{"id":"00cb2a81-8f75-4095-b335-1fd80a4c35e2","arxiv_id":"2601.11391","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A Python-LAMMPS framework that extracts interfacial velocity and temperature slip coefficients from equilibrium molecular dynamics fluctuations.","lead":"NAVIS is a Python-LAMMPS toolkit that computes the Navier friction coefficient and Kapitza resistance at solid-fluid interfaces from equilibrium molecular dynamics. It packages previously published linear-response methods into reusable scripts for nanofluidic researchers who need slip boundary conditions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Method-3's claimed robustness rests on an untested one-term Maxwellian memory-kernel ansatz; a multi-exponential kernel would bias the extracted zero-frequency friction and Kapitza values.","rationale":"The reader's weakest assumption and my load-bearing concern coincide: the one-term Maxwellian memory-kernel ansatz is the theoretical linchpin of the NAVIS methodology. Method-1 is model-free but is the one reported to become statistically unreliable at long lag times; Method-2 and Method-3 both assume the exponential form. If the ansatz fails, the central quantitative outputs are biased in a way that the stability plots cannot reveal, because the plots only show insensitivity to lag time within the assumed model class. This is a correctness risk, not merely a deviation from community consensus: the equations (3)-(8) and (10)-(11) explicitly depend on the ansatz. However, the concern is not a demonstrated error. The prior NEMD comparison for water-graphene, the physically plausible dominance of a single interfacial relaxation time, and the internal consistency of the derivation give the method meaningful support. The manuscript's own statement that the figures are reproduced from previous studies means the current code's output is not independently demonstrated, but that is secondary to the theoretical assumption. The appropriate verdict remains CONDITIONAL: the method is plausible and well-founded, provided the ansatz is tested and the code is made verifiable. A synthetic two-exponential test directly settles whether the concern lands, and a flexible model-free fit on real data would be a valuable additional check.","tokens_in":14717,"tokens_out":7817,"duration_ms":87204,"concrete_test":"Generate exact, noise-free Laplace-space data from a two-exponential friction kernel ζ(t)=B1 e^{-λ1 t}+B2 e^{-λ2 t}, with parameters chosen to mimic plausible interfacial relaxation (e.g., B2/B1≈0.3, λ2/λ1≈10). Use the exact relation C̃_uxFx(s)=–ζ̃(s)C̃_uxux(s) with a specified C̃_uxux(s), then run Algorithm 4 Method-3 over s∈[0,1] with s_mid=0.5. Compare the recovered ξ0 to the true zero-frequency value ζ̃(0)=B1/λ1+B2/λ2. If the recovery error is large relative to the statistical error bars in Fig. 3(b), the Maxwellian ansatz is the dominant source of bias and the claim of robustness for arbitrary interfaces is not established. If the recovery error is small, the ansatz is vindicated for this class of kernels.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central derivation in Sec. 2.1 assumes the friction memory kernel is exactly ζ(t)=B1 e^{-λ1 t} (Eq. 4), and Sec. 2.2 assumes G_k(t)=k1 e^{-μ1 t} (Eq. 10). Under this ansatz, the ratio Q(s) ≡ –C̃_uxux(s)/C̃_uxFx(s) equals (s+λ1)/B1, which is exactly linear in s. Both 'robust' estimators depend on this linearity: Method-2 fits Q(s) to a straight line, and Method-3 computes B1 and λ1 from integrals of Q(s) over two finite s-intervals (s=0→0.5 and 0.5→1), implicitly extrapolating Q(0). If the true memory kernel has additional relaxation modes, Q(s) is not linear, and the finite-s extrapolation to s=0 yields a biased ξ0 = B1/(A λ1) rather than the true ζ̃(0). The same issue applies to R_k via Eq. (10)-(11). The manuscript provides no diagnostic for the ansatz, no test of sensitivity to the chosen slab thickness Δ, and no model-free comparison in this work; the NEMD comparison in Fig. 3(d) is reprinted from a prior study for a single interface. Thus the key advertised property — Method-3 being robust across correlation lag times — may reflect the rigid fitting form rather than the physics.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents NAVIS, a LAMMPS-Python toolkit that implements the authors' previously developed equilibrium-MD methods for computing the Navier friction coefficient and the Kapitza resistance at solid-fluid interfaces. The manuscript gives a condensed account of the linear-response theory, provides LAMMPS input snippets and Python post-processing algorithms, and illustrates the workflow on a water-graphene system (hydrodynamic slip) and a water-CNT system (thermal slip). The central advertised claim is that the 'Method-3' variant, together with the analogous thermal method, yields interfacial transport coefficients that are statistically stable across correlation lag times and avoids the Green-Kubo plateau/divergence problem. The figures are explicitly stated to be reproduced from the authors' earlier papers.","tokens_in":15111,"tokens_out":6939,"duration_ms":73243,"significance":"If