betaFeedback

WorksPhysics and Astronomy

Thermodynamic cost of locally maintaining a steady state

Abstract

The Gibbs states of local Hamiltonians have short-range conditional correlations. To maintain a steady state with long-range conditional correlations, one might expect that a stochastic process must ``pay'' either in nonlocal moves or entropy production. Surprisingly, we find strictly local Gaussian processes that obey detailed balance and have unique steady states with arbitrarily long-range conditional correlations. We show that such processes must still produce entropy, if implemented using a local quadratic Hamiltonian coupled locally to equilibrium, rapidly relaxing baths. We lower-bound this entropy production in terms of the steady state's conditional mutual information (CMI); if the CMI decays exponentially with distance, the associated length-scale is called the Markov length ξξ. The entropy production per site per relaxation time obeys σ^≥cξ2\hatσ\ge c ξ^2 for large ξξ, and with cc an implementation-dependent constant. We also bound the tradeoff between coupling range and entropy production. We contrast these Gaussian processes with jump processes on discrete state spaces, for which, under certain irreducibility assumptions, steady states with long-range CMI are only possible if detailed balance is violated. We conjecture (and present evidence) that rapidly mixing local jump processes cannot stabilize steady states with long-range CMI.

Introduction

Gibbs states of local Hamiltonians are unusual states, and the canonical processes that stabilize them — Glauber dynamics and its variants — are unusual processes. In both cases what is unusual can be stated precisely. By the Hammersley–Clifford theorem, local Gibbs states are precisely the full-support distributions with the Markov property: if we divide the system into regions AA, BB, CC such that the Hamiltonian decomposes as HAB+HBCH_{AB} + H_{BC}, then conditioning on the state of BB removes all correlations between AA and CC — the conditional mutual information (CMI) I(A:C∣B)I(A:C \mid B) vanishes. The analogous constraint for processes like Glauber dynamics is detailed balance with respect to the target steady state. Detailed-balance processes model the interaction of the system with a heat bath; in the steady state the system has reached equilibrium with the bath, and no entropy is produced. One might expect that any local process producing no entropy in its steady state is of this Glauber type — in detailed balance with respect to some local Hamiltonian — and hence has a Markov steady state. Processes that violate detailed balance generate entropy, quantifiable in terms of the violation [1]. In many settings, however, the process is a means of creating a steady state, so the question of most interest is: how much entropy must be produced to maintain a given steady state with a local process?

The present work addresses this question, both for Gaussian processes (where explicit calculations are available) and for interacting lattice systems (where our results are more heuristic). The answer has some surprising features. First, discrete and continuous state spaces seem fundamentally different. One might have expected, on intuitive grounds, that a local process satisfying detailed balance stabilizes a local Gibbs state. While true for discrete jump processes with irreducible frozen blocks and full-support steady states, this is not true for general Gaussian processes: we present an explicit counterexample — a hyperuniform steady state with exactly Coulomb conditional correlations. Even though the counterexample satisfies detailed balance — so its entropy production is invisible within stochastic thermodynamics — we show that any implementation of it as a complete, locally interacting Gaussian system with bounded quadratic stiffness and finite-range linear couplings to heat baths must nevertheless produce entropy, and we lower-bound that entropy production in terms of the steady state’s long-range CMI.1 Suppose the CMI decays at long distances with a characteristic *Markov length ξ\xi: I(A:C∣B)∼ιe−∣B∣/ξI(A:C \mid B) \sim\iota e^{-|B|/\xi} with O(1) amplitude ι\iota. For Gaussian field theories coupled to multiple baths at different temperatures we show that, in cutoff units (we set the ultraviolet cutoff Λ=1\Lambda= 1 throughout), ξ≲rThottest/Tcoldest\xi\lesssim r\sqrt{T_{\mathrm{hottest}}/T_{\mathrm{coldest}}}, where rr is the coupling range. For our regulated counterexample at fixed range the entropy production is at least crξ2cr\xi^2 per site per characteristic relaxation time, and we give an explicit nearest-neighbor implementation attaining this scaling.2 In particular, the algebraic-CMI endpoint of this family requires infinite entropy production within this implementation class. That these processes are out of equilibrium is witnessed by a violation of the fluctuation-dissipation theorem: if they are local, their response to perturbations of the Hamiltonian necessarily reveals a mode-dependent temperature.

For lattice models with a finite local state space, the situation is strikingly different: assuming irreducibility of frozen blocks and full support, one can show that any process with a long-range CMI steady state must violate detailed balance; one can, moreover, quantify this violation (and thus the entropy production) under a uniform entropy-relaxation assumption for the frozen blocks. Whether this violation is achievable, we have not been able to prove. We conjecture that for strictly classical processes it is not: shot noise appears to fundamentally limit the achievable Markov length. We show, moreover, that for binary finite-state hidden Markov models with a strictly positive hidden transition matrix and an active odd-flip rule, a long-range CMI is not achievable; this suggests, but does not establish, that rapid-mixing processes cannot reach steady states with long-range CMI. Adapting a very recent result of Ref. [2], however, we observe that quantum spin systems with a discrete state space can violate this constraint, and, with a local open-boundary completion, maintain unique rapidly mixing steady states with long-range CMI. The thermal implementation we give has an infinite entropy production rate at this endpoint.

The rest of this work is structured as follows. § 2 briefly reviews entropy production and CMI. § 3 addresses linear stochastic field theories, for which all relevant quantities are Gaussian integrals. § 4 turns to local jump processes on discrete state spaces, focusing on processes that relax rapidly to their unique steady state. Finally, § 5 summarizes our results and lists open questions.

Context

What entropy production means

In the fundamental picture, the system is in contact with several thermal baths at distinct temperatures. In steady state the system itself is unchanged, but energy flows through it, from hot baths to cold ones. For two baths at temperatures T1>T2T_1 > T_2 coupled to a single overdamped degree of freedom in a unit-stiffness trap, with mobilities M1,M2\mathcal{M}_1,\mathcal{M}_2, the steady heat current is J=M1M2(T1−T2)/(M1+M2)J = \mathcal{M}_1\mathcal{M}_2(T_1-T_2)/(\mathcal{M}_1+\mathcal{M}_2) [we will define this carefully for many-body systems later], and the entropy production [3] is the Clausius rate3

σ=J(1T2−1T1)=M1M2(T1−T2)2(M1+M2)T1T2≥0.(1)\sigma= J\left(\frac{1}{T_2}-\frac{1}{T_1}\right) = \frac{\mathcal{M}_1\mathcal{M}_2(T_1-T_2)^2}{(\mathcal{M}_1+\mathcal{M}_2)T_1T_2} \ge0. \tag*{(1)}

This physical entropy production depends on the details of how the system is coupled to its baths. What we normally observe is the marginal process acting on the system alone, and this does not fix the implementation uniquely. Without further qualifications, therefore, “the entropy production needed to maintain a given state” is ill-defined. At one extreme, coupling the system to a mixture of two otherwise identical heat baths at different temperatures can give rise to a marginal process that is detailed balanced at some intermediate temperature: thus, a given physical implementation can produce “superfluous” entropy that can be eliminated without changing either the steady state or any of the observed rates. At the other extreme, any full-rank state is e−He^{-H} for some HH, and Glauber dynamics with respect to that HH maintains it with a single bath at zero entropy production. When we talk about the entropy production needed to maintain a steady state we really mean a constrained infimum over all local implementations that maintain e−He^{-H} as the system’s steady state.

The standard entropy production in stochastic thermodynamics is a different constrained infimum: over all implementations (local or not) that reproduce each rate of the marginal process. This infimum is [1, 4, 5, 6]

σ=∑x,uπ(x)Wu(x)ln⁡π(x)Wu(x)π(x+u)W−u(x+u)(2)\sigma= \sum_{x,u} \pi(x) W_u(x) \ln\frac{\pi(x) W_u(x)}{\pi(x+u) W_{-u}(x+u)} \tag*{(2)}

and is attained by assigning each inverse pair of transitions to a separate bath, whereupon the bath temperature and exchanged heat enter only through the combination q/T=ln⁡[Wu(x)/W−u(x+u)]q/T = \ln[W_u(x)/W_{-u}(x+u)] (splitting a pair across several baths only adds to the total, by the log-sum inequality). The concept we focus on and the standard one are similar in both being infima, but under inequivalent constraints. A given marginal process that is promised to have a local implementation must satisfy both constraints, and must therefore produce at least as much entropy as the more stringent of these bounds.

Conditional mutual information

Almost everywhere in this work we will be concerned with the classical CMI, I(A:C∣B)I(A:C|B), which is simply the average mutual information of the conditional distribution: ∑bpB(b)I(A:C)P(AC∣b)\sum_b p_B(b) I(A:C)_{P(AC|b)}. In terms of Shannon entropies one can write I(A:C∣B)=SAB+SBC−SB−SABCI(A:C|B) = S_{AB} + S_{BC} - S_B - S_{ABC}, where subscripts denote subsystems. This latter definition carries over to quantum systems if one replaces Shannon entropies with von Neumann entropies. Our main partitioning geometry is the annular one shown in Figure 1(a), where BB separates AA and CC by some distance ∣B∣|B| at its narrowest. For a full-support classical distribution on a finite graph, vanishing CMI for every separating tripartition is equivalent to a Gibbs representation by interactions on the graph cliques—this is the Hammersley-Clifford theorem. In particular, a range-rr Hamiltonian gives zero CMI for every complete shielding partition wider than rr. Hammersley-Clifford has not been fully generalized yet for quantum Hamiltonians, but see Refs. [7, 8, 9, 10]. In both classical and quantum settings, a small CMI implies efficient local reconstruction: specifically, if I(A:C∣B)≤e−∣B∣/ξI(A:C|B) \leq e^{-|B|/\xi} for fixed ξ\xi, one can approximately retrieve ρABC\rho_{ABC} after AA is erased, by a recovery channel acting on BB, with error controlled by the CMI [7, 11, 12]. Thus the CMI plays a central part in the theory of quantum error correction, mixed-state phases of matter [13, 14], and learnability transitions [15, 16]—including retrieval in neural-network diffusion models [17]. The ubiquity of the CMI in these settings makes it natural to ask how one might control it, and what resources such control would require.

