betaFeedback

WorksPhysics and Astronomy

Extended Onsager's variational principle for nonlinear chemical reaction kinetics in active matter

Abstract

Onsager's variational principle (OVP) constructs dissipative dynamics from free energy, dissipation, and input power, but deterministic rate laws alone do not identify fluctuations or chemical fuel consumption. We examine these distinctions for nonlinear reaction kinetics in active matter. Comparing Gaussian diffusion and Poisson reaction events clarifies when statistical fluctuations match a physical Rayleighian: the reaction dissipation potential must be consistent with counting statistics, local detailed balance links rate bias to chemical affinity, and kinetic scales remain constitutive inputs. Under these conditions, the normalized Rayleighian excess gives the leading rate-fluctuation cost in the frozen-state, many-event regime. Three models illustrate the consequences. Spatially uniform reactions exhibit protocol-dependent nonreciprocal response compatible with fixed-state OVP; different Rayleighian representations can select the same rate without specifying the same fluctuations. In an active emulsion, a finite fuel pool accounts for free-energy transfer and dissolution of a chemically maintained domain after replenishment stops. In a membrane-cytosol model, distinct reaction-channel realizations yield identical protein-density trajectories but different chemical input power. Under a separate kinetic assumption, an exact total-density equation still requires an initial rate and does not determine chemical work or fluctuations. These results clarify the physical and statistical inputs required for constructing reaction-transport models and interpreting their fluctuations.

Introduction

Chemical reactions maintain many active systems away from equilibrium and couple internal chemistry to motion, transport, and structure. Examples include the cytoskeleton and chemically maintained biomolecular condensates[1, 3] [2]. Reactions change composition and material properties, while mechanical forces and intermolecular interactions can change reaction rates. The modeling challenge is to combine these mechanisms under shared conservation laws and energy exchanges, and to identify which physical inputs determine the dynamics, its fluctuations, and the chemical power that sustains it.

Onsager’s variational principle (OVP) organizes dissipative dynamics through a scalar Rayleighian[4, 5, 6, 7]. For an isothermal system, its physical decomposition is

R=Φ+F˙−Pin.(1)\mathcal{R} = \Phi+ \dot{\mathcal{F}} - \mathcal{P}_{\mathrm{in}}. \tag*{(1)}

Here, Φ\Phi is the dissipation potential, F\mathcal{F} the retained free energy, and Pin\mathcal{P}_{\mathrm{in}} the input power; a dot denotes a time derivative. All three terms have power units, and positive Pin\mathcal{P}_{\mathrm{in}} denotes supply to the retained system. Their physical meaning comes from specified resistance, storage, and work mechanisms. Minimization over admissible rates at fixed state and controls selects the potential-generated dissipative dynamics. Interpreting the Rayleighian excess R−Rmin⁡\mathcal{R}-\mathcal{R}_{\min}, the difference from its fixed-state minimum, as a fluctuation cost requires matching to a specified stochastic mechanism[8].

Using established nonlinear variational and reaction-fluctuation structures[9, 10, 13, 14], we construct and compare coupled active-matter models sharing passive and fueled conversion channels. The comparisons make three distinctions explicit. First, a two-channel model shows how nonreciprocal net-affinity response under specified controls is compatible with partial-affinity response symmetries and a fixed-state dissipation potential. Second, reproducing the selected rate r∗r_* and dissipation power P\mathcal{P} does not determine the Rayleighian excess or its fluctuation interpretation. Third, the membrane–cytosol model shows that distinct passive and fueled channel realizations can have identical total attachment and detachment rates, and hence identical protein-density trajectory statistics, but different chemical input power, showing that composition dynamics alone cannot identify fuel consumption. The coupled applications further track free-energy exchange with a finite fuel pool and distinguish the loss of reaction-channel information from the elimination of dynamical state variables.

