PHYSICAL REVIEW LETTERS 124, 210605 (2020)

Universality Classes of Spin Transport in One-Dimensional Isotropic Magnets:

The Onset of Logarithmic Anomalies
Jacopo De Nardis,1 Marko Medenjak,2 Christoph Karrasch,3 and Enej Ilievski4,5
1
Department of Physics and Astronomy, University of Ghent, Krijgslaan 281, 9000 Gent, Belgium
2
Institut de Physique Théorique Philippe Meyer, École Normale Supérieure, PSL University, Sorbonne Universités,
CNRS, 75005 Paris, France
3
Technische Universität Braunschweig, Institut für Mathematische Physik, Mendelssohnstraße 3,
38106 Braunschweig, Germany
4
Faculty for Mathematics and Physics, University of Ljubljana, Jadranska ulica 19, 1000 Ljubljana, Slovenia
5
Institute for Theoretical Physics Amsterdam and Delta Institute for Theoretical Physics, University of Amsterdam,
Science Park 904, 1098 XH Amsterdam, Netherlands
(Received 3 February 2020; accepted 7 May 2020; published 27 May 2020)

We report a systematic study of finite-temperature spin transport in quantum and classical one-
dimensional magnets with isotropic spin interactions, including both integrable and nonintegrable models.
Employing a phenomenological framework based on a generalized Burgers’ equation in a time-dependent
stochastic environment, we identify four different universality classes of spin fluctuations. These comprise,
aside from normal spin diffusion, three types of superdiffusive transport: the Kardar-Parisi-Zhang
universality class and two distinct types of anomalous diffusion with multiplicative logarithmic corrections.
Our predictions are supported by extensive numerical simulations on various examples of quantum and
classical chains. Contrary to common belief, we demonstrate that even nonintegrable spin chains can
display a diverging spin diffusion constant at finite temperatures.

DOI: 10.1103/PhysRevLett.124.210605

Introduction.—Obtaining a theoretical framework that is comprising both the Drude weights [27–30] and diffusion
able to explain how macroscopic laws of transport emerge constants [20,31–34], together with a myriad of other
from the microscopic deterministic dynamics presents one applications [35–41]. This provided a coherent picture
of the central challenges of condensed matter physics. This for earlier numerical results (see, e.g., Refs. [42–46] and
transcends purely academic interest, as many problems of references therein). Despite these developments, the dis-
quantum transport remain unresolved in the presence of covery of superdiffusive spin transport and KPZ scaling
strong interactions [1,2]. One viable strategy to improve our [cf. Eq. (4)] in integrable spin chains with isotropic
understanding of transport phenomena is to identify uni- interactions, originally discovered numerically in the
versality classes and study certain representative instances Heisenberg spin-1=2 chain in Refs. [47,48] and further
which can either be solved exactly, or at least simulated surveyed in Refs. [49–53], came as a surprise. Although
numerically in an efficient manner [3–6]. In this respect, recent numerical works [52,54–56], in combination with
interacting many-particle systems confined to one spatial scaling arguments explaining the dynamical exponent [50],
dimension, in the realms of both quantum and classical constitute convincing evidence in support of the KPZ
models, take a special role as they often exhibit anomalous universality, a rigorous analytical account of this phenome-
features [7,8]. One of the prominent examples of a non- non is still lacking. Most recent studies suggest that
equilibrium universality class is given by the Kardar-Parisi- anomalous spin transport occurs only in integrable systems
Zhang (KPZ) equation [9], which is widespread in the area invariant under non-Abelian [SU(2) or SO(3)] Lie groups.
of growing one-dimensional interfaces [10,11]. The KPZ In contrast, normal spin diffusion is expected to be immedi-
and Lévy universality classes, which also occur in systems ately recovered upon breaking integrability [52,56]. The
of classical particles, can be understood in the scope of the occurrence of normal diffusion in nonintegrable symmetric
nonlinear fluctuating hydrodynamics [12–16]. chains was also suggested in the numerical study [57] (see
In recent years, the advent of the generalized hydro- also Refs. [53,56]) after a long-lasting controversy [58–63].
dynamics [17,18], studies of quantum chaos and its relation With the aim to resolve these controversies and giving
to transport [19–23], and of noisy quantum systems a comprehensive description of spin transport in these
[24–26], reinvigorated the field of transport laws in spin systems, in this Letter we address two key open questions:
chain models. In integrable quantum chains, a closed-form (i) Do nonintegrable homogeneous spin chains always
universal expression for the conductivity matrix was found, display normal diffusive spin transport? (ii) What are all

