{"id":"e28281c4-b950-418a-806a-0ca4c61f97a8","arxiv_id":"2412.03226","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors derive information-thermodynamic fluctuation theorems for mixed jump-diffusion systems and apply them to show that self-oscillations in a single-electron shuttle are fueled by energy transfer rather than information.","lead":"This paper builds a thermodynamic framework for systems where fast random switches, like electrons hopping on and off a dot, are coupled to a mechanical oscillator. It shows that in a nanoelectromechanical shuttle, the oscillator's self-sustained motion is powered mostly by energy flow, not by information flow.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The IFTs and adiabatic vanishing-flow results are proven only for position-dependent jump rates; for velocity-dependent rates the parity-consistent LDB is not established, so the claimed general extension is narrower than the abstract suggests.","rationale":"The reader's weakest_assumption identifies exactly the position-only jump-rate condition, and the paper itself flags the consequence in Sec. VII. I re-examined the central proof in Appendix C: the modified jump rates in Eq. (C19) and the escape-rate calculation are consistent, and my initial suspicion of a sign error in Eq. (C21) resolves once the index convention λ_{a,b} = rate from b to a is respected. The IFT chain then closes correctly, so I do not see an internal inconsistency in the main theoretical argument. The remaining serious limitation is scope: the IFTs and the adiabatic flow-vanishing hierarchy are established for λ(x), not λ(x,v), and velocity-dependent rates would require a parity-modified LDB whose consequences for Eq. (39) are, by the authors' own admission, unverified. For the electron-shuttle application this limitation is inactive, but for a paper whose abstract announces a general extension to 'Markov jump processes coupled to underdamped diffusion,' the assumption is load-bearing. The reader's CONDITIONAL verdict already captures this, together with the separate numerical-transparency concerns about error bars and uncontrolled higher-order TS terms; my analysis does not move that verdict.","tokens_in":43720,"tokens_out":32892,"duration_ms":312251,"concrete_test":"Simulate a minimal two-state dot coupled to an underdamped oscillator with velocity-dependent rates that are odd under time reversal, e.g. λ_{1,0}(x,v)=Γ_b e^{-x/λ} f_L(E(x)) e^{κ v/λ} and λ_{0,1}(x,v)=Γ_b e^{-x/λ} [1-f_L(E(x))] e^{-κ v/λ} for κ≠0, chosen so that the parity-corrected LDB log[λ_{1,0}(x,v)/λ_{0,1}(x,-v)] = -β[φ(1)-φ(0)-w] holds. From ≥10^5 stochastic trajectories, compute ⟨e^{-σ_X/k_B}⟩ and ⟨e^{-σ_Y/k_B}⟩; if either deviates from 1 beyond statistical error, the position-only assumption is load-bearing for Eq. (39) and the generality claim must be restricted accordingly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing premise is the position-only dependence of the jump rates λ_{y',y}^ρ(x), stated in Sec. II A. It enters the two pillars of the paper: (i) the proof of the integrated fluctuation theorems in Appendix C constructs modified dynamics under time reversal; if λ depended on v, the local detailed balance condition Eq. (5) would acquire odd-parity corrections, and the auxiliary-process construction used for Eq. (39) is not justified. (ii) The adiabatic result that steady-state energy and information flows vanish at leading order, Eqs. (55)–(56), and the subsequent O(ε²) shuttle formulas, Eqs. (101)–(102), rely on the fast conditional distribution relaxing to the velocity-independent π_y(x). The paper's own Sec. VII concedes that velocity-dependent jump rates could drive the slow underdamped subsystem out of equilibrium even under the adiabatic approximation and that 'further analysis is required to verify the validity of the generalized integrated fluctuation theorem (IFT) under these conditions.' Thus the central theoretical claim is rigorously established only for a restricted class of models; the electron-shuttle application is safe because its tunnel rates, Eq. (78), depend only on x, but the title-level generality is conditional on an assumption that is not generic for MJ-UD systems.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a stochastic-thermodynamics framework for autonomous systems in which an underdamped diffusive subsystem X is coupled to a Markov jump subsystem Y. The authors derive trajectory-level energy and entropy balances, identify internal energy and information flows, establish local second laws for each subsystem, and prove integrated fluctuation theorems for the partial entropy productions (Eq. (39)). They then analyze two asymptotic regimes: a timescale-separated (adiabatic) limit in which the leading-order energy and information flows vanish and the first non-zero contributions are computed at O(epsilon^2), and a large-mass limit in which the dynamics reduces to a deterministic mean-field description. The framework is applied to a single-electron