{"id":"6d62cd4f-46c3-4075-a2d5-6d34e77d8600","arxiv_id":"2502.10179","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Local detailed balance implies a hierarchy of identities relating nonlinear response memory kernels to cumulants of the conjugate observable, extending the fluctuation dissipation theorem to higher orders.","lead":"This paper derives exact identities connecting nonlinear response memory kernels to equilibrium and non-equilibrium fluctuation cumulants in driven systems, with the fluctuation dissipation theorem as the lowest order member. The identities give simulators and experimentalists new consistency checks for nonlinear response measurements, and they hold for both overdamped and underdamped Brownian dynamics.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Identity (4) omits the cubic term D''' of the time-symmetric action; for generic nonlinear coupling this term enters Gamma(0,3) and cannot cancel on the RHS, so the central claim is unsupported.","rationale":"The reader's weakest_assumption focuses on local detailed balance and the convergence of the Volterra expansion. My concern is different and more load-bearing: even granting local detailed balance, the expansion of the time-symmetric action D is algebraically incomplete at third order. The third-order identity (4) is the first place where D''' can appear, and the paper's explicit kernel expressions omit it. This is not a convergence question; it is a missing term of the same order as the other terms in Eq. (4). The simulation tests cannot detect the omission because the chosen trap potential makes D''' = 0. If the D''' term is indeed nonzero and uncancelled, the general nonlinear-coupling claim is false, although the lower-order identities and the restricted quadratic-coupling case may still be valid. A direct re-derivation or a simulation with quartic x-y coupling would settle the point. I therefore disagree with the reader's identification of the weakest assumption and recommend rejecting the general claim as stated; a conditional acceptance would require either proving that D''' contributions cancel or explicitly restricting the theorem to the class where D''' = 0.","tokens_in":15673,"tokens_out":27708,"duration_ms":305088,"concrete_test":"Re-derive Gamma(0,3) keeping all third-order terms in D, including the missing -(1/6)<D''' ; B_0>_c term, for an overdamped Langevin model with U(y,x) = y^4/4 + (g/2) x^2 y^2 + (1/2) kappa (y-x)^2, and check whether Eq. (4) still holds. If the D''' contribution survives, the identity is false. Equivalently, simulate this model following the paper's protocol, extract the four kernels in Eq. (4) with the Laplace/symmetrized procedure, and compare; a mismatch confirms the concern.","verdict_should_be":"REJECT","load_bearing_attack":"The derivation of the third-order identity (4) expands the time-symmetric action only through second order in the protocol velocity: D = D' + D''/2 + ... (Eqs. (16)-(17)), and the explicit kernel Gamma(0,3) in Appendix A contains no D''' term. This is not a harmless truncation. For a generic nonlinear coupling U(y_s,x_s), the symmetric part of the Onsager-Machlup action contains a cubic term D''' proportional to an integral of three powers of x_dot with coefficient D_{s1,s2,s3}; for example, an overdamped particle with U = y^4/4 + (g/2)x^2 y^2 produces such a term from expanding [(dU/dy)^2] in x_s - x_t. The third-order response of the mean, Gamma(0,3), receives a contribution -(1/6) <D''' ; B_0>_c from the cubic term in e^{-D+S/2}, but Eq. (22) and Eq. (A4) do not include it. The kernels on the right-hand side of Eq. (4), Gamma(1,2), Gamma(2,1), and Gamma(3,0), are of lower order in x_dot and therefore cannot contain D'''; no cancellation is possible. The simulations use U_ext = (1/2) kappa (y-x)^2, for which dU/dy is linear in x and D''' = 0, so the numerical tests do not probe this term. Thus the central claim, stated for arbitrary nonlinear coupling, is not established and appears false for generic U.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives identities for nonlinear memory kernels appearing in Volterra expansions of non-equilibrium cumulants, with the stated goal of covering protocols that couple nonlinearly to the system. The identities, Eqs. (2)-(4), relate the nonlinear response of a mean observable to cumulants of the conjugate force and to each other, generalizing the fluctuation-dissipation theorem. The derivation uses a path-integral expansion around the equilibrium state at the final protocol value, assuming local detailed balance and taking the initial time to minus infinity with x_{t0}=x_t. The authors test the identities numerically for overdamped and underdamped coupled Brownian particles with a harmonic trap potential U_ext = (1/2) κ (y-x)^2.","tokens_in":16065,"tokens_out":10201,"duration_ms":98301,"significance":"If the identities held as stated, they would be a substantial extension of the fluctuation-dissipation theorem to nonlinear response, connecting each order of response to equilibrium and non-equilibrium cumulants of the conjugate observable, and would be applicable to a broad class of driven systems with nonlinear coupling. The paper also provides explicit kernel expressions in Appendix A and simulation tests for both overdamped and underdamped