0031-9007=20=124(21)=210605(6) 210605-1 © 2020 American Physical Society

 PHYSICAL REVIEW LETTERS 124, 210605 (2020)

possible types of spin transport exhibited by rotationally and classical spin systems with isotropic interactions
invariant chains? valid on large spatiotemporal scales. Here we propose an
To address these questions, we carry out a systematic effective description, which is inherently classical in nature,
study of magnetization transport in classical and quantum by employing the continuum theory for a classical spin
spin systems in the nonmagnetized sector of thermal field, which is in turn treated within a hydrodynamic
equilibrium states where the global rotational symmetry approximation. Microscopic details are included implicitly
remains unbroken. Aiming at complete classification of through an appropriate phenomenological noise.
admissible transport laws, we build on a recent work by As our starting point, we consider the most general
Bulchandani [53] and devise a simple phenomenological form of a manifestly SO(3)-symmetric Hamiltonian equa-
model that helps us to single out two novel nonequilibrium tion of motion,
universality classes of spin transport. We corroborate our
δH
findings with extensive numerical simulations of classical S⃗ t ¼ −S⃗ × ⃗ S⃗ x ; S⃗ xx ; …;
¼ F ½S; ⃗ ¼ 1;
jSj ð1Þ
and quantum chains. Quite unexpectedly, we find that the δS⃗
answer to question (i) is negative: Surprisingly, the spin
diffusion constant is found to diverge logarithmically in time specified by some functional F involving scalar and vector
in rotationally invariant nonintegrable chains, thus refuting a products of the spin field and derivatives thereof. We can
widely held belief that nonintegrable models cannot display include classical lattice models as well, which are analyzed
infinite diffusion constants at finite temperatures. through their continuum counterparts. To additionally
Moreover, we find that rotationally invariant integrable incorporate quantum spin chains, we first perform a
chains can display different universal transport dynamics, mean-field average [64] of the microscopic spin
from KPZ-type to normal diffusive. Hamiltonian. It has to be stressed that such a correspon-
Therefore, from our perspective the results summarized dence cannot retain all quantitative features of spin
in Fig. 1 provide a comprehensive classification of dynamics. Nonetheless, we shall argue, in the spirit of
magnetization transport in homogeneous rotationally Ref. [53], that correspondence is still meaningful to capture
invariant quantum and classical spin models, nonintegrable the correct large-time transport behavior.
and integrable, and give a comprehensive response to The outlined effective theory applies in thermal equi-
question (ii). librium in the nonmagnetized sector (i.e., at half-filling)
We proceed by first introducing the formalism of an where the global rotational invariance of the underlying
effective spin field theory which we relate to the KPZ invariant measure is unbroken. This is of paramount
equation in a time-dependent noisy environment. To importance for the anomalous character of magnetization
systematically test our predictions, we subsequently con- dynamics, as the addition of finite chemical potential
centrate on a number of simple representative examples. (or external magnetic field), which dynamically breaks
Spin-field theory of isotropic magnets.—In order to the non-Abelian symmetry, leads to restoration of normal
capture various universality classes of spin superdiffusion, spin diffusion (accompanied in integrable systems by a
the task at hand is to devise an effective theory for finite spin Drude weight [66] or ballistic current). On large
finite-temperature magnetization dynamics in quantum spatiotemporal scales, the evolution of a spin field is
accurately captured by a “hydrodynamic soft mode”
carrying a negligible energy density, which can be con-
veniently described in terms of two intrinsic geometric
quantities, curvature κ ¼ ðS⃗ x · S⃗ x Þ1=2 and torsion τ ¼ κ −2 S⃗ ·
ðS⃗ x × S⃗ xx Þ. The soft mode pertains to the long-wavelength
[k ∼ Oð1=lÞ] limit of the spin procession (about a dis-
tinguished axis fixed by the perturbation which breaks the
gauge invariance assumed subsequently to be the z axis)
at constant
pffiffiffiffiffiffiffiffiffiffiffiffi ffi latitude Sz ¼ h, with Sx ðx; tÞ  iSy ðx; tÞ ¼
1 − h exp ½iðkx þ wtÞ, frequency wðkÞ ¼ −k2 h, and
2