the robustness claims hold, NAVIS would be a useful community resource: it packages a linear-response EMD protocol for extracting intrinsic interfacial friction and thermal resistance from a single equilibrium simulation with wall thermostatting, avoiding NEMD extrapolation issues. The open-source GPL release, the explicit LAMMPS snippets, and the step-by-step algorithms are strengths. However, the evidence presented in this manuscript is weaker than the advertised central claim: the figures are reprints from prior studies, no test is shown that the released code reproduces those figures, and no test is provided for the single-exponential memory-kernel ansatz on which Methods 2 and 3 rely. The claimed robustness of Method-3 may be an artifact of that assumed functional form rather than a generic property of the estimator. With sensitivity tests and a reproducible demo, the paper could become a solid software contribution.","major_comments":[{"comment":"Eqs. (4) and (10) assume the friction and thermal memory kernels are exactly one-term Maxwellians, ζ(t)=B1 e^{-λ1 t} and G_k(t)=k1 e^{-μ1 t}. Methods 2 and 3, and the analog for the Kapitza resistance, all depend on this form: Q(s) ≡ -C~uxux(s)/C~uxFx(s) is then exactly linear in s, and Method-3 extrapolates Q(s) to s=0 using finite-s intervals (chosen in Sec. 4.1 as s1=0, s2=s3=0.5, s4=1). If the true memory kernel has additional relaxation modes, Q(s) is nonlinear and the extracted ξ0 = B1/(A λ1) and Rk = μ1/k1 are biased. The manuscript provides no diagnostic for the ansatz, no convergence check, no sensitivity study of the slab thickness Δ, and no model-free comparison within this work. The advertised robustness of Method-3 in Fig. 3(b,c) could therefore reflect the rigid fitting form rather than the underlying physics. Please add a quantitative test, e.g., fit a two-exponential kern","section":"Secs. 2.1, 2.2, 4.1"},{"comment":"The paper is a software contribution, but the validation does not exercise the released code. Section 4 states that the figures 'have been reproduced using data from our previous studies,' and Figs. 3(d) and 4 are reprinted. No output of the NAVIS code is shown, no sample input/output files or unit tests are provided, and no check that the Python implementation reproduces the printed correlation functions is reported. A reader cannot verify that the GitHub implementation corresponds to the described Algorithms 2-5. I request a reproducible test case (input files, run commands, expected numeric values for ξ0 and Rk) and a demonstration that the code reproduces at least one of the reprinted figures.","section":"Secs. 4, Algorithms 2-5"},{"comment":"The generality claims rest on a single system per property — water-graphene for friction and water-CNT for Kapitza — and several free parameters are chosen without sensitivity analysis. The interfacial slab thickness Δ is set to 3.165 Å in both cases but its influence on the extracted coefficients is not investigated, despite Δ defining the interfacial region over which the correlation functions are computed. Likewise, Algorithm 2 introduces no_of_sets (the number of independent correlation blocks) as a free parameter, and the paper gives no guidance on how the results depend on it. Since these parameters directly affect the correlation data entering Eqs. (3) and (11), the claim that the method yields an intrinsic interfacial property is not fully established. Please add a sensitivity study for Δ and no_of_sets, and state the criterion used to select them.","section":"Secs. 4.1, 4.2, Algorithm 2"}],"minor_comments":[{"comment":"There are several typos and spelling inconsistencies: 'Kaptiza' in the Section 2.2 heading, 'noticable' and 'nanometeric' in the Introduction, 'analagous' in Section 2.2, and the title is rendered 'NA VIS' in the abstract but 'NAVIS' elsewhere.","section":"Throughout"},{"comment":"Refs. 31 and 49 are the same Barrat-Chiaruttini article; Ref. 48 (Puech et al.) is not the source of the Green-Kubo method attributed to Barrat and Chiaruttini in Sec. 4.2. The reference list should be cleaned up.","section":"References"},{"comment":"The GitHub link is broken across lines ('https://github.com/sleebaslv/ NAVIS') and no version or commit hash is given. Please provide a single clickable URL and a version identifier so that the deposited code is reproducible.","section":"Program Summary"},{"comment":"The relationship between the LAMMPS quantities computed in Listing 2 (vz times the z-force tally) and the heat flux J_q used in Eq. (9) is not explained. Please give the explicit definition of J_q used, or point to the exact equation in the referenced prior work, so that users can verify they are computing the intended correlation function.","section":"Listing 2 / Sec. 3.2"},{"comment":"The word 'efficient' in the title is not quantified. No runtime or performance benchmark is reported; a brief wall-clock time for the two example systems would help users gauge the cost of the EMD approach.","section":"Sec. 4"}],"recommendation":"major_revision","confidential_remarks":"This is a software-paper submission