shuttle, comparing stochastic simulations with the timescale-separated and mean-field predictions, and concluding that in the self-oscillating regime energy flows dominate information flows and that the shuttle acts as a sensor rather than an information engine.","tokens_in":43914,"tokens_out":7184,"duration_ms":70160,"significance":"If the results hold, the paper substantially extends bipartite information thermodynamics to a previously untreated class of mixed Markov-jump/underdamped-diffusion systems. The IFT proof in Appendix C constructs modified forward/backward dynamics for each subsystem and is a nontrivial contribution. The adiabatic and large-mass perturbation expansions are derived systematically from the master equation rather than assumed ad hoc, and the electron-shuttle application yields concrete, falsifiable predictions (Eqs. (101)-(102), Fig. 5) that are compared with direct stochastic simulations of the same model. The authors are also explicit about the main modeling restriction (position-dependent jump rates) in Sec. VII, which is a strength of the presentation. However, the advertised generality of the framework is wider than what is actually proven, and the quantitative shuttle analysis relies in part on uncontrolled higher-order terms, so the central claims need to be carefully delimited.","major_comments":[{"comment":"The load-bearing assumption that the jump rates lambda_{y',y}^rho(x) depend only on the position of the underdamped particle is not generic for MJ-UD systems, and the paper's central theoretical claims are proven only under this restriction. For velocity-dependent rates, the parity-consistent local detailed balance condition would acquire odd-parity corrections, and the modified-process constructions used in Appendix C (Eqs. (C11) and (C19)) would need to be re-derived. The authors themselves concede in Sec. VII that 'further analysis is required to verify the validity of the generalized integrated fluctuation theorem (IFT) under these conditions.' Since the abstract and introduction present the result as a general extension of information thermodynamics to mixed MJ-UD dynamics, the claims should either be extended to velocity-dependent rates or the abstract should explicitly state the position-only restriction. This is a genuine limitation, not merely a presentational one, because the adiabatic vanishing-flow results of Eqs. (55)-(56) and the subsequent O(epsilon^2) formulas depend directly on the velocity-independence of the fast conditional distribution.","section":"Sec. II A, Eq. (5); Sec. VII"},{"comment":"The quantitative shuttle results, including the advertised conclusion that energy flows dominate information flows in the self-oscillating regime, are obtained by solving the full timescale-separated Fokker-Planck equation (Eq. (82)) and using the full steady-state solution P_TS(E) from Eq. (D8), which contains uncontrolled higher-order terms in epsilon. Appendix D (Fig. D.2) shows that the exact O(epsilon^2) expressions for the flows, Eqs. (D13), agree with stochastic simulations only for small voltages, and the range of validity shrinks with increasing electromechanical coupling. The main-text comparisons in Figs. 5 and 8 therefore rely on a resummation that is not controlled, even though it is described as 'retaining these higher order terms.' The paper should either justify this resummation or present the controlled O(epsilon^2) results as the primary quantitative evidence, with the full-solution results clearly labeled as an uncontrolled but empirically better approximation.","section":"Sec. VI B 1 and Appendix D"},{"comment":"The large-mass mean-field dynamics is derived to O(m^{-1}) using a delta-function ansatz for the marginal distribution, but the exact heat flow Q_X in Eq. (90) contains a diffusive contribution -gamma/(m beta) that is of order m^{-1} and is omitted in the mean-field thermodynamic expressions (e.g., Eq. (74) and Eq. (76)). The paper acknowledges this in Sec. V, yet Fig. 6(a) compares the mean-field energy flow with stochastic simulations that include the full 1/m correction. The order of accuracy of each thermodynamic quantity computed from the mean-field dynamics should be stated explicitly, and the comparison should be adjusted or the missing 1/m term included, otherwise the apparent quantitative agreement in Fig. 6(a) mixes different orders in the expansion.","section":"Sec. V and Sec. VI A 1, Eq. (76), Eq. (90)"}],"minor_comments":[{"comment":"The notation d_tE_X and d_tE_Y in Eq. (12) is not defined before first use; the reader must infer that these are the rates of change of the bare energies E_X and E_Y. Please define these quantities explicitly.","section":"Sec. III A, Eq. (12)"},{"comment":"Equation (62) contains a misplaced parenthesis: the term D_eff(x) partial_v^2 is written inside the drift bracket with an extra closing parenthesis. This makes the Fokker-Planck structure harder to read and should be corrected.","section":"Sec. IV B, Eq. (62)"},{"comment":"The