dispersion EðkÞ ¼ 12 k2 ð1 − h2 Þ, which in terms of the curve

−1
FIG. 1. Universality classes of spin dynamics in rotationally p ffiffiffiffiffiffiffiffiffiffiffiffiffi helices with constant (on scale ∼l ) κ ¼
describes
2
invariant spin systems. The dynamical exponents z and b 1 − h k and τ ¼ hk [67]. Hydrodynamic modulation
characterizing the time-asymptotic behavior of the spin dynami- on a characteristic scale l, leading to scaling κ ∼ τ ∼
⃗
cal correlations hSðx; ⃗
tÞ · Sð0; 0Þi ∼ t−1=z log−b ðtÞ, as predicted by Oð1=lÞ, indicates that the energy density E ¼ κ 2 =2 ∼ l−2
the generalized noisy Burgers’ equation for the hydrodynamic
pffiffiffiffiffiffiffiffi is suppressed compared to the dynamics of torsion τ
evolution of the torsion component τt þ ½τn þ Dτx þ γðtÞηx ¼ (see also Ref. [53]). Further noticing that energy fluctua-
0 with a time-dependent noisy environment γðtÞ ∼ t−ζ . tions are diffusive (known to hold in both integrable and

210605-2
 PHYSICAL REVIEW LETTERS 124, 210605 (2020)

nonintegrable systems [33,68]), they can be effectively For n ¼ 2, Eq. (2) is just the ordinary noisy Burgers’
decoupled from the fluctuations of the torsion field pro- equation equivalent to the KPZ equation, up to a change of
vided the latter are superdiffusive. Based on this, we can variable [9]. A recent work [69] examined the properties of
conclude that torsion τ remains the only relevant scalar such KPZ equations with time-dependent noise term
field at large times and that, since τ ∼ h at small h, the of the form (3), finding nonuniversal large-time behavior
⃗
finite-temperature spin-spin fluctuations hSðx; ⃗
tÞ · Sð0; 0Þi for ζ > 1=2 and universal KPZ dynamics (with modified
are proportional to fluctuations of the torsional mode dynamical exponents) for ζ ≤ 1=2. Exactly at the “critical”
hτðx; tÞτð0; 0Þi. We shall assume that such a decoupling point ζ ¼ 1=2 corresponding precisely to diffusive spread-
mechanism holds generically for equations of the form (1). ing of microscopic excitations, in Ref. [69] the authors
Generalized noisy Burgers’s equation.—To account for deduce a modified diffusive scaling x ∼ t1=2 log2=3 ðt=t0 Þ
thermal fluctuations, we invoke the standard arguments (for the particular case the scaling should be understood
of the nonlinear fluctuating hydrodynamics (NLFHD) [14], as a lower bound and not a rigorous statement [70], as
where the microscopic degrees of freedom of the under- numerics are not able to distinguish slightly different
lying Hamiltonian dynamics are effectively taken into exponents; see also additional numerical data in Ref. [71]).
account through an appropriate stochastic term and effec- Keeping this in mind, the statistics of spin fluctuations in
tive diffusion (i.e., dissipation). This brings us to the this case is expected to exhibit a crossover from an effective
generalized noisy Burgers’ equation of the form KPZ dynamics at short-intermediate times t ≃ t0 ,
pffiffiffi
τt þ ðτn þ Dτx þ γ ηÞx ¼ 0; ð2Þ ⃗
hSðx; ⃗
tÞ · Sð0; 0Þi ≃ t−2=3 f KPZ ðΓKPZ xt−2=3 Þ; ð4Þ

