This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Topological Classification of Gapped/Gap-preserving Rational Space-Time Crystal Systems

Shu-Xuan Wang [email protected] Department of Modern Physics, University of Science and Technology of China, Hefei, 230026, China    Shaolong Wan [email protected] Department of Modern Physics, University of Science and Technology of China, Hefei, 230026, China
Abstract

The traditional systems researched in condensed matter physics always have spatial translation symmetry. However for space-time crystal systems, the spatial translation symmetry is no longer preserved and the lattice potential have space-time translation symmetry instead. We show that a rational space-time crystal system is equal to a traditional floquet system. Then, we find a way to solve the floquet equation analytically and construct an effective Hamiltonian of the rational space-time crystal system. By this effective Hamiltonian, we obtain the topological classification of gapped and gap-preserving rational space-time crystal systems. Our works reveal the correlation between space-time crystal systems and floquet systems and give a systematic method to explore the properties of rational space-time crystal systems.

I Introduction

Topological phase is an important topic in condensed matter physics [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22]. There are numerous novel phenomena about topological phase. For a topological non-trivial system, there exist some states localized at boundaries, whose energies are isolated in the spectrum, which are called topological states [2, 16]. The topological states can apply conductance even if the bulk of the system is gapped, and this phenomenon can be described by an effective field theory, topological field theory [2, 3]. If the system has some crystalline symmetries, the topological states may no longer be localized at the boundary, but localized at the corner, these topological states are called high-order topological states [7, 8, 9]. Theoretically, for a given system, we can use the topological invariants to describe whether this system is topological non-trivial or not [2, 3, 16], and a topological non-trivial system can not be transformed to a topological trivial system adiabatically without an energy gap closing. Another fascinating direction is classification of the topological phase. For systems with different internal or crystalline symmetries, the topological invariants used to describe these topological system may be different. For traditional Hermitian and non-Hermitian topological systems, the classification and topological invariants can be given by KK theory according to the symmetries and the type of the energy gap of the systems [1, 22].

In the past, the topological systems researched always have the spatial translation symmetry. Recently, space-time crystal system has been noticed [23, 24]. Spatial translation symmetry is no longer preserved in space-time crystal systems. Instead, space-time translation symmetry is kept in these systems. Hence, Bloch theorem is invalid and generalized floquet-Bloch theorem is used in space-time crystal systems  [23]. Similar to traditional systems, it is natural to expect that space-time crystal systems can have non-trivial topological phase, which can be described by topological invariants.

In this article, we show that a rational space-time crystal system can be transformed to a traditional floquet system, which respect the spatial transition symmetry. Solving the floquet equation, we can obtain the quasi-spectrum of the rational space-time crystal system and construct an effective Hamiltonian. Then, topological classification of rational space-time crystal systems is reduced to the topological classification of the effective Hamiltonian. Thus, by researching the effective Hamiltonian, we obtain the topological classification of gapped/gap-preserving rational space-time crystal systems.

This article is organized as follows. In Sec.𝐈𝐈\mathbf{II}, we give the definition of rational space-time crystal system and show that a rational space-time crystal system is equal to a traditional floquet system. In Sec.𝐈𝐈𝐈\mathbf{III}, we solve the floquet equation and construct the effective Hamiltonian. In Sec.𝐈𝐕\mathbf{IV}, we give the transformation of the effective Hamiltonian under internal symmetries. In Sec.𝐕\mathbf{V}, we discuss the type of energy gap about Hermitian and non-Hermitian rational space-time crystal systems and give the topological classification of gapped/gap-preserving rational space-time crystal systems. In Sec.𝐕𝐈\mathbf{VI}, we give examples of topological classification about Hermitian and non-Hermitian space-time crystal system. Conclusion and discussion is given in Sec.𝐕𝐈𝐈\mathbf{VII}.

II Space-Time translation symmetry and Hamiltonian of Rational Space-Time Crystal

For (d+1)(d+1)-dimensional space-time crystal, the lattice potential may not satisfy the translation symmetry. Instead, it satisfies space-time translation symmetry [23, 24]:

V(𝐫,t)=V(𝐫+𝐬i,t+τi)V(\mathbf{r},t)=V(\mathbf{r}+\mathbf{s}_{i},t+\tau_{i}) (1)

for i=0,1,,di=0,1,\cdots,d. Since 𝐬i\mathbf{s}_{i} is a dd-dimensional spatial vector, only dd vectors of {𝐬i}\{\mathbf{s}_{i}\} are independent. Hence, there exist mim_{i}\in\mathbb{R} for i=0,1,2,,di=0,1,2,\cdots,d, such that111at least one number of {mi}\{m_{i}\} is nonzero.

i=0dmi𝐬i=0.\sum_{i=0}^{d}m_{i}\mathbf{s}_{i}=0. (2)

We call the space-time crystal system is rational space-time crystal system if mim_{i}\in\mathbb{Z} for i=0,1,2,,di=0,1,2,\cdots,d and τiτj\frac{\tau_{i}}{\tau_{j}} is rational number for all i,j=0,1,2,,di,j=0,1,2,\cdots,d. For the rational space-time crystal system, we can define T=i=0dmiτiT=\sum_{i=0}^{d}m_{i}\tau_{i}. Then we have

V(𝐫,t)=V(𝐫+i=0dmi𝐬i,t+i=0dmiτi)=V(𝐫,t+T).V(\mathbf{r},t)=V(\mathbf{r}+\sum_{i=0}^{d}m_{i}\mathbf{s}_{i},t+\sum_{i=0}^{d}m_{i}\tau_{i})=V(\mathbf{r},t+T). (3)

It means the potential of the rational space-time crystal system has a period, TT, about time. Thus, the discrete space-time translation basis vectors can be written as

𝐚0=(𝟎,T),𝐚i=(𝐬i,αiT),\mathbf{a}_{0}=(\mathbf{0},-T),\qquad\mathbf{a}_{i}=(\mathbf{s}_{i},\alpha_{i}T), (4)

with αi=τiTmod 1=(j=0dmjτjτi)1mod 1(12,12]\alpha_{i}=\frac{\tau_{i}}{T}\ \mathrm{mod}\ 1=(\sum_{j=0}^{d}m_{j}\frac{\tau_{j}}{\tau_{i}})^{-1}\ \mathrm{mod}\ 1\in(-\frac{1}{2},\frac{1}{2}] for i=1,2,,di=1,2,\cdots,d. Since mjm_{j} is integer and τjτi\frac{\tau_{j}}{\tau_{i}} is rational for all i=1,2,,di=1,2,\cdots,d and j=0,1,2,,dj=0,1,2,\cdots,d, αi\alpha_{i} is rational number for i=1,2,,di=1,2,\cdots,d. The reciprocal vectors corresponding to the discrete space-time translation basis vectors are

𝐛0=(i=1dαi𝐠i,Ω),𝐛i=(𝐠i,0),\mathbf{b}_{0}=(\sum_{i=1}^{d}\alpha_{i}\mathbf{g}_{i},\Omega),\qquad\mathbf{b}_{i}=(\mathbf{g}_{i},0), (5)

with Ω=2πT\Omega=\frac{2\pi}{T}, 𝐠i𝐬j=2πδi,j\mathbf{g}_{i}\cdot\mathbf{s}_{j}=2\pi\delta_{i,j}. Hence, 𝐚i𝐛j=2πδi,j\mathbf{a}_{i}\cdot\mathbf{b}_{j}=2\pi\delta_{i,j} 222We take 𝐠=diag{𝟏d×d,1}\mathbf{g}=diag\{\mathbf{1}_{d\times d},-1\} as the space-time metric to make it consistent with Refs.[23],[24] . Then, we take {𝐬i}\{\mathbf{s}_{i}\} as lattice vectors, and the tight-binding Hamiltonian of the system is [24]

H=n=einΩtHn,\displaystyle H=\sum_{n=-\infty}^{\infty}e^{-in\Omega t}H_{n}, (6)
Hn=𝐑i,𝐑jeinδ𝐤𝐑it𝐑i𝐑j(n)c𝐑ic𝐑j,\displaystyle H_{n}=\sum_{\mathbf{R}_{i},\mathbf{R}_{j}}e^{in\mathbf{\delta k}\cdot\mathbf{R}_{i}}t_{\mathbf{R}_{i}-\mathbf{R}_{j}}^{(n)}c_{\mathbf{R}_{i}}^{\dagger}c_{\mathbf{R}_{j}}, (7)

where δ𝐤=i=1dαi𝐠i\mathbf{\delta k}=\sum_{i=1}^{d}\alpha_{i}\mathbf{g}_{i}, c𝐑ic_{\mathbf{R}_{i}} and c𝐑ic_{\mathbf{R}_{i}}^{\dagger} are creation and annihilation operator about the site locating at 𝐑i\mathbf{R}_{i}, and t𝐑i𝐑j(n)t_{\mathbf{R}_{i}-\mathbf{R}_{j}}^{(n)} is the hopping amplitude between site 𝐑i\mathbf{R}_{i} and 𝐑j\mathbf{R}_{j}, which only depends on the value of 𝐑i𝐑j\mathbf{R}_{i}-\mathbf{R}_{j}.

Since αi\alpha_{i} is rational number, we can take αi=piqi\alpha_{i}=\frac{p_{i}}{q_{i}}, where pip_{i} and qiq_{i} are coprime for i=1,2,,di=1,2,\cdots,d. In this case, we can treat i=1dqi\prod_{i=1}^{d}q_{i} sites as i=1dqi\prod_{i=1}^{d}q_{i} orbitals in a unit cell. Hence, the Hamiltonian can be rewritten as (See Appendix A for details):

H=n=einΩt𝐑~i,𝐑~j𝐜𝐑~i𝐌𝐑~i𝐑~j(n)𝐜𝐑~j=n=einΩtH~n,\begin{split}H&=\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\sum_{\tilde{\mathbf{R}}_{i},\tilde{\mathbf{R}}_{j}}\mathbf{c}_{\tilde{\mathbf{R}}_{i}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}_{i}-\tilde{\mathbf{R}}_{j}}^{(n)}\mathbf{c}_{\tilde{\mathbf{R}}_{j}}\\ &=\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n}\end{split}, (8)

where 𝐌ij(n)\mathbf{M}_{i-j}^{(n)} is a (i=1dqi)×(i=1dqi(\prod_{i=1}^{d}q_{i})\times(\prod_{i=1}^{d}q_{i}) matrix only dependent on 𝐑~i𝐑~j\tilde{\mathbf{R}}_{i}-\tilde{\mathbf{R}}_{j} and 𝐜\mathbf{c} (𝐜\mathbf{c}^{\dagger}) is a (i=1dqi)(\prod_{i=1}^{d}q_{i})-dimensional column vector composed by annihilation (creation) operators.