The practical value of OVP lies in constructing and coupling models. Free energy specifies interactions and storage, dissipation potentials specify resistance, and input power identifies external supply. These ingredients also underlie active-gel constitutive models[1] [2]. Related continuum descriptions couple drift, diffusion, and reaction in semiconductors[15], and ionic transport, electrostatics, and interfacial kinetics in electrochemistry[16]. Varying the Rayleighian under shared constraints incorporates reaction and transport consistently into their coupled equations and power balance. An equivalent direct construction gives the same equations; the advantage is their systematic assembly rather than a different prediction for already fixed inputs. Restricting trial fields before variation also provides a variational approximation method[17, 18].

Chemical reaction kinetics makes the distinction between deterministic dynamics and fluctuations particularly clear. The selected net rate rρ,∗=rρ+−rρ−r_{\rho,*}=r_\rho^+-r_\rho^- determines only the difference between the mean forward and reverse rates rρ±r_\rho^\pm, not their sum or the associated counting statistics. Different Rayleighians can therefore select the same deterministic rate while implying different fluctuation costs under their respective stochastic models. Onsager–Machlup theory connects near-equilibrium linear relaxation and Gaussian fluctuations[19, 20]. For nonlinear chemical kinetics, the question is when the same Rayleighian retains both physical rate selection and a statistical meaning. A nonideal free energy changes chemical driving forces; a nonquadratic dissipation potential changes the force–rate relation at fixed kinetic scales. These are distinct constitutive choices, and both must be specified when reactions couple to phase separation.

Nonlinear variational chemistry has substantial precedents, including thermokinetic and dissipation-potential approaches[13, 21] and formulations using separate forward and reverse rates[22]. The Gyarmati–Li generalized reciprocal relations propose symmetry of nonlinear differential response for specified conjugate forces and rates[13, 21]. Bataille, Edelen, and Kestin found chemical-kinetic counterexamples when net rates are differentiated with respect to net affinities at fixed complementary thermodynamic variables[23]. Lengyel obtained separate symmetric blocks for partial-affinity responses[22]; these statements concern different derivatives and are not contradictory. Reciprocal linear response can also persist in selected microscopically irreversible or driven models, whereas other driven states are nonreciprocal[24, 25, 26]. Neither outcome establishes a universal finite-affinity symmetry. For a common convex dissipation potential and its conjugate, the Gyarmati-type functional and Rayleighian excess coincide up to sign. Their variational equivalence must therefore be distinguished from identifying the physical inputs and fluctuation law of a reaction model.

The statistical connection also builds on established results. Grabert and Green extended short-time transition probabilities and path integrals to nonlinear irreversible processes[27]. Large-deviation formulations related trajectory probabilities to generalized dissipative dynamics[9, 28, 29]; for reversible chemical jump models, the event statistics yield nonquadratic reaction potentials[9, 10, 14]. For driven Markov systems, local detailed balance and time-symmetric kinetic activity enter nonlinear Onsager–Machlup structures[30, 31, 32]; flux-based variational formulations extend beyond detailed balance[33]. Kobayashi et al.[34] developed the Hessian geometry of the dual nonquadratic reaction potentials and associated entropy-production decompositions, and discussed quadratic alternatives representing the same kinetics. Energetic variational approaches provide related constructions coupling nonlinear reactions and diffusion[35, 36, 37]. The extension considered here uses these established structures, rather than proposing a distinct general extremum principle.

The retained variables and reaction channels carry different information. A description is dynamically sufficient at a stated resolution when its present variables and imposed controls determine the subsequent transition law without unresolved slow memory. An observed membrane density can fail this test when attachment depends on an evolving cytosolic pool[12, 38]. Conditional local equilibrium of eliminated thermal degrees of freedom permits retained free energies without requiring chemical equilibrium. Resolving reaction channels is a separate requirement: channels with the same retained-state change can exchange different chemical work, so the retained-state dynamics can close without channel labels while concealing fuel consumption[39, 40]. Neither dynamical closure nor timescale separation alone establishes local detailed balance[41]. The slow variables, thermodynamically distinct channels, and nonlinear dissipation with its event statistics are therefore complementary modeling choices.