where D ¼ γ=χ τ is the phenomenological diffusion con- to the asymptotic scaling of the form
stant, χ τ is static susceptibility of τ, ηðx; tÞ a white noise
with unit variance, γ is the effective variance of the noisy G½ΓG xt−1=2 log−2=3 ðt=t0 Þ
environment, and parameter n ≥ 2 specifies the degree of ⃗
hSðx; ⃗
tÞ · Sð0; 0Þi ≃ ; ð5Þ
t1=2 log2=3 ðt=t0 Þ
nonlinearity. From the general scaling relations τðx; tÞ ≃
t−1=2z fðxt−1=z Þ and hτðx; tÞτð0; 0Þi ≃ t−1=z gðxt−1=z Þ (here 2
and below, the bracket refers to the average with respect with Gaussian profile GðxÞ ≃ e−x in the limit t ≫ t0 .
to the canonical invariant measure), one however deduces The anomalous form (5) implies a divergent behavior
that z ¼ ðn þ 1Þ=2, which implies that nonlinearities of DðtÞ ∼ ½logðtÞ4=3 .
degree n ≥ 4 (with z > 2) are subdiffusive and thus Integrable isotropic magnets.—Our main example is the
R
irrelevant at large times. Although NLFHD has no pre- Heisenberg continuous magnet Hð2Þ ¼ 12 dxS⃗ x · S⃗ x (using
dictive power in this case, one generically expects to find standard notation S⃗ x ¼ ∂ x S,
⃗ etc., for partial derivatives),
normal spin transport. also known as the isotropic Landau-Lifshitz model [75,76],
The final key ingredient is to impose the structure of the which is a paradigmatic example of an integrable classical
noise, reflecting the nature of fluctuating modes which are field theory. The time evolution is governed by the non-
relevant on a hydrodynamic scale. In the spirit of conven- ð2Þ
tional NLFHD (see, e.g., Ref. [14] for application to linear partial differential equation S⃗ t ¼ FLL ¼ S⃗ × S⃗ xx .
This equation possesses an infinite family of
anharmonic chains), in the presence of long-lived ballis- R
tically propagating normal modes of Euler hydrodynamics, Hamiltonians HðnÞ in involution, e.g., H ð3Þ ¼ 12 dxS⃗ ·
R
we adopt a time-independent white noise γ ¼ γ 0 . The same ðS⃗ x × S⃗ xx Þ and Hð4Þ ¼ − 12 dx½S⃗ xx · S⃗ xx − 54 ðS⃗ x · S⃗ x Þ2 .
applies to integrable systems which exhibit extensively Invoking the “decoupling hypothesis” and the phenom-
many local conserved fields, as previously suggested in enological noisy environment, the torsional mode in each
Ref. [53]. On the contrary, generic spin systems do not HðnÞ is governed by the generalized Burgers’ equations (2)
support ballistic modes and excitations dissipate through with nonlinearity of degree n (see also Ref. [71] for the
the system. In this case, the variance γ is expected to obey details). Adopting a constant value for γ, the dynamics falls
the diffusive scaling and decay with time into the KPZ class at the lowest order n ¼ 2. This is,
however, no longer the case for n > 2 where the quadratic
γ ≡ γðtÞ ∼ ½t0 =ðt0 þ tÞζ with ζ ¼ 1=2; ð3Þ nonlinearity is absent. For n ≥ 4, the nonlinearity is
dominated by diffusive processes, and Hamiltonians
where t0 denotes an unknown model- and temperature- Hðn>3Þ thus do not display any enhancement of normal
dependent scale. The picture behind this is that fluctuations diffusion. The cubic n ¼ 3 case is however marginally
excited by the spatiotemporal variation of the chemical irrelevant in the dynamic renormalisation group sense. This
potential should dissipate away diffusively as their density type of nonlinearity has been previously examined in the
decays to zero with exponent ζ ¼ 1=2. study of Toom interface [77–79] and argued to result in a

210605-3
 PHYSICAL REVIEW LETTERS 124, 210605 (2020)

