Extending the microcanonical-bath derivation of the Langevin equation beyond a single particle in a fast-equilibrating bath

The Langevin equation for a Brownian particle is conventionally derived microscopically by eliminating bath variables represented by a canonical ensemble, as in the Zwanzig–Mori projection formalism. Eichhorn (Entropy 26, 277, 2024) took a different route: the bath is an isolated Hamiltonian system that, at every mesoscopic time step, is described by the microcanonical ensemble at its current energy, and this construction alone was shown to yield the underdamped Langevin equation — with friction given by a microcanonical autocorrelation of the interaction forces, the Einstein fluctuation–dissipation relation traced to the microcanonical definition of temperature, the Stratonovich stochastic-calculus prescription identified, and the Fokker–Planck equation recovered. The derivation is, however, restricted to a single structureless particle in a bath that re-equilibrates within every mesoscopic time step, and the Gaussianity of the noise is only argued heuristically through an intermediate-time-scale central-limit-theorem reasoning. The problem is to settle these regimes within (or a justified modification of) the microcanonical framework, thereby fixing the range of validity of this route to the Markovian Langevin equation.

Background

The Langevin equation is the standard mesoscopic model of Brownian motion: the momentum $P$ of a small 'system', such as a colloidal sphere suspended in a fluid, evolves under deterministic friction and a rapidly fluctuating random force, while its position $Q$ drifts with that velocity. In the underdamped form used here, $\dot Q_i = P_i/M$ and $\dot P_i = -\sum_j \gamma_{ij}P_j/M + f_i +$ noise, with particle mass $M$, friction tensor $\gamma$, and conservative force $f$; the noise is taken to be Gaussian white noise whose strength relative to $\gamma$ is fixed by the Einstein (fluctuation–dissipation) relation $D = k_B T\,\gamma^{-1}$, which guarantees relaxation to equilibrium at the bath temperature $T$.

Microscopic derivations of this equation classically represent the bath canonically. Zwanzig's ensemble method (1960) and Mori's projection-operator formalism (1965) eliminate the bath variables to produce a generalized Langevin equation in which friction is a memory integral over autocorrelations of the interaction forces; the memory-free Markovian Langevin equation emerges only under additional weak-coupling or time-scale-separation assumptions. These routes presuppose a bath described by the canonical ensemble — the equilibrium state of a system weakly coupled to an unbounded reservoir at fixed temperature.

Eichhorn (2024) started instead from the bath's own perspective. Since the bath is assumed to remain in equilibrium at all times and is isolated from the rest of the universe, exchanging energy only with the (slow) system, it can be described at each mesoscopic time step $\Delta t$ by the microcanonical ensemble — the equilibrium measure assigning equal statistical weight to every phase-space point on the shell of the bath's current total energy. The construction rests on two assumptions: (i) the bath is in such an equilibrium state at every mesoscopic time, and (ii) there is a vast separation between the bath correlation time $\tau$ (of the order of the molecular collision time) and all system time scales, $\tau \ll \Delta t$. It also explicitly disregards perturbation-induced slow collective degrees of freedom in the bath. Under these assumptions, Hamilton's equations for bath and system together, averaged over the microcanonical bath ensemble, reproduce the standard underdamped Langevin equation: the friction coefficient is a time integral over the autocorrelation of the interaction forces, evaluated as a microcanonical average (a Green–Kubo-type formula); the Einstein relation follows from the microcanonical definition of temperature; the mid-point (Stratonovich) prescription emerges as the correct stochastic calculus; and the associated Fokker–Planck equation is recovered.

Two restrictions of this derivation motivate the present problem. First, it covers a single structureless Brownian particle. Second, its noise characterization is incomplete: only the leading moments and correlations of the momentum increments over $\Delta t$ are computed, while their Gaussianity is inferred from a central-limit-theorem argument that presumes an intermediate time scale $\Delta\tilde t$ with $\tau \ll \Delta\tilde t \ll \Delta t$ on which increments are effectively independent and identically distributed. This is an assumption about the deterministic bath dynamics — a mixing-type property — that is not part of the framework and fails, for example, for integrable baths.

The source itself identifies the regimes its derivation excludes, into which the approach should be extended or against which its validity should be delimited: (i) many Brownian particles, where a fluid bath mediates long-ranged 'hydrodynamic' interactions between the suspended particles (the flow field set up by one moving particle drags the others), while the straightforward generalization covers only direct pairwise potentials with these mediated interactions neglected; (ii) the presence of magnetic fields, where a charged particle feels the non-dissipative Lorentz force; (iii) other configuration spaces, namely rotational Brownian motion (diffusion of a particle's orientation, whose coordinates live on the rotation group rather than in $\mathbb{R}^3$) and, more generally, Brownian motion on manifolds; (iv) other Hamiltonian system–bath complexes, for example classical spin systems whose system variables are classical spins evolving under non-canonical (Poisson-bracket) equations of motion; and (v) baths whose perturbation by the system creates slow collective modes, for which the quasi-static equilibrium-bath assumption fails. Adjacent work on canonical oscillator baths — the universal slow-variable Langevin equation of Itami and Sasa, and the rotating/magnetic-bath results of Matevosyan and Allahverdyan, who found long-range-correlated noise and non-Gibbsian stationary states in precisely such regimes — indicates that several of these extensions genuinely strain the white-noise Markovian Langevin structure, which sharpens the question of where that structure survives.

