Reports on Progress in Physics

REVIEW Related content

- Light-induced gauge fields for ultracold
Time crystals: a review atoms
N Goldman, G Juzelinas, P Öhberg et al.

To cite this article: Krzysztof Sacha and Jakub Zakrzewski 2018 Rep. Prog. Phys. 81 016401 - Non-standard Hubbard models in optical
lattices: a review
Omjyoti Dutta, Mariusz Gajda, Philipp
Hauke et al.

- Keldysh field theory for driven open

View the article online for updates and enhancements. quantum systems
L M Sieberer, M Buchhold and S Diehl

Recent citations
- Time Crystal Platform: From Quasicrystal
Structures in Time to Systems with Exotic
Interactions
Krzysztof Giergiel et al

- Driving-induced metamorphosis of
transport in arrays of coupled resonators
Huanan Li et al

- Towards Quantum Simulation with Circular

Rydberg Atoms
T. L. Nguyen et al

This content was downloaded from IP address 128.111.121.54 on 05/04/2018 at 12:46

 Reports on Progress in Physics

Rep. Prog. Phys. 81 (2018) 016401 (25pp) https://doi.org/10.1088/1361-6633/aa8b38

Review

Time crystals: a review

Krzysztof Sacha and Jakub Zakrzewski1
Instytut Fizyki imienia Mariana Smoluchowskiego and Mark Kac Complex Systems Research Center,
Uniwersytet Jagielloński, ulica Profesora Stanisława Łojasiewicza, 11 PL-30-348 Kraków, Poland

E-mail: [email protected]

Received 24 April 2017, revised 9 August 2017

Accepted for publication 8 September 2017
Published 23 November 2017

Corresponding Editor Professor Maciej Lewenstein

Abstract
Time crystals are time-periodic self-organized structures postulated by Frank Wilczek in 2012.
While the original concept was strongly criticized, it stimulated at the same time an intensive
research leading to propositions and experimental verifications of discrete (or Floquet) time
crystals—the structures that appear in the time domain due to spontaneous breaking of
discrete time translation symmetry. The struggle to observe discrete time crystals is reviewed
here together with propositions that generalize this concept introducing condensed matter
like physics in the time domain. We shall also revisit the original Wilczek’s idea and review
strategies aimed at spontaneous breaking of continuous time translation symmetry.

Keywords: time crystals, many-body interacting systems, spontaneous breaking of time

symmetry, many-body localization, periodic driving

(Some figures may appear in colour only in the online journal)

Contents
4. Condensed matter physics in time crystals.....................15
1. Introduction.......................................................................2 4.1. Space periodic systems versus periodically
2. Time crystals: original idea and perspectives...................2 driven systems........................................................15
2.1. Origin of space crystals: spontaneous breaking of 4.2. Anderson localization in the time domain.............18
space translation symmetry......................................2 4.2.1. Anderson localization in time crystals........18
2.2. Origin of time crystals: spontaneous breaking 4.2.2. Anderson localization in time without
of continuous time translation symmetry.................3 crystalline structures...................................19
2.3. Excited states as a resource?.............................6 4.3. Mott insulator in the time domain......................... 20
3. Spontaneous breaking of discrete time translation 4.4. Many-body localization with temporal
symmetry: idea and experiments......................................7 disorder................................................................. 21
3.1. Discrete time crystal for atoms bouncing on an 4.5. Time crystals with properties of multi-
oscillating mirror......................................................7 dimensional systems............................................. 21
3.2. Discrete time crystals in spin systems...................10 5. Conclusions.................................................................... 21
3.3. Experiments...........................................................13 Acknowledgments.......................................................... 22
3.4. Effect of disorder...................................................14 References...................................................................... 22

1
Author to whom any correspondence has been addressed.

1361-6633/18/016401+25$33.00 1 © 2017 IOP Publishing Ltd Printed in the UK

Rep. Prog. Phys. 81 (2018) 016401 Review

1. Introduction energy state. It is already known that the original Wilczek idea
cannot be realized (Bruno 2013c, Watanabe and Oshikawa
Crystals are everywhere ranging from jewelery to salt in the 2015). However, it became an inspiration to other physicists
kitchen. They are built of atoms in a regular arrangement and a novel research field has been opened.
in space and they have distinct mechanical properties as Exploration of this time territory started immediately and
well as heat or electric conductance properties. Crystals are soon a new version of time crystals was proposed, that is, a
formed due to mutual interactions between atoms that self- discrete or Floquet time crystals (Sacha 2015b). That time
organize and build regular structures in space. This kind of crystals are related to quantum self-organization of motion of
self-organization is a quantum mechanical phenomenon and a many-body system that is periodically driven by an exter-
is related to spontaneous space translation symmetry break- nal force. Soon other propositions of Floquet time crystals
ing (Strocchi 2005). It is often neglected that a probability emerged in driven spin systems (Else et al 2016, Khemani
density for measurement of a position of a single particle in a et al 2016b). While one would expect that in a stationary state
system of mutually interacting particles cannot reveal a crys- a system should follow an external driving it turns out that due
talline structure unless continuous space translation symmetry to the mutual interaction between particles, the system prefers
is broken. Quantum mechanics tells us that the center of mass to move on its own and spontaneously switches to a periodic
of interacting atoms behaves like a free massive particle and motion with a different period than a period of driving. This
in the ground state it must be totally delocalized. The trans- kind of time crystallization has been recently realized in labo-
lationally invariant Hamiltonian leads to quantum eigenstates ratories (Choi et al 2017, Zhang et al 2017).
that are also transitionally invariant. To observe a crystalline In condensed matter physics properties of space crystals
structure in the single particle probability density, this trans- are often analyzed with the help of space periodic potentials—
lational symmetry must be broken—in other words we do not it is assumed that a space crystal is already formed, i.e. a crys-
know where a crystal is unless we perform a measurement talline structure has emerged. Systems that are periodically
or a system is perturbed by an infinitesimally weak external driven in time by an external force not only can reveal spon-
perturbation. Let us repeat, this process occurs spontaneously taneous breaking of a discrete time translation symmetry but
and in the real world one deals with the symmetry broken state can also show solid state phenomena in the time domain. This
already. After the symmetry is broken the quantum effects do research has been already initiated and shows that Anderson
not give up. However, they are extremely slow and even if a localization or Mott insulator phase in the time domain can be
space crystal were isolated it would take thousand of years observed (Sacha 2015a).
to see quantum blurring of the crystalline structure. It should In this review article we describe the present status of the
be noted that if one was able to measure relative distances time crystal research. We start with an introduction to the
between particles (not particles positions themselves) in a phenom­enon of spontaneous space translation symmetry
solid state system prepared in the ground state, signatures of breaking in solid state systems. Then, we switch to the con-
regular arrangement of atoms would be observed and such a sideration of possibility of spontaneous time translation sym-
measurement would not break the continuous space transla- metry breaking and the idea of time crystal. We will report
tional symmetry. on the on-going research in time-independent systems and
Spontaneous symmetry breaking is a general property of in systems that are periodically driven and are able to reveal
Nature that surrounds us and it is responsible for a wide class discrete time crystal phenomena. Next, the research on a pos-
of phenomena like magnetization of ferromagnetic materials sible realization of condensed matter phenomena in the time
(rotational symmetry breaking) or Higgs mechanism (break- domain will be described.
ing of gauge symmetries) (Strocchi 2005). It is related to a
situation where equations describing a system possess a sym-
metry but a system chooses spontaneously a solution that 2. Time crystals: original idea and perspectives
breaks this symmetry. The effect can be identified with a
vulnerability of exact symmetric eigenstates to any infinitesi- We introduce here a concept of time crystals which is related to
mally weak perturbation. spontaneous breaking of time translation symmetry in analogy
In 2012 Frank Wilczek proposed an idea of time crystals to spontaneous space translation symmetry breaking in the for-
(Wilczek 2012). He posed a question whether there exist mation of space crystals. We begin with the description of space
systems which break spontaneously time translation symme- crystals that allows us to explain phenomena that can also be
try. In other words if it is possible that a many-body system observed in the time domain when a many-body system switches
self-organizes in time and starts spontaneously to undergo a spontaneously to periodic motion realizing the time crystal.
periodic motion. This kind of self-organization is a quantum
effect and should not be mixed with classical self-organiza-
2.1. Origin of space crystals: spontaneous breaking of space
tion processes that occur in many systems in nature ranging
translation symmetry
from flashing Asian fireflies to synchronized clapping of an
audience that expresses appreciation for an impressive perfor- Formation of space crystals relies on periodic self-organization
mance (Glass and Mackey 1988, Neda et al 2000). Wilczek of atoms due to their mutual interactions. Under certain condi-
considered a time-independent system and suggested that it tions atoms arrange themselves in a periodic lattice that man-
can spontaneously turn to periodic motion even in the lowest ifests itself in a periodic behavior in space of a probability

2
Rep. Prog. Phys. 81 (2018) 016401 Review

density for a measurement of a single particle (an electron or an The ground state of a system corresponds to the total momen-
ion). Strictly speaking such a state cannot be the ground state tum P = 0 and is entirely delocalized in the configuration
of a many-body atomic system because it breaks the trans- space. Measurement of particle positions leads to the local-
lation symmetry (assuming a typical nondegenerate ground ization of the system and to an emergence of crystalline
state). To see this, let us consider a solid state Hamiltonian, structures. It should be stressed that if one performed the mea-
N N surement not of individual positions of particles but rather the
 p2i 1
H = + Uij (ri − rj ), relative distances between them, such a measurement would
(1)
2mi 2 also reveal crystalline properties of the system without break-
i=1 i=j
ing the continuous space translation symmetry.
that describes N interacting particles in a finite volume V with When the continuous space translation symmetry is bro-
periodic boundary conditions. If we shift all positions ri by ken to a discrete symmetry one can ask about the lifetime of
the same vector R , the Hamiltonian does not change because the symmetry broken state because it is no longer the system
it depends on relative distances between particles only. It eigenstate. In the thermodynamic limit defined as N → ∞,
means that the system possesses continuous (i.e. vector R can the volume V → ∞ but the particle density N/V = const., the
be arbitrary) space translation symmetry; the corresponding energy of the symmetry broken state is infinitely close to the
(unitary) space translation operator TR commutes with H and ground state energy and its lifetime exceeds easily thou-
eigenstates ψn (r1 , . . . , rN ) of H are also eigenstates of TR , sands of years. It can be estimated assuming that the center
TR ψn (r1 , . . . , rN ) = eiϕ ψn (r1 , . . . , rN ).
(2) of mass is described by a wave-packet localized on a length
scale σ ≈ 10−11m, then, the corresponding kinetic energy is
Taking into account equation (2) and the fact that h2
Ek = 2mNσ 2 . If such a system is isolated, the quantum spread-

TR ψn (r1 , . . . , rN ) = ψn (r1 + R, . . . , rN + R),

(3) ing of the wave-packet leads to a delocalization of a space
it is easy to show that the probability density for the measure- crystal. However, in order see that the crystalline structure is
ment of a single particle must be uniform in space if a system blurred, the center of mass must be delocalized on a length scale
is prepared in the ground state (or any other nondegenerate of the order of the lattice constant of a crystal, i.e. a ≈ 10−10
eigenstate), m, that takes time of the order of t = a/(h/mNσ) ≈ 5 · 104
 years for mN = 1 kg. Fortunately, jewelery of our grandmoth-
ρ(r1 + R) = d(r2 + R) . . . d(rN + R) ers is not isolated from the outside world and ‘Diamonds are
forever’ (due to the measurement process).
2 Kinetic energy Ek is extremely close to the ground state
× |ψn (r1 + R, . . . , rN + R)|
 energy and about 10−26 smaller than energy of an optical
2
= dr2 . . . drN |ψn (r1 , . . . , rN )| = ρ(r1 ), (4) photon in our example. It implies that breaking of the sym-
 metry practically costs no energy and can be induced by an
infinitesimally small perturbation and, therefore, it occurs
and no discrete structure is visible. However, crystalline prop-
spontaneously.
erties can be observed in the two-point correlation function,
 An alternative method to predict the spontaneous break-
2 ing of a space translation symmetry is to apply a symmetry
ρ2 (r1 , r2 ) = dr3 . . . drN |ψn (r1 , r2 , r3 . . . , rN )| ,
(5)
breaking perturbation, calculate the ground state of a system
because the symmetry does not forbid ρ2 (r1 , r2 ) to be non- for finite N, take the thermodynamic limit and finally turn off
uniform in space. If, for a fixed r1, the two-point correlation the perturbation which leads to the symmetry broken state
function reveals periodic behavior as a function of r2, then (Anderson 1997, Kaplan et al 1989, Koma and Tasaki 1993).
spontaneous breaking of the continuous space translation sym-
metry to a discrete translation symmetry can be predicted. The 2.2. Origin of time crystals: spontaneous breaking
correlation function ρ2 (r1 , r2 ) tells us what is the probability of continuous time translation symmetry
density for a measurement of the next particle provided the first
particle has been detected at r1. Thus, it is enough to measure a If a time-independent many-body system is prepared in an
single particle in order to see if a crystalline structure emerges. eigenstate ψn corresponding to energy eigenvalue En, the
The measurement can be intentional, i.e. performed by appara- probability density for detection of particles at a fixed posi-
tus in a laboratory, or simply due to the coupling of a system to tion in the configuration space obviously does not change in
its environment and the resulting possible particle losses. time. It is a direct consequence of continuous time translation
The breaking of the space translation symmetry is related symmetry of a system. Indeed, time-independent Hamiltonian
to the localization of the centre of mass of a system (van Wezel H commutes with the time translation operator which is sim-
and van den Brink 2007). In the centre of mass coordinate ply the evolution operator Tt = e−iHt (we assume  = 1) and
frame it is apparent that the centre of mass degree of freedom eigenstates of H are also eigenstates of Tt , thus,
decouples from the relative positions’ degrees of freedom, i.e. |ψn (t)|2 = |Tt ψn (0)|2 = |e−iEn t ψn (0)|2 = |ψn (0)|2 .
(7)
the Hamiltonian (1) can be written as
In 2012 Frank Wilczek proposed (Wilczek 2012) that
P2 time translation symmetry can be spontaneously broken in
(6)
H= + relative degrees of freedom.
2mN an analogue way to space translation symmetry breaking in
3
Rep. Prog. Phys. 81 (2018) 016401 Review