logarithmic-type correction. We shall corroborate on this

scenario later on.
The quantum Heisenberg hierarchy.—We proceed by
examining the spin diffusion constant in integrable quan-
P ˆ ˆ
tum spin-S Heisenberg chains, Ĥð2Þ ¼ j S⃗ j · S⃗ jþ1 ,
together with its hierarchy of local conservation laws
P ðnÞ ðnÞ
ĤðnÞ ¼ j ĥj with n-site densities ĥj (hats denotes
quantum operators). To our knowledge, these models
exhaust all homogeneous SU(2)-symmetric integrable
quantum spin chains with short-range interactions (exclud-
ing cases symmetric under higher-rank Lie groups).
Taking full advantage of quantum integrability and the
underlying quasiparticle picture [17,18], we have computed FIG. 2. Left: Spin diffusion constant as a function of logðtÞ,
the exact spin diffusion constant being the dominant extracted from numerical simulations of C0 ðtÞ, shown for a few
contribution to spin transport in the zero magnetization representative classical nonintegrable lattice spin models at
sector where the Drude weigh vanishes. The details of this infinite temperature. The dashed curves are fitting lines as
computation are spelled out in Ref. [71], where we show functions of logðtÞ. The Heisenberg and next-to-nearest-neighbor
ð2Þ
that Dðn≥4Þ < ∞. For the marginal case n ¼ 3, we however Hamiltonians belong to the Dlog class, with diverging diffusion
find a logarithmic divergence Dð3Þ ðtÞ ∼ logðtÞ. This is our DðtÞ → ðlog tÞ4=3 , whereas the nonintegrable lattice discretiza-
ðnÞ
first example of a logarithmically enhanced diffusion tions of HðnÞ (denoted by H lattice ) exhibit normal diffusion D for
ð3Þ n ≥ 3, with DðtÞ → const. Right: Plot of the spin diffusion
labeled by Dlog in Fig. 1. As a proof of principle, the
constant DðδÞ as a function of the interaction anisotropy δ for the
same conclusion can be independently reached by follow- classical anisotropic chain Hδ and the quantum nonintegrable
ing the lines of our phenomenological program, using spin-1 chain Ĥδ for two values of inverse temperature β ¼ 1=T.
that the appropriate classical mean-field continuous limits Dashed lines are fitting curves DðδÞ ¼ −a2 log ðc2 δÞ, with fitting
of the Hamiltonians Ĥ ðnÞ are the higher Landau-Lifshitz parameters a and c.
Hamiltonians HðnÞ (cf. Refs. [80,81]), which are sub-
sequently reduced to effective Burgers’ equations of the ð2Þ
form (2). constant (Dlog universality class in Fig. 1). The continuous
Nonintegrable classical and quantum spin chains.—To ð3Þ
counterpart of nonintegrable Hamiltonian Hlattice is instead
demonstrate how nonintegrable chains also fall in the described by a time-inhomogeneous cubic n ¼ 3 Burgers’
classification scheme of Fig. 1, we next examine the no- equation. Despite the lack of theoretical prediction for the
P
integrable classical Heisenberg chain HHeis ¼ j S⃗ j · S⃗ jþ1 , late-time dynamics in this case, our numerics, Fig. 2,
where the most recent works align in favor of normal spin indicates that spin dynamics is purely diffusive. Finally,
diffusion [57] (see also Refs. [53,56,82]). Performing ð4Þ
the simulation of H lattice nicely conforms with normal
numerical simulations with the method of Ref. [57] (which diffusion as expected from Eq. (2) with a higher degree
conserves exactly energy density and jS⃗ j j at all times; see of nonlinearity.
Ref. [71] for more details on the simulations) and checking As an independent test, we additionally considered the
P
in addition the next-nearest-neighbor scalar interaction,
P anisotropic model Hδ ¼ j S⃗ j · S⃗ jþ1 þ δŜzj · Ŝzjþ1 , with
HNNN ¼ j S⃗ j · S⃗ jþ1 þ 0.8S⃗ j · S⃗ jþ2 along with the (non- interaction anisotropy δ acting as a “regulator” which
integrable) lattice discretization of H ð3Þ and Hð4Þ flows restores normal diffusion; see also additional numerical
ðnÞ
(denoted by Hlattice ), we computed the time-dependent on- data in Ref. [71]. By computing the diffusion constant DðδÞ
site correlator C0 ðtÞ ≡ hSzL=2 ðtÞSzL=2 ð0Þi by flipping the and monitoring its value by approaching δ → 0þ , we find
behavior (cf. Fig. 2) compatible with a mild divergence
center spin S⃗ L=2 in a chain of length L ¼ 103 and average
over an ensemble of Oð106 Þ random configurations and DðδÞ ∼ j logðδÞjd ð6Þ
over time, with different time intervals Δt ≤ 0.05. The
time-dependent diffusion constant DðtÞ has been extracted with an exponent d ≈ 1. A reliable extraction of transport
with the aid of the diffusive ansatz C0 ðtÞ ∼ ½tDðtÞ−1=2 . The coefficient in quantum spin chains is a very demanding
numerical results shown in Fig. 2 (and additionally in task, as the rapid growth of entanglement entropy renders
Ref. [71]) confirm that both HHeis and HNNN are compatible time dependent matrix renormalization group simulations
with effective Burgers’ equation (2) with n ¼ 2 and with fixed bond dimension uncertain at large times.
diffusive noise ζ ¼ 1=2, yielding (contrary to claims in Despite such inherent issues, which are rarely acknowl-
Refs. [19,57]) a logarithmically divergent spin diffusion edged in the literature, a recent numerical study of isotropic

210605-4
 PHYSICAL REVIEW LETTERS 124, 210605 (2020)