Figure 1. (a) The annular tripartition used throughout: a buffer BB of width ww separates a region AA from the rest of the system CC; in d=1d=1 the annulus degenerates to the two buffer segments of the ring geometry used in panel (b). (b) Exact Gaussian conditional mutual information of the near-Coulomb state (5), computed from covariance determinants, with an infinitesimal regulator (C→C+T1C \to C + T_1, T1=10−6T_1 = 10^{-6}) keeping the CMI finite. d=1d=1: ring of N=512N=512 sites, blocks LA=64L_A=64 and 16, two-sided buffer of width ww per side. Markers: exact evaluation; lines: the determinant formula (23) of Appendix A. d=2d=2: torus of 40×4040 \times40 sites, AA a disk of radius 5, BB an annulus of width ww.

An important distinction between CMI and other standard information-theoretic quantities is that CMI does not necessarily decrease under marginalizing: a Markov chain acting on BB can either increase or decrease I(A:C∣B)I(A:C|B) [13, 18, 19, 20, 21]. In particular, marginalizing a Gibbs state will generically leave it non-Gibbs [22]: for example, the critical Gibbs states of a short-range Ising Hamiltonian has short-range CMI but long-range MI, so a Markov chain that simply deletes BB will replace the short-range CMI with the long-range MI. This is why it will be crucial in the rest of this work to focus on the case where one conditions completely on the state of BB.

What our results could not be

To orient the rest of the discussion, we note a few conjectures one might have been tempted to make but that are ruled out by elementary considerations or existing results.

  1. Entropy production is a rate and the Markov length is not, so any relation between them involves entropy produced per timescale. For steady states the natural timescale is the one on which local deviations from the steady state (of the kind the process continually generates) relax, not the global mixing time: the stationary entropy budget concerns the continual relaxation of fluctuations, rather than preparation from an arbitrary initial state. Because the relaxation time diverges at an equilibrium critical point, our bounds constrain critical processes only weakly; indeed, arbitrarily weak detailed-balance-violating perturbations of a critical point — producing arbitrarily little entropy — generate long-range CMI, as shown very recently in Ref. [23].

  1. Our results assume a unique steady state for the specified finite-volume dynamics; uniformity over volumes and frozen boundary conditions is stated separately. Dynamics with multiple steady states — conserving model-B dynamics, say — generate long-range steady-state CMI for free: under high-temperature charge-conserving equilibrium dynamics, any initial state of definite charge evolves to a distribution that is random apart from the global charge constraint, which suffices for long-range CMI, with no entropy produced.

  1. Our results concern steady states, not transients. In the recent literature on finite-time phase transitions — e.g., spontaneous strong-to-weak symmetry breaking of product states under stochastic dynamics (see [24] for a recent review) — the CMI diverges at the finite-time transition, but not in the steady state of the associated process.

  1. For discrete-time Markov chains it is straightforward to write down “prepare-and-replace” dynamics with long-range CMI in the steady state4: run stochastic dynamics from a fixed configuration up to the time at which the CMI diverges (as above), reset, and alternate indefinitely. The entropy accounting of such discrete-time protocols is deferred to future work.

  1. Any bound must rely, not just on the Markov length, but also on the amount of CMI that is present at large distances. An illustrative example was very recently studied in Ref. [23], consisting of two non-reciprocally coupled Gaussian fields near criticality. Infinitesimal coupling suffices to mix the correlation and Markov lengths, but can clearly only produce entropy at an infinitesimal rate.

  1. There can be no general converse, from detailed-balance violation to non-Gibbsian steady states: many processes violate detailed balance yet mix to the infinite-temperature state — for example, independent three-state spins whose only allowed transitions are 0→1→2→00 \to1 \to2 \to0, at equal rates.

Relation to previous work

The question addressed here—minimal entropy production needed to maintain a target state—was previously explored in Ref. [25]; our setup differs from this work, and other previous literature on entropy production, by explicitly imposing a spatial restriction on the allowed processes. The field theories we consider are closely related to those explored in Refs. [26, 27]; in particular, their “model AB” exhibits the phenomenon of a local process satisfying detailed balance with respect to a nonlocal parent Hamiltonian. The distinction between resolved and hidden entropy production is also central to coarse-grained stochastic thermodynamics [6, 28]. The steady state of the counterexample belongs to a family of states that have been explored in the context of hyperuniformity [29, 30, 31, 32]. The Markov property for quantum systems has seen considerable recent attention [9, 33, 34], though primarily for dynamics that is explicitly detailed balance with respect to a local parent Hamiltonian.

Finally, two very recent works (posted while this paper was nearing completion) address similar themes: Ref. [23] showed that the steady-state Markov length can diverge for equilibrium critical points subject to weak nonequilibrium perturbations (which automatically produce arbitrarily little entropy), and Ref. [2] presented an example of a quantum steady state with long-range Markov length, which offers an interesting contrast to the classical discrete setting, where these states seem not to exist.

Single-component Gaussian processes

Setup

In this section we consider statistically translation-invariant Ornstein–Uhlenbeck theories of the form

ϕ˙=−Bϕ+η,⟨η(x,t)ηT(x′,t′)⟩=2D δ(x−x′)δ(t−t′)(3)\dot{\phi}=-\mathcal{B}\phi+\eta,\qquad\langle\eta(x,t)\eta^{T}(x',t')\rangle=2\mathcal{D}\,\delta(x-x')\delta(t-t') \tag*{(3)}

where ϕ\phi is an nn-component real field; B(k)\mathcal{B}(k) and D(k)\mathcal{D}(k) are matrix-valued; B\mathcal{B} is stable, i.e., has only eigenvalues with positive real part; and D\mathcal{D} is symmetric positive semidefinite. Equivalently, the process is specified by its MSR action, S=∫dt ϕ^T(∂t+B)ϕ−Dϕ^2S=\int dt\,\hat{\phi}^{T}(\partial_t+\mathcal{B})\phi-\mathcal{D}\hat{\phi}^{2}. Real-space locality enforces that B(k),D(k)\mathcal{B}(k),\mathcal{D}(k) polynomials in kk, and reality of ϕ\phi together with parity symmetry kills odd powers of kk in the small-kk expansion. We define the drift F[ϕ]=−BϕF[\phi]=-\mathcal{B}\phi. Wavenumbers run over the band ∣k∣≤Λ|k|\leq\Lambda, with Λ\Lambda the ultraviolet (lattice) cutoff; we set Λ=1\Lambda=1. The stationary covariance solves BC+CBT=2D\mathcal{B}C+C\mathcal{B}^{T}=2\mathcal{D}; the stationary measure is π∝e−12ϕTKϕ\pi\propto e^{-\frac{1}{2}\phi^{T}K\phi} with K=C−1K=C^{-1}, and naturally specifies an effective Hamiltonian. We can write

BC=D+Q,QT=−Q⟺B=(D+Q)K(4)\mathcal{B}C=\mathcal{D}+Q,\qquad Q^{T}=-Q\qquad\Longleftrightarrow\qquad\mathcal{B}=(\mathcal{D}+Q)K \tag*{(4)}

with QQ the circulation [35, 36]: Q=0Q=0 iff detailed balance holds.

For Gaussian states, correlations and mutual information are read off from the covariance CC, while conditional correlations and CMI are read off from its inverse, the “precision matrix” (or “effective Hamiltonian”) K=C−1K=C^{-1}. Poles of KK are zeroes of CC and vice versa. Rational K(k)K(k) with complex poles gives exponentially decaying CMI, with a length scale ξ−1=min⁡∣Im⁡kpole∣\xi^{-1}=\min|\operatorname{Im}k_{\mathrm{pole}}|; real-axis singularities give power-law or non-decaying CMI. We use ξ\xi for this precision-decay scale below; in the one-dimensional setting, the Markov length is ξ/2\xi/2.

The counterexample

Equipped with these definitions, we can introduce the model that will illustrate many of the themes of this paper: a one-component field on a chain, with ultralocal drift and conserved nearest-neighbor noise,

ϕ˙i=−ϕi+ηi,ηi=ζi−ζi−1,⟨ζiζj⟩=2δij⟹D(k)=4sin⁡2(k/2).(5)\dot{\phi}_{i}=-\phi_{i}+\eta_{i},\qquad\eta_{i}=\zeta_{i}-\zeta_{i-1},\qquad\langle\zeta_{i}\zeta_{j}\rangle=2\delta_{ij}\qquad\Longrightarrow\qquad\mathcal{D}(k)=4\sin^{2}(k/2). \tag*{(5)}

Since B=1\mathcal{B}=1, the stationary covariance is C=DC=\mathcal{D}, and BC=D\mathcal{B}C=\mathcal{D} is symmetric: Q=0Q=0, and detailed balance holds. (The k=0k=0 mode is noiseless but damped, so it pins to zero, and the steady state is unique.) All modes relax at the same rate. Correlations vanish identically beyond one site but

K(k)=14sin⁡2(k/2)(6)K(k) = \frac{1}{4\sin^2(k/2)} \tag*{(6)}

has a real-axis pole at k=0k=0, so the Markov length is infinite. For consecutive finite intervals A,B,CA,B,C in the infinite chain, the CMI is (App. § 7)

I(A:C∣B)=−12log⁡(1−∣A∣∣A∣+∣B∣+1∣C∣∣C∣+∣B∣+1),(7)I(A:C|B)=-\frac{1}{2}\log\left(1-\frac{|A|}{|A|+|B|+1}\frac{|C|}{|C|+|B|+1}\right), \tag*{(7)}

so it scales as ∼1/∣B∣2\sim1/|B|^2 for ∣A∣,∣C∣|A|,|C| fixed and small, ∣B∣|B| large, and ∼12log⁡(L/∣B∣)\sim\frac{1}{2}\log(L/|B|) for ∣A∣=∣C∣=L≫∣B∣|A|=|C|=L\gg|B|. The case of the full system requires care: the state space is continuous, and the CMI naively diverges unless regulated by adding infinitesimal noise. Figure 1 plots the regulated ring and the analogous two-dimensional geometry. The effective Hamiltonian is the confining one-dimensional lattice Coulomb potential, Kij∼−∣i−j∣/2K_{ij}\sim-|i-j|/2 [16]. This example illustrates the counterintuitive fact that a completely local process can satisfy detailed balance with respect to a nonlocal effective Hamiltonian.

