Martingale properties of entropy production and a generalized work theorem with decoupled forward and backward processes

Tom Chou; Xiangting Li

REVIEW 2 major objections 4 minor 36 references

Martingale properties of entropy production and a generalized work theorem with decoupled forward and backward processes

T0 review · 2 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Pairing a Langevin trajectory with any positive backward density solving the reversed Fokker-Planck equation yields the conditional identity $E[e^{\theta_\tau}|x_0]=1$ for every bounded stopping time and every initial state.

desk verdict A genuinely useful generalization of martingale work theorems, with a correct central proof but a needed fix for the Novikov hypotheses and a factor-2 typo in Eq. (39). read the letter →

arxiv 2411.08311 v2 pith:MDDX7BGK submitted 2024-11-13 cond-mat.stat-mech math-phmath.MPmath.PR

Xiangting Li , Tom Chou This is my paper · ORCID

classification cond-mat.stat-mechmath-phmath.MPmath.PR MSC 60G4460H1082C31 PACS 05.40.-a05.70.Ln

keywords martingaleworktheoremfluctuationentropyproductionLangevindynamicsbackwardFokker-Planckequationstoppingtimefluctuation-dissipationrelation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper establishes that the exponential work-fluctuation identity is a martingale property that does not require the standard time-reversed ensemble: any positive density $\psi$ that solves the backward Fokker-Planck equation with an arbitrary initial condition yields $E[e^{\theta_\tau}|x_0]=1$ for every bounded stopping time $\tau$ and every initial state $x_0$, where $\theta_t = -\beta W_t + \beta \Delta H(x_t,t) + \beta \ln(\psi(x_t,t)/\psi(x_0,0))$. The proof uses the stochastic-calculus chain rule instead of path integrals, so it carries over to underdamped dynamics and to initial states that have no smooth density, such as a delta peak. This generalizes and strengthens the earlier ensemble-averaged work theorem, and a stationary choice of $\psi$ gives a simplified identity requiring only the stationary distribution. If correct, the result lowers the computational burden of free-energy estimates and makes work theorems applicable to a wider class of nonequilibrium and biological systems.

What carries the argument

The machinery is an exponential martingale built from a backward density. Given a probability density $\psi(x,t)$ that satisfies the backward Fokker-Planck equation $-\partial_t\psi = (1/\gamma)\nabla\cdot[(\nabla H-f)\psi] + (1/(\beta\gamma))\Delta\psi$ with an arbitrary initial condition, stochastic calculus gives $d e^{\theta_t} = e^{\theta_t} a_t\cdot dB_t$ with $a_t = \sqrt{2/(\beta\gamma)}[\beta(\nabla H-f)+\nabla\ln\psi]$, so the drift term cancels exactly when the diffusion coefficient equals $1/(\beta\gamma)$. The square-exponential condition on $a_t$ ensures that $e^{\theta_t}$ is a true martingale rather than a local martingale, and optional stopping then delivers $E[e^{\theta_\tau}|x_0]=1$ for any bounded stopping time. The same construction extends to underdamped dynamics by working with the phase-space Fokker-Planck equation and adding space inversion to the backward process.

What would settle it

Run the linear overdamped process in Eq. (51) with $k=\gamma=1$, $\beta=1$, and $D=1/(\beta\gamma)=1$, starting from a Gaussian distribution; take $\psi$ as the backward evolution of any Gaussian initial condition under the backward Fokker-Planck equation, stopped at a time $t_1$ earlier than any singularity of $\psi$, and estimate $E[e^{\theta_{t_1}}|x_0]$ from $10^7$ simulated trajectories binned by initial value. The paper's Eq. (24) predicts every bin equals 1; a bin whose estimate deviates by several standard errors would falsify the conditional work theorem.