The spatial translation symmetry is respected in Eq.(8). Hence, a rational space-time crystal system is equal to a floquet system.

III Qusi-spectrum and effect Hamiltonian of Rational Space-time crystal system

We start from the Hamiltonian given in Eq.(8). The Schrödinger equation about the Hamiltonian, HH, is

H|ψ(t)=it|ψ(t).H|\psi(t)\rangle=i\partial_{t}|\psi(t)\rangle. (9)

Since HH has a period TT, the solution of Eq.(9) has the form[27, 18, 28]

|ψ(t)=eiωt|uω(t),|\psi(t)\rangle=e^{-i\omega t}|u_{\omega}(t)\rangle, (10)

where ω\omega is the quasienergy and |uω(t)|u_{\omega}(t)\rangle is quasistate with period TT. Since

|uω(t)=|uω(t+T),|u_{\omega}(t)\rangle=|u_{\omega}(t+T)\rangle, (11)

|uω(t)|u_{\omega}(t)\rangle can be decomposed as

|uω(t)=m=eimΩt|um,ω,|u_{\omega}(t)\rangle=\sum_{m=-\infty}^{\infty}e^{-im\Omega t}|u_{m,\omega}\rangle, (12)

with Ω=2πT\Omega=\frac{2\pi}{T}. Substituting Eqs.(8),(10),(12) into Eq.(9), we obtain

m,n(H~nei(m+n)Ωtδm,neimΩtmΩ)|um,ω=ωmeimΩt|um,ω.\sum_{m,n}(\tilde{H}_{n}e^{-i(m+n)\Omega t}-\delta_{m,n}e^{-im\Omega t}m\Omega)|u_{m,\omega}\rangle\\ =\omega\sum_{m}e^{-im\Omega t}|u_{m,\omega}\rangle. (13)

Then, we can get the equation about |um,ω|u_{m,\omega}\rangle,

n=H~mn|un,ωmΩ|um,ω=ω|um,ω.\sum_{n=-\infty}^{\infty}\tilde{H}_{m-n}|u_{n,\omega}\rangle-m\Omega|u_{m,\omega}\rangle=\omega|u_{m,\omega}\rangle. (14)

The eigenvalue problem about Eq.(14) has a property, if ω\omega is the quasienergy of |um,ω|u_{m,\omega}\rangle, ω+nΩ\omega+n\Omega is the quasienergy of |um+n,ω|u_{m+n,\omega}\rangle [19]. We call the set of quasienergies of |um,ω|u_{m,\omega}\rangle the mmth sector of the quasi-spectrum.

We define an enlarged Hamiltonian,

Henlarged=(H~0ΩH~1H~2H~1H~0H~1H~2H~1H~0+Ω),H_{enlarged}=\begin{pmatrix}\ddots&&&&\\ &\tilde{H}_{0}-\Omega&\tilde{H}_{-1}&\tilde{H}_{-2}&\\ &\tilde{H}_{1}&\tilde{H}_{0}&\tilde{H}_{-1}&\\ &\tilde{H}_{2}&\tilde{H}_{1}&\tilde{H}_{0}+\Omega&\\ &&&&\ddots\end{pmatrix}, (15)

and an enlarged vector,

𝐮enlargedT(ω)=(,|u1,ωT,|u0,ωT,|u1,ωT,).\mathbf{u}_{enlarged}^{T}(\omega)=(\cdots,|u_{-1,\omega}\rangle^{T},|u_{0,\omega}\rangle^{T},|u_{1,\omega}\rangle^{T},\cdots). (16)

Then, Eq,(14) can be written as

Henlarged𝐮enlarged(ω)=ω𝐮enlarged(ω),H_{enlarged}\mathbf{u}_{enlarged}(\omega)=\omega\mathbf{u}_{enlarged}(\omega), (17)

and the set of eigenvalues of Eq.(17) is the 0th sector of the quasi-spectrum. Since the index of sector, mm, varies from -\infty to \infty in Eq.(16), the boundary condition of HenlargedH_{enlarged} about sector is infinite boundary condition (IBC) and HenlargedH_{enlarged} is an infinite-dimensional matrix. Thus, we assume that

|um,ω=eimθ|u0,ω,|u_{m,\omega}\rangle=e^{-im\theta}|u_{0,\omega}\rangle, (18)

where θ[0,2π]\theta\in[0,2\pi]. Then, Eq.(17) can be reduced as

n=H~mnei(nm)θ|um,ω=(ω+mΩ)|um,ω.\sum_{n=-\infty}^{\infty}\tilde{H}_{m-n}e^{-i(n-m)\theta}|u_{m,\omega}\rangle=(\omega+m\Omega)|u_{m,\omega}\rangle. (19)

For the 0th sector,

n=H~neinθ|u0,ω=ω|u0,ω.\sum_{n=-\infty}^{\infty}\tilde{H}_{-n}e^{-in\theta}|u_{0,\omega}\rangle=\omega|u_{0,\omega}\rangle. (20)

Hence, we define an effective Hamiltonian

Heff(θ)=n=H~neinθ.H_{eff}(\theta)=\sum_{n=-\infty}^{\infty}\tilde{H}_{-n}e^{-in\theta}. (21)

With θ\theta varying from 0 to 2π2\pi, all eigenvalues of Heff(θ)H_{eff}(\theta) compose the 0th sector of the quasi-spectrum. In this work, we only consider the 0th sector, because 0th sector gives the physical spectrum (Appendix B).

IV Internal Symmetry of the Effective Hamiltonian

Now, we discuss the internal symmetries of the system. We use 𝒯\mathcal{T}, 𝒞\mathcal{C} and 𝒮\mathcal{S} to denote the time-reversal operator, particle-hole operator and chiral operator. Here, we consider fermion systems. If the system has time-reversal symmetry (TRS), the effective Hamiltonian has the relation

𝒯Heff(θ)𝒯1=Heff(θ),\mathcal{T}H_{eff}(\theta)\mathcal{T}^{-1}=H_{eff}(\theta), (22)

Under periodic boundary condition (PBC) about the spatial dimension, we can transform the Hamiltonian to the momentum space, and the relation Eq.(22) becomes

𝒯Heff(𝐤,θ)𝒯1=Heff(𝐤,θ),\mathcal{T}H_{eff}(\mathbf{k},\theta)\mathcal{T}^{-1}=H_{eff}(-\mathbf{k},\theta), (23)

where

Heff(𝐤,θ)=n=H~n(𝐤)einθ.H_{eff}(\mathbf{k},\theta)=\sum_{n=-\infty}^{\infty}\tilde{H}_{-n}(\mathbf{k})e^{-in\theta}. (24)

For particle-hole symmetry (PHS) and chiral symmetry (CS), the relations are (See Appendix C for details)

𝒞Heff(𝐤,θ)𝒞1=Heff(𝐤,θ)\displaystyle\mathcal{C}H_{eff}(\mathbf{k},\theta)\mathcal{C}^{-1}=-H_{eff}(-\mathbf{k},\theta) (25)
𝒮Heff(𝐤,θ)𝒮1=Heff(𝐤,θ).\displaystyle\mathcal{S}H_{eff}(\mathbf{k},\theta)\mathcal{S}^{-1}=-H_{eff}(\mathbf{k},\theta). (26)

These relations can be generalized to the internal symmetries of non-Hermitian cases [22] (Appendix C). These relations are

𝒯+Heff(𝐤,θ)𝒯+1=Heff(𝐤,θ)\displaystyle\mathcal{T}_{+}H_{eff}(\mathbf{k},\theta)\mathcal{T}^{-1}_{+}=H_{eff}(-\mathbf{k},\theta) (27)
𝒞Heff(𝐤,θ)𝒞1=Heff(𝐤,θ)\displaystyle\mathcal{C}_{-}H_{eff}(\mathbf{k},\theta)\mathcal{C}^{-1}_{-}=-H_{eff}(-\mathbf{k},\theta) (28)
𝒞+Heff(𝐤,θ)𝒞+1=Heff(𝐤,θ)\displaystyle\mathcal{C}_{+}H_{eff}^{\dagger}(\mathbf{k},\theta)\mathcal{C}^{-1}_{+}=H_{eff}(-\mathbf{k},\theta) (29)
𝒯Heff(𝐤,θ)𝒯1=Heff(𝐤,θ)\displaystyle\mathcal{T}_{-}H_{eff}^{\dagger}(\mathbf{k},\theta)\mathcal{T}^{-1}_{-}=-H_{eff}(-\mathbf{k},\theta) (30)
𝒮Heff(𝐤,θ)𝒮1=Heff(𝐤,θ)\displaystyle\mathcal{S}H_{eff}(\mathbf{k},\theta)\mathcal{S}^{-1}=-H_{eff}(\mathbf{k},\theta) (31)
ΓHeff(𝐤,θ)Γ1=Heff(𝐤,θ)\displaystyle\Gamma H_{eff}^{\dagger}(\mathbf{k},\theta)\Gamma^{-1}=-H_{eff}(\mathbf{k},\theta) (32)
ηHeff(𝐤,θ)η1=Heff(𝐤,θ).\displaystyle\eta H_{eff}^{\dagger}(\mathbf{k},\theta)\eta^{-1}=H_{eff}(\mathbf{k},\theta). (33)

Here, 𝒯+\mathcal{T}_{+} and 𝒞+\mathcal{C}_{+} are operators of TRS and TRS, 𝒞\mathcal{C}_{-} and 𝒯\mathcal{T}_{-} are operators of PHS and PHS, 𝒮\mathcal{S} is the operator of CS, Γ\Gamma is the operator of sublattice symmetry (SLS), and η\eta is the operator of pseudo-Hermitian symmetry.

Under internal symmetry transformations, θ\theta is always invariant. Hence, we treat θ\theta as a parameter but not a momentum-like variable even if θ\theta has a period, 2π2\pi.