These three modeling choices guide the analysis summarized in Figure 1. Section II establishes the statistical and thermodynamic foundations by comparing Gaussian and Poisson fluctuations and identifying the conditions under which a physical Rayleighian also determines their fluctuation costs. We then examine three models sharing passive and fueled conversion channels. Section III distinguishes nonlinear response across controlled states from fixed-state variational symmetry, and deterministic rates from their fluctuation costs. Section IV couples the reactions to nonideal phase separation and a finite fuel pool. Section V examines how channel aggregation and state-variable reduction affect membrane–cytosol dynamics and chemical-power accounting. Section VI discusses the broader modeling implications and limitations. Appendices A–C provide supporting response relations and stochastic matching conditions.

Figure 1. Nonlinear response, fluctuation costs, and chemically driven organization. (a) Near equilibrium, LijL_{ij} relates selected molar reaction-rate densities ri,∗r_{i,*} to molar affinities Aj\mathcal{A}_j and satisfies the Onsager reciprocal relations (ORR), Lij=LjiL_{ij}=L_{ji}. The solid curve schematically represents the response of a selected reaction-rate density ri,∗r_{i,*} to a molar affinity Aj\mathcal{A}_j under a chosen control protocol; the dashed line shows its near-equilibrium linear approximation. The questioned equality concerns a proposed extension to finite-driving differential response across compositions or controls; it need not hold and is distinct from the symmetry of a fixed-state trial-rate Hessian (Sec. III). (b) The physical Rayleighian combines the dissipation potential Φ\Phi, retained free-energy rate F˙\dot{F}, and chemical input power PinP_{\mathrm{in}}. For physically matched event statistics, the normalized Rayleighian excess L=(R−Rmin⁡)/(2kBT)\mathcal{L}=(\mathcal{R}-\mathcal{R}_{\min})/(2k_{\mathrm{B}}T) gives the leading rate-fluctuation cost per unit time. Here, the statistics are resolved Poisson counting with local detailed balance (LDB) in the frozen-state, many-event regime[9, 10]. kBk_{\mathrm{B}} is Boltzmann’s constant and TT the bath temperature. (c) A qualitative reaction–transport cycle couples A⇌BA\rightleftharpoons B conversion and transport between dilute and dense phases to fuel consumption F→WF\to W. Here, FF, WW denote fuel and waste. Adapted from the mechanism in Figure 1(a) of Haugerud et al. [11] under CC BY 4.0. (d) Membrane binding beneath a solution layer of fixed height hh. The bulk molar concentration Cb(x,z,t)C_b(x,z,t) gives the cytosolic amount per unit area c=∫0hCb dzc=\int_0^h C_b\,\mathrm{d}z, which exchanges with the membrane-bound molar surface density mm. The coordinates xx, zz are lateral and perpendicular to the membrane; rapid perpendicular mixing is assumed. The arrows r±r^{\pm} denote total one-way attachment and detachment rate densities. The passive and fueled channels, with net rates r0,rFr_0,r_F, produce the same conversion between cytosolic and bound protein states CC, MM, but carry different chemical work. The membrane schematic is informed by Weyer, Roth, and Frey [12].

Extending Onsager–Machlup theory to nonlinear chemical reaction kinetics

In this section, we first revisit Onsager and Machlup’s linear Gaussian theory[19, 20] to establish the deterministic and statistical roles of the physical Rayleighian. Our construction of nonlinear chemical kinetics rests on three complementary modeling choices: the retained slow variables, thermodynamically distinct reaction channels, and their nonlinear dissipation and event statistics. For a resolved Markovian reaction model, we derive the nonquadratic reaction-rate fluctuation cost from frozen-state Poisson counting statistics in the many-event regime and impose local detailed balance (LDB) to identify the statistical bias with the chemical affinity. The resulting physical Rayleighian retains the classical form of Eq. (1), R=Φ+F˙−Pin\mathcal{R}=\Phi+\dot{\mathcal{F}}-\mathcal{P}_{\mathrm{in}}, now with a nonquadratic dissipation potential Φ\Phi. It also preserves both roles: its minimization over admissible trial rates at fixed state and controls selects the deterministic kinetics, while its normalized excess (R−Rmin⁡)/(2kBT)(\mathcal{R}-\mathcal{R}_{\min})/(2k_{\mathrm{B}}T) gives the leading reaction-rate fluctuation cost per unit time, where Rmin⁡\mathcal{R}_{\min} is the instantaneous minimum.