There are two very different ways of reading this process. The first is the “obvious” one: ϕ\phi is coupled to two baths at different temperatures; the cold bath is at T=0T=0, and obeys no constraints, while the hot bath is at nonzero TT, and is restricted to charge-conserving moves. (The conserving bath also gives a kk-dependent damping term, but this is subleading at small kk and can be neglected.) Also available, though, is an equilibrium reading, according to which the fundamental field is Ei=∑j≤iϕjE_i=\sum_{j\le i}\phi_j, its Hamiltonian is H=12∑iEi2H=\frac{1}{2}\sum_i E_i^2 (nonlocal in the ϕ\phi variables), and it undergoes single-site Glauber updates at temperature T=1T=1, E˙i=−∂H/∂Ei+ζi\dot{E}_i=-\partial H/\partial E_i+\zeta_i, where ζi\zeta_i is white noise. Then ϕ˙i=E˙i−E˙i−1=−ϕi+ζi−ζi−1\dot{\phi}_i=\dot{E}_i-\dot{E}_{i-1}=-\phi_i+\zeta_i-\zeta_{i-1}, precisely Eq. (5). Thus, the process (5) itself is equally compatible with strongly nonequilibrium local dynamics and equilibrium dynamics with respect to a nonlocal Hamiltonian.

These two interpretations of § 3.2 describe the same differential equation, so any passive correlation measurement will yield identical answers in the two cases. In particular, C(k,ω)∼k2C(k,\omega)\sim k^2 as ω→0\omega\to0. However, the response to perturbations of the Hamiltonian distinguishes between them. We sketch the main ideas here; a derivation in the Keldysh framework is presented in App. § 8. For a Gaussian theory, the response function R(k,ω)\mathcal{R}(k,\omega) of a mode probes its intrinsic dynamics, independent of the occupation factors [37]. The effective temperature of mode kk, Teff(k,ω)≡ωC(k,ω)/[2Im⁡R(k,ω)]T_{\mathrm{eff}}(k,\omega)\equiv\omega C(k,\omega)/[2\operatorname{Im}\mathcal{R}(k,\omega)]. On the Coulomb interpretation, the low-momentum modes are increasingly stiff, so their response is suppressed as R∼k2/(λ−iω)\mathcal{R}\sim k^2/(\lambda-i\omega), similar to their fluctuations: thus, TeffT_{\mathrm{eff}} approaches a constant, as it should since the system is in equilibrium with respect to a nonlocal Hamiltonian. On the nonequilibrium interpretation, the stiffness of the low-kk modes is a kk-independent constant, so R∼1/(λ−iω)\mathcal{R}\sim1/(\lambda-i\omega) at small kk. This yields Teff(k)∼k2T_{\mathrm{eff}}(k)\sim k^2. Thus the fluctuation-dissipation theorem is violated on the nonequilibrium interpretation. Crucially, the Coulomb interpretation requires a singular Hamiltonian stiffness, KC∼k−2\mathcal{K}_C\sim k^{-2}. For a local quadratic Hamiltonian with uniformly bounded stiffness 0<K(k)≤K∗0<\mathcal{K}(k)\le\mathcal{K}_*, the static response is R(0,k)=1/K(k)≥1/K∗\mathcal{R}(0,k)=1/\mathcal{K}(k)\ge1/\mathcal{K}_*. Matching the vanishing fluctuations then forces Teff(k)=K(k)C(k)T_{\mathrm{eff}}(k)=\mathcal{K}(k)C(k) to vanish as k→0k\to0. Thus any such local interpretation must be nonequilibrium.

Generalizations

Even for a single-component field, the mechanism outlined in the previous section can be extended to give increasingly nonlocal CMI, by forcing the noise to come with higher powers of kk, e.g., through dipole moment conservation [38]. This mechanism can lead to a ladder of hyperuniform steady states, with S(k)∼k2nS(k)\sim k^{2n} for any integer n≥1n\ge1; all these steady states have diverging CMI, and all the associated processes are local and obey detailed balance. More generally, for any rational function K(k)≡ν(k)/θ(k)K(k) \equiv\nu(k)/\theta(k), one can straightforwardly construct a local process that satisfies detailed balance and has a steady state with precision K(k)K(k), as follows. Take ν\nu and θ\theta to be even polynomials with ν(k)>0\nu(k)>0 on the band and θ(k)≥0\theta(k)\geq0 on the band, and choose

B(k)=ν(k),D(k)=θ(k).(8)\mathcal{B}(k)=\nu(k), \qquad\mathcal{D}(k)=\theta(k). \tag*{(8)}

Both are polynomials, so the process is local. Since ν>0\nu>0, every mode is damped: the steady state exists and is unique, with covariance C=D/B=θ/ν=K−1C=\mathcal{D}/\mathcal{B}=\theta/\nu=K^{-1}, as desired. And BC=θ=D\mathcal{B}C=\theta=\mathcal{D} is symmetric, so Q=0Q=0: detailed balance holds by construction. The counterexample is the case ν=1\nu=1, θ=4sin⁡2(k/2)\theta=4\sin^2(k/2). The representation is not unique: multiplying ν\nu and θ\theta by a common positive even polynomial yields a different local reversible process with the same steady state. (The construction generalizes verbatim to nn-component fields: θ\theta the common denominator, D=θ 1\mathcal{D}=\theta\,\mathbf{1}, and B=θK\mathcal{B}=\theta K positive definite on the band.)

In continuum notation the counterexample reads

ϕ˙=−λϕ+∇⋅η,D=Mk2,C(k)=Mk2λ,K(k)=λMk2,(9)\dot{\phi}=-\lambda\phi+\nabla\cdot\boldsymbol{\eta}, \qquad\mathcal{D}=Mk^2, \qquad C(k)=\frac{Mk^2}{\lambda}, \qquad K(k)=\frac{\lambda}{Mk^2}, \tag*{(9)}

with λ\lambda the non-conserving relaxation rate and MM the conserved-noise mobility. We use this form whenever dimensional constants matter, as in the bounds of § 3.4.

Entropy production

We have seen that the counterexample violates the fluctuation-dissipation theorem. We now turn to lower-bounding this violation, and the associated entropy cost, more generally. For simplicity we will continue to focus on the one-component, translation-invariant setting. For the explicit bounds we take H=12∑iϕi2H=\frac{1}{2}\sum_i\phi_i^2 and finite-range linear couplings to Markovian baths. The first thing we need to establish is that long Markov lengths cannot arise for a local field ϕ\phi in equilibrium with a local bath. This is slightly less trivial than it sounds: the composite system obeys the Markov property by assumption, but we are concerned with the marginal over the bath, and the Markov length can grow (indeed diverge) under marginalization. For a particular kk mode, after integrating out the bath, the precision matrix for the ϕ\phi fields is Keff(k)≡K0(k)−h2(k)Gb(k)K_{\mathrm{eff}}(k) \equiv K_0(k) - h^2(k)G_b(k), where Gb(k)G_b(k) is the bath’s correlation function at kk. Stability of the state forces h2≤K0Gb−1h^2 \le K_0G_b^{-1} for every kk. The Markov length of the system’s marginal is set by the pole of Keff(k)K_{\mathrm{eff}}(k) that lies closest to the real axis. By assumption, K0,hK_0,h are local, and do not have poles near the real axis. The danger is that GbG_b might: in other words, integrating out a gapless mode can generate effective long-range interactions. Provided the bath has a finite correlation length ξbath\xi_{\mathrm{bath}}, however, the nearest pole of Gb(k)G_b(k) is also gapped away from the real axis, so Keff(k)K_{\mathrm{eff}}(k) has no singularities near the axis, and ϕ\phi cannot have a large Markov length. To summarize, marginalizing over the bath cannot generate a Markov length that is larger than the bath’s correlation length. (Note that a similar result for local state spaces was proved in the high-temperature limit in Ref. [21] — if we take the system and its bath to be in equilibrium at sufficiently high temperature and apply the erasure channel to the bath, Theorem 2 of that work states that the system’s marginal retains a finite Markov length.)

Any remaining, strictly local, route to a large Markov length must come from nonequilibrium effects. To parameterize these we will couple the system to a set of arbitrarily many baths, indexed by ν\nu. Each bath is in equilibrium at a temperature TνT_\nu and a characteristic, momentum-dependent Golden Rule Rate Mν(k)\mathcal{M}_\nu(k) at which it couples to the system: Mν(k)=g(k)2ν(k)\mathcal{M}_\nu(k) = g(k)^2\nu(k), where g(k)g(k) is the interaction between system mode kk and the bath, and ν(k)\nu(k) is the bath density of states. Using each bath’s FDT, and suppressing mode indices, the marginal process has B=∑νMν\mathcal{B} = \sum_\nu\mathcal{M}_\nu, D=∑νMνTν\mathcal{D} = \sum_\nu\mathcal{M}_\nu T_\nu. Matching the dissipation B=λ\mathcal{B} = \lambda, we get that D=λTeff\mathcal{D} = \lambda T_{\mathrm{eff}}, where Teff(k)=Ek(T)T_{\mathrm{eff}}(k) = E_k(T) and Ek(f)≡λ−1∑νMν(k)f(Tν)E_k(f) \equiv\lambda^{-1}\sum_\nu\mathcal{M}_\nu(k)f(T_\nu) denotes the average over baths weighted by their couplings to mode kk. Rearranging the Clausius entropy-production formula gives

σk=λ[Teff(k)Ek[1/T]−1],(10)\sigma_k = \lambda\left[T_{\mathrm{eff}}(k)E_k[1/T]-1\right], \tag*{(10)}

which is non-negative for every kk by Jensen’s inequality.