spin-S chains [52] concluded in favor of normal spin Finally, It is reasonable to expect that the divergence (6)
diffusion (z ¼ 2) in nonintegrable spin chains (S ≥ 1). could be seen in a real experimental setting, and we hope
Here we prefer to facilitate a direct comparison with that this can be successfully addressed in the near future.
classical spin chains. To this end, we carried out time
We thank G. Barraquand, B. Bertini, V. Bulchandani,
dependent matrix renormalization group simulations
P. Le Doussal, Ž. Krajnik, J. Moore, T. Prosen, and H.
[83–85] of the quantum anisotropic spin-1 chain Ĥ δ and, Spohn for extensive and illuminating discussions, and the
restricting ourselves to only moderately small δ, numeri- organizers of the workshop “Thermalization, Many-Body-
cally extracted the spin diffusion constant from the time- Localization and Generalized Hydrodynamics” at the ICTS
dependent dc conductivity with a diffusive tail σðtÞ ≃ Bangalore for hospitality, where parts of this work were
ðχ=TÞD þ ct−1=2 with spin susceptibility χ and fitting carried out. J. D. N. is supported by the Research
parameters D and c; see additional numerical data in Foundation Flanders (FWO). E. I. is supported by the
Ref. [71]. The data shown in Fig. 2 indicate that the spin Slovenian Research Agency under the Programme
dynamics in the nonintegrable spin-1 chain mirrors that of P1-0402 and NWO Talent Programme Veni Grant
its classical counterparts and thus experiences the same No. 680-47-454 by the Netherlands Organisation for
ð2Þ
divergence (6) (Dlog class). Notice that in the quantum Scientific Research. C. K. acknowledges support by the
integrable S ¼ 1=2 chain, the divergence in the δ → 0þ Deutsche Forschungsgemeinschaft through the Emmy
limit is different as it diverges polynomially as D ∼ δ−1=2 Noether program (Grant No. KA 3360/2-1).
[20,32], signaling the onset of the KPZ dynamical exponent
at δ ¼ 0.
Conclusions.—We have proposed a phenomenological [1] V. J. Emery and S. A. Kivelson, Phys. Rev. Lett. 74, 3253
description of finite-temperature spin transport in one- (1995).
dimensional quantum and classical systems with isotropic [2] O. Gunnarsson, M. Calandra, and J. E. Han, Rev. Mod.
interactions. We have predicted four different classes, Phys. 75, 1085 (2003).
[3] K. Damle and S. Sachdev, Phys. Rev. B 57, 8307 (1998).
including in particular two distinct types of logarithmically
[4] G. Berkolaiko and J. Kuipers, Phys. Rev. E 85, 045201(R)
enhanced diffusion, which have not been previously dis- (2012).
closed in the context of many-body deterministic systems. [5] C. W. J. Beenakker, Phys. Rev. Lett. 70, 1155 (1993).
We have conjectured that in homogeneous SU(2) or SO(3) [6] N. Kulvelis, M. Dolgushev, and O. Mülken, Phys. Rev. Lett.
spin systems with short-range interactions, this list is 115, 120602 (2015).
exhaustive (cf. Fig. 1). While the outlined approach can [7] S. Lepri, R. Livi, and A. Politi, Phys. Rev. Lett. 78, 1896
adequately capture the qualitative features of spin dynamic (1997).
(dynamical exponents and logarithmic corrections), it does [8] S. Lepri, Phys. Rep. 377, 1 (2003).
not give access to exact values of transport coefficients [9] M. Kardar, G. Parisi, and Y.-C. Zhang, Phys. Rev. Lett. 56,
889 (1986).
or couplings of the effective hydrodynamical equations (2). [10] I. Corwin, Random Matrices Theory Appl. 01, 1130001
Important refinements in this direction are left to future (2012).
works. [11] K. A. Takeuchi, Physica (Amsterdam) 504A, 77 (2018).
Our findings shine some light on the puzzling observa- [12] H. Spohn, Large Scale Dynamics of Interacting Particles
tions in our previous work [51], which discusses spin (Springer, Berlin, 1991).
dynamics in the context of Haldane antiferromagnets. The [13] M. Prähofer and H. Spohn, J. Stat. Phys. 115, 255 (2004).
observed short-time behavior with approximate exponent [14] H. Spohn, J. Stat. Phys. 154, 1191 (2014).
z ¼ 3=2 (numerically detected also in Ref. [86]), later [15] M. Kulkarni, D. A. Huse, and H. Spohn, Phys. Rev. A 92,
043612 (2015).
argued in Ref. [52] to be merely a pronounced transient
[16] V. Popkov, A. Schadschneider, J. Schmidt, and G. M.
regime which crossovers into normal diffusion, indeed Schütz, Proc. Natl. Acad. Sci. U.S.A. 112, 12645 (2015).
plays nicely with the expected transient scenario assisted by [17] O. A. Castro-Alvaredo, B. Doyon, and T. Yoshimura, Phys.
an effective time-dependent noise. Nonetheless, we argue Rev. X 6, 041065 (2016).
now that despite the broken integrability, the spin diffusion [18] B. Bertini, M. Collura, J. De Nardis, and M. Fagotti, Phys.
constant does not saturate at asymptotically large times. Rev. Lett. 117, 207201 (2016).
This can be reconciled with the predictions of Ref. [51] [19] A. Das, S. Chakrabarty, A. Dhar, A. Kundu, D. A. Huse, R.
based on the effective low-energy quantum field theory Moessner, S. S. Ray, and S. Bhattacharjee, Phys. Rev. Lett.
provided that the transient is regulated by a temperature- 121, 024101 (2018).
[20] S. Gopalakrishnan, D. A. Huse, V. Khemani, and R.
dependent timescale diverging in the T → 0 limit, as a Vasseur, Phys. Rev. B 98, 220303(R) (2018).
consequence of the diverging effective lifetime t0 of [21] D. E. Parker, X. Cao, A. Avdoshkin, T. Scaffidi, and E.
the microscopic degrees of freedom in Eq. (3), despite Altman, Phys. Rev. X 9, 041017 (2019).
different mechanisms resembling the situation in gapless [22] V. Alba, J. Dubail, and M. Medenjak, Phys. Rev. Lett. 122,
one-dimensional systems [87,88]. 250603 (2019).

