{"id":"e696cba3-9687-4325-9b54-d806bc183579","arxiv_id":"1908.07202","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Starting from Nosé-Hoover thermostatted walls, the paper derives linear response in terms of boundary heat flows, yielding Green's formula for thermal conductivity and the McLennan-Zubarev steady-state distribution.","lead":"Microscopic theory of heat flow from thermostatted boundary walls: response of any observable is expressed via boundary heat flow correlations, and the bulk thermal conductivity emerges as a Green-Kubo formula. The paper gives a new derivation of the McLennan-Zubarev nonequilibrium distribution and shows how surface-to-bulk relations link thermostat driving to bulk dissipation.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unproven microscopic relaxation from P_lc to P_st is the load-bearing gap in the claimed first-principles McLennan-Zubarev derivation (Eqs. 122-123); the paper itself lists the needed numerical check.","rationale":"The reader's verdict of CONDITIONAL is appropriate, but I disagree with the identification of Eq. (34) as the weakest assumption. The fast-thermostat identification is used for the practical replacement of J_K by I_K in Eqs. (63)-(66), but the derivation of Green's expression in Eq. (45) is exact using the thermostat heat flows and the surface-to-bulk relation, so the linear-response central claim does not rest on Eq. (34). The genuinely load-bearing step is in Section IV E: the steady-state McLennan-Zubarev form is obtained by assuming, not proving, that P_lc relaxes to P_st in a microscopic time. The paper itself flags this missing support in Summary remark (6), and the convergence of D(t_lc) is a nontrivial ergodicity issue. This concern does not invalidate the linear-response theory, so the verdict should remain CONDITIONAL rather than move to REJECT, but the condition should be framed as a proof or numerical demonstration of the P_lc to P_st relaxation and of the convergence of D(t_lc). The circularity in the derivation of Eq. (66) noted by the reader is real but secondary, since Eq. (45) already establishes Green's expression without solving the hydrodynamic equations.","tokens_in":24336,"tokens_out":24032,"duration_ms":278410,"concrete_test":"Run nonequilibrium molecular dynamics, for example the 1D Nosé-Hoover chain suggested in remark (6) or a Lennard-Jones film, initialized from an approximation to P_lc, and monitor the ratio P(Γ,t)/P_lc against exp[D(t)] on individual trajectories. Check whether the ratio becomes stationary at a time t_lc much shorter than the diffusion time H^2/D, and whether the resulting P_st is independent of t_lc, tau_h, and the applied gradient. For an analytic check, use a harmonic chain with an exactly solvable nonequilibrium steady state and compare the exact stationary covariance matrix with the covariance of P_lc exp[D(t_lc)] for increasing t_lc; if agreement requires t_lc comparable to H^2/D or varies with tau_h, the first-principles claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central linear-response result is in good shape: Eq. (41) follows from the linearized Liouville equation for the thermostat perturbation, Eq. (25) is an exact surface-to-bulk identity, and the time-derivative terms drop from Eq. (45) by time-reversal symmetry, so the Green-Kubo expression for lambda does not actually require the J_K approx I_K identification of Eq. (34). The load-bearing gap is the nonlinear claim. Section IV E asserts that an initial local-equilibrium distribution P_lc (Eq. (114)) relaxes to the steady state as microscopic events in a microscopic time and then writes P_st = P_lc exp[D(t_lc)] (Eqs. (122)-(123)) with t_lc larger than that relaxation time. This is an unproven timescale and ergodicity assumption: D(t) is a time integral of the boundary heat flux G(t), so the representation is only meaningful as a limit or with the epsilon regularization of Eqs. (125)-(126), and no argument shows that the convergence time is independent of tau_h, H, or the applied gradient. The paper's own remark (6) says a numerical study of the P_lc to P_st relaxation should be informative, conceding the missing support. If that relaxation is slow, Eqs. (122)-(123) do not describe the actual nonequilibrium steady state.