In general, achieving a large Markov length requires D(k)\mathcal{D}(k) to have a zero near the real axis. Once the range of the process is capped, the only possible origin for this zero is from Teff(k)T_{\mathrm{eff}}(k). Assume for simplicity that the zero is on the imaginary axis, and the behavior around it is quadratic: D(k)=12D′′(0)(κ2+k2)\mathcal{D}(k) = \frac{1}{2}\mathcal{D}^{\prime\prime}(0)(\kappa^2 + k^2), κ2=2D(0)/D′′(0)\kappa^2 = 2\mathcal{D}(0)/\mathcal{D}^{\prime\prime}(0). The precision length is ξ=1/κ\xi= 1/\kappa (with the CMI convention specified in § 3.1). It is clear that λTcoldest≤D(k)≤λThottest\lambda T_{\mathrm{coldest}} \le\mathcal{D}(k) \le\lambda T_{\mathrm{hottest}}; moreover, if the coupling range is rr, D(k)\mathcal{D}(k) is a trigonometric polynomial of degree rr in kk. By Bernstein’s inequality, a degree-rr polynomial that is bounded on the band can have curvature at most ∣D′′∣≤r2λThottest|\mathcal{D}^{\prime\prime}| \le r^2\lambda T_{\mathrm{hottest}}: a range-rr coupling cannot bend faster than r2r^2 times its height. This gives the scaling of the bound. Keeping the full Taylor series and centering between the noise floor and ceiling gives the sharper result

ξ≤r2arctanh⁡Tcoldest/Thottest≤r2ThottestTcoldest.(11)\xi\le\frac{r}{2\operatorname{arctanh}\sqrt{T_{\mathrm{coldest}}/T_{\mathrm{hottest}}}} \le\frac{r}{2}\sqrt{\frac{T_{\mathrm{hottest}}}{T_{\mathrm{coldest}}}}. \tag*{(11)}

The argument above bounds the achievable Markov length in terms of the bath temperatures. Next, we fix the range and Markov length, and estimate the minimum entropy production consistent with these. We proceed as follows. Define g(k)≡E[1/T]g(k) \equiv E[1/T], so the mode-wise entropy production is σk=λ(g(k)D(k)−1)\sigma_k = \lambda(g(k)\mathcal{D}(k)-1), with g(k)≥1/D(k)g(k) \ge1/\mathcal{D}(k). Any choice of bath defines some valid g(k)g(k), and for range rr, g(k)g(k) must be an order-rr polynomial in kk, no matter how many baths we couple to the system. To get a bound, we relax the problem and optimize over all polynomials g(k)g(k) rather than optimizing over bath assignments: we compute the “majorant bound”

Lr(m)=min⁡g real trigonometric polynomialdeg⁡g≤r, g(k)≥1/D(k)∫−ππdk2π[D(k)g(k)−1].(12)L_r(m) = \min_{\substack{g\ \mathrm{real\ trigonometric\ polynomial}\\ \deg g \le r,\ g(k)\ge1/\mathcal{D}(k)}} \int_{-\pi}^{\pi} \frac{dk}{2\pi}\left[\mathcal{D}(k)g(k)-1\right]. \tag*{(12)}

This lower-bounds the entropy production regardless of the choice of bath. The physical intuition is particularly simple when mm is small. The target

1D(k)≃1m+k2(13)\frac{1}{D(k)} \simeq\frac{1}{m+k^2} \tag*{(13)}

has a narrow, tall peak: height 1/m1/m and width m\sqrt{m}. A polynomial with only rr Fourier harmonics cannot follow an arbitrarily narrow peak. To stay above it, the polynomial must also be large elsewhere. That unavoidable excess is entropy production. A helpful way to rewrite this bound is to note that D(k)g(k)=0D(k)g(k)=0 at k=−imk=-i\sqrt{m}. Minimizing g(k)D(k)−1g(k)D(k)-1 subject to these constraints gives the closed-form bound

Lr(m)=2sinh⁡((r+2)κ)/sinh⁡κ−(r+2).(14)L_r(m)=\frac{2}{\sinh((r+2)\kappa)/\sinh\kappa-(r+2)}. \tag*{(14)}

For fixed range and large ξ\xi,

Lr(m)∼12ξ2(r+1)(r+2)(r+3).(15)L_r(m)\sim\frac{12\xi^2}{(r+1)(r+2)(r+3)}. \tag*{(15)}

Thus fixed-range thermal implementations necessarily pay a cost growing as ξ2\xi^2. In particular, at nearest-neighbor range,

L1(m)=2m(m+4),(16)L_1(m)=\frac{2}{m(m+4)}, \tag*{(16)}

and an explicit two-bath construction attains it. We choose the following two baths:

Tc=m,Th=m+4,Mc=λ(1−s),Mh=λs.(17)T_c=m,\qquad T_h=m+4,\qquad\mathcal{M}_c=\lambda(1-s),\qquad\mathcal{M}_h=\lambda s. \tag*{(17)}

Their coupling amplitudes are proportional to (1+eik)/2(1+e^{ik})/2 and (1−eik)/2(1-e^{ik})/2, so both are nearest-neighbor couplings. Their entropy production is

σ^k=16s(1−s)m(m+4),σ^site=2m(m+4)∼ξ22.(18)\hat{\sigma}_k=\frac{16s(1-s)}{m(m+4)},\qquad\hat{\sigma}_{\mathrm{site}}=\frac{2}{m(m+4)}\sim\frac{\xi^2}{2}. \tag*{(18)}

Here σ^=σ/λ\hat{\sigma}=\sigma/\lambda and ξ=[2arsinh⁡(m/2)]−1\xi=[2\operatorname{arsinh}(\sqrt{m}/2)]^{-1}. At nearest-neighbor range, this choice saturates the bound. Figure 2 compares the cost of the explicit implementation with the lower bound from optimization; the bounds coincide at r=1r=1. The positive lower bound at every finite rr assumes strictly finite-range Markovian baths. As discussed earlier in this section, exponentially decaying CMI can instead arise at equilibrium by marginalizing over a bath with finite correlation length.

Figure 2. The range–dissipation tradeoff for C=m+4sin⁡2(k/2)\mathcal{C}=m+4\sin^2(k/2), at m=0.01m=0.01 and 0.0010.001. Dashed curves: the lower bound from optimizing over (14). Solid curves: entropy production of the explicit bath construction (17). The bounds coincide at nearest-neighbor range. Both curves use the lattice momentum measure; the lower bound is not asserted to be physically attainable at higher range.

Jump processes on discrete spaces

In this section we consider jump processes that are defined on discrete systems (e.g., Ising spins or lattice gases): the local state space has a finite number of states, and these are locally coupled by jump moves of bounded support: a configuration xx is updated to x+ux+u, where uu differs from xx only locally, and the probability of the jump happening only depends on the local neighborhood of uu (within some range rr). We will see that results of the previous section are surprisingly fragile to discreteness.

Detailed balance + locality implies locality of HH. The counterexample cannot arise for full-support jump processes with irreducible frozen blocks [39] (i.e., a sufficiently large block has a unique steady state if one runs its dynamics freezing the boundaries): the locality of HH follows from that of the moves. Suppose the stationary measure is ∝e−H(x)\propto e^{-H(x)}. Detailed balance gives Wu(x)/W−u(x+u)=e−[H(x+u)−H(x)]W_u(x)/W_{-u}(x+u)=e^{-[H(x+u)-H(x)]}. Since the lhs is local, so is the rhs: every HH-difference between pairs of connected configurations is local. If all locally different pairs of configurations were connected by local moves, then this would immediately imply the locality of HH. This last implication follows if we make one of two assumptions. If we assume that the model consists of single-site updates, conditioned on some finite window around the site, such that the rate is strictly positive for any state on the window, then any pair of configurations is connected by local moves and HH is manifestly local. More generally, assume that the chain of moves supported inside AA is irreducible for every fixed exterior. If BB shields AA from CC, the frozen generator on AA depends only on bb. Denote its stationary state as μb(A)\mu_b(A). Detailed balance means the generator of the dynamics is frustration-free, so the global steady state is also the steady state of the frozen chain, so the conditional distribution P(A∣b,c)P(A|b,c) is also a steady state for each cc; uniqueness then gives P(A∣b,c)=μb(A)P(A\mid b,c)=\mu_b(A), and hence I(A:C∣B)=0I(A:C\mid B)=0. Thus, there is a much tighter relation between detailed balance and Hammersley-Clifford in this setting: under this irreducibility assumption, a full-support local detailed-balance process must stabilize a local Hamiltonian.

If one tries to derive a continuum process as the limit of a jump process, the argument above fails because each continuum increment sums over a large number of jumps, so channel-by-channel detailed balance is lost, and the weaker notion that survives for the continuum increments is unable to guarantee a local parent Hamiltonian (as the counterexample shows).

