are candid about the missing pieces—they explicitly say the τ–λs comparison \"did not prove any confirmation\" and that time-step concatenation is not achieved. This is a building block for quantum CFD, and a serious referee could guide it to a solid revision. My recommendation: engage with it, require the adaptive-limit analysis and a reproducibility package, and see the revision.","headline":"A genuinely new deterministic lattice-gas construction with a clean quantum encoding, but the LBM-equivalence claim rests on equilibrium matching and one benchmark because the adaptive hydrodynamic limit is never derived.","tokens_in":16382,"tokens_out":2167,"would_cite":true,"duration_ms":20121,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["76M28","81P68"],"pacs":["47.11.-j","03.67.Ax"],"model":"deepseek-v4-flash","headline":"A deterministic lattice gas with locally adapted collision fractions reproduces LBM equilibrium and low-speed dynamics, and admits a quantum version with a linear collision operator on log(N)+3 qubits.","keywords":["lattice gas automata","integer lattice gas","lattice Boltzmann method","adaptive collision rate","quantum lattice gas","D1Q3 model","linear combination of unitaries"],"falsifier":"Measure the effective viscosity of ALGA from the decay rate of the cosine wave over many time steps, at several $|u|$ values inside the validity range, and compare with the LBM viscosity $(\\tau - 0.5)/3$ for the best matching relaxation time; if the fitted viscosity varies with amplitude or the density profiles deviate systematically from LBM at matched times, the claim of identical simulation results is falsified.","tokens_in":15354,"feed_emoji":"⚛️","tokens_out":13978,"duration_ms":110224,"temperature":0.7,"pith_summary":"Starting from a one-dimensional integer lattice gas (ILGA), this paper replaces the random collision selection of Monte Carlo lattice gas automata with a fixed fraction of the possible crunches and splittings at each site. The crunch fraction $\\lambda_c$ is then adapted locally through Eq. (17), so that the equilibrium populations match those of the lattice Boltzmann method (LBM) in the $|u|\\approx 0$ limit. The resulting adaptive lattice gas algorithm (ALGA) is deterministic and, on a cosine-wave benchmark, reproduces the shock formation and density/momentum profiles of LBM. The same collision rule is encoded in quantum amplitudes via Eq. (18), making the collision operator linear; the quantum version uses $\\log(N)+3$ qubits and costs $O(\\log^2 N)$ per evolution step before measurement and reinitialization. If the claim holds, ALGA offers a quantum-friendly, sampling-free classical scheme that inherits LBM's low-Mach behavior.","feed_headline":"A dice-free lattice gas matches the lattice Boltzmann method","feed_subtitle":"Per-cell collision fractions replace random sampling and unlock a compact quantum circuit.","key_machinery":"The load-bearing object is the adapted crunch fraction $\\lambda_c(x,t) = \\lambda_s (2 - 3u^2)/(1 - 3|u| + 3u^2)$, which turns a linear-in-velocity equilibrium into LBM's quadratic equilibrium by making the collision intensity a local function of momentum density. Its quantum counterpart is the amplitude encoding $|\\Psi\\rangle = (1/M)\\sum_x |x\\rangle(n_0|00\\rangle + n_1|01\\rangle + n_2|10\\rangle + \\lambda_c \\min(n_0,n_1)|11\\rangle)$, which buries the nonlinear min term in an amplitude so the collision update becomes the linear operator $\\hat{C}$ of Eq. (19). That operator is implemented through singular-value decomposition and a linear combination of unitaries, with streaming carried out by a quantum shift circuit at $O(\\log^2 N)$ cost. The $\\min(n_0,n_1)$ term is the nonlinear engine in both the classical and quantum versions.","core_discovery":"The paper's central claim is that a D1Q3 (one-dimensional, three-velocity) integer lattice gas in which a fraction $\\lambda_s$ of splittings and a locally adapted fraction $\\lambda_c(x,t)$ of crunches are performed reproduces the equilibrium distribution functions of LBM for $|u|\\approx 0$, and therefore simulates the same macroscopic phenomena. The adaptation rule $\\lambda_c = \\lambda_s (2 - 3u^2)/(1 - 3|u| + 3u^2)$ is obtained by substituting LBM's Maxwell–Boltzmann equilibrium populations into the ILGA equilibrium condition, and is validated by equilibrium measurements and by a cosine-wave shock simulation matching LBM. The paper further claims that encoding the populations, including the nonlinear min term, in quantum amplitudes makes the collision operator the linear matrix of Eq. (19); using singular-value decomposition and a linear combination of unitaries, the per-step quantum cost is $O(\\log^2 N)$ with $\\log(N)+3$ qubits, but the present implementation needs measurement and reinitialization every time step, so the full $T$-step cost remains $\\Omega(NT)$.","pith_inferences":["The paper leaves implicit that a second-order Chapman–Enskog expansion with the adaptive $\\lambda_c(x,t)$ would settle whether the effective viscosity is LBM's; the numerical hint $\\tau \\propto (1+\\lambda_s)/\\lambda_s$ is a concrete starting point for deriving that formula.","The encoding's trick of hiding the nonlinear min term in an amplitude so the collision operator stays linear suggests a general recipe for quantum-friendly lattice gases; its scalability will be tested when the same idea is extended to 2D/3D