In addition, it is worth to point out that TRS is not independent of the space-time translation symmetry. TRS gives constraints about space-time translation symmetry. A breif discussion about these constraints is given in Appendix D.

V Topological Classification of According to the Effective Hamiltonian

Since Heff(θ)H_{eff}(\theta) can give the physical spectrum of the rational space-time crystal system, the topological classification of the rational space-time crystal system can be reduced to classifying Heff(θ)H_{eff}(\theta).

V.1 Hermitian Case

For Hermitian rational space-time crystal system, the Hamiltonian HH in Eq.(8) is Hermitian. Thus, H~n=H~n\tilde{H}_{n}=\tilde{H}_{-n}^{\dagger}. According to Eq.(21), the effective Hamiltonian Heff(θ)H_{eff}(\theta) corresponding to HH is also Hermitian.

There are 33 types for Hermitian rational space-time crystal systems:

Definition V.1.1.

Gapped — The Hermitian effective Hamiltonian under PBC, Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), is gapped for all θ[0,2π]\theta\in[0,2\pi].

Definition V.1.2.

Half-gapped — The Hermitian effective Hamiltonian under PBC, Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), is gapless for some specific values of θ\theta.

Definition V.1.3.

Gapless — The Hermitian effective Hamiltonian under PBC, Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), is gapless for all θ[0,2π]\theta\in[0,2\pi].

Here, we discuss the topological classification of the gapped systems. Since we have obtained the representation of internal symmetries on the effective Hamiltonian, a Hermitian Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) can be classified by tenfold AZ class [1]. Since θ\theta is not a momentum-like variable, we treat Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) as a dd-dimensional Hamiltonian if 𝐤\mathbf{k} is a dd-dimensional momentum. Thus, the topological invariant about Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta), given by the traditional tenfold classification [1], contains θ\theta. For gapped system, the gap of Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) will not be closed when θ\theta varies from 0 to 2π2\pi. This means that if we treat Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) as a dd-dimensional Hamiltonian with a parameter θ\theta, Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) experience no topological transition when θ\theta changes from 0 to 2π2\pi. Hence, 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta) is well defined for all θ[0,2π]\theta\in[0,2\pi], and for arbitrary θ1,θ2[0,2π]\theta_{1},\theta_{2}\in[0,2\pi], 𝐢𝐧𝐯(θ1)=𝐢𝐧𝐯(θ2)\mathbf{inv}(\theta_{1})=\mathbf{inv}(\theta_{2}). Thus, 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta) is a good topological invariant for gapped rational Hermitian space-time crystal system.

V.2 Non-Hermitian case

For non-Hermitian rational space-time crystal system, the effective Hamiltonian is no longer Hermitian. For a given θ0[0,2π]\theta_{0}\in[0,2\pi], according to Ref.[22], there are 33 types of the complex energy gap, which are line gap, point gap and gapless. Hence, there are 33 types for non-Hermitian rational space-time systems:

Definition V.2.1.

Gap-preserving — The non-Hermitian effective Hamiltonian, under PBC Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), has line gap for all θ[0,2π]\theta\in[0,2\pi], or Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) has point gap for all θ[0,2π]\theta\in[0,2\pi].

Definition V.2.2.

Gap-breaking — For non-Hermitian effective Hamiltonian under PBC, Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), there exist θ1,θ2[0,2π]\theta_{1},\theta_{2}\in[0,2\pi] and θ1θ2\theta_{1}\not=\theta_{2}, such that the gap type of Heff(𝐤,θ1)H_{eff}(\mathbf{k},\theta_{1}) and Heff(𝐤,θ2)H_{eff}(\mathbf{k},\theta_{2}) are different.

Definition V.2.3.

Gapless — The non-Hermitian effective Hamiltonian under PBC, Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), is gapless for all θ[0,2π]\theta\in[0,2\pi].

In this subsection, we discuss the topological classification of gap-preserving systems. Similar to the Hermitian case, the non-Hermitian rational space-time crystal system can be classified by 38-fold classification of the effective Hamiltonian Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) and the type of energy gap according to Ref.[22]. Thus, for Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) with dd-dimensional 𝐤\mathbf{k}, the topological invariant of Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta), 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta), is the topological invariant given by Ref.[22] for dd-dimensional non-Hermitian system, which contains a parameter θ\theta. For gap-preserving system, the energy gap is always preserved when θ\theta changes form 0 to 2π2\pi. This means, if we treat Heff(𝐤,θ)H_{eff}(\mathbf{k},\theta) as dd-dimensional Hamiltonian with a parameter θ\theta, this Hamiltonian will not experience a topological transition when θ\theta varies from 0 to 2π2\pi. Thus, 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta) is well defined for all θ[0,2π]\theta\in[0,2\pi] for gap-preserving non-Hermitian rational space-time crystal system, and the value of 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta) is independent of θ\theta. We choose 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta) as the topological invariant of gap-preserving non-Hermitian rational space-time crystal system.

VI Examples

VI.1 Hermitian case

Consider a (1+1)(1+1)-dimensional space-time crystal system with two orbitals. The lattice potential has the form

V(r,t)=V0(r)+V1(r,t),V(r,t)=V_{0}(r)+V_{1}(r,t), (34)

with

V0(r)=V0(r+a),V1(r,t)cos(πar2πTt),V_{0}(r)=V_{0}(r+a),\qquad V_{1}(r,t)\propto cos(\frac{\pi}{a}r-\frac{2\pi}{T}t), (35)

where aa is the lattice constant. Thus, this system has space-time translation symmetry,

V(r,t)=V(r+a,t+T2),V(r,t)=V(r+2a,t+T).V(r,t)=V(r+a,t+\frac{T}{2}),\quad V(r,t)=V(r+2a,t+T). (36)

Apparently, this system is a rational space-time crystal system. We choose

𝐚0=(0,T),𝐚1=(a,T2)\mathbf{a}_{0}=(0,-T),\qquad\mathbf{a}_{1}=(a,\frac{T}{2}) (37)

as basis of discrete space-time translation symmetry. The reciprocal vectors of them are 22footnotemark: 2

𝐛0=(πa,2πT),𝐛1=(2πa,0).\mathbf{b}_{0}=(\frac{\pi}{a},\frac{2\pi}{T}),\qquad\mathbf{b}_{1}=(\frac{2\pi}{a},0). (38)

Now, we take a=1a=1 and T=2π5T=\frac{2\pi}{5}. Then, we have δk=π\delta k=\pi and Ω=5\Omega=5 for this system. The tight-binding Hamiltonian of this system is given by Eqs.(6) and (7). According to Ref.[24], for such a lattice potential given by Eqs.(34) and (35), only H1H_{-1}, H0H_{0} and H1H_{1} are nonzero. Hence, we assume that,

H0=nt0(0)cn,Acn,B+t1(0)cn,Bcn+1,A+H.c.H_{0}=\sum_{n}t_{0}^{(0)}c_{n,A}^{\dagger}c_{n,B}+t_{1}^{(0)}c_{n,B}^{\dagger}c_{n+1,A}+H.c. (39)
H1=nt1(1)einπcn,Acn,B,H_{1}=\sum_{n}t_{1}^{(1)}e^{in\pi}c_{n,A}^{\dagger}c_{n,B}, (40)

and

H1=H1,H_{-1}=H_{1}^{\dagger}, (41)

where AA and BB are indexes of orbitals and n[1,L]n\in[1,L] is the index of site and LL is the length of the system. In matrix form,

H0=(0t0(0)t0(0)0t1(0)t1(0)0t0(0)t0(0)0t1(0)t1(0)0)2L×2L,H_{0}=\begin{pmatrix}0&t_{0}^{(0)}&&&&\\ t_{0}^{(0)}&0&t_{1}^{(0)}&&&\\ &t_{1}^{(0)}&0&t_{0}^{(0)}&&\\ &&t_{0}^{(0)}&0&t_{1}^{(0)}&\\ &&&t_{1}^{(0)}&0&\ddots\\ &&&&\ddots&\ddots\end{pmatrix}_{2L\times 2L}, (42)

and

H1=(0t1(1)000t1(1)00)2L×2L.H_{1}=\begin{pmatrix}0&-t_{1}^{(1)}&&&&\\ &0&0&&&&\\ &&0&t_{1}^{(1)}&&\\ &&&0&0&\\ &&&&\ddots&\ddots\\ &&&&&\ddots\end{pmatrix}_{2L\times 2L}. (43)

Thus, According to Eq.(8) and Eq.(21),

Heff(θ)=(𝐡0𝐡1𝐡1𝐡0𝐡1𝐡1𝐡0𝐡1)2L×2L,H_{eff}(\theta)=\begin{pmatrix}\mathbf{h}_{0}&\mathbf{h}_{1}&&&\\ \mathbf{h}_{1}^{\dagger}&\mathbf{h}_{0}&\mathbf{h}_{1}&&\\ &\mathbf{h}_{1}^{\dagger}&\mathbf{h}_{0}&\mathbf{h}_{1}&\\ &&\ddots&\ddots&\ddots\\ &&&\ddots&\ddots\end{pmatrix}_{2L\times 2L}, (44)

where

𝐡0=(0t0(0)t1(1)eiθ00t0(0)t1(1)eiθ0t1(0)00t1(0)0t0(0)+t1(1)eiθ00t0(0)+t1(1)eiθ0),\mathbf{h}_{0}=\\ \begin{pmatrix}0&t_{0}^{(0)}-t_{1}^{(1)}e^{i\theta}&0&0\\ t_{0}^{(0)}-t_{1}^{(1)}e^{-i\theta}&0&t_{1}^{(0)}&0\\ 0&t_{1}^{(0)}&0&t_{0}^{(0)}+t_{1}^{(1)}e^{i\theta}\\ 0&0&t_{0}^{(0)}+t_{1}^{(1)}e^{-i\theta}&0\end{pmatrix}, (45)

and

𝐡1=(000000000000t1(0)000).\mathbf{h}_{1}=\begin{pmatrix}0&0&0&0&\\ 0&0&0&0&\\ 0&0&0&0&\\ t_{1}^{(0)}&0&0&0&\end{pmatrix}. (46)

