betaFeedback

WorksPhysics and Astronomy

Floquet insulators and lattice fermions

Abstract

Floquet insulators are periodically driven quantum systems that can host novel topological phases as a function of the drive parameters. These new phases exhibit features reminiscent of fermion doubling in discrete-time lattice fermion theories. We make this suggestion concrete by mapping the spectrum of a noninteracting (1+1)D Floquet insulator for certain drive parameters onto that of a discrete-time lattice fermion theory with a time-independent Hamiltonian. The resulting Hamiltonian is distinct from the Floquet Hamiltonian that generates stroboscopic dynamics. It can take the form of a discrete-time Su-Schrieffer-Heeger model with half the number of spatial sites of the original model, or of a (1+1)D Wilson-Dirac theory with one quarter of the spatial sites.

Zero-energy fermionic modes, or zero modes, are among the earliest manifestations of topology in quantum many-body theory. In systems with a gapped bulk, they appear localized at topological defects like solitons and vortices, or as dispersive modes at boundaries between phases with different bulk topological invariants [1]. They can carry fractional charges under global symmetries and manifest fractional or non-Abelian statistics when braided [2, 3, 4, 5, 6, 7], and their existence enforces degeneracies in the many-body spectrum [8, 9].

Quantum systems driven periodically in time, known as Floquet systems, furnish intrinsically nonequilibrium generalizations of topological and conventionally ordered phases [10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21]. Despite lacking energy conservation, they retain a notion of eigenstates and and eigenvalues when observed at “stroboscopic times” that are integer multiples of the driving period TT. Instead of an energy spectrum which can in principle be unbounded in the thermodynamic limit, stroboscopic dynamics and eigenstates are characterized by a bounded spectrum of quasienergies −π/T≤ϵ<π/T-\pi/T \le\epsilon< \pi/T that are only conserved modulo 2π/T2\pi/T. The periodic nature of quasienergy furnishes a generalization of zero modes. For example, fermionic Floquet systems known as Floquet insulators can exhibit localized “π\pi modes” whose existence implies a “π\pi pairing” between many body states at quasienergy ϵ\epsilon and ϵ+π/T\epsilon+\pi/T [22, 23, 24, 25, 26, 27].

Another setting in which π\pi modes arise is spacetime lattice regularizations of fermionic quantum field theories. In particular, discretizing the Dirac operator on a spacetime lattice leads to the so-called fermion doubling problem, where the total number of fermionic degrees of freedom increases by a factor of 2D2^D, where DD is the spacetime dimension [28, 29]. The extra modes, known as “doublers,” are undesirable and several methods, including Wilson, Kogut-Susskind, and domain-wall fermions [30, 31, 32, 33, 34, 35], are commonly used to dispense with them. Nevertheless, it is natural to ask whether the doubler modes

Image not extracted

Figure 1. Schematic of the mapping between stroboscopic Floquet dynamics and time-independent lattice Hamiltonians. The spectrum of the stroboscopic dynamics corresponding to Eq. (5), shown in (a), can be mapped onto that of a discrete-time Su-Schrieffer-Heeger (b) or Wilson-Dirac (c) theory. The number of spatial sites in (b) is halved with respect to (a) to accommodate the additional degrees of freedom due to fermion doubling. The number of sites are further halved in (c) to accommodate the spinful nature of the Wilson-Dirac fermions (depicted via arrows inside the lattice sites).

associated with zero modes, which occur at frequency π/τ\pi/\tau where τ\tau is the temporal lattice spacing, are related to the π\pi modes that can appear in Floquet insulators.

In this paper, we answer this question affirmatively for a particular (1+1)(1+1)D Floquet-insulator model. Specifically, we show that the quasienergy spectrum of a continuous-time Floquet model of spinless complex fermions can be mapped directly onto the spectrum of discrete-time lattice fermions with a time-independent Hamiltonian. The lattice spacing of this discrete time theory equals the driving time period TT of the Floquet system. The mapping is made easier by the observation that the Dirac equation takes the form of a Schroedinger equation for the corresponding fermion Hamiltonian. The lattice fermion model can take the form of a Su-Schrieffer-Heeger (SSH) model with half the spatial lattice sites of the Floquet model, or of a Wilson-Dirac model with one quarter of the sites. Furthermore, for appropriately chosen solutions of the mapping between spectra, the phase diagrams of the Floquet and lattice models match. In particular, the topological phase of the Floquet model, which exhibits localized zero and π\pi modes with open boundary conditions, coincides with the topological phase of the related lattice Hamiltonians, where π\pi modes appear as doublers of zero modes. While previous studies [36, 37] have used Floquet systems to provide new perspectives on lattice fermion doubling, here we focus on exposing a direct correspondence between the two.

**Lattice fermion doubling.—**A continuous-time free-fermion theory in Minkowski space with Hamiltonian HH has the fermion/Dirac operator γ0(i∂t−H)\gamma_0(i\partial_t-H) from which we can extract the Schrödinger operator i∂t−Hi\partial_t-H. The eigenvalues of this operator are ±p02−E2\pm\sqrt{p_0^2-E^2}, where p0p_0 is the Fourier variable conjugate to time tt and EE are the eigenvalues of HH [38]. The zeroes of these eigenvalues at p0=Ep_0=E correspond to poles of the fermion propagator (i∂t−H)−1(i\partial_t-H)^{-1}. Discretizing time in this theory leads to fermion doubling [28, 29]. To see this, let τ\tau be the lattice spacing in the time direction. The poles of the discrete-time theory now satisfy 1τsin⁡(p0τ)=E\frac{1}{\tau}\sin(p_0\tau)=E, which leads to two solutions:

p0=1τsin⁡−1(Eτ)andp0=πτ−1τsin⁡−1(Eτ)(1)p_0=\frac{1}{\tau}\sin^{-1}(E\tau)\quad\text{and}\quad p_0=\frac{\pi}{\tau}-\frac{1}{\tau}\sin^{-1}(E\tau) \tag*{(1)}

Thus, for every pole of the continuous-time propagator there are two poles of the discrete-time propagator [39, 40, 41, 42]. In particular, for a zero mode of the continuous-time theory, the discrete-time theory has modes at p0=0p_0=0 and p0=πτp_0=\frac{\pi}{\tau}; the new π\pi mode arises purely due to time discretization.

A corresponding phenomenon is also observed in Euclidean-time lattice field theory [43] [44]. There, the Schrödinger operator in continuous time has eigenvalues ±p02+E2\pm\sqrt{p_0^2+E^2}, which, unlike the Minkowski theory, vanishes only at p0=E=0p_0=E=0. In this case, upon discretizing time, fermion doubling manifests itself as a degeneracy of Schrödinger eigenvalues. For instance, any eigenstate ∣p0,E⟩|p_0,E\rangle with eigenvalue 1τsin⁡2(p0τ)+E2\frac{1}{\tau}\sqrt{\sin^2(p_0\tau)+E^2} has a partner ∣πτ−p0,E⟩|\frac{\pi}{\tau}-p_0,E\rangle with the same eigenvalue. In congruence with the Minkowski-space analysis, if the continuous-time fermion propagator has a pole at p0=E=0p_0=E=0, then there is another pole at p0=π/τp_0=\pi/\tau. More interestingly, the states ∣E,E⟩|E,E\rangle, which correspond to the energy eigenstates of HH, have the same spatial profiles in the position eigenbasis as ∣πτ−E,E⟩|\frac{\pi}{\tau}-E,E\rangle. Therefore, if HH has a spatially localized zero mode, the corresponding discrete-time Schrödinger equation has two solutions with the same spatial profile.

In this paper, we will consider real-time (i.e., Minkowski) theories as opposed to imaginary-time (Euclidean) ones, as the comparison between the Floquet and lattice spectra is more natural for the former. However, there is no fundamental obstruction to performing a similar analysis for Euclidean theories, and such an approach may be desirable when generalizing beyond the case of noninteracting fermions, which is our focus here. We will briefly mention how the comparison between Floquet systems and lattice fermions in Minkowski spacetime can be adapted for Euclidean spacetime lattices.