Problem Statement

Within the microcanonical-bath framework of Eichhorn (2024) — an isolated Hamiltonian bath that, at each mesoscopic time step $\Delta t \gg \tau$, is described by the microcanonical ensemble at its current energy, evolved together with the system under Hamilton's equations — determine whether the memory-free white-noise Langevin equation with Einstein relation remains derivable, and in what form, in each of the following regimes that the source derivation excludes:

(a) Gaussianity of the fluctuations: replace the heuristic central-limit-theorem argument by a proof under explicitly stated dynamical assumptions on the bath (e.g. quantitative decay of correlations), or a disproof by an explicit microcanonical bath whose mesoscopic momentum increments are non-Gaussian while carrying the derived moments.

(b) Many Brownian particles: extend the derivation beyond the case in which bath-mediated ('hydrodynamic') interactions between the suspended particles are neglected, i.e. include the long-ranged coupling mediated by the bath.

(c) Magnetic fields: extend the derivation to a charged system particle subject to a static magnetic field (Lorentz force).

(d) Other configuration spaces: extend the derivation to rotational Brownian motion and, more generally, to system configuration spaces that are manifolds.

(e) Other Hamiltonian system–bath complexes: extend the derivation to system–bath splits whose 'system' is not a point particle, for example classical spin systems.

(f) Slow collective bath modes: extend the approach to, or delimit its validity against, baths in which perturbations by the system create slow collective degrees of freedom, so that the quasi-static equilibrium-bath assumption fails.

The common scientific objective is to fix the range of validity of the microcanonical route to the Markovian Langevin equation — and to characterize the effective dynamics (for example, generalized Langevin equations with memory, colored noise, or non-Gaussian fluctuations) that replaces the Markovian structure in regimes where it breaks down. A complete answer settles each of (a)–(f) by a derivation, a refutation, or a reduction to a characterized alternative effective dynamics within (or a justified modification of) the microcanonical framework.

References

1. R. Eichhorn, 'Derivation of the Langevin Equation from the Microcanonical Ensemble', Entropy 26(4), 277 (2024). DOI: 10.3390/e26040277, URL: https://doi.org/10.3390/e26040277
2. Yun Jeong Kang, Jae Won Jung, Sung Kyu Seo, Kyungsik Kim, 'On the Stochastic Motion Induced by Magnetic Fields in Random Environments', Entropy 27(4), 330 (2025). DOI: 10.3390/e27040330, URL: https://doi.org/10.3390/e27040330
3. Yun Jeong Kang, Sung Kyu Seo, Kyungsik Kim, 'Analytical Solutions for a Charged Particle with White, Thermal, and Active Noises in the Presence of a Uniform Magnetic Field', Entropy 28(7), 766 (2026). DOI: 10.3390/e28070766, arXiv:2602.18273, URL: https://doi.org/10.3390/e28070766
4. Ashot Matevosyan, Armen E. Allahverdyan, 'System-bath approach to rotating Brownian motion', Physical Review E 112, 064112 (2025). DOI: 10.1103/s7m5-4wvy, arXiv:2507.03202, URL: https://doi.org/10.1103/s7m5-4wvy
5. Ashot Matevosyan, Armen E. Allahverdyan, 'Lasting effects of static magnetic field on classical Brownian motion', Physical Review E 107, 014125 (2023). DOI: 10.1103/PhysRevE.107.014125, arXiv:2202.13160, URL: https://doi.org/10.1103/PhysRevE.107.014125
6. Ashot Matevosyan, 'Weak (non)conservation and stochastic dynamics of angular momentum', Journal of Statistical Mechanics: Theory and Experiment 2024, 053206 (2024). DOI: 10.1088/1742-5468/ad3cce, arXiv:2405.18660, URL: https://doi.org/10.1088/1742-5468/ad3cce
7. Masato Itami, Shin-ichi Sasa, 'Universal Form of Stochastic Evolution for Slow Variables in Equilibrium Systems', Journal of Statistical Physics 167, 46–63 (2017). DOI: 10.1007/s10955-017-1738-6, arXiv:1608.00371, URL: https://doi.org/10.1007/s10955-017-1738-6
8. Kunimasa Miyazaki, Kazuhiko Seki, 'Brownian motion of spins revisited', The Journal of Chemical Physics 108, 7052–7059 (1998). DOI: 10.1063/1.476123, URL: https://doi.org/10.1063/1.476123
9. Robert Zwanzig, 'Ensemble Method in the Theory of Irreversibility', The Journal of Chemical Physics 33, 1338–1341 (1960). DOI: 10.1063/1.1731409, URL: https://doi.org/10.1063/1.1731409
10. Hazime Mori, 'Transport, Collective Motion, and Brownian Motion', Progress of Theoretical Physics 33, 423–455 (1965). DOI: 10.1143/PTP.33.423, URL: https://doi.org/10.1143/PTP.33.423
11. Robert Zwanzig, 'Nonlinear generalized Langevin equations', Journal of Statistical Physics 9, 215–220 (1973). DOI: 10.1007/BF01008729, URL: https://doi.org/10.1007/BF01008729