Watch

Extended reading notes

Core claim

The central claim is the conditional identity $E[e^{\theta_\tau}|x_0]=1$, valid for overdamped and underdamped Langevin dynamics, for any bounded stopping time $\tau$, and for any choice of backward density $\psi$ that solves the backward Fokker-Planck equation with an arbitrary initial condition. The exponent is $\theta_t = -\beta W_t + \beta[H(x_t,t)-H(x_0,0)] + \beta \ln(\psi(x_t,t)/\psi(x_0,0))$, so it combines negative dimensionless work, dimensionless energy change, and the log-density ratio of the backward process along the forward trajectory. Because the equality holds for every starting $x_0$, it is strictly stronger than the usual averaged version in which the work theorem is integrated over the initial distribution, and it remains meaningful when the initial distribution is a delta function. Choosing $\psi$ to be the stationary distribution of the forward dynamics yields a simplified stationary work theorem; choosing $\psi$ to be the time-reversed forward density reproduces the earlier ensemble result. The derivation also shows that if the diffusion constant violates the fluctuation-dissipation relation $D = 1/(\beta\gamma)$, the exponential acquires a nonzero drift and the theorem fails, with the deviation expressed by an explicit formula.

Load-bearing premise

The load-bearing premise is that the chosen backward density $\psi$ stays positive, smooth, and satisfies the square-exponential condition up to the stopping time, so that $e^{\theta_t}$ is a genuine martingale rather than a local martingale; if $\psi$ acquires zeros or singularities (as in the Gaussian example of Eq. (50) at time $t_2$), or the drift grows too fast, the identity $E[e^{\theta_\tau}|x_0]=1$ can break down.

Editorial extensions

If this is right

For any bounded stopping time $\tau$ and any initial state $x_0$, $E[e^{\theta_\tau}|x_0]=1$, so work equalities need not be averaged over the initial ensemble; they apply to experiments that prepare a single initial state, including delta-peaked initial conditions where trajectory entropy is undefined.
Choosing $\psi$ as the stationary distribution of the forward dynamics turns the identity into $E[e^{-\beta(W_\tau - \Delta H_\tau + T\Delta S^*_\tau)}]=1$, which requires solving only a time-independent problem and therefore has lower computational cost than solving the time-dependent Fokker-Planck equation.
The same martingale construction works for underdamped Langevin dynamics with phase-space variables, including the deterministic zero-friction limit, where the information-entropy production vanishes along trajectories.
When the fluctuation-dissipation relation fails ($D \neq 1/(\beta\gamma)$), the exponential develops a nonzero drift and the generalized work theorem is violated; the explicit drift formula gives a quantitative measure of how far a process is from the equilibrium framework.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because $\psi$ is arbitrary, the family of identities can be used as an importance-sampling tool: choosing $\psi$ to reduce estimator variance would let one estimate free-energy differences from fewer trajectories than the standard choice requires.
The conditional form suggests a per-initial-state fluctuation theorem that may sharpen single-molecule measurements: binning repeated trajectories by their initial configuration should yield $E[e^{\theta_\tau}|x_0]=1$ in every bin, a testable prediction that the ensemble-averaged version cannot make.
The violation formula for $D \neq 1/(\beta\gamma)$ offers a model-free probe of active or non-thermal systems: measuring $E[e^{\theta_t}]$ as a function of time directly estimates how strongly the effective diffusion deviates from the fluctuation-dissipation relation, independent of any specific microscopic model.
The authors' observation that faster erasure costs more work suggests a concrete next step: apply the stationary-choice identity with a stopping time tied to erasure completion to obtain a finite-time bound on the erasure work, extending the classical $k_B T\ln 2$ limit.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

This paper is worth a serious look. The overdamped martingale identity in Eq. (24) is correct as far as the stochastic calculus goes, and it genuinely extends earlier work by conditioning on the initial state x0 and allowing arbitrary initial conditions for the backward process. The stationary case is a nice practical improvement, and the underdamped extension is real. The derivation via Itô calculus is cleaner than the usual path-integral route and avoids some of the rigor gaps.

The main soft spot is the Novikov condition. The authors assert it but never verify it for any ψ, and the theorem is stated for "any bounded stopping time" without restrictions on ψ. Their own Gaussian ψ-family in the drift-diffusion example has a singularity at t2; for a bounded stopping time that gets within ε of t2, the integrand a_t grows like 1/ε and the Novikov expectation is infinite. So the identity can fail for such stopping times. The paper should state explicit hypotheses—ψ positive and smooth on [0,τ], and an exponential integrability condition—and then the claim follows. This is a genuine gap in the theorem statement, but not a flaw in the core argument; the Discussion already acknowledges the Novikov requirement.