dynamics, which are useful. However, as detailed below, the central third-order identity is not established for the claimed generality, because the derivation omits the cubic term of the time-symmetric action. Since the numerical tests use a potential for which this term vanishes, they do not probe the problem, and the main claim is therefore unsupported in its present form.","major_comments":[{"comment":"The derivation of the third-order identity (4) ignores the cubic term D''' of the time-symmetric action. The expansion of D is truncated at second order in Eqs. (16)-(17), and the explicit kernel Γ(0,3) in Eq. (A4) contains no contribution from D'''. This is not a harmless truncation: for a generic potential U(y_s,x_s) that is nonlinear in x_s, the symmetric part of the Onsager-Machlup action contains a term of third order in \\dot{x}. For example, in the underdamped action of Appendix B, the difference of (∂_y U)^2 between x_s and x_t produces terms cubic in x_s - x_t when U has third derivatives in x, such as U = y^4/4 + (g/2) x^2 y^2. This D''' term contributes directly to Γ(0,3) through the expansion of e^{-D+S/2}, but it is absent from Eq. (A4) and from Eq. (22). Since the right-hand side of Eq. (4) contains only kernels Γ(1,2), Γ(2,1), Γ(3,0), which are of lower order in \\dot{x}, no cancellation can remove this contribution. Thus Eq. (4) appears false for generic nonlinear coupling. The numerical tests use U_ext = (1/2) κ (y-x)^2, for which dU_ext/dy is linear in x and hence D''' = 0, so the simulations do not detect the problem.","section":"§III.B and Appendix A, Eq. (4) vs Eq. (A4)"},{"comment":"The expansion of the mean response to third order is incomplete for the same reason: it includes the term (1/3)⟨S''' ; O_t⟩_eq but no analogous ⟨D''' ; O_t⟩_eq term. The identities (23)-(26) for equilibrium correlation functions also omit any D''' contribution. The paper asserts that the form of D is not needed to obtain the identities, but at third order the form of D is essential. This missing term propagates to the cumulant relations in Section IV, in particular Eqs. (29) and (39), which therefore are not justified for the stated general setting.","section":"§III.B, Eq. (22) and Eqs. (23)-(26)"},{"comment":"The algebraic steps leading from the kernel expressions in Appendix A to the identities (2)-(4) are not shown. Because the identities rest on delicate cancellations of the symmetric action, and because the kernel expressions themselves appear to omit D''' (see above), a step-by-step verification is necessary. Without it, the claim that the identities 'follow from the kernel expressions' cannot be checked, and the reader cannot distinguish a genuine cancellation from an accidental omission.","section":"§III.B, 'The derivation follows from the kernel expressions given in Appendix A'"}],"minor_comments":[{"comment":"There is a typo: 'indentities' should be 'identities'.","section":"§II, text before Eq. (2)"},{"comment":"The notation for symmetrization over time arguments on the left-hand side is not spelled out: the kernels are symmetric in their s-arguments by construction, but the identities as written require an explicit symmetrization with respect to interchanges of s and t indices. This should be defined precisely to avoid ambiguity.","section":"Eqs. (3) and (4)"},{"comment":"The description 'kernels with all time arguments integrated over' is imprecise; it would be clearer to state that the Laplace-transformed kernels are evaluated at z=0 for all arguments.","section":"§V.B, figure captions"}],"recommendation":"reject","confidential_remarks":"The central higher-order identity appears to be false for generic nonlinear coupling because the cubic term of the time-symmetric action is omitted. This is a load-bearing error, and the numerical tests do not cover the failing regime. If the authors restrict the claim to potentials that are linear in the protocol or otherwise ensure D'''=0, the paper's scope would be substantially reduced, but the current title and abstract claim nonlinear coupling as a key feature. I therefore recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: one good new result, one overreach. The second-order identity (3) is new, follows from local detailed balance, and checks out in both overdamped and underdamped simulations. That part deserves credit. The first identity (2) is just FDT, as stated.\n\nThe soft spot is identity (4). The derivation expands the entropy S to third order in the protocol velocity, but the time-symmetric action D only to second order. For a generic nonlinear coupling U(y,x), D has a cubic piece D'''. The example U = y^4/4 + (g/2)x^2 y^2 produces one when you expand [(∂U/∂y)^2] in x_s − x_t. That piece contributes directly to Γ(0,3) and cannot be canceled by the kernels on the right side of (4), which are of lower order in x_dot. The paper's simulations use Uext = (1/2)κ(y−x)^2, for which D''' = 0, so the numerics never test this. As written, identity (4) is not established for arbitrary nonlinear coupling and is probably false in general. The second-order identity survives because D'' is included; the missing convergence proof for the Volterra expansion is a lesser concern.