210605-5
 PHYSICAL REVIEW LETTERS 124, 210605 (2020)

[23] E. Leviatan, F. Pollmann, J. H. Bardarson, D. A. Huse, and [58] G. Müller, Phys. Rev. Lett. 60, 2785 (1988).
E. Altman, arXiv:1702.08894. [59] R. W. Gerling and D. P. Landau, Phys. Rev. B 42, 8214 (1990).
[24] T. Jin, A. Krajenbrink, and D. Bernard, arXiv:2001.04278. [60] J.-M. Liu, N. Srivastava, V. S. Viswanath, and G. Müller, J.
[25] D. Bernard and P. L. Doussal, arXiv:1912.08458. Appl. Phys. 70, 6181 (1991).
[26] D. Bernard and T. Jin, Phys. Rev. Lett. 123, 080601 (2019). [61] O. F. de Alcantara Bonfim and G. Reiter, Phys. Rev. Lett.
[27] E. Ilievski and J. De Nardis, Phys. Rev. Lett. 119, 020602 69, 367 (1992).
(2017). [62] M. Böhm, R. W. Gerling, and H. Leschke, Phys. Rev. Lett.
[28] V. B. Bulchandani, R. Vasseur, C. Karrasch, and J. E. 70, 248 (1993).
Moore, Phys. Rev. B 97, 045407 (2018). [63] N. Srivastava, J.-M. Liu, V. S. Viswanath, and G. Müller, J.
[29] B. Doyon and H. Spohn, SciPost Phys. 3, 039 (2017). Appl. Phys. 75, 6751 (1994).
[30] E. Ilievski and J. De Nardis, Phys. Rev. B 96, 081118(R) [64] For spin-S quantum chains linear in spin generators, this is
(2017). achieved by taking the expectation value on the SU(2) spin-
[31] J. De Nardis, D. Bernard, and B. Doyon, Phys. Rev. Lett. coherent states. Spin chains with nonlinear terms, where one
121, 160603 (2018). encounters path-integral anomalies [65], have to be ex-
[32] J. De Nardis, D. Bernard, and B. Doyon, SciPost Phys. 6, cluded from our framework.
049 (2019). [65] J. H. Wilson and V. Galitski, Phys. Rev. Lett. 106, 110401
[33] M. Medenjak, J. De Nardis, and T. Yoshimura, arXiv: (2011).
1911.01995. [66] J. Herbrych, P. Prelovšek, and X. Zotos, Phys. Rev. B 84,
[34] B. Doyon, arXiv:1912.01551. 155125 (2011).
[35] V. B. Bulchandani, R. Vasseur, C. Karrasch, and J. E. [67] While in the Landau-Lifshitz hierarchy of commuting
Moore, Phys. Rev. Lett. 119, 220604 (2017). flows helices are stationary states of the full nonlinear
[36] B. Doyon, T. Yoshimura, and J.-S. Caux, Phys. Rev. Lett. Hamiltonian dynamics, their fate in generic spin systems is
120, 045301 (2018). less clear. We nonetheless expect they become stabilized by
[37] B. Doyon, J. Dubail, R. Konik, and T. Yoshimura, Phys. effective decoupling of the curvature field at large times.
Rev. Lett. 119, 195301 (2017). [68] A. Das, K. Damle, A. Dhar, D. A. Huse, M. Kulkarni, C. B.
[38] M. Schemmer, I. Bouchoule, B. Doyon, and J. Dubail, Phys. Mendl, and H. Spohn, J. Stat. Phys., (2019),https://doi.org/
Rev. Lett. 122, 090601 (2019). 10.1007/s10955-019-02397-y.
[39] G. Misguich, K. Mallick, and P. L. Krapivsky, Phys. Rev. B [69] G. Barraquand, P. L. Doussal, and A. Rosso, Phys. Rev. E
96, 195151 (2017). 101, 040101 (2020).
[40] G. Misguich, N. Pavloff, and V. Pasquier, SciPost Phys. 7, [70] P. Le Doussal and G. Barraquand (private communication).
25 (2019). [71] See Supplemental Material at http://link.aps.org/
[41] O. Gamayun, Y. Miao, and E. Ilievski, Phys. Rev. B 99, supplemental/10.1103/PhysRevLett.124.210605 for details
140301(R) (2019). on numerical simulations and analytical calculations and
[42] J. Sirker, R. G. Pereira, and I. Affleck, Phys. Rev. B 83, extra numerical data, which includes Refs. [72–74].
035115 (2011). [72] L. Da Rios, Rend. Circ. Mat. Palermo 22, 117 (1906).
[43] C. Karrasch, J. Hauschild, S. Langer, and F. Heidrich- [73] M. Barros, A. Ferrández, P. Lucas, and M. A. Meroño, J.
Meisner, Phys. Rev. B 87, 245128 (2013). Geom. Phys. 31, 217 (1999).
[44] C. Karrasch, J. E. Moore, and F. Heidrich-Meisner, Phys. [74] B. Fuchssteiner, Prog. Theor. Phys. 70, 1508 (1983).
Rev. B 89, 075139 (2014). [75] L. Takhtajan, Phys. Lett. 64A, 235 (1977).
[45] R. Steinigeweg, J. Gemmer, and W. Brenig, Phys. Rev. B [76] L. D. Faddeev and L. A. Takhtajan, Hamiltonian Methods in
91, 104404 (2015). the Theory of Solitons (Springer, Berlin, 1987).
[46] R. Steinigeweg, F. Jin, D. Schmidtke, H. De Raedt, K. [77] B. Derrida, J. L. Lebowitz, E. R. Speer, and H. Spohn, Phys.
Michielsen, and J. Gemmer, Phys. Rev. B 95, 035155 (2017). Rev. Lett. 67, 165 (1991).
[47] M. Žnidaric, Phys. Rev. Lett. 106, 220601 (2011). [78] P. Devillard and H. Spohn, J. Stat. Phys. 66, 1089 (1992).
[48] M. Ljubotina, M. Žnidarič, and T. Prosen, Nat. Commun. 8, [79] M. Paczuski, M. Barma, S. N. Majumdar, and T. Hwa, Phys.
16117 (2017). Rev. Lett. 69, 2735 (1992).
[49] E. Ilievski, J. De Nardis, M. Medenjak, and T. Prosen, Phys. [80] V. A. Kazakov, A. Marshakov, J. A. Minahan, and K.
Rev. Lett. 121, 230602 (2018). Zarembo, J. High Energy Phys. 05 (2004) 024.
[50] S. Gopalakrishnan and R. Vasseur, Phys. Rev. Lett. 122, [81] T. Bargheer, N. Beisert, and N. Gromov, New J. Phys. 10,
127202 (2019). 103023 (2008).
[51] J. De Nardis, M. Medenjak, C. Karrasch, and E. Ilievski, [82] N. Li, Phys. Rev. E 100, 062104 (2019).
Phys. Rev. Lett. 123, 186601 (2019). [83] S. R. White, Phys. Rev. Lett. 69, 2863 (1992).
[52] M. Dupont and J. E. Moore, Phys. Rev. B 101, 121106 (2020). [84] U. Schollwöck, Ann. Phys. (Amsterdam) 326, 96 (2011).
[53] V. B. Bulchandani, Phys. Rev. B 101, 041411 (2020). [85] C. Karrasch, J. H. Bardarson, and J. E. Moore, Phys. Rev.
[54] M. Ljubotina, M. Žnidarič, and T. Prosen, Phys. Rev. Lett. Lett. 108, 227206 (2012).
122, 210602 (2019). [86] J. Richter, N. Casper, W. Brenig, and R. Steinigeweg, Phys.
[55] A. Das, M. Kulkarni, H. Spohn, and A. Dhar, Phys. Rev. E Rev. B 100, 144423 (2019).
100, 042116 (2019). [87] Y. Huang, C. Karrasch, and J. E. Moore, Phys. Rev. B 88,
[56] Ž. Krajnik and T. Prosen, J. Stat. Phys. 179, 110 (2020). 115126 (2013).
[57] D. Bagchi, Phys. Rev. B 87, 075133 (2013). [88] R. Vasseur and J. E. Moore, J. Stat. Mech. (2016) 064010.

210605-6