Under PBC,

Heff(k,θ)=𝐡0+𝐡1eik+𝐡1eik=(0t0(0)t1(1)eiθ0t1(0)eikt0(0)t1(1)eiθ0t1(0)00t1(0)0t0(0)+t1(1)eiθt1(0)eik0t0(0)+t1(1)eiθ0).H_{eff}(k,\theta)=\mathbf{h}_{0}+\mathbf{h}_{1}e^{-ik}+\mathbf{h}_{1}^{\dagger}e^{ik}=\\ \begin{pmatrix}0&t_{0}^{(0)}-t_{1}^{(1)}e^{i\theta}&0&t_{1}^{(0)}e^{ik}\\ t_{0}^{(0)}-t_{1}^{(1)}e^{-i\theta}&0&t_{1}^{(0)}&0\\ 0&t_{1}^{(0)}&0&t_{0}^{(0)}+t_{1}^{(1)}e^{i\theta}\\ t_{1}^{(0)}e^{-ik}&0&t_{0}^{(0)}+t_{1}^{(1)}e^{-i\theta}&0\end{pmatrix}. (47)

This system has chiral symmetry,

𝒮Heff(𝐤,θ)𝒮1=Heff(𝐤,θ),\mathcal{S}H_{eff}(\mathbf{k},\theta)\mathcal{S}^{-1}=-H_{eff}(\mathbf{k},\theta), (48)

where 𝒮=𝐈2×2σz\mathcal{S}=\mathbf{I}_{2\times 2}\otimes\sigma_{z}, 𝐈2×2\mathbf{I}_{2\times 2} is the 22-dimensional identity matrix and σz\sigma_{z} is the Pauli matrix. Thus this system is 𝐀𝐈𝐈𝐈\mathbf{AIII} class.

For the case t0(0)=0t_{0}^{(0)}=0, t1(0)=1t_{1}^{(0)}=1 and t1(1)=1t_{1}^{(1)}=1, the system is gapless (Fig.1), and for the case t0(0)=2t_{0}^{(0)}=2, t1(0)=2t_{1}^{(0)}=2 and t1(1)=1t_{1}^{(1)}=1, the system is half-gapped (Fig.1).

Refer to caption
Refer to caption
Figure 1: (a).The blue area is the spectrum of the effective Hamiltonian Heff(k,θ)H_{eff}(k,\theta) given by Eq.(47) with t0(0)=0t_{0}^{(0)}=0, t1(0)=1t_{1}^{(0)}=1 and t1(1)=1t_{1}^{(1)}=1 under PBC. (b).The blue area is the spectrum of the effective Hamiltonian Heff(k,θ)H_{eff}(k,\theta) given by Eq.(47) with t0(0)=2t_{0}^{(0)}=2, t1(0)=2t_{1}^{(0)}=2 and t1(1)=1t_{1}^{(1)}=1 under PBC.

If Heff(k,θ)H_{eff}(k,\theta) give by Eq.(47) is gapped, the topological invariant of system belonged to 𝐀𝐈𝐈𝐈\mathbf{AIII} class is winding number,

𝐢𝐧𝐯(θ)=i2(2π)ππTr[𝒮Heff1(k,θ)ddkHeff(k,θ)]dk.\mathbf{inv}(\theta)=\frac{i}{2(2\pi)}\int_{-\pi}^{\pi}Tr[\mathcal{S}H_{eff}^{-1}(k,\theta)\frac{d}{dk}H_{eff}(k,\theta)]\mathrm{d}k. (49)

If we take t0(0)=4t_{0}^{(0)}=4, t1(0)=1t_{1}^{(0)}=1 and t1(1)=1t_{1}^{(1)}=1, the system is gapped and topological trivial. In this case, 𝐢𝐧𝐯(θ)=0\mathbf{inv}(\theta)=0 for all θ\theta (Fig.2). For the case t0(0)=2t_{0}^{(0)}=2, t1(0)=4t_{1}^{(0)}=4 and t1(1)=1t_{1}^{(1)}=1, this system is gapped and topological non-trivial and 𝐢𝐧𝐯(θ)=1\mathbf{inv}(\theta)=1 for all θ\theta (Fig.3). For topological non-trivial system, there exist topological state in the spectrum under OBC (Fig.3). For topological trivial system, there exist no topological state in the OBC spectrum (Fig.2). This is consistent with the traditional Hermitian bulk-boundary correspondence.

Refer to caption
Refer to caption
Refer to caption
Figure 2: For the case t0(0)=4t_{0}^{(0)}=4, t1(0)=1t_{1}^{(0)}=1 and t1(1)=1t_{1}^{(1)}=1, (a).The blue area is the spectrum of the effective Hamiltonian Heff(k,θ)H_{eff}(k,\theta) given by Eq.(47) under PBC. (b).The blue area is the spectrum of he effective Hamiltonian Heff(θ)H_{eff}(\theta) given by Eq.(44) under OBC with L=50L=50. (c).The red line is the winding number 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta) of this this system.
Refer to caption
Refer to caption
Refer to caption
Figure 3: For the case t0(0)=2t_{0}^{(0)}=2, t1(0)=4t_{1}^{(0)}=4 and t1(1)=1t_{1}^{(1)}=1, (a).The blue area is the spectrum of the effective Hamiltonian Heff(k,θ)H_{eff}(k,\theta) given by Eq.(47) under PBC. (b).The blue area is the spectrum of he effective Hamiltonian Heff(θ)H_{eff}(\theta) given by Eq.(44) under OBC with L=50L=50. (c).The red line is the winding number 𝐢𝐧𝐯(θ)\mathbf{inv}(\theta) of this this system.

VI.2 non-Hermitian case

We consider a (1+1)(1+1)-dimensional non-Hermitian space-time crystal system with single orbital. For simplicity, we still assume the the lattice potential has the form given by Eqs.(34) and (35) with a=1a=1 and T=2π5T=\frac{2\pi}{5}. Thus, δk=π\delta k=\pi and Ω=5\Omega=5 for this system, and only H1H_{-1}, H0H_{0} and H1H_{1} are nonzero for the tight-binding Hamiltonian given by Eq.(6). Hence, we assume that,

H0=nt1(0)cn+1cn+t1(0)cn1cn\displaystyle H_{0}=\sum_{n}t_{1}^{(0)}c_{n+1}^{\dagger}c_{n}+t_{-1}^{(0)}c_{n-1}^{\dagger}c_{n} (50)
H1=nγei(n+1)πcn+1cn+γei(n1)πcn1cn\displaystyle H_{1}=\sum_{n}\gamma e^{i(n+1)\pi}c_{n+1}^{\dagger}c_{n}+\gamma e^{i(n-1)\pi}c_{n-1}^{\dagger}c_{n} (51)
H1=nγei(n+1)πcn+1cn+γei(n1)πcn1cn,\displaystyle H_{-1}=\sum_{n}\gamma e^{i(n+1)\pi}c_{n+1}^{\dagger}c_{n}+\gamma e^{i(n-1)\pi}c_{n-1}^{\dagger}c_{n}, (52)

where n[1,L]n\in[1,L] is the index of site and LL is the length of the system. Under PBC, according to Eq.(8) and Appendix A,

H~0=(0t1(0)+t1(0)eikt1(0)+t1(0)eik0)\displaystyle\tilde{H}_{0}=\begin{pmatrix}0&t_{-1}^{(0)}+t_{1}^{(0)}e^{ik}\\ t_{1}^{(0)}+t_{-1}^{(0)}e^{-ik}&0\end{pmatrix} (53)
H~1=(0γγeikγ+γeik0)\displaystyle\tilde{H}_{1}=\begin{pmatrix}0&-\gamma-\gamma e^{ik}\\ \gamma+\gamma e^{-ik}&0\end{pmatrix} (54)
H~1=(0γγeikγ+γeik0).\displaystyle\tilde{H}_{-1}=\begin{pmatrix}0&-\gamma-\gamma e^{ik}\\ \gamma+\gamma e^{-ik}&0\end{pmatrix}. (55)

Thus, the effective Hamiltonian under PBC is

Heff(k,θ)=H~0+H~1eiθ+H~1eiθ=(0f1(k,θ)f2(k,θ)0),\begin{split}H_{eff}(k,\theta)&=\tilde{H}_{0}+\tilde{H}_{1}e^{i\theta}+\tilde{H}_{-1}e^{-i\theta}\\ &=\begin{pmatrix}0&f_{1}(k,\theta)\\ f_{2}(k,\theta)&0\end{pmatrix},\end{split} (56)

where

f1(k,θ)=(t1(0)2γcos(θ))+(t1(0)2γcos(θ))eik\displaystyle f_{1}(k,\theta)=(t_{-1}^{(0)}-2\gamma cos(\theta))+(t_{1}^{(0)}-2\gamma cos(\theta))e^{ik} (57)
f2(k,θ)=(t1(0)+2γcos(θ))+(t1(0)+2γcos(θ))eik.\displaystyle f_{2}(k,\theta)=(t_{1}^{(0)}+2\gamma cos(\theta))+(t_{-1}^{(0)}+2\gamma cos(\theta))e^{-ik}. (58)

We find that

ΓHeff(k,θ)Γ1=Heff(k,θ),\Gamma H_{eff}^{\dagger}(k,\theta)\Gamma^{-1}=-H_{eff}(k,\theta), (59)

with Γ=σzKP\Gamma=\sigma_{z}KP, where σz\sigma_{z} is the Pauli matrix, KK is the complete conjugation operator and PP is the operator acting on creation and annihilation operators,

PckP1=ck,PckP1=ck.Pc_{k}^{\dagger}P^{-1}=c_{k},\qquad Pc_{k}P^{-1}=c_{k}^{\dagger}. (60)

Thus, this system belongs to 𝐀\mathbf{A} class with sublattice symmetry. For Heff(k,θ)H_{eff}(k,\theta) given in Eq.(56), the eigenenergy is

E2(k,θ)=Det(Heff(k,θ))=f1(k,θ)f2(k,θ)=eik[(eikt1(0)+t1(0))24(1+eik)2γ2cos2(θ)].E^{2}(k,\theta)=-Det(H_{eff}(k,\theta))=f_{1}(k,\theta)f_{2}(k,\theta)\\ =e^{-ik}[(e^{ik}t_{1}^{(0)}+t_{-1}^{(0)})^{2}-4(1+e^{ik})^{2}\gamma^{2}cos^{2}(\theta)]. (61)
Refer to caption
Figure 4: The blue area is the values of E2(k,θ)E^{2}(k,\theta) with t1(0)=1t_{1}^{(0)}=1, t1(0)=6t_{-1}^{(0)}=6 and γ=12\gamma=\frac{1}{2} for k,θ[0.2π]k,\theta\in[0.2\pi].