the formation of space crystals. He coined the term time crys- placed at a certain point in the configuration space reacts peri-
tal for that phenomenon. If it exists then a time-independent odically because particles are returning to this point.
many-body system, prepared in the ground state, can switch to A potential example fulfilling our requirements seems
a periodic motion in time under an infinitesimally weak per- to be a superconducting device where an external magnetic
turbation. Experimentally it could be observed as a periodic field induces current of Cooper pairs. However, the flow of
behaviour in time of the probability density for a measure- Cooper pairs is uniform and if, at a certain position in space,
ment of a system at a fixed point of the configuration space. It we detect particles at a certain moment of time, the next detec-
means that switching from space to time crystals we have to tion event will not be correlated temporally with the previ-
exchange the role of space and time. In the space crystal case ous one and no periodic crystalline structure in time will be
we expect periodic behaviour in space at a fixed instant of observed (Wilczek 2012, Yamamoto 2015). The original idea
time (i.e. at the moment when we perform a measurement of of Wilczek was more involved. It is known that interacting
a system) while in the time crystal case we fix the position in particles can form spontaneously inhomogeneous structures
space and ask whether a detector clicks periodically in time. in space. On the other hand a charged particle on a ring (one-
More formal definitions of time crystals may be formulated, dimensional (1D) problem with periodic boundary condi-
see e.g. Else et al (2016) and Khemani et al (2016a)—we shall tions) subjected to magnetic flux can reveal non-vanishing
stick to this intuitive one. probability current along the ring in the ground state if the
The idea of time crystals was proposed in two variants. flux is properly chosen. Combining these two observations,
Shapere and Wilczek (2012b) showed that a classical system it should be possible to observe a spontaneous process where
can reveal periodic motion in the lowest energy state, see also a many-body system prepared in the ground state switches to
(Ghosh 2014), while Wilczek himself (Wilczek 2012) pre- periodic evolution where an inhomogeneous particles density
sented an idea of quantum time crystal. In the classical case moves around a ring. Wilczek concentrated on bosons inter-
if we ask whether a time-independent system can move and acting via attractive contact potential on an Aharonov-Bohm
at the same time possess the lowest energy, the answer seems ring and we will elaborate on this system in a moment. Very
to be obviously No! Indeed, in order to find the lowest energy soon another proposition (Li et al 2012b) was suggested that
for a classical particle we have to find an extremal value of a ions on a ring may spontaneously form a space crystal when
Hamiltonian, kinetic energy of ions is much smaller than Coulomb potential
energy between them. Such a Wigner crystal, in the presence
∂H ∂H
= 0, = 0, (8) of a magnetic flux, reveals periodic motion along the ring even
∂p ∂x
 if the system is initially prepared in the ground state. Both
but that means no motion of a particle is possible because the proposals were immediately criticized by Bruno (2013a and
first condition in (8) implies that the Hamilton equation, 2013b) who pointed out that such scenario is impossible, for
replies see Li et al (2012a) and Wilczek (2013b). Soon after
dx ∂H
ẋ = = = 0. (9) Patrick Bruno showed that under quite general conditions
dt ∂p
 spontaneous breaking of continuous time translation symme-
However, if we assume the energy of a particle of the form try is not possible in any time-independent system prepared in
the ground state (Bruno 2013c). Before we present the argu-
ẋ4 ẋ2
E= − , (10) ments of Bruno and other researchers we first describe details
 4 2 of Wilczek idea (Wilczek 2012).
one sees that the lowest energy corresponds to particle motion Let us consider N bosons on a ring of unit length with
with velocity ẋ = ±1. This apparent contradiction with attractive contact Dirac δ-interactions and in the presence of a
the conclusion based on the Hamilton equation (9) can be magnetic flux α (a ring problem in the presence of a magnetic
resolved when we realize that the energy (10) cannot be con- flux is dubbed Aharonov–Bohm ring). The particle mass and
verted to the Hamiltonian smoothly. That is, the Hamiltonian  are assumed to be equal to unity and the parameter g0 that
is a multi-valued function of the momentum with cusps determines the strength of attractive interactions is negative.
corre­sponding precisely to energy minima at ẋ = ±1 where The Hamiltonian of the system,
the Hamilton equations are not defined (Shapere and Wilczek
N

2012b). ( pi − α)2 g0 
H=
(11) + δ(xi − xj ),
Encouraged by the classical analysis we can now look for 2 2
i=1 i=j
a quantum time-independent many-body system that in the
ground state can spontaneously switch to periodic motion possesses continuous time and space translation symmetries
and reveal crystalline properties in the time domain. Shapere and consequently probability density corresponding to any
and Wilczek (2012a) proposed how to quantize the classical eigenstate is invariant under any translation in time and any
single-particle system described by energy (10). However, in translation of all particles along the ring.
the following we will concentrate on a more general problem Let us assume for a moment that the magnetic flux α = 0
of many-body systems with more conventional kinetic energy and let us apply the mean field approximation. In the mean
terms that could reveal periodic motion in measurements field approach all bosons are supposed to occupy the same
repeated many times on the same realization of a system. single particle state φ, i.e. form a Bose–Einstein condensate.
That is, we would like to consider a situation where a detector Many-body eigenstates are then of the form of a product

4
Rep. Prog. Phys. 81 (2018) 016401 Review

state φ(x1 )φ(x2 ) . . . φ(xN ). In order to obtain the ground state equation (15) to vanish and once the bright soliton is formed in
within the mean field approximation one has to find the mini- the process of spontaneous breaking of space translation sym-
mal value of the energy of the system within the Hilbert sub- metry it will move too. We see it is not the case in the crucial
space spanned by product states that reduces to the solution limit, i.e. where the total number of particles increases. The
of the Gross–Pitaevskii equation (Pethick and Smith 2002), idea of Li and co-workers was similar (Li et al 2012b). Instead
  of particles interacting via an attractive contact potential, they
1 2 2 envision ions on an Aharanov–Bohm ring. A spontaneous
(12) − ∂x + g0 (N − 1)|φ| φ = µφ.
2 breaking of space translation symmetry results in a formation
If the attractive particle interactions are sufficiently strong, i.e. of a space crystal which has been supposed to move in the
g0 (N − 1) < −π 2, it is energetically favorable to group par- presence of a magnetic flux even if an experiment started with
ticles together and the mean field solution φ breaks the space the ground state. Again, the same line of arguments leads to
translation symmetry and becomes inhomogeneous in space the conclusion that it is not possible.
(Carr et al 2000). The φ function is given by the Jacobi ellip- The impossibility of realization of the time crystal idea
tic function but when |g0 |N  1, it is well approximated by a was also considered by Nozières (2013) who investigated a
bright soliton solution superfluid ring in a magnetic field and presented arguments
  that a charge density wave cannot reveal rotation induced by
g0 (N − 1) diamagnetic currents. However, general analysis of the impos-
(13) φ(x) ≈ cosh−1 (x − xCM ) ,
2 sibility of spontaneous time translation symmetry breaking in
where xCM is a parameter that describes the centre of mass the ground state was performed by Bruno (2013c) and later by
position (Carr et al 2000, Pethick and Smith 2002). Thus, Watanabe and Oshikawa (2015). Bruno considered a general
the mean field approach predicts that bosons form a Bose– many-body system on an Aharonov-Bohm ring that is sub-
Einstein condensate where all particles occupy a localized jected to a perturbation that rotates periodically along the ring.
wavefunction (13). When we return to the many-body descrip- He showed that for the ground state that breaks rotational sym-
tion we obtain that the many-body ground state has the space metry, the moment of inertia of the system is always positive
translation symmetry but this state is strongly vulnerable to and imposing rotation increases energy. Bruno also consid-
any perturbation. In fact to break the space translation invari- ered the thermal equilibrium state in the rotating frame, where
ance it is enough to measure a position of a single particle the Hamiltonian becomes time-independent, and reached a
(Delande et al 2013). similar conclusion. However, for a periodically driven system
Wilczek expected that in the presence of a properly cho- analysis of thermal equilibrium in a rotating frame should
sen magnetic flux α, one would not only observe spontane- be performed with caution because it actually implies that a
ous localization of density of particles in the configuration system is assumed to be in contact with a reservoir of rotat-
space but such a density will also move periodically along ing particles. Moreover, it was shown that spontaneous emis-
the ring and the motion sustains forever in the limit when sion of photons may correspond to a jump of an electron in
N → ∞, g0 → 0 but g0 (N − 1) = const., i.e. in the limit an atom upwards in energy if the process is described in the
where the mean field prediction remains unchanged, see (13). rotating frame (Delande and Zakrzewski 1998). This obstacle
This is actually false as immediately pointed out by Bruno was overcome by Watanabe and Oshikawa (2015) who did
(2013a). Probably, the simplest way to demonstrate it, is to not assume the presence of a symmetry broken perturbation
switch to the centre of mass coordinate frame (Syrwid et al but focused on analysis of a correlation function. They proved
2017). Then, the Hamiltonian (11) reads that the two-point correlation function does not reveal any
time dependence when a many-body system is prepared in the
(P − Nα)2 ground state or thermal equilibrium state in the limit when
(14)
H= + relative degrees of freedom,
2N volume V of a system goes to infinity.
where P is the centre of mass momentum, i.e. the total momen- In the case of space crystals, spontaneous space transla-
tum of the system, which is a conserved quantity. The centre of tion symmetry breaking is demonstrated in the limit when a
mass and the relative positions’ degrees of freedom decouple number of particles N and V tend to infinity but the density
and eigenstates of the system are determined by an indepen- of particles is constant. Then, any perturbation is sufficient to
dent choice of the centre of mass momentum Pj = 2πj (where break the symmetry and a symmetry broken state lives for-
j is integer) and the relative degrees of freedom quant­um num- ever. In the case of time crystals, it is not necessary to take
bers. The ground state corresponds to V → ∞ because we do not have to assume periodic or any
other behaviour of a system in space. We may set V = const.,
∂H j increase N but keep the product of the coupling constant g0
= 2π − α ≈ 0.
(15)
∂Pj N and N fixed, see equation (11).
In the limit when N → ∞, equation (15) can be fulfilled Evolution of the phase of a Bose–Einstein condensate
exactly. This is very bad news because it means there is no described within ground canonical ensemble can be consid-
probability current related to the centre of mass degree of free- ered in the context of time crystals. In the ground canonical
dom if the system is prepared in the ground state. Wilczek formalism, the order parameter of a Bose–Einstein condensate
idea relied on the assumption that if the flux α is chosen reveals periodic oscillations ψ̂(r, t) = ψ0 e−iµt, where ψ̂ is a
properly, quantized values of particle momenta do not allow bosonic field operator and μ is a chemical potential (Castin

5
Rep. Prog. Phys. 81 (2018) 016401 Review

and Dum 1998, Pethick and Smith 2002). Such a periodic ∂H

time evolution can be measured provided the condensate is = 2π − α,
(16)
∂PN
coupled to another condensate. If a system is strictly isolated,
i.e. when a number of particles is conserved, there is no refer- does not vanish provided the flux α = 2π . Thus, for α = 2π ,
ence frame and no time dependence can be detected. Volovik let us choose among all eigenstates with the total momentum
(2013) analyzed a class of such systems (see also Nicolis and equal PN the eigenstate |ψ0  with the lowest energy. Although
Piazza (2012), Wilczek (2013a), Castillo et al (2014) and such an eigenstate is not the ground state, it realizes Wilczek’s
Thies (2014)) and introduced two relaxation times: energy idea. That is, the probability density related to this eigenstate
relaxation time τE and relaxation time τN related to a total is invariant under time translation transformation but if attrac-
number of particles in our example. If τN  τE , a system with tive particle interactions are sufficiently strong, in the limit
an average number of particles N relatively quickly relaxes to when N → ∞, any perturbation can make the density inho-
a minimal energy state at fixed N and then slowly relaxes to the mogeneous. The inhomogeneous density should rotate around
equilibrium state where no oscillations are present. Although the ring with the period T = (2π − α)−1 that is determined by
in the intermediate time τN  t  τE one can observe break- equation (16) (remember that the ring has a unit length).
ing of time translation symmetry, a system is not strictly in the Numerical simulations (Syrwid et al 2017) have confirmed
equilibrium state. While τN can tend to infinity, the strict limit that the density-density correlation function,
τN = ∞ cannot be taken if one wants to observe the oscilla- ρ2 (x, t) ∝ ψ0 |ψ̂ † (x, t)ψ̂(x, t) ψ̂ † (x1 , 0)ψ̂(x1 , 0)|ψ0 ,
(17)
tion because then there is no reference frame to measure them.
Similar class of systems is considered by Else et al (2017), is inhomogeneous as a function of x and reveals periodic
however, the authors assume isolated systems. U(1) symme- rotation around the ring with the lifetime tc that increases lin-
tries that are spontaneously broken are not exact symmetries early with N if g0 → 0 but g0 (N − 1) = constant , i.e. in the
but they are related to effective Hamiltonians. Spontaneous limit when the mean field prediction remains unchanged, see
breaking of U(1) symmetry results in an order parameter figure 1. The density-density correlation function (17) corre-
that, according to an effective Hamiltonian, oscillates forever. sponds actually to the probability density for the measurement
However, at infinitely long times a system approaches a ther- of a particle at space point x and at time t provided that at t = 0
mal state determined by a full Hamiltonian where no oscilla- another particle was already detected at x1. It illustrates the
tions are observed. nature of a spontaneous process: in order to learn whether the
The results presented indicate that in a many-body system symmetry is broken one has to perform measurement and a
spontaneous breaking of time translation symmetry cannot be minimal possible information, i.e. information about the posi-
observed if a system is prepared strictly in the equilibrium tion of a single particle, is sufficient to break the symmetry.
state. Wilczek idea in its original version turned out to be Measurement of the position of a single particle
impossible for realization but it became an inspiration to other results in a localization of the centre of mass of the sys-
researchers. Particular progress has been achieved for systems tem on a length scale of the order of the bright soliton size
that break discrete time translation symmetry—the problem σCM (0) ≈ [g0 (N − 1)]−1 = const., see equation (13).
described in section 3. Before considering it let us consider The localized centre of mass probability density spreads,
the possibility to utilize excited states of many-body systems. σCM (t) ∝ t/N , in the course of time evolution determined by a
free particle like Hamiltonian (14). Thus, time moment tc when
σCM (tc ) becomes comparable to the length of the ring scales
2.3. Excited states as a resource? linearly with N. That agrees with the numerical results. If more
Let us analyze if the spontaneous breaking of time transla- particles are measured initially, the centre of mass probability
tion symmetry can be observed for a many-body system pre- density can be localized on a length scale σCM (0) ∝ N −1/2 . In
pared in an excited eigenstate. In other words we would like to such a case, the lifetime of periodic evolution of the symmetry
answer the question if a time independent many-body system broken state scales like N 1/2 (Syrwid et al 2017).
prepared in an excited eigenstate is able to self-organize in It is not simple to prepare in a laboratory a many-body
time and switch to periodic motion under infinitesimally weak system in a specific excited eigenstate. However, ultra-cold
perturbation. We additionally require that such a situation atomic gases constitutes a perfect playground for realization,
should not be a theoretical issue only but should be realizable control and detection not only many-body ground states but
experimentally. also collectively excited states (Pethick and Smith 2002). The
Analysis of Wilczek model, i.e. the system described by described spontaneous breaking of continuous time transla-
the Hamiltonian (11), Syrwid et al (2017) suggests that indeed tion symmetry in excited states (Syrwid et al 2017) can be
the spontaneous breaking of continues time translation sym- observed in ultra-cold atoms trapped in a toroidal potential
metry can occur for an excited eigenstate. For any chosen in the presence of an artificial gauge field (Goldman et al
value of the magnetic flux α, the ground state of the system 2014). The latter can be realized, e.g. by imposing rotation
corresponds to such a total momentum Pj that the probability of a thermal cloud during the evaporative cooling because
current associated with the centre of mass motion vanishes then the system dissipates to the lowest energy state in the
if N → ∞, see equation (15). However, if the system is pre- rotating frame and the Coriolis force mimics a magnetic field
pared in an eigenstate with total momentum PN = 2πN , the (Pethick and Smith 2002). Proper choice of an artificial gauge
probability current related to the center of mass motion, potential allows one to obtain the flux α = 2π for which the

6
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 1. Time evolution of the density-density correlation function (17), i.e. the probability density for the measurement of the second
particle provided the first particle has been measured at t = 0 at position x = 0.5. The measurement breaks the continuous translation
symmetry making the probability density nonuniform. During a subsequent evolution, as expressed in the different panels, the density
moves along a ring with the period T = 1/2π but also spreads with the characteristic time tc increasing with the particle number N. All
results are obtained for g0 (N − 1) = −15. Reproduced from Syrwid et al (2017) with permission from K. Sacha.