The underdamped derivation has a factor-2 typo in Eq. (39): the first line has −γ/m v·∇v lnψ dt, but completing the square with the noise term requires −2γ/m. The second line and Appendix A.2 have the correct coefficient, so the theorem is fine after correction. A referee should ask for this fix.

The "violation of work theorems" discussion is a bit loosely framed—when D≠1/(βγ), the model no longer satisfies the fluctuation-dissipation relation, so the failure of the equality is a failure of the theorem's assumptions, not a contradiction. That's a minor wording issue.

Overall, the central argument holds up. The paper is for the stochastic thermodynamics community, specifically people who work on martingale fluctuation theorems. It deserves a serious referee and publication after revision. I'd engage with it.

Referee Report

2 major / 4 minor

Summary. The paper develops a stochastic-calculus derivation of exponential martingales associated with entropy production for overdamped and underdamped Langevin dynamics. The central object is a backward density ψ(x,t) that solves a backward Fokker-Planck equation with an arbitrary initial condition, and the process θ_t = −βW_t + βΔH(x_t,t) + ln(ψ(x_t,t)/ψ(x_0,0)). The authors claim that e^{θ_t} is an exponential martingale and that, for any bounded stopping time τ, E[e^{θ_τ}|x_0] = 1 for every initial state x_0 (Eq. (24)). This is presented as a generalization of the Manzano et al. result, with a conditional version in Eq. (27), a stationary version in Eq. (28), and an underdamped counterpart in Eqs. (39)-(41). The paper also discusses violations of the fluctuation-dissipation relation and gives numerical examples based on a drift-diffusion model of kinetic proofreading and an Ornstein-Uhlenbeck process.

Significance. The conditional martingale identity is a genuine conceptual strengthening if it holds under stated conditions: it removes the initial-distribution average, allows decoupling of forward and backward processes, and gives a concrete route to work theorems for singular initial distributions. The Ito-calculus derivation is self-contained and avoids path integrals, and the appendix treatment of general diffusion coefficients is useful. The numerical examples illustrate the scope of the claimed identities. However, the theorem as stated is not fully proved: the Novikov condition is invoked but never verified, and the main-text underdamped derivation contains an algebraic inconsistency. These are repairable, but they are load-bearing for the central claims.

major comments (2)

[§2.2, Eqs. (11), (24); §3, Eq. (50)] The main identity is stated for every bounded stopping time τ and every backward density ψ solving Eq. (16), but the proof only establishes that exp(θ_t) is a nonnegative local martingale. The Novikov condition (11) is asserted but never verified for any of the ψ families considered, and no sufficient conditions on H, f, and ψ are given. This is not a purely technical gap: for the Gaussian family ψ(x,t)=p(x,t_2−t) in Eq. (50), ψ is singular at t=t_2, and for a bounded stopping time such as τ=t_2−ε the integrand a_s = sqrt(2/(γβ))[β(∇H−f)+∇lnψ] has quadratic variation of order (sup_s B_s^2)/ε over a short interval, so E[exp(1/2 ∫_0^τ a_s^2 ds)] is infinite for ε sufficiently small. In that situation the optional stopping equality can fail; for a nonnegative local martingale one only has E[e^{θ_τ}|x_0] ≤ 1. The theorem should be restated with explicit hypotheses: ψ strictly positive and smooth on [0,τ], and a verified exponential integrability condition for the chosen ψ. The Discussion's acknowledgment that the Novikov condition is an "inconvenient aspect" does not repair the theorem statement, which currently claims "∀ bounded stopping time" without these qualifications.
[§2.3, Eqs. (38)–(40)] The underdamped rearrangement is algebraically incorrect. Equation (38) contains a drift term −(γ/m) v·∇_v lnψ dt after cancellation of the two nγ/m terms, but the claimed representation dθ_t = a_t·dB_t − (1/2)||a_t||^2 dt with a_t = sqrt(2γβ)v_t + sqrt(2γ/(βm^2))∇_v lnψ_t has a cross term −(2γ/m) v·∇_v lnψ dt. The correct Ito expansion of lnψ from Eq. (36) indeed gives the coefficient −2γ/m, as shown in Appendix A.2, Eq. (A21), at D=1/(βγ); Equation (38) omits one contribution coming from the ∂_tψ term. Consequently Eq. (39) is missing the factor 2, and Eq. (40) does not follow from Eq. (39) as written. The appendix calculation repairs the result, but the main-text derivation of the underdamped martingale must be corrected.