**Floquet insulator model.—**As a simple example of a Floquet insulator, we consider a continuous-time (1+1)(1+1)D model on a spatial lattice of 2N2N sites defined by the evolution operator

U(t)={e−iH0tfor 0<t<t0,e−iH1(t−t0)e−iH0t0for t0≤t<t0+t1,(2)U(t)=\begin{cases} e^{-iH_0t} & \text{for }0<t<t_0,\\ e^{-iH_1(t-t_0)}e^{-iH_0t_0} & \text{for }t_0\le t<t_0+t_1, \end{cases} \tag*{(2)}

with

H0=2∑j=0N−1(a2j†a2j+1+H.c.)H_0=2\sum_{j=0}^{N-1}(a^\dagger_{2j}a_{2j+1}+\mathrm{H.c.})
{#eq:3}

H_1=2\sum_{j=0}^{N-1}(a^\dagger_{2j+1}a_{2j+2}+\mathrm{H.c.}),

{#eq:3}

where aia_i is a fermion annihilation operator on site i=0,…,2N−1i=0,\ldots,2N-1. Unless otherwise specified, we consider periodic boundary conditions (PBC) such that a2N≡a0a_{2N}\equiv a_0. The Hamiltonians in Eq. (3) are the “trivial” and “topological” parts of the (static) SSH model [2, 45],

HSSH=u2H1+v2H0.(3)H_{\mathrm{SSH}}=\frac{u}{2}H_1+\frac{v}{2}H_0. \tag*{(3)}

The energy spectrum of this model with PBC is given by

ESSH(k)=±u2+v2+2uvcos⁡(2k),(4)E_{\mathrm{SSH}}(k)=\pm\sqrt{u^2+v^2+2uv\cos(2k)}, \tag*{(4)}

where 0≤k<π0\le k<\pi is the crystal momentum and we have set the spatial lattice spacing to 11. The SSH model has a symmetry-protected topological (SPT) and a trivial phase that have identical bulk spectra. The two gapped phases can be distinguished by their energy spectra in open boundary conditions (OBC): in the SPT phase there are two spatially localized zero modes pinned to the middle of the energy gap, one at each end of the chain, while in the trivial phase there are none. Expanding the model around k=π2k=\frac{\pi}{2} yields a Dirac fermion theory with a mass proportional to u−vu-v; the transition between the SPT and trivial phases occurs at the massless point u=vu=v. H0H_0 and H1H_1 can be viewed as representatives of the trivial and SPT phases, respectively.

The Floquet model (2) is closely related to a simple (1+1)(1+1)D model discussed in Refs. [24, 27, 46], with the small difference that Eq. (2) is formulated in terms of complex rather than Majorana fermion operators [47]. The analysis of both models is nearly identical, and they have the same phase diagram. Since energy is no longer conserved in the time-dependent model (2), phases are classified according to spectral properties of the Floquet operator

UF=U(T)=e−iH1t1e−iH0t0≡e−iHFT,(5)U_F=U(T)=e^{-iH_1t_1}e^{-iH_0t_0}\equiv e^{-iH_FT}, \tag*{(5)}

where T=t0+t1T=t_0+t_1 is the driving period and HF=iTln⁡UFH_F=\frac{i}{T}\ln U_F is known as the Floquet or stroboscopic Hamiltonian. The quasienergies −π/T≤ϵ<π/T-\pi/T\leq\epsilon<\pi/T are the eigenvalues of HFH_F. They can be obtained analytically because the model (5) is quadratic [47]. There are four phases, which we label trivial, 00, π\pi, and 0π0\pi (see Figure 2). The 00 and π\pi phases have localized boundary modes with ϵ=0\epsilon=0 and π/T\pi/T, respectively; the 0π0\pi phase has both zero and π\pi modes, while the trivial phase has neither. The phase boundaries in Figure 2 are the lines in the t0t_0-t1t_1 plane where the quasienergy gap closes. The trivial and 00 phases are adiabatically connected to the trivial and topological phases of the SSH model, respectively. The phases with π\pi modes are “intrinsically Floquet” phases in the sense that they cannot arise in the absence of the drive.

Image not extracted

Figure 2. Phase diagram of the Floquet model (5). The phases are labeled by the presence or absence of zero and π\pi modes localized to boundaries. This work focuses on the vertical dashed line at t0T=π4\frac{t_0}{T}=\frac{\pi}{4}, which passes through the trivial and 0π0\pi phases.

Floquet to lattice mapping.—It is natural to speculate that the π\pi modes in the Floquet system can be viewed as doubler modes of an appropriate discrete-time lattice fermion theory with time lattice constant τ=T\tau=T. To confirm this, we will identify a discrete-time theory with Hamiltonian HH and spectrum EE such that the poles in Eq. (1) are in one-to-one correspondence with quasienergy eigenvalues ϵ\epsilon. Note that, in order for this identification to be possible, it is necessary that the (single-particle) spectrum of HFH_F exhibit a π\pi-pairing such that for every quasienergy ϵ\epsilon there is a partner at π/T−ϵ\pi/T-\epsilon. For the model (5), this feature arises along the line t0T=π4\frac{t_0}{T}=\frac{\pi}{4}, which connects the trivial and 0π0\pi phases. On this line, the quasienergy spectrum of HFH_F with PBC takes the form [47]:

ϵ(k)=±cos⁡−1[−cos⁡(2η)cos⁡(2k)],(6)\epsilon(k)=\pm\cos^{-1}\left[-\cos(2\eta)\cos(2k)\right], \tag*{(6)}

where η=t1T−π4\eta=\frac{t_1}{T}-\frac{\pi}{4} measures the distance from the gap closure at t1T=π4\frac{t_1}{T}=\frac{\pi}{4}. To enable a one-to-one mapping onto the poles (1), we will partition the quasienergy spectrum ϵ\epsilon into two subsets containing quasienergies ϵ~\tilde{\epsilon} and πT−ϵ~\frac{\pi}{T}-\tilde{\epsilon}. We will then seek a model HH whose energy spectrum E~\tilde{E} satisfies

E=1Tsin⁡(Tϵ~),(7)E=\frac{1}{T}\sin(T\tilde{\epsilon}), \tag*{(7)}

which stems from identifying p0=ϵ~p_0=\tilde{\epsilon} in Eq. (1). A suitable choice of ϵ~\tilde{\epsilon} consists of the values ϵ(k)\epsilon(k) for π4≤k<3π4\frac{\pi}{4}\leq k<\frac{3\pi}{4} (see Figure 3). The corresponding construction in Euclidean spacetime aims to identify the eigenvalues of the states ∣E,E⟩|E,E\rangle and ∣πT−E,E⟩|\frac{\pi}{T}-E,E\rangle with 1T2sin⁡(Tϵ~)\frac{1}{T}\sqrt{2}\sin(T\tilde{\epsilon}).

Image not extracted

Figure 3. Spectrum of the Floquet model and its mapping onto the lattice spectrum. (a) Quasienergy spectrum of Eq. (5) along the line t0T=π4\frac{t_0}{T}=\frac{\pi}{4} [see Eq. (6)]. The quasienergy values highlighted in blue are those corresponding to crystal momenta kk in the interval [π4,3π4][\frac{\pi}{4},\frac{3\pi}{4}] (grey vertical lines), denoted ϵ~\tilde{\epsilon} in the text. (b) Sine-transformed quasienergy spectrum entering Eq. (7). (c) Pole positions p0p_0 with momenta assigned according to the SSH mapping. The orange bands denote doublers, which correspond to the quasienergies πT−ϵ~\frac{\pi}{T}-\tilde{\epsilon}.