\n\nNet: the framework is useful and identity (3) is a real result, but the headline claim of nonlinear identities through third order is load-bearing and currently unsupported. A serious referee should engage, and the authors should either prove that D''' cancels or restrict the third-order claim to couplings with D''' = 0. I would not cite it as is.","headline":"The third-order identity (4) omits the cubic term D''' of the time-symmetric action, so the general claim is unsupported; the second-order identity (3) is the solid new result.","tokens_in":16497,"tokens_out":9849,"would_cite":false,"duration_ms":97049,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["05.40.-a","05.70.Ln"],"model":"deepseek-v4-flash","headline":"Under a single local-balance condition, nonlinear response kernels are fixed by covariance and higher cumulants.","keywords":["nonlinear memory kernels","Volterra series","fluctuation-dissipation theorem","local detailed balance","nonlinear response","non-equilibrium cumulants","driven Brownian particles"],"falsifier":"A direct test would use a driven probe in a bath whose hidden degrees of freedom are not equilibrated, such as a colloidal particle in an active or sheared suspension: extract the memory kernels from force cumulants under a constant-speed protocol and check whether Eq. (3) holds to within sampling error; a systematic mismatch that grows with protocol velocity would refute the universality of the identities, while a nonzero second-order deviation for Gaussian observables would challenge Eq. (31).","tokens_in":15526,"feed_emoji":"⚖️","tokens_out":9443,"duration_ms":88748,"temperature":0.7,"pith_summary":"This paper aims to show that the nonlinear response of a system driven by a time-dependent protocol is not free: order by order, the memory kernels that appear in a Volterra expansion of response are fixed by covariances and higher cumulants of the observable conjugate to the driving. The two new identities, Eqs. (3) and (4), relate the second- and third-order response kernels to symmetrized combinations of the covariance kernels and equilibrium third and fourth cumulants, with the fluctuation-dissipation theorem as the first-order case. They hold whenever local detailed balance applies, even when the protocol couples nonlinearly to the system, and they can be rewritten as a direct series relation between nonequilibrium cumulants. The authors verify the identities in simulations of driven overdamped and underdamped Brownian particles in a nonlinear bath, using both equilibrium cumulants and directly measured force statistics.","feed_headline":"One balance law fixes nonlinear response from fluctuations","feed_subtitle":"Two new identities extend the fluctuation-dissipation theorem; simulated Brownian particles confirm them.","key_machinery":"The machinery is a Volterra expansion of nonequilibrium cumulants in powers of the protocol velocity $\\dot{x}_s$ around the equilibrium state at the final protocol value, combined with a decomposition of the path action into a time-antisymmetric part $S$ and a time-symmetric part $D$. Local detailed balance identifies $S$ with the entropy production from the first law, Eq. (13), which is known explicitly in terms of the conjugate observable $F(y_s,x_s)$; $D$ is not known in general but cancels when the response expressions are compared. The explicit kernel formulas in Appendix A, together with the identities (23)-(26) for equilibrium correlations of $S$ and $D$, then give Eqs. (2)-(4). Because the cancellations do not require an explicit form of $D$, the identities are insensitive to dynamical details. The definition of the kernels includes the case $m=0$ (mean response), and the observable $B$ may equal $F$.","core_discovery":"The central claim is that under local detailed balance the memory kernels $\\Gamma^{(m,n)}$ defined by Eq. (1) are constrained by exact identities, so that every order of nonlinear response is expressible through lower-order response kernels and equilibrium fluctuation cumulants. Explicitly, $\\Gamma^{(0,1)}=\\Gamma^{(1,0)}$ (the fluctuation-dissipation theorem), $\\Gamma^{(0,2)}_{s_1,s_2}=\\frac12\\sum_{\\pi\\in S_2}\\Gamma^{(1,1)}_{s_{\\pi(1)};s_{\\pi(2)}}-\\frac12\\Gamma^{(2,0)}_{s_1,s_2}$, and $\\Gamma^{(0,3)}$ is the corresponding symmetrized combination of $\\Gamma^{(1,2)}$, $\\Gamma^{(2,1)}$, and $\\Gamma^{(3,0)}$. In cumulant form, the mean $\\beta\\langle B_t\\rangle$ equals $\\beta\\langle B\\rangle_{\\rm eq}$ plus integrals of the covariance, third, and fourth cumulants of the conjugate observable $F$, Eq. (28), and the deviation from the fluctuation-dissipation form starts at second order and is controlled by non-Gaussian fluctuations, Eq. (31). This remains true when the energy $U(y,x)$ depends nonlinearly on the protocol $x$, because the expansion includes the resulting derivatives of $F$ in the action. The identities are tested for $B\\equiv F$ in a two-particle model with a periodic interaction potential, for overdamped and underdamped dynamics.","pith_inferences":["Editorial inference: if these identities survive at higher orders and in multidimensional protocols, the response functions of a driven system cease to be independent material properties; each order is slaved to fluctuation statistics at the same total order, giving nonequilibrium thermodynamics a predictive closure similar to equilibrium fluctuation