notation Delta-hat-I_X(Y)(t) in Eq. (39) is ambiguous because the subscript X(Y) appears both as a label and as a placeholder. Writing the two identities with explicit subscripts (e.g., for σ_X and σ_Y separately) would improve clarity.","section":"Sec. III D, Eq. (39)"},{"comment":"The mean-field equations (Eq. (81)) are written as closed equations for (x,v,p_1), but the derivation in Sec. V shows that they hold only up to O(m^{-1}) and that the diffusive term of order m^{-2} is neglected. The text around Eq. (81) should remind the reader of this limitation, since the subsequent comparison in Fig. 6 uses finite masses.","section":"Sec. VI A 1, Eq. (81)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the core derivations are valuable. The main concerns are the mismatch between the stated generality and the position-dependent-rate assumption, and the uncontrolled higher-order terms used in the main numerical validation. Both are addressable by restricting the claims or providing the missing extensions, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nWhat you should know: this is a real contribution, not a repackaging. The authors build the first general information-thermodynamics framework for systems with one Markov jump part and one underdamped diffusive part. The new results that matter are the integrated fluctuation theorems for partial entropy production (Eq. 39), the adiabatic vanishing of energy/information flows at leading order, and the systematic next-order corrections in both the slow-UD and large-mass limits. The electron-shuttle analysis is a genuine application, and the finding that energy flow dominates information flow in the self-oscillating regime is clearly presented.\n\nThe paper is honest about its main restriction: jump rates depend only on position, not velocity. That assumption is load-bearing. The IFT proof and the adiabatic zero-flow results both rely on it. The authors flag in Sec. VII that velocity-dependent rates need a careful re-examination of local detailed balance and that the generalized IFT may not hold. So the title-level generality is conditional, and the abstract oversells slightly. That said, every system they actually apply it to satisfies the assumption, and it is a natural starting class.\n\nWhat I would push on in review: the numerical validation. The comparisons with stochastic simulations are plausible and the TS predictions track the data, but there is no code, no error bars, and the steady-state solution of the modified Fokker-Planck equation retains uncontrolled higher-order terms. Appendix D shows that truncating at O(ε²) actually does worse than keeping higher-order terms, which is fine as an empirical observation, but it means the quantitative checks are not as clean as the text suggests. A referee should ask for convergence checks or a controlled expansion to support the claimed accuracy.\n\nThe derivations look right to me: standard stochastic thermodynamics, modified-process proofs, LDB. The self-citations are to prior work by the same group and are legitimate building blocks. The paper cites relevant competing work.\n\nBottom line: if you work on stochastic thermodynamics, information engines, or NEMS, read it. Bring it to a reading group and cite it if you touch mixed MJ-UD dynamics. I would send it to referees. The core framework is likely correct, and the caveats are stated rather than hidden. A serious referee should focus on the numerical quantitative claims and on whether the position-only assumption can be lifted.","headline":"Solid extension of information thermodynamics to mixed jump–underdamped systems, with honest limitations; worth refereeing despite a narrower scope than the abstract suggests.","tokens_in":44466,"tokens_out":2346,"would_cite":true,"duration_ms":19956,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82C31","82C70","60J75"],"pacs":["05.40.-a","05.70.Ln","85.85.+j"],"model":"deepseek-v4-flash","headline":"Mixed Markov-jump and underdamped-diffusion systems satisfy integral fluctuation theorems for each subsystem's partial entropy production.","keywords":["information thermodynamics","Markov jump process","underdamped diffusion","partial entropy production","integrated fluctuation theorem","local second law","electron shuttle","nanoelectromechanical systems"],"falsifier":"Simulate a two-state dot coupled to an underdamped oscillator with jump rates that include an explicit velocity dependence, for example $\\lambda(x,v)=\\lambda_0\\exp(\\pm x/\\lambda + a v)$, and check two predictions: the integrated fluctuation theorems either fail or change their information-flow term, and the adiabatic energy and information flows no longer vanish at leading order. Alternatively, for the shuttle at fixed coupling, measure the steady-state ratio $T\\dot I/\\dot E$ as a function of voltage; the paper predicts it vanishes in the self-oscillating regime, so a controlled experiment or numerical simulation showing that it remains order-one would falsify