models with multiple nonlinear terms.","If the min term can be computed unitarily or supplied analytically, the paper's concatenation argument implies the $T$-step cost drops to $O(T \\log^2 N)$, at the price of one ancilla per step and a $2^{-T}$ success probability; that trade-off is the natural next design problem."],"forward_implications":["In the $|u|\\approx 0$ limit, ALGA inherits LBM's low-Mach behavior: the cosine-wave benchmark matches LBM's mass and momentum profiles, and stability is lost at the same high-velocity threshold as LBM.","Because ALGA is deterministic, the ensemble averaging needed by Monte Carlo lattice gas automata is not required; with smooth initial data a single run reproduces the averaged results.","The quantum encoding uses $\\log(N)+3$ qubits, with an $O(1)$ collision via a linear combination of unitaries and $O(\\log^2 N)$ streaming from the quantum shift circuit.","In the constant-$\\lambda_c$ case, time-step concatenation is possible when the min term is always attained by the same population; with local adaptation it is possible if the local momentum is known analytically and has one sign everywhere, removing measurement and reinitialization.","Without such concatenation, measurement and reinitialization dominate: the full algorithm costs $\\Omega(NT)$, so a quantum advantage awaits efficient tomography or unitary computation of the nonlinear term."],"supporting_citations":[{"why":"Supplies the D1Q3 Monte Carlo lattice gas model and its nonlinear equilibrium distributions that ALGA starts from.","marker":"[31]"},{"why":"Shows an integer lattice gas with a Monte Carlo collision operator recovers LBM results, the precedent ALGA seeks to make deterministic.","marker":"[30]"},{"why":"Provides the amplitude-encoding of lattice Boltzmann populations that the quantum encoding in Eq. (18) adapts.","marker":"[18]"},{"why":"Gives the quantum shift circuit used for streaming at $O(\\log^2 N)$ cost.","marker":"[33]"},{"why":"Provides the singular-value decomposition used to implement the non-unitary collision operator.","marker":"[37]"},{"why":"Supplies the linear-combination-of-unitaries method for implementing the diagonal factor of the collision operator.","marker":"[38]"},{"why":"Defines the BGK single-relaxation-time collision model whose equilibrium ALGA reproduces.","marker":"[35]"}],"fun_headline_variants":["Adaptive lattice gas matches LBM without random dice","Quantum lattice gas with adaptive collisions matches LBM","Lattice gas adapts to replicate LBM equilibrium","Adaptive collisions let lattice gas mimic LBM","No dice needed: adaptive lattice gas equals LBM"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole result assumes that matching LBM's equilibrium populations makes the large-scale behavior match too; the paper tests this only on one cosine-wave setup, because the governing equations for the adaptive rule are not derived.","fun_headline_variants_meta":{"raw":{"variants":["Adaptive lattice gas matches LBM without random dice","Quantum lattice gas with adaptive collisions matches LBM","Lattice gas adapts to replicate LBM equilibrium","Adaptive collisions let lattice gas mimic LBM","No dice needed: adaptive lattice gas equals LBM"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000254,"raw_usage":{"total_tokens":1546,"prompt_tokens":899,"completion_tokens":647,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":515,"completion_tokens_details":{"reasoning_tokens":573}},"tokens_in":515,"tokens_out":647,"duration_ms":6302,"temperature":1.0,"reasoning_tokens":573,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:07:50.294457+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the effective viscosity of ALGA from the decay rate of the cosine wave over many time steps, at several $|u|$ values inside the validity range, and compare with the LBM viscosity $(\\tau - 0.5)/3$ for the best matching relaxation time; if the fitted viscosity varies with amplitude or the density profiles deviate systematically from LBM at matched times, the claim of identical simulation results is falsified.","supporting_citations":[{"cited_title":"Succi, F","cited_arxiv_id":null,"evidence_quote":"Supplies the D1Q3 Monte Carlo lattice gas model and its nonlinear equilibrium distributions that ALGA starts from."},{"cited_title":"Sanavio and S","cited_arxiv_id":null,"evidence_quote":"Shows an integer lattice gas with a Monte Carlo collision operator recovers LBM results, the precedent ALGA seeks to make deterministic."},{"cited_title":"In this simulation we are well within the limits for which the collisions take place in each site","cited_arxiv_id":null,"evidence_quote":"Provides the amplitude-encoding of lattice Boltzmann populations that the quantum encoding in Eq. (18) adapts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the quantum shift circuit used for streaming at $O(\\log^2 N)$ cost."},{"cited_title":"Blommel and A","cited_arxiv_id":null,"evidence_quote":"Provides the singular-value decomposition used to implement the non-unitary collision operator."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the linear-combination-of-unitaries method for implementing the diagonal factor of the collision operator."},{"cited_title":"Chopard, A","cited_arxiv_id":null,"evidence_quote":"Defines the BGK single-relaxation-time collision model whose equilibrium ALGA reproduces."}],"review_version":1}