Noncommuting zero-noise and zero-frequency limits in particle-hole symmetric fluids Ewan McCulloch, Romain Vasseur, Sarang Gopalakrishnan Abstract In charged fluids obeying particle-hole symmetry, such as the Dirac fluid in graphene, charge transport is diffusive despite the presence of ballistically propagating sound waves: sound waves “hydrodynamically decouple” from the slower charge fluctuations. For quasi-one-dimensional fluids, we show that this symmetry-protected charge diffusion is not smoothly connected to the normal diffusion that arises when momentum conservation is broken by noise (or static impurities). Instead, the charge diffusion constant is a discontinuous function of noise, which (in the weak-noise limit) depends only on the ratio of momentum and energy relaxation rates. In the special limit of momentum-conserving noise (e.g., spatially uniform fluctuations of the Hamiltonian), the diffusion constant diverges in the presence of noise. We describe the resulting superdiffusion in terms of coupled Burgers equations. We present a general mechanism—hydrodynamic recoupling —by which weak noise can induce singular changes in transport coefficients. Our results highlight the limits of zero-noise extrapolation for predicting dynamical quantities like diffusion constants. Hydrodynamics dictates how systems relax to equilibrium under generic classical or quantum dynamics 27 ; 5 ; 13 ; 44 ; 30 . The predictions of hydrodynamics depend only on the symmetries of the system; long-lived deviations from equilibrium are due to fluctuations of the associated conserved charge densities. Within hydrodynamics, charge fluctuations generically relax diffusively (if momentum is not conserved) or spread ballistically (if it is). In the momentum-conserving case, charge density fluctuations are carried by ballistically propagating sound waves. It has long been appreciated that discrete symmetries can change this picture: for example, charge conjugation (or particle-hole) symmetry can prevent sound waves from coupling to a particular conserved charge 7 ; 12 ; 31 ; 18 . In this situation—which occurs most famously in graphene at charge neutrality 6 ; 32 —ballistic energy transport coexists with diffusive charge transport. This symmetry-protected charge diffusion qualitatively differs from the “regular” hydrodynamic diffusion that occurs in the presence of momentum relaxation: as we recently showed 16 ; 34 ; 17 , symmetry-protected diffusion (in quasi-1D geometries) leads to anomalous noise signatures, such as non-gaussian full counting statistics 28 . Figure 1: Singular behavior of the diffusion constant of quasi-one-dimensional two-component fluids as momentum and energy relaxation rates ( \gamma_{e} and \gamma_{p} respectively) are tuned to zero. At point \mathfrak{D} , momentum and energy are conserved. As this point is approached along a ray in the (\gamma_{p},\gamma_{e}) plane, the diffusion constant converges to a ray-dependent value (Eq. ( 2 )), which diverges as \gamma_{p}\to 0 . When \gamma_{p}=0 , charge transport is superdiffusive. The right panel shows the mechanism for a discontinuous diffusion constant: the autocorrelation function of the current develops a long timescale, due to repeated collisions of a charge element (marked in black) with sound waves. At the point “ \mathfrak{D} ” sound waves are ballistic and only collide once with each charge element (the resulting current autocorrelator is shown in gray). Energy (momentum) relaxation causes repeated collisions, giving rise to positive (negative) correlations. These correlations dominate the diffusion constant, which is proportional to the integrated current autocorrelation function. In the present work, we explore how symmetry-protected diffusion crosses over into regular diffusion when one weakly breaks the underlying conservation laws. Specifically, we consider a minimal hydrodynamic model with three conserved quantities—which for concreteness we will term energy, momentum, and charge—as well as a discrete particle-hole (or charge conjugation) symmetry. Our results are most striking for quasi-one-dimensional wire geometries, so we will focus on these. When all conservation laws are strictly satisfied, charge transport is diffusive, with a diffusion constant \mathfrak{D} . We introduce momentum and energy relaxation rates \gamma_{p},\gamma_{e} respectively, and explore how the diffusion constant D(\gamma_{e},\gamma_{p}) evolves as these two relaxation rates are tuned to zero. For generic noise, both relaxation mechanisms will be present, with some nonuniversal ratio. Our main conclusion is that the zero-noise limit is singular (Fig. 1 ): in the symmetric limit where energy and momentum are conserved, we have \mathfrak{D}\equiv D(\gamma_{e}=0,\gamma_{p}=0)=D_{\rm Fick}+D_{\rm conv}. (1) We will describe these terms in detail below. The contribution D_{\rm Fick} is regular as a function of \gamma_{e},\gamma_{p} , with (generically) a nonzero value when \gamma_{e}=\gamma_{p}=0 . However, at the symmetric point, there is an additional convective contribution D_{\rm conv} from the ballistic sound waves to the charge diffusion constant 15 ; 35 ; 16 ; 34 ; 19 . In the momentum and energy-breaking case, the diffusion constant instead scales as D(\gamma_{e},\gamma_{p})=D_{\rm Fick}+D_{\rm conv}(\gamma_{e}/\gamma_{p})^{1/2}+\ldots, (2) with \ldots representing corrections that are analytic in \gamma_{e},\gamma_{p} (and disappearing as \gamma_{e},\gamma_{p}\to 0 ) 1 . This result has important implications for recent computational approaches (on both classical and quantum computers) that aim to compute diffusion constants by zero-noise extrapolation 45 ; 29 ; 4 ; 46 ; 42 : the system we are considering presents an explicit example where the diffusion constant converges to some value at low noise, but this converged value is unrelated to the true zero-noise diffusion constant \mathfrak{D} . Previous work had established a discontinuity of this type in the fine-tuned case of integrable spin chains 37 subject to dephasing noise; we show here that the conclusion is in fact completely general, relying neither on integrability nor on the noise model. Another key result of our work is that the diffusion constant diverges when \gamma_{p}=0 and \gamma_{e}\neq 0 : i.e., when only momentum and charge are conserved, and particle-hole symmetry forbids them from coupling linearly to one another. Thus, weak momentum-conserving noise (e.g., global fluctuations of the Hamiltonian parameters) can cause a singular enhancement of charge diffusion, leading to superdiffusive charge transport. The coupled momentum and charge dynamics can be described using a recent hydrodynamic theory 8 ; 39 of coupled Burgers equations. This theory was introduced to describe certain fine-tuned limits of coupled asymmetric exclusion processes; we show how it emerges without fine-tuning. Three-mode hydrodynamics .—We briefly review the three-mode hydrodynamics of particle-hole symmetric fluids in quasi-one-dimensional geometries 16 ; 34 ; 48 ; 49 . In this work we will be concerned with linear response about equilibrium states at nonzero temperature. We consider systems with three conserved densities: energy e , momentum p , and charge n . These transform under discrete symmetries as follows: e,n are even under time-reversal and spatial reflection, while p is odd under these symmetries; e,p are even under charge conjugation, while n is odd under this symmetry. (The chemical potential conjugate to n is pinned by charge conjugation symmetry.) Given these constraints, we arrive at the following hydrodynamic equations 16 : \displaystyle\partial_{t}e+\partial_{x}p=0, \displaystyle\partial_{t}p+c^{2}\partial_{x}e=\ldots, (3) \displaystyle\partial_{t}n+\partial_{x}j_{n}=0, \displaystyle j_{n}=C_{2}pn-D_{\rm Fick}\partial_{x}n+\eta+\ldots In these equations, \eta denotes Gaussian white noise whose strength is set by the fluctuation-dissipation theorem, c is the speed of sound, and the coefficients C_{2},D_{\rm Fick} are O(1) nonuniversal constants. For convenience we set C_{2}=1 . Also, \ldots denotes terms of higher order in the gradient expansion, which can be neglected for the purposes of understanding the hydrodynamics of charge 16 . Among these neglected terms is the renormalization of the speed of sound c by equilibrium charge fluctuations. This renormalization can be absorbed into the definition of c , so that the equations for e,p form a closed system describing sound waves that that carry energy and momentum (but not charge). Charge diffusion constant .