minor comments (4)

[§4, Discussion] The sentence mentioning the "generalized Legendre-Frenchel transform" contains a typo (Fenchel) and the reference [35] is listed twice as "[35, 35, 36]".
[§2.3, Eq. (36)] The explanation of the two nγ/m terms in Eq. (38) is terse; writing out the divergence ∇_v·[(−γv + ∇_xU − f)ψ] and the Ito correction from (dv)^2 explicitly would help the reader verify the cancellation.
[§3, Fig. 4] The numerical claim that E[e^{θ_τ}] = 1 for all choices of t_2 would be more convincing with error bars or a convergence criterion, since the expectation involves rare large deviations of the exponential.
[§2.2, Eq. (27)] The conditional statement Eq. (27) contains the ratio ρ(x_0,0) in the denominator; the paper should state explicitly that the identity is meaningful only when ρ(x_0,0)>0, and explain how the Dirac-delta application mentioned in the text avoids this issue.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Eq. (24) is derived from the stated SDEs and backward PDE; the only flagged issue is a technical Novikov gap, not circularity.

full rationale

Eq. (24) is derived, not assumed. The paper defines θt through Eq. (18), lets ψ be any positive solution of the backward Fokker–Planck equation Eq. (16), and by Itô expansion (Eqs. 19–23) obtains de^{θt}=e^{θt}a_t·dB_t with zero drift; optional stopping then gives E[e^{θτ}|x0]=1. The backward PDE is a hypothesis from which the martingale property is proved, not the conclusion restated. Manzano's Eq. (8) is recovered only as the special case ψ(x,t1)=ρ(x,t1), so the paper is not renaming a known result. The only self-citation [26] is used to motivate the numerical proofreading example and is not load-bearing. The unverified Novikov condition is explicitly flagged in Eq. (11) and again in the Discussion as an inconvenient aspect; it is a technical gap, not a circular step. There are no fitted parameters presented as predictions. Hence score 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No free parameters are fitted; the central identity is parameter-free. The key input is the freely chosen backward density ψ, which is a mathematical degree of freedom rather than a fitted number. The main assumptions are regularity of ψ, the Novikov condition, and the fluctuation-dissipation relation.

assumptions (4)

standard math Stochastic calculus: Itô's formula, exponential martingales, optional stopping theorem under the Novikov condition.
Used throughout Eqs. (19) through (23) and Eqs. (37) through (40); standard background material.
domain assumption The Novikov condition E[exp(1/2 ∫ θ_s^2 ds)] < ∞ is assumed for each chosen backward process ψ and each finite horizon, ensuring e^{θt} is a true martingale.
Invoked after Eqs. (23) and (40) but never verified for the stationary, time-reversed, or decoupled ψ used in the theorems and examples.
domain assumption The backward process ψ solves the backward Fokker-Planck equation (16) and stays positive and twice differentiable along forward trajectories until the stopping time.
Required for the Itô expansion of dψ and for ln ψ; the Discussion notes that bounded or discrete spaces and Novikov conditions remain open issues.
domain assumption The fluctuation-dissipation relation D = 1/(βγ) is assumed in the main theorems (Eqs. (12) and (32)); the D ≠ 1/(βγ) section relaxes this at the cost of the martingale property.
The forward SDEs set the noise amplitude through β and γ; this is the physical equilibrium-bath assumption.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Martingale properties of entropy production and a generalized work theorem with decoupled forward and backward processes." pith.science (2026). https://pith.science/paper/MDDX7BGK

@misc{pith2026241108311,
  author       = {Pith},
  title        = {Pith review of: Martingale properties of entropy production and a generalized work theorem with decoupled forward and backward processes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MDDX7BGK}},
  note         = {Machine review of arXiv:2411.08311}
}

read the original abstract

By decoupling forward and backward stochastic trajectories, we construct a family of martingales and work theorems for both overdamped and underdamped Langevin dynamics. Our results are made possible by an alternative derivation of work theorems that uses tools from stochastic calculus instead of path-integration. We further strengthen the equality in work theorems by evaluating expectations conditioned on an arbitrary initial state value. These generalizations extend the applicability of work theorems and offer new interpretations of entropy production in stochastic systems. Lastly, we discuss the violation of work theorems in far-from-equilibrium systems.