ground state is the desire state |ψ0  with the total momentum where |un (t) must fulfill periodic boundary condition in time.
PN = 2πN . Then, turning off the gauge potential leaves the Eigenvalues En are real and they are called quasi-energies of a
system in the state |ψ0  which is no longer the ground state system. These consequences of the Floquet theorem (Shirley
of the Hamiltonian with α = 0 and breaking of time trans- 1965) are in full analogy to the Bloch theorem known in con-
lation symmetry under any weak perturbation is expected. densed matter physics where space periodic Hamiltonian are
This observation will be in contrast with the same experiment common models of solid state systems. There, momentum of
but performed for g0 (N − 1) > −π 2 where no spontaneous a particle is not conserved because of the presence of an exter-
breaking of space or time translation symmetry occurs. nal potential but due to its space periodicity, quasi-momenta
To summarize this section, any time-independent many- are well defined and eigenstates of a particle are plane waves
body system has the continuous time translation symme- modulated with periodicity of a potential (Ashcroft and
try. It implies that probability density corresponding to any Mermin 1976). In the case of time periodic Hamiltonians
eigenstate does not change in time. Moreover, no spontane- we deal with analogues situation, time evolution of a single
ous breaking of time translation symmetry can be observed Floquet state e−iEn t |un (t) nearly agrees with time evolution
if a system is in the ground state or the thermal equilibrium of an eigenstate of a time-independent system but a time-
state. However, it was illustrated with the help of Wilczek’s dependent phase e−iEn t is modulated with periodicity of a
model that the continuous time translation symmetry can be Hamiltonian. Quasi-energy spectrum is not bounded from
spontaneously broken if a system is prepared in an excited below. It is actually periodic with a period 2 pi/T and it is
eigenstate. Such a phenomenon could be in principle realized sufficient to consider only a single Floquet zone in order to
in ultra-cold atoms laboratory. The idea of time crystals initi- fully describe a system. This is in analogy to a Brillouin zone
ated a novel field of research and inspired many physicists in condensed matter physics (Ashcroft and Mermin 1976).
(Chernodub 2013, Mendonca and Dodonov 2014, Robicheaux Periodically driven systems break continuous time transla-
and Niffenegger 2015, Yamamoto 2015, Faizal et al 2016). tion symmetry but they possess discrete time translation sym-
metry, i.e. a Hamiltonian H(t + T) = H(t) commutes with
time translation operator TT related to evolution of a system
3. Spontaneous breaking of discrete time
by period T and Floquet eigenstates are also eigenstates of TT ,
translation symmetry: idea and experiments
TT |un (0) = e−iEn T |un (T) = e−iEn T |un (0).
(20)
Time independent systems possess continuous time
translation symmetry. This symmetry is broken when a Thus, if we choose a point in the configuration space and ask
Hamiltonian becomes explicitly time dependent H(t). how probability density for detection of a single or many par-
Then, energy is not conserved but if a Hamiltonian is time ticles at this point changes in time, the answer is it is periodic
periodic, H(t + T) = H(t), there exists a kind of stationary with a period T if a system is prepared in a Floquet eigenstate.
states that are time periodic so-called Floquet eigenstates An interesting question arises: can a many-body periodically
|un (t + T) = |un (t) (Shirley 1965). Time evolution of any driven system prepared in a Floquet eigenstate spontaneously
quantum state can be written as a superposition of Floquet self-organize in time and start evolving with a period that is
states because they form a complete basis at any time, not equal T? The answer is yes and this phenomenon, called a
 discrete or Floquet time crystal, has been recently realized in
(18) |ψ(t) = cn e−iEn t |un (t). laboratories (Zhang et al 2017, Choi et al 2017).
n

Substituting e−iEn t |un (t) in the time-dependent Schrödinger 3.1. Discrete time crystal for atoms bouncing on an
equation results in an eigenvalue problem for the so-called oscillating mirror
Floquet Hamiltonian HF, that is hermitian with respect to the
scalar product involving integration over time, The first example of the spontaneous breaking of discrete
time translation symmetry to another discrete symmetry was
HF |un (t) = (H(t) − i∂t ) |un (t) = En |un (t),
(19) given in 2015 (Sacha 2015b). As an illustration of the effect a

7
Rep. Prog. Phys. 81 (2018) 016401 Review

system of ultra-cold atoms bouncing on an oscillating mirror

in the presence of the gravitational field is considered. In order
to explain the phenomenon we have to first describe the corre­
sponding single particle problem, a classical version of which,
called a bouncer has been introduced by Pustyl’nikov as a
model for Fermi acceleration (Zaslavsky 1970, Pustyl’nikov
1978). This famous model of classical chaos has been studied
experimentally (see e.g. Pierański (1983)), its quantum ver-
sion has been often studied (Dembiński et al 1993, Buchleitner
et al 2002), moreover, it was realized experimentally for cold
atoms bouncing on a mirror formed by an evanescent wave
(Steane et al 1995).
A single particle bouncing on an oscillating mirror is
described (in 1D approximation and in the frame oscillating
with a mirror and in the so-called gravitational units) by the
following quantum Hamiltonian (Buchleitner et al 2002),
1
H = − ∂z2 + V(z) + λz cos(ωt),
(21)
2
where V(z) = z for z  0 and V(z) = ∞ for z < 0. In the
frame oscillating with a mirror, the position of a mirror is Figure 2. Time evolution of the probability density for a particle
fixed (at z = 0) but the gravitational constant depends peri- bouncing on an oscillating mirror and prepared in a Floquet
odically on time. In the absence of the mirror oscillations eigenstate that reveals evolution of two localized wave-packet
along a classical 2:1 resonant orbit. Different panels correspond to
(λ = 0) all classical trajectories of a particle are periodic different time moments as indicated in the figure. Initially (t = 0)
with a period increasing with the energy of the particle. When two localized wave-packets overlap but because they propagate in
mirror oscillations are turned on, classical motion becomes opposite directions one can see interference fringes. In the course
irregular but some of periodic orbits survive. They are stable of the time evolution one wave-packet moves towards the mirror
resonant orbits living in regular parts of the classical phase (located at z = 0) bounces off the mirror and returns. The other
wave-packet moves towards the classical turning point in the
space. There is 1:1 resonant orbit where a particle moves gravitational field and also returns. Despite the fact that each of the
periodically with a period equal to the period of the mirror wave-packets evolves with a period 2T, at t = T we end up with the
oscillations T = 2π/ω. There exist 2:1 resonant orbit where a initial situation because at this moment of time the wave-packets
particle bounces on a mirror with a period twice longer than exchange their roles.
that of the oscillations as well as higher order resonances s : 1.
Switching to the quantum description, a motion of a particle splitting J is related to a tunneling process. By superposing
is described by Floquet eigenstates. It may be surprising but u1 (z, t) and e−iωt/2 u2 (z, t) we can eliminate one of the two
classical-like motion of a particle on resonant orbits can be localized wave-packets. The remaining wave-packet evolves
also observed in the quantum world. For example, suitable along 2:1 resonant trajectory but at the same time slowly tun-
choice of parameters results in a Floquet eigenstate that is nels to a position of the other missing wave-packet—the full
represented by a localized wave-packet moving periodically tunneling process is completed after a period π/J which is
along 1:1 resonant orbit. This is an example of the so-called very long as compared to T. It is worth noting that in order
non-spreading wave-packet motion that was discovered more to observe the described resonant behaviour, the resonant
than 20 years ago (Henkel and Holthaus 1992, Bialynicki- condition does not need to be strictly fulfilled. That is, if the
Birula et al 1994, Delande and Buchleitner 1994, Buchleitner resonance condition corresponds to an unperturbed classical
and Delande 1995, Holthaus 1995, Zakrzewski et al 1995) for orbit with a sufficiently high energy, in the quantum descrip-
a review see Buchleitner et al (2002). tion the nearest in energy unperturbed quantum states will
We will focus on the 2:1 resonance case (Sacha 2015b). A form the Floquet eigenstates that we look for. In other words
single localized wave-packet moving along classical 2:1 reso- small changes of the driving frequency ω do not change the
nant orbit cannot form a Floquet eigenstate because it moves behaviour and no fine tuning is necessary because the motion
with a period twice longer than T. However, a superposition of localized wave-packets is protected by local constants of
of two such wave-packets that move with the period 2T but motion related to regular parts of the classical phase space
after T exchange their positions can form a proper Floquet (Buchleitner et al 2002).
eigenstate and indeed there exists such a state, see figure 2. When we consider N particles bouncing on an oscillating
Two localized wave-packets can actually form two mutually mirror we observe a similar behaviour if particles are bos-
orthogonal superpositions. Therefore, there exist two Floquet ons and they do not interact. That is, in the Hilbert subspace
eigenstates, u1 (z, t) and u2 (z, t), consisting of two localized spanned by Fock states |n1 , n2 , where n1 and n2 = N − n1
wave-packets moving along the 2:1 resonant trajectory. The are numbers of particles occupying single particle Floquet
eigenstates u1 (z, t) and u2 (z, t) correspond to nearly degener- states u1 and e−iωt/2 u2 , respectively, the lowest quasi-energy
ate (modulo ω/2 ) quasi-energies E2 − E1 = ω/2 + J . A tiny eigenstate corresponds to |N, 0. In order to describe particles

8
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 3. Left panel: schematic plot of a system of N atoms

bouncing on an oscillating mirror and prepared in a many-body Figure 4. Left panel: schematic plot of a system of N atoms bouncing
Floquet state (24). Each of two atomic clouds moves with a period on an oscillating mirror as in figure 3 but after the measurement of
2T but they exchange their positions after time T so that the entire the position of a single atom—atomic cloud visible in the plot moves
Floquet state is periodic with a period T. Right panel shows time with a period 2T. Right panel shows the results of the measurements
evolution of the corresponding single particle probability density of positions of 100 atoms, i.e. at t = 0 one measures positions of
(25) for N = 104, g0 N = −0.5, ω = 1.1 and λ = 0.06 . Reprinted 100 atoms, let the remaining atoms evolve and after T/2 one again
figure with permission from Sacha (2015b), Copyright (2015) by measures positions of 100 atoms and so on. The histograms presented
the American Physical Society. in the right panel indicate that time periodic evolution of the system,
after the spontaneous time translation symmetry breaking, can be
observed in a single experimental realization. The initial number of
atoms N = 104 and the other parameters as in figure 3. Reprinted
interacting via δ-contact potential one has to consider the fol- figure with permission from Sacha (2015b), Copyright (2015) by the
lowing many-body Floquet Hamiltonian American Physical Society.
 2T  ∞  
g0 matches the non-interacting result, |ψ0  ≈ |N, 0. However,
HF = dt dz ψ̂ H(t) + ψ̂ † ψ̂ − i∂t ψ̂, when
0 0 2
J † †
 1
N|U − 2U12 | > J,
(23)
≈− ĉ1 ĉ1 − ĉ2 ĉ2 + (U − 2U12 )
2 4  it is energetically favorable to collect all bosons in a single
× (ĉ†1 )2 ĉ22 + (ĉ†2 )2 ĉ21 + 2ĉ†1 ĉ1 ĉ†2 ĉ2 + const., (22) localized wave-packet and consequently the ground state of
 the Hamiltonian (22) is a Schrödinger cat-like state which
where U and U12 are equal to integrals over time and space is clear if one writes such a many-body Floquet eigenstate
of products of the probability densities of two evolving local- in another Fock basis |ñ1 , ñ2  where ñ1 and ñ2 = N − ñ1
ized wave-packets multiplied by g0. In equation (22) we are occupations of the first localized wave-packet,
restricted to the Hilbert subspace spanned by the previously √
φ1 = (u1 + e−iωt/2 u2 )/ 2 , and√the other localized wave-
described Fock states |n1 , n2  and the bosonic field operator
packet, φ2 = (u1 − e−iωt/2 u2 )/ 2 , respectively. Then, the
ψ̂(z, t) ≈ u1 ĉ1 + e−iωt/2 u2 ĉ2 where ĉ1 and ĉ2 are standard
many-body ground state reads
annihilation operators. Such a restriction is valid provided the
interaction energy is very small and couplings to the comple- |N, 0 + |0, N
(24)
|ψ0  ≈ √ .
mentary Hilbert subspace can be neglected. It will be the case 2
because the interaction energy per particle will be of the order
The corresponding single particle probability density,
of the tunneling splitting J that is extremely small as com-
pared to any other energy scale of the system. Eigenstates of ρ1 (z, t) = ψ0 |ψ̂ † (z, t)ψ̂(z, t)|ψ0 
the Hamiltonian (22) correspond to many-body Floquet states N 
that are also eigenstates of the time translation operator TT ≈ |φ1 (z, t)|2 + |φ2 (z, t)|2 ,
(25)
2
what is apparent when one realizes that there are two classes 
of the eigenstates of (22). Eigenstates from the first class are is plotted in figure 3 at different time moments. The discrete
spanned by Fock states with only even occupations of the time translation symmetry is preserved in time evo­lution
e−iωt/2 u2 (z, t) mode while eigenstates from the other class by of the many-body Floquet eigenstate |ψ0  but this state is
Fock states with the odd occupations only (Sacha 2015b). extremely vulnerable to any perturbation. After a measure-
Assuming attractive (g0 < 0 ) interactions that are very ment of a position x1 of a single particle, the symmetry is
weak, i.e. N|U − 2U12 | < J , the ground state |ψ0  of (22) gone because the quantum state of the remaining particles
9
Rep. Prog. Phys. 81 (2018) 016401 Review