For the case t1(0)=1t_{1}^{(0)}=1, t1(0)=6t_{-1}^{(0)}=6 and γ=12\gamma=\frac{1}{2}, E2(k,θ)=Det(Heff(k,θ))0E^{2}(k,\theta)=-Det(H_{eff}(k,\theta))\not=0 for all k,θ[0,2π]k,\theta\in[0,2\pi] (Fig.4). According to the definition 11 in Ref.[22], Heff(k,θ)H_{eff}(k,\theta) has point gap for all θ[0,2π]\theta\in[0,2\pi]. Thus, this system is gap-preserving if t1(0)=1t_{1}^{(0)}=1, t1(0)=6t_{-1}^{(0)}=6 and γ=12\gamma=\frac{1}{2}. In this case, this system has two topological invariants [22],

𝐢𝐧𝐯1(θ)=i2πππf11(k,θ)ddkf1(k,θ)dk\displaystyle\mathbf{inv}_{1}(\theta)=\frac{-i}{2\pi}\int_{-\pi}^{\pi}f_{1}^{-1}(k,\theta)\frac{d}{dk}f_{1}(k,\theta)\mathrm{d}k (62)
𝐢𝐧𝐯2(θ)=i2πππf21(k,θ)ddkf2(k,θ)dk.\displaystyle\mathbf{inv}_{2}(\theta)=\frac{-i}{2\pi}\int_{-\pi}^{\pi}f_{2}^{-1}(k,\theta)\frac{d}{dk}f_{2}(k,\theta)\mathrm{d}k. (63)

If t1(0)=t1(0)t_{1}^{(0)}=t_{-1}^{(0)}, E2(π,θ)=0E^{2}(\pi,\theta)=0 for arbitrary θ[0,2π]\theta\in[0,2\pi]. Thus, this system is gapless in this case.

If we take t1(0)=1t_{1}^{(0)}=-1, t1(0)=3t_{-1}^{(0)}=3 and γ=12\gamma=\frac{1}{2},

E2(k,θ)=eik[(eik+3)2(1+eik)2cos2(θ)].E^{2}(k,\theta)=e^{-ik}[(-e^{ik}+3)^{2}-(1+e^{ik})^{2}cos^{2}(\theta)]. (64)

If θ=π2\theta=\frac{\pi}{2}, E2(k,π2)=eik(eik+3)20E^{2}(k,\frac{\pi}{2})=e^{-ik}(-e^{ik}+3)^{2}\not=0 for all k[π,π]k\in[-\pi,\pi]. If θ=0\theta=0, E2(k,0)=eik[(eik+3)2(1+eik)2]E^{2}(k,0)=e^{-ik}[(-e^{ik}+3)^{2}-(1+e^{ik})^{2}] and E2(0,0)=0E^{2}(0,0)=0. Hence, Heff(k,π2)H_{eff}(k,\frac{\pi}{2}) has point gap and Heff(k,0)H_{eff}(k,0) is gapless. This means this system is gap-breaking for the case t1(0)=1t_{1}^{(0)}=-1, t1(0)=3t_{-1}^{(0)}=3 and γ=12\gamma=\frac{1}{2}.

VII Conclusion and Discussion

In this work, we show that space-time crystal system can be divided into two classes, rational space-time crystal system and irrational space-time crystal system. A rational space-time crystal system is equal to a traditional floquet system. Then we find a way to solve the floquet equation analytically. Hence, we can obtain the quasi-spectrum and the effective Hamiltonian of the system analytically, and we point out that the physical spectrum of the system is just the 0th sector of the quasi-spectrum. Then, we discuss the properties of the effective Hamiltonian under internal symmetries and give the types of the Hermitian and non-Hermitian effective Hamiltonian according to the energy gap. There are three types of Hermitian effective Hamiltonian, gapped, half-gapped and gapless. And there are also three types of non-Hermitian effective Hamiltonian, gap-preserving, gap-breaking and gapless. For gapped Hermitian rational space-time crystal systems, the topological classification of them can be reduced to the Hermitian tenfold AZ classes, and for gap-preserving non-Hermitian rational space-time crystal system under PBC, the topological classification of them can be reduced to the non-Hermitian 3838-fold classes with three types of the complex energy gap. Our works give a new perspective of space-time crystal system and pave a way to research rational space-time crystal system systematically.

In the future, the topological classification about the half-gapped and gapless Hermitian rational space-time crystal systems and the topological classification about the gap-breaking and gapless non-Hermitian rational space-time crystal systems will be considered for the complete topological classification of rational space-time crystal systems. Furthermore, the properties of irrational space-time crystal systems will also be discussed in our future work .

VIII acknowledgements

The authors thank useful discussion with Yongxu Fu, Haoshu Li and Zhiwei Yin. This work was supported by NSFC Grant No. 11275180.

Appendix A The Hamiltonian After Taking the New Unit Cell

Consider a site locating at 𝐑\mathbf{R}. Under the basis {𝐬i}\{\mathbf{s}_{i}\}, we can decompose 𝐑\mathbf{R} as 𝐑=i=1dri𝐬i\mathbf{R}=\sum_{i=1}^{d}r_{i}\mathbf{s}_{i} with rir_{i}\in\mathbb{Z}. Thus, we can use a column vector to represent 𝐑\mathbf{R},

𝐑=(r1,r2,,rd)T.\mathbf{R}=(r_{1},r_{2},\cdots,r_{d})^{T}. (65)

The integer rir_{i} can be decomposed to ri=liqi+hir_{i}=l_{i}q_{i}+h_{i} with li,hil_{i},h_{i}\in\mathbb{Z} and 0hi<qi0\leq h_{i}<q_{i}. Then, 𝐑\mathbf{R} can be rewritten as

𝐑=(l1q1+h1,l2q2+h2,,ldqd+hd)T.\mathbf{R}=(l_{1}q_{1}+h_{1},l_{2}q_{2}+h_{2},\cdots,l_{d}q_{d}+h_{d})^{T}. (66)

After taking {qi𝐬i}\{q_{i}\mathbf{s}_{i}\} as the group of lattice vectors, there are i=1dqi\prod_{i=1}^{d}q_{i} orbitals in a unit cell. In this case, we treat the site locating at 𝐑\mathbf{R} under original coordinate as an orbital locating at

𝐑~=(l1,l2,,ld)T\tilde{\mathbf{R}}=(l_{1},l_{2},\cdots,l_{d})^{T} (67)

with orbital indexes (h1,h2,,hd)(h_{1},h_{2},\cdots,h_{d}). Then, we define

𝐜𝐑~=(c𝐑~(0,0,,0),c𝐑~(1,0,,0),,c𝐑~(q11,q21,,qd1))T\mathbf{c}_{\tilde{\mathbf{R}}}=(c_{\tilde{\mathbf{R}}}^{(0,0,\cdots,0)},c_{\tilde{\mathbf{R}}}^{(1,0,\cdots,0)},\cdots,c_{\tilde{\mathbf{R}}}^{(q_{1}-1,q_{2}-1,\cdots,q_{d}-1)})^{T} (68)

as the annihilation operator about the new unit cell locating at R~\tilde{R} and (0,0,,0),(1,0,,0),,(q11,q21,,qd1)(0,0,\cdots,0),(1,0,\cdots,0),\cdots,(q_{1}-1,q_{2}-1,\cdots,q_{d}-1) are orbital indexes. For

𝐑i=𝐑~i+m=1dhi,m𝐬m\displaystyle\mathbf{R}_{i}=\tilde{\mathbf{R}}_{i}+\sum_{m=1}^{d}h_{i,m}\mathbf{s}_{m} (69)
𝐑j=𝐑~j+m=1dhj,m𝐬m,\displaystyle\mathbf{R}_{j}=\tilde{\mathbf{R}}_{j}+\sum_{m=1}^{d}h_{j,m}\mathbf{s}_{m}, (70)

we define

(𝐌𝐑~i𝐑~j(n))(hi,1,hi,2,,hi,d),(hj,1,hj,2,,hj,d)=einδ𝐤𝐑it𝐑i𝐑j(n).(\mathbf{M}_{\tilde{\mathbf{R}}_{i}-\tilde{\mathbf{R}}_{j}}^{(n)})_{(h_{i,1},h_{i,2},\cdots,h_{i,d}),(h_{j,1},h_{j,2},\cdots,h_{j,d})}\\ =e^{in\mathbf{\delta k}\cdot\mathbf{R}_{i}}t_{\mathbf{R}_{i}-\mathbf{R}_{j}}^{(n)}. (71)

For 𝐑~=(l1,l2,,ld)T\tilde{\mathbf{R}}=(l_{1},l_{2},\cdots,l_{d})^{T}, δ𝐤𝐑~=2π(i=1dpili)\mathbf{\delta k}\cdot\tilde{\mathbf{R}}=2\pi(\sum_{i=1}^{d}p_{i}l_{i}). Thus, einδ𝐤𝐑~=1e^{in\mathbf{\delta k}\cdot\tilde{\mathbf{R}}}=1, and

einδ𝐤𝐑i=einδ𝐤(m=1dhi,m𝐬m).e^{in\mathbf{\delta k}\cdot\mathbf{R}_{i}}=e^{in\mathbf{\delta k}\cdot(\sum_{m=1}^{d}h_{i,m}\mathbf{s}_{m})}. (72)

This means the matrix 𝐌𝐑~i𝐑~j(n)\mathbf{M}_{\tilde{\mathbf{R}}_{i}-\tilde{\mathbf{R}}_{j}}^{(n)} only depends on the value of 𝐑~i𝐑~j\tilde{\mathbf{R}}_{i}-\tilde{\mathbf{R}}_{j}. Hence, the Hamiltonian in Eq.(6) and Eq.(7) can be written as the Hamiltonian in Eq.(8), which respect the spatial translation symmetry.