Onsager–Machlup theory and the physical Rayleighian

Gaussian fluctuations and irreversible dynamics. Onsager and Machlup connected near-equilibrium Gaussian fluctuations with linear irreversible relaxation[19]. Following Wang, Qian, and Xu[8], we revisit this connection using an additive Itô stochastic model for NαN_{\alpha} independent retained slow coordinates α={α1,…,αNα}\boldsymbol{\alpha}=\{\alpha_{1},\ldots,\alpha_{N_{\alpha}}\}:

dα=LX(α,t) dt+2B dW.(2)\mathrm{d}\boldsymbol{\alpha}=L\boldsymbol{X}(\boldsymbol{\alpha},t)\,\mathrm{d}t+\sqrt{2}B\,\mathrm{d}\boldsymbol{W}. \tag*{(2)}

Here, LL is a constant response matrix and X(α,t)\boldsymbol{X}(\boldsymbol{\alpha},t) is a prescribed generalized force. The drift is linear in X\boldsymbol{X} but may be nonlinear in α\boldsymbol{\alpha}. The components of W\boldsymbol{W} are independent standard Wiener processes, and the constant noise-amplitude matrix BB defines the symmetric diffusion matrix D=BBTD=BB^{\mathsf{T}}, which is assumed positive definite. We first derive the transition density without imposing Onsager reciprocity on LL or the Einstein relation between DD and LL.

For the Markov model in Eq. (2), approximate the force over a sufficiently short interval [tn,tn+1][t_n,t_{n+1}] by its initial value Xn=X(αn,tn)\boldsymbol{X}_n=\boldsymbol{X}(\boldsymbol{\alpha}_n,t_n). This gives the Euler–Maruyama update

Δαn=LXnΔt+2ΔtBξn,(3)\Delta\boldsymbol{\alpha}_n=L\boldsymbol{X}_n\Delta t+\sqrt{2\Delta t}B\boldsymbol{\xi}_n, \tag*{(3)}

where Δαn=αn+1−αn\Delta\boldsymbol{\alpha}_n=\boldsymbol{\alpha}_{n+1}-\boldsymbol{\alpha}_n, Δt=tn+1−tn\Delta t=t_{n+1}-t_n, and ξn=[W(tn+1)−W(tn)]/Δt\boldsymbol{\xi}_n=[\boldsymbol{W}(t_{n+1})-\boldsymbol{W}(t_n)]/\sqrt{\Delta t}. These Gaussian vectors have zero mean and unit covariance, with each ξn\boldsymbol{\xi}_n independent of the history up to tnt_n. For this update, the conditional increment has mean LXnΔtL\boldsymbol{X}_n\Delta t and covariance 2DΔt2D\Delta t. Writing α˙n=Δαn/Δt\dot{\boldsymbol{\alpha}}_n=\Delta\boldsymbol{\alpha}_n/\Delta t for the measured step rate, the conditional transition density is

pEM(αn+1,tn+1∣αn,tn)=exp⁡[−Δt LD(α˙n)](4πΔt)Nα/2det⁡D,(4)p_{\mathrm{EM}}(\boldsymbol{\alpha}_{n+1},t_{n+1}\mid\boldsymbol{\alpha}_n,t_n)=\frac{\exp[-\Delta t\,\mathcal{L}_D(\dot{\boldsymbol{\alpha}}_n)]}{(4\pi\Delta t)^{N_{\alpha}/2}\sqrt{\det D}}, \tag*{(4)}

with the local fluctuation cost

LD(α˙)=14(α˙−LXn)TD−1(α˙−LXn).(5)\mathcal{L}_D(\dot{\boldsymbol{\alpha}})=\frac{1}{4}(\dot{\boldsymbol{\alpha}}-L\boldsymbol{X}_n)^{\mathsf{T}}D^{-1}(\dot{\boldsymbol{\alpha}}-L\boldsymbol{X}_n). \tag*{(5)}

Here, α˙\dot{\boldsymbol{\alpha}} denotes a trial rate and LD\mathcal{L}_D has units of inverse-time; the measured step rate α˙n\dot{\boldsymbol{\alpha}}_n does not imply differentiability of a Brownian trajectory. (4) is exact for the frozen update but generally only a short-time approximation to the transition density of Eq. (2). The Gaussian cost in Eq. (5) selects LXnL\boldsymbol{X}_n as the most probable step rate, but does not by itself identify the physical dissipation and energy exchanges.