SSH mapping.—We now seek a Hamiltonian whose spectrum satisfies Eq. (7). Since UFU_F is built using SSH-type Hamiltonians (3), we first consider an SSH Hamiltonian with dispersion given by Eq. (4). In this case, Eq. (7) can be solved by

ESSH(k′)=1Tsin⁡[Tϵ(k′2+π4)],(8)E_{\mathrm{SSH}}(k')=\frac{1}{T}\sin\left[T\epsilon\left(\frac{k'}{2}+\frac{\pi}{4}\right)\right], \tag*{(8)}

where 0≤k′<π0\leq k'<\pi, where

u=1±sin⁡(2η)2Tandv=1T−u.(9)u=\frac{1\pm\sin(2\eta)}{2T}\qquad\text{and}\qquad v=\frac{1}{T}-u. \tag*{(9)}

(Note that we have assumed u,v≥0u,v\geq0 for simplicity.) The two possible assignments of uu satisfying Eq. (8) have no impact on the bulk energy spectrum, but can nevertheless be physically distinguished solely by the SSH model with OBC. The SSH model (3) is in the SPT phase when u>vu>v and the trivial phase when u<vu<v. Since the topological 0π0\pi phase of HFH_F occurs when η>0\eta>0, we must pick the branch of Eq. (9) such that u>vu>v when η>0\eta>0; this is accomplished by picking the “++” branch. Let H~SSH\widetilde{H}_{\mathrm{SSH}} denote the resulting time-independent SSH Hamiltonian. By construction, the doubled spectrum (1) of H~SSH\widetilde{H}_{\mathrm{SSH}} when defined on a discrete-time lattice with spacing τ=T\tau=T [ Figure 3(c) ] matches the quasienergy spectrum of the Floquet Hamiltonian HFH_F [ Figure 3(a) ]. Thus, H~SSH\widetilde{H}_{\mathrm{SSH}} cannot be defined on the same spatial lattice as HFH_F; rather, if the original Floquet model is defined on a lattice of 2N2N sites, HSSHH_{\mathrm{SSH}} must be defined on NN sites. To see this, recall that Eq. (8) maps the interval π4≤k<3π4\frac{\pi}{4}\leq k<\frac{3\pi}{4} to the interval 0≤k′<π0\leq k'<\pi. The kk interval contains half of the NN allowed crystal momentum values for HFH_F, so k′k' can only take N2\frac{N}{2} values (note that this requires NN to be even). This corresponds to NN sites because the SSH model has a two-site unit cell. Thus, H~SSH\widetilde{H}_{\mathrm{SSH}} should not be interpreted as a mere rewriting of HFH_F. Instead, the models are related by the nontrivial procedure of fermion doubling.

Wilson-Dirac mapping

To make more direct contact with lattice field theory, we now show that Eq. (7) can also be satisfied by the spectrum of a Wilson-Dirac (WD) Hamiltonian. The corresponding solution is inspired by the observation that the spectrum of a WD Hamiltonian with PBC can be mapped onto that of an SSH Hamiltonian with PBC. The WD Hamiltonian on a 1D spatial lattice with NN sites can be written (here working in OBC for simplicity)

HWD=∑x,x′=0N−1ψˉx[Rγ1(−i∇x,x′)−R2∇x,x′2+mδx,x′]ψx′,(10)H_{\mathrm{WD}}=\sum_{x,x'=0}^{N-1}\bar{\psi}_{x}\left[R\gamma_{1}(-i\nabla_{x,x'})-\frac{R}{2}\nabla^{2}_{x,x'}+m\delta_{x,x'}\right]\psi_{x'}, \tag*{(10)}

where ∇x,x′=(δx′,x+1−δx′,x−1)/2\nabla_{x,x'}=(\delta_{x',x+1}-\delta_{x',x-1})/2 is a symmetric spatial derivative, ∇x,x′2=δx′−1,x+1+δx′+1,x−1−2\nabla^{2}_{x,x'}=\delta_{x'-1,x+1}+\delta_{x'+1,x-1}-2 is the second derivative, ψx\psi_x is a two-component Dirac spinor with associated gamma matrices γ0\gamma_0, γ1\gamma_1, and ψˉx=ψx†γ0\bar{\psi}_x=\psi_x^\dagger\gamma_0. The energy spectrum of this model with PBC is given by

EWD(p)=±R2sin⁡2p+[m+R(1−cos⁡p)]2,(11)E_{\mathrm{WD}}(p)=\pm\sqrt{R^{2}\sin^{2}p+[m+R(1-\cos p)]^{2}}, \tag*{(11)}

where −π≤p<π-\pi\leq p<\pi. Solving Eq. (7) yields

EWD(p′)=1Tsin⁡[Tε(p′4+π2)],(12)E_{\mathrm{WD}}(p')=\frac{1}{T}\sin\left[T\varepsilon\left(\frac{p'}{4}+\frac{\pi}{2}\right)\right], \tag*{(12)}

where −π≤p′<π-\pi\leq p'<\pi, provided that

m=±sin⁡(2η)TandR=12T−m2,(13)m=\pm\frac{\sin(2\eta)}{T}\qquad\text{and}\qquad R=\frac{1}{2T}-\frac{m}{2}, \tag*{(13)}

where we have assumed R≥0R\geq0 for simplicity. Following the SSH case, we now ask which of these branches corresponds to a topological phase when η>0\eta>0. With OBC, the WD Hamiltonian (11) exhibits localized edge modes when m<0m<0 [47]. Demanding that this condition coincide with positive η\eta selects the “−-” branch. We denote the resulting time-independent Wilson-Dirac Hamiltonian by H~WD\widetilde{H}_{\mathrm{WD}}.

We have thus found a second time-independent Hamiltonian, H~WD\widetilde{H}_{\mathrm{WD}}, whose doubled spectrum when defined on a discrete-time lattice coincides with the quasienergy spectrum of HFH_F. We note here that H~WD\widetilde{H}_{\mathrm{WD}} must be defined on a lattice with N/2N/2 sites (i.e., one quarter of the sites in the original Floquet model). Like the SSH case, the crystal momentum interval 0≤k<π/20\leq k<\pi/2 corresponding to ε~\tilde{\varepsilon} contains N/2N/2 points; thus, so does the new crystal momentum interval −π≤p′<π-\pi\leq p'<\pi. However, unlike the SSH model, the WD model has a one-site unit cell, so there is one site per crystal momentum value.

Discussion

The equalities (8) and (12) relate the discrete-time spectra of static SSH and WD models to the quasienergy spectrum of Eq. (5) with PBC. In each case, the “correct” branch of the respective model parameter solutions is chosen by matching the presence of zero modes as a function of η\eta. However, these zero modes are properties of the OBC spectrum. Given that the PBC spectra match by construction, we expect that the OBC spectra match up to ∼1/N\sim1/N corrections. In Figure 4(a) we demonstrate this numerically for the SSH mapping by plotting the system size dependence of the maximum difference between the discrete-time frequency spectrum p0p_0 and the quasienergy spectrum ε\varepsilon, where each spectrum is constructed as an ordered list.

Image not extracted

Figure 4. Finite-size scaling of the maximum difference between discrete-time spectra p0p_0 and quasienergy spectra ε\varepsilon (a) for OBC and (b) for OBC with a domain wall where η\eta switches sign. In both panels we compare to the SSH spectrum and fix η=π/8\eta=\pi/8, considering system sizes N=100,200,…,900N=100,200,\ldots,900. Dashed lines indicate best fits to power laws in 1/N1/N that are consistent with the expected scaling.

Another test of equivalence is to compare spectra in the presence of a domain wall in η\eta. For the SSH model, this corresponds to a standard domain wall in the mass profile [2], while for the WD model it amounts to a domain wall in both the mass and the Wilson parameter RR. For the Floquet model, a domain wall in η\eta entails an abrupt change in the coefficient of H1H_1 from 22 to 2(π4−η)(π4+η)2(\frac{\pi}{4}-\eta)(\frac{\pi}{4}+\eta) at some point in space. In Figure 4(b) we examine the finite-size dependence of the difference between the discrete-time SSH and quasienergy spectra in the presence of a domain wall in the middle of the chain, again finding the expected ∼1/N\sim1/N scaling. These results build confidence that the mapping we develop defines a notion of equivalence in the thermodynamic limit between models irrespective of boundary conditions or the presence of topological defects.

Outlook

We have demonstrated spectral equivalence between a simple Floquet insulator model in (1+1)(1+1) dimensions and two canonical fermion models (SSH and WD) defined on a two-dimensional spacetime lattice. The mapping between spectra relies on a notion of pairing in the single-particle spectrum, which occurs along the line t0T=π4\frac{t_0}{T}=\frac{\pi}{4}. This pairing allows us to discard half of the degrees of freedom in the Floquet model to define a related static model, which recovers the discarded eigenvalues via fermion doubling after time discretization. One interesting question for future work is whether the equivalence can be extended off the line t0T=π4\frac{t_0}{T}=\frac{\pi}{4} [48]. We note that π\pi pairing also occurs along the line t1T=π4\frac{t_1}{T}=\frac{\pi}{4} for PBC but is lost for OBC, as evidenced by the separate existence of 0 and π\pi modes with OBC. A separate direction is to consider extensions of these mappings to (2+1)D systems, where new phases like anomalous Floquet topological insulators emerge [23, 49]. In any dimension, it is also worth considering whether a similar equivalence between bulk topological invariants in Floquet systems [23, 50, 51, 52, 53] and lattice field theories can emerge [54]. Finally, it will be interesting to consider whether an equivalence to lattice field theory can be formulated for interacting Floquet systems, which experience unbounded heating that requires the introduction of prethermalization or localization physics to enable long-lived phenomena [24, 55, 56, 57].