Appendix B The Physical Spectrum of The System

For the effective Hamiltonian Heff(θ)H_{eff}(\theta), we use |uωθ|u_{\omega}^{\theta}\rangle to represent the eigenvector of Heff(θ)H_{eff}(\theta) with eigenvalue ω\omega. For a given θ\theta, {|uωθ}\{|u_{\omega}^{\theta}\rangle\} are orthogonal and complete,

uωθ|uωθ=δω,ω,ω|uωθuωθ|=𝟏.\langle u_{\omega^{\prime}}^{\theta}|u_{\omega}^{\theta}\rangle=\delta_{\omega^{\prime},\omega},\qquad\sum_{\omega}|u_{\omega}^{\theta}\rangle\langle u_{\omega}^{\theta}|=\mathbf{1}. (73)

According to Eq.(10) and Eq.(12), the physical state corresponding to the quasistate |uωθ|u_{\omega}^{\theta}\rangle is

|ψωθ(t)=limleiωt12lm=lleimΩteimθ|uωθ.\displaystyle|\psi_{\omega}^{\theta}(t)\rangle=\lim_{l\rightarrow\infty}e^{-i\omega t}\frac{1}{\sqrt{2l}}\sum_{m=-l}^{l}e^{-im\Omega t}e^{-im\theta}|u_{\omega}^{\theta}\rangle. (74)

The expected value of energy about the state |ψωθ(t)|\psi_{\omega}^{\theta}(t)\rangle under long time average is

E¯=limt012t0t0t0ψωθ(t)|H(t)|ψωθ(t)dt=limt012t0t0t0ψωθ(t)|it|ψωθ(t)dt\begin{split}\bar{E}&=\lim_{t_{0}\rightarrow\infty}\frac{1}{2t_{0}}\int_{-t_{0}}^{t_{0}}\langle\psi_{\omega}^{\theta}(t)|H(t)|\psi_{\omega}^{\theta}(t)\rangle\mathrm{d}t\\ &=\lim_{t_{0}\rightarrow\infty}\frac{1}{2t_{0}}\int_{-t_{0}}^{t_{0}}\langle\psi_{\omega}^{\theta}(t)|i\partial_{t}|\psi_{\omega}^{\theta}(t)\rangle\mathrm{d}t\end{split} (75)

According to Eq.(74),

ψωθ(t)|it|ψωθ(t)=liml12lm=lln=ll(ω+mΩ)×ei(mn)Ωtei(mn)θuωθ|uωθ.\langle\psi_{\omega}^{\theta}(t)|i\partial_{t}|\psi_{\omega}^{\theta}(t)\rangle=\lim_{l\rightarrow\infty}\frac{1}{2l}\sum_{m=-l}^{l}\sum_{n=-l}^{l}(\omega+m\Omega)\\ \times e^{-i(m-n)\Omega t}e^{-i(m-n)\theta}\langle u_{\omega}^{\theta}|u_{\omega}^{\theta}\rangle. (76)

Since

uωθ|uωθ=1\displaystyle\langle u_{\omega}^{\theta}|u_{\omega}^{\theta}\rangle=1 (77)
limt012t0t0t0ei(mn)Ωt𝑑t=δm,n,\displaystyle\lim_{t_{0}\rightarrow\infty}\frac{1}{2t_{0}}\int_{-t_{0}}^{t_{0}}e^{-i(m-n)\Omega t}{d}t=\delta_{m,n}, (78)

Eq.(75) can be reduced as

E¯=liml12lm=lln=ll(ω+mΩ)δm,nei(mn)θ=ω.\begin{split}\bar{E}&=\lim_{l\rightarrow\infty}\frac{1}{2l}\sum_{m=-l}^{l}\sum_{n=-l}^{l}(\omega+m\Omega)\delta_{m,n}e^{-i(m-n)\theta}\\ &=\omega.\end{split} (79)

This means the expected value of energy about the state |ψωθ(t)|\psi_{\omega}^{\theta}(t)\rangle, which is the physical state corresponding to the quasitate |uωθ|u_{\omega}^{\theta}\rangle with quasienergy ω\omega in 0th sector of the quasi-spectrum, under long time average is still ω\omega. Hence, the 0th sector of the quasi-spectrum gives the physical spectrum of the system.

If Heff(θ)H_{eff}(\theta) is non-Hermitian, the eigenvectors of Heff(θ)H_{eff}(\theta) are biorthogonal and complete [20, 21],

uω,Lθ|uω,Rθ=δω,ω,ω|uω,Rθuω,Lθ|=𝟏,\langle u_{\omega^{\prime},L}^{\theta}|u_{\omega,R}^{\theta}\rangle=\delta_{\omega^{\prime},\omega},\qquad\sum_{\omega}|u_{\omega,R}^{\theta}\rangle\langle u_{\omega,L}^{\theta}|=\mathbf{1}, (80)

where RR (LL) means this state is a right (left) eigenstate of Heff(θ)H_{eff}(\theta). In this case,

|ψω,Rθ(t)=limleiωt12lm=lleimΩteimθ|uω,Rθ\displaystyle|\psi_{\omega,R}^{\theta}(t)\rangle=\lim_{l\rightarrow\infty}e^{-i\omega t}\frac{1}{\sqrt{2l}}\sum_{m=-l}^{l}e^{-im\Omega t}e^{-im\theta}|u_{\omega,R}^{\theta}\rangle (81)
|ψω,Lθ(t)=limleiωt12lm=lleimΩteimθ|uω,Lθ,\displaystyle|\psi_{\omega,L}^{\theta}(t)\rangle=\lim_{l\rightarrow\infty}e^{-i\omega t}\frac{1}{\sqrt{2l}}\sum_{m=-l}^{l}e^{-im\Omega t}e^{-im\theta}|u_{\omega,L}^{\theta}\rangle, (82)

and

E¯=limt012t0t0t0ψω,Lθ(t)|H(t)|ψω,Rθ(t)dt=limt012t0t0t0ψω,Lθ(t)|it|ψω,Rθ(t)dt=ω.\begin{split}\bar{E}&=\lim_{t_{0}\rightarrow\infty}\frac{1}{2t_{0}}\int_{-t_{0}}^{t_{0}}\langle\psi_{\omega,L}^{\theta}(t)|H(t)|\psi_{\omega,R}^{\theta}(t)\rangle\mathrm{d}t\\ &=\lim_{t_{0}\rightarrow\infty}\frac{1}{2t_{0}}\int_{-t_{0}}^{t_{0}}\langle\psi_{\omega,L}^{\theta}(t)|i\partial_{t}|\psi_{\omega,R}^{\theta}(t)\rangle\mathrm{d}t\\ &=\omega.\end{split} (83)

Hence, for non-Hermitian rational space-time crystal system, the 0th sector of the quasi-spectrum still gives the physical spectrum.

Appendix C The Representation of Internal Symmetry on the Effective Hamiltonian

C.1 For Hermitian case

Under PBC, the Hamiltonian in Eq.(8) is written as

H=𝐤n=einΩtH~n(𝐤)=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤𝐌𝐑~(n)𝐜𝐤.\begin{split}H&=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n}(\mathbf{k})\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{\mathbf{k}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)}\mathbf{c}_{\mathbf{k}}.\end{split} (84)

Under time-reversal transformation, operator 𝐜𝐤\mathbf{c}_{\mathbf{k}} and unit imaginary number ii are transformed as

𝒯𝐜𝐤𝒯1=U𝒯𝐜𝐤,𝒯i𝒯1=i,\mathcal{T}\mathbf{c}_{\mathbf{k}}\mathcal{T}^{-1}=U_{\mathcal{T}}\mathbf{c}_{-\mathbf{k}},\qquad\mathcal{T}i\mathcal{T}^{-1}=-i, (85)

where U𝒯U_{\mathcal{T}} is a unitary matrix. Thus, under TRS, the Hamiltonian HH is transformed as

H𝒯=𝒯H𝒯1=𝐤n=einΩt𝒯H~n(𝐤)𝒯1=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤U𝒯𝐌𝐑~(n)U𝒯𝐜𝐤=𝐤n=einΩtH~n,𝒯(𝐤),\begin{split}H_{\mathcal{T}}&=\mathcal{T}H\mathcal{T}^{-1}=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{in\Omega t}\mathcal{T}\tilde{H}_{n}(\mathbf{k})\mathcal{T}^{-1}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{T}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)*}U_{\mathcal{T}}\mathbf{c}_{-\mathbf{k}}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n,\mathcal{T}}(-\mathbf{k}),\end{split} (86)

where

H~n,𝒯(𝐤)=𝐑~ei𝐤𝐑~𝐜𝐤U𝒯𝐌𝐑~(n)U𝒯𝐜𝐤,\tilde{H}_{n,\mathcal{T}}(-\mathbf{k})=\sum_{\tilde{\mathbf{R}}}e^{i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{T}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(-n)*}U_{\mathcal{T}}\mathbf{c}_{-\mathbf{k}}, (87)

and * means complex conjugation. Thus, the effective Hamiltonian corresponding to H𝒯H_{\mathcal{T}} is

𝒯Heff(𝐤,θ)𝒯1=Heff,𝒯(𝐤,θ)=n=einθH~n,𝒯(𝐤).\mathcal{T}H_{eff}(\mathbf{k},\theta)\mathcal{T}^{-1}=H_{eff,\mathcal{T}}(\mathbf{k},\theta)=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n,\mathcal{T}}(\mathbf{k}). (88)

If the system has TRS, H=H𝒯H=H_{\mathcal{T}}. Hence, for such a system, H~n,𝒯(𝐤)=H~n(𝐤)\tilde{H}_{n,\mathcal{T}}(-\mathbf{k})=\tilde{H}_{n}(\mathbf{k}). Thus,

Heff,𝒯(𝐤,θ)=n=einθH~n(𝐤)=Heff(𝐤,θ).H_{eff,\mathcal{T}}(\mathbf{k},\theta)=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n}(-\mathbf{k})=H_{eff}(-\mathbf{k},\theta). (89)