Physical Rayleighian and deterministic rate. We now describe the deterministic energetics of the retained variables α\boldsymbol{\alpha}, with the free energy, resistance, and applied forces specified independently of the noise covariance. For an isothermal system, let X=−∇αF+Fa\boldsymbol{X}=-\nabla_{\boldsymbol{\alpha}}\mathcal{F}+\boldsymbol{F}_a, where F(α)\mathcal{F}(\boldsymbol{\alpha}) is the free energy of the retained system, with no explicit time dependence, and Fa(α,t)\boldsymbol{F}_a(\boldsymbol{\alpha},t) is a prescribed nonconservative force. For a trial rate α˙\dot{\boldsymbol{\alpha}}, the rate of free energy change and the input power are F˙=∇αF⋅α˙\dot{\mathcal{F}}=\nabla_{\boldsymbol{\alpha}}\mathcal{F}\cdot\dot{\boldsymbol{\alpha}} and Pin=Fa⋅α˙\mathcal{P}_{\mathrm{in}}=\boldsymbol{F}_a\cdot\dot{\boldsymbol{\alpha}}, respectively, so that F˙−Pin=−X⋅α˙\dot{\mathcal{F}}-\mathcal{P}_{\mathrm{in}}=-\boldsymbol{X}\cdot\dot{\boldsymbol{\alpha}}.

For time-reversal-even coordinates, we impose Onsager reciprocity, L=LTL=L^{\mathsf{T}}. This symmetry follows from microscopic reversibility in equilibrium linear response without time-reversal-odd controls[4, 5, 42]; beyond that regime, it is a constitutive assumption. The second law requires nonnegative dissipation; we assume LL to be positive definite so that L−1L^{-1} defines the resistance matrix. Integrating the linear resistance force L−1α˙L^{-1}\dot{\boldsymbol{\alpha}} over the trial rates gives the dissipation potential, normalized by Φ(0)=0\Phi(\boldsymbol{0})=0,

Φ(α˙)=12α˙TL−1α˙.\Phi(\dot{\boldsymbol{\alpha}})=\frac{1}{2}\dot{\boldsymbol{\alpha}}^{\mathsf{T}}L^{-1}\dot{\boldsymbol{\alpha}}.

Since ∂Φ/∂α˙=L−1α˙\partial\Phi/\partial\dot{\boldsymbol{\alpha}}=L^{-1}\dot{\boldsymbol{\alpha}}, the overdamped force balance between resistance and driving force is the stationarity condition of the physical Rayleighian

R(α˙;α,t)=Φ−X⋅α˙=Φ+F˙−Pin,(6)\mathcal{R}(\dot{\boldsymbol{\alpha}};\boldsymbol{\alpha},t)=\Phi-\boldsymbol{X}\cdot\dot{\boldsymbol{\alpha}}=\Phi+\dot{\mathcal{F}}-\mathcal{P}_{\mathrm{in}}, \tag*{(6)}

which has the classical form of Eq. (1)[8, 18, 19]. Its strict convexity gives the unique minimizing rate at fixed state and controls:

α˙∗=LX.(7)\dot{\boldsymbol{\alpha}}_{*}=L\boldsymbol{X}. \tag*{(7)}

This recovers the deterministic drift in Eq. (2). Substitution into Eq. (6) gives Rmin⁡(α,t)=−12XTLX\mathcal{R}_{\min}(\boldsymbol{\alpha},t)=-\frac{1}{2}\boldsymbol{X}^{\mathsf{T}}L\boldsymbol{X}. Along the selected deterministic evolution, the free-energy balance is