the energy-dominance claim.","tokens_in":43490,"feed_emoji":"⚙️","tokens_out":6591,"duration_ms":60570,"temperature":0.7,"pith_summary":"This paper extends information thermodynamics to autonomous systems in which one part jumps between discrete states while the other moves through an underdamped, inertial continuum. The authors establish local second laws for each part, with the missing entropy accounted for by information flows between the parts, and prove integrated fluctuation theorems for the partial entropy production of each subsystem. They then analyze two limit regimes—fast jumps with slow inertial motion, and large oscillator mass—and apply the framework to the single-electron shuttle. The shuttle analysis concludes that self-oscillations are sustained mainly by energy transfer from the dot to the oscillator, with information flows saturating and becoming relatively negligible at large voltage.","feed_headline":"Each half of a jump-diffusion system obeys its own fluctuation theorem","feed_subtitle":"Local second laws with information flows now cover mixed jump-diffusion machines, and the electron shuttle runs on energy, not information.","key_machinery":"The load-bearing object is the decomposition of the total entropy production into partial contributions $\\delta\\sigma_X$ and $\\delta\\sigma_Y$, assigned to the two subsystems' own dynamics, together with the information flows $\\dot I_X$ and $\\dot I_Y$ that measure each subsystem's rate of change of mutual information. The proof of the integrated fluctuation theorems constructs modified time-reversed dynamics: for the jump part, rates are reweighted by the instantaneous conditional distribution, and for the underdamped part, the drift is reversed and reweighted, so that the log-ratio of forward to modified-reverse path probabilities equals the partial entropy production. A second mechanism is the systematic $\\epsilon$-expansion of the joint master equation around the adiabatic limit, which yields a closed Fokker-Planck equation for the slow underdamped marginal with effective position-dependent damping $\\gamma_{\\rm eff}(x)$ and diffusion $D_{\\rm eff}(x)$, capturing the first nonzero energy and information flows. A third is the large-mass expansion, where the conditional distribution collapses to the dot marginal and the closed dynamics reduces to the mean-field shuttle equations.","core_discovery":"The central claim is that the bipartite information-thermodynamics formalism extends to a subsystem described by Markov jump dynamics coupled to a subsystem described by underdamped diffusion, in which position-dependent jump rates satisfy local detailed balance. Concretely, the paper derives the trajectory-level partial entropy productions $\\sigma_X$ and $\\sigma_Y$ and shows they obey the integrated fluctuation theorems $\\langle e^{-\\sigma_Y/k_B}\\rangle=\\langle e^{-[\\Sigma_Y-\\Delta I_Y]/k_B}\\rangle=1$ and $\\langle e^{-\\sigma_X/k_B}\\rangle=1$, so each subsystem's average partial entropy production is nonnegative once the information flow into or out of it is included. The paper further shows that in the adiabatic limit, where the jump dynamics is infinitely fast, a purely conservative underdamped subsystem remains at equilibrium and all steady-state energy and information flows vanish; nonvanishing flows first appear at next order in the timescale separation, through effective damping and diffusion terms that break detailed balance. In the large-mass limit, the underdamped subsystem becomes deterministic and the coupled dynamics reduces to a mean-field approximation, where energy flows survive but information flows vanish. Applied to the single-electron shuttle, these results yield a measurement-feedback picture in which the oscillator acts as a sensor of the dot state, and a thermodynamic efficiency $\\eta=\\dot E/T\\dot\\sigma_{\\rm tot}$ for converting electrical work into mechanical oscillations.","pith_inferences":["A velocity-dependent jump rate would likely break the adiabatic flow-hierarchy result; testing this extension directly would show whether the position-only assumption is essential or merely convenient.","The integrated fluctuation theorems for partial entropy production could be combined with standard uncertainty-relation arguments to derive thermodynamic uncertainty relations for the shuttle's current or oscillation amplitude, a step the paper does not take.","In non-isothermal conditions, the information-to-energy ratio should grow because the cost and gain scale with different temperatures, so a temperature-biased shuttle may act as a genuine information engine.","The saturation of the information flow at large voltage suggests a calibrated nanomechanical shuttle could serve as a direct probe of information-thermodynamic bounds, comparing the measured $T\\dot I$ against the local second-law inequality."],"forward_implications":["The local second laws become exact once information flow is included, so apparent violations of the second law in the