For PHS,

𝒞𝐜𝐤𝒞1=U𝒞𝐜𝐤,𝒞i𝒞1=i,\mathcal{C}\mathbf{c}_{\mathbf{k}}\mathcal{C}^{-1}=U_{\mathcal{C}}^{*}\mathbf{c}_{-\mathbf{k}}^{\dagger},\qquad\mathcal{C}i\mathcal{C}^{-1}=i, (90)

where, U𝒞U_{\mathcal{C}} is a unitary matrix. Thus,

H𝒞=𝒞H𝒞1=𝐤n=einΩt𝒞H~n(𝐤)𝒞1=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤U𝒞T𝐌𝐑~(n)U𝒞𝐜𝐤=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤U𝒞𝐌𝐑~(n)TU𝒞𝐜𝐤=𝐤n=einΩtH~n,𝒞(𝐤),\begin{split}H_{\mathcal{C}}&=\mathcal{C}H\mathcal{C}^{-1}=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\mathcal{C}\tilde{H}_{n}(\mathbf{k})\mathcal{C}^{-1}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}U_{\mathcal{C}}^{T}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)}U_{\mathcal{C}}^{*}\mathbf{c}_{-\mathbf{k}}^{\dagger}\\ =&-\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{C}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)T}U_{\mathcal{C}}\mathbf{c}_{-\mathbf{k}}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n,\mathcal{C}}(-\mathbf{k}),\end{split} (91)

where

H~n,𝒞(𝐤)=𝐑~ei𝐤𝐑~𝐜𝐤U𝒞𝐌𝐑~(n)TU𝒞𝐜𝐤,\tilde{H}_{n,\mathcal{C}}(-\mathbf{k})=-\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{C}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)T}U_{\mathcal{C}}\mathbf{c}_{-\mathbf{k}}, (92)

and we assume that TrH=0TrH=0. The effective Hamiltonian corresponding to H𝒞H_{\mathcal{C}} is

𝒞Heff(𝐤,θ)𝒞1=Heff,𝒞(𝐤,θ)=n=einθH~n,𝒞(𝐤).\mathcal{C}H_{eff}(\mathbf{k},\theta)\mathcal{C}^{-1}=H_{eff,\mathcal{C}}(\mathbf{k},\theta)=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n,\mathcal{C}}(\mathbf{k}). (93)

If the system has PHS, H=H𝒞H=-H_{\mathcal{C}}. Thus, H~n,𝒞(𝐤)=H~n(𝐤)\tilde{H}_{n,\mathcal{C}}(-\mathbf{k})=-\tilde{H}_{n}(\mathbf{k}) and

𝒞Heff(𝐤,θ)𝒞1=n=einθH~n(𝐤)=Heff(𝐤,θ).\mathcal{C}H_{eff}(\mathbf{k},\theta)\mathcal{C}^{-1}=-\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n}(\mathbf{-k})=-H_{eff}(-\mathbf{k},\theta). (94)

Since CS is the combination of TRS and PHS,

𝒮Heff(𝐤,θ)𝒮1=Heff(𝐤,θ).\mathcal{S}H_{eff}(\mathbf{k},\theta)\mathcal{S}^{-1}=-H_{eff}(\mathbf{k},\theta). (95)

C.2 For Non-Hermitian case

We consider TRS firstly. If a non-Hermitian system has TRS,

𝒯+H𝒯+1=H,\mathcal{T}_{+}H\mathcal{T}_{+}^{-1}=H, (96)

with

𝒯+𝐜𝐤𝒯+1=U𝒯+𝐜𝐤,𝒯+i𝒯+1=i,\mathcal{T}_{+}\mathbf{c}_{\mathbf{k}}\mathcal{T}_{+}^{-1}=U_{\mathcal{T}_{+}}\mathbf{c}_{-\mathbf{k}},\qquad\mathcal{T}_{+}i\mathcal{T}_{+}^{-1}=-i, (97)

and U𝒯+U_{\mathcal{T}_{+}} is a unitary matrix. TRS in non-Hermitian system has the same form as TRS in Hermitian system. Thus, for non-Hermitian system with TRS,

𝒯+Heff(𝐤,θ)𝒯+1=Heff,𝒯+(𝐤,θ)=Heff(𝐤,θ).\mathcal{T}_{+}H_{eff}(\mathbf{k},\theta)\mathcal{T}_{+}^{-1}=H_{eff,\mathcal{T}_{+}}(\mathbf{k},\theta)=H_{eff}(-\mathbf{k},\theta). (98)

For non-Hermitian PHS,

𝒞𝐜𝐤𝒞1=U𝒞𝐜𝐤,𝒞i𝒞1=i,\mathcal{C}_{-}\mathbf{c}_{\mathbf{k}}\mathcal{C}_{-}^{-1}=U_{\mathcal{C}_{-}}^{*}\mathbf{c}_{-\mathbf{k}}^{\dagger},\qquad\mathcal{C}_{-}i\mathcal{C}_{-}^{-1}=i, (99)

with a unitary matrix U𝒞U_{\mathcal{C}_{-}}, and

𝒞H𝒞1=H.\mathcal{C}_{-}H\mathcal{C}_{-}^{-1}=-H. (100)

The form of PHS in Hermitian system and non-Hermitian system are the same. Hence, for non-Hermitian PHS,

𝒞Heff(𝐤,θ)𝒞1=Heff,𝒞(𝐤,θ)=Heff(𝐤,θ).\mathcal{C}_{-}H_{eff}(\mathbf{k},\theta)\mathcal{C}_{-}^{-1}=H_{eff,\mathcal{C}_{-}}(\mathbf{k},\theta)=-H_{eff}(-\mathbf{k},\theta). (101)

Now, we consider TRS in non-Hermitian system. Under TRS

𝒞+𝐜𝐤𝒞+1=U𝒞+𝐜𝐤,𝒞+i𝒞+1=i,\mathcal{C}_{+}\mathbf{c}_{\mathbf{k}}\mathcal{C}_{+}^{-1}=U_{\mathcal{C}_{+}}\mathbf{c}_{-\mathbf{k}},\qquad\mathcal{C}_{+}i\mathcal{C}_{+}^{-1}=-i, (102)

with a unitary matrix U𝒞+U_{\mathcal{C}_{+}}. However, if a non-Hermitian system has TRS, the Hamiltonian has the relation

𝒞+H𝒞+1=H.\mathcal{C}_{+}H^{\dagger}\mathcal{C}_{+}^{-1}=H. (103)

For the Hamiltonian HH given in Eq.(84) ,

H𝒞+=𝒞+H𝒞+1=𝐤n=einΩt𝒞+H~n(𝐤)𝒞+1=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤U𝒞+𝐌𝐑~(n)U𝒞+𝐜𝐤=𝐤n=einΩtH~n,𝒞+(𝐤),\begin{split}H_{\mathcal{C}_{+}}&=\mathcal{C}_{+}H\mathcal{C}_{+}^{-1}=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{in\Omega t}\mathcal{C}_{+}\tilde{H}_{n}(\mathbf{k})\mathcal{C}_{+}^{-1}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{C}_{+}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)*}U_{\mathcal{C}_{+}}\mathbf{c}_{-\mathbf{k}}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n,\mathcal{C}_{+}}(-\mathbf{k}),\end{split} (104)

where

H~n,𝒞+(𝐤)=𝐑~ei𝐤𝐑~𝐜𝐤U𝒞+𝐌𝐑~(n)U𝒞+𝐜𝐤.\tilde{H}_{n,\mathcal{C}_{+}}(-\mathbf{k})=\sum_{\tilde{\mathbf{R}}}e^{i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{C}_{+}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(-n)*}U_{\mathcal{C}_{+}}\mathbf{c}_{-\mathbf{k}}. (105)

Thus, the effective Hamiltonian corresponding to H𝒞+H_{\mathcal{C}_{+}} is

𝒞+Heff(𝐤,θ)𝒞+1=Heff,𝒞+(𝐤,θ)=n=einθH~n,𝒞+(𝐤).\mathcal{C}_{+}H_{eff}(\mathbf{k},\theta)\mathcal{C}_{+}^{-1}=H_{eff,\mathcal{C}_{+}}(\mathbf{k},\theta)=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n,\mathcal{C}_{+}}(\mathbf{k}). (106)

According to Eqs.(84),(103) and (104), H~n,𝒞+(𝐤)=H~n(𝐤)\tilde{H}_{n,\mathcal{C}_{+}}(-\mathbf{k})=\tilde{H}_{-n}^{\dagger}(\mathbf{k}). Hence,

𝒞+Heff(𝐤,θ)𝒞+1=n=einθH~n(𝐤)=Heff(𝐤,θ).\mathcal{C}_{+}H_{eff}(\mathbf{k},\theta)\mathcal{C}_{+}^{-1}=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{n}^{\dagger}(-\mathbf{k})=H_{eff}^{\dagger}(-\mathbf{k},\theta). (107)

If a non-Hermitian system has PHS,

𝒯H𝒯1=H,\mathcal{T}_{-}H^{\dagger}\mathcal{T}_{-}^{-1}=-H, (108)

for the Hamiltonian HH and

𝒯𝐜𝐤𝒯1=U𝒯𝐜𝐤,𝒯i𝒯1=i,\mathcal{T}_{-}\mathbf{c}_{\mathbf{k}}\mathcal{T}_{-}^{-1}=U_{\mathcal{T}_{-}}^{*}\mathbf{c}_{-\mathbf{k}}^{\dagger},\qquad\mathcal{T}_{-}i\mathcal{T}_{-}^{-1}=i, (109)

with a unitary matrix U𝒯U_{\mathcal{T}_{-}}. For HH given in Eq.(84),

H𝒯=𝒯H𝒯1=𝐤n=einΩt𝒯H~n(𝐤)𝒯1=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤U𝒯T𝐌𝐑~(n)U𝒯𝐜𝐤=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤U𝒯𝐌𝐑~(n)TU𝒯𝐜𝐤=𝐤n=einΩtH~n,𝒯(𝐤),\begin{split}H_{\mathcal{T}_{-}}&=\mathcal{T}_{-}H\mathcal{T}_{-}^{-1}=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\mathcal{T}_{-}\tilde{H}_{n}(\mathbf{k})\mathcal{T}_{-}^{-1}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}U_{\mathcal{T}_{-}}^{T}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)}U_{\mathcal{T}_{-}}^{*}\mathbf{c}_{-\mathbf{k}}^{\dagger}\\ &=-\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{T}_{-}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)T}U_{\mathcal{T}_{-}}\mathbf{c}_{-\mathbf{k}}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n,\mathcal{T}_{-}}(-\mathbf{k}),\end{split} (110)

where

H~n,𝒯(𝐤)=𝐑~ei𝐤𝐑~𝐜𝐤U𝒯𝐌𝐑~(n)TU𝒯𝐜𝐤.\tilde{H}_{n,\mathcal{T}_{-}}(-\mathbf{k})=-\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{-\mathbf{k}}^{\dagger}U_{\mathcal{T}_{-}}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)T}U_{\mathcal{T}_{-}}\mathbf{c}_{-\mathbf{k}}. (111)