F˙=Pin−D,(8)\dot{\mathcal{F}}=\mathcal{P}_{\mathrm{in}}-\mathcal{D}, \tag*{(8)}

where D\mathcal{D} is the nonnegative dissipation power,

References

  1. [1]K. Kruse, J.-F. Joanny, F. Jülicher, J. Prost, and K. Sekimoto, Eur. Phys. J. E 16, 5 (2005). F. Jülicher, S. W. Grill, and G. Salbreux, Rep. Prog. Phys. 81, 076601 (2018).
  2. [3]C. A. Weber, D. Zwicker, F. Jülicher, and C. F. Lee, Rep. Prog. Phys. 82, 064601 (2019).
  3. [4]L. Onsager, Phys. Rev. 37, 405 (1931).
  4. [5]L. Onsager, Phys. Rev. 38, 2265 (1931).
  5. [6]M. Doi, J. Phys.: Condens. Matter 23, 284118 (2011).
  6. [7]M. Doi, Prog. Polym. Sci. 112, 101339 (2021).
  7. [8]H. Wang, T. Qian, and X. Xu, Soft Matter 17, 3634 (2021).
  8. [9]N. ShirAishi, An Introduction to Stochastic Thermodynamics: From Basic to Advanced, Fundamental Theories of Physics, Vol. 212 (Springer, Singapore, 2023).DOI
  9. [10]A. Mielke, R. I. A. Patterson, M. A. Peletier, and D. R. M. Renger, SIAM J. Appl. Math. 77, 1562 (2017).
  10. [11]S. Haugerud, H. D. Vuijk, J. Boeckhoven, and C. A. Weber, Phys. Rev. Lett. 136, 218401 (2026).
  11. [12]H. Wever, T. A. Roth, and E. Frey, Nat. Phys. 22, 94 (2026).
  12. [13]I. Gyarmati, Non-equilibrium Thermodynamics: Field Theory and Variational Principles (Springer, Berlin, Heidelberg, 1970).
  13. [14]M. Liero, A. Mielke, M. A. Peletier, and D. R. M. Renger, Discrete Contin. Dyn. Syst. S 10, 1 (2017).
  14. [15]W. van Roosbroeck, The Bell System Technical Journal 29, 560 (1950).
  15. [16]M. Z. Bazant, K. Thornton, and A. Ajdari, Physical Review E 70, 021506 (2004).
  16. [17]M. Doi, Chin. Phys. B 24, 020505 (2015).
  17. [18]M. Doi, J. Zhou, Y. Di, and X. Xu, Phys. Rev. E 99, 063303 (2019).
  18. [19]L. Onsager and S. Machlup, Phys. Rev. 91, 1505 (1953).
  19. [20]S. Machlup and L. Onsager, Phys. Rev. 91, 1512 (1953).
  20. [21]J. C. M. Li, J. Chem. Phys. 37, 1592 (1962).
  21. [22]S. Lengyel, J. Chem. Phys. 88, 1617 (1988).
  22. [23]J. Bataille, D. G. B. Edelen, and J. Kestin, J. Non-Equilib. Thermodyn. 3, 153 (1978).
  23. [24]D. Gabrielli, G. Jona-Lasinio, and C. Landim, Physical Review Letters 77, 1202 (1996).
  24. [25]X. Wang, J. Dobnikar, and D. Frenkel, Physical Review Letters 129, 238002 (2022).
  25. [26]M. De Corato and I. Pagonabarraga, The Journal of Chemical Physics 157, 084901 (2022).
  26. [27]H. Grabert and M. S. Green, Phys. Rev. A 19, 1747 (1979).
  27. [28]A. Mielke, M. A. Peletier, and D. R. M. Renger, Potential Anal. 41, 1293 (2014).
  28. [29]A. Mielke, M. A. Peletier, and D. R. M. Renger, J. Non-Equilib. Thermodyn. 41, 141 (2016).
  29. [30]C. Maes and K. Netočný, EPL 82, 30003 (2008).
  30. [31]M. Kaiser, R. L. Jack, and J. Zimmer, Journal of Statistical Physics 170, 1019 (2018).
  31. [32]C. Maes and K. Netočný, J. Non-Equilib. Thermodyn. 51, 279 (2026).