Figures

Figures reproduced from arXiv: 2411.08311 by the authors.

**Figure 2.** Decoupling of the forward and backward processes. (a) [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** (a) Schematic representation of the continuum kinetic pr [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Numerical results for the quantities E[e θτ ] (red), E[θτ ] (blue), and E[∆Στ ] (green) for different choices of t2. The red line is at unity, consistent with our work theorem. By contrast, E[θτ ] and E[∆Στ ] vary with t2, shown by the blue and green curves, respective…

**Figure 5.** Figure 5: Numerical results of the exponentiated entropy produc [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 35 canonical work pages

[1]

Jarzynski

C. Jarzynski. Nonequilibrium equality for free energy diﬀerences . Physical Review Letters, 78(14):2690– 2693, 1997

work page 1997
[2]

Jarzynski

C. Jarzynski. Equilibrium free-energy diﬀerences from nonequilib rium measurements: A master- equation approach. Physical Review E , 56(5):5018–5035, November 1997

work page 1997
[3]

Sevick, R

E.M. Sevick, R. Prabhakar, Stephen R. Williams, and Debra J. Sear les. Fluctuation theorems. Annual Review of Physical Chemistry , 59(1):603–633, May 2008

work page 2008
[4]

G. E. Crooks. Entropy production ﬂuctuation theorem and the nonequilibrium work relation for free energy diﬀerences. Physical Review E , 60(3):2721–2726, 1999

work page 1999
[5]

Sekimoto

K. Sekimoto. Langevin equation and thermodynamics. Progress of Theoretical Physics Supplement , 130:17–27, 1998

work page 1998
[6]

U. Seifert. Entropy production along a stochastic trajectory and an integral ﬂuctuation theorem. Phys- ical Review Letters , 95(4):040602, 2005

work page 2005
[7]

Mesoscopic nonequilibrium thermodynamics of single mac romolecules and dynamic entropy- energy compensation

Hong Qian. Mesoscopic nonequilibrium thermodynamics of single mac romolecules and dynamic entropy- energy compensation. Physical Review E , 65(1):016102, December 2001

work page 2001
[8]

Fluctuation theorem for stochastic dynamics

Jorge Kurchan. Fluctuation theorem for stochastic dynamics. Journal of Physics A: Mathematical and General, 31(16):3719–3729, April 1998

work page 1998

Show all 36 references

[9]

Pekola, and ´Edgar Rold´ an

Gonzalo Manzano, Diego Subero, Olivier Maillet, Rosario Fazio, Jukk a P. Pekola, and ´Edgar Rold´ an. Thermodynamics of gambling demons. Physical Review Letters , 126(8):080603, 2021

2021
[10]

Sagawa and M

T. Sagawa and M. Ueda. Generalized Jarzynski equality under n onequilibrium feedback control. Physical Review Letters, 104(9):090602, 2010

2010
[11]

Second law of thermodynamics at stopping times

Izaak Neri. Second law of thermodynamics at stopping times. Physical Review Letters , 124(4):040601, January 2020

2020
[12]

Cataliotti, Andrea Trombettoni, Nicole Fabbri, and Stefano Gherardini

Santiago Hern´ andez-G´ omez, Francesco Poggiali, Paola Capp ellaro, Francesco S. Cataliotti, Andrea Trombettoni, Nicole Fabbri, and Stefano Gherardini. Energy excha nge statistics and ﬂuctuation theorem for non-thermal asymptotic states. arXiv, quant-ph:2404.05310v1, Apr 2024

2024 arXiv
[13]

Wiener and Feynman-path integrals and their a pplications

Sergio Albeverio. Wiener and Feynman-path integrals and their a pplications. In Proceedings of Symposia in Applied Mathematics , volume 52, pages 163–194. American Mathematical Society, 1997