whose central risk is that the advertised 'robustness' of Method-3 is an artifact of the single-exponential Maxwellian ansatz, and that the released code has not been demonstrated to reproduce the reprinted figures. The authors' prior papers contain the substantive methodological validation; this manuscript needs an independent reproducibility test and a sensitivity analysis before it can be accepted as a self-contained toolkit paper. I would also suggest the editor verify that the GitHub repository is public and contains the exact scripts described in the manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: NAVIS is a genuinely useful pedagogical wrapper around the EMD methods that Hansen, Varghese, Alosious and co-workers published over the last fifteen years. But the submitted manuscript does not demonstrate that the shipped code actually reproduces the figures it shows, and the advertised \"robust\" Method-3 leans on an untested single-exponential memory kernel. Neither flaw sinks the underlying approach, but both need to be fixed before I'd trust the toolkit.\n\nWhat's new: the code, the LAMMPS snippets, and the walkthrough. The theory is unchanged from Refs [16,30,33], and the authors say so plainly. That honesty is to their credit. The algorithms are clear enough that a competent grad student could implement them from the pseudo-code alone. The comparison with the Green-Kubo plateau problem is well framed, and the NEMD-EMD agreement in earlier work is a strong external anchor.\n\nSoft spots: First, there is no verification run. The paper states the figures are \"reproduced using data from our previous studies\" — so the reader never sees NAVIS output. The GitHub link is not a versioned/archived artifact, and the CPC link is still TBD. For a software paper, that is the central gap. Second, the claim that Method-3 is stable across correlation lag times may be an artifact of the one-term Maxwellian ansatz (Eq. 4 and Eq. 10). If the true memory kernel has multiple relaxation modes, the finite-s extrapolation to s=0 is biased. The paper gives no diagnostic, no sensitivity to slab thickness Δ, and no model-free check within the present work. The prior NEMD validation in Ref [16] for a single system is reassuring, but it is not a proof of the ansatz. Third, minor: the Kapitza \"excellent agreement\" plot has no error bars, and reference [28] is missing authors; refs [2] and [5] are duplicated.\n\nWho it's for: researchers who want a ready-made equilibrium simulation pipeline for interfacial slip and are willing to test it on their own systems. The paper deserves a serious referee, but the referee should be asked to check that the code runs and reproduces at least one of the reprinted figures, and to push the authors to discuss kernel-dependence. If the repo is pinned with validation data, I'd be happy to see it accepted; as is, it's a conditional.","headline":"A useful but unverified software wrapper around validated EMD slip methods; Method-3's stability claim needs a kernel-model caveat.","tokens_in":15548,"tokens_out":5963,"would_cite":false,"duration_ms":58207,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"NAVIS extracts the intrinsic friction and Kapitza resistance at a solid-fluid interface from fluctuations in a single equilibrium molecular dynamics run, avoiding the Green-Kubo plateau problem by fitting a single-exponential memory kernel","keywords":["Navier friction coefficient","Kapitza resistance","equilibrium molecular dynamics","velocity slip","thermal slip","interfacial slab","Laplace-domain analysis","nanochannel"],"falsifier":"Run the same equilibrium simulation with several slab thicknesses Δ (e.g. 2, 3, 4, and 5 Å) and check whether ξ0 and Rk remain constant; or fit the ratio of Laplace-transformed correlation functions at many s values and look for systematic deviation from the single-exponential form. A visibly non-exponential ratio, or a drift of ξ0 with Δ, would falsify the central extraction claim.","tokens_in":14598,"feed_emoji":"💧","tokens_out":4012,"duration_ms":46127,"temperature":0.7,"pith_summary":"The paper presents NAVIS, a Python toolkit that computes the intrinsic Navier friction coefficient and Kapitza resistance of a solid-fluid interface from fluctuations measured in an equilibrium molecular dynamics simulation. The key move is to restrict all correlation functions to a thin slab of fluid next to the wall, making the result a local interface property rather than a bulk property. To bypass the divergence of Green-Kubo integrals in confined systems, the toolkit assumes the friction and thermal memory kernels are single exponentials and extracts zero-frequency coefficients in the Laplace domain. The paper argues that its preferred Method-3 yields friction values that do not drift with correlation lag time and that its Kapitza method similarly stays flat while the standard Green-Kubo analogue diverges.","feed_headline":"Compute wall slip and Kapitza resistance from one equilibrium run","feed_subtitle":"NAVIS fits a single-exponential memory kernel in Laplace space, so results stay stable with correlation