—The second line of Eq. ( 3 ) expresses the constitutive relation for the charge current. By particle-hole symmetry the two leading symmetry-allowed terms are the Fick’s law term (with coefficient D_{\rm Fick} , and accompanying noise \eta ) and a nonlinear “convective” coupling of charge to sound waves. Since (at this order) the motion of sound waves is unaffected by charge fluctuations, these fluctuations can be regarded as a source of extrinsic multiplicative noise. The charge diffusion constant can be defined in terms of the autocorrelation function of the charge density: 2\mathfrak{D}\,t\simeq\chi^{-1}\sum_{x}x^{2}\,\langle n(x,t)\,n(0,0)\rangle (where \chi is the charge susceptibility). Here, the stochastic average \langle\ldots\rangle involves two sources of noise: (i) the noise term \eta and (ii) the multiplicative “velocity” noise that arises because sound waves advect thermal fluctuations of energy and momentum density, imparting random kicks to the charge pattern. We will now compute this diffusion constant in two steps. First, we will artificially set D_{\rm Fick}=0 . In this limit, the density obeys the first-order differential equation \partial_{t}n+\partial_{x}(pn)=0 . This equation can be solved by dividing the density into small packets, each with an evolving position: n(x,t)=\int dx_{0}\,n(x_{0},0)\,\delta(x-X(x_{0},t)) , with characteristics solving \partial_{t}X=p(X(x_{0},t),t) . A crucial simplification is that the ballistic sound waves move parametrically faster than the packets of charge; therefore, one can replace p(X,t) with p(x_{0},t) up to subleading corrections 16 . Making this approximation, one arrives at a rigid shift of the charge pattern, n(x,t)\simeq n_{0}(x-X(t)),\quad X(t)\equiv\int_{0}^{t}d\tau\,p(x_{0},\tau), (4) where p(x_{0},\tau) is obtained by propagating the momentum along characteristics p(x,t)=p_{r}(x-ct)+p_{l}(x+ct) . The structure factor is then given by \langle n(x,t)n(0,0)\rangle\simeq\hskip-0.7113pt\int\hskip-1.42262ptdXP_{t}(X)\langle n_{0}(x-X)n_{0}(0)\rangle\simeq\chi P_{t}(x), (5) where P_{t}(x)\equiv\langle\delta(x-X(t))\rangle is the distribution of the random displacement X(t) . Therefore, in the pure convection limit, the charge diffusion constant \mathfrak{D}=D_{\rm conv} is equal to the diffusion constant of random displacement X (via the second moment of the structure factor). To complete the calculation, we restore Fickian processes, as follows. Consider conditioning on a particular history of sound waves and a particular initial charge configuration. On this history, the charge pattern is rigidly translated by X(t) . Beyond this rigid translation, the charge dynamics is due to Fickian processes. If one treats the elements of charge as particles, one can describe these two effects as follows: the particle configuration is randomly rearranged over the scale \sqrt{D_{\rm Fick}t} in addition to being shifted by \sqrt{D_{\rm conv}t} . The full diffusion constant comes from convolving these effects 16 ; 49 , n(x,t)=\int dx^{\prime}G_{\rm Fick}(x-x^{\prime},t)n_{0}(x^{\prime}-X(t)) : for each fixed X , the momentum-space charge structure factor at momentum k is S_{X}(k,t)=\chi\exp[-D_{\mathrm{Fick}}k^{2}t+ikX(t)] ; averaging this structure factor over the Gaussian variable X(t) gives the final result S(k,t)=\chi\exp[-(D_{\mathrm{Fick}}+D_{\mathrm{conv}})k^{2}t] , which yields Eq. ( 1 ). Since the convective and Fickian effects can be treated separately, we will focus below on the fate of the convective contribution when relaxation processes are allowed; the Fickian contribution can be restored at the end of the calculation. Relaxing momentum and energy .—We now consider the effects of relaxing energy and momentum at rates \gamma_{e},\gamma_{p} respectively. At D_{\rm Fick}=0 , the diffusion constant is still given by Eq. ( 5 ), but finding X(t) requires one to solve the hydrodynamics of energy and momentum in the presence of relaxation processes. The relevant equations can be written as: \displaystyle\partial_{t}e+\partial_{x}p=-\gamma_{e}e+\eta_{e}, (6) \displaystyle\partial_{t}p+c^{2}\partial_{x}e=-\gamma_{p}p+\eta_{p}+D_{p}\partial_{x}^{2}p+\partial_{x}\eta_{p}, where the dissipative coefficients \gamma_{e},\gamma_{p},D_{p} are accompanied by noise terms \eta_{e},\eta_{p},\eta_{p} to preserve the stationary measure. The diffusion constant is determined by the variance of the random displacement X(t) , which we express in Fourier space as \langle X(t)^{2}\rangle=\int\frac{dk}{2\pi}\int\frac{d\omega}{2\pi}\frac{\sin^{2}(\omega t/2)}{(\omega/2)^{2}}S_{pp}(k,\omega), (7) where S_{pp}(k,\omega) is the Fourier transform of \langle p(x,t)p(0,0)\rangle (see 1 ). The diffusive broadening of the sound peaks is subleading to the Euler-scale advection in the weak symmetry breaking limit ( \gamma_{e},\gamma_{p}\to 0 ). For simplicity, we now set D_{p}=0 , (see the supplemental materials 1 for the general case). Fourier transforming Eqs. ( 6 ) and solving for p(k,\omega) yields p(k,\omega)=\frac{(\gamma_{e}-i\omega)\,\eta_{p}(k,\omega)-ic^{2}k\,\eta_{e}(k,\omega)}{(\gamma_{e}-i\omega)(\gamma_{p}-i\omega)+c^{2}k^{2}}. (8) Using the noise correlators and the nascent delta-function {\rm sinc}^{2}(ta)^{2}\to\pi\,\delta(a)/t as t\to\infty , we find \langle X^{2}\rangle=2D(\gamma_{e},\gamma_{p})t , with D(\gamma_{e},\gamma_{p})=\int\frac{dk}{4\pi}\,\frac{2\gamma_{p}\chi_{p}\,\gamma_{e}^{2}+2\gamma_{e}\chi_{e}\,c^{4}k^{2}}{(\gamma_{e}\gamma_{p}+c^{2}k^{2})^{2}}\sim\sqrt{\frac{\gamma_{e}}{\gamma_{p}}}, (9) where \chi_{e} and \chi_{p} are energy and momentum static susceptibilities. This expression has two immediate and striking consequences. First, the diffusion constant along a general ray in (\gamma_{e},\gamma_{p}) space converges to a finite ray-dependent value that does not match continuously to \mathfrak{D} ; thus the zero-noise limit is singular. Second, it suggests that when energy (momentum) remains as a residual conserved quantity, the diffusion constant vanishes (diverges). This approach can be generalized to compute the charge structure factor and optical conductivity 1 . In the low frequency regime, fluctuating hydrodynamics predicts \sigma(\omega)=\chi\int\frac{dk}{2\pi}S_{pp}(k,\omega). (10) In the d.c. limit, we recover \sigma=\chi D with D given by eq. ( 9 ). As illustrated in Fig. 2 , the zero-noise and zero-relaxation rate results do not commute. The discontinuity of the d.c. conductivity is associated with broad regimes of anomalous super- or subdiffusive charge transport. To gain an intuition for Eqs. ( 9 ), ( 10 ), we now consider these two limiting cases, for which the analysis simplifies. Figure 2: Charge optical conductivity. The charge optical conductivity \sigma(\omega) at weak energy and momentum relaxation, predicted by fluctuating hydrodynamics, Eq. ( 10 ). For convenience, we have set D_{\textrm{F}ick}=0 , and c=\chi=\chi_{e}=\chi_{p}=D_{p}=1 . (Main) When both energy and momentum conservation are broken, the conductivity has a \gamma_{e}/\gamma_{p} dependent zero frequency limit. We show \sigma(\omega) for several values of \gamma_{e} and \gamma_{p} at fixed ratio \gamma_{p}/\gamma_{e}=10 (red) and \gamma_{p}/\gamma_{e}=0.1 (blue). As the relaxation rates are reduced at fixed ratio (from light to dark), the crossover to a small frequency plateau shifts to smaller frequencies, reflecting the noncommuting zero noise and zero frequency limits. (Inset) When only momentum is relaxed, \gamma_{e}=0 , the convective contribution to the conductivity vanishes as \sigma(\omega)\sim\sqrt{\omega} in the zero frequency limit (red). Whereas, when only energy is relaxed, \gamma_{p}=0) , the conductivity diverges as \sigma(\omega)\sim 1/\sqrt{\omega} (blue). Energy-conserving limit .—First, we consider setting \gamma_{e}=0 , for example by considering static disorder instead of noise. In this case, energy evolves diffusively, with a diffusion constant set by D_{e}\sim 1/\gamma_{p} . By Lorentz covariance, momentum density is equal to the energy current: consequently, X(t) is the integrated current of a diffusive quantity across a fixed point, and as such scales as X(t)\sim t^{1/4} . Thus, the convective contribution to charge transport becomes subdiffusive with dynamical exponent z=4 for any nonzero \gamma_{p} when \gamma_{e}=0 . In this regime, the convective contribution to charge transport is subleading to the Fickian term, and the diffusion constant is simply D_{\rm Fick} as \gamma_{p}\to 0 . Intuitively, the subdiffusive behavior can be understood as follows (see also 43 ; 37 ). In time t , a given packet of charge is repeatedly encountered by the same sound waves as they scatter back and forth. Since the backscattering of sound waves conserved energy, its momentum changes sign at each backscatter, so that successive encounters with a wave give opposite kicks. As a result, each sound wave contributes only an O(1) random kick to the net displacement of a charge packet over time t . Since the sound waves themselves diffuse, only O(t^{1/2}) independent sound waves cross the origin in time t . As a sum of t^{1/2} random kicks, the net displacement grows as X(t)\sim t^{1/4} . This limit clearly illustrates the crucial difference between the \gamma_{p}=0 and \gamma_{p}\neq 0 cases: in the former case each charge packet encounters each sound wave exactly once (by hydrodynamic decoupling), but for any nonzero \gamma_{p} the charge packet encounters each sound wave infinitely many times (in the quasi-1D geometry we are considering). Momentum-conserving limit .