Hyperuniform states cannot be reached rapidly. A natural attempt to discretize the counterexample might run as follows. Consider a system with three states per site, {−1,0,1}\{-1,0,1\}, indexed by their charge qq, subject to the following two baths: a non-conserving zero-temperature bath that favors the state 00, and drives both other states to it; and a finite-temperature bath that conserves Q=∑iqiQ=\sum_i q_i, and allows nearest-neighbor moves such as 0,±1↔±1,00,\pm1\leftrightarrow\pm1,0 and 00↔+1,−100\leftrightarrow+1,-1. This fails to generate a hyperuniform steady state: processes like 0,0→1,−1→0,−10,0\to1,-1\to0,-1 generate shot-noise fluctuations of the total charge, so the structure factor at zero momentum (which measures var⁡(Q)\operatorname{var}(Q)) is nonzero.

The same obstruction applies whenever local charge-changing activity is nonzero and the local drift has summable correlations. Suppose that the jump process includes some charge-changing terms re(x)r_e(x), each of which changes the charge by ΔQe\Delta Q_e (which in the above example is ±1\pm1). Define the signed drift bQ(x)=∑ere(x)ΔQeb_Q(x) = \sum_e r_e(x)\Delta Q_e and the noise activity vQ(x)=∑ere(x)(ΔQe)2v_Q(x) = \sum_e r_e(x)(\Delta Q_e)^2. For a steady state with summable correlations of the local drift observables, var⁡(bQ)≤CN\operatorname{var}(b_Q) \le CN for some constant CC. In the steady state, ⟨bQ⟩=0\langle b_Q\rangle= 0 and ⟨Q2⟩\langle Q^2\rangle is invariant. There are two contributions to its evolution, coming from the deterministic drift and the accumulation of variance from random increments. Balancing these gives −2Cov⁡(Q,bQ)=⟨vQ⟩≡Nr2-2\operatorname{Cov}(Q,b_Q) = \langle v_Q\rangle\equiv Nr_2. By Cauchy–Schwarz, Nr2≤2var⁡(Q)var⁡(bQ)Nr_2 \le2\sqrt{\operatorname{var}(Q)\operatorname{var}(b_Q)}, and assembling these observations gives us the bound

var⁡(Q)N≥r224C2>0,(19)\frac{\operatorname{var}(Q)}{N} \ge\frac{r_2^2}{4C^2} > 0, \tag*{(19)}

so shot noise blocks hyperuniformity under these assumptions when r2>0r_2 > 0. The shot-noise floor is set by the minimum charge increment ΔQ\Delta Q, and one recovers the counterexample in the previous section for a suitable ΔQ→0\Delta Q \to0 limit.

Is rapid mixing compatible with long-range CMI? Under the assumptions above, the argument ruled out the possibility of getting a long-range CMI through hyperuniformity. One might wonder whether there are other means to achieve a long-range CMI in the unique steady state of a jump process that mixes rapidly (as we were able to with a rapidly relaxing continuous process). We conjecture that this is impossible, based on the following heuristic argument. Since the process mixes rapidly and its steady state is unique, the steady state cannot have long-range order. All known examples where long-range conditional correlations persist in the absence of long-range order have some form of “string order,” i.e., there is a product of observables supported in ABCABC that is ordered, though each individual factor is random. String order allows AA and CC to be correlated conditional on the state of BB, even at early times (the string is always extended so it is not subject to light cones). However, in the steady state of a system with asynchronous stochastic jumps, the string order parameter in region BB will be flipped by single events in BB unless it is protected by some global symmetry of the dynamics (as it is for spontaneous strong-to-weak symmetry breaking [24]), or by some other structure such as synchronous updates designed to ensure that these local events cancel. So any string decorrelates in its bulk, and long-range CMI cannot be achieved through this means. This step of the heuristic is in fact a theorem, proved in App. § 9: for a one-dimensional local jump process with a translation-invariant, spatially ergodic stationary law, persistent string order (with arbitrary bounded endpoint decorations) forces the string-flipping activity to vanish exactly; with full local support this is an exact conservation law of the dynamics, contradicting uniqueness of the stationary state on periodic chains. The remaining gap between this statement and the main conjecture is the first step of the heuristic — that long-range CMI at short times must arise through a string. App. § 10 closes this gap for binary steady states with an exact finite-state hidden-Markov representation, a strictly positive hidden transition matrix, and an active odd-flip rule, where exponential decay of the CMI itself can be proved; beyond that class, we are unable to rule out currently unknown mechanisms that avoid the “string order” route.

Discussion

In terms of entropy production, the discrete and continuous classical worlds are surprisingly different: in the former, assuming full support and irreducible frozen blocks, a local process can only be in detailed balance with respect to a local Hamiltonian; in the latter, this restriction is gone, and the entropy production mandated by locality is “hidden” from the standard definitions of stochastic thermodynamics. The two worlds also differ in the nonequilibrium steady states they admit: in the discrete world, long-range CMI seems unachievable in the unique steady state of any rapidly mixing process, while the continuous world admits many straightforward examples with rapid drift relaxation. Interestingly, in this respect the discrete quantum world looks more like the continuous classical one, admitting modes with no noise floor and rapid-mixing processes that generate long-range CMI [2]. (Interestingly, our thermal implementation of this quantum example also requires infinite entropy production to sustain its steady state; a thermally regularized version pays entropy logarithmically in the Markov length, σ^≃2p(1−p)ln⁡ξ\hat{\sigma} \simeq2p(1-p)\ln\xi — see App. § 11.)

Finally, we comment briefly on interacting mechanisms that evade the trade-offs of § 3.4. The linear noise floor from the cold bath can be circumvented by tightly coupling the system to a nonequilibrium ratchet, which in turn is coupled to heat baths [40], such that every “model A” damping event coincides with a heat quantum ΔE\Delta E transported from a hot to a cold bath, as in autonomous Maxwell demons [41, 42]. In this setting, wrong-way transitions are suppressed by e−Ae^{-A} with A=ΔE(1/Tc−1/Th)≤Eloc/TcA = \Delta E(1/T_c - 1/T_h) \le E_{\mathrm{loc}}/T_c — an effective temperature exponentially below the coldest reservoir, costing AA per corrective event: effectively, this mechanism replaces the scaling Tc→exp⁡(−1/Tc)T_c \to\exp(-1/T_c), allowing for an exponentially larger ξ\xi at fixed TcT_c. However, all such strategies we have found require discrete events, so they cannot increase ξ\xi past the shot-noise floor of Eq. (19). Thus the general question of minimizing σ\sigma at fixed ξ\xi for interacting systems remains open.

Our results raise many natural questions for future work. The most immediate one is to settle the conjecture on achieving long-range CMI in discrete jump processes (and to understand entropy production in such processes, if they exist). Other natural questions include generalizing our results to find a relation between quantum CMI and KMS detailed balance (cf. the quantum Gibbs sampler literature [43]); identifying steady-state phase transitions at which the Markov length diverges, and understanding their thermodynamic signatures; extending our entropy-production floors beyond Gaussian theories to generic interacting field theories; and more generally incorporating our spatial locality restrictions (and Hammersley-Clifford) into the framework of stochastic thermodynamics.

Acknowledgements

I used Claude Fable 5 and ChatGPT Astra 6 to develop the calculations and proofs in this work. I thank Yu-Hsueh Chen, Matthew Fisher, Sam Garratt, Tarun Grover, Hyunsoo Ha, Jake Hauser, Tim Hsieh, David Huse, Jérôme Lloyd, Nick O’Dea, Shengqi Sang, Romain Vasseur, Cenke Xu, Carolyn Zhang, and Frank Zhang for helpful discussions.

CMI of the counterexample

This appendix computes the conditional mutual information plotted in Figure 1. The state is the regulated counterexample on a ring of NN sites, C(k)=T1+4sin⁡2(k/2)C(k) = T_1 + 4\sin^2(k/2); equivalently,

ϕi=Ei−Ei−1+T1ui(20)\phi_i = E_i - E_{i-1} + \sqrt{T_1}u_i \tag*{(20)}

with EiE_i and uiu_i independent standard Gaussians — heights with unit prior variance, plus a white regulator. We take BB to consist of two segments (or an annulus) of width ww.

For jointly Gaussian variables, I(A:C∣B)=−12∑iln⁡(1−ρi2)I(A:C \mid B) = -\frac{1}{2}\sum_i \ln(1-\rho_i^2), where ρi\rho_i are the “canonical correlations,” i.e., the singular values of the matrix R≡ΣAA−1/2ΣACΣCC−1/2\mathbf{R} \equiv\Sigma_{AA}^{-1/2}\Sigma_{AC}\Sigma_{CC}^{-1/2}, and the Σ\Sigma’s are conditional covariance matrices. For the equal-width buffers in App. A, both canonical correlations have simple exact formulas, including at nonzero regulator. Write a=LAa=L_A, c=LCc=L_C, N=a+c+2wN=a+c+2w, and define

κ=2arsinh⁡(T1/2),u=(a−1)κ2,v=(c−1)κ2,t=(w+1)κ.(21)\kappa=2\operatorname{arsinh}\left(\sqrt{T_1}/2\right),\qquad u=\frac{(a-1)\kappa}{2},\qquad v=\frac{(c-1)\kappa}{2},\qquad t=(w+1)\kappa. \tag*{(21)}

For a,c≥2a,c\ge2, the two nonzero canonical correlations are

ρ02=cosh⁡ucosh⁡vcosh⁡(u+t)cosh⁡(v+t),ρ12=sinh⁡usinh⁡vsinh⁡(u+t)sinh⁡(v+t).(22)\rho_0^2=\frac{\cosh u\cosh v}{\cosh(u+t)\cosh(v+t)},\qquad\rho_1^2=\frac{\sinh u\sinh v}{\sinh(u+t)\sinh(v+t)}. \tag*{(22)}

All remaining canonical correlations vanish, and

I(A:C∣B)=−12log⁡(1−ρ02)−12log⁡(1−ρ12)=12ln⁡sinh⁡((LA+2w+1)κ)sinh⁡((LC+2w+1)κ)4sinh⁡2((w+1)κ)sinh⁡2(Nκ/2).(23)I(A:C\mid B)=-\frac{1}{2}\log(1-\rho_0^2)-\frac{1}{2}\log(1-\rho_1^2)=\frac{1}{2}\ln\frac{\sinh((L_A+2w+1)\kappa)\sinh((L_C+2w+1)\kappa)}{4\sinh^2((w+1)\kappa)\sinh^2(N\kappa/2)}. \tag*{(23)}

In the small-regulator regime Nκ≪1N\kappa\ll1,

I(A:C∣B)=12ln⁡(LA+2w+1)(LC+2w+1)N2(w+1)2T1+O(T1N2).(24)I(A:C\mid B)=\frac{1}{2}\ln\frac{(L_A+2w+1)(L_C+2w+1)}{N^2(w+1)^2T_1}+O(T_1N^2). \tag*{(24)}

We expect a similar decomposition to hold for the annular geometry in d=2d=2: the sum over two channels is replaced by one over multipoles of the enclosed charge. We have not evaluated this in closed form: the plotted CMI in the main text is computed directly from covariance determinants.