Acknowledgments

TI acknowledges support from the National Science Foundation under grant DMR-2143635. SS and LS acknowledge support from the U.S. Department of Energy, Nuclear Physics Quantum Horizons program through the Early Career Award DE-SC0021892.

Supplemental materials for “Floquet insulators and lattice fermions”

Thomas Iadecola; Srimoyee Sen; Lars Sivertsen

QUASIENERGY SPECTRUM OF UFU_{\mathrm{F}}

In this section we will derive the quasienergy spectrum of the bulk Floquet Hamiltonian defined as

HF=−Im⁡[log⁡UF]=−Im⁡[log⁡(exp⁡(−iH1t1)exp⁡(−iH0t0))](14)H_{\mathrm{F}}=-\operatorname{Im}[\log U_{\mathrm{F}}]=-\operatorname{Im}[\log(\exp(-iH_{1}t_{1})\exp(-iH_{0}t_{0}))] \tag*{(14)}

where the SSH-Hamiltonians HiH_i are defined as

H0=2∑j=0N−1(a2j†a2j+1+H.c.)H1=2∑j=0N−1(a2j+1†a2j+2+H.c.).(15)\begin{aligned} H_0 &= 2\sum_{j=0}^{N-1}(a^\dagger_{2j}a_{2j+1}+\text{H.c.})\\ H_1 &= 2\sum_{j=0}^{N-1}(a^\dagger_{2j+1}a_{2j+2}+\text{H.c.}). \tag*{(15)} \end{aligned}

Since we are considering the bulk spectrum, we have in the above and will in what follows assume periodic boundary conditions, a2N=a0a_{2N}=a_0. We define

a2j=c2j,a2j+1=d2j+1.(16)a_{2j}=c_{2j},\qquad a_{2j+1}=d_{2j+1}. \tag*{(16)}

We proceed by performing a Fourier transform

c2j=1N∑ke2ikjck,d2j+1=1N∑ke+ik(2j+1)dk,(17)\begin{aligned} c_{2j} &= \frac{1}{\sqrt{N}}\sum_k e^{2ikj}c_k,\\ d_{2j+1} &= \frac{1}{\sqrt{N}}\sum_k e^{+ik(2j+1)}d_k, \tag*{(17)} \end{aligned}

where 0≤k<π0\leq k<\pi. The Hamiltonian written in momentum space are of the form

H0=2∑k(ck†dkeik+H.c.),H1=2∑k(ck†dke−ik+H.c.)(18)\begin{aligned} H_0 &= 2\sum_k(c^\dagger_k d_k e^{ik}+\text{H.c.}),\\ H_1 &= 2\sum_k(c^\dagger_k d_k e^{-ik}+\text{H.c.}) \tag*{(18)} \end{aligned}

which can be rewritten in matrix form as

H0=2(0eike−ik0)=2σxcos⁡k−2σysin⁡k≡2σ⋅n0,H1=2(0e−ikeik0)=2σxcos⁡k+2σysin⁡k≡2σ⋅n1.(19)\begin{aligned} H_0 &= 2\begin{pmatrix}0 & e^{ik}\\ e^{-ik} & 0\end{pmatrix}=2\sigma_x\cos k-2\sigma_y\sin k\equiv2\boldsymbol{\sigma}\cdot\mathbf{n}_0,\\ H_1 &= 2\begin{pmatrix}0 & e^{-ik}\\ e^{ik} & 0\end{pmatrix}=2\sigma_x\cos k+2\sigma_y\sin k\equiv2\boldsymbol{\sigma}\cdot\mathbf{n}_1. \tag*{(19)} \end{aligned}

Here σi\sigma_i are the Pauli spin matrices, and

n0=(cos⁡k−sin⁡k),n1=(cos⁡ksin⁡k),σ=(σxσy).(20)\mathbf{n}_0=\begin{pmatrix}\cos k\\-\sin k\end{pmatrix},\qquad \mathbf{n}_1=\begin{pmatrix}\cos k\\\sin k\end{pmatrix},\qquad \boldsymbol{\sigma}=\begin{pmatrix}\sigma_x\\\sigma_y\end{pmatrix}. \tag*{(20)}

Using the relation

exp⁡(−iθ(σ⋅n))=cos⁡(θ)−i(σ⋅n)sin⁡(θ)(21)\exp(-i\theta(\boldsymbol{\sigma}\cdot\mathbf{n}))=\cos(\theta)-i(\boldsymbol{\sigma}\cdot\mathbf{n})\sin(\theta) \tag*{(21)}

we advance towards finding the eigenvalues of the Floquet operator by inserting the expressions for H0H_0 and H1H_1 into (S1):

UF=exp⁡(−iH1t1)exp⁡(−iH0t0)=exp⁡(−2it1(σ⋅n1))exp⁡(−2it0(σ⋅n0))(22)U_{\mathrm{F}}=\exp(-\mathrm{i}H_1t_1)\exp(-\mathrm{i}H_0t_0) =\exp(-2\mathrm{i}t_1(\boldsymbol{\sigma}\cdot\mathbf{n}_1))\exp(-2\mathrm{i}t_0(\boldsymbol{\sigma}\cdot\mathbf{n}_0)) \tag*{(22)}

From here one can find the eigenvalues of UFU_{\mathrm{F}}

λ=14[cos⁡(2k−2t0−2t1)+2cos⁡(2t0−2t1)−cos⁡(2k+2t0−2t1)−cos⁡(2k−2t0+2t1)+2cos⁡(2t0+2t1)+cos⁡(2k+2t0+2t1)]±i{1−116[cos⁡(2k−2t0−2t1)+2cos⁡(2t0−t1)−cos⁡(2k+2t0−2t1)−cos⁡(2k−2t0+2t1)+2cos⁡(2t0+2t1)+cos⁡(2k+2t0+2t1)]2}1/2=e±iθ.(23)\begin{aligned} \lambda={}&\frac{1}{4}\left[\cos(2k-2t_0-2t_1)+2\cos(2t_0-2t_1)-\cos(2k+2t_0-2t_1)\right.\\ &\left.-\cos(2k-2t_0+2t_1)+2\cos(2t_0+2t_1)+\cos(2k+2t_0+2t_1)\right]\\ &\pm\mathrm{i}\left\{1-\frac{1}{16}\left[\cos(2k-2t_0-2t_1)+2\cos(2t_0-t_1)-\cos(2k+2t_0-2t_1)\right.\right.\\ &\left.\left.-\cos(2k-2t_0+2t_1)+2\cos(2t_0+2t_1)+\cos(2k+2t_0+2t_1)\right]^2\right\}^{1/2}\\ ={}&e^{\pm\mathrm{i}\theta}. \tag*{(23)} \end{aligned}

We can now obtain the eigenvalues of the Floquet Hamiltonian HFH_{\mathrm{F}} given by

λHF=±θ=±arccos⁡{14[cos⁡(2k−2t0−2t1)+2cos⁡(2t0−2t1)−cos⁡(2k+2t0−2t1)−cos⁡(2k−2t0+2t1)+2cos⁡(2t0+2t1)+cos⁡(2k+2t0+2t1)]}.(24)\begin{aligned} \lambda_{H_{\mathrm{F}}}&=\pm\theta\\ &=\pm\arccos\left\{\frac{1}{4}\left[\cos(2k-2t_0-2t_1)+2\cos(2t_0-2t_1)-\cos(2k+2t_0-2t_1)\right.\right.\\ &\left.\left.-\cos(2k-2t_0+2t_1)+2\cos(2t_0+2t_1)+\cos(2k+2t_0+2t_1)\right]\right\}. \tag*{(24)} \end{aligned}