relations.","Editorial inference: the identities can serve as a fluctuation-based detector of local equilibration — in a system with hidden degrees of freedom that are themselves driven, Eq. (3) should fail in a velocity-dependent way, so measuring the mismatch gives a quantitative probe of how far the hidden bath is from equilibrium.","Editorial inference: applying the same cumulant relation to a multidimensional protocol, such as several trap coordinates or time-dependent particle interactions, would test whether the scalar time-ordering in Eqs. (15) is the only structure needed, since the symmetrization over time arguments would have to be supplemented by rotations in protocol space."],"forward_implications":["The fluctuation-dissipation theorem is the first term of a hierarchy; the second and third response orders become measurable from covariance and cumulant data, including equilibrium third and fourth cumulants.","For any observable $B$, the mean response can be reconstructed without knowing the dissipative time-symmetric part of the dynamics, since only equilibrium averages and cumulants of the conjugate force enter.","Equations (31) and (39) give a quantitative measure of the failure of the fluctuation-dissipation theorem at second order, tied to the equilibrium third cumulant, so non-Gaussian fluctuations are directly responsible for the breakdown.","If $B$ and $F$ have Gaussian statistics, the displayed higher-order corrections vanish, so the mean response equals the integrated covariance to the stated order; non-Gaussianity is the source of the discrepancy.","The pattern of Eqs. (2)-(4) suggests a general-order identity, Eq. (40), although the paper explicitly verifies only up to third order."],"supporting_citations":[{"why":"Supplies the Volterra expansion formalism and the explicit overdamped kernel expressions used as a cross-check in the simulations.","marker":"[56]"},{"why":"Provides the response-theory route and the limiting choice $x_{t0}=x_t$ that simplifies the action to Eq. (13).","marker":"[50]"},{"why":"Defines local detailed balance and lists counterexamples; this is the assumption on which all identities rest.","marker":"[61]"},{"why":"Introduces the decomposition of the path action into time-symmetric and time-antisymmetric parts used in the derivation.","marker":"[62]"},{"why":"Shows similar kernel relations that motivate the conjectured general-order identity, Eq. (40).","marker":"[64]"},{"why":"States the fluctuation-dissipation theorem recovered as the lowest-order identity.","marker":"[59]"}],"fun_headline_variants":["Nonlinear response tied to equilibrium fluctuations by exact identities","Memory kernels obey new identities from detailed balance","Beyond FDT: exact identities link nonlinear response to cumulants","Driven Brownian particles confirm new response identities","Volterra kernels constrained by detailed balance identities"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is local detailed balance—that the energy lost by the driven system is dissipated into hidden degrees of freedom that remain equilibrated, so the antisymmetric part of the action equals the entropy production of the first law—and, secondarily, that the Volterra expansion in protocol velocity converges.","fun_headline_variants_meta":{"raw":{"variants":["Nonlinear response tied to equilibrium fluctuations by exact identities","Memory kernels obey new identities from detailed balance","Beyond FDT: exact identities link nonlinear response to cumulants","Driven Brownian particles confirm new response identities","Volterra kernels constrained by detailed balance identities"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000142,"raw_usage":{"total_tokens":1172,"prompt_tokens":952,"completion_tokens":220,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":147}},"tokens_in":568,"tokens_out":220,"duration_ms":2715,"temperature":1.0,"reasoning_tokens":147,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T19:05:46.286235+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test would use a driven probe in a bath whose hidden degrees of freedom are not equilibrated, such as a colloidal particle in an active or sheared suspension: extract the memory kernels from force cumulants under a constant-speed protocol and check whether Eq. (3) holds to within sampling error; a systematic mismatch that grows with protocol velocity would refute the universality of the identities, while a nonzero second-order deviation for Gaussian observables would challenge Eq. (31).","supporting_citations":[{"cited_title":"Caspers and M","cited_arxiv_id":null,"evidence_quote":"Supplies the Volterra expansion formalism and the explicit overdamped kernel expressions used as a cross-check in the simulations."},{"cited_title":"Kr¨ uger and C","cited_arxiv_id":null,"evidence_quote":"Provides the response-theory route and the limiting choice $x_{t0}=x_t$ that simplifies the action to Eq. (13)."},{"cited_title":"Maes, SciPost Phys","cited_arxiv_id":null,"evidence_quote":"Defines local detailed balance and lists counterexamples; this is the assumption on which all identities rest."},{"cited_title":"Holsten and M","cited_arxiv_id":null,"evidence_quote":"Shows similar kernel relations that motivate the conjectured general-order identity, Eq. (40)."}],"review_version":1}