immediately collapses to one of the terms in the sum (24), i.e. While mainly observed for one-dimensional spin chains,
to |N − 1, 0 or |0, N − 1 depending on a result x1 of the mea- MBL is considered by now to be a generic phenomenon in
surement. Then, time evolution shows that the original discrete disordered many-body systems. The latter are often studied
time translation symmetry has been broken, i.e. the system using strong periodic driving—being a natural current exten-
evolves with the period 2T. The resulting state |N − 1, 0 (or sion of strongly periodically driven single particle systems
|0, N − 1) is robust against any further perturbation—one intensively studied in the last millennium (consider e.g. laser-
can perform many measurements and still the period of the atom interactions or microwave resonance techniques). Until
time evo­lution remains 2T, see figure 4. Tunneling time from quite recently it has been a common understanding that peri-
the state |N − 1, 0 to |0, N − 1 or vice versa, in the limit odic driving of a many body system must supply (on average)
N → ∞, g0 → 0 but g0 N = const., increases like eαN /N with energy to the system and lead to heating (see e.g. D’Alessio
a positive constant α and very quickly becomes so long that it and Rigol (2014), Lazarides et al (2014a) and Lazarides et al
is not-measurable (Sacha 2015b). (2014b)) for arbitrary initial states—see, however, Abanin
Not only the ground and first excited states of (22) possess et al (2015), Chandran and Sondhi (2016), Kuwahara et al
a Schrödinger cat-like structure. With N → ∞ more and more (2016) and Mori et al (2016). Since any periodically driven
eigenstates come in pairs of cat states (Ziń et al 2008, Oleś system may be described by Floquet time-periodic eigenvec-
et al 2010) and experimental preparation of any initial state tors (Shirley 1965, Sambe 1973), the inevitable heating sug-
where most of atoms occupy a single localized wave-packet gests a lack of possibility to prepare an initial state in a form of
will result in time evolution with the period 2T that practi- a single, or few Floquet eigenstates. Interestingly, this believe
cally never ends. This is in contrast to the same experiment is wrong for simple driven single particle problems where
performed for N|U − 2U12 | < J where one will observe tun- specific localized examples—the so-called nonspreading
neling of atoms to another localized wave-packet after time wavepackets (Buchleitner et al 2002) were mentioned already
period of the order of 1/J . in the previous section. They have been also demonstrated
The presented example constitutes an illustration that experimentally (Maeda and Gallagher 2004). Let us note that
spontaneous self-organization in time of a periodically driven lifetime of states that are superpositions of few Floquet eigen-
many-body system is possible, i.e. spontaneous breaking of states is infinite. These wavepackets were studied, however,
discrete time translation symmetry to another discrete sym- in relatively simple small systems while the heating argument
metry may occur. In the following section we shall discuss was usually presented in the thermodynamic limit.
a similar behaviour observed for a very different model—a Numerical experiments revealed that in the presence of
system of driven interacting spins prepared in such a way that diso­rder MBL persists for driven systems provided the fre-
during a single period T all the spins are flipped (being thus quency of the driving is high enough (Abanin et al 2015,
brought to the same orientation after 2T). Then, the corre­ Lazarides et al 2015, Ponte et al 2015) so the long-time system
sponding Floquet eigenstates are macroscopic Schrödinger behavior may be analyzed by an effective Hamiltonian result-
cat-like states of two possible orientations of spins. A meas- ing from proper time averaging over the period of the drive.
urement of the direction of even a single spin leads to a col- Soon it has been realized that periodically driven systems can
lapse of any of these cat-like states to a short range correlated not only remain many-body localized but that one can distin-
time crystal state. guish different ‘phases’ by means of appropriate correlators as
discussed for spin systems by Khemani et al (2016b). Among
those, in the present context, important is the phase named
3.2. Discrete time crystals in spin systems
as π-spin glass. While the original analysis (Khemani et al
Interestingly, the original presentation of discrete time trans- 2016b) uses (Jordan-Wigner transformation based) link to ear-
lation symmetry breaking (Sacha 2015b) was not originally lier works on driven Majorana edge modes for noninteracting
noticed. It took more than a year to rediscover that phenom­ systems (Jiang et al 2011, Bastidas et al 2012, Thakurathi et al
enon in quite a different setting (Else et al 2016, 2017, 2013) let us stay within the language of Schrödinger cat-like
Khemani et al 2016b, von Keyserlingk and Sondhi 2016b) delocalized Floquet states as discussed in the previous section.
involving, importantly, the disorder induced effects. The system considered (Khemani et al 2016b) is a driven
Recent years brought an intensive study of many-body spin model with a binary periodic drive corresponding to
localization (MBL)—a phenomenon which seems to break a the unitary evolution over the period T = t1 + t2 given by
common wisdom about disordered many body systems. The U = exp(−it2 Hx )exp(−it1 Hz ) with
latter were supposed to generically thermalize if evolved from  
Hz = hi σiz + Jz σiz σi+1
z
some nonstationary initial state. Starting from a seminal work
i i
of Basko et al (2006) it seems it does not have to be the case,   
a sufficiently strong disorder may bring a many body system Hx = Ji σix σi+1
x
+ Jz σiz σi+1
z
. (26)
to a localized, non-ergodic phase (Oganesyan and Huse 2007, i

Žnidarič et al 2008). The phenomenon received a lot of atten- Aiming at discussing physics in the many-body localization
tion with literally hundreds of publications in last 10 years (for regime, the authors analyse statistics of quasi-energies (eigen-
excellent early reviews see Huse et al (2014) and Nandkishore values of U) looking in particular at the average r̄ ratio between
and Huse (2015)). the smallest and the largest adjacent energy gaps (Oganesyan

10
Rep. Prog. Phys. 81 (2018) 016401 Review

U = U2 U1 = exp(−it2 H2 ) exp(−it1 H1 ) with the period

T = t1 + t2 and

H1 = hi σix
i
 
H2 = Ji σiz σi+1
z
, (29)
i

where hi may be random while Ji are uniformly drawn from

[Jz − δJ, Jz + δJ] (to obtain many-body localization). This is
a ‘minimal’ spin system with Jz controlling the interactions.
Again for hi t1 = π/2 the action of exp(−it1 H1 ) corresponds to
Figure 5. Disorder averaged spectral function (28) for different flipping the spins along, this time, z direction, i.e. U1 = Πr iσrx .
values of the interaction term in (26). Solid lines, corresponding The second unitary is diagonal in z components. Clearly the
to π-spin glass phase, reveal a robust peak at −π/T with T being
the period of the drive—in (26), hi t1 and Ji t2 are chosen randomly quasi-energy eigenstates will be now cat-like states
and uniformly in the intervals [1.512, 1.551] and [0.393, 1.492],
respectively. Reprinted figure with permission from Khemani et al 1
∝ |{di }, p = ± = √ (|{mi }z ± |{−mi }z ) ,
(30)
|±
(2016b), Copyright (2016) by the American Physical Society. 2
and Huse 2007): rn = min[δn , δn−1 
]/ max[δn , δn−1

], with where di is the expectation value of σiz σi+1
z
and p is the Ising par-
δn = n − n−1 , and n ∈ [0, 2π) are the quasi-energies. In the

ity of the state. While the argument is presented for hi t1 = π/2
MBL phase, one expects r̄ to be close to the Poisson limit the numerical evidence (Khemani et al 2016b) reveals that the
r̄Poisson = 2 ln 2 − 1 ≈ 0.386 (Atas et al 2013). That allows phase obtained is stable and robust. It is shown that broken
one to choose the proper values for Jz and suitable distribu- symmetry phases are stable against weak local deformations
tions for random values of hi t1 and Ji t2 in (26). of Floquet drives and it is stressed that the order revealed by
Observe that Hx combined with Hz enjoys similar sym- π-spin glass (i.e. Floquet time crystal) is always spatio-tem-
metries to Ising model. The Floquet quasi-energy eigenstates poral and never purely temporal as related to general symme-
are also eigenstates of parity P = Πi σiz . Observe also that for try properties—for details see von Keyserlingk et al (2016).
hi t1 = π/2 the role of the first term in Hz will be to flip the The authors stress that absolutely stable states are generally
spins in the x direction. Neglecting for a moment the other possible by a combination of Floquet periodicity, broken sym-
terms in U, its action in two periods of the drive brings the metries and MBL—on the other hand no proof is given that
spins to the same orientation. One then expects that many- MBL is a necessary condition for observing absolutely stable
body localized (when disorder and interactions are included) phases. It is shown that signatures of stable time crystals may
Floquet eigenstates will resemble Schrödinger cat states with be obtained starting from generic short-range initial states.
frozen domain walls deep in the spin glass phase In a number of papers (Else and Nayak 2016, Potter et al
1 2016, Po et al 2016, Roy and Harper 2016, von Keyserlingk
|± = √ (| ↑↓↑ . . . ↑x ± | ↓↑↓ . . . ↓x ) .
(27) and Sondhi 2016a, Harper and Roy 2017) a further classifica-
2
tions of possible phases in periodically driven system have
They will come in pairs with different parity being almost been attempted with the stress on novel interesting topologi-
degenerate in energy (modulo π/T ) revealing the existence of cal opportunities offered by Floquet Hamiltonians. This goes
π/T phases resulting from discrete symmetry breaking (that beyond the modest scope of this review centered on time crys-
destroys vulnerable cats). Figure 5 presents the disorder aver- tals so we only mention those papers for an interested reader.
aged spectral function of the spin raising operator, σi+ on an An independent clarification of time crystals for spin sys-
arbitrary site i in the Floquet basis tems has been made by Else et al (2016). They consider a gen-
1  eral time-periodic Hamiltonian and formulate simple criteria
A(ω) = L
(28) φα (0)|σi+ |φβ (0) δ(ω − (α − β )), for the occurrence of a discrete time translation symmetry
2
αβ breaking (DTTSB) in terms of Floquet eigenstate properties.
where α are Floquet eigenvalues with corresponding vec- A simple statement is that Floquet eigenstates cannot be short-
tors |φα  with L being the system size. A peak at −π/T is range correlated. Note that, as nicely pointed out (Else et al
robust against changes of interaction strengths, disappearing 2016), this condition implies lack of MBL in Floquet systems
only above a critical interaction strength. The effective period exhibiting DTTSB if MBL is strictly understood as an integra-
doubling observed may be interpreted as a potential Floquet ble disordered system with a complete set of local integrals of
realization of a time crystal (Khemani et al 2016b). motion (LIOMs) (Serbyn et al 2013, Huse et al 2014). Non
The same group presented a nice general group analy- local Floquet states can not be all the common eigenstates of
sis of symmetry broken phases in Floquet systems (von the Hamiltonian and the quasi-local operators so the system of
Keyserlingk et al 2016, von Keyserlingk and Sondhi 2016b) LIOMs (together with the Floquet Hamiltonian) cannot form
in a similar setting. This time the authors chose to flip the a complete set of observables. As shown (Else et al 2016)
role of x and z axes considering the model defined by the condition of non local character of Floquet eigenstates is

11
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 6. The time-dependence of the disorder averaged

magnetizations σiα  ≡ σα  as indicated in the figure.
Magnetization along z axis shows persistent oscillations with twice
the period T of the driving—compare (31)–(33). Exact π spin flip is
assumed as in (31). Reprinted figure with permission from Else et al
(2016), Copyright (2016) by the American Physical Society.

equivalent to the statement that for any time instant, τ and

any state |ψ(τ ) with short-range correlations there exist an
operator Ô with the average in state ψ(τ ) being not time- Figure 7. Phase space diagram (a) in the plane of Jz-ε with Jz
periodic with period T despite T-periodicity of the Hamiltonian being the strength of the interactions while ε the deviation from the
itself. Note that this property is operational, i.e. it suggests perfect spin flip. The discrete (Floquet) time crystal phase appears
that DTTSB can be detected by measurements on such an only in the presence of the interactions and is robust with respect to
operator. changes of ε. The time crystal phase is identified by the magnitude
and variance of the Fourier peak, of the spin (z component)
The example studied, a simple spin system (very close to correlation function, at ω/2 as well as the peak position, where
that considered by Khemani et al (2016b), von Keyserlingk ω = 2π/T . Panel (b) illustrates that in the presence of interactions
et al (2016) and von Keyserlingk and Sondhi (2016b)) applies the peak is locked to ω/2 while the peak position changes
within a single period two operations, a global rotation around with ε without interactions. Panels (c) and (d) show the Fourier
x axis by π in time t1 = π/2 with the unitary transform shapes without and with the interactions, respectively,
corresponding to ε in (b). Reprinted figure with permission from
 
 Yao et al (2017), Copyright (2017) by the American Physical
(31) U1 = exp it1 σx , Society.
i

followed by an evolution for time t2 in a disordered reasoning can be easily extended to a still soluble case of
Hamiltonian HMBL, t1 = π/2 and hzi ∈ [0, hz ] (Else et al 2016).
The most important feature is that those TTSB solutions are
U2 = exp [−iHMBL t2 ] ,
(32) robust against perturbations such as non-zero h or t1 leading to
with T = t1 + t2 and a rotation around x axis by an angle slightly different from π.
  In effect σz  shows pronounced oscillations with the period
(33) H MBL = Ji σiz σi+1
z
+ hzi σiz + hxi σix . of 2T (compare figure 6) revealing DTTSB. The numerical
i studies use time-evolving block decimation (TEBD) scheme
The parameters Ji , hzi, and hxi are uniformly and randomly (Vidal 2004) for a system of 200 spins and h = 0.3. It is dif-
chosen, Ji ∈ [J/2, 3J/2], hzi ∈ [0, hz ], and hxi ∈ [0, h] with, ficult to separate the effects due to interactions and those due
typically J = hz = 1 and h  J . At the prize of repeating the to randomness since the latter is inherent in the model studied.
reasoning presented above, we mention that the system is sol- Quite a similar system is studied by Yao and coworkers
uble for h = 0 , the eigenstates of HMBL being the product of (Yao et al 2017) who stress the robustness of the observed
eigenvectors of individual σiz, |{mi }z, σkz |{mi }z = mk |{mi }z effect against perturbations. They concentrate on small devia-
with mk = ±1. Explicitly tions, say ε, of the ordered phase rotation, i.e. t1 = π/2 −  in
U1 (31) and actually draw a phase diagram in the interactions
HMBL |{mi }z = [E2 ({mi }) + E1 ({mi })] |{mi }z ,
(34) Jz-ε plane—see figure 7. The system they study is obtained
 
with E2 ({mi }) = i Ji mi mi+1 and E1 ({mi }) = i hzi mi . from (31)–(33) by taking Ji ∈ [0.8Jz , 1.2Jz ] and hxi = 0 with
Suppose for simplicity that all hzi = 0 . For t1 = π/2 in (31) t2 = 1 (we are grateful to Norman Yao for providing us with
the action of U1 is to flip all the spins {mi } → {−mi } while updated information in the parameters used). The data lead-
U2 adds a global phase. Clearly |{mi }z cannot be a Floquet ing to panel figure 7(a) are obtained after averaging over 100
eigenstate—but both Schrodinger-cat like combinations disorder realizations.
|± = (|{mi }z ± |{−mi }z are T-periodic Floquet eigen- The main claim of the paper is that there is a phase trans­
states. Such cat states are fragile to perturbations and DTTSB ition from DTTSB phase to trivial paramagnet which is of the
selects one of the pair (at random). Both |{mi }z and |{−mi }z Ising type. Importantly, Yao and coworkers (Yao et al 2017)
evolve periodically at exact resonance with a period 2T. This consider not only the abstract spin system but discuss small
12
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 8. Comparison of the Fourier transform of the spin z component correlation function without (left column) and in the presence
of (right column) the disorder for the short ranged system (31)–(33) in the top row and for the long ranged interactions corresponding to
the chain of ions (35) in the bottom. Reprinted figure with permission from Yao et al (2017), Copyright (2017) by the American Physical
Society.

systems with long range interactions providing quantitative disorder addressing each ion. This is accompanied by π/2
predictions for controlled ions chain. In that case the unitary pulses to transfer the disorder from σiz to σix . In effect the fol-
evolution over the period T = t1 + t2 is similar to the one lowing Floquet unitary operator is realized:
given above with interactions between nearest neighbours
U(T) = exp(−iH3 t3 ) exp(−iH2 t2 ) exp(−iH1 t1 ) with
replaced by the power law decaying long range interactions:  y
U(T) = U1 (t1 )U2 (t2 ) with U1 (t1 ) = exp(−it1 i σix ) and H1 = Ω(1 − ) σi ,
   i
 Jij  
 H2 = Jij σix σjx ,
U
(35) 
2 (t2 ) = exp −it2
 σz σz + hzi σiz  .
rijα i j i i,j
i=j

Interestingly the DTTSB time crystal behaviour seems to be H3 = hi σix . (36)
i
even more robust with long range interactions as also evi-
dent from additional material in supplementary information The Rabi rotation Ω = π/2t1 so for  = 0 spins are flipped (as
to that paper. It shows also a difference between a dis­ordered in theory propositions above). Non-zero, controllable ε allows
and clean cases showing in the character of the Fourier for checking the robustness of the Floquet time crystal formed.
transform of spin-spin correlation function as reproduced in The long range interactions Jij = J0 /|i − j|α with α = 1.5,
figure 8. site dependent disorder corresponds to hi uniformly chosen
in the interval [0, W] (Zhang et al 2017). While each of the
steps in (36) looks simple and straightforward the impressive
3.3. Experiments experimental strategy has to be carried out to actually imple-
A detailed proposition of ion chain experiment has proven ment (36) with the desired precision, the interested reader is
successful. The joint effort with experimental team of Chris advised to consult the original work and the supplementary
Monroe resulted in the report of an experimental realization of information (Zhang et al 2017). A single Floquet period lasts
discrete time-crystals (Zhang et al 2017). The ion chain con- T = t1 + t2 + t3 ≈ (15 + 27 + 33)µs being limited by a time
sists of 10 171Yb+ ions, each providing two sublevels of 2S1/2 needed for sufficient interactions between spins as well as by
state: |F = 1, mF = 0 and |F = 0, mF = 0 denoted as effec- the fact that disorder hi is applied to the ions consecutively.
tive 1/2-spin states | ↓z and | ↑z , respectively. Operations on The system is initiated in the separable state of all spins
a single spin are performed using optical Raman transitions. pointing downwards in the x-direction |Ψ = | ↓↓ . . . ↓x with
Interactions between spins are due to spin-dependent optical | ↓x = √12 (| ↓z + | ↑z ). After evolution with a variable num-
dipole forces. The AC Stark shifts due to off-resonant tightly ber of Floquet periods the magnetization along x is measured
focused laser beam allow one to introduce programmable yielding the time correlation function

13
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 9. Observed magnetization signal as a function of time (in the units of Floquet period T)—top row. Bottom row shows the
corresponding Fourier transforms (FTs). In the absence of interactions the spins oscillate with beats corresponding to the perturbation ε
from perfect synchronization, those beats may be affected by disorder (panel b). Disorder maximal amplitude is Wt3 = π as indicated in
the figure. In the presence of interactions (panel c) the self-organization of the system is restored as indicated by a single peak in the FT. As
shown in panel (d) time crystal survives even in the presence of quite large perturbations. Adapted from Zhang et al (2017) with permission
from Macmillan Publishers Ltd: Nature. Copyright (2017).


σix (t) = Ψ|σix (t)σix (0)|Ψ.
(37) H(t) = [Ωx (t)σix + Ωy (t)σiy + ∆i σiz ]
i
A typical duration of the experiment is about 100 Floquet   Jij  
periods. For optimally chosen t1 (i.e. for  = 0 in (36)) + σix σjx + σiy σjy − σiz σjz , (38)
and in the absence of the disorder and interactions one ij
rij3
expects the magnetization to restore after integer multiples where σiµ are Pauli spin-1/2 operators, Ωx and Ωy are Rabi fre-
of 2T because a single period evolution flips the spins. For quencies of microwave drivings, ∆i are disorder on-site fields
 = 0 , but small, spins rotation reveals beating that is indi-
which may be only qualitatively estimated, Jij are dipolar,
cated by two peaks in the Fourier transform of the time direction dependent coefficients with couplings decaying as a
evolution of the magnetization—the peaks are located sym- third power of the distance between vacancies.
metrically around 1/2T , see figure 9(a). Imperfections like While the physical system is very different from the one
the presence of disorder introduce dephasing among the spins, discussed previously, the assumed experimental scheme is a bit
figure 9(b). The crucial observation is that the presence of similar as shown in the top of figure 10. Initially spins are polar-
interactions restores the orderly behaviour, the Fourier trans- ized along x-axis in the initiation stage. Then during a single
form of the correlation function reveals a single peak at half period T two strategies are repeated: driving along x with Ωx for
the Floquet frequency, compare figure 9(c). The self-organi- τ1 followed by driving along y with Ωy for τ2 (the Floquet period
zation of the system may be understood as a manifestation is T = τ1 + τ2). As compared with the former experiment here
of time crystal behaviour. The initial product state in the the disorder and interactions are not switched off/on periodi-
experiment | ↓↓ . . . ↓x may be thought of as a superposi- cally but are given by a particular system realization. Almost π
tion of two Floquet eigenstates that are Schrödinger cat-like pulses are realized with Ωy , θ = Ωy τ2 ≈ π . A rotation along x
states |± = √12 (| ↓↓ . . . ↓x ± | ↑↑ . . . ↑x ). The Floquet with τ1 serves as an effective ‘interaction’ time. For relatively
eigenstates themselves evolve with the period T but they are short τ1 = 92 ns if θ = π a clear 2T-signal is observed in mag-
extremely fragile. Even if the system was prepared in one of netization (or its Fourier transform)—however deviations from
the Floquet eigenstates, it would collapse to a product state π destroy this synchronization, compare figure 10(c). Only when
under infinitesimally weak perturbation and starts evolving τ1 time is significantly increased (τ1 = 989ns) the Fourier trans-
with the period 2T that indicates spontaneous breaking of the form peak remains locked at 1/2T regardless of the detuning of
original discrete time translation symmetry. θ from the optimal value equal π. Thus the system self-organizes
Back to back report in the journal Nature (Choi et al 2017) into a time crystal like behaviour with 2T periods.
describes Floquet time-crystalline order observed in a dis­ It is worth stressing that the same work reports also the 3T-
ordered dipolar many body system—somehow as far from the behavior that may be realized modifying the driving scheme
controlled 10 ions chain as one could imagine. The system is such a way that after 3T the system of spins rotates closely
studied is an ensemble of nitrogen vacancy spin impurities to the initial configuration.
in diamond. Altogether about 106 impurities contribute to the
signal. Each nitrogen vacancy has unit spin but a two-level 3.4. Effect of disorder
system is isolated, by applying an external magnetic field, and
it can be manipulated using microwave radiation. In effect the Interestingly authors of Choi et al (2017) stress that their
following Hamiltonian is realized: experiment is not performed in the MBL regime, see also

14
Rep. Prog. Phys. 81 (2018) 016401 Review

discrete time crystals (Sacha 2015b) obtain their stability via a

semiclassical connection to the classical stability of motion in a
stable resonant s : 1 elliptical island. Recently another realiza-
tions of clean time crystals in cold atoms have been proposed
(Huang et al 2017). A series of clean quasi one-dimensional
models is shown to exhibit Floquet time crystalline behaviour,
robust in the strongly interacting regime due to emergent int­
egrals of motion in the dynamical systems studied. It will be
interesting to consider the connection between the last two
approaches. Last but not least it has been shown that clean
Floquet time crystal behaviour may be obtained in the well
known kicked Lipkin–Meshkov–Glick model (Russomanno
et al 2017). The stability of the time-crystal phase can be in
this model directly analyzed in the limit of infinite size, dis-
cussing the properties of the corresponding classical phase
space structure. Thus the role played by disorder and MBL
in time crystal phenomenon is by no means obvious. There
exists the numerical evidence (compare figure 8) that for some
spin systems disorder makes time crystal behaviour stronger,
still it does not seem to be necessary.
The real area of speculation (or careful analysis) opens
when one considers the thermodynamic limit. It is possible
that exponentially slow heating may occur, under favorable
circumstances in periodically driven many body systems
(Abanin et al 2015, 2017). The resistance of driven systems
to expected heating has been even verified experimentally
(Bordia et al 2017). Thus one may consider that before the
inevitable thermalization, even in the absence of MBL, for
some quite long time one may observe, in a finite system, a
time crystal behavior. Such a behavior is called a pre-ther-
malization and the corresponding time crystal properties
have been considered in such a regime (Else et al 2017). The
authors of diamond experiment (Choi et al 2017) argue that
their system cannot be considered as ‘pre-thermalizing’ yet
Figure 10. Panel (a) shows scheme of the experiment on nitrogen
vacancies. Initially spins are polarized along the positive x direction;
they observe the time crystal features in the experiment. It
during the period τ1 they evolve in the presence of interactions and may well be, therefore that for quite large yet finite systems
microwave field polarization aligned along the x axis; next the spins the time crystal behavior remains robust for sufficiently long
are rotated around the y axis by angle θ ≈ π and the entire procedure times that other mechanisms such as decoherence (Lazarides
is repeated many times. Panels (b)-(d) show observed magnetization and Moessner 2017) take over. Clearly more studies of robust-
(left column) and its Fourier transform (right column). (b) and (c)
correspond to short interaction time, the system is sensitive to tuning
ness and limitations of time-crystal phenomenon in different
showing trivial spin-flip signal at resonance when θ = π (b) which settings is desirable (Moessner and Sondhi 2017).
becomes modified for rotation detuned θ = 1.034π (c). For much
longer interaction time, the signal becomes robust and locked at
half the frequency corresponding to the period of the drive even for 4. Condensed matter physics in time crystals
rotation detuned from exact resonance, θ = 1.034π . Adapted from
Zhang et al (2017) with permission from Macmillan Publishers Ltd:
Nature. Copyright (2017).
Here we go beyond the demonstration of the time crystal
behaviour and present examples in which condensed matter
Zeng and Sheng (2017), which seems to contradict the claim phenomena may be observed in the time domain. The work
that MBL is a prerequisite for time crystal behavior (Else et al in this direction is still in its initial stages and one may expect
2016, Khemani et al 2016b, von Keyserlingk et al 2016, Yao further developments in the near future beyond early propos-
et al 2017). Moreover, a recent study of long-range interact- als revised below.
ing systems (Ho et al 2017) further elaborates on that point—
linking the stability of time crystals in the system studied to
4.1. Space periodic systems versus periodically driven
quant­um critical regime. The authors, using perturbative argu-
systems
ments for their 3D system, suggest, that there is a sharp cross-
over between a region where the discrete time crystal is stable In condensed matter physics it is often assumed that a
due to quantum criticality and the regime where discrete time space crystal is already formed and its properties are ana-
crystal eventually dies. Early, disorder free propositions of lyzed with the help of a space periodic Hamiltonian

15
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 11. Left panel: phase space structure corresponding to

a semiclassical version of the Hamiltonian HRWA (40) when the
operators are substituted by complex numbers, â → a and ↠→ a∗,
for s = 10, µ = 4.2 · 10−3 and λ → 0 , see (41). Periodic structure
visible in this panel was associated with a phase space crystal in
Guo et al (2013). Right panel: quasi-energy levels g(m) of (41), that
form a band structure, presented in the reduced Brillouin zone, i.e.
mτ = m2π/s ∈ [−π, π[, for s = 10, µ = 3.2 · 10−3 and λ = 1/205.
Reprinted figure with permission from Guo et al (2013), Copyright Figure 12. Particle bouncing on an oscillating mirror. Panel (a): plot
(2013) by the American Physical Society. of a Floquet eigenstate, related to localized wave-packets evolving
along s : 1 classical resonant orbit with s = 4, in configuration space
H(x + λ) = H(x). For ultra-cold atoms in an optical lattice for two slightly different moments of time, t = 0.25T (solid line)
the space periodicity is imposed by external laser fields, so and t = 0.3T (dashed line). The eigenstate is a superposition of 4
a space periodic Hamiltonian arises without the spontaneous wave-packets that move on the classical orbit. Each wave-packet
symmetry breaking. Counterparts of space periodic systems evolves with the period 4T but after T they exchange their positions
what makes the whole eigenstate periodic with the period T. There
but in the time domain are periodically driven systems where are 4 such Floquet eigenstates that are localized on the orbit. Proper
H(t + T) = H(t). As mentioned already there is a vast litera- superpositions of the eigenstates allows one to extract 4 individual
ture on the properties of periodically driven (sometimes called wave-packets, φj, that are numbered in (a) and (b). Crystalline
Floquet) systems. We concentrate here on a simple question structures are not visible in the configuration space, however, they
whether systems with time periodicity may reveal non-trivial emerge in the time domain. Panel (b) shows time evolution of the 4
wave-packets whose superpositions form the eigenstates localized
crystalline structures in time. on the resonant orbit, at z = 121 that is close to the turning point
Guo et al (2013) have considered a single particle in a per- of a particle in the classical description. These wave-packets are
turbed harmonic oscillator potential that is periodically driven, analogues of Wannier states corresponding to the lowest energy
band of a particle in a spatially periodic potential. Tunneling
p2 Ω 2 z2 ν between wave-packets that are neighbours in the time domain are
(39) H= + + z4 + 2V0 zs cos ωt the leading tunneling processes. Reprinted from Sacha et al (2015a)
2 2 2
with permission from Macmillan Publishers Ltd: Nature Scientific
and have shown that the resonant driving, i.e. when ω ≈ sΩ Reports. Copyright (2015).
with integer s  1, results in the band structure of the quasi-
energy spectrum of the system. This result has been demon- with
strated within the rotating wave approximation (RWA). That  s/2
2λV0 Ω|ω/s − Ω|
is, applying the unitary transformation U(t) = ei(ωt/s)â â,
†
(42)
µ= .
|ω/s − Ω| 3ν
where â, ↠are the usual harmonic oscillator annihilation and
creation operators, and dropping fast oscillating terms, the Looking at (41) it is easy to realize that HRWA is invariant under
effective Hamiltonian becomes the discrete rotation θ̂ → θ̂ + 2π/s , eigenstates of the system
 in the angular representation are Bloch-like waves um (θ)eimθ
ω † 3ν † †
HRWA = Ω − â â + â â(â â + 1) where um (θ + 2π/s) = um (θ) and the spectrum splits into
s 4Ω2 bands as illustrated in figure 11. The performed RWA is valid
 V0
+ (â†s + âs ). (40) provided Ω( ωs − Ω)2 < 32 ν what indicates the necessity of the
(2Ω)s/2
anharmonic term in the Hamiltonian (39) (Guo et al 2013).
The key observation is that HRWA displays a new dis- Guo et al called such a behaviour a phase space crystal. We
crete symmetry that is not visible in the original will see that a phase space crystal is actually a time crystal.
Hamiltonian (39), i.e. it commutes with a unitary operator That is, any resonantly driven system can be reduced to a solid
†
e−i2πâ â/s . It is convenient to introduce the radial r̂ and the state-like Hamiltonian by means of the secular approximation
√ √
angular θ̂ operators via â = e−iθ̂ r̂/ 2λ and ↠= r̂eiθ̂ / 2λ approach (Lichtenberg and Lieberman 1992, Buchleitner et al
where λ = 3ν/(4Ω2 |ω/s − Ω|) is an effective Planck con- 2002) and a periodic structure in the phase space is always
reproduced in the time domain (Sacha and Delande 2016).
stant. Then, HRWA = λ−1 | ωs − Ω|ĝ where
Let us first illustrate this conjecture on the problem of a
1 µ  iθ̂ s  −iθ̂ s  single particle bouncing on an oscillating mirror in the pres-
ĝ = (r̂2 + λ − 1)2 +
(41) r̂e + r̂e ,
4 2 ence of the gravitational field, see the Hamiltonian (21). It was

16
Rep. Prog. Phys. 81 (2018) 016401 Review

said that a suitable choice of the system parameters allows one also longer range tunnelings, e.g. between next-to-nearest
to realize single particle Floquet eigenstates where localized neighbour wave-packets but they are at least two orders of
wave-packets evolve along classical resonant periodic orbits. magnitude weaker in a typical situation and, similarly as in the
We have already analyzed the case of the 2:1 resonance. Now tight-binding model, can be neglected.
let us consider a general s : 1 resonance (Sacha 2015a). With We have demonstrated that the description of a resonant
a suitable choice of parameters it is possible to realize Floquet motion of a particle bouncing on an oscillating mirror can be
eigenstates that consist of s well localized wave-packets reduced to a tight-binding model known in condensed mat-
which evolve along s : 1 resonant orbit each of them with a ter physics (Guo et al 2013, Sacha 2015a). We have already
period s times longer than the period T of the mirror oscil- mentioned that such an approach can be applied to any res-
lations. They exchange their positions so that the Floquet onantly driven dynamical system ranging from ultra-cold
eigenstates are periodic with the fundamental period T, see atomic gasses to a Rydberg electron perturbed by a micro-
an example for s = 4 in figure 12(a). There are s such Floquet wave field (Sacha and Delande 2016). Indeed, assume that
eigenstates u1 (z, t), . . . , us (z, t) and they are related to quasi- a single particle integrable classical system described by an
energies E1 , . . . , Es which are nearly degenerate modulo unperturbed Hamiltonian H0 is driven periodically, i.e. there
ω/s . When s → ∞ these quasi-energies start forming a quasi- is another term in a system Hamiltonian, H = H0 + H1 (t),
energy band. Moreover, one can extract s individual localized which H1 (t + T) = H1 (t). It is convenient to describe clas-
wave-packets φ1 (z, t), . . . , φs (z, t) by a proper superpositions sical motion in action-angle variables of the unperturbed sys-
of s Floquet eigenstates. We anticipate that when s → ∞ these tem (Lichtenberg and Lieberman 1992). For example in the
wave-packets become counterparts of Wannier states known case of a 1D system, we can perform canonical transformation
in solid state physics that are localized in sites of a spatially from Cartesian position and momentum to canonically con-
periodic potential and are obtained from Bloch eigenstates by jugate action I and angle θ for which H0 = H0 (I). Then, the
a proper superposition (Kohn 1959). unperturbed motion can be solved immediately, I = constant
Now we have all elements in order to show that our sys- ∂H (I)
and θ = ∂I0 t + θ0 . The latter equation describes a position
tem can reveal crystalline structure in the time domain. In of a particle on its trajectory. Now, let us assume that we drive
figure 12(a) an example of a Floquet eigenstate is presented in the system harmonically with frequency ω which is close to a
the configuration space for two different moments of time so one multiple of frequency of an unperturbed motion,
can observe at which direction each of the localized wave-pack-
ets is propagating. However, the configuration space is not the ∂H0 (Is )
ω≈s
(45) ,
domain where we can see crystalline structures. Let us choose a ∂Is
point in the configuration space, e.g. close to the turning point of with an integer s. Then, in the moving frame,
a classical particle, and plot how the probability density for the
detection of a particle at this point changes in time, figure 12(b). ωt
Θ=θ− ,
(46)
We can see that each wave-packet arrives at this point periodi- s
cally and in the same way. That is guaranteed by the Floquet P = I − Is ,
(47)
theorem. Thus, the time domain is a proper ‘space’ where a crys-
talline structure emerges. This conjecture becomes clearer when are slowly varying. Averaging the original Hamiltonian
one restricts the description of a particle to the Hilbert subspace over time we obtain an effective time-independent (secular)
spanned by s evolving localized wave- packets and calculates Hamiltonian (Lichtenberg and Lieberman 1992)
the corresponding quasi-energy of a particle, P2
 s2π/ω  ∞ (48)
H≈ + V0 cos(sΘ),
2m
E= dt dz ψ ∗ (H(t) − i∂t ) ψ
0 0 where m and V0 are the effective mass and the amplitude of the
s
1 effective Hamiltonian respectively. Thus, in the moving frame
≈− (Jj a∗j+1 aj + c.c.), (43) any resonant motion with s  1 reduces to a solid state prob-
2
j=1 lem of a particle in a space periodic potential (48) (Lichtenberg
 and Lieberman 1992, Guo et al 2013, Sacha 2015a, Guo and
s
where ψ(z, t) ≈ j=1 aj φj (z, t) was substituted (Sacha Marthaler 2016, Sacha and Delande 2016).
2015a). Equation (43) is nothing but the kinetic energy term of In order to obtain a quantum effective description one
the tight-binding model known from solid state physics. The can either quantize a classical secular Hamiltonian (48)
localized wave-packets, φj, play a role of Wannier states and (Buchleitner et al 2002) or apply the quantum version of the
 s2π/ω  ∞ secular approximation from the very beginning (Berman and
(44) Zaslavsky 1977). By restricting to the lowest energy band of a
Jj = −2 dt dz φ∗j+1 (H(t) − i∂t ) φj
0 0 quantum version of (48) one ends up with a tight-binding model
are tunneling amplitudes. They are related to tunneling of a of the form of equation (43). Interestingly, it is also possible to
particle between neighbouring wave-packets (but neighbour- consider higher energy bands of the secular Hamiltonian (48)
ing in the time domain) see figure 12(b). All the tunneling and investigate multi-band physics (Guo et al 2013, Sacha
amplitudes have the same absolute values |Jj | = J . There are 2015a, Guo and Marthaler 2016, Sacha and Delande 2016).

17
Rep. Prog. Phys. 81 (2018) 016401 Review

Note that if a wave-function of a particle in the moving

frame, ψ(Θ), reveals some crystalline structure, in the labo-
ratory frame this wave-function, ψ(θ − ωt/s), will show
crystalline properties also versus time t if one fixes position
close to a classical orbit because the transformation (46) is
linear (Sacha and Delande 2016). On the other hand there is
no guarantee that such a wave-function will reveal periodic
behaviour as a function of a Cartesian position in space for
fixed time because canonical transformation between action-
angle variables and Cartesian position and momentum is non-
linear in general. This is the case for example for a particle
bouncing on an oscillating mirror where in the time domain
one observes a crystalline structure but does not see it in the
configuration space, figure 12.

4.2. Anderson localization in the time domain

Energy eigenstates of a single particle in the presence of a space Figure 13. Floquet eigenstate of the system (49), i.e. a particle
bouncing on an oscillating mirror (with a period T) in the presence
periodic potential are given by Bloch waves that are extended
of time fluctuating perturbation that is repeated with a period
in space. However, when a potential is not strictly periodic sT with s = 4. The perturbation is chosen so that j ’s in (50) are
because some disorder is present, eigenstates may become random numbers corresponding to a Lorentzian distribution. Then,
exponentially localized around different points of configura- equation (49) constitutes the Lloyd model of 1D lattice where all
tion space due to destructive interference between different eigenstates are Anderson localized and the exact expression for
the localization length is known (Haake 2001). From the Lloyd
multiple scattering paths (Anderson 1958). Regardless of the
model we know that the eigenstates are superpositions of the
magnitude of the disorder, provided it is random and decorre- 
wave-packets, j aj φj (z, t), with |aj |2 ∝ e−| j0 −j|/l where l is the
lated, Anderson localization is inevitable in one-dimensional
Anderson localization length and j0 is a number of the wave-packet
(1D) systems as well as for time-reversal invariant spinless around which a given eigenstate is localized. The wave-packets φj
two dimensional (2D) systems (Abrahams et al 1979, Müller arrive at a given position z in equidistant intervals in time, thus,
and Delande 2011). In the three dimensional (3D) world the the localization length in time is lt = lT . Solid lines in (a) and (b)
situation is more complicated, typically disordered system show one of the 4 eigenstates localized on the 4-resonant orbit at
reveal the so-called mobility edge. Eigenstates with energies z = 121 versus time in the linear (a) and logarithmic (b) scales.
Dash lines present behaviour of the eigenstates in the absence of the
below the mobility edge are localized and those above are disorder in time, i.e. when H  = 0. Despite the fact that the system
extended (Abrahams et al 1979, Müller and Delande 2011). is rather small, the characteristic exponential decay of the humps
Anderson localization can be also observed in the is clearly visible in (b)—the fitted exponential profile (dash-dotted
momentum space being called then the dynamical localiza- line) corresponds to lt = 0.18T . Reprinted from Sacha et al (2015a)
tion (Fishman et al 1982, Haake 1991, Moore et al 1995, with permission from Macmillan Publishers Ltd: Nature Scientific
Reports. Copyright (2015).
Stöckmann 1999, Lemarié et al 2009). Interestingly, periodi-
cally driven single particle systems can reveal Anderson local- It leads to additional terms in the tight-binding energy (43)
ization in the momentum space despite the classical diffusive which now takes the form
behaviour. The diffusion in the phase space is suppressed by
s s
quantum interference effects. 1 
E≈−
(49) (Jj a∗j+1 aj + c.c.) + j |aj |2 ,
In the present section it will be shown that yet another kind 2
j=1 j=1
of Anderson localization is possible—Anderson localization
in the time domain due to the presence of disorder in time where
(Sacha 2015a), i.e. when the disorder is added on top of the  s2π/ω  ∞
periodically changing force. (50)
j = dt dz H  (t) |φj |2 ,
0 0
4.2.1. Anderson localization in time crystals. The simple idea are random numbers that can belong to any distribution by
to realize Anderson localization in the time domain is to add a proper engineering of the fluctuating part, H  (t), of the
random vibrations to a mirror that oscillates with a period T in Hamiltonian (Sacha 2015a). Energy (49) constitutes actually
the bouncing particle problem that has been already described a 1D Anderson model which possesses exponentially local-
above. It is assumed that a bouncing particle is resonantly ized eigenstates. In figure 13 an example of a localized eigen-
driven by a periodically oscillating mirror and s : 1 resonance state is shown in the case when j ’s belong to a Lorentzian
condition is fulfilled. Moreover, there is an additional term distribution.
H  (t) in the system Hamiltonian that introduces a weak tem- The entire Hamiltonian is time periodic with a long period
poral disorder— H  (t) fluctuates in time but fulfills periodic sT. It guarantees existence of Floquet eigenstates periodic also
boundary condition on a long time scale, H  (t + sT) = H  (t). with a period sT. Anderson localization in the time domain

18
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 14. Comparison of Anderson localization in space and

time crystals. Left panel shows Anderson localization of a particle
in a 1D space crystal with periodic boundary conditions when a
disorder is added to a space periodic potential. Due to the periodic
boundary conditions when one travels along a ring again and
again, one observes periodically a localized wave-function. Right
panel illustrates Anderson localization in a time crystal. For a Figure 15. Probability density of a Floquet eigenstate corresponding
fixed position in configuration space, probability density for a to the Hamiltonian (51) versus time for a fixed position in the
measurement of a particle at this fixed point is localized around a laboratory frame. The presented Floquet state reveals Anderson
certain time moment. Such a behaviour is repeated periodically with localization in the time domain with localization length of the order
a long period sT. of 0.17/ω . The upper plot is on a linear scale, the lower plot on a
logarithmic scale, showing approximate exponential localization.
requires that the localization length in time has to be much Reprinted figure with permission from Sacha and Delande (2016) by
smaller than sT. If a detector is situated close to a classical the American Physical Society. Copyright (2016).
turning point in the configuration space, it is expected to click 
with probability that is localized exponentially around a cer- g(θ) = gn einθ ,
tain moment of time and such a behaviour is repeated periodi- (52)
n=0
cally with a period sT. It is in a complete analogy to Anderson
where gn = i(−1)n /(πn). The function f (t + 2π/ω) = f (t)
localization in a 1D space crystal with periodic boundary con-
is periodic but between t = 0 and 2π/ω it performs random
ditions in space (a ring topology). In figure 14 a comparison of
fluctuations, i.e.
Anderson localization in a time crystal and a 1D space crystal is 
presented. In the space crystal case when a disorder is present, f (t) = fk eikωt ,
a particle localizes around a certain space point and moving (53)
k=0
periodically around the ring, one observes periodically a local-
where fk = f−k∗ are independent random variables. Switching
ized density profile. The larger the ring, the larger the space
to the moving frame, Θ = θ − ωt and P = p − ω, one can see
crystal. In the time crystal case, probability for a measurement
that new position Θ and momentum P are slowly varying vari-
of a particle at fixed space point is exponentially localized in
ables if resonant condition is fulfilled, P ≈ 0 . Then, averaging
time and this behaviour is repeated periodically with a period
the Hamiltonian over the fast time variable results in a time-
sT and the higher s : 1 resonance the larger the time crystal.
independent effective Hamiltonian,
4.2.2. Anderson localization in time without crystalline P2
structures. Anderson localization in configuration space (54) H≈ + U(Θ),
2
takes place when a disorder is present in a space crystal. 
However, in order to observe Anderson localization it is not U(Θ) = V0 gk f−k eikΘ + constant,
necessary to start with a space crystal. One can begin with (55)
k=0
a purely dis­ordered time-independent potential characterized
where a particle moves in the presence of a time-independent
by a finite correlation length and Anderson localization can be
disordered potential U(Θ). This potential can be thought of
still observed (Kuhn et al 2007). It turns out that an analogous
as the coherent addition of resonant terms between spatial
situation takes place in the time domain that will be illustrated
harmonics of the potential and the corresponding temporal
with a simple model (Sacha and Delande 2016).
harmonics of the disordered driving amplitude (Sacha and
Let us consider a single particle on a ring described by the
Delande 2016). Statistical properties of Fourier components
following classical Hamiltonian,
of U(Θ) can be engineered by a proper choice of a distribu-
p2 tion for fk. The coefficients gk drops like 1/k and in order to
(51) H= + V0 g(θ) f (t),
2 deal with the disordered potential characterized by a small
correlation length, the drop of gk has to be compensated by an
where θ is an angle that denotes a position of a particle on a
increase of absolute values of fk.
ring and V0 is amplitude of the perturbation. The perturbation 2 2
is assumed to be regular in space (for fixed time t) and dis­ Assume that |gk f−k | ∝ e−k /(2k0 ) and Arg( f−k ) are random
ordered in time (for fixed θ). In the following it is assumed numbers chosen uniformly in the interval [0, 2π[. With a suit-
that g(θ) = θ/π for θ ∈ [−π, π[ but because −π and π cor- able choice of the amplitude V0 and the correlation length
√
respond to the same point on the ring, g(θ) has a discontinuity of the disordered potential, which is equal to 2/k0 , eigen-
and its Fourier expansion reads, states of the Hamiltonian (54) are Anderson localized even for
19
Rep. Prog. Phys. 81 (2018) 016401 Review

Figure 16. Panel (a): schematic plot of the initial stage of the
experiment: ultra-cold bosonic atoms are prepared in a strong
optical lattice and shallow trapping potentials. It is assumed that the
amplitude of the lattice potential is so strong that slices of the atomic Figure 17. Comparison of Mott insulator phase in the space and
cloud are formed that consist of well defined numbers of particles time domains. Panel (a) illustrates a space periodic potential in
and do not have mutual phase coherence. Panel (b): final stage of 1D with the periodic boundary conditions, the larger the system,
the experiment: the optical lattice is turned off, a modulated in time the greater number of potential sites along the ring. Particles in
sawtooth-like potential is turned on and atoms are kicked so that the potential wells illustrate a Mott insulator state where in each
their momentum along the z axis fulfills the resonance condition potential site there is a well defined number of particles. Panel (b)
with the modulation frequency. Atoms will fly over the time refers to a crystal structure in the time domain with the time periodic
modulated sawtooth potential and do not spread along z due to the conditions corresponding to the period sT. In the Mott insulator
predicted Anderson localization even though their energy is much regime, at a fixed position, a well defined numbers of particles
greater than the amplitude of the sawtooth potential. Reproduced arrive every time interval T. The greater s, the larger time crystal is.
from Delande et al (2017) with permission from K. Sacha. Reprinted from Sacha et al (2015a) with permission from Macmillan
Publishers Ltd: Nature Scientific Reports. Copyright (2015).
energies higher than the standard deviation of the disordered
Resonant bouncing of a single particle is described by quasi-
potential (Sacha and Delande 2016). These eigenstates when
energy of the form of a tight-binding model, equation (43).
plotted versus time for a fixed space point in the laboratory
In the many-body case a similar simplification can be used if
frame show up Anderson localization—an example is pre-
the interaction energy per particle is not much greater than a
sented in figure 15.
width of the quasi-energy band of the system (43). Close to
The model (51) can be realized in ultra-cold atomic gases
a s : 1 resonance a system of ultra-cold bosons with contact
as suggested in Delande et al (2017). Instead of trapping atoms
interactions may be then described in a restricted Hilbert sub-
in a ring-shape potential one can prepare a 1D sawtooth-shape
space spanned by Fock states |n1 , . . . , ns , where ni denotes
potential that is modulated in time in a random manner. It is
occupation of a localized wave-packet φi evolving along the
assumed that ultra-cold atoms are initially prepared in a time-
s : 1 resonant orbit. The corresponding many-body Floquet
independent optical lattice potential in a Mott insulator phase
Hamiltonian reads
(Pethick and Smith 2002) that results in a sequence of inde-
 s2π/ω  ∞  
pendent slices of an atomic cloud, see figure 16. Then, the g0
optical lattice is turned off, a modulated sawtooth potential HF = dt dz ψ̂ H(t) + ψ̂ † ψ̂ − i∂t ψ̂,
0 0 2
is turned on and atoms are kicked so that their velocity ful- s  s
fills resonance condition with the modulated sawtooth poten- 1 † 1
≈− (Jj âj+1 âj + h.c.) + Uij â†j â†j âj âj , (56)
tial. As a result of Anderson localization in time, the slices of 2 2
j=1 j=1
the atomic cloud keep their shape—they do not spread even 
though they fly with energy much greater than the amplitude s
of the modulated sawtooth potential. where the bosonic field operator ψ̂ ≈ j=1 φj âj and |Jj | = J ,
Other systems that seem promising for realization of see (43). The coefficients
Anderson localization in time are Rydberg atoms perturbed by  s2π/ω  ∞
fluctuating microwave field (Giergiel and Sacha 2017). A highly (57)
Uij = g0 dt dz |φi |2 |φj |2 ,
excited electron experiences resonant linearly polarized micro- 0 0
wave field that performs random fluctuations. An additional
static electric field along the microwave polarization axis stabi- describe effective interactions between particles that occupy
lizes a Kepler orbit an electron is traveling along. It results in an different localized wave-packets. Note that despite the fact the
effective Hamiltonian of the form of equation (54) in the frame original interactions between particles are contact, the effec-
moving with an electron because other degrees of freedom of the tive interactions can have long range character (Anisimovas
3D system are frozen due to the presence of the external fields. et al 2015, Eckardt and Anisimovas 2015, Sacha 2015a, Guo
One can observe Anderson localization of an electron along a et al 2016) because all localized wave-packets pass each other
classical Kepler orbit that corresponds to Anderson localization in evolution along a 1D resonant orbit. The on-site interac-
in time if a detector is placed at, e.g. nucleus. tions are dominating i.e. |Uii | > |Uij | for i = j.
For sufficiently strong repulsive (g0 > 0 ) contact interac-
4.3. Mott insulator in the time domain tions, when Uii  NJ/s , the ground state of the system (56)
is a single Fock state |N/s, . . . , N/s where well defined num-
In the present section we have so far considered single par- bers of atoms occupy each localized wave-packet, long-time
ticle systems. However, not only a single atom but a bunch phase coherence is lost, a gap opens between the ground state
of atoms can bounce on an oscillating mirror (Sacha 2015a). level and excited levels and consequently a Mott insulator

20
Rep. Prog. Phys. 81 (2018) 016401 Review

phase is realized in the time domain (Sacha 2015a), see sche- above it are not. In the laboratory frame, when one asks how
matic picture in figure 17. the probability density for a measurement of a particle at some
fixed value of, e.g. θ changes in time, it turns out it is local-
ized around a certain moment of time provided the relevant
4.4. Many-body localization with temporal disorder
Floquet eigenstate corresponds to energy of (59) below the
It is quite natural now to join the results of the previous parts mobility edge (Delande et al 2017).
and consider an interacting many-particle system driven reso-
nantly, with at the same time added perturbation that fluctu-
ates in time but is repeated with some large period sT. That 5. Conclusions
extends the analysis of section 4.2.1 on interacting particle
In the present article we have reviewed current state of the
systems. Combining with the many body language of the pre-
art of investigations of time crystals originally proposed by
vious section we are led to a genuine example that may result
Wilczek (Wilczek 2012), i.e. the phenomena that are related
in a many-body localization induced by temporal disorder.
to self-organization of quantum many-body systems in time.
Work in this direction is in progress (Mierzejewski et al 2017).
This kind of self-organization is a truly quantum effect and
should be distinguished from classical self-organization phe-
4.5. Time crystals with properties of multi-dimensional nomena where non-linear oscillators synchronize their motion
systems if a coupling between them is sufficiently strong (Glass and
Mackey 1988, Neda et al 2000). The formation of time crys-
It is difficult to predict at which direction, research on time
tals is quite analogous to the formation of space crystals. In
crystals will develop. There are open questions and new ideas
the space crystal case, a center of mass of a solid state system
are still coming. One of very intriguing topics concerns a
localizes due to spontaneous breaking of translational sym-
possibility of realization of multi-dimensional time crystals.
metry and the resulting quantum state is no longer invariant
While it is definitely not possible to augment dimensionality
under this symmetry. However, a remnant of the symmetry
of time, one can envisage 1D time crystals but with proper-
remains, i.e. a discrete space translation symmetry emerges
ties of, e.g. 3D space crystals. One proposal was already sug-
which results in a periodic structure of the probability density
gested in the literature that we shortly describe now.
for measurements of positions of atoms.
The problem of a single particle on a 1D ring subjected to
Time crystals are also related to spontaneous breaking
time-fluctuating perturbation (51) was generalized to the 3D
of a translation symmetry but the translation in time. Time-
case (Delande et al 2017). Consider a particle on a 3D torus
independent systems are invariant under continuous time
described by the following Hamiltonian:
translation transformation which means that if a system is
p2θ + p2ψ + p2φ prepared in an eigenstate, the corresponding probability den-
H= + V0 g(θ)g(ψ)g(φ) f1 (t) f2 (t) f3 (t). (58) sity does not evolve in time. Frank Wilczek anticipated that
 2
it is possible to prepare a many-body system in a state that
The function g(x) is the same as in equation (52). Time- under an infinitesimally weak perturbation, a system reveals
dependent functions fi (t) are periodic with periods 2π/ωi but periodic motion (Wilczek 2012). Subsequent studies revealed
between t = 0 and 2π/ωi they perform random fluctuations, that such a spontaneous formation of periodic structures in
 (i)
i.e. fi (t) = k=0 fk eikωi t where fk(i) = f−k
(i)∗
are random num- time is not possible if a system is prepared in the ground state
bers. Each degree of freedom is assumed to be in a resonance or in a thermal equilibrium state (Bruno 2013c, Watanabe
with one of the time-periodic functions fi (t) so that in the mov- and Oshikawa 2015). This is different from the space crystals
ing frame where Θ = θ − ω1 t , Ψ = ψ − ω2 t and Φ = φ − ω3 t case where spontaneous formation of crystalline structures is
and for the conjugate momenta PΘ = pθ − ω1 ≈ 0, observed in a thermal equilibrium state.
PΨ = pψ − ω2 ≈ 0 and PΦ = pφ − ω3 ≈ 0 , the effective sec- While the original time crystal proposition turned out to
ular Hamiltonian reads, be impossible for realization, Wilczek’s vision triggered a
new research field and became an inspiration to other scien-
P2 + P2Ψ + P2Φ
(59) Heff = Θ + Veff (Θ, Ψ, Φ), tists. Recently a proposition appeared (Syrwid et al 2017) that
2 avoids the above limitations suggesting to prepare the time
with Veff = V0 h1 (Θ)h2 (Ψ)h3 (Φ) where crystal using excited states. The main research interest concen-
 (i) trated, however, on a possibility to break discrete time transla-
hi (x) = gk f−k eikx , tion symmetry present in periodically driven systems, i.e. on
(60)
k=0
the so called discrete (Floquet) time crystals. This attempt has
provided ωi are irrational numbers. The effective potential in been described here all the way from the birth of the idea to
(59) is time-independent and it is a product of three indepen- its experimental realizations (Sacha 2015b, Else et al 2016,
dent disordered potentials along each degree of freedom. From Khemani et al 2016b, Choi et al 2017, Nayak 2017, Yao et al
the general theory of Anderson localization in 3D (Abrahams 2017, Zhang et al 2017). Discrete time crystals are related to
et al 1979) one expects the existence of the mobility edge. self-reorganization of periodically driven many-body systems.
That is, for properly engineered time fluctuating functions Under certain conditions a many-body system spontaneously
fi (t) and for a fixed amplitude V0, all eigenstates of (59) with switches its period of motion despite the fact that an external
energies below the mobility edge are Anderson localized and driving would like a system to follow its periodic changes.
21
Rep. Prog. Phys. 81 (2018) 016401 Review

Another research direction concerns realization of condensed in periodically driven many-body systems Phys. Rev. B
matter physics in time crystals. Space crystals can be electric 95 014112
Abanin D A, De Roeck W and François H 2015 Exponentially slow
conductors or insulators when transport properties in space
heating in periodically driven many-body systems Phys. Rev.
are studied. It turns out that time crystals are able to possess Lett. 115 256803
analogues properties in time. It was already demonstrated that Abrahams E, Anderson P W, Licciardello D C and
time crystals can be Anderson or Mott insulators (Sacha 2015a). Ramakrishnan T V 1979 Scaling theory of localization:
While time is a single degree of freedom and it is not possible to absence of quantum diffusion in two dimensions Phys. Rev.
Lett. 42 673–6
realize multi-dimensional time crystals, time crystals with prop-
Anderson P W 1958 Absence of diffusion in certain random lattices
erties of multi-dimensional space crystals can be imagined and Phys. Rev. 109 1492–505
in fact they were already proposed (Delande et al 2017). Anderson P W 1997 Basic Notions Of Condensed Matter Physics
There are several possible ways of further developments. (Boulder: Westview Press)
One may, for example, define and classify space-time crystals Anisimovas E, Žlabys G, Anderson B M, Gediminas J and
Eckardt A 2015 Role of real-space micromotion for bosonic
exhibiting mixed periodicities in both space and time (Xu and
and fermionic floquet fractional chern insulators Phys. Rev. B
Wu 2017). One may also pose a question whether time crystal 91 245135
is robust with respect to coupling to an external environ­ment. Ashcroft N W and Mermin N D 1976 Solid State Physics (HRW
Time translation symmetry breaking leading to time crystal International Editions) (Holt: Rinehart and Winston)
formation may be considered as an effect of the measure- Atas Y Y, Bogomolny E, Giraud O and Roux G 2013 Distribution
of the ratio of consecutive level spacings in random matrix
ment—see section 2—or external perturbation. Still some
ensembles Phys. Rev. Lett. 110 084101
argue (Lazarides and Moessner 2017) that coupling to the Basko D M, Aleiner I L and Altschuler B L 2006 Ann. Phys., NY
environment must eventually destroy the time crystal while 321 1126
specific example of effective time-crystal creation by deco- Bastidas V M, Emary C, Schaller G and Brandes T 2012
herence have also been introduced (Nakatsugawa et al 2017). Nonequilibrium quantum phase transitions in the ising model
Phys. Rev. A 86 063627
Time crystals are also very attractive candidates for quantum
Berman G P and Zaslavsky G M 1977 Theory of quantum nonlinear
simulators that seem to be more flexible than those suggested resonance Phys. Lett. A 61 295–6
so far. In the time crystals case, time is an additional degree of Bialynicki-Birula I, Kaliński M and Eberly J H 1994 Lagrange
freedom, thus, one has an additional knob to control quantum equilibrium points in celestial mechanics and nonspreading
simulation. We believe that a current strong activity in the field of wave packets for strongly driven rydberg electrons Phys. Rev.
Lett. 73 1777–80
time crystals will reveal novel phenomena that are difficult to dis-
Bordia P, Luschen H, Schneider U, Knap M and Bloch I 2017
cover in condensed matter systems or simply overlooked so far. Periodically driving a many-body localized quantum system
Bearing in mind that a time degree of freedom adds an additional Nat. Phys. 13 460–4
dimension, more possibilities for new discoveries are opened. Bruno P 2013a Comment on ‘quantum time crystals’ Phys. Rev.
Lett. 110 118901
Bruno P 2013b Comment on ‘space-time crystals of trapped ions’
Acknowledgments Phys. Rev. Lett. 111 029301
Bruno P 2013c Impossibility of spontaneously rotating time
We are grateful to Dominique Delande for discussions of vari- crystals: a no-go theorem Phys. Rev. Lett. 111 070402
Buchleitner A and Delande D 1995 Nondispersive electronic
ous aspects of time crystals. This work was performed within wave packets in multiphoton processes Phys. Rev. Lett.
the EU Horizon2020 FET project QUIC (nr. 641122). We 75 1487–90
acknowledge support of the National Science Centre, Poland Buchleitner A, Delande D and Zakrzewski J 2002 Non-dispersive
via project No.2016/21/B/ST2/01095 (KS) and 2016/21/B/ wave packets in periodically driven quantum systems
ST2/01086 (JZ). Phys. Rep. 368 409–547
Carr L D, Clark Charles W and Reinhardt W P 2000 Stationary
solutions of the one-dimensional nonlinear Schrödinger
Note added in proof. Recently new relevant works equation. II. case of attractive nonlinearity Phys. Rev. A
62 063611
appeared that has not been discussed in the present review Castillo E, Koch B and Palma G 2014 On the dynamics of
article (Gong et al 2017, Iemini et al 2017, Wang et al 2017, fluctuations in time crystals (arXiv:1410.2261 [quant-ph])
Flicker 2017a, Flicker 2017b). Castin Y and Dum R 1998 Low-temperature bose-einstein
condensates in time-dependent traps: beyond the u(1)
symmetry-breaking approach Phys. Rev. A 57 3008–21
Chandran A and Sondhi S L 2016 Interaction-stabilized steady
ORCID iDs states in the driven o(n) model Phys. Rev. B 93 174305
Chernodub M N 2013 Rotating casimir systems: magnetic-
Jakub Zakrzewski https://orcid.org/0000-0003-0998-9460 field-enhanced perpetual motion, possible realization in
doped nanotubes, and laws of thermodynamics Phys. Rev. D
87 025021
References Choi S et al 2017 Observation of discrete time-crystalline order in a
disordered dipolar many-body system Nature 543 221–5
D’Alessio L and Rigol M 2014 Long-time behavior of isolated
Abanin D A, De Roeck W, Ho W W and François H 2017 Effective periodically driven interacting lattice systems Phys. Rev. X
hamiltonians, prethermalization, and slow energy absorption 4 041048