In particular, note that for t0=π4t_0=\frac{\pi}{4} we have

λHF=±arccos⁡[−sin⁡(2t1)cos⁡(2k)].(25)\lambda_{H_{\mathrm{F}}}=\pm\arccos\left[-\sin(2t_1)\cos(2k)\right]. \tag*{(25)}

The arccos function has the feature

arccos⁡(−x)=π−arccos⁡(x),(26)\arccos(-x)=\pi-\arccos(x), \tag*{(26)}

and therefore, for a fixed value of t1t_1, there is a π\pi pairing in the eigenvalues of HFH_{\mathrm{F}}, i.e. if there is an eigenvalue λHF=κ\lambda_{H_{\mathrm{F}}}=\kappa, there is another one at π/T−κ\pi/T-\kappa. This follows as kk takes values from 00 to π\pi. Furthermore, note that if we instead let t1=π/4t_1=\pi/4, we get the exact same result, except t0→t1t_0\to t_1. Hence there is also π\pi pairing along the t1=π/4t_1=\pi/4 line. However, this pairing disappears with OBC, as the phases along this line have either zero or π\pi modes on the boundary, not both—the line t0/T=π/4t_0/T=\pi/4 is therefore the only region of the phase diagram with π\pi-pairing independent of boundary conditions, so it is there that we expect the cleanest relationship between the Floquet and lattice-fermion models.

Phases of UFU_{\mathrm{F}}

In this section we discuss the phase diagram of the Floquet operator

UF=e−iH1t1e−iH0t0,(27)U_{\mathrm{F}}=e^{-\mathrm{i}H_1t_1}e^{-\mathrm{i}H_0t_0}, \tag*{(27)}

where

H0=2∑j=0N−1(a2j†a2j+1+H.c.)H1=2∑j=0N−2(a2j+1†a2j+2+H.c.).(28)\begin{aligned} H_0&=2\sum_{j=0}^{N-1}(a^\dagger_{2j}a_{2j+1}+\mathrm{H.c.})\\ H_1&=2\sum_{j=0}^{N-2}(a^\dagger_{2j+1}a_{2j+2}+\mathrm{H.c.}). \tag*{(28)} \end{aligned}

Note that we have written the model with open boundary conditions to facilitate distinguishing the four phases by their edge modes. In this section we will work in units such that T=1T=1. This model is directly analogous to the “Class D” toy model of Ref. [1], with the exception that the latter is composed of Majorana fermions instead of complex fermions. We follow closely the analysis of Ref. [1]. The strategy is to characterize UFU_{\mathrm{F}} along the representative lines t0=0t_0=0, t1=0t_1=0, t0=π/2t_0=\pi/2, and t1=π/2t_1=\pi/2. Standard continuity arguments then imply that any point in parameter space that can be reached from these lines without closing the quasienergy gap must be in the same phase.

Trivial and 0 phases

The trivial and 0 phases are closely related to their undriven counterparts. The trivial phase is smoothly connected to the line t1=0t_1=0, where UFU_{\mathrm{F}} reduces to evolution under the trivial Hamiltonian H0H_0, which hybridizes all lattice sites in pairs. The 0 phase is smoothly connected to the line t0=0t_0=0, where UFU_{\mathrm{F}} reduces to evolution under the topological Hamiltonian H1H_1, which hybridizes bulk sites while leaving the edge sites 0 and 2N−12N-1 decoupled.

0π phase

To understand the 0π phase, it is useful to rewrite H0H_0 and H1H_1 via a Jordan-Wigner transformation as

H0=∑j=0N−1(X2jX2j+1+Y2jY2j+1).(29)H_0=\sum_{j=0}^{N-1}(X_{2j}X_{2j+1}+Y_{2j}Y_{2j+1}). \tag*{(29)}
H1=∑j=0N−2(X2j+1X2j+2+Y2j+1Y2j+2).(30)H_1=\sum_{j=0}^{N-2}(X_{2j+1}X_{2j+2}+Y_{2j+1}Y_{2j+2}). \tag*{(30)}

We then consider UFU_{\mathrm{F}} along the line t1=π/2t_1=\pi/2:

UF=e−iπ2H1e−iH0t0(31)U_{\mathrm{F}}=e^{-i\frac{\pi}{2}H_1}e^{-iH_0t_0} \tag*{(31)}
=∏j=0N−2Z2j+1Z2j+2e−iH0t0(32)=\prod_{j=0}^{N-2}Z_{2j+1}Z_{2j+2}e^{-iH_0t_0} \tag*{(32)}
=Uedgee−it0(X0X1+Y0Y1+X2N−2X2N−1+Y2N−2Y2N−1)Q Ubulk(33)=U_{\mathrm{edge}}e^{-it_0(X_0X_1+Y_0Y_1+X_{2N-2}X_{2N-1}+Y_{2N-2}Y_{2N-1})}Q\,U_{\mathrm{bulk}} \tag*{(33)}

where Uedge=Z0Z2N−1U_{\mathrm{edge}}=Z_0Z_{2N-1}, Q=∏j=0N−1Z2jZ2j+1Q=\prod_{j=0}^{N-1}Z_{2j}Z_{2j+1}, Ubulk=∏j=1N−2e−it0(X2jX2j+1+Y2jY2j+1)U_{\mathrm{bulk}}=\prod_{j=1}^{N-2}e^{-it_0(X_{2j}X_{2j+1}+Y_{2j}Y_{2j+1})}, and where we have used the identity

e−iπ2H1=∏j=0N−2e−iπ2(X2j+1X2j+2+Y2j+1Y2j+2)(34)e^{-i\frac{\pi}{2}H_1}=\prod_{j=0}^{N-2}e^{-i\frac{\pi}{2}(X_{2j+1}X_{2j+2}+Y_{2j+1}Y_{2j+2})} \tag*{(34)}
=∏j=0N−2e−iπ2X2j+1X2j+2e−iπ2Y2j+1Y2j+2(35)=\prod_{j=0}^{N-2}e^{-i\frac{\pi}{2}X_{2j+1}X_{2j+2}}e^{-i\frac{\pi}{2}Y_{2j+1}Y_{2j+2}} \tag*{(35)}
=−∏j=0N−2X2j+1X2j+2Y2j+1Y2j+2(36)=-\prod_{j=0}^{N-2}X_{2j+1}X_{2j+2}Y_{2j+1}Y_{2j+2} \tag*{(36)}
=∏j=0N−2Z2j+1Z2j+2.(37)=\prod_{j=0}^{N-2}Z_{2j+1}Z_{2j+2}. \tag*{(37)}

Note that QQ constitutes a global symmetry as it commutes with H0H_0 and H1H_1. Next, we transform Eq. (33) by the unitary operator

W=eit02(X0X1+Y0Y1+X2N−2X2N−1+Y2N−2Y2N−1),(38)W=e^{i\frac{t_0}{2}(X_0X_1+Y_0Y_1+X_{2N-2}X_{2N-1}+Y_{2N-2}Y_{2N-1})}, \tag*{(38)}

which commutes with QQ, UedgeU_{\mathrm{edge}}, and UbulkU_{\mathrm{bulk}}, to obtain

U~F=W†UFW=W†Uedgee−it0(X0X1+Y0Y1+X2N−2X2N−1+Y2N−2Y2N−1)WQUbulk(39)\widetilde{U}_{F}=W^{\dagger}U_{F}W=W^{\dagger}U_{\mathrm{edge}}e^{-it_{0}(X_{0}X_{1}+Y_{0}Y_{1}+X_{2N-2}X_{2N-1}+Y_{2N-2}Y_{2N-1})}WQU_{\mathrm{bulk}} \tag*{(39)}
=UedgeWe−it0(X0X1+Y0Y1+X2N−2X2N−1+Y2N−2Y2N−1)WUbulk(40)=U_{\mathrm{edge}}We^{-it_{0}(X_{0}X_{1}+Y_{0}Y_{1}+X_{2N-2}X_{2N-1}+Y_{2N-2}Y_{2N-1})}WU_{\mathrm{bulk}} \tag*{(40)}
=UedgeQUbulk,(41)=U_{\mathrm{edge}}QU_{\mathrm{bulk}}, \tag*{(41)}

where we used that UedgeU_{\mathrm{edge}} anticommutes with the generator of WW. This demonstrates that the quasienergy spectrum of UFU_{F} factors into the product of the spectra of the commuting unitaries UedgeU_{\mathrm{edge}}, QQ, and UbulkU_{\mathrm{bulk}}. Eigenstates can be labeled uniquely by their eigenvalues under the commuting quantities UedgeU_{\mathrm{edge}}, QQ, UbulkU_{\mathrm{bulk}}, and under the additional bulk symmetry operator Pbulk=∏j=1N−2X2jX2j+1P_{\mathrm{bulk}}=\prod_{j=1}^{N-2}X_{2j}X_{2j+1} and edge symmetry operator Pedge=X0X1X2N−2X2N−1P_{\mathrm{edge}}=X_{0}X_{1}X_{2N-2}X_{2N-1}.