—In the energy-conserving limit, a charge packet encountered anticorrelated kicks from encountering the same sound wave first moving right, then left. When momentum is conserved and energy is not, one has precisely the opposite effect: a sound wave carrying positive energy (relative to the reference state) backscatter as one carrying negative energy. Both of these waves kick the charge packet in the same direction, so charge diffusion is enhanced rather than suppressed. Assuming linear diffusion of momentum, there are {\cal O}(t^{1/2}) independent sound waves that cross the origin, each doing so {\cal O}(t^{1/2}) times and therefore each contributing a (random signed) kick also of the size {\cal O}(t^{1/2}) , leading to a total displacement X(t)\sim t^{1/2}\times t^{1/4}=t^{3/4} . This clearly contradicts our assumption of hydrodynamic decoupling (since in the linear theory momentum is diffusive). To arrive at a consistent theory we need to revisit the hydrodynamic equations for the two remaining conserved quantities, momentum and charge. Including all symmetry allowed terms, these equations take the form: \displaystyle\partial_{t}p+\partial_{x}(C_{1}p^{2}+C_{2}n^{2})=D_{p}\partial_{x}^{2}p+\partial_{x}\eta_{p}, \displaystyle\partial_{t}n+\partial_{x}(np)=D_{\rm Fick}\partial_{x}^{2}n+\partial_{x}\eta_{n}. (11) Eq. ( Noncommuting zero-noise and zero-frequency limits in particle-hole symmetric fluids ) describes a “degenerate” limit of two coupled degenerate Burgers equations, recently introduced in Ref. 8 (see also 10 ) and explored further in Refs. 40 ; 41 (where the equations take “cyclic” form yielding a Gaussian stationary state). Previously considered realizations of Eq. ( Noncommuting zero-noise and zero-frequency limits in particle-hole symmetric fluids ) required fine-tuning to achieve the degenerate and cyclic conditions; in the present case, these conditions follow from the symmetries of the problem. Momentum and charge are strongly nonlinearly coupled, and both quantities scale with the Kardar-Parisi-Zhang (KPZ) 21 dynamical exponent x\sim t^{2/3} . Crossover timescales .—We now turn to the crossover timescales in the general case where \gamma_{p},\gamma_{e}>0 . To discuss these timescales, it is convenient to invoke the Kubo formula that relates the d.c. conductivity (or equivalently the diffusion constant) to the autocorrelation function of the charge current, D\propto L^{-1}\int_{0}^{\infty}dt\,\langle J_{n}(t)J_{n}(0)\rangle , where L is the system size and J_{n}=\int dx\,j_{n}(x) [cf. Eq. ( 3 )]. Neglecting the Fickian term, and assuming hydrodynamic decoupling, we have \langle j_{n}(x,t)j_{n}(0,0)\rangle\simeq\langle p(x,t)p(0,0)\rangle\langle n(x,t)n(0,0)\rangle . Since charge density is parametrically slower than momentum, the charge autocorrelation function is approximately a delta function, so D\propto\int_{0}^{\infty}dt\langle p(0,t)p(0,0)\rangle . The timescale on which this integral converges (and on which the asymptotic diffusion constant is achieved) is thus proportional to how non-Markovian the effective sound-wave “bath” is. In the absence of relaxation processes, \langle p(x,t)p(x,0)\rangle decorrelates rapidly, since sound waves move ballistically. Energy relaxation without momentum relaxation causes positive correlations, \langle p(x,t)p(0,0)\rangle\sim(\gamma_{e}t)^{-1/2} on timescales starting with the first back-scattering event \sim\gamma_{e}^{-1} (since momentum is now a diffusive mode with a large return probability). Conversely, momentum relaxation without energy relaxation causes sound waves to back-scatter, giving rise to anticorrelations in \langle p(x,t)p(x,0)\rangle\sim-\gamma_{p}^{-1/2}t^{-3/2} (since a returning sound wave is carrying the same energy but opposite momentum) 36 . The slower of the two relaxation processes acts as a long-time cutoff: beyond that timescale, sound waves cease to be defined 1 , and n is the only surviving hydrodynamic mode. For example, when energy relaxation is faster, the power-law dependence of \langle p(x,t)p(0,0)\rangle is cut off on a timescale \sim\gamma_{p}^{-1} , so the net contribution to the diffusion constant is \sqrt{\gamma_{p}/\gamma_{e}} . Even though the current autocorrelation function is always instantaneously close to its value in the absence of relaxation processes, the emergent long timescale gives rise to a singular correction to diffusion (Fig. 1 ). We dub this phenomenon “hydrodynamic recoupling”, as the sound waves that decoupled in the \gamma_{e}=\gamma_{n}=0 limit “recouple” in a singular way acting as a non-Markovian bath when \gamma_{e},\gamma_{n}>0 . On the diagonal ray \gamma_{e}=\gamma_{p} backscattering of sound waves is equally likely to reverse the wave’s momentum as not, so that consecutive collisions with a charge element provide uncorrelated kicks. Therefore, charge diffusion in the weak dissipation limit along this special ray corresponds to the fully conserved case—in both cases the kicks contributing to X(t) are effectively Markovian 1 —fixing the coefficient of \sqrt{\gamma_{e}/\gamma_{p}} to D_{\rm conv} in Eq. ( 2 ). Finally we remark that the scaling form ( 9 ) will be modified for \gamma_{p}/\gamma_{e}\ll 1 , because the nonlinearities in Eq. ( Noncommuting zero-noise and zero-frequency limits in particle-hole symmetric fluids ) must be included in the solution of Eq. ( 6 ). Charge transfer statistics .—While our discussion of charge transport has so far centered on the charge structure factor, we note that equilibrium charge transfer across a cut provides an equivalent probe of the charge diffusion constant 33 ; 47 : the asymptotic growth of the charge-transfer variance obeys 1 \langle\Delta Q(t)^{2}\rangle\simeq 2\chi\sqrt{Dt/\pi} . With this in mind, we briefly comment on how higher cumulants of charge transfer cross over as energy and momentum relaxation terms are added. For standard diffusion, these higher cumulants scale with the usual macroscopic fluctuation form t^{1/2} 9 ; 38 ; 3 ; 33 , while for particle–hole symmetric diffusion the n -th cumulant scales as t^{n/4} 24 ; 23 ; 25 ; 16 ; 34 ; 48 ; 49 . This enhancement of charge-transfer fluctuations reflects the spatiotemporal correlations in the effective noise “bath” generated by sound waves. When energy and momentum relaxation are both present, one generically expects a crossover to the standard t^{1/2} scaling, on a timescale set by the slower of the two relaxation rates. When only energy is conserved and D_{\rm Fick}=0 , the higher moments remain anomalous at all times, as computed in Refs. 16 ; 34 ; 48 ; 49 . The dynamics of higher moments in the complementary limit where only momentum is conserved is more subtle, and we defer a detailed analysis to future work. Figure 3: Singular charge diffusion at weak symmetry breaking. Numerical results from the stochastic gas model in the regime of weak energy and momentum relaxation. (Left) Time-dependent charge diffusion constant D(t) extracted from equilibrium current fluctuations (see 1 ), shown versus rescaled time t/\min(\gamma_{e},\gamma_{p}) for two fixed ratios r\equiv\gamma_{e}/\gamma_{p} : for r=4.25 (red) we show \gamma_{e}=0.006,0.009,0.012 (from dark to light); for r=0.04 (blue) we show \gamma_{e}=0.015,0.0,0.03 (dark to light). The black curve corresponds to the fully conserved case \gamma_{e}=\gamma_{p}=0 (for comparison to the weak dissipation data, we rescale time axis by 0.02\,t ). The plateau value depends nontrivially on the approach direction to (\gamma_{e},\gamma_{p})=(0,0) , and can be either enhanced or suppressed relative to the fully conserved case. These simulations used systems of length L=2000 . (Upper right) Total-current autocorrelator L^{-1}\langle J(t)J(0)\rangle for r=2 (red), r=0.02 (blue), and fully conserved case \gamma_{e}=\gamma_{p}=0 (black), illustrating how the sign structure of current correlations (correlated versus anticorrelated) depends on the relative strength of the two relaxation channels. These simulations used L=1000 . (Lower right) Weak-dissipation diffusion constant D(r) obtained from the long-time plateau, plotted as a function of r=\sqrt{\gamma_{e}/\gamma_{p}} , showing agreement with the predicted square-root singularity in Eq. (9). These simulations used L=2000 and simulation times of t=1000 . Numerical evidence .