time; water-graphene and water-nanot","key_machinery":"The load-bearing object is the one-term Maxwellian memory kernel, ζ(t)=B1 e^{-λ1 t} for momentum and G_k(t)=k1 e^{-μ1 t} for heat, together with the Laplace-domain relation C̃_{u_x F_x}(s)=-(B1/(s+λ1)) C̃_{u_x u_x}(s) and its thermal analogue. The extraction compares Laplace transforms of measured slab-velocity/force and temperature/heat-flux correlations. Method-3 integrates these transforms over two intervals in s-space to solve for B1 and λ1 algebraically, removing the need to truncate the correlation function at long times. The dynamic fluid slab, defined as the layer within a few molecular diameters of the wall, is what makes the computed coefficients intrinsic to the interface rather t","core_discovery":"The central claim is that interfacial hydrodynamic and thermal resistance can be obtained from a single equilibrium simulation by correlating fluctuations in a molecularly thin slab adjacent to the wall, with the extraction stabilized by modelling the friction memory kernel as ζ(t)=B1 e^{-λ1 t} and the thermal kernel as G_k(t)=k1 e^{-μ1 t}. The zero-frequency friction coefficient ξ0=B1/(Aλ1) and the Kapitza resistance Rk=μ1/k1 follow from Laplace-transforming the measured correlation functions and fitting or algebraically inverting the single-exponential form. For a water-graphene channel, Method-3 gives friction coefficients that stay constant as the correlation lag time is varied, unlike t","pith_inferences":["If the single-exponential ansatz is exact, the same Laplace-domain algebraic inversion could be extended to multi-exponential kernels by adding more integration intervals and solving for additional relaxation times, providing a built-in shape test.","The slab thickness Δ is a free parameter; a natural convergence check is to verify that ξ0 and Rk are independent of Δ while Δ lies within the first density peak, which the current paper does not report.","Method-3 uses four quadrature points in Laplace space; shifting the integration intervals s1..s4 would reveal how sensitive the result is to the shape of the memory kernel and to the numerical Laplace transform.","The nonequilibrium agreement is shown at one field strength; comparing at several field strengths and temperatures would confirm the linear-response regime and sharpen the validity of the exponential ansatz."],"forward_implications":["Users can obtain velocity-slip and thermal-slip parameters from one equilibrium run with wall thermostatting, without imposing a flow or a temperature gradient.","Because the correlation functions are restricted to a thin slab, the method can deliver separate intrinsic friction coefficients for two different walls in the same simulation.","The Kapitza-resistance version works in cylindrical confinement without needing an auxiliary inner cylinder to generate a temperature gradient.","Method-3's stability across correlation lag times removes the need to visually pick a plateau region, a standard bottleneck in Green-Kubo analysis.","The packaged Python algorithms for correlation and Laplace transformation are reusable for arbitrary solid-fluid interfaces, not just the two water-carbon examples demonstrated."],"fun_headline_variants":["One equilibrium run yields slip and Kapitza resistance","Stable friction via single-exponential Laplace fit","NAVIS: friction and thermal slip from one LAMMPS run","Single-exponential kernel keeps friction stable over time","Equilibrium MD with exponential kernel nails slip and Kapitza"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The derivation assumes the friction memory kernel is exactly a single decaying exponential, ζ(t)=B1 e^{-λ1 t}, and the thermal kernel analogously a single exponential; if the true kernel has a second timescale or a different shape, the extracted zero-frequency coefficient is biased, and the paper gives no test of this shape assumption.","fun_headline_variants_meta":{"raw":{"variants":["One equilibrium run yields slip and Kapitza resistance","Stable friction via single-exponential Laplace fit","NAVIS: friction and thermal slip from one LAMMPS run","Single-exponential kernel keeps friction stable over time","Equilibrium MD with exponential kernel nails slip and Kapitza"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000203,"raw_usage":{"total_tokens":1259,"prompt_tokens":816,"completion_tokens":443,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":560,"completion_tokens_details":{"reasoning_tokens":365}},"tokens_in":560,"tokens_out":443,"duration_ms":5838,"temperature":1.0,"reasoning_tokens":365,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T10:00:36.861410+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same equilibrium simulation with several slab thicknesses Δ (e.g. 2, 3, 4, and 5 Å) and check whether ξ0 and Rk remain constant; or fit the ratio of Laplace-transformed correlation functions at many s values and look for systematic deviation from the single-exponential form. A visibly non-exponential ratio, or a drift of ξ0 with Δ, would falsify the central extraction claim.","supporting_citations":[],"review_version":1}