To see how 00 and π\pi modes emerge, let us denote a reference eigenstate of U~F\widetilde{U}_{F} by ∣uedge,pedge,pbulk,q,E⟩\lvert u_{\mathrm{edge}},p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangle, where EE is an energy eigenvalue of Hbulk=∑j=1N−2(X2jX2j+1+Y2jY2j+1)H_{\mathrm{bulk}}=\sum_{j=1}^{N-2}(X_{2j}X_{2j+1}+Y_{2j}Y_{2j+1}) and where uedge,pedge,pbulk,q=±1u_{\mathrm{edge}},p_{\mathrm{edge}},p_{\mathrm{bulk}},q=\pm1 are eigenvalues of UedgeU_{\mathrm{edge}}, PedgeP_{\mathrm{edge}}, PbulkP_{\mathrm{bulk}}, and QQ, respectively. (Note that all of these operators mutually commute with one another; we will therefore call them symmetries of UFU_{F}.) Denote the quasienergy of this reference state by ϵ=iln⁡(uedgeqe−iEt0)\epsilon=i\ln(u_{\mathrm{edge}}q e^{-iEt_{0}}). We can toggle between the four joint eigenstates of the edge operators UedgeU_{\mathrm{edge}} and PedgeP_{\mathrm{edge}} by acting with the operators Z0Z_{0}, which anticommutes with PedgeP_{\mathrm{edge}} and commutes with all other symmetries, and X0X1X_{0}X_{1}, which anticommutes with UedgeU_{\mathrm{edge}} and commutes with all other symmetries. This generates the following “0π0\pi-multiplet” of states:

StateQuasienergy
∣uedge,pedge,pbulk,q,E⟩\lvert u_{\mathrm{edge}},p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangleϵ\epsilon
Z0∣uedge,pedge,pbulk,q,E⟩=∣uedge,−pedge,pbulk,q,E⟩Z_{0}\lvert u_{\mathrm{edge}},p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangle=\lvert u_{\mathrm{edge},-p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangle}ϵ\epsilon
X0X1∣uedge,pedge,pbulk,q,E⟩=∣−uedge,pedge,pbulk,q,E⟩X_{0}X_{1}\lvert u_{\mathrm{edge}},p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangle=\lvert-u_{\mathrm{edge}},p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangleϵ+π\epsilon+\pi
Y0X1∣uedge,pedge,pbulk,q,E⟩=∣−uedge,−pedge,pbulk,q,E⟩Y_{0}X_{1}\lvert u_{\mathrm{edge}},p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangle=\lvert-u_{\mathrm{edge}},-p_{\mathrm{edge}},p_{\mathrm{bulk}},q,E\rangleϵ+π\epsilon+\pi

Table 1.

We refer to Z0Z_{0} as a zero-mode operator since it commutes with UFU_{F}, and X0X1X_{0}X_{1} as a π\pi-mode operator because it anticommutes with UFU_{F}. Note that there is another zero- and π\pi-mode operator at the other end of the chain: Z2N−1=UedgeZ0Z_{2N-1}=U_{\mathrm{edge}}Z_{0} and X2N−2X2N−1=PedgeX0X1X_{2N-2}X_{2N-1}=P_{\mathrm{edge}}X_{0}X_{1}. Thus we could have alternatively chosen to label states in this multiplet by, say, eigenvalues of Z0Z_{0} and Z2N−1Z_{2N-1} instead of UedgeU_{\mathrm{edge}} and PedgeP_{\mathrm{edge}}, and to toggle between these eigenvalues with X0X1X_{0}X_{1} and X2N−2X2N−1X_{2N-2}X_{2N-1}.

π\pi phase

The π\pi phase can be treated similarly to the 0π0\pi phase. Here we consider the representative point t0=π/2t_{0}=\pi/2, where

UF=e−iH1t1e−iπ2H0(42)U_{F}=e^{-iH_{1}t_{1}}e^{-i\frac{\pi}{2}H_{0}} \tag*{(42)}
=e−iH1t1Q(43)=e^{-iH_{1}t_{1}}Q \tag*{(43)}
=e−iH1t1(∏j=0N−2Z2j+1Z2j+2)Z0Z2N−1(44)=e^{-iH_{1}t_{1}}\left(\prod_{j=0}^{N-2}Z_{2j+1}Z_{2j+2}\right)Z_{0}Z_{2N-1} \tag*{(44)}
=e−iH1(t1+π2)Z0Z2N−1(45)=e^{-iH_{1}(t_{1}+\frac{\pi}{2})}Z_{0}Z_{2N-1} \tag*{(45)}
=UbulkUedge,(46)=U_{\mathrm{bulk}}U_{\mathrm{edge}}, \tag*{(46)}

where we have used Eq. (34). Note that the resulting model also has the global symmetry P=∏j=02N−1X2jX2j+1P=\prod_{j=0}^{2N-1}X_{2j}X_{2j+1} with eigenvalues p=±1p=\pm1. We can repeat the analysis of the 0π0\pi phase to demonstrate the existence of π\pi modes. To do this, let ∣uedge,p,E⟩\lvert u_{\mathrm{edge}},p,E\rangle be a reference eigenstate labeled by the symmetry eigenvalues uedge,p=±1u_{\mathrm{edge}},p=\pm1 and an energy eigenvalue EE of H1H_{1}. The associated quasienergy is ϵ=iln⁡(uedgee−iEt1)\epsilon=i\ln(u_{\mathrm{edge}}e^{-iEt_{1}}). The eigenvalues of UedgeU_{\mathrm{edge}} can be toggled by the π\pi-mode operator X0X_{0}, leading to the following “π\pi-doublet”:

StateQuasienergy
∣uedge,p,E⟩\lvert u_{\mathrm{edge}},p,E\rangleϵ\epsilon
X0∣uedge,p,E⟩=∣−uedge,p,E⟩X_{0}\lvert u_{\mathrm{edge}},p,E\rangle=\lvert-u_{\mathrm{edge}},p,E\rangleϵ+π\epsilon+\pi

Table 2.

Mapping to two copies of the Majorana model

A simpler way to verify that the phase diagram of UFU_{\mathrm{F}} matches that of the Majorana model of Ref. [1] is to observe that the former amounts to two copies of the latter. To see this we write the complex fermions aja_j in terms of Majorana fermions αj,βj\alpha_j,\beta_j in the standard way:

aj=12(αj+iβj).(47)a_j = \frac{1}{2}(\alpha_j+i\beta_j). \tag*{(47)}

Substituting this into Eq. (28) gives

H0=∑j=0N−1(iα2jβ2j+1−iβ2jα2j+1)H_0 = \sum_{j=0}^{N-1}(i\alpha_{2j}\beta_{2j+1}-i\beta_{2j}\alpha_{2j+1})
H1=∑j=0N−2(iα2j+1β2j+2−iβ2j+1α2j+2).(48)H_1 = \sum_{j=0}^{N-2}(i\alpha_{2j+1}\beta_{2j+2}-i\beta_{2j+1}\alpha_{2j+2}). \tag*{(48)}

This is precisely two copies of the aforementioned Majorana model, with one of the copies time-reversed with respect to the other. We note that the two copies are not strictly independent, since performing a U(1) transformation on the complex fermions leads to a nontrivial mixing of the Majorana operators:

aj→eiθaj=(cos⁡θ αj+sin⁡θ βj)+i(sin⁡θ αj−cos⁡θ βj).(49)a_j \to e^{i\theta}a_j = (\cos\theta\,\alpha_j+\sin\theta\,\beta_j)+i(\sin\theta\,\alpha_j-\cos\theta\,\beta_j). \tag*{(49)}

This mixing is necessary in order for the two copies of the Majorana model to manifest the U(1)U(1) symmetry of the complex-fermion model.