—To support the analytical picture developed above, we introduce a minimal stochastic gas model that realizes the same linear fluctuating hydrodynamics as symmetry-broken graphene at charge neutrality. The particle number of the gas plays the role of the energy density e , while energy in the stochastic gas is always strongly broken. Charge is represented by an additional label q_{i}\in\{-1,0,+1\} carried by the particles. The model is a controlled deformation of a multicomponent one-dimensional ideal gas. Particles move ballistically between collisions and stochastic symmetry-breaking events. At each collision the relative velocity is stochastically redrawn while preserving ideal-gas static correlations. Number conservation is broken by allowing momentum-conserving particle splitting and merging events at rate \gamma_{e} (we reuse \gamma_{e} since energy density in the Dirac fluid plays the role of particle density in the stochastic gas), while momentum conservation is broken by allowing stochastic sign flips of the center-of-mass velocity at collisions, at rate \gamma_{p} . In all moves, the total charge is conserved; however, at collisions the charge labels of the participating particles are randomly redrawn, subject to charge conservation, with probability p_{\rm mix} . This process generates a tunable Fickian contribution to charge transport. These deformations can be implemented so that the stationary state remains the grand-canonical (multicomponent) ideal gas, which we set to total particle density \rho=1 (with equal species split) and temperature k_{B}T=1 , while the long-wavelength dynamics is governed by the same linear hydrodynamic equations as Eq. ( 3 ). The symmetry-breaking rates \gamma_{p},\gamma_{e} are related to the microscopic probabilities (p_{\rm flp},p_{\rm merge}) by O(1) nonuniversal coefficients which we determine empirically. Full details of the microscopic rules and the empirical determination of the symmetry breaking rates are given in the supplementary material 1 . Figure 4: Pure momentum breaking ( \gamma_{e}=0 ). (Main panel) Equilibrium charge-transfer variance {\rm Var}(\Delta Q(t)) (plotted as {\rm Var}(Q_{\rm eq}(t)) ) with charge-label swaps disallowed ( p_{\rm swap}=0 , so D_{\rm Fick}=0 ). For \gamma_{p}=0 the transfer is diffusive at long times, while any nonzero \gamma_{p} produces a crossover on the timescale \gamma_{p}^{-1} to subdiffusive growth {\rm Var}(\Delta Q)\sim t^{1/4} (guide lines). These simulations used systems length L=1500 . (Inset) Restoring charge-label swaps ( p_{\rm swap}=0.05 ) yields a time-dependent diffusion constant D(t) with two clear plateaus: an early-time plateau corresponding to the sum of the Fickian and convective contributions, and a late-time plateau set by the bare Fickian diffusion constant D_{\rm Fick} . This simulation used L=6000 . We first consider the case where both number and momentum conservation are weakly broken. Holding the ratio \gamma_{e}/\gamma_{p} fixed and taking the weak-dissipation limit, we observe that the charge diffusion constant exhibits a singular dependence on the symmetry-breaking rates. Two representative ratios are shown in Fig. 3 , illustrating that the diffusion constant can be either enhanced or suppressed relative to the fully conserved case. Correspondingly, the current-current correlator has a positive (correlated) or negative (anti-correlated) long-time tail. We further scan over different ratios \gamma_{e}/\gamma_{p} at weak symmetry breaking and extract the charge diffusion constant. The numerical results are consistent with the predicted singular square-root dependence in Eq. ( 9 ). Next, we consider pure momentum breaking, setting \gamma_{e}=0 . To isolate the convective effect, we first eliminate Fickian charge diffusion by disallowing any charge-label swaps. In this limit, \gamma_{p} controls a crossover from diffusive to subdiffusive charge transfer, as shown in Fig. 4 (main panel). At short times the charge transfer grows diffusively, because sound packets propagate ballistically for long times before backscattering and therefore impart effectively uncorrelated kicks. The crossover to subdiffusive behavior occurs on the timescale \gamma_{p}^{-1} , when a sound packet revisit the same charge parcel with reversed velocity, producing kicks of opposite sign. We then restore Fickian diffusion by allowing charge-label exchanges. The resulting time-dependent diffusion constant exhibits two plateaus: an initial one corresponding to the fully conserved dynamics and a late-time plateau set by the bare Fickian contribution (Fig. 4 inset). Finally, we consider pure number breaking ( \gamma_{p}=0 ). In this regime momentum remains conserved and eventually enters the KPZ universality class. Numerically, we observe superdiffusive charge transfer with variance growing approximately as t^{\alpha} with \alpha\approx 0.69 , while the momentum transfer variance grows with exponent \alpha\approx 0.62 . Both charge and momentum transfer variance are shown in Fig. 5 (left). Figure 5: Pure number/energy breaking ( \gamma_{p}=0 ). (Left) Variances of equilibrium charge (black) and momentum transfer (red), \langle\Delta Q(t)^{2}\rangle and \langle\Delta P(t)^{2}\rangle , in the number-breaking, momentum-conserving regime (here \gamma_{e}=0.1 ). Over the accessible time window the data show effective power-law growth with exponents \alpha\approx 0.69 for charge transfer and \alpha\approx 0.62 for momentum transfer (guide lines), consistent with an intermediate-time regime between the linear hydrodynamics estimate ( \Delta Q\sim t^{0.75} , \Delta P\sim t^{0.5} ), and the expected KPZ scaling ( \Delta Q\sim\Delta P\sim t^{2/3} ). (Right) Total-current autocorrelator L^{-1}\langle J(t)J(0)\rangle , which decays approximately as t^{-1/2} , indicating nonintegrable long-time tails and superdiffusive charge transport in this regime. These simulations used L=1000 . This behavior can be understood as an intermediate-time regime. At early times, the momentum mode is described by the linearized hydrodynamics and appears diffusive. The convective contribution to the charge transfer then scales superdiffusively with \Delta Q(t)^{2}\sim X(t)\sim t^{3/4} (i.e., from repeated returns of backscattered sound waves). At longer times, however, momentum fluctuations cross over to KPZ scaling governed by the coupled Burgers equations in Eq. ( Noncommuting zero-noise and zero-frequency limits in particle-hole symmetric fluids ). In this regime, both momentum and charge transfer variances are expected to scale as t^{2/3} . One therefore expects intermediate-time effective exponents between 1/2 and 2/3 for momentum transfer and between 3/4 and 2/3 for charge transfer, consistent with the exponents observed numerically in Fig. 5 (left). Figure 5 (right) shows the corresponding current–current correlator, which decays approximately as t^{-1/2} , consistent with nonintegrable long-time tails and superdiffusive charge transport. Breaking particle-hole symmetry .—So far, we have assumed that particle-hole symmetry is exact. We now consider the consequences of weakly breaking it. In the three-mode hydrodynamics, particle-hole symmetry breaking allows a direct linear coupling between particle and energy current, i.e., j_{n}\sim C_{2}n_{0}p , where n_{0} is the net charge (at the symmetric point, n_{0}=0 ). Solving the Euler-scale linear hydrodynamics in this case, one finds a propagating (ballistic) mode and a purely diffusive mode; the particle density n couples to both modes. Thus the structure factor is the sum of a diffusive component (which dominates the local autocorrelation function) and a ballistic component (which dominates the d.c. conductivity). If one breaks weakly momentum conservation while retaining energy conservation, both modes become diffusive, but with very different diffusion constants. Thus, again, the diffusion constants associated with the conductivity and the local autocorrelation function are unrelated quantities. Discussion .