The effective Hamiltonian corresponding to H~n,𝒯\tilde{H}_{n,\mathcal{T}_{-}} is

𝒯Heff(𝐤,θ)𝒯1=Heff,𝒯(𝐤,θ)=n=einθH~n,𝒯(𝐤).\mathcal{T}_{-}H_{eff}(\mathbf{k},\theta)\mathcal{T}_{-}^{-1}=H_{eff,\mathcal{T}_{-}}(\mathbf{k},\theta)=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n,\mathcal{T}_{-}}(\mathbf{k}). (112)

Since H𝒯=HH_{\mathcal{T}_{-}}=-H^{\dagger}, H~n,𝒯(𝐤)=Hn(𝐤)\tilde{H}_{n,\mathcal{T}_{-}}(-\mathbf{k})=-H_{-n}^{\dagger}(\mathbf{k}). Hence,

Heff,𝒯(𝐤,θ)=n=einθHn(𝐤)=Heff(𝐤,θ).H_{eff,\mathcal{T}_{-}}(\mathbf{k},\theta)=-\sum_{n=-\infty}^{\infty}e^{-in\theta}H_{n}^{\dagger}(-\mathbf{k})=-H_{eff}^{\dagger}(-\mathbf{k},\theta). (113)

Since CS is the combination of TRS and PHS, or the combination of TRS and PHS,

𝒮Heff(𝐤,θ)𝒮1=Heff(𝐤,θ).\mathcal{S}H_{eff}(\mathbf{k},\theta)\mathcal{S}^{-1}=-H_{eff}(\mathbf{k},\theta). (114)

SLS is the generalization of CS in non-Hermitian system. For SLS,

Γ𝐜𝐤Γ1=UΓ𝐜𝐤,ΓiΓ1=i,\Gamma\mathbf{c}_{\mathbf{k}}\Gamma^{-1}=U_{\Gamma}^{*}\mathbf{c}_{\mathbf{k}}^{\dagger},\qquad\Gamma i\Gamma^{-1}=-i, (115)

where UΓU_{\Gamma} is a unitary matrix with UΓ2=𝟏U_{\Gamma}^{2}=\mathbf{1}. If HH satisfies SLS,

ΓHΓ1=HΓ=H.\Gamma H\Gamma^{-1}=H_{\Gamma}=-H^{\dagger}. (116)

According to Eq.(115),

HΓ=ΓHΓ1=𝐤n=einΩtΓH~n(𝐤)Γ1=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤UΓT𝐌𝐑~(n)UΓ𝐜𝐤=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤UΓ𝐌𝐑~(n)UΓ𝐜𝐤=𝐤n=einΩtH~n,Γ(𝐤),\begin{split}H_{\Gamma}&=\Gamma H\Gamma^{-1}=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{in\Omega t}\Gamma\tilde{H}_{n}(\mathbf{k})\Gamma^{-1}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{\mathbf{k}}U_{\Gamma}^{T}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)*}U_{\Gamma}^{*}\mathbf{c}_{\mathbf{k}}^{\dagger}\\ &=-\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{\mathbf{k}}^{\dagger}U_{\Gamma}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)\dagger}U_{\Gamma}\mathbf{c}_{\mathbf{k}}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n,\Gamma}(\mathbf{k}),\end{split} (117)

where

H~n,Γ(𝐤)=𝐑~ei𝐤𝐑~𝐜𝐤UΓ𝐌𝐑~(n)UΓ𝐜𝐤.\tilde{H}_{n,\Gamma}(\mathbf{k})=\sum_{\tilde{\mathbf{R}}}e^{i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{\mathbf{k}}^{\dagger}U_{\Gamma}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(-n)\dagger}U_{\Gamma}\mathbf{c}_{\mathbf{k}}. (118)

Then,

ΓHeff(𝐤,θ)Γ1=Heff,Γ(𝐤,θ)=n=einθH~n,Γ(𝐤).\Gamma H_{eff}(\mathbf{k},\theta)\Gamma^{-1}=H_{eff,\Gamma}(\mathbf{k},\theta)=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n,\Gamma}(\mathbf{k}). (119)

If HH satisfies Eq.(116), H~n,Γ(𝐤)=H~n(𝐤)\tilde{H}_{n,\Gamma}(\mathbf{k})=-\tilde{H}_{-n}^{\dagger}(\mathbf{k}). Hence,

ΓHeff(𝐤,θ)Γ1=n=einθH~n(𝐤)=Heff(𝐤,θ).\Gamma H_{eff}(\mathbf{k},\theta)\Gamma^{-1}=-\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{n}^{\dagger}(\mathbf{k})=-H_{eff}^{\dagger}(\mathbf{k},\theta). (120)

For pseudo-Hermitian symmetry, the operator η\eta is a linear operator [29]. Thus,

η𝐜𝐤η1=Uη𝐜𝐤,ηiη1=i,\eta\mathbf{c}_{\mathbf{k}}\eta^{-1}=U_{\eta}\mathbf{c}_{\mathbf{k}},\qquad\eta i\eta^{-1}=i, (121)

where UηU_{\eta} is a unitary and Hermitian matrix. If HH is pseudo-Hermitian,

ηHη1=H.\eta H\eta^{-1}=H^{\dagger}. (122)

We know that

Hη=ηHη1=𝐤n=einΩtηH~n(𝐤)η1=𝐤n=einΩt𝐑~ei𝐤𝐑~𝐜𝐤Uη𝐌𝐑~(n)Uη𝐜𝐤=𝐤n=einΩtH~n,η(𝐤),\begin{split}H_{\eta}&=\eta H\eta^{-1}=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\eta\tilde{H}_{n}(\mathbf{k})\eta^{-1}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{\mathbf{k}}^{\dagger}U_{\eta}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)}U_{\eta}\mathbf{c}_{\mathbf{k}}\\ &=\sum_{\mathbf{k}}\sum_{n=-\infty}^{\infty}e^{-in\Omega t}\tilde{H}_{n,\eta}(\mathbf{k}),\end{split} (123)

where

H~n,η(𝐤)=𝐑~ei𝐤𝐑~𝐜𝐤Uη𝐌𝐑~(n)Uη𝐜𝐤,\tilde{H}_{n,\eta}(\mathbf{k})=\sum_{\tilde{\mathbf{R}}}e^{-i\mathbf{k}\cdot\tilde{\mathbf{R}}}\mathbf{c}_{\mathbf{k}}^{\dagger}U_{\eta}^{\dagger}\mathbf{M}_{\tilde{\mathbf{R}}}^{(n)}U_{\eta}\mathbf{c}_{\mathbf{k}}, (124)

and the effective Hamiltonian corresponding to H~n,η\tilde{H}_{n,\eta} is

Heff,η(𝐤,θ)=ηHeff(𝐤,θ)η1=n=einθH~n,η(𝐤)H_{eff,\eta}(\mathbf{k},\theta)=\eta H_{eff}(\mathbf{k},\theta)\eta^{-1}=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{-n,\eta}(\mathbf{k}) (125)

If HH satisfies Eq.(122), H~n,η(𝐤)=H~n(𝐤)\tilde{H}_{n,\eta}(\mathbf{k})=\tilde{H}_{-n}^{\dagger}(\mathbf{k}). Hence,

ηHeff(𝐤,θ)η1=n=einθH~n(𝐤)=Heff(𝐤,θ).\eta H_{eff}(\mathbf{k},\theta)\eta^{-1}=\sum_{n=-\infty}^{\infty}e^{-in\theta}\tilde{H}_{n}^{\dagger}(\mathbf{k})=H_{eff}^{\dagger}(\mathbf{k},\theta). (126)

All above give the effective Hamiltonian transformed under internal symmetries.

Appendix D The Constraints of Space-Time Translation Symmetry Given by TRS

Consider a (d+1)(d+1)-dimensional lattice potential V(𝐫,t)V(\mathbf{r},t) with space-time translation symmetry, whose space-time translation vectors are {(𝐬i,τi)}\{(\mathbf{s}_{i},\tau_{i})\} for i=1,2,,d+1i=1,2,\cdots,d+1. That means

V(𝐫,t)=V(𝐫+𝐬i,t+τi)V(\mathbf{r},t)=V(\mathbf{r}+\mathbf{s}_{i},t+\tau_{i}) (127)

for i=1,2,,d+1i=1,2,\cdots,d+1. If V(𝐫,t)V(\mathbf{r},t) is invariant under TRS, we have

V(𝐫,t)=𝒯V(𝐫,t)=V(𝐫,t),V(\mathbf{r},t)=\mathcal{T}V(\mathbf{r},t)=V^{*}(\mathbf{r},-t), (128)

where 𝒯\mathcal{T} is the time-reversal operator and “*” means complex conjugation. Since V(𝐫,t)V(\mathbf{r},t) satisfies Eq.(127) and Eq.(128) simultaneously,

V(𝐫,t)=V(𝐫,t)=V(𝐫+𝐬i,t+τi)=V(𝐫+𝐬i,tτi).V(\mathbf{r},t)=V^{*}(\mathbf{r},-t)=V^{*}(\mathbf{r}+\mathbf{s}_{i},-t+\tau_{i})=V(\mathbf{r}+\mathbf{s}_{i},t-\tau_{i}). (129)

Eq.(129) shows that for a (d+1)(d+1)-dimensional space-time crystal system with space-time translation vectors {(𝐬i,τi)}\{(\mathbf{s}_{i},\tau_{i})\} (i=1,2,,d+1i=1,2,\cdots,d+1), if this system is invariant under TRS, {(𝐬i,τi)}\{(\mathbf{s}_{i},-\tau_{i})\} (i=1,2,,d+1i=1,2,\cdots,d+1) are also space-time translation vectors. Under these constraints, TRS is compatible with space-time translation symmetry.

References