marginalized dynamics are resolved and each subsystem's entropy production is nonnegative.","In the fast-jump, slow-oscillator limit, any steady-state energy and information flow is at least of order $\\epsilon^2$ for conservative interactions, so the slow inertial subsystem can be driven out of equilibrium only by going beyond the adiabatic approximation.","In the large-mass limit, the mean-field dynamics captures the limit-cycle attractor and the leading energy flow, but predicts zero information flow, which only appears through the higher-order diffusion term.","For the single-electron shuttle, the oscillator acts as a sensor that acquires mutual information from the dot, while the self-oscillating regime is maintained predominantly by energy flow, with the ratio $T\\dot I/\\dot E$ tending to zero at large voltage.","The framework provides a path to information-thermodynamic analyses of hybrid CMOS and mechanical clock circuits, where the derived fluctuation theorems and local second laws can be used to bound performance."],"supporting_citations":[{"why":"Establishes the bipartite information-thermodynamics decomposition for Markov jump processes that this paper extends.","marker":"[12]"},{"why":"Extends the same decomposition to underdamped diffusion, supplying the underdamped-side information-flow structure.","marker":"[17]"},{"why":"Derives integral fluctuation theorems for partial entropy productions in bipartite underdamped systems, the direct analogue generalized here.","marker":"[21]"},{"why":"Provides the corresponding integral fluctuation theorem for bipartite Markov jump processes.","marker":"[22]"},{"why":"Supplies the adiabatic perturbation method used to expand the master equation in the timescale-separation parameter.","marker":"[63]"},{"why":"Gives the electron-shuttle model and its effective Fokker-Planck description, the benchmark for the shuttle analysis.","marker":"[61]"},{"why":"Introduces the mean-field shuttle dynamics that the large-mass limit recovers systematically.","marker":"[64]"},{"why":"Provides the measurement-feedback perspective on nanoelectromechanical clocks that motivates the information-flow analysis.","marker":"[47]"}],"fun_headline_variants":["Each half of a mixed jump-diffusion system gets its own second law","Jump and diffusion halves each satisfy their own fluctuation theorem","Energy beats information in the electron shuttle's self-oscillations","Two fluctuation theorems for coupled jump-diffusion machines","Partial entropy laws split cleanly in jump-diffusion systems"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes the jump rates depend only on the underdamped particle's position, never on its velocity; if real rates carry velocity dependence, the adiabatic limit can drive the slow subsystem out of equilibrium and the derived flow hierarchy stops holding.","fun_headline_variants_meta":{"raw":{"variants":["Each half of a mixed jump-diffusion system gets its own second law","Jump and diffusion halves each satisfy their own fluctuation theorem","Energy beats information in the electron shuttle's self-oscillations","Two fluctuation theorems for coupled jump-diffusion machines","Partial entropy laws split cleanly in jump-diffusion systems"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000494,"raw_usage":{"total_tokens":2469,"prompt_tokens":1036,"completion_tokens":1433,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":652,"completion_tokens_details":{"reasoning_tokens":1347}},"tokens_in":652,"tokens_out":1433,"duration_ms":9820,"temperature":1.0,"reasoning_tokens":1347,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T22:37:44.168600+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a two-state dot coupled to an underdamped oscillator with jump rates that include an explicit velocity dependence, for example $\\lambda(x,v)=\\lambda_0\\exp(\\pm x/\\lambda + a v)$, and check two predictions: the integrated fluctuation theorems either fail or change their information-flow term, and the adiabatic energy and information flows no longer vanish at leading order. Alternatively, for the shuttle at fixed coupling, measure the steady-state ratio $T\\dot I/\\dot E$ as a function of voltage; the paper predicts it vanishes in the self-oscillating regime, so a controlled experiment or numerical simulation showing that it remains order-one would falsify the energy-dominance claim.","supporting_citations":[{"cited_title":"Freitas, J.-C","cited_arxiv_id":null,"evidence_quote":"Supplies the adiabatic perturbation method used to expand the master equation in the timescale-separation parameter."},{"cited_title":"Siniscalchi, F","cited_arxiv_id":null,"evidence_quote":"Gives the electron-shuttle model and its effective Fokker-Planck description, the benchmark for the shuttle analysis."},{"cited_title":"Lassagne, Y","cited_arxiv_id":null,"evidence_quote":"Provides the measurement-feedback perspective on nanoelectromechanical clocks that motivates the information-flow analysis."}],"review_version":1}