—We have presented a hydrodynamic argument and unambiguous numerical evidence suggesting that the charge diffusion constant in a particle-hole symmetric fluid changes discontinuously upon adding perturbations that relax energy and/or momentum. Moreover, the diffusion constant in the weak noise limit takes a nonuniversal value that depends on the ratio of energy and momentum relaxing processes. An important consequence of this is that techniques using zero-noise extrapolation on the d.c. diffusion constant, applied to this setting, will yield answers that are unrelated to the true zero-noise value. Thus our result illustrates a potential failure mode of zero-noise extrapolation techniques that are routinely used in classical and quantum simulations. Of course, in the present case, the solution is straightforward—one has to compute the time-dependent diffusion constant and extrapolate it suitably—but requires detailed knowledge of the underlying hydrodynamic theory. A natural question is whether our results can be extended to two dimensions, where, again, a random walker returns infinitely often to the origin. In this setting, the analysis is complicated by the presence of shear modes, which give rise to singularities in the diffusion constant even absent momentum or energy relaxation. Whether an unambiguous signature of the convective mechanism can be found in higher dimensions (e.g., in nonlinear response) is an interesting question for future work. It would be interesting to test our predictions against experiments, both in classical colloidal systems (where multicomponent-fluid models are easy to implement) and in quantum systems such as graphene ribbons and multicomponent Bose or Fermi gases. The quantity we have been exploring is low-frequency charge transport, which is straightforward to explore in each of these settings. An especially interesting open question (which is beyond the scope of our methods) is how the convective effects we have been discussing change in the low-temperature limit, where the dominant energy fluctuations are quantum rather than thermal in character. Acknowledgments .—The authors thank Jacopo De Nardis, Enej Ilievski, and Luka Paljk for helpful discussions. S.G. acknowledges support from NSF DMR-2516303. The simulations presented in this article were performed on computational resources managed and supported by Princeton Research Computing, a consortium of groups including the Princeton Institute for Computational Science and Engineering (PICSciE) and Research Computing at Princeton University. Supplementary Materials: Noncommuting zero-noise and zero-frequency limits in particle-hole symmetric fluids Ewan McCulloch, 1,2 Romain Vasseur, 3 Sarang Gopalakrishnan 2 1 Laboratoire de Physique de l’École Normale Supérieure, CNRS, ENS & Université PSL; 24 rue Lhomond, 75005 Paris, France 2 Department of Electrical and Computer Engineering, Princeton University, Princeton, NJ 08544, USA 3 Department of Theoretical Physics, University of Geneva, 24 quai Ernest-Ansermet, 1211 Genève, Switzerland I Charge transfer variance and the diffusion constant In this appendix we relate the long-time growth of equilibrium charge-transfer fluctuations across a cut to the equilibrium diffusion constant D . We consider a conserved charge density n(x,t) in equilibrium, obeying the continuity equation \partial_{t}n(x,t)+\partial_{x}j(x,t)=0 , with static susceptibility is \chi\equiv\int dx\,\langle n(x,0)n(0,0)\rangle . We assume that the charge density is transported diffusively with diffusion constant D . That is to say, we assume that the retarded density–density correlator has a diffusive pole, G^{R}_{nn}(k,\omega)\xrightarrow[k,\omega\to 0]{}\frac{\chi Dk^{2}}{-i\omega+Dk^{2}}. (S1) This assumption specifies the universality class of transport but does not impose a microscopic constitutive relation such as Fick’s law. In particular, it allows for non-Gaussian full counting statistics arising from nonlinear or convective couplings to additional slow modes, since such effects enter only through higher cumulants. Using charge conservation, n(k,\omega)=\frac{k}{\omega}j(k,\omega) , the diffusive pole ( S1 ) implies \mathrm{Im}\,G^{R}_{jj}(k,\omega)=\chi D\,\omega+O(\omega k^{2}) as k,\omega\to 0 . It follows immediately that 26 ; 13 D=\frac{1}{\chi}\lim_{\omega\to 0}\lim_{k\to 0}\frac{1}{\omega}\,\mathrm{Im}\,G^{R}_{jj}(k,\omega), (S2) so that the diffusion constant entering the pole of G^{R}_{nn} is precisely the diffusion constant defined by the Kubo formula 26 . We define the integrated charge transfer across the origin, \Delta Q(t)\equiv\int_{0}^{t}d\tau\,j(0,\tau) . Using the continuity equation, this may be written exactly as \Delta Q(t)=\int_{0}^{\infty}dx\,\big[n(x,t)-n(x,0)\big] . The variance of the equilibrium charge transfer is therefore \langle\Delta Q(t)^{2}\rangle=2\int_{0}^{\infty}dx\int_{0}^{\infty}dx^{\prime}\,\Big[\langle n(x,0)n(x^{\prime},0)\rangle-\langle n(x,t)n(x^{\prime},0)\rangle\Big]. (S3) Using Eq. ( S1 ), the long-wavelength density correlator takes the form \langle n(x,t)n(0,0)\rangle=\chi\int\frac{dk}{2\pi}\,e^{ikx-Dk^{2}t} . Substituting this into the expression for the charge-transfer variance and performing the spatial integrals gives \langle\Delta Q(t)^{2}\rangle=2\chi\int_{0}^{\infty}\frac{dk}{\pi}\,\frac{1-e^{-Dk^{2}t}}{k^{2}}. (S4) Evaluating the remaining integral yields the universal asymptotic form 9 ; 38 ; 3 ; 33 \langle\Delta Q(t)^{2}\rangle\underset{t\to\infty}{\sim}\frac{2\chi}{\sqrt{\pi}}\,\sqrt{Dt}. (S5) II Singular charge diffusion from linear hydrodynamics with symmetry breaking In the main text we introduced an intuitive picture of charge transport based on isolated ballistically propagating sound waves that are stochastically backscattered by weak symmetry-breaking processes. The purpose of this appendix is to provide a complementary derivation of the same result within linear fluctuating hydrodynamics. While the toy model emphasizes individual ballistic excursions, the hydrodynamic calculation encodes their cumulative effect through long-wavelength equilibrium correlations. We focus on the convective contribution to equilibrium charge transport, which arises from the nonlinear coupling between charge and the energy current. At charge neutrality this provides a distinct contribution to equilibrium charge-transfer fluctuations, in addition to ordinary Fickian diffusion. In the weak symmetry-breaking regime, this contribution may be isolated by a decoupling assumption as described in the main text: to leading order in the symmetry-breaking rates, the sound modes evolve independently and act to advect the entire charge profile by a stochastic displacement 16 . Under this assumption, the convective contribution can be obtained by probing the momentum (energy current) p at a fixed spatial location. The resulting stochastic displacement is X(t)=\int_{0}^{t}dt^{\prime}\,p(0,t^{\prime}). (S6) The long-time statistics of X(t) determine the convective contribution to the equilibrium charge-transfer variance and hence to the equilibrium charge diffusion constant. Linear fluctuating hydrodynamics with symmetry breaking We consider linear fluctuating hydrodynamics for graphene energy transport in one dimension, formulated in terms of the energy density e(x,t) and the energy current p(x,t) . (The stochastic gas model introduced earlier provides a minimal microscopic realization of the same hydrodynamic structure, with e and p identified respectively with particle number and momentum density.) Weak symmetry breaking relaxes both energy and momentum, while diffusive corrections encode gradient-scale dissipation. The equations are \displaystyle\partial_{t}e(x,t)+\partial_{x}p(x,t) \displaystyle=-\gamma_{e}\,e(x,t)+\eta_{e}(x,t), (S7) \displaystyle\partial_{t}p(x,t)+c^{2}\partial_{x}e(x,t) \displaystyle=-\gamma_{p}\,p(x,t)+D_{p}\partial_{x}^{2}p(x,t)+\eta_{p}^{\rm OU}(x,t)+\partial_{x}\eta_{p}^{\rm cons}(x,t), (S8) where c is the sound velocity, \gamma_{e} and \gamma_{p} are the symmetry-breaking relaxation rates, and D_{p} is the Fickian diffusion constant for the energy current. The noise correlators are fixed by the fluctuation-dissipation theorem 27 , and are given by \displaystyle\langle\eta_{e}(x,t)\,\eta_{e}(x^{\prime},t^{\prime})\rangle \displaystyle=2\gamma_{e}\chi_{e}\,\delta(x-x^{\prime})\delta(t-t^{\prime}), \displaystyle\langle\eta_{p}^{\rm OU}(x,t)\,\eta_{p}^{\rm OU}(x^{\prime},t^{\prime})\rangle \displaystyle=2\gamma_{p}\chi_{p}\,\delta(x-x^{\prime})\delta(t-t^{\prime}), \displaystyle\langle\eta_{p}^{\rm cons}(x,t)\,\eta_{p}^{\rm cons}(x^{\prime},t^{\prime})\rangle \displaystyle=2D_{p}\chi_{p}\,\delta(x-x^{\prime})\delta(t-t^{\prime}), (S9) with all cross-correlators vanishing. These choices ensure a Gaussian stationary state. Eqs. ( S7 )–( S8 ) constitute the most general linear fluctuating