22
Rep. Prog. Phys. 81 (2018) 016401 Review

Delande D and Buchleitner A 1994 Classical and quantum chaos in Huse D A, Nandkishore R and Oganesyan V 2014a Phenomenology
atomic systems Advances In Atomic, Molecular, and Optical of fully many-body-localized systems Phys. Rev. B 90 174202
Physics vol 34, ed B Benjamin and W Herbert (New York: Iemini F, Russomanno A, Keeling J, Schirò M, Dalmonte M and
Academic) pp 85–123 Fazio R 2017 Boundary time crystals (arXiv:1708.05014
Delande D, Morales-Molina L and Sacha K 2017 Three- [quant-ph])
dimensional localized-delocalized Anderson transition in the Jiang L, Kitagawa T, Alicea J, Akhmerov A R, Pekker D, Refael G,
time domain (arXiv:1702.03591 [quant-ph]) Cirac J I, Demler E, Lukin M D and Zoller P 2011 Majorana
Delande D, Sacha K, Marcin P, Avazbaev Sanat K and Zakrzewski J fermions in equilibrium and in driven cold-atom quantum wires
2013 Many-body Anderson localization in one-dimensional Phys. Rev. Lett. 106 220402
systems New J. Phys. 15 045021 Kaplan T, Horsch P and von der Linden W 1989 Order parameter in
Delande D and Zakrzewski J 1998 Spontaneous emission of quantum antiferromagnets J. Phys. Soc. Japan 58 3894
nondispersive rydberg wave packets Phys. Rev. A 58 466–77 Khemani V, von Keyserlingk C W and Sondhi S L 2016a Defining
Dembiński S T, Makowski A J and Pepłowski P 1993 Quantum time crystals via representation theory (arXiv:1612.08758
bouncer with chaos Phys. Rev. Lett. 70 1093–6 [cond-mat.stat-mech])
Eckardt A and Anisimovas E 2015 High-frequency approximation Khemani V, Lazarides A, Moessner R and Sondhi S L 2016b
for periodically driven quantum systems from a floquet-space Phase structure of driven quantum systems Phys. Rev. Lett.
perspective New J. Phys. 17 093039 116 250401
Else D V, Bauer B and Nayak C 2016 Floquet time crystals Kohn W 1959 Analytic properties of bloch waves and wannier
Phys. Rev. Lett. 117 090402 functions Phys. Rev. 115 809–21
Else D V, Bauer B and Nayak C 2017 Prethermal phases of matter Koma T and Tasaki H 1993 Symmetry breaking in heisenberg
protected by time-translation symmetry Phys. Rev. X 7 011026 antiferromagnets Commun. Math. Phys. 158 191
Else D V and Nayak C 2016 Classification of topological phases in Kuhn R C, Sigwarth O, Miniatura C, Delande D and Müller C A
periodically driven interacting systems Phys. Rev. B 93 201103 2007 Coherent matter wave transport in speckle potentials
Faizal M, Khalil M M and Das S 2016 Time crystals from minimum New J. Phys. 9 161
time uncertainty Eur. Phys. J. C 76 30 Kuwahara T, Mori T and Saito K 2016 Floquet-magnus theory and
Fishman S, Grempel D R and Prange R E 1982 Chaos, quantum generic transient dynamics in periodically driven many-body
recurrences, and Anderson localization Phys. Rev. Lett. quantum systems Ann. Phys. 367 96–124
49 509–12 Lazarides A and Moessner R 2017 The fate of a discrete time
Flicker F 2017a Pell and Clapeyron words as stable trajectories in crystal in an open system (arXiv:1703.02547 [cond-mat.
dynamical systems (arXiv:1707.09333 [nlin.CD]) stat-mech])
Flicker F 2017b Time quasicrystals in dissipative dynamical Lazarides A, Das A and Moessner R 2014a Equilibrium states of
systems (arXiv:1707.09371 [nlin.CD]) generic quantum systems subject to periodic driving Phys. Rev.
Ghosh S 2014 Spontaneous generation of a crystalline ground state E 90 012110
in a higher derivative theory Physica A 407 245–51 Lazarides A, Das A and Moessner R 2014b Periodic
Giergiel K and Sacha K 2017 Anderson localization of a rydberg thermodynamics of isolated quantum systems Phys. Rev. Lett.
electron along a classical orbit Phys. Rev. A 95 063402 112 150401
Glass L and Mackey M C 1988 From Clocks to Chaos: the Rhythms Lazarides A, Das A and Moessner R 2015 Fate of many-body
of Life (Princeton, NJ: Princeton University Press) localization under periodic driving Phys. Rev. Lett.
Goldman N, Juzeliunas G, Öhberg P and Spielman I B 2014 Light- 115 030402
induced gauge fields for ultracold atoms Rep. Prog. Phys. Lemarié G, Chabé J, Szriftgiser P, Garreau J C, Grémaud B and
77 126401 Delande D 2009 Observation of the anderson metal-insulator
Gong Z, Hamazaki R and Ueda M 2017 Discrete time-crystalline transition with atomic matter waves: Theory and experiment
order in cavity and circuit QED systems (arXiv:1708.01472 Phys. Rev. A 80 043626
[cond-mat.stat-mech]) Li T, Gong Z-X, Yin Z-Q, Quan H T, Yin X, Zhang P, Duan L-M
Guo L, Liu M and Marthaler M 2016 Effective long-distance and Zhang X 2012a Reply to comment on ‘space-time crystals
interaction from short-distance interaction in a periodically of trapped’ (arXiv:1212.6959)
driven one-dimensional classical system Phys. Rev. A Li T, Gong Z-X, Yin Z-Q, Quan H T, Yin X, Zhang P, Duan L-M
93 053616 and Zhang X 2012b Space-time crystals of trapped ions
Guo L and Marthaler M 2016 Synthesizing lattice structures in Phys. Rev. Lett. 109 163001
phase space New J. Phys. 18 023006 Lichtenberg A J and Lieberman M A 1992 Regular and chaotic
Guo L, Marthaler M and Schön G 2013 Phase space crystals: a new dynamics Applied Mathematical Sciences (New York:
way to create a quasienergy band structure Phys. Rev. Lett. Springer)
111 205303 Maeda H and Gallagher T F 2004 Nondispersing wave packets
Haake F 1991 Quantum Signatures of Chaos (Berlin: Springer) Phys. Rev. Lett. 92 133004
Haake F 2001 Quantum Signatures of Chaos (Physics and Mendonca J T and Dodonov V V 2014 Time crystals in ultracold
Astronomy Online Library) (New York: Springer) matter J. Russ. Laser Res. 35 93–100
Harper F and Roy R 2017 Floquet topological order in interacting Mierzejewski M, Giergiel K and Sacha K 2017 Many-body
systems of bosons and fermions Phys. Rev. Lett. 118 115301 localization caused by temporal disorder (arXiv:1706.09791
Henkel J and Holthaus M 1992 Classical resonances in quantum [cond-mat.dis-nn])
mechanics Phys. Rev. A 45 1978–86 Moessner R and Sondhi S L 2017 Equilibration and order in
Ho W W, Choi S, Lukin M D and Abanin D A 2017 Critical time quantum floquet matter Nat. Phys. 13 424–8
crystals in dipolar systems (arXiv:1703.04593 [cond-mat. Moore F L, Robinson J C, Bharucha C F, Sundaram B and
dis-nn]) Raizen M G 1995 Atom optics realization of the quantum
Holthaus M 1995 On the classical-quantum correspondence for δ-kicked rotor Phys. Rev. Lett. 75 4598–601
periodically time dependent systems Chaos Solitons Fractals Mori T, Kuwahara T and Saito K 2016 Rigorous bound on energy
5 1143–67 (quantum chaos: present and future) absorption and generic relaxation in periodically driven
Huang B, Wu Y-H and Liu W V 2017 Clean floquet time crystals: quantum systems Phys. Rev. Lett. 116 120401
models and realizations in cold atoms (arXiv:1703.04663 Müller C A and Delande D 2011 Disorder and interference:
[cond-mat.quant-gas]) localization phenomena Lecture Notes of the Les Houches

23
Rep. Prog. Phys. 81 (2018) 016401 Review

Summer School in Singapore: Ultracold Gases and Quantum Steane A, Szriftgiser P, Desbiolles P and Dalibard J 1995 Phase
Information vol 91 (Oxford: Oxford University Press) ch 9 modulation of atomic de broglie waves Phys. Rev. Lett.
Nakatsugawa K, Fujii T and Tanda S 2017 Effective quantum time 74 4972–5
crystal by decoherence (arXiv:1703.04002 [quant-ph]) Stöckmann H-J 1999 Quantum Chaos an Introduction (Cambridge:
Nandkishore R and Huse D A 2015 Many-body-localization and Cambridge University Press)
thermalization in quantum statistical mechanics Ann. Rev. Strocchi F 2005 Symmetry Breaking (The Lecture Notes in Physics)
Condens. Matter. Phys. 6 15 (Berlin: Springer)
Nayak C 2017 Condensed-matter physics: marching to a different Syrwid A, Zakrzewski J and Sacha K 2017 Time crystal behavior of
quantum beat Nature 543 185–6 (news and views) excited eigenstates (arXiv:1702.05006 [quant-ph])
Neda Z, Ravasz E, Brechet Y, Vicsek T and Barabasi A-L 2000 Thakurathi M, Patel A A, Sen D and Dutta A 2013 Floquet
Self-organizing processes: the sound of many hands clapping generation of majorana end modes and topological invariants
Nature 403 849–50 Phys. Rev. B 88 155133
Nicolis A and Piazza F 2012 Spontaneous symmetry probing Thies M 2014 Semiclassical time crystal in the chiral Gross–Neveu
J. High Energy Phys. JHEP12(2012)25 model (arXiv:1411.4236 [hep-th])
Nozières P 2013 Time crystals: can diamagnetic currents drive a van Wezel J and van den Brink J 2007 Spontaneous symmetry
charge density wave into rotation? Europhys. Lett. 103 57008 breaking in quantum mechanics Am. J. Phys. 75 160401
Oganesyan V and Huse D A 2007 Localization of interacting Vidal G 2004 Efficient simulation of one-dimensional quantum
fermions at high temperature Phys. Rev. B 75 155111 many-body systems Phys. Rev. Lett. 93 040502
Oleś B, Ziń P, Chwedeńczuk J, Sacha K and Trippenbach M 2010 von Keyserlingk C W, Khemani V and Sondhi S L 2016 Absolute
Bose–Einstein condensate in a double well potential in the stability and spatiotemporal long-range order in floquet
vicinity of a critical point Laser Phys. 20 671–7 systems Phys. Rev. B 94 085112
Pethick C and Smith H 2002 Bose–Eistein Condensation in Dilute von Keyserlingk C W and Sondhi S L 2016 Phase structure of one-
Gases (Cambridge: Cambridge University Press) dimensional interacting floquet systems. I. Abelian symmetry-
Pierański P 1983 Jumping particle model. Period doubling cascade protected topological phases Phys. Rev. B 93 245145
in an experimental system J. Physique 44 573–8 von Keyserlingk C W and Sondhi S L 2016 Phase structure of one-
Po H C, Fidkowski L, Morimoto T, Potter A C and Vishwanath A dimensional interacting floquet systems. II. Symmetry-broken
2016 Chiral floquet phases of many-body localized bosons phases Phys. Rev. B 93 245146
Phys. Rev. X 6 041070 Volovik G E 2013 On the broken time translation symmetry in
Ponte P, Papić Z, François H and Abanin D A 2015 Many-body macroscopic systems: precessing states and off-diagonal long-
localization in periodically driven systems Phys. Rev. Lett. range order JETP Lett. 98 491
114 140401 Wang R R W, Xing B, Carlo G G and Poletti D 2017 Period-
Potter A C, Morimoto T and Vishwanath A 2016 Classification of doubling in period-1 steady states (arXiv:1708.09070
interacting topological floquet phases in one dimension [quant-ph])
Phys. Rev. X 6 041001 Watanabe H and Oshikawa M 2015 Absence of quantum time
Pustyl’nikov L D 1978 Stable and oscillating motions in crystals Phys. Rev. Lett. 114 251603
nonatonomous dynamical systems Trans. Moskow Math. Soc. Wilczek F 2012 Quantum time crystals Phys. Rev. Lett.
14 1–101 109 160401
Robicheaux F and Niffenegger K 2015 Quantum simulations of Wilczek F 2013a Superfluidity and space-time translation symmetry
a freely rotating ring of ultracold and identical bosonic ions breaking Phys. Rev. Lett. 111 250402
Phys. Rev. A 91 063618 Wilczek F 2013b Wilczek reply Phys. Rev. Lett. 110 118902
Roy R and Harper F 2016 Abelian floquet symmetry-protected Xu S and Wu C 2017 Space-time crystal and space-time group
topological phases in one dimension Phys. Rev. B 94 125105 symmetry (arXiv:1703.03388 [cond-mat.str-el])
Russomanno A, Iemini F, Dalmonte M and Fazio R 2017 Floquet Yamamoto N 2015 Generalized bloch theorem and chiral transport
time crystal in the lipkin-meshkov-glick model Phys. Rev. B phenomena Phys. Rev. D 92 085011
95 214307 Yao N Y, Potter A C, Potirniche I-D and Vishwanath A 2017
Sacha K 2015 Anderson localization and Mott insulator phase in the Discrete time crystals: rigidity, criticality, and realizations
time domain Sci. Rep. 5 10787 Phys. Rev. Lett. 118 030401
Sacha K 2015 Modeling spontaneous breaking of time-translation Zakrzewski J, Delande D and Buchleitner A 1995 Nonspreading
symmetry Phys. Rev. A 91 033617 electronic wave packets and conductance fluctuations
Sacha K and Delande D 2016 Anderson localization in the time Phys. Rev. Lett. 75 4015–8
domain Phys. Rev. A 94 023633 Zaslavsky G V 1970 Statistical Irreversibility in Nonlinear Systems
Sambe H 1973 Steady states and quasienergies of a quantum- (Moskow: Nauka)
mechanical system in an oscillating field Phys. Rev. A Zeng T-S and Sheng D N 2017 Prethermal time crystals in
7 2203–13 one-dimensional periodically driven Floquet system
Serbyn M, Papić Z and Abanin D A 2013 Local conservation laws (arXiv:1707.00404 [cond-mat.str-el])
and the structure of the many-body localized states Phys. Rev. Zhang J et al 2017 Observation of a discrete time crystal
Lett. 111 127201 Nature 543 217–20 (letter)
Shapere A and Wilczek F 2012 Branched quantization Phys. Rev. Ziń P, Chwedeńczuk J, Oleś B, Sacha K and Trippenbach M 2008
Lett. 109 200402 Critical fluctuations of an attractive Bose gas in a double-well
Shapere A and Wilczek F 2012 Classical time crystals Phys. Rev. potential Europhys. Lett. 83 64007
Lett. 109 160402 Žnidarič M, Prosen T and Prelovšek P 2008 Many-body localization
Shirley J H 1965 Solution of the schrödinger equation with a in the heisenberg xxz magnet in a random field Phys. Rev. B
hamiltonian periodic in time Phys. Rev. 138 B979–87 77 064426

24
Rep. Prog. Phys. 81 (2018) 016401 Review

Krzysztof Sacha received his D.Phil. degree in 1998 from Jagiellonian University in Kraków and joined there the Faculty of
Physics, Astronomy and Computer Sciences. He was a fellow of Alexander von Humboldt Foundation at Philipps-Universität
Marburg and a fellow of Fulbright Program at Los Alamos National Laboratory. He worked on quantum chaos, strong field
ionization of atoms and ultra-cold atomic gases.

Jakub Zakrzewski, MSc 1981 from Jagiellonian University, PhD (physics) from Institute of Physics, Polish Academy of Sciences,
Warsaw 1985, since 1984 at Marian Smoluchowski Institute of Physics, Faculty of Physics, Mathematics and Applied Computer
Science, Jagiellonian University, (with two year break for a post-doc at Department of Chemistry, USC, Los Angeles) since 1996 as
professor of physics, since 2009 head of Atomic Optics Department. His scientific interests include quantum optics, strong laser
atom interactions, quantum chaos, cold atomic gases, strongly interacting many-body systems which he was happy to contribute to
while in Krakow or during frequent visits to Paris to Laboratoire Kastler-Brossel at Ecole Normale Superieure and University Pierre
et Marie Curie as invited professor as well as to ICFO, Barcelona.

25