Keldysh analysis of response in the counterexample

In this appendix we show, using the Keldysh formalism, how the two implementations of the counterexample naturally give rise to distinct response functions, even though the processes are identical absent an external perturbation. We use the fact that coupling a system to an equilibrium bath at temperature TT and mobility Γ\Gamma adds the following term to the Keldysh action:

ΔSa=∫t[X^ΓaδHδX−ΓaTaX^2].(25)\Delta S_a=\int_t\left[\hat{X}\Gamma_a\frac{\delta H}{\delta X}-\Gamma_aT_a\hat{X}^2\right]. \tag*{(25)}

The two readings of the counterexample have different Hamiltonians, and different system-bath couplings, so as to cancel out in the marginal process. The source is always added to HH as H0+hXH_0+hX.

Implementation A: long-range Hamiltonian, one bath. Coulomb Hamiltonian KC(k)=1/k^2\mathcal{K}_C(k)=1/\hat{k}^2, one current-coupled (conserving) bath Γ(k)=Γ0k^2\Gamma(k)=\Gamma_0\hat{k}^2 at temperature TT. Assembling one block:

SkA[h]=∫t[ϕ^(∂t+Γ0)ϕ−Γ0Tk^2ϕ^2−Γ0k^2hϕ^];(26)S_k^A[h]=\int_t\left[\hat{\phi}(\partial_t+\Gamma_0)\phi-\Gamma_0T\hat{k}^2\hat{\phi}^2-\Gamma_0\hat{k}^2h\hat{\phi}\right]; \tag*{(26)}

the conserving mobility cancels the Coulomb kernel in the drift (a local current driven by the gradient of a long-range potential has a local divergence), so BA=Γ0\mathcal{B}_A=\Gamma_0, DA=Γ0Tk^2\mathcal{D}_A=\Gamma_0T\hat{k}^2, S(k)=Tk^2S(k)=T\hat{k}^2, and the single bath is in detailed balance: σ=0\sigma=0.

Implementation B: trivial Hamiltonian, two baths. K=1\mathcal{K}=1; a cold density-coupled bath (Γc,Tc)(\Gamma_c,T_c) and a hot conserving bath (Γhk^2,Th)(\Gamma_h \hat{k}^{2},T_h):

SkB[h]=∫t[ϕ^(∂t+Γc+Γhk^2)ϕ−(ΓcTc+ΓhThk^2)ϕ^2−(Γc+Γhk^2)hϕ^].(27)S_k^B[h]=\int_t\left[\hat{\phi}(\partial_t+\Gamma_c+\Gamma_h\hat{k}^{2})\phi-(\Gamma_cT_c+\Gamma_hT_h\hat{k}^{2})\hat{\phi}^{2}-(\Gamma_c+\Gamma_h\hat{k}^{2})h\hat{\phi}\right]. \tag*{(27)}

In the limit Tc→0T_c\to0, Γh→0\Gamma_h\to0 with ΓhTh\Gamma_hT_h fixed (the cold block loses its noise, the hot block its friction), and under the dictionary Γc=Γ0\Gamma_c=\Gamma_0, ΓhTh=Γ0T\Gamma_hT_h=\Gamma_0T, the h=0h=0 parts of (26) and (27) become identical term by term.

Response. The source terms do not agree: implementation A carries −Γ0k^2hϕ^-\Gamma_0\hat{k}^{2}h\hat{\phi}, implementation B carries −Γ0hϕ^-\Gamma_0h\hat{\phi}. Hence

RA(ω,k)=Γ0k^2−iω+Γ0,RB(ω,k)=Γ0−iω+Γ0,(28)\mathcal{R}_A(\omega,k)=\frac{\Gamma_0\hat{k}^{2}}{-i\omega+\Gamma_0},\qquad\mathcal{R}_B(\omega,k)=\frac{\Gamma_0}{-i\omega+\Gamma_0}, \tag*{(28)}

with identical C(ω,k)\mathcal{C}(\omega,k).

Persistent string order forces a conserved charge

This appendix proves the claim of § 4 that in the steady state of a local jump process, string order cannot survive unless it is protected by an exact conservation law of the dynamics. (If a conservation law exists, it prevents rapid mixing.) The strategy relies on translational invariance and spatial ergodicity, which is a consequence of the short-range correlations in the steady state. We also assume that the marginal distribution in the steady state has full support on the interaction range—i.e., for our two-site interaction window we assume that all four bitstrings occur with nonzero probability. Spatial ergodicity states that there are no long-range connected correlations in space. Translation-invariant, spatially ergodic distributions obey Birkhoff’s theorem, which states that for any bounded local observable gg, the space-average of any observable converges to its ensemble average. To keep the notation simple we will state the theorem for bits with two-site updates, and consider the parity string Sn(x)=(−1)x1+x2+⋯+xnS_n(x)=(-1)^{x_1+x_2+\cdots+x_n}, but the argument manifestly works for any finite-size alphabet and any string that is a product over sites. The theorem states:

Theorem 1. (informal). Under the assumptions above, ⟨Sn⟩→0\langle S_n\rangle\to0 as n→∞n\to\infty in the steady state.

Proof. We sketch the proof. Again, to keep the combinatorics transparent we assume that all updates act on nearest-neighbor pairs of sites. Define ri(xi,xi+1)r_i(x_i,x_{i+1}) as the probability that an update acting on that configuration will flip the value of SnS_n (for i,i+1i,i+1 inside the string), and define a=⟨ri⟩s.s.a=\langle r_i\rangle_{\mathrm{s.s.}} where the average is over the steady-state probability distribution (and is independent of ii by assumption). The Markov process generator L\mathcal{L}, acting on SnS_n, gives

LSn=−2Sn∑i=1n−1ri+Gn,(29)\mathcal{L}S_n=-2S_n\sum_{i=1}^{n-1}r_i+G_n, \tag*{(29)}

where GnG_n are boundary terms of order unity that can be bounded by a constant GG. Since ⟨LSn⟩s.s.=0\langle\mathcal{L}S_n\rangle_{\mathrm{s.s.}}=0 by stationarity [we will drop the subscript from now on], we can bound

∣⟨Sn(n−1∑i=1n−1ri)⟩∣≤G/n.(30)\left|\left\langle S_n\left(n^{-1}\sum_{i=1}^{n-1}r_i\right)\right\rangle\right|\leq G/n. \tag*{(30)}

Define rˉn≡n−1∑i=1n−1ri\bar{r}_n \equiv n^{-1}\sum_{i=1}^{n-1}r_i. Since the magnitude of SnS_n is at most 1, we can also bound

∣⟨Snrˉn⟩−a⟨Sn⟩∣≤∣rˉn−a∣.(31)\left|\langle S_n\bar{r}_n\rangle-a\langle S_n\rangle\right|\leq|\bar{r}_n-a|. \tag*{(31)}

By Birkhoff’s theorem the latter quantity goes to zero, so for large nn, a−1⟨Sn⟩→∣⟨Snrˉn⟩∣→0a^{-1}\langle S_n\rangle\to|\langle S_n\bar{r}_n\rangle|\to0 by Eq. (30). So either a=0a=0 or ⟨Sn⟩=0\langle S_n\rangle=0. Since aa is an average over non-negative rates, and the steady state has support on all configurations, if a=0a=0 on average the rate at which parity flips must be identically zero. If this is true then parity is conserved and the dynamics is non-ergodic, contradicting our assumptions.

The value of this proof, and the reason it improves on the argument in the main text, is that we made no assumptions about the correlation between the value of the string and the local rates in the region it covers.

Short-range CMI for steady states with hidden Markov structure

This appendix states and proves the one-dimensional obstruction announced in § 4:

Theorem 2 (finite-state case). Let π\pi be a translation-invariant stationary law on {±1}Z\{\pm1\}^{\mathbb{Z}} of a finite-range jump process with bounded rates and at least one nonzero bulk rule flipping an odd number of spins, and suppose π\pi has an exact edge-emitting hidden-Markov representation with DD hidden states and strictly positive hidden transition matrix, Tab≥δ>0T_{ab}\geq\delta>0. Then the two-sided CMI across a buffer of width ℓ\ell obeys Cℓ≤Ce−cℓC_\ell\leq Ce^{-c\ell}, with (C,c)(C,c) depending only on (D,δ)(D,\delta) for deterministic readout — and only on (D,δ,η)(D,\delta,\eta) for stochastic readout whose nonzero labeled transition probabilities are ≥η\geq\eta — uniformly in the jump rates, their range, and the entropy production.