hydrodynamic theory for energy density and energy current consistent with translation invariance, time-reversal symmetry, and spatial parity. (At linear order, no additional gradient terms are allowed in the energy continuity equation since p is defined as the energy current.) Fourier transforming Eqs. ( S7 )–( S8 ) in space and time yields \begin{pmatrix}-i\omega+\gamma_{e}&ik\\ ic^{2}k&-i\omega+\gamma_{p}+D_{p}k^{2}\end{pmatrix}\begin{pmatrix}e(k,\omega)\\ p(k,\omega)\end{pmatrix}=\begin{pmatrix}\eta_{e}(k,\omega)\\ \eta_{p}^{\rm OU}(k,\omega)+ik\,\eta_{p}^{\rm cons}(k,\omega)\end{pmatrix}. (S10) Inverting this matrix, we obtain p(k,\omega)=\frac{ic^{2}k\,\eta_{e}+(-i\omega+\gamma_{e})\bigl(\eta_{p}^{\rm OU}+ik\,\eta_{p}^{\rm cons}\bigr)}{(-i\omega+\gamma_{e})(-i\omega+\gamma_{p}+D_{p}k^{2})+c^{2}k^{2}}. (S11) Using the noise correlators above, the energy-current structure factor takes the form S_{pp}(\omega,k)\equiv\langle p(k,\omega)p(-k,-\omega)\rangle=\frac{2\gamma_{e}c^{4}\chi_{e}k^{2}+2(\gamma_{p}+D_{p}k^{2})\chi_{p}\bigl(\omega^{2}+\gamma_{e}^{2}\bigr)}{\left[\omega^{2}-c^{2}k^{2}-\gamma_{e}(\gamma_{p}+D_{p}k^{2})\right]^{2}+\omega^{2}(\gamma_{e}+\gamma_{p}+D_{p}k^{2})^{2}}. (S12) As discussed in the main text and above, the long-time growth of \langle X(t)^{2}\rangle defines the convective contribution to the equilibrium charge diffusion constant. Using translational invariance, \langle X(t)^{2}\rangle=\int_{0}^{t}dt_{1}\int_{0}^{t}dt_{2}\int\frac{dk\,d\omega}{(2\pi)^{2}}\,e^{-i\omega(t_{1}-t_{2})}S_{pp}(\omega,k). (S13) In the long-time limit, the double time integral yields \int_{0}^{t}dt_{1}\int_{0}^{t}dt_{2}\,e^{-i\omega(t_{1}-t_{2})}\;\xrightarrow[t\to\infty]{}\;2\pi t\,\delta(\omega), (S14) so that \langle X(t)^{2}\rangle\simeq 2Dt , with D given by D(\gamma_{e},\gamma_{p})=\int\frac{dk}{2\pi}\,S_{pp}(k,0)=\int\frac{dk}{2\pi}\,\frac{2\gamma_{e}\bigl(c^{4}\chi_{e}k^{2}+\chi_{p}\gamma_{e}(\gamma_{p}+D_{p}k^{2})\bigr)}{\bigl[\gamma_{e}(\gamma_{p}+D_{p}k^{2})+c^{2}k^{2}\bigr]^{2}}. (S15) Performing this integral exactly gives, D(\gamma_{e},\gamma_{p})=\frac{1}{4}\,\sqrt{\frac{\gamma_{e}}{\gamma_{p}}}\;\frac{c^{4}\chi_{e}+\chi_{p}\bigl(c^{2}+2\gamma_{e}D_{p}\bigr)}{\bigl(c^{2}+\gamma_{e}D_{p}\bigr)^{3/2}}. (S16) Double scaling limit We now evaluate Eq. ( S16 ) in the joint weak-dissipation limit where both symmetry-breaking rates are taken to zero at fixed ratio, \gamma_{e}=r\,\gamma_{p} with r>0 fixed, and \gamma_{e}\to 0 . In this limit \gamma_{e}D_{p}\ll c^{2} , and the exact expression simplifies, D(\gamma_{e},\gamma_{p})=\frac{c^{2}\chi_{e}+\chi_{p}}{4c}\sqrt{r}\left(1+O\!\Bigl(\frac{\gamma_{e}D_{p}}{c^{2}}\Bigr)\right). (S17) Thus, to leading order in the double scaling limit, D(\gamma_{e},\gamma_{p})\;\xrightarrow[\gamma_{e},\gamma_{p}\to 0]{\;}\frac{c^{2}\chi_{e}+\chi_{p}}{4c}\sqrt{\frac{\gamma_{e}}{\gamma_{p}}}. (S18) Equation ( S18 ) shows that in the weak-dissipation limit the convective contribution approaches a finite value that depends on the ray r=\gamma_{p}/\gamma_{e} along which the fully conserved point is approached, reflecting the noncommutativity of the limits \gamma_{e}\to 0 and \gamma_{p}\to 0 . Fully conserved case We now evaluate the convective contribution in the fully conserved limit \gamma_{e}=\gamma_{p}=0 , setting \gamma_{e}=\gamma_{p}=0 directly in Eq. ( S12 ). The variance of the walk X(t) is then given by \langle X(t)^{2}\rangle=\int_{0}^{t}dt_{1}\int_{0}^{t}dt_{2}\int\frac{dk\,d\omega}{(2\pi)^{2}}\,e^{-i\omega(t_{1}-t_{2})}\,\frac{2D_{p}\,\chi_{p}\,k^{2}\,\omega^{2}}{\left(\omega^{2}-c^{2}k^{2}\right)^{2}+\omega^{2}\left(D_{p}k^{2}\right)^{2}}. (S19) Performing the k integral gives I(\omega)\equiv\int_{-\infty}^{\infty}\frac{dk}{2\pi}\,S_{pp}(\omega,k)=\frac{\chi_{p}}{c}\,\left(\frac{1+\sqrt{1+\left(\frac{D_{p}|\omega|}{c^{2}}\right)^{2}}}{2\left[1+\left(\frac{D_{p}|\omega|}{c^{2}}\right)^{2}\right]}\right)^{1/2}. (S20) Using the nascent delta function identity in Eq. ( S14 ) gives \langle X(t)^{2}\rangle\;\xrightarrow[t\to\infty]{}t\int\frac{d\omega}{2\pi}\,\delta(\omega)\,I(\omega) . Since I(0)=\chi_{p}/c , we obtain \langle X(t)^{2}\rangle\simeq\frac{\chi_{p}}{c}\,t , and hence, comparing with \langle X(t)^{2}\rangle\simeq 2D_{\rm conv}t , the convective diffusion constant is D_{\rm conv}=\frac{\chi_{p}}{2c}. (S21) In the strictly conserved theory, the static susceptibilities \chi_{e} and \chi_{p} are not independent. Requiring a time-translation-invariant Gaussian equilibrium fixes the ratio of susceptibilities to the speed of sound. In our normalization this gives \chi_{p}=c^{2}\chi_{e} , so that Eq. ( S21 ) may be rewritten as D_{\rm conv}=(c^{2}\chi_{e}+\chi_{p})/4c , i.e., the weak dissipation limit on the ray \gamma_{e}=\gamma_{p} . Throughout this work we assume that the weak symmetry-breaking noise is implemented so as to respect the stationary ensemble of the fully conserved case (as in the deformed ideal gas studied in our numerics). Otherwise, the weak-dissipation limit and the fully conserved limit could differ for a trivial reason: the effective baths would impose different equal-time fluctuations (different \chi_{e},\chi_{p} ), and the diffusion constant would jump simply because the stationary state changes, rather than due to the non-Markovian effects that are the focus of this work. Charge structure factor in the energy conserving case Consider the rigid shift form of the charge density in Eq. ( 4 ). Its spatial Fourier transform gives n_{k}(t)\equiv\int dx\,e^{-ikx}\,n(x,t)=e^{-ikX(t)}. (S22) Assuming X(0)=0 (translational invariance), the corresponding dynamical structure factor is S_{nn}(k,t)\equiv\langle n_{k}(t)n_{-k}(0)\rangle=\chi\big\langle e^{-ikX(t)}\big\rangle. (S23) For \gamma_{e}=0 , X(t) is sum of independent random kicks from {\cal O}(\sqrt{t}) sound waves, and is therefore a Gaussian random variable. The generating function in Eq. ( S23 ) takes the form S_{nn}(k,t)=\chi\exp\!\left[-\frac{k^{2}}{2}\,\langle X(t)^{2}\rangle\right]. (S24) For \gamma_{e}=0 and D_{p}=0 the long-time growth of \langle X(t)^{2}\rangle is controlled by the slow diffusive mode with diffusion constant D_{E}=c^{2}/\gamma_{p} , yielding \langle X(t)^{2}\rangle\simeq\frac{2\alpha^{2}\chi_{p}}{c\sqrt{\pi}}\sqrt{t/\gamma_{p}}. (S25) Combining Eqs. ( S24 ) and ( S25 ), we obtain the stretched-exponential decay S_{nn}(k,t)\simeq\chi\exp\!\left[-\frac{\alpha^{2}\chi_{p}}{c\sqrt{\pi}}\sqrt{\frac{t}{\gamma_{p}}}\;k^{2}\right]. (S26) Equivalently, n(x,t) is subdiffusive with dynamical exponent z=4 . This structure factor is characteristic of the “XNOR” or “tracer diffusion” universality class discussed in Refs. 43 ; 37 ; 11 , that we derived here from a purely hydrodynamic perspective. Optical conductivity Going back to the general case, the formalism outlined above can also be used to compute the optical conductivity \sigma(\omega) given by the Kubo formula \sigma(\omega)=\lim_{k\to 0}\int_{0}^{\infty}dt{\rm e}^{i\omega t}S_{jj}(k,t), (S27) with S_{jj}(k,t) the current-current correlator. Using the continuity equation \partial_{t}n+\partial_{x}j=0 , we can compute the k\to 0 limit of the current-current correlator using the general form S_{nn}(k,t)=\chi\exp\!\left[-D_{\rm Fick}k^{2}t-\frac{k^{2}}{2}\,\langle X(t)^{2}\rangle\right]. (S28) of the structure factor. Note that the variable X(t) remains Gaussian in the general case, as it is a linear combination of Gaussian variables given by linearized hydrodynamics. We find \sigma(\omega)=\chi D_{\rm Fick}+\frac{\chi}{2}\int_{0}^{\infty}dt{\rm e}^{i\omega t}\frac{d^{2}}{dt^{2}}\langle X(t)^{2}\rangle, (S29) where the first term corresponds to the Fickian contribution, whereas the second one is due to convective terms. Finally, using the general expression ( 7 ) and focusing on the real part of the conductivity, we obtain \sigma(\omega)=\chi D_{\rm Fick}+\chi\int\frac{dk}{2\pi}S_{pp}(k,\omega). (S30) In the d.c. limit \omega\to 0 , we recover \sigma=\chi D with D=D_{\rm Fick}+\int\frac{dk}{2\pi}S_{pp}(k,0) the diffusion constant. III Exact quasiparticle representation of linear fluctuating hydrodynamics We show here that the toy model of reflected sound waves used in the main text is in fact an exact kinetic (Boltzmann) representation of the linear fluctuating hydrodynamics Eqs. ( S7 )–( S8 ). In the absence of the viscous D_{p}\partial_{x}^{2}p term, the linearized equations reduce to a telegrapher-type dynamics for chiral sound packets 14 ; 20 . The sound packets can be represented in terms of quasiparticles which carry both a propagation direction and a particle/hole character, and the stochastic quasiparticle dynamics reproduces the hydrodynamic equations without approximation. We introduce quasiparticle densities f_{a,s}(x,t) , where s=\pm labels the propagation direction with velocity v=s\,c , and a=\pm labels particle-like ( + ) versus hole-like ( - ) waves. The hydrodynamic fields are represented exactly as e(x,t)=\sum_{a,s}a\,f_{a,s}(x,t),\quad p(x,t)=c\sum_{a,s}a\,s\,f_{a,s}(x,t). (S31) The kinetic evolution consists of ballistic propagation together with two independent symmetry-breaking Poisson processes, corresponding to the two relaxation rates in Eqs. ( S7 )–( S8 ): (i) a number/energy-breaking event, which converts particle/hole type and reverses direction, (a,s)\xrightarrow{\gamma_{e}}(-a,-s) , and (ii) a momentum-breaking event, which reverses direction without changing particle/hole type, (a,s)\xrightarrow{\gamma_{p}}(a,-s) . These processes are encoded by the linear kinetic equation \displaystyle\partial_{t}f_{a,s}+sc\,\partial_{x}f_{a,s} \displaystyle=-(\gamma_{e}+\gamma_{p})\,f_{a,s}+\gamma_{e}\,f_{-a,-s}+\gamma_{p}\,f_{a,-s}+\zeta_{a,s}(x,t), (S32) where \zeta_{a,s} is Gaussian white noise. Choosing the noise covariance so that the stationary ensemble for e and p is Gaussian with susceptibilities \chi_{e} and \chi_{p} yields exactly the fluctuation–dissipation relations in Eq. ( S9 ), namely \langle\eta_{a}(x,t)\,\eta_{a}(x^{\prime},t^{\prime})\rangle=2\gamma_{a}\chi_{a}\,\delta(x-x^{\prime})\delta(t-t^{\prime}),\quad(a=e,p) where we have dropped the noise corresponding to the viscous term D_{p}\partial_{x}^{2}p for simplicity. With the identifications above, taking the linear combinations defining e and p in terms of quasiparticles (Eq. ( S31 )) reduces Eq. ( S32 ) to the two-mode fluctuating linear hydrodynamics equation in Eqs. ( S7 )–( S8 ) (after neglecting the viscous/Fickian term). In this form, the quasiparticle picture becomes precise: between stochastic events the quasiparticles propagate ballistically at \pm c , while energy-breaking events convert particle-like quasiparticles to hole-like quasiparticles (and reverse direction), and energy-current-breaking events reverse direction without changing the particle/hole type. The correlated or anticorrelated momentum kicks discussed in the main text are simply the statistics of repeated returns generated by these two Poisson processes. We note that the kinetic description in Eq. ( S32 ) is not unique 2 : different microscopic implementations of the Poisson processes (e.g. redistributing relaxation between the two flip channels or between explicit flips and quasiparticle birth–death noise) lead to the same closed linear equations for e and p with identical transport coefficients and noise correlators. This “gauge freedom” of the Boltzmann representation does not affect any hydrodynamic observables. \gamma_{e}=\gamma_{p} and the fully conserved case A useful simplification occurs on the diagonal line \gamma_{e}=\gamma_{p} . Although the quasiparticle labels (a,s) in Eq. ( S32 ) make the two flip processes look like reflections , one can exploit the non-uniqueness of the kinetic representation to choose variables in which no quasiparticle ever reflects: instead, right- and left-movers simply propagate ballistically and decay, with the stationary state maintained by birth noise. To see this explicitly, define the signed (particle–hole) densities for each propagation direction, g_{s}(x,t)\;\equiv\;\sum_{a=\pm}a\,f_{a,s}(x,t)\;=\;f_{+,s}(x,t)-f_{-,s}(x,t),\qquad s=\pm. (S33) In terms of these fields the hydrodynamic variables are e=g_{+}+g_{-} , and p/c=g_{+}-g_{-} . Multiplying Eq. ( S32 ) by a and summing over a gives a closed equation for g_{s} : \displaystyle\partial_{t}g_{s}+sc\,\partial_{x}g_{s} \displaystyle=-(\gamma_{e}+\gamma_{p})\,g_{s}+(\gamma_{p}-\gamma_{e})\,g_{-s}+\xi_{s}(x,t), (S34) where \xi_{s}\equiv\sum_{a}a\,\zeta_{a,s} is Gaussian white noise (its covariance is fixed by demanding the correct equal-time susceptibilities for e and p ). On the diagonal \gamma_{e}=\gamma_{p}\equiv\gamma , the coupling between opposite directions cancels identically and we obtain \partial_{t}g_{s}+sc\,\partial_{x}g_{s}=-2\gamma\,g_{s}+\xi_{s}(x,t). (S35) Thus, in this gauge there is no backscattering : right-movers remain right-moving, left-movers remain left-moving, and the only effect of the symmetry-breaking processes is an exponential decay of each chiral density. The stationary state is maintained by the birth–death noise \xi_{s} . This representation makes it particularly transparent why, at D_{p}=0 , the convective diffusion constant is independent of the common relaxation rate \gamma for any \gamma\geq 0 . Since there are no reflections, any quasiparticle crosses the origin at most once. Furthermore, the fluctuation–dissipation condition fixes the birth noise amplitude to be proportional to the decay rate, meaning that the total number of quasiparticles crossing the origin does not depend on \gamma . As a result, the time-integrated current autocorrelation \int_{0}^{\infty}dt\,\langle p(t)p(0)\rangle , and hence D(\gamma,\gamma) , is independent of \gamma , including the fully conserved case, \gamma=0 . This once again shows that the weak dissipation limit along the ray \gamma_{e}=\gamma_{p} corresponds to the fully conserved case. Reintroducing D_{p}>0 changes D(\gamma,\gamma) at fixed \gamma\geq 0 , but does not alter the diffusion constant in the weak dissipation limit. IV Stochastic symmetry–broken ideal gas We define a stochastic gas as a controlled deformation of a one–dimensional ideal gas of impenetrable point particles. This formulation makes explicit which conservation laws are broken and allows a direct demonstration of microscopic reversibility. We begin with a 1D ideal gas of identical particles of unit mass. Particles move ballistically between collisions, \dot{x}_{i}=v_{i} , and undergo elastic collisions. In this limit particle number, momentum, and energy are conserved, and the equilibrium measure is that of an ideal gas. We now systematically break the conservation laws by modifying the collision rules and adding “splitting” and “merging” rules. First, consider binary ( 2\!\to\!2 ) collisions. To break energy conservation while retaining momentum and particle number conservation, we conserve the center–of–mass velocity V=(v_{1}+v_{2})/2 while discarding the relative velocity u=v_{1}-v_{2} and drawing a new relative velocity u^{\prime} from a Rayleigh distribution \rho(u^{\prime})\propto|u^{\prime}|\,e^{-u^{\prime 2}/4T}. (S36) The outgoing velocities are then v_{1,2}^{\prime}=V\pm u^{\prime}/2 . This construction is closely analogous to the momentum–conserving Lowe–Andersen thermostat 22 , in which a stochastic bath interaction thermalizes the relative motion at temperature T while preserving the pair center–of–mass velocity (and hence total momentum). Unlike Lowe–Andersen, which applies such updates to nearby pairs (within a cutoff distance) at a Poisson rate independent of their relative speed, here the update is applied at actual collision events whose rate is proportional to the relative speed |u|=|v_{1}-v_{2}| , compensates exactly for the velocity bias in collision rates. Microscopic detailed balance for these collisions can be checked explicitly. For a collision taking incoming velocities (v_{1},v_{2}) to outgoing velocities (v_{1}^{\prime},v_{2}^{\prime}) , detailed balance requires P_{\mathrm{eq}}(v_{1},v_{2})\,W\!\left[(v_{1},v_{2})\to(v_{1}^{\prime},v_{2}^{\prime})\right]=P_{\mathrm{eq}}(v_{1}^{\prime},v_{2}^{\prime})\,W\!\left[(v_{1}^{\prime},v_{2}^{\prime})\to(v_{1},v_{2})\right], (S37) where P_{\mathrm{eq}} is the equilibrium measure and W the transition rate. Writing velocities in center–of–mass and relative coordinates, the Maxwell equilibrium weight factorizes as P_{\mathrm{eq}}(v_{1},v_{2})\propto e^{-v_{1}^{2}/2T}e^{-v_{2}^{2}/2T}=e^{-V^{2}/T}\,e^{-u^{2}/4T}, (S38) and similarly for the outgoing velocities with u replaced by u^{\prime} . The transition rate for the forward process is given by W\!\left[(v_{1},v_{2})\to(v_{1}^{\prime},v_{2}^{\prime})\right]\propto|u|\rho(u^{\prime}) , while the reverse process is given by W\!