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a microscopic theory of heat flow in a classical fluid confined between Nosé-Hoover thermostatted boundary walls. In the linear regime it derives a response formula, Eq. (41), expressing the deviation of any observable in terms of equilibrium time-correlation functions with the thermostat heat flows J_K, and uses the surface-to-bulk identities of Eq. (25) to rewrite the response in a bulk form whose dissipative part yields Green's expression for the thermal conductivity, Eq. (45). In the nonlinear regime the paper derives a phase-space dissipation function, presents fluctuation theorems, and claims a first-principles derivation of a McLennan-Zubarev-type steady-state distribution, Eqs. (122)-(123), starting from a local-equilibrium distribution P_lc. The paper also provides hydrodynamic calculations of the relaxation after a boundary temperature change and a canonical-ensemble analysis of long-range correlations.","tokens_in":24616,"tokens_out":9568,"duration_ms":96021,"significance":"If the results hold, the paper establishes a clean formal link between boundary-driven nonequilibrium molecular dynamics and bulk Green-Kubo transport coefficients, and it gives a microscopic route to the McLennan-Zubarev steady-state form. The linear-response part is elegant: Eq. (41) follows from the linearized Liouville equation, and Eq. (25) is an exact surface-to-bulk identity. The causality prediction in Eq. (47) is a falsifiable statement that could be tested numerically. The nonlinear steady-state claim, however, rests on an unproven relaxation assumption, so the paper's strongest claim is not yet fully supported.","major_comments":[{"comment":"The central nonlinear claim P_st(Γ) = P_lc(Γ) exp[D(t_lc)] is not established. The argument relies on the assertion that P_lc relaxes to the steady state in a microscopic time, but no proof, timescale estimate, or numerical evidence is provided. This matters because D(t) = -γ_a ∫_0^t ds G(-s) is a time integral of the boundary heat flux whose steady-state mean is nonzero; hence exp[D(t_lc)] depends on the arbitrarily chosen t_lc unless a convergent regularization is employed, and the equivalence of Eqs. (125)-(126) to the unregularized form is not shown. The paper's own Remark (6) asks for a numerical study of the P_lc → P_st relaxation, effectively conceding that the necessary support is missing. Without this support, the claim of a first-principles derivation of the McLennan-Zubarev form should be softened to a conjecture or the required relaxation argument must be supplied.","section":"Section IV E, Eqs. (122)-(123) and Remark (6)"},{"comment":"The boundary-fluctuation formula for the thermal conductivity, Eq. (66), is derived from the hydrodynamic solution in Appendix A, which starts from the heat equation (A1) with a given conductivity λ. Consequently, Eq. (66) follows from an input that already contains λ and cannot serve as an independent derivation of λ from boundary fluctuations. This circularity does not affect the main Green-Kubo expression, Eq. (45), which is obtained directly from the linear-response calculation and the surface-to-bulk identity, but the text should explicitly identify Eq. (66) as a consistency check of the hydrodynamic description rather than a microscopic derivation.","section":"Section III C, Eqs. (63)-(66) and Appendix A"},{"comment":"The nonlinear analysis assumes a linear inverse-temperature profile δβ̄(z) = δβ_bot − γ_a z in Eq. (113), and this profile is used to define both the local-equilibrium distribution P_lc(Γ) in Eq. (114) and the dissipation function D(t) in Eq. (112). For large boundary temperature differences, the steady-state temperature profile is generally nonlinear when transport coefficients depend on temperature. The paper does not state the regime of validity of the linear-profile assumption in the nonlinear regime, nor does it generalize Ψ(Γ) to an arbitrary β(z). The claim that Eqs. (122)-(123) represent the steady-state distribution should be restricted to the linear-profile regime or generalized accordingly.","section":"Section IV D, Eqs. (112)-(114)"}],"minor_comments":[{"comment":"There are several typographical errors: the running header contains 'Wa lls' and 'Fluctuat ion', the abstract contains 'quantiti es', Eq. (48) uses the notation 'ẋζ_K' where dζ_K/dt is intended, and Eq. (125) states 'ǫe^{-ǫt} = -d(e^{-ǫ})/dt' but the final exponential should be e^{-ǫt}. Please proofread carefully.","section":"Throughout the text, Eqs. (48), (125)"},{"comment":"The statement that ⟨B(t)J_K(0)⟩_e is 'nearly equal' to ⟨B(t)I_K(0)⟩_e for t ≫ τ_h is an uncontrolled approximation. The main linear-response result in Eq. (41) and the Green-Kubo formula in Eq. (45) are exact in terms of J_K and do not depend on this identification; making that point explicit would prevent readers from believing the central results require the fast-thermostat assumption.","section":"Section III A, after Eq. (42)"},{"comment":"Reference [20] contains a typo, 'Fourierfs law' instead of 'Fourier's law'.","section":"Reference list, Ref. [20]"},{"comment":"The switching between backward trajectories in Eq. (124) and forward trajectories in Eqs. (120) and (123) is somewhat terse; a short sentence explicitly listing which definition of D(t) and G(t) is used in each of Eqs. (119), (120), (121), and (125)-(126) would improve readability.","section":"Section IV E, Eqs. (119)-(121)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious theory contribution from an established researcher. The linear-response part is sound and likely to be of interest to the statistical mechanics community. The nonlinear steady-state claim, however, is not yet proven; the author's own