Two-state hidden layer. First we consider a much simpler version of the general problem. We take a system with hidden bits YiY_i on every edge, visible spins XiX_i on every vertex ii (right of edge ii), and a deterministic function Xi=f(Yi,Yi+1)X_i=f(Y_i,Y_{i+1}) coupling them (this is called “deterministic readout”). The underlying probability distribution is over YiY_i, which obey a Markov chain in the spatial direction, but XiX_i “reads” this Markov chain through the deterministic function. In this simple case there are only sixteen functions that could exist, and they fall into four classes: (i) two constant functions, which lead to product distributions; (ii) four cases where Xi=±Yi,±Yi+1X_i=\pm Y_i,\pm Y_{i+1}, where XX inherits the Markov property; (iii) eight functions for which at least one output XiX_i reveals YiY_i and Yi+1Y_{i+1}, so (by the Markov property) conditioning on XiX_i decouples the left and right states on YY and thus (by data processing) on XX, so the CMI vanishes; and (iv) two functions Xi=±YiYi+1X_i=\pm Y_iY_{i+1}. These parity functions are the only case that might admit long-range CMI. The mechanism for long-range CMI is, however, precisely string order: ⟨YiYi+n⟩=⟨∏j∈[i,i+n−1]Xj⟩\langle Y_iY_{i+n}\rangle=\langle\prod_{j\in[i,i+n-1]}X_j\rangle. A zero-dimensional Markov chain with entirely positive rates certainly mixes, so the result in the previous section rules out this case, and (for a two-dimensional hidden state space) thus rules out long-range CMI.

General case: strategy. Either visible observations erase hidden-state ambiguity, or a hidden parity survives. In the first case, there are some configurations in which the hidden variables are revealed to the visible ones, and this decorrelates the hidden Markov chain and cuts off the CMI. This intuition is converted to an exponential bound by Travers’s result [46]. In the second case, either the hidden parity produces string order (in which case we rule it out as before) or it is completely hidden, in which case it does not affect any properties of the visible layer and thus does not contribute CMI.

Predictive memory. Set

hn=H(Xn∣X1n−1),h=lim⁡nhn.h_n = H(X_n \mid X_1^{n-1}), \qquad h = \lim_n h_n.

The entropy chain rule gives

Cℓ(X)=∑n>ℓ(hn−h).C_\ell(X) = \sum_{n>\ell}(h_n-h).

Thus exponential convergence of hnh_n suffices. All these quantities are finite: past and future communicate through a hidden state, so their mutual information is at most log⁡D\log D.

The binary support dichotomy. Two hidden states are mergeable if they can emit the same word and reach the same hidden state. They are immediately incompatible if they cannot emit the same first symbol. Either every pair merges within two steps or is immediately incompatible, or

Xi=σs(Si−1)s(Si)X_i = \sigma s(S_{i-1})s(S_i)

for a nonconstant hidden sign ss.

Indeed, take a pair a,ba,b violating the first alternative. Failure to merge in one step, together with T>0T>0 and the binary alphabet, forces their labels to every destination jj to be deterministic and opposite, say f(j)f(j) and −f(j)-f(j). Immediate compatibility makes ff nonconstant. Failure to merge in two steps then forces every row indexed by f=+1f=+1 to be opposite every row indexed by f=−1f=-1. All positive rows therefore equal some row gg, and all negative rows equal −g-g. Comparing with row aa gives g=f(a)fg=f(a)f, hence

f(i,j)=f(a)f(i)f(j),f(i,j)=f(a)f(i)f(j),

the claimed parity form.

In the nonexceptional case, Travers’s theorem gives exponential convergence of hnh_n [46 Theorem 2 and Remark 3]. For uniformity, his finite-block coupling uses a block length bounded by some K(D)K(D): there are only finitely many labeled support patterns. Positive block entries are at least ηK(D)\eta^{K(D)}, and stationary hidden weights are at least δ\delta. His explicit estimates therefore give uniform constants.

An asymmetric parity sector produces string order. Absorb σ\sigma into the visible spins and write

Yi=s(Si),Xi=Yi−1Yi.Y_i=s(S_i), \qquad X_i=Y_{i-1}Y_i.

If the law of YY is not invariant under Y↦−YY\mapsto-Y, some bounded odd function GG of a finite block has mean g≠0g\ne0. The functions Y0G(Y−qq)Y_0G(Y_{-q}^q) and YnG(Yn−qn+q)Y_nG(Y_{n-q}^{n+q}) are sign-even, so they depend only on the corresponding visible endpoint blocks. Consequently there are bounded decorations fL,fRf_L,f_R such that

On=fL(X−q+10)(∏j=1nXj)fR(Xn+1n+q)=G(Y−q0)G(Ynn+q).O_n=f_L(X_{-q+1}^0)\left(\prod_{j=1}^{n}X_j\right)f_R(X_{n+1}^{n+q})=G(Y_{-q}^0)G(Y_n^{n+q}).

Mixing of the positive hidden chain gives EOn→g2\mathbb{E}O_n\to g^2.

Let the update window have width RR, including every spin read or flipped, and let rir_i be the total rate of odd-spin flips in its translate beginning at ii. Write a=Eria=\mathbb{E}r_i. Interior odd flips negate OnO_n, while only finitely many windows touch its endpoints. Stationarity therefore gives

0=E[LOn]=−2E[On∑i=1n−R+1ri]+O(1).0=\mathbb{E}[LO_n]=-2\mathbb{E}\left[O_n\sum_{i=1}^{n-R+1}r_i\right]+O(1).

The visible process is spatially ergodic, so n−1∑iri→an^{-1}\sum_i r_i \to a in L1L^1. Since OnO_n is uniformly bounded, division by nn gives ag2=0ag^2=0. But every finite visible word has positive probability in the parity representation: choose an initial sign, propagate it using the visible word, and choose hidden states in the required sign classes. Thus the nonzero odd rule implies a>0a>0, a contradiction. The law of YY must be sign-symmetric.

An invisible symmetry does not affect anything. The only case that remains is the trivial one: the hidden layer has an exact symmetry that nothing in the visible layer ever detects (a “strong symmetry” of the spatial evolution). In this case it produces no structure in the visible layer and can be ignored.

Entropy production of a quantum steady state with long-range CMI

Model. Write cj=(aj+ibj)/2c_j=(a_j+ib_j)/2 and dj=(aj+ibj+1)/2d_j=(a_j+ib_{j+1})/2, with aj,bja_j,b_j Majoranas. The open-chain fermionic model of Ref. [2], with boundary damping completed, has jumps

1−p cj (1≤j≤N),p dj (1≤j<N),p2b1,p2aN,(32)\sqrt{1-p}\,c_j\ (1\le j\le N),\qquad\sqrt{p}\,d_j\ (1\le j<N),\qquad\frac{\sqrt{p}}{2}b_1,\qquad\frac{\sqrt{p}}{2}a_N, \tag*{(32)}

where 0<p<10<p<1 and D[L]ρ=LρL†−{L†L,ρ}/2\mathcal{D}[L]\rho=L\rho L^\dagger-\{L^\dagger L,\rho\}/2. For a local spin realization, introduce auxiliary fermions gj=(γj+iδj)/2g_j=(\gamma_j+i\delta_j)/2, replace each jump FF anchored at jj by iγjFi\gamma_jF, and add unit-rate jumps −iγjδj-i\gamma_j\delta_j (j≤Nj\le N), iγjγj+1i\gamma_j\gamma_{j+1} (j<Nj<N), and γ1\gamma_1. Jordan–Wigner ordering g1,c1,…,gN,cNg_1,c_1,\ldots,g_N,c_N gives jumps on at most four consecutive spins; γ1\gamma_1 is the leftmost Pauli XX and removes the parity degeneracy. The presence of boundary terms ensures a unique steady state in the spin representation; rapid mixing and long-range CMI were already established. The origin of the long-range CMI is that the Lindbladian contains no terms repopulating the k=0k=0 mode of the cc fermions. Indeed, in steady state, ⟨ck†ck⟩∼k2\langle c_k^\dagger c_k\rangle\sim k^2, so the steady state is hyperuniform, analogous to our classical example.

Implementation. We implement the system-bath coupling in the “collision model” framework: system fermionic operators are alternately coupled to ancillas coupling respectively to the cc and dd fermions. Each ancilla has Hamiltonian ϵτz\epsilon\tau^z, and is initialized at temperature TT, i.e., excited-state probability f=(eϵ/T+1)−1f=(e^{\epsilon/T}+1)^{-1}. The ancilla couples to the system via V(t)=g(t)(F⊗s++F†⊗s−)V(t)=g(t)(F\otimes s^+ +F^\dagger\otimes s^-), where s+=∣1⟩⟨0∣s^+=|1\rangle\langle0| raises the ancilla. The time-dependent drive is needed to allow jump operators to try to “cool” fermions in distinct bases. In the weak-coupling limit the baths have the following jump operators:

(1−p)(1−f) cj,(1−p)f cj†,p(1−f) dj,pf dj†.(33)\sqrt{(1-p)(1-f)}\,c_j,\qquad\sqrt{(1-p)f}\,c_j^\dagger,\qquad\sqrt{p(1-f)}\,d_j,\qquad\sqrt{pf}\,d_j^\dagger. \tag*{(33)}

At f=0f=0, both channels empty the k=0k=0 mode: pumping vanishes there, but loss remains finite. For f≠0f\ne0 the Markov length scales as ξ∼p(1−p)/f\xi\sim\sqrt{p(1-p)/f}. Given this data and the steady-state occupation factors, the energy current into each bath is just the change in its occupation, and the entropy production per site and relaxation time is

σ=ϵT∫−ππdk2π(Jc+Jd)=p(1−p)(1−2f)log⁡1−ff=2p(1−p)log⁡ξ+O(1).(34)\sigma=\frac{\epsilon}{T}\int_{-\pi}^{\pi}\frac{dk}{2\pi}(J_c+J_d)=p(1-p)(1-2f)\log\frac{1-f}{f}=2p(1-p)\log\xi+O(1). \tag*{(34)}

This argument was for an explicit bath but it manifestly applies to any bath that cools the k=0k=0 mode to vacuum at unit rate: the combination of rapid dissipation and occupation near zero forces the cold bath to rapidly absorb heat, leading to a diverging entropy production rate in general.