\left[(v_{1}^{\prime},v_{2}^{\prime})\to(v_{1},v_{2})\right]\propto|u^{\prime}|\rho(u) . Substituting these expressions into Eq. ( S37 ), both sides of the detailed balance condition reduces to |u||u^{\prime}|\,e^{-u^{\prime 2}/4T-u^{2}/4T} . To break particle–number conservation we introduce splitting and merging events. At a collision, two particles may merge into one with probability p_{\mathrm{merge}} , with the child inheriting the total momentum v=v_{1}+v_{2} . Independently, each particle may split into two at rate r_{\mathrm{split}} , producing children with velocities v_{1,2}=\frac{v}{2}\pm\frac{u}{2} , where u is again drawn from the Rayleigh distribution Eq. ( S36 ). Detailed balance between splitting and merging is verified by writing the local balance condition, P_{\mathrm{eq}}(v)\,W_{\mathrm{split}}(v\to v_{1},v_{2})=P_{\mathrm{eq}}(v_{1},v_{2})\,W_{\mathrm{merge}}((v_{1},v_{2})\to v), (S39) and using the Maxwell equilibrium weights, the Rayleigh distribution for the stochastic outcome (v_{1},v_{2}) in the splitting process, and the velocity bias term |u| in the merging process. Explicitly, P_{\mathrm{eq}}(v)W_{\mathrm{split}}(v\to v_{1},v_{2})\propto e^{-v^{2}/2T}\times|u|e^{-u^{2}/4T} , while P_{\mathrm{eq}}(v_{1},v_{2})W_{\mathrm{merge}}((v_{1},v_{2})\to v)\propto e^{-v_{1}^{2}/2T}e^{-v_{2}^{2}/2T}\times|u| . Using v_{1}^{2}+v_{2}^{2}=v^{2}/2+u^{2}/2 , the two expressions are identical. The ratio r_{\mathrm{split}}/p_{\mathrm{merge}} fixes the mean density, yielding a grand–canonical equilibrium ensemble. Finally, momentum conservation may also be broken by allowing the center–of–mass velocity to flip sign with probability p_{\mathrm{flip}} during 2\!\to\!2 collisions. Since the Maxwell distribution is even in V , this move is self–inverse and satisfies detailed balance trivially. Putting these ingredients together, each elementary stochastic move—Rayleigh relative velocity redraws at collisions, splitting and merging events, and center–of–mass velocity flips—satisfies microscopic detailed balance with respect to the same grand–canonical Gibbs measure, P(\{x_{i},v_{i}\})\propto\sum_{N}P_{\mathrm{GC}}(N)\prod_{i=1}^{N}e^{-v_{i}^{2}/2T}. (S40) As a result, equal–time snapshots of the system have ideal–gas static correlations: particle positions are Poisson distributed and velocities are independently Maxwell distributed at temperature T , despite the absence of energy, particle–number, or momentum conservation in the dynamics. We always break energy conservation in the stochastic gas. This choice is motivated by graphene hydrodynamics, where energy and energy current play roles analogous to particle number and particle current in the present model. The construction above is therefore a minimal (and simulable) model for these hydrodynamic slow modes, both near and far from equilibrium. To make closer contact with graphene hydrodynamics, we next introduce an analogue of electric charge and show that the same detailed–balance structure is preserved. Extension to charged particles We now extend the stochastic gas defined above by endowing each particle with a discrete charge label q\in\{-1,0,+1\} in order to introduce a conserved analogue of electric charge, while preserving microscopic reversibility and the ideal–gas character of the stationary state. The introduction of charge labels does not modify how velocities are updated in any stochastic event. In particular, for binary collisions, splitting events, merging events, and center–of–mass velocity flips, the post–event velocities are sampled exactly as for the chargeless case above, with no dependence on the charge labels. Since the velocity update rules are unchanged, the arguments establishing detailed balance and invariance of the Maxwell distribution at temperature T carry over here also. The velocity sector of the dynamics is therefore completely unaffected by the presence of charge labels. In the 2\to 2 collisions, we randomly redraw the charge labels from the same charge sector with probability p_{\rm mix} , otherwise the ordering (in space) of the labels is preserved. For example, if two particles collide with labels (+1,-1) , then with probability p_{\rm mix} , the outgoing labels are drawn randomly from \{(0,0),(+1,-1),(-1,+1)\} . Up to here, the charge sector is entirely a spectator to the particle dynamics. However, the particle dynamics are affected by the charge conservation law at merging and splitting events, which now include additional constraints. In a splitting event, a parent particle of charge q produces two children with charges (q_{1},q_{2}) satisfying q_{1}+q_{2}=q . The allowed splittings are +1\to(+1,0)\ \text{or}\ (0,+1),\qquad 0\to(0,0),\ (+1,-1),\ (-1,+1),\qquad-1\to(-1,0)\ \text{or}\ (0,-1), (S41) with no splitting events producing (+1,+1) or (-1,-1) . Merging events are defined as the exact inverses of these splittings: two particles may merge only if their charges sum to a value in \{-1,0,+1\} , and when a merge is allowed the child inherits the total charge q=q_{1}+q_{2} and the total momentum. The assignment probabilities of charge labels in splitting events determine the stationary single–particle charge distribution. We denote this distribution by \pi(q) , with \pi(+1)=\pi(-1) imposed by charge neutrality. The equilibrium measure for the charged gas then factorizes as P_{\mathrm{eq}}(\{v_{i},q_{i}\})\propto\prod_{i}e^{-v_{i}^{2}/2T}\,\pi(q_{i}). (S42) Microscopic detailed balance for charge–conserving splitting and merging follows by construction. Writing the local balance condition P_{\mathrm{eq}}(v,q)\,W_{\mathrm{split}}\!\left((v,q)\to(v_{1},q_{1}),(v_{2},q_{2})\right)=P_{\mathrm{eq}}(v_{1},q_{1};v_{2},q_{2})\,W_{\mathrm{merge}}\!\left((v_{1},q_{1}),(v_{2},q_{2})\to(v,q)\right), (S43) the splitting and merging weights take the explicit form W_{\mathrm{split}}=r_{\mathrm{split}}(q)\rho(u)p_{\mathrm{split}}(q\to q_{1},q_{2}) , and W_{\mathrm{merge}}=|u|p_{\mathrm{merge}}(q_{1},q_{2}) , where u=v_{1}-v_{2} , \rho(u) is the Rayleigh distribution, p_{\rm merge}(q_{1},q_{2}) is the probability that a merge of particles with labels q_{1},q_{2} occurs at a collision, and p_{\rm split}(q\to q_{1},q_{2}) is the probability of a particle with label q splitting into children with labels q_{1},q_{2} conditioned on a splitting event occuring. As in the chargeless case, the velocity–dependent factors cancel identically, and verifying detailed balance reduces to a purely charge–sector condition, \pi(q)\,r_{\mathrm{split}}(q)\,p_{\mathrm{split}}(q\to q_{1},q_{2})=\pi(q_{1})\pi(q_{2})\,p_{\mathrm{merge}}(q_{1},q_{2}),\qquad q_{1}+q_{2}=q. (S44) The probability of charge labels outcomes at splitting events therefore determine the stationary single–particle charge distribution \pi(q) . A natural and sufficient choice is to take the conditional splitting probabilities p_{\mathrm{split}}(q\to q_{1},q_{2})=\frac{\pi(q_{1})\pi(q_{2})}{\sum_{q_{1}^{\prime}+q_{2}^{\prime}=q}\pi(q_{1}^{\prime})\pi(q_{2}^{\prime})},\qquad q_{1}+q_{2}=q, (S45) with the merging probabilities p_{\mathrm{merge}}(q_{1},q_{2}) chosen accordingly so that Eq. ( S44 ) holds. With this choice, every allowed splitting process is paired with a merging process of equal weight. We work at charge neutrality \pi(+1)=\pi(-1)=(1-\pi(0))/2 and choose \pi(0)=1/3 for convenience, for which the conditional splitting probabilities reduce to p_{\mathrm{split}}(0\to q_{1},q_{2})=1/3 for the three allowed channels and p_{\mathrm{split}}(\pm 1\to q_{1},q_{2})=1/2 for the two allowed channels. For disallowed charge configurations, such as (+1,+1) or (-1,-1) , both the splitting process and the reverse merging process are forbidden. These channels therefore satisfy detailed balance trivially and do not affect the equilibrium measure. Calibration of symmetry-breaking rates The symmetry-breaking rates \gamma_{p} and \gamma_{e} entering the hydrodynamic theory are not, in general, equal to the bare microscopic probabilities p_{\rm flp} and p_{\rm merge} ; the mapping involves O(1) nonuniversal coefficients. Rather than attempting to compute these coefficients microscopically, we calibrate the mapping empirically using the measured diffusion constant D(p_{\rm flp},p_{\rm merge}) . Operationally, we have access to (i) the diffusion constant \mathfrak{D} in the fully conserved limit (no symmetry breaking), obtained numerically, and (ii) a set of measurements of D along a scan of ratios p_{\rm merge}/p_{\rm flp} in the weak-breaking regime. We then define \gamma_{p}\equiv p_{\rm flp} and \gamma_{e}\equiv a\,p_{\rm merge} , where the calibration constant a is chosen so that D(\gamma,\gamma)=\mathfrak{D} , consistent with the result from linear hydrodynamics (Appendix II) that the diagonal ray \gamma_{e}=\gamma_{p} reproduces the fully conserved diffusion constant. This choice simply fixes the relative units of \gamma_{e} and \gamma_{p} , ensuring that the line \gamma_{e}=\gamma_{p} in our plots corresponds, in the weak dissipation limit, to the charge diffusion constant \mathfrak{D} . For the data shown in Fig. 2 this procedure yields a\simeq 0.213 . We use this conversion, from (p_{\rm merge},p_{\rm flip}) to (\gamma_{e},\gamma_{p}) , for all figures shown in the main text.