Remark (6) acknowledges that the relaxation P_lc → P_st requires numerical investigation. I recommend major revision: the nonlinear claim should be either proven under controlled assumptions or explicitly labeled as a conjecture, and the circular derivation of Eq. (66) should be clarified. The paper is within the scope of the journal and, with these revisions, could be a valuable contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this if you work on boundary-driven heat conduction. The paper's real contribution is the surface-to-bulk identity, Eq. (25), which expresses the thermostat heat flows as bulk integrals plus a time-derivative piece. That lets the boundary response be converted into bulk flux correlations, and the route from the Liouville equation to Green's expression for the thermal conductivity is clean. The author is careful with time reversal, and the hydrodynamic evaluation of the response functions in Appendices A and B is thorough. The stress-test note is right that the fast-thermostat replacement J_K ≈ I_K is not load-bearing for Eq. (45): the time-derivative terms drop from the time integral, so the Green-Kubo result does not actually depend on that approximation. The linear half of the paper is in good shape.\n\nThe soft spot is the nonlinear section. The derivation of the steady-state distribution in McLennan-Zubarev form, Eqs. (122)-(123), assumes that the local-equilibrium distribution relaxes to the steady state in a microscopic time, independent of the system size, the gradient, and the thermostat time. That is an unproven timescale assumption. D(t) is a time integral of the boundary heat flux; as t grows it generically does not converge as a phase-space function, so writing P_st = P_lc exp[D(t_lc)] needs a limit or an epsilon regularization. The author knows this—remark (6) says a numerical study of the P_lc-to-P_st relaxation would be informative—but it is still a load-bearing gap in the claim of a first-principles derivation. The circularity in Eq. (66) is, I think, a minor issue: the surface-fluctuation formula is derived using hydrodynamic expressions that already contain λ, so it is a consistency check rather than an independent derivation. The paper should say that plainly, but it does not damage the Green-Kubo derivation.\n\nThe citation pattern is fine, including the author's own prior work, and the formal manipulations are internally consistent. No simulations are provided, which is acceptable for a theory paper but leaves the nonlinear claims unsupported. My overall take: the linear-response part deserves a serious referee and likely publication after revision; the nonlinear part should be reframed as a conjecture or supported by a numerical test. I would send it to peer review rather than desk reject it.","headline":"The linear-response machinery from boundary thermostats to Green-Kubo is solid and worth engaging; the nonlinear McLennan-Zubarev derivation rests on an unproven fast-relaxation step that the author himself flags.","tokens_in":25127,"tokens_out":3489,"would_cite":true,"duration_ms":42326,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["44.10.+i","05.60.-k","05.70.Ln"],"model":"deepseek-v4-flash","headline":"Boundary heat flows from thermostatted walls yield Fourier's law and the McLennan-Zubarev steady state from first principles.","keywords":["thermal conductivity","Green-Kubo formula","boundary thermostats","linear response theory","McLennan-Zubarev distribution","fluctuation theorems","Nosé-Hoover thermostat","surface-to-bulk relation"],"falsifier":"Run a molecular-dynamics simulation of a Lennard-Jones fluid between two Nosé-Hoover thermostatted walls, measure the steady-state heat flux and temperature gradient to obtain $\\lambda$ directly, and separately compute the equilibrium time-correlation integral in Eq. (45) from the wall heat flows $J_K$ (or from the bulk flux $J_z^h$) over a range of thermostat times $\\tau_h$. If the correlation-integral value disagrees with the direct $\\lambda$ unless $\\tau_h$ is artificially small, the fast-thermostat identification $J_K\\simeq I_K$ and the surface-to-bulk reduction would be falsified for realistic thermostat parameters.","tokens_in":24062,"feed_emoji":"🔥","tokens_out":10431,"duration_ms":94096,"temperature":0.7,"pith_summary":"Onuki develops a microscopic statistical mechanics of heat conduction in a fluid film heated from thermostatted boundary walls, starting from the equations of motion rather than postulating a temperature profile. He shows that the linear response of any observable to a small boundary temperature change is fixed by equilibrium time-correlation functions between that observable and the heat flows from the two thermostats. A surface-to-bulk identity rewrites those wall heat flows as bulk space integrals, splitting the response into a local-equilibrium part and a dissipative part; the dissipative part yields Fourier's law with Green's expression for the thermal conductivity. In the nonlinear regime the same