References

  1. [1]U. Seifert, Stochastic thermodynamics, fluctuation theorems and molecular machines, Rep. Prog. Phys. 75, 126001 (2012).DOI
  2. [2]A.-R. Negari, Y. Zou and T. H. Hsieh, Instability of Markov chains under infinitesimal coherent dissipation, ArXiv:2610.18983 (2026).arxiv.org/abs/2610.18983
  3. [3]C. Godrèche and J.-M. Luck, Characterising the nonequilibrium stationary states of Ornstein–Uhlenbeck processes, J. Phys. A 52, 035002 (2019).DOI
  4. [4]J. Schnakenberg, Network theory of microscopic and macroscopic behavior of master equation systems, Rev. Mod. Phys. 48, 571 (1976).DOI
  5. [5]M. Esposito and C. Van den Broeck, Three faces of the second law. I. Master equation formulation, Phys. Rev. E 82, 011143 (2010).DOI
  6. [6]A. Gomez-Marin, J. M. R. Parrondo and C. Van den Broeck, Lower bounds on dissipation upon coarse graining, Phys. Rev. E 78, 011107 (2008).
  7. [7]P. Hayden, R. Jozsa, D. Petz and A. Winter, Structure of states which satisfy strong subadditivity of quantum entropy with equality, Commun. Math. Phys. 246, 359 (2004).
  8. [8]T. Kuwahara, K. Kato and F. G. S. L. Brandão, Clustering of conditional mutual information for quantum Gibbs states above a threshold temperature, Phys. Rev. Lett. 124, 220601 (2020).DOI
  9. [9]C.-F. Chen and C. Rouzé, Quantum gibbs states are locally markovian, arXiv preprint arXiv:2504.02208 (2025).arxiv.org/abs/2504.02208
  10. [10]A. Bakshi, A. Liu, A. Moitra and E. Tang, A dobrushin condition for quantum markov chains: Rapid mixing and conditional mutual information at high temperature, In Proceedings of the 58th Annual ACM Symposium on Theory of Computing, pp. 1508–1516 (2026).arxiv.org/abs/2510.08542
  11. [11]O. Fawzi and R. Renner, Quantum conditional mutual information and approximate markov chains, Communications in Mathematical Physics 340(2), 575 (2015).
  12. [12]M. Junge, R. Renner, D. Sutter, M. M. Wilde and A. Winter, Universal recovery maps and approximate sufficiency of quantum relative entropy: M. junge et al., In Annales Henri Poincaré, vol. 19, pp. 2955–2978. Springer (2018).
  13. [13]S. Sang and T. H. Hsieh, Stability of mixed-state quantum phases via finite Markov length, Phys. Rev. Lett. 134, 070403 (2025).
  14. [14]S. Sang, L. A. Lessa, R. S. K. Mong, T. Grover, C. Wang and T. H. Hsieh, Mixed-state phases from local reversibility, Phys. Rev. X 16, 041014 (2026), doi:10.1103/1kr5-jlcb.DOI
  15. [15]T. A. Kumar, Y. Zou, A.-R. Negari, R. G. Melko and T. H. Hsieh, Unlearnable phases of matter, ArXiv:2602.11262 (2026).arxiv.org/abs/2602.11262
  16. [16]J. Hauser, K. Su, H. Ha, J. Lloyd, T. G. Kiely, R. Vasseur, S. Gopalakrishnan, C. Xu and M. Fisher, Strong-to-weak symmetry breaking in open quantum systems: From discrete particles to continuum hydrodynamics, arXiv preprint arXiv:2602.16045 (2026).arxiv.org/abs/2602.16045
  17. [17]F. Hu, G. Liu, Y. F. Zhang and X. Gao, Local diffusion models and phases of data distributions, arXiv preprint arXiv:2508.06614 (2025).arxiv.org/abs/2508.06614
  18. [18]S. Sang, Z. Li, T. H. Hsieh and B. Yoshida, Ultrafast entanglement dynamics in monitored quantum circuits, PRX Quantum 4, 040332 (2023), doi:10.1103/PRXQuantum.4.040332.DOI
  19. [19]Y. Zhang and S. Gopalakrishnan, Nonlocal growth of quantum conditional mutual information under decoherence, Phys. Rev. A 110, 032426 (2024), doi:10.1103/PhysRevA.110.032426.DOI
  20. [20]S.-u. Lee, C. Oh, Y. Wong, S. Chen and L. Jiang, Universal spreading of conditional mutual information in noisy random circuits, Phys. Rev. Lett. 133, 200402 (2024), doi:10.1103/PhysRevLett.133.200402.DOI
  21. [21]Y. Zhang and S. Gopalakrishnan, Conditional mutual information and information-theoretic phases of decohered Gibbs states, ArXiv:2502.13210 (2025).arxiv.org/abs/2502.13210
  22. [22]A. G. D. van Enter, R. Fernández and A. D. Sokal, Regularity properties and pathologies of position-space renormalization-group transformations: scope and limitations of Gibbsian theory, J. Stat. Phys. 72, 879 (1993).DOI
  23. [23]Y.-H. Chen and T. Grover, Markov length can diverge in systems whose universal physics is spatially markovian, arXiv:2610.11235 (2026).arxiv.org/abs/2610.11235
  24. [24]C. Wang, Strong-to-weak spontaneous symmetry breaking, arXiv:2606.02555 (2026).arxiv.org/abs/2606.02555
  25. [25]J. M. Horowitz and J. L. England, Information-theoretic bound on the entropy production to maintain a classical nonequilibrium distribution using ancillary control, Entropy 19, 333 (2017).arxiv.org/abs/1707.00367
  26. [26]C. Nardini, É. Fodor, E. Tjhung, F. van Wijland, J. Tailleur and M. E. Cates, Entropy production in field theories without time-reversal symmetry, Phys. Rev. X 7, 021007 (2017).DOI
  27. [27]Y. I. Li and M. E. Cates, Steady state entropy production rate for scalar Langevin field theories, J. Stat. Mech. p. 013211 (2021).DOI
  28. [28]M. Esposito, Stochastic thermodynamics under coarse graining, Phys. Rev. E 85, 041125 (2012).pubmed.ncbi.nlm.nih.gov/22680437
  29. [29]D. Hexner and D. Levine, Noise, diffusion, and hyperuniformity, Phys. Rev. Lett. 118, 020601 (2017).DOI
  30. [30]S. Torquato, Hyperuniform states of matter, Phys. Rep. 745, 1 (2018).DOI
  31. [31]R. Maire and L. Chaix, Hyperuniformity and conservation laws in non-equilibrium systems, J. Chem. Phys. 163, 214507 (2025), ArXiv:2509.04242.DOI
  32. [32]M. Casiulis, S. Anand and S. Martiniani, Hyperuniform systems are maximally irreversible, ArXiv:2607.07411 (2026).arxiv.org/abs/2607.07411
  33. [33]A. Bakshi, A. Liu, A. Moitra and E. Tang, A dobrushin condition for quantum markov chains: Rapid mixing and conditional mutual information at high temperature, ArXiv:2510.08542 (2025).arxiv.org/abs/2510.08542
  34. [34]P. Rosa-Ruiz, M. Scandi, Á. Capel and Á. M. Alhambra, Static features from mixing in short- and long-range lindbladians: Markov property and correlations, ArXiv:2606.28054 (2026).
  35. [35]K. Tomita and H. Tomita, Irreversible circulation of fluctuation, Prog. Theor. Phys. 51, 1731 (1974).
  36. [36]R. Graham, Macroscopic potentials, bifurcations and noise in dissipative systems, In Noise in Nonlinear Dynamical Systems. Cambridge University Press (1989).
  37. [37]A. Kamenev, Field Theory of Non-Equilibrium Systems, Cambridge University Press, Cambridge (2011).
  38. [38]A. Gromov, A. Lucas and R. M. Nandkishore, Fracton hydrodynamics, Phys. Rev. Research 2, 033124 (2020).DOI
  39. [39]F. Martinelli, Lectures on glauber dynamics for discrete spin models, In Lectures on probability theory and statistics: Ecole d’été de probabilités de saint-flour xxvii-1997, pp. 93–191. Springer (2004).
  40. [40]C. Stahl, E. Lake and V. Khemani, Brownian ratchets and pumps universally simulate many-body active dynamics, ArXiv:2607.01231 (2026).arxiv.org/abs/2607.01231
  41. [41]T. L. Hill, Free Energy Transduction in Biology, Academic Press, New York (1977).DOI
  42. [42]P. Strasberg, G. Schaller, T. Brandes and M. Esposito, Thermodynamics of a physical model implementing a Maxwell demon, Phys. Rev. Lett. 110, 040601 (2013).pubmed.ncbi.nlm.nih.gov/25166147
  43. [43]C.-F. Chen, M. J. Kastoryano and A. Gilyén, An efficient and exact noncommutative quantum Gibbs sampler, ArXiv:2311.09207 (2023).arxiv.org/abs/2311.09207
  44. [46]N. F. Travers, Exponential bounds for convergence of entropy rate approximations in hidden Markov models satisfying a path-mergeability condition, Stoch. Proc. Appl. 124, 4149 (2014), Theorem and remark numbering as in arXiv:1211.6181v2.arxiv.org/abs/1211.6181

Paper details

vlnc:W_yykpwgm5
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Claude Fable 5 (bare), ChatGPT Astra 6

View statement
Looking for related papers…
Contents