  32. [33]R. I. A. Patterson, D. R. M. Renger, and U. Sharma, J. Stat. Phys. 191, 18 (2024).
  33. [34]T. J. Kobayashi, D. Loutchko, A. Kamimura, and Y. Sugiyama, Phys. Rev. E 84, 033208 (2022).
  34. [35]Y. Wang, C. Liu, P. Liu, and B. Eisenberg, Phys. Rev. E 102, 062147 (2020).
  35. [36]Y. Wang and C. Liu, Entropy 24, 721 (2022).
  36. [37]W. Hao, C. Liu, Y. Wang, and Y. Yang, Math. Biosci. 385, 109453 (2025).
  37. [38]F. Brauns, J. Halatek, and E. Frey, Phys. Rev. X 10, 041036 (2020).
  38. [39]M. Polettini and M. Esposito, J. Chem. Phys. 141, 024117 (2014), see also erratum, J. Chem. Phys. 142, 229901 (2015).
  39. [40]R. Rao and M. Esposito, Phys. Rev. X 6, 041064 (2016).
  40. [41]D. Hartich and A. Godec, Phys. Rev. Research 5, L032017 (2023).
  41. [42]H. B. G. Casimir, Rev. Mod. Phys. 17, 343 (1945).
  42. [43]A. W. C. Lau and T. C. Lubensky, Phys. Rev. E 76, 011123 (2007).
  43. [44]J. D. Goddard, Ind. Eng. Chem. Res. 54, 4078 (2015).
  44. [46]H. Mori, Prog. Theor. Phys. 33, 423 (1965).
  45. [47]L. C. G. Rogers and J. W. Pitman, The Annals of Probability 9, 573 (1981).
  46. [48]J. Schmakenberg, Rev. Mod. Phys. 48, 571 (1976).
  47. [49]S. Lengyel and I. Gyarmati, J. Chem. Phys. 75, 2384 (1981).
  48. [50]U. Seifert, Eur. Phys. J. E 34, 26 (2011).
  49. [51]D. T. Gillespie, J. Chem. Phys. 113, 297 (2000).
  50. [52]R. I. A. Patterson and D. R. M. Renger, Math. Phys. Anal. Geom. 22, 21 (2019).
  51. [53]T. R. Gingrich, J. M. Horowitz, N. Perunov, and J. L. England, Physical Review Letters 116, 120601 (2016).
  52. [54]T. Schmidl and U. Seifert, J. Chem. Phys. 126, 044101 (2007).
  53. [55]C. Maes, SciPost Phys. Lect. Notes 32 (2021).
  54. [56]D. F. Anderson and T. G. Kurtz, in Design and Analysis of Biomolecular Circuits, edited by H. K. Oeppel, D. Densmore, G. Setti, and M. di Bernardo (Springer, New York, 2011) pp. 3–42.
  55. [58]M. Z. Bazant, M. S. Kilic, B. D. Storey, and A. Ajdari, Advances in Colloid and Interface Science 152, 48 (2009).
  56. [59]M. Z. Bazant, Acc. Chem. Res. 46, 1144 (2013).
  57. [60]J. D. Goddard, Combust. Sci. Technol. 195, 3627 (2023).
  58. [61]J. Bauermann, S. Laha, P. M. McCall, F. Jülicher, and C. A. Weber, J. Am. Chem. Soc. 144, 19294 (2022).
  59. [62]K. Yasuda, B. Zheng, Z. Xiong, Z. Hou, K. Ishimoto, X. Xu, D. Andelman, and S. Komura, Chin. Phys. B 35, 260750 (2026).
  60. [63]U. Seifert, Rep. Prog. Phys. 75, 126001 (2012).
  61. [64]J. M. Horowitz, J. Chem. Phys. 143, 044111 (2015).
  62. [65]M. Tena-Solsona, B. Rieß, R. K. Grötsch, F. C. Löhner, C. Wanzke, B. Käsdorf, A. R. Bausch, P. Müller-Buschbaum, O. Lieleg, and J. Boeckhoven, Nature Communications 8, 15895 (2017).
  63. [66]A. Mielke, Nonlinearity 24, 1329 (2011).