route delivers the steady-state phase-space distribution in the McLennan-Zubarev form, with a local-equilibrium distribution weighted by an integrated bulk heat-flux dissipation function, and several fluctuation theorems. If the derivation stands, it connects boundary-driven nonequilibrium simulations to bulk Green-Kubo transport coefficients.","feed_headline":"Boundary heat flow yields Fourier's law from first principles","feed_subtitle":"Wall-heat-flow correlations reproduce Green's conductivity formula and the steady-state distribution.","key_machinery":"The load-bearing object is the surface-to-bulk relation in Eqs. (13) and (25), which writes each thermostat heat flow $J_K$ as a sum of a boundary-layer energy time-derivative, a time-derivative of an integral of the bulk heat variable $\\hat q = \\hat e - h\\hat n$, and the integrated bulk heat flux $G(t)=\\int dr J_z^h(r,t)$. This identity is what turns surface-driven linear response into bulk correlation functions: inserted into $\\chi_{BK}(t)=\\langle B(t)J_K(0)\\rangle_e$, it produces the split into a local-equilibrium part (equal-time correlations with $\\hat q$) and a dissipative part (time integral of $G$), whose long-time limit gives Green's formula for the thermal conductivity. The argument also relies on the fast-thermostat identification $J_K\\simeq I_K$, meaning the Nosé-Hoover friction variable $\\zeta_K$ relaxes on a time $\\tau_h$ much shorter than bulk microscopic times, so that the thermostat heat flow equals the mechanical heat flow from the bound wall particles to the fluid.","core_discovery":"The central claim is that all boundary-driven heat response in a classical fluid between Nosé-Hoover thermostatted walls is governed by the thermostat heat flows $J_K$ (with $K$ = top or bottom), defined as the energy per time delivered from each thermostat to the particles. In the linear regime the deviation of any variable $B$ is $\\delta\\bar B(t) = -\\int_0^t ds\\,\\sum_K \\chi_{BK}(s)\\,\\delta\\beta_K(t-s)$ with $\\chi_{BK}(t)=\\langle B(t)J_K(0)\\rangle_e$. The key step is the surface-to-bulk relation, $J_{\\rm top}= d/dt[\\int dr\\,(z/H)\\hat q + H_{\\rm top}] - G/H$ and its mirror for the bottom, where $G(t)=\\int dr J_z^h(r,t)$ is the integrated bulk heat flux; this converts wall quantities into bulk ones. The dissipative part of the response is proportional to $G$, and in the limit $t\\to\\infty$ its time integral yields Green's expression $\\lambda = (1/V k_B T^2)\\int_0^\\infty dt\\int dr\\int dr'\\,\\langle J_z^h(r,t)J_z^h(r',0)\\rangle_e$. In the nonlinear regime, the steady-state distribution takes the McLennan-Zubarev form $P_{\\rm st}=P_{\\rm lc}\\exp[D(t_{\\rm lc})]$, where $P_{\\rm lc}$ is the local-equilibrium distribution built from a linear inverse-temperature profile and $D(t)$ is the integrated bulk heat-flux dissipation. The author also derives transient and steady fluctuation theorems, including the exact transient relation for stepwise boundary temperature changes.","pith_inferences":["The same surface-to-bulk construction should extend to simultaneous heat and momentum boundary drives, yielding a unified derivation of Green-Kubo shear and bulk viscosities from wall force correlations; the paper develops only the thermal case and cites the shear analogue.","If the thermostat relaxation time is not short compared with bulk microscopic times, Eq. (48) provides a correction term in $J'_K$; one could test numerically whether using $J'_K$ restores the Green-Kubo value of $\\lambda$ when $\\tau_h$ is varied over a range.","The local-equilibrium weight $\\exp[D(t)]$ is expressed solely through the integrated bulk heat flux, suggesting that for coarse-grained or multi-component descriptions the same entropy-production functional controls the steady-state ensemble; this is an extension beyond the one-component film treated here.","Near the critical point, where the piston effect makes adiabatic heating large, the separation of time scales $t_1 \\ll t_D$ in the dissipation function could be measured directly in experiments or simulations to test the hydrodynamic part of the calculation."],"forward_implications":["If the derivation is right, the thermal conductivity of a fluid can be extracted from equilibrium correlations of the wall thermostat heat flows alone, giving a practical route that matches boundary-driven nonequilibrium simulations with bulk Green-Kubo theory.","Steady-state averages of hydrodynamic variables follow from a two-part formula: a local-equilibrium part set by equal-time canonical correlations, including long-range $V^{-1}$ terms that yield the constant-pressure temperature derivative, and a dissipative part tied to the heat-flux integral.","The nonlinear steady-state distribution is the McLennan-Zubarev form, so nonlinear corrections to transport coefficients can be computed starting from a temperature-profile-dependent local-equilibrium ensemble rather than from the equilibrium one.","The causality prediction $\\chi_{aK}(z,t)=0$ for $0