1997
[14]

Mathematical Feynman path integrals and their application s

Sonia Mazzucchi. Mathematical Feynman path integrals and their application s. World Scientiﬁc, 2009

2009
[15]

Mathematical aspects of Feynman path integ rals, divergences, quantum ﬁelds and diagrams, and some more general reﬂections

Sergio Albeverio. Mathematical aspects of Feynman path integ rals, divergences, quantum ﬁelds and diagrams, and some more general reﬂections. In When Form Becomes Substance: Power of Gestures, Diagrammatical Intuition and Phenomenology of Space , pages 267–282. Springer, 2022

2022
[16]

Fluctuation theorem as a s pecial case of Girsanov theorem, 2023

Annwesha Dutta and Saikat Sarkar. Fluctuation theorem as a s pecial case of Girsanov theorem, 2023

2023
[17]

Brown and David A

Aidan I. Brown and David A. Sivak. Theory of nonequilibrium free e nergy transduction by molecular machines. Chemical Reviews, 120(1):434–459, August 2019

2019
[18]

Leighton, Jannik Ehrich, and David A

Matthew P. Leighton, Jannik Ehrich, and David A. Sivak. Inform ation arbitrage in bipartite heat engines. Physical Review X , 14(4):041038, November 2024

2024
[19]

Experimental demonstration of information-to-energy conversion and validatio n of the generalized Jarzynski equality

Shoichi Toyabe, Takahiro Sagawa, Masahito Ueda, Eiro Muneyu ki, and Masaki Sano. Experimental demonstration of information-to-energy conversion and validatio n of the generalized Jarzynski equality. Nature Physics, 6(12):988–992, November 2010. 18

2010
[20]

Shuoming An, Jing-Ning Zhang, Mark Um, Dingshun Lv, Yao Lu, Ju nhua Zhang, Zhang-Qi Yin, H. T. Quan, and Kihwan Kim. Experimental test of the quantum Jarzynsk i equality with a trapped-ion system. Nature Physics, 11(2):193–199, December 2014

2014
[21]

Eﬃciency ﬂuctu- ations and noise induced refrigerator-to-heater transition in info rmation engines

Govind Paneru, Sandipan Dutta, Takahiro Sagawa, Tsvi Tlusty , and Hyuk Kyu Pak. Eﬃciency ﬂuctu- ations and noise induced refrigerator-to-heater transition in info rmation engines. Nature Communica- tions, 11(1):1012, February 2020

2020
[22]

Martingales for physicists: a treatise on stochastic the rmodynamics and beyond

´Edgar Rold´ an, Izaak Neri, Rapha¨ el Chetrite, Shamik Gupta, Simone Pigolotti, Frank J¨ ulicher, and Ken Sekimoto. Martingales for physicists: a treatise on stochastic the rmodynamics and beyond. Advances in Physics , 72(1–2):1–258, April 2023

2023
[23]

Brownian motion and stochastic calculus , volume 113

Ioannis Karatzas and Steven Shreve. Brownian motion and stochastic calculus , volume 113. Springer, 2014

2014
[24]

Brian D. O. Anderson. Reverse-time diﬀusion equation models. Stochastic Processes and their Appli- cations, 12(3):313–326, 1982

1982
[25]

J. J. Hopﬁeld. Kinetic proofreading: A new mechanism for reduc ing errors in biosynthetic processes requiring high speciﬁcity. Proceedings of the National Academy of Sciences , 71(10):4135–4139, 1974

1974
[26]

Reliable ligand discrimination in stochast ic multistep kinetic proofreading: First passage time vs

Xiangting Li and Tom Chou. Reliable ligand discrimination in stochast ic multistep kinetic proofreading: First passage time vs. product counting strategies. PLoS Computational Biology , 20(6):e1012183, 2024

2024
[27]

Mello, Divya N

David Hathcock, Qiwei Yu, Bernardo A. Mello, Divya N. Amin, Gera ld L. Hazelbauer, and Yuhai Tu. A nonequilibrium allosteric model for receptor-kinase complexes : The role of energy dissipation in chemotaxis signaling. Proceedings of the National Academy of Sciences , 120(42):...

2023
[28]

J´ er´ emie Klinger and Grant M. Rotskoﬀ. Universal energy-speed-accuracy trade-oﬀs in driven nonequi- librium systems. arXiv, cond-mat.stat-mech:2402.17931v2, Feb 2024

2024 arXiv
[29]

Dynamics of information erasu re and extension of Landauer’s bound to fast processes

Salambˆ o Dago and Ludovic Bellon. Dynamics of information erasu re and extension of Landauer’s bound to fast processes. Physical Review Letters , 128(7):070604, 2022

2022
[30]

Ray, and James P

Jinghao Lyu, Kyle J. Ray, and James P. Crutchﬁeld. Learning en tropy production from underdamped Langevin trajectories. Physical Review E , 110(6):064151, December 2024

2024
[31]

Casey O. Barkan. On the convergence of phase space distribu tions to microcanonical equilibrium: dynamical isometry and generalized coarse-graining. Journal of Physics A: Mathematical and General , 57(47):475001, 2024

2024
[32]

Finite -time Landauer principle

Karel Proesmans, Jannik Ehrich, and John Bechhoefer. Finite -time Landauer principle. Physical Review Letters, 125(10):100602, 2020

2020
[33]

Stochastic thermodynamics of ch emical reaction networks

Tim Schmiedl and Udo Seifert. Stochastic thermodynamics of ch emical reaction networks. The Journal of Chemical Physics , 126(4):044101, 2007

2007
[34]

Large deviations and ﬂuct uation theorems for cycle currents deﬁned in the loop-erased and spanning tree manners: A comparat ive study

Yuhao Jiang, Bingjie Wu, and Chen Jia. Large deviations and ﬂuct uation theorems for cycle currents deﬁned in the loop-erased and spanning tree manners: A comparat ive study. Physical Review Research, 5(1):013207, March 2023

2023
[35]

Ying-Jen Yang and Ken A. Dill. Deriving the forces of nonequilibria f rom two laws. arXiv, cond- mat.stat-mech:2410.09277v1, Oct 2024

2024 arXiv
[36]

Universal non-equilibrium response t heory beyond steady states

Jiming Zheng and Zhiyue Lu. Universal non-equilibrium response t heory beyond steady states. arXiv, cond-mat.stat-mech:2403.10952v4, Mar 2024. 19

2024 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

Jarzynski

C. Jarzynski. Nonequilibrium equality for free energy diﬀerences . Physical Review Letters, 78(14):2690– 2693, 1997

work page 1997

[2] [2]

Jarzynski

C. Jarzynski. Equilibrium free-energy diﬀerences from nonequilib rium measurements: A master- equation approach. Physical Review E , 56(5):5018–5035, November 1997

work page 1997

[3] [3]

Sevick, R

E.M. Sevick, R. Prabhakar, Stephen R. Williams, and Debra J. Sear les. Fluctuation theorems. Annual Review of Physical Chemistry , 59(1):603–633, May 2008

work page 2008

[4] [4]

G. E. Crooks. Entropy production ﬂuctuation theorem and the nonequilibrium work relation for free energy diﬀerences. Physical Review E , 60(3):2721–2726, 1999

work page 1999

[5] [5]

Sekimoto

K. Sekimoto. Langevin equation and thermodynamics. Progress of Theoretical Physics Supplement , 130:17–27, 1998

work page 1998

[6] [6]

U. Seifert. Entropy production along a stochastic trajectory and an integral ﬂuctuation theorem. Phys- ical Review Letters , 95(4):040602, 2005

work page 2005

[7] [7]

Mesoscopic nonequilibrium thermodynamics of single mac romolecules and dynamic entropy- energy compensation

Hong Qian. Mesoscopic nonequilibrium thermodynamics of single mac romolecules and dynamic entropy- energy compensation. Physical Review E , 65(1):016102, December 2001

work page 2001

[8] [8]

Fluctuation theorem for stochastic dynamics

Jorge Kurchan. Fluctuation theorem for stochastic dynamics. Journal of Physics A: Mathematical and General, 31(16):3719–3729, April 1998

work page 1998