Domain wall in η\eta in the Wilson-Dirac model

Here we consider a spatially dependent Wilson-Dirac Hamiltonian with a domain wall in the parameter η=t1T−π4\eta=\frac{t_1}{T}-\frac{\pi}{4} that tunes between the trivial and 0π0\pi phases. To implement this we pick the following domain wall configuration with ++ and −- subscripts denoting the regions x1>0x_1>0 and <0<0

m±=−sin⁡(2η±)R±=1+sin⁡(2η±)2η+=−η−>0.(50)\begin{aligned} m_{\pm} &= -\sin(2\eta_{\pm})\\ R_{\pm} &= \frac{1+\sin(2\eta_{\pm})}{2}\\ \eta_{+} &= -\eta_{-}>0. \tag*{(50)} \end{aligned}

We can now look for edge modes on the domain wall by solving the equation of motion

(−iRγ1∇1+m−R2∇12)ψ=0(51)\left(-iR\gamma_1\nabla_1+m-\frac{R}{2}\nabla_1^2\right)\psi=0 \tag*{(51)}

with γ0=σ2\gamma_0=\sigma_2 and γ1=−iσ1\gamma_1=-i\sigma_1 and mm and RR are spatially dependent, i.e. taking the above specified values for x1>0x_1>0 and <0<0. We can solve this EOM with an ansatz of

ψ=(11)φ(52)\psi=\begin{pmatrix}1\\1\end{pmatrix}\varphi \tag*{(52)}

which reduces the EOM to

−φ(x1+1)−φ(x1−1)2+mRφ−φ(x1+1)+φ(x1−1)−2φ(x1)2=0-\frac{\varphi(x_1+1)-\varphi(x_1-1)}{2}+\frac{m}{R}\varphi-\frac{\varphi(x_1+1)+\varphi(x_1-1)-2\varphi(x_1)}{2}=0
⟹φ(x1+1)=(1+mR)φ.(53)\Longrightarrow\varphi(x_1+1)=\left(1+\frac{m}{R}\right)\varphi. \tag*{(53)}

So, we find normalizable solutions for all of the parameter space of interest, i.e. π4>η>0\frac{\pi}{4}>\eta>0 given by

φ(x1)=(1+mR)x1={(1+m+R+)x1,for x1>0,(1+m−R−)x1,for x1<0.\begin{aligned} \varphi(x_1)&=\left(1+\frac{m}{R}\right)^{x_1}\\ &=\begin{cases} \left(1+\frac{m_+}{R_+}\right)^{x_1}, & \text{for }x_1>0,\\ \left(1+\frac{m_-}{R_-}\right)^{x_1}, & \text{for }x_1<0. \end{cases} \end{aligned}

This corresponds to a localized mode with localization length

ξ±=−1ln⁡(1−∣m±∣R±)(54)\xi_{\pm} = \frac{-1}{\ln\left(1-\frac{\lvert m_{\pm}\rvert}{R_{\pm}}\right)} \tag*{(54)}

to either side of the domain wall.

