Charge transport information scrambling and quantum operator-coherence in a many-body system with U1 symmetry Lakshya Agarwal1Subhayan Sahu2 3and Shenglong Xu1

2025-04-30 0 0 3.64MB 18 页 10玖币
侵权投诉
Charge transport, information scrambling and quantum operator-coherence in a
many-body system with U(1) symmetry
Lakshya Agarwal,1Subhayan Sahu,2, 3 and Shenglong Xu1
1Department of Physics & Astronomy, Texas A&M University, College Station, Texas 77843, USA
2Condensed Matter Theory Center and Department of Physics,
University of Maryland, College Park, MD 20742, USA
3Perimeter Institute for Theoretical Physics, Waterloo, Ontario N2L 2Y5, Canada
In this work, we derive an exact hydrodynamical description for the coupled, charge and operator
dynamics, in a quantum many-body system with U(1) symmetry. Using an emergent symmetry
in the complex Brownian SYK model with charge conservation, we map the operator dynamics in
the model to the imaginary-time dynamics of an SU(4) spin-chain. We utilize the emergent SU(4)
description to demonstrate that the U(1) symmetry causes quantum-coherence to persist even after
disorder-averaging, in sharp contrast to models without symmetries. In line with this property, we
write down a ‘restricted’ Fokker-Planck equation for the out-of-time ordered correlator (OTOC) in
the large-Nlimit, which permits a classical probability description strictly in the incoherent sector
of the global operator-space. We then exploit this feature to describe the OTOC in terms of a
Fisher-Kolmogorov-Petrovsky-Piskun (FKPP)-equation which couples the operator with the charge
and is valid at all time-scales and for arbitrary charge-density profiles. The coupled equations
obtained belong to a class of models also used to describe the population dynamics of bacteria
embedded in a diffusive media. We simulate them to explore operator-dynamics in a background
of non-uniform charge configuration, which reveals that the charge transport can strongly affect
dynamics of operators, including those that have no overlap with the charge.
CONTENTS
I. Introduction 1
II. Summary of the main results 2
III. Charged operator basis 4
IV. General procedure and emergent SU(4)
algebra 5
V. Map from operator strings to SU(4) basis 6
VI. Quantum operator coherence due to U(1)
symmetry 7
VII. The large-Nformalism and the ‘restricted’
Fokker-Planck equation 7
VIII. The OTOC at large N9
A. Free-fermionic chain 9
B. Charge-dependent FKPP equation 9
1. Constant charge density 10
2. Charge-transport and late-time
behavior 10
IX. A case study: operator dynamics in a domain
wall density background 11
A. Asymmetric butterfly velocity 11
B. Conserved vs. non-conserved operator 11
C. Charge dynamics influences the dynamics of
non-conserved operators 12
X. Conclusion and Discussions 12
XI. Acknowledgement 12
References 13
Appendix A. Anti-commuting fermions on
multiple time-contours 15
Appendix B. Charge-resolution of
null-eigenstates 15
Appendix C. Solving the restricted Fokker-Planck
equation 16
Appendix D. Analytic solution of the
free-fermionic chain 16
1. Charge transport at finite-N16
2. OTOC at finite-N18
I. INTRODUCTION
In many-body dynamics, a fine-grained approach to
study thermalization [14] comes from the study of
operator-dynamics [57]. This is also related to the con-
cept of information scrambling, where local information
flows to non-local degrees of freedom and is not retriev-
able via local measurements [813]. The notions of op-
erator growth and information scrambling can be pre-
cisely measured using the out-of-time-ordered correlator
(OTOC):
F(W(t), V ) = 1
tr(I)tr(W(t)VW(t)V)(1)
This quantity is often used to measure the size of the
evolving operator W(t) and shows interesting features
arXiv:2210.14828v3 [cond-mat.str-el] 30 Nov 2022
2
such as Lyapunov growth and butterfly velocity [11,14
27]. The OTOC is experimentally accessible on various
platforms [2838], where it is often used as a measure of
information scrambling.
Previous work on random circuits and noisy-driven
models has established several related effective classical
models of operator dynamics, including biased random
walk [5], reaction diffusion process [39] and population
dynamics [40]. These classical models are used to fit and
understand experimental data of the OTOC on quantum
platforms [34,41]. The models are obtained by mapping
the unitary operator dynamics to a classical stochastic
process by disorder average. While this approach is fea-
sible for usual noisy/random models without symmetry,
adding charge conservation makes the problem more dif-
ficult as it causes quantum coherence to persist at the
operator level, even after disorder-averaging (Fig. 1(e)).
However, such a classical description is important to ob-
tain for charged models, as thermalization in systems
with charge conservation has recently become accessible
on quantum simulators [42].
There is extensive literature on operator-dynamics [43
55] and spectral form factors [5661] in the presence
of symmetries. The primary mechanism through which
charge, or other conserved quantities, can influence oper-
ator growth is by confining the access of growing opera-
tors to a specific (symmetry) sector of the Hilbert space.
In addition, within extended systems, the conservation
law leads to transport of local conserved quantities, which
couples to the operator dynamics. Therefore the correct
semi-classical picture of operator dynamics in the pres-
ence of a symmetry should at least contain two dynamical
variables, the operator size and the local density of the
conserved quantity.
In this work, we study operator dynamics in the pres-
ence of charge transport. We derive the required semi-
classical equations which couple the charge and the oper-
ator, that are valid even in inhomogeneous and dynamical
charge-density backgrounds:
tρ=2
rρ
tξ=2
rξ+ 2g2ξ(ξ2ρ(1 ρ)) (2)
Here ρ(r, t) is the charge density, which obeys the diffu-
sion equation, while ξ(r, t) is the analog of operator-size
in models with symmetries that measures local scram-
bling of the operator. The dynamics of ξis described by a
diffusion-reaction equation that depends on the dynami-
cal charge density. In particular, the local density bounds
the range of ξfrom 0 to pρ(1 ρ). We note that this
class of equations has been independently studied in the
context of bacterial population growth in diffusive media,
where it is denoted as the ‘Diffusive Fisher–Kolmogorov
equation’ [62].
To derive these equations, we use the complex Brow-
nian SYK model on a lattice with U(1) symmetry. SYK
models, which are all-to-all interacting models with ran-
dom couplings [63], are well-known for their connection
to black-hole physics [64,65] and for their exactly solv-
able nature in the large-Nlimit [1720]. Their Brownian
limit, which considers time-dependent couplings, has also
been studied in various contexts [48,50,6670]. The dy-
namics of Brownian models can be mapped to a stochas-
tic process [39,69,71], or the imaginary time dynamics of
bosonic-models [67]. It has also been shown that Brown-
ian SYK models, in particular, give rise to emergent sym-
metry structures after the disorder averaging procedure
which can be used to compute the OTOC both at large
finite Nand in the infinite-Nlimit [50]. In this work,
we show that the extended complex Brownian model is
mapped to a quantum SU(4) spin chain with inter-site
Heisenberg coupling and intra-site interaction. In the
large Nlimit, the microscopic quantum model is reduced
to the semi-classical equations in Eq. (2). We provide a
complete picture of the coupled dynamics between oper-
ators and charge. Our approach can also be extended to
systems with other symmetries.
The rest of the paper is organized as follows: Sec. II
provides a brief summary of the main results. In Sec. III,
we discuss the operator-basis which respects U(1) sym-
metry and can therefore be used to precisely characterize
the four dynamical charges involved in the computation
of the OTOC. In Sec. IV, we describe the details of the
complex SYK model, and the mapping of the effective
imaginary-time evolution in the Brownian model to an
SU(4) spin-chain. In Sec. Vwe describe the mapping of
operator-states of complex fermions to GT-patterns in
the SU(4) algebra. We also map the physical charges to
the weights of the algebra. Sec. VI covers the primary dis-
tinguishing feature of operator dynamics restricted by a
U(1) symmetry, namely the presence of transitions which
introduce quantum operator coherence. We describe the
procedure to compute the infinite-Nlimit in Sec. VII,
within which we also describe the general structure of the
‘restricted’ Fokker-Planck equation obtained for quan-
tum many-body models with charge conservation. In
Sec. VIII, we derive the equation governing the OTOC,
namely the charge-dependent FKPP equation. This re-
veals some interesting features of operator-dynamics in
the presence of symmetries, as we show in section Sec. IX,
where we examine operator-dynamics in different charge
domain-wall backgrounds and observe the strong influ-
ence of charge dynamics on even non-conserved opera-
tors. Sec. Xconcludes this work with a brief summary
and some discussions on a contrasting view of charge vs.
energy conservation.
II. SUMMARY OF THE MAIN RESULTS
The primary result of our work is captured by Eq (2).
The equation effectively models the evolution of the
OTOC depicted in Eq. (1), for a charge-conserving
fermionic model defined on a chain (Fig. 1(a)). The steps
via which this connection is established are:
To begin, one specifies the charge density on each
3
1 2
L
N
r
Coherent Incoherent
FIG. 1. (a) The Brownian SYK model with Lclusters of
Ncomplex fermions, with on-site interaction Jand near-
est neighbor hopping K. The total fermion number operator
Pnr,i is the conserved operator, leading to a U(1) symmetric
circuit. (b-d) The time evolution of charge ρ(b),the OTOC
F(W(t), χr) (c), and OTOC F(W(t), nr) (d), when the ini-
tial operator Whas an initial charge distribution as shown
in (b). In each of the graphs the darkness of the plots in-
creases with increasing times. The charge (b) and conserved
part of the operator (d) have diffusive behavior, while the
OTOC F(W(t), χr) (c) and the uncharged part of OTOC
F(W(t), nr) propagates ballistically. (e) Schematic of the
time evolution of two copies of the operator, which is involved
in the OTOC computation. Due to the U(1) symmetry, the
super-operator develops coherence, which makes the hydro-
dynamic description hard. In this work, we argue that for
certain probing operators, the correction from this induced
coherence is 1/N suppressed. In the N→ ∞ limit, we di-
rectly obtain the hydrodynamical equations 2, which lead to
the dynamical evolution in (b-d).
site r. This fixes the profile of ρ(r, t = 0).
Because the initial charge on every site has been
fixed, the simplest initial operator is the projec-
tion operator onto a symmetry sector with a given
charge density on every site.
In models where the charge is conserved on each
site, this projector is static. However, once the
charge is allowed to flow from site to site, this op-
erator becomes dynamical as well.
Just as there are multiple states within each charge
sector, there are also multiple operators [50]. The
variable ξcontrols the choice of the operator once
the charge is fixed. For example, the maximal value
of ξ=pρ(1 ρ) represents a simple operator such
as a projector, while ξ= 0 represents a local scram-
bled operator within the charge sector. This is con-
sistent with ξ= 0 being a stable solution of Eq. (2),
as all simple operators will eventually evolve into
the most complex one.
Fixing both ρ(t= 0) and ξ(t= 0) also completely
fixes the initial global operator W(t= 0) in Eq. (1).
Hence the operator Wcarries information about
two separate modes, the charge, and the ‘complex-
ity’ of the operator within the charge subspace. Fol-
lowing this, the operator is evolved using Eq. (2),
where it is observed that the mode ξspreads bal-
listically while ρis governed by diffusion.
The evolved operator is then measured via the use
of a probing operator V. The choice of whether Vis
chosen to be a conserved (has overlap with charge)
or non-conserved operator determines which modes
the OTOC detects:
F(t)|V=χr0=ξ(r0, t)
F(t)|V=nr0=ξ2(r0, t) + ρ2(r0, t)(3)
The operator χr0refers to the local creation operator
on site r0of the chain, while nr0is the local number oper-
ator, which has overlap with the charge. This formalism
distinguishes the time-ordered Green’s function from the
OTOC, as the former can only detect the charged mode
and not the fixed-charge operator transitions (Fig. 1).
It is important to mention that Wis a non-local oper-
ator as it fixes the initial charge profile (Sec. III). This
is necessary if one wishes to obtain a precise descrip-
tion of how information dynamics are related to charge
dynamics, as local operators do not have a well-defined
charge. This is perhaps also related to other works which
have observed that locality has non-trivial implications
in the presence of symmetries [72]. The equations in
Eq. (2) correctly reproduce all known features of charged
chaotic models, such as the charge-dependent Lyapunov
exponent/butterfly velocity [47,48], and the late-time
diffusive tail of the OTOC [44,45,73]. Moreover, since
they are valid for arbitrary initial charge/operator pro-
file, we simulate the coupled equations in inhomogeneous
backgrounds to obtain new phenomena.
The formalism developed in this work encodes the mi-
croscopic operator dynamics in terms of transitions be-
tween states in a particular irrep of the SU(4) algebra.
Since the OTOC is computed on four time-contours, this
allows us to track the dynamics of the four corresponding
conserved charges in terms of the weights of the SU(4)
algebra, and explains the emergence of the coupled equa-
tions describing the OTOC in terms of a single charged
mode (ρ) and a non-conserved ballistic mode (ξ). This
derivation reveals new features of operator dynamics in
4
the presence of conservation laws. Namely, in contrast
with models that do not have continuous symmetries, the
operator dynamics in the model with a U(1) symmetry
allow for transitions that introduce quantum coherence at
the operator level. Therefore, while usual noisy/random
models are well described by a classical stochastic pro-
cess, for U(1) symmetric models the time-evolution of the
operator can only be modeled by a probability in a fixed
subspace, and in the infinite-Nlimit, this gives rise to a
Fokker-Planck equation which only describes a conserved
quantity in the ‘incoherent’ sector of the operator-states.
III. CHARGED OPERATOR BASIS
In this work we will be concerned with computing the
OTOC, which can be viewed through the lens of operator
spreading. Due to the charge conservation, we work with
a specific choice of operator basis. We assume the degrees
of freedom are represented by complex fermions, which
satisfy the anti-commutation relations:
{χ
i, χj}=δij ;{χi, χj}= 0
The relevant operators we consider are left and right
eigen-operators with respect to the U(1) symmetry:
X
i
niO=qaO;OX
i
ni=qbO(4)
where (n=χχ) is the number operator. It can be
easily checked that for a model with U(1) conservation,
the charges (qa, qb) are a constant of motion during the
dynamics of the operator string. For a single fermionic
operator, we choose the relevant four dimensional eigen-
operator basis and the charges take the following respec-
tive values:
χ: (1,0) χ: (0,1) n: (1,1) ¯n: (0,0) (5)
where ¯n=In. Following this insight, we will work with
operator strings Swhen the system contains Nfermions,
where each element of the string is picked from the U(1)
operator basis defined above:
S= 2N/2s1s2···sN, si∈ {χ
i, χi, ni,¯ni}.(6)
This ensures that the entire operator string is also an
eigen-operator of the global U(1) symmetry. The factor
of 2N/2in the string is picked to ensure the following
orthogonal and completeness relations
1
trItr(SS0) = δ(S0,S),1
trIX
SS
mnSpq =δmq δnp.(7)
The procedure to compute the OTOC, on the other
hand, begins by rearranging the correlator to write it in
FIG. 2. Figure adapted from [50]. In the first part of (a),
the process of resolving the identity in the output state to
ensure a proper overlap with the input state is depicted. For
the Brownian model, disorder-averaging after this procedure
turns the OTOC into an amplitude, computed after the states
are evolved in imaginary time using an effective Hamiltonian
H. (b) and (c) depict the null-eigenstates on four-time con-
tours, which are a consequence of the dynamics being Unitary.
While (b) is simply the double-copy of the identity, (c) rep-
resents the fully-scrambled steady state PS|S⊗ Si if the
dynamics are also chaotic.
the operator state language:
F(W(t), V ) = 1
trItr(W(t)VW(t)V) = trIhout|U|ini
|ini=1
trIXW
mnWpq |mnpqi
|outi=1
trIXV
mqVpn |mnpqi
U=UUUU
(8)
Hence the computation of the OTOC involves four copies
of the Unitary and in a charge-conserving model, four
corresponding conserved charges as well. These con-
served charges, which we label as (qa, qb, qc, qd), are the
left and right charges of the double-copy of the operator-
state involved in the OTOC. Therefore, an exact dynam-
ical description of the OTOC would depend on the evo-
lution of these four independent charges as well.
To complete the formalism and ensure a proper over-
lap between the input and output state, we insert the
摘要:

Chargetransport,informationscramblingandquantumoperator-coherenceinamany-bodysystemwithU(1)symmetryLakshyaAgarwal,1SubhayanSahu,2,3andShenglongXu11DepartmentofPhysics&Astronomy,TexasA&MUniversity,CollegeStation,Texas77843,USA2CondensedMatterTheoryCenterandDepartmentofPhysics,UniversityofMaryland,Col...

展开>> 收起<<
Charge transport information scrambling and quantum operator-coherence in a many-body system with U1 symmetry Lakshya Agarwal1Subhayan Sahu2 3and Shenglong Xu1.pdf

共18页,预览4页

还剩页未读, 继续阅读

声明:本站为文档C2C交易模式,即用户上传的文档直接被用户下载,本站只是中间服务平台,本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私,请立即通知玖贝云文库,我们立即给予删除!

相关推荐

分类:图书资源 价格:10玖币 属性:18 页 大小:3.64MB 格式:PDF 时间:2025-04-30

开通VIP享超值会员特权

  • 多端同步记录
  • 高速下载文档
  • 免费文档工具
  • 分享文档赚钱
  • 每日登录抽奖
  • 优质衍生服务
/ 18
客服
关注