  64. [67]M. Arroyo, N. Walani, A. Torres-Sánchez, and D. Kaurin, in The Role of Mechanics in the Study of Lipid Bilayers, CISM International Centre for Mechanical Sciences, Vol. 577, edited by D. J. Steigmann (Springer, Cham, 2018) pp. 287–332.DOI
  65. [68]F. Avanzini, E. Penocchio, G. Falasco, and M. Esposito, J. Chem. Phys. 154, 094114 (2021).
  66. [69]F. Avanzini, T. Aslyamov, É. Fodor, and M. Esposito, J. Chem. Phys. 161, 174108 (2024).
  67. [70]M. E. Cates, É. Fodor, T. Markovich, C. Nardini, and E. Tjhung, Entropy 24, 254 (2022).
  68. [71]X. Zhao, S. Liese, A. Honigmann, F. Jülicher, and C. A. Weber, New J. Phys. 26, 103025 (2024).
  69. [72]S. Liese, X. Zhao, C. A. Weber, and F. Jülicher, Proc. Natl. Acad. Sci. U.S.A. 122, e2403083122 (2025).
  70. [73]D. Zhou and E. Frey, “Dynamical between chemical potential dynamics and reaction–diffusion systems,” arXiv:2605.15158 (2026), preprint, version 1.arxiv.org/abs/2605.15158
  71. [74]W. E. and B. Yu, Communications in Mathematics and Statistics 6, 1 (2018).
  72. [75]Y. Wang, B. Zou, J. Su, D. Wang, and X. Xu, Soft Matter 18, 6015 (2022).
  73. [76]Z. Li, B. Zou, H. Wang, J. Su, D. Wang, and X. Xu, Communications in Computational Physics 37, 353 (2025).
  74. [77]K. Yasuda, K. Ishimoto, and S. Komura, Phys. Rev. E 110, 044104 (2024).
  75. [78]B. Zheng, Z. Xiong, C. Li, Z. Hou, Z. Zhang, X. Xu, L.-S. Lin, K. Ishimoto, K. Yasuda, and S. Komura, Phys. Rev. E 114, L012103 (2026).
  76. [79]R. Wittkowski, A. Tiribocchi, J. Stenhammar, R. J. Allen, D. Marenduzzo, and M. E. Cates, Nature Communications 5, 4351 (2014).
  77. [80]E. Tjhung, C. Nardini, and M. E. Cates, Physical Review X 8, 031080 (2018). G. Pisegna and R. Golestanian, “Active phase separation from concentration-dependent diffusion in chemical networks,” (2026), arXiv:2610.30958 [cond-mat.stat-mech].
  78. [82]N. Nardin, É. Fodor, E. Tjhung, F. van Wijland, J. Tailleur, and M. E. Cates, Phys. Rev. X 7, 021007 (2017).
  79. [83]T. Markovich, É. Fodor, E. Tjhung, and M. E. Cates, Physical Review X 11, 021057 (2021).
  80. [84]D. Andrieux and P. Gaspard, J. Stat. Mech. 2007, P02006 (2007).
  81. [85]U. Basu, M. Krüger, A. Lazarescu, and C. Maes, Physical Chemistry Chemical Physics 17, 6653 (2015).
  82. [86]D. T. N. Chen, A. W. C. Lau, L. A. Hough, M. F. Islam, M. Goudian, T. C. Lubensky, and A. G. Yodh, Phys. Rev. Lett. 99, 148302 (2007).
  83. [87]C. Maggi, M. Paoluzzi, L. Angelani, and R. Di Leonardo, Scientific Reports 7, 17588 (2017).
  84. [88]I. A. Martínez, É. Roldán, J. M. R. Parrondo, and D. Petrov, Phys. Rev. E 87, 032159 (2013).
  85. [89]C. Maes and S. Steffenoni, Phys. Rev. E 91, 022128 (2015).

Paper details

vlnc:W_c85uzpks
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

ChatGPT

View statement
Looking for related papers…
Contents