References

  1. [1]J. C. Y. Teo and C. L. Kane, Topological defects and gapless modes in insulators and superconductors, Phys. Rev. B 82, 115120 (2010).DOI
  2. [2]R. Jackiw and C. Rebbi, Solitons with fermion number 1, Phys. Rev. D 13, 3398 (1976).
  3. [3]R. Jackiw and P. Rossi, Zero modes of the vortex-fermion system, Nuclear Physics B 190, 681 (1981).
  4. [4]C.-Y. Hou, C. Chamon, and C. Mudry, Electron fractionalization in two-dimensional graphene-like structures, Phys. Rev. Lett. 98, 186809 (2007).
  5. [5]G. Moore and N. Read, Nonabelions in the fractional quantum hall effect, Nuclear Physics B 360, 362 (1991).DOI
  6. [6]N. Read and D. Green, Paired states of fermions in two dimensions with breaking of parity and time-reversal symmetries and the fractional quantum hall effect, Phys. Rev. B 61, 10267 (2000).DOI
  7. [7]D. A. Ivanov, Non-abelian statistics of half-quantum vortices in p-wave superconductors, Phys. Rev. Lett. 86, 268 (2001).DOI
  8. [8]P. Fendley, Parafermionic edge zero modes in Zn-invariant spin chains, Journal of Statistical Mechanics: Theory and Experiment 2012, P11020 (2012).DOI
  9. [9]P. Fendley, Strong zero modes and eigenstate phase transitions in the XYZ/interacting majorana chain, Journal of Physics A: Mathematical and Theoretical 49, 30LT01 (2016).DOI
  10. [10]J. Cayssol, B. Dóra, F. Simon, and B. Moessner, Floquet topological insulators, physica status solidi (RRL) - Rapid Research Letters 7, 101 (2013).DOI
  11. [11]M. S. Rudner and N. H. Lindner, Band structure engineering and non-equilibrium dynamics in floquet topological insulators, Nature Reviews Physics 2, 229 (2020).DOI
  12. [12]K. Sacha and J. Zakrzewski, Time crystals: a review, Reports on Progress in Physics 81, 016401 (2017).DOI
  13. [13]D. V. Else, C. Monroe, C. Nayak, and N. Y. Yao, Discrete time crystals, Annual Review of Condensed Matter Physics 11, 467 (2020).DOI
  14. [14]V. Khemani, R. Moessner, and S. L. Sondhi, A brief history of time crystals (2019).arxiv.org/abs/1910.10745
  15. [15]J. Zhang, P. W. Hess, A. Kyprianidis, P. Becker, A. Lee, J. Smith, G. Pagano, I.-D. Potirniche, A. C. Potter, A. Vishwanath, N. Y. Yao, and C. Monroe, Observation of a discrete time crystal, Nature 543, 217 (2017).DOI
  16. [16]S. Choi, J. Choi, R. Landig, G. Kucsko, H. Zhou, J. Isoya, F. Jelezko, S. Onda, H. Sumiya, V. Khemani, C. von Keyserlingk, N. Y. Yao, E. Demler, and M. D. Lukin, Observation of discrete time-crystalline order in a disordered dipolar many-body system, Nature 543, 221 (2017).DOI
  17. [17]A. Kyprianidis, F. Machado, W. Morong, P. Becker, K. S. Collins, D. V. Else, L. Feng, P. W. Hess, C. Nayak, G. Pagano, N. Y. Yao, and C. Monroe, Observation of a prethermal discrete time crystal, Science 372, 1192 (2021).DOI
  18. [18]X. Mi, M. Ippoliti, C. Quintana, A. Greene, Z. Chen, J. Gross, F. Arute, K. Arya, J. Atalaya, R. Bab- bush, J. C. Bardin, J. Basso, A. Bengtsson, A. Bilmes, A. Bourassa, L. Brill, M. Broughton, B. B. Buck- ley, D. A. Buell, B. Burkett, N. Bushnell, B. Chiaro, R. Collins, W. Courtney, D. Debroy, S. Demura, A. R. Derk, A. Dunworth, D. Eppens, C. Erickson, E. Farhi, A. G. Fowler, B. Foxen, C. Gidney, M. Giustina, M. P. Harrigan, S. D. Harrington, J. Hilton, A. Ho, S. Hong, T. Huang, A. Huff, W. J. Huggins, L. B. Ioffe, S. V. Isakov, J. Y. J. Ivland, E. Jeffrey, Z. Jiang, C. Jones, D. Kafri, T. Khattar, S. Kim, A. Kitaev, P. V. Klimov, A. N. Korotkov, F. Kostritsa, D. Landhuis, P. Laptev, J. Lee, K. Lee, A. Locharla, E. Lucero, O. Martin, J. R. McClean, T. McCourt, M. McEwen, K. C. Miao, M. Mohseni, S. Montazeri, W. Mruzckiewicz, O. Naaman, M. Neeley, C. Neill, M. Newman, M. Y. Niu, T. E. O’Brien, A. Opremcak, E. Ostby, B. Pato, A. Petukhov, N. C. Rubin, D. Sank, K. J. Satzinger, V. Shvarts, Y. Su, D. Strain, M. Szalay, M. D. Trevithick, B. Villalonga, T. White, Z. J. Yao, P. Yeh, J. Yoo, A. Zalcmann, H. Neven, S. Boixo, V. Smelyanskiy, A. Megrant, J. Kelly, Y. Chen, S. L. Sondhi, R. Moessner, K. Kechedzhi, V. Khemani, and P. Roushan, Time-crystalline eigenstate order on a quantum processor, Nature 601, 531 (2021).DOI
  19. [19]P. T. Dumitrescu, J. G. Bohnet, J. P. Gaebler, A. Hankin, D. Hayes, A. Kumar, B. Nevenhuis, R. Vasseur, and A. C. Potter, Dynamical topological phase realized in a trapped-ion quantum simulator, Nature 607, 463 (2022).DOI
  20. [20]X. Zhang, W. Jiang, J. Deng, K. Wang, J. Chen, P. Zhang, W. Ren, H. Dong, S. Xu, Y. Gao, F. Jin, X. Zhu, Q. Guo, H. Li, C. Song, A. V. Gorshkov, T. Iadecola, F. Liu, Z.-X. Gong, Z. Wang, D.-L. Deng, and H. Wang, Digital quantum simulation of floquet symmetry-protected topological phases, Nature 607, 468 (2022).
  21. [21]J. W. McIver, B. Schulte, F.-U. Stein, T. Matsuyama, G. Jotzu, G. Meier, and A. Cavalleri, Light-induced anomalous hall effect in graphene, Nature Physics 16, 38 (2019).
  22. [22]M. Thakurathi, A. A. Patel, D. Sen, and A. Dutta, Floquet generation of majorana end modes and topological invariants, Phys. Rev. B 88, 155133 (2013).DOI
  23. [23]M. S. Rudner, N. H. Lindner, E. Berg, and M. Levin, Anomalous edge states and the bulk-edge correspondence for periodically driven two-dimensional systems, Phys. Rev. X 3, 031005 (2013).arxiv.org/abs/1212.3324
  24. [24]V. Khemani, A. Lazarides, R. Moessner, and S. L. Sondhi, Phase structure of driven quantum systems, Phys. Rev. Lett. 116, 250401 (2016).DOI
  25. [25]D. V. Else, B. Bauer, and C. Nayak, Floquet time crystals, Phys. Rev. Lett. 117, 090402 (2016).DOI
  26. [26]D. V. Else and C. Nayak, Classification of topological phases in periodically driven interacting systems, Phys. Rev. B 93, 201103 (2016).DOI
  27. [27]C. W. von Keyserlingk and S. L. Sondhi, Phase structure of one-dimensional interacting floquet systems. i. abelian symmetry-protected topological phases, Phys. Rev. B 93, 245145 (2016).
  28. [28]H. B. Nielsen and M. Ninomiya, Absence of Neutrinos on a Lattice. 1. Proof by Homotopy Theory, Nucl. Phys. B 185, 20 (1981), [Erratum: Nucl.Phys.B 195, 541 (1982)].
  29. [29]H. B. Nielsen and M. Ninomiya, Absence of Neutrinos on a Lattice. 2. Intuitive Topological Proof, Nucl. Phys. B 193, 173 (1981).
  30. [30]D. B. Kaplan, A Method for simulating chiral fermions on the lattice, Phys. Lett. B 288, 342 (1992), arXiv:hep-lat/9206013.DOI
  31. [31]Y. Shamir, Chiral fermions from lattice boundaries, Nucl. Phys. B 406, 90 (1993), arXiv:hep-lat/9303005.DOI
  32. [32]P. H. Ginsparg and K. G. Wilson, A remnant of chiral symmetry on the lattice, Phys. Rev. D 25, 2649 (1982).DOI
  33. [33]H. Neuberger, Exactly massless quarks on the lattice, Phys. Lett. B 417, 141 (1998), arXiv:hep-lat/9707022.DOI
  34. [34]H. Neuberger, More about exactly massless quarks on the lattice, Phys. Lett. B 427, 353 (1998), arXiv:hep-lat/9801031.DOI
  35. [35]J. Kogut and L. Susskind, Hamiltonian formulation of wilson’s lattice gauge theories, Phys. Rev. D 11, 395 (1975).DOI
  36. [36]X.-Q. Sun, M. Xiao, T. Bzdušek, S.-C. Zhang, and S. Fan, Three-dimensional chiral lattice fermion in floquet systems, Phys. Rev. Lett. 121, 196401 (2018).DOI
  37. [37]M. DeMarco and X.-G. Wen, A single right-moving free fermion mode on an ultra-local 1 + 1d spacetime lattice (2018), arXiv:1805.03663 [hep-lat].arxiv.org/abs/1805.03663
  38. [38]Unless indicated otherwise, we restrict our focus in this work to single-particle spectral properties rather than many-body ones.
  39. [39]D. Bödeker, G. D. Moore, and K. Rummukainen, Chern-simons number diffusion and hard thermal loops on the lattice, Phys. Rev. D 61, 056003 (2000).DOI
  40. [40]J. Ambjørn, T. Askgaard, H. Porter, and M. E. Shaposhnikov, Sphaleron transitions and baryon asymmetry: A Numerical real time analysis, Nucl. Phys. B 353, 346 (1991).DOI
  41. [41]G. Aarts and J. Smit, Real time dynamics with fermions on a lattice, Nucl. Phys. B 555, 355 (1999), arXiv:hep-ph/9812413.DOI
  42. [42]Z.-G. Mou, P. M. Safin, and A. Tranberg, Ensemble fermions for electroweak dynamics and the fermion pre-heating temperature, JHEP 11, 097, arXiv:1307.7924 [hep-ph].
  43. [43]L. H. Karsten, Lattice Fermions in Euclidean Space-time, Phys. Lett. B 104, 315 (1981).DOI
  44. [44]Typical lattice Monte Carlo studies of quantum field theories are performed in Euclidean space-time. For real-time formulations path integrals for quantum field theories and quantum mechanical systems see [58–63].
  45. [45]W. P. Su, J. R. Schrieffer, and A. J. Heeger, Solitons in polyacetylene, Phys. Rev. Lett. 42, 1698 (1979).
  46. [46]A. C. Potter, T. Morimoto, and A. Vishwanath, Classification of interacting topological floquet phases in one dimension, Phys. Rev. X 6, 041001 (2016).DOI
  47. [47]See Supplemental Material.
  48. [48]T. Iadecola, S. Sen, and L. Sivertsen, in preparation.
  49. [49]H. C. Po, L. Fidkowski, A. Vishwanath, and A. C. Potter, Radical chiral floquet phases in a periodically driven kitaev model and beyond, Phys. Rev. B 96, 245116 (2017).DOI
  50. [50]T. Kitagawa, E. Berg, M. Rudner, and E. Demler, Topological characterization of periodically driven quantum systems, Phys. Rev. B 82, 235114 (2010).DOI
  51. [51]F. Nathan and M. S. Rudner, Topological singularities and the general classification of floquet–bloch systems, New Journal of Physics 17, 125014 (2015).DOI
  52. [52]D. Carpentier, P. Delplace, M. Fruchart, and K. Gawedzki, Topological index for periodically driven time-reversal invariant 2d systems, Phys. Rev. Lett. 114, 106806 (2015).pubmed.ncbi.nlm.nih.gov/25815957
  53. [53]M. Fruchart, Complex classes of periodically driven topological lattice systems, Phys. Rev. B 93, 115429 (2016).DOI
  54. [54]M. F. L. Golterman, K. Jansen, and D. B. Kaplan, Chern-Simons currents and chiral fermions on the lattice, Phys. Lett. B 301, 219 (1993), arXiv:hep-lat/9209003.DOI
  55. [55]P. Ponte, A. Chandran, Z. Papić, and D. A. Abanin, Periodically driven ergodic and many-body localized quantum systems, Annals of Physics 353, 196 (2015).DOI
  56. [56]L. D’Alessio and M. Rigol, Long-time behavior of isolated periodically driven interacting lattice systems, Phys. Rev. X 4, 041048 (2014).DOI
  57. [57]D. V. Else, B. Bauer, and C. Nayak, Prethermal phases of matter protected by time-translation symmetry, Phys. Rev. X 7, 011026 (2017).DOI

Paper details

vlnc:W_6euhs6v8
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents