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

Dynamical Symmetries and Symmetry-Protected Selection Rules in Periodically Driven Quantum Systems

Georg Engelhardt1    Jianshu Cao2 [email protected] 1Beijing Computational Science Research Center, Beijing 100193, People’s Republic of China
2Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA
Abstract

In recent experiments, the light-matter interaction has reached the ultrastrong coupling limit, which can give rise to dynamical generalizations of spatial symmetries in periodically driven systems. Here, we present a unified framework of dynamical-symmetry-protected selection rules based on Floquet response theory. Within this framework, we study rotational, parity, particle-hole, chiral, and time-reversal symmetries and the resulting selection rules in spectroscopy, including symmetry-protected dark states (spDSs), symmetry-protected dark bands (spDBs), and symmetry-induced transparency (siT). Specifically, dynamical rotational and parity symmetries establish spDS and spDB conditions. Particle-hole symmetry introduces spDSs for symmetry-related Floquet states and also a siT at quasienergy crossings. Chiral symmetry and time-reversal symmetry alone do not imply spDS conditions but can be combined to define a particle-hole symmetry. These symmetry conditions arise from destructive interference due to the synchronization of symmetric quantum systems with the periodic driving. Our predictions reveal new physical phenomena when a quantum system reaches the strong light-matter coupling regime, which is important for superconducting qubits, atoms and molecules in optical fields or plasmonic cavities, and optomechanical systems.

pacs:

Introduction. Over the last few decades, the light-matter interaction strength has been pushed to the ultrastrong coupling regime in optomechanical systems [1], quantum dots, atoms and molecules in optical or plasmonic cavities [2, 3, 4, 5, 6], and superconducting quantum circuits [7, 8, 9]. As standard nonlinear perturbation theory [10] becomes unfeasible under these conditions, Floquet response theory has been developed recently, describing systems that are subject to a strong but time-periodic driving field (of frequency Ω\Omega), and a weak but arbitrary probe field [11, 12, 13, 14]. For a monochromatic probe of frequency ωp\omega_{p}, system observables generate response frequencies ωp+nΩ\omega_{p}+n\Omega termed Floquet bands [15].

Spatial symmetries give rise to appealing physical properties. Inversion symmetry results in selection rules for dipole transitions; particle-hole, chiral, and time-reversal symmetries establish the so-called periodic table, a classification scheme for topological insulators [16, 17, 18], and symmetries have an essential impact on transport properties [19, 20, 21, 22, 23, 24, 25]. For periodically driven systems, these spatial symmetries can be generalized to dynamical symmetries that can give rise to a generalized periodic table for topological insulators [26, 27] and new control mechanisms  [28, 29, 30, 31, 32, 33, 34, 35]. Dynamical symmetries have been used to control the coherent destruction of tunneling effect [36] and induce selection rules for high harmonic generation [37, 38, 39, 40].

Table 1: Overview of the spectroscopic signatures of dynamical rotational symmetry (RS), particle-hole symmetry (PHS), parity symmetry (PS), chiral symmetry (CS), and time-reversal symmetry (TRS). The signatures include symmetry-protected dark states (spDSs), symmetry-protected dark bands (spDBs), symmetry-induced transparency (siT), and accidental dark states (aDSs). The rightmost column lists example models.
Symmetry Effect Example
RS spDS benzene ring (Fig. 1)
spDB
PS spDS two-level system (Fig. 3)
spDB
PHS spDS dimer (Fig. 2)
2 ×\times PHS siT two-level system (Fig. 3)
TRS none
CS none
none aDS all (Figs. 1,2 ,3)

In this Letter, we introduce a unified conceptual framework of selection rules based on general dynamical symmetries of periodically driven quantum systems as described by Floquet response theory [41]. Physically, the synchronization of symmetric quantum systems with the periodic driving gives rise to destructive interference effects in Floquet space and thus to forbidden transitions between Floquet states. This set of forbidden transitions defines the symmetry-protected selection rules that are robust against symmetry-preserving parameter variations. Specifically, there are four types of forbidden transitions ordered in increasing degree of complexity: (i) accidental dark states (aDSs), appearing for a specific combination of system parameters; (ii) symmetry-protected dark states (spDSs), which refer to the symmetry-protected absence of a complete transition line similar to symmetry-protected excitations of topological band structures; (iii) symmetry-protected dark bands (spDBs), which refer to the absence of a complete Floquet band due to a combination of spDSs; (iv) symmetry-induced transparency (siT), which refers to the vanishing transition intensity at the degeneracy of quasienergies. Except for the aDS, which is not symmetry related, we establish symmetry-protected selection rules for important dynamical symmetries, which are classified in Table 1.

Floquet response theory. We apply a semiclassical approach based on the general Hamiltonian

H^(t)=H^0(t)+0𝑑ω[λV^(a^ω+a^ω)+ωa^ωa^ω],\hat{H}(t)=\hat{H}_{0}(t)+\int_{0}^{\infty}d\omega\left[\lambda\hat{V}\left(\hat{a}_{\omega}+\hat{a}_{\omega}^{\dagger}\right)+\omega\hat{a}_{\omega}^{\dagger}\hat{a}_{\omega}\right], (1)

where H^0(t)=H^0(t+τ)\hat{H}_{0}(t)=\hat{H}_{0}(t+\tau) describes a system driven by a periodic classical electromagnetic field of frequency Ω=2π/τ\Omega=2\pi/\tau. The probe field is given by a continuum of photonic operators aωa_{\omega}^{\dagger} with frequencies ω\omega, which are coupled via the dipole transition operator V^\hat{V} with strength λ\lambda to the driven system. The physical properties of H^0(t)\hat{H}_{0}(t) are determined by the Floquet equation

[H^0(t)iddt]|uμ(t)=ϵμ|uμ(t),\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right]\left|u_{\mu}(t)\right>=\epsilon_{\mu}\left|u_{\mu}(t)\right>, (2)

where |uμ(t)=|uμ(t+τ)\left|u_{\mu}(t)\right>=\left|u_{\mu}(t+\tau)\right> and ϵμ\epsilon_{\mu} are the corresponding Floquet states and quasienergies that generalize the concept of eigenstates and eigenenergies of time-independent systems. It is implicitly assumed that the driven system is weakly dissipative such that, for long times, it approaches the stationary state

ρ(t)=μpμ|uμ(t)uμ(t)|,\rho(t)=\sum_{\mu}p_{\mu}\left|u_{\mu}(t)\right>\left<u_{\mu}(t)\right|, (3)

which is diagonal in the Floquet basis and thus synchronizes with the driving ρ(t)=ρ(t+τ)\rho(t)=\rho(t+\tau). Equation (3) is consistent with the Floquet-Redfield equation [42, 43, 44, 45] describing periodically driven open quantum systems. In the model calculations, we assume the special distribution pμeβϵμp_{\mu}\propto e^{-\beta\epsilon_{\mu}}, i.e., a Floquet-Gibbs distribution, but all our predictions hold even if the Floquet-Gibbs distribution breaks down [45]. Strongly dissipative systems could be addressed by generalizing our approach to non-Hermitian Hamiltonians [18] or via the polaron transformation [46].

The interaction of H^0(t)\hat{H}_{0}(t) with the probe field is treated using the input-output formalism and a perturbation expansion for small λ\lambda. The input field consists of a bichromatic probe field (of frequencies ωp,1\omega_{p,1} and ωp,2=ωp,1+nΩ\omega_{p,2}=\omega_{p,1}+n\Omega, integer nn). As shown separately [14], the intensity change of the output field at frequency ωp,2\omega_{p,2} proportional to the coherence a^ωp,2a^ωp,1\left<\hat{a}_{\omega_{p,2}}^{\dagger}\hat{a}_{\omega_{p,1}}\right> is given by ΔIcoh(ωp,2)=iχ~n(ωp,1)a^ωp,2a^ωp,1+c.c.\Delta I_{\rm coh}(\omega_{p,2})=-i\tilde{\chi}_{n}(\omega_{p,1})\left<\hat{a}_{\omega_{p,2}}^{\dagger}\hat{a}_{\omega_{p,1}}\right>+\text{c.c.}, where the susceptibility χ~n(ωp,1)\tilde{\chi}_{n}(\omega_{p,1}) can be evaluated using Floquet response theory and reads

χ~n(ωp,1)\displaystyle\tilde{\chi}_{n}(\omega_{p,1}) =\displaystyle= iλ2ν,μ,mVν,μ(nm)Vμ,ν(m)(pνpμ)ϵμϵν+mΩωp,1iγν,μ(m).\displaystyle i\lambda^{2}\sum_{\nu,\mu,m}\frac{V_{\nu,\mu}^{(-n-m)}V_{\mu,\nu}^{(m)}\left(p_{\nu}-p_{\mu}\right)}{\epsilon_{\mu}-\epsilon_{\nu}+m\Omega-\omega_{p,1}-i\gamma_{\nu,\mu}^{(m)}}. (4)

The index nn denotes the Floquet band, which describes nonelastic scattering of the probe field, and the dynamical dipole matrix elements read

Vλ,μ(n)=1τ0τuλ(t)|V^|uμ(t)einΩt𝑑t.V_{\lambda,\mu}^{(n)}=\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\lambda}(t)\right|\hat{V}\left|u_{\mu}(t)\right>e^{-in\Omega t}dt. (5)

The parameters γν,μ(m)\gamma_{\nu,\mu}^{(m)} have been added phenomenologically and denote dephasing rates.

Unified conceptual framework of dynamical-symmetry-protected selection rules. We consider the following class of symmetry operations [26]:

Σ^[H^0(tS+βSt)iddt]Σ^1=αS[H^0(t)iddt],\hat{\Sigma}\left[\hat{H}_{0}(t_{S}+\beta_{S}t)-i\frac{d}{dt}\right]\hat{\Sigma}^{-1}=\alpha_{S}\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right], (6)

where Σ^\hat{\Sigma} is a time-independent spatial operator. By specifying Σ^\hat{\Sigma}, tSt_{S}, and (αS,βs=±1)(\alpha_{S},\beta_{s}=\pm 1), one can define a set of dynamical symmetries. Applying Eq. (6) to the Floquet equation Eq. (2), one can identify relations between Floquet states μ\mu and μ\mu^{\prime}:

|uμ(t)=πμ(S)Σ^|uμ(tS+βSt).\left|u_{\mu^{\prime}}(t)\right>=\pi^{(S)}_{\mu}\hat{\Sigma}\left|u_{\mu}(t_{S}+\beta_{S}t)\right>. (7)

These relations can be used to evaluate the dynamical dipole elements in Eq. (5). Imposing an invariance condition for the transition dipole operator Σ^V^Σ^=αV(S)V^\hat{\Sigma}^{\dagger}\hat{V}\hat{\Sigma}=\alpha_{V}^{(S)}\hat{V} and using Eq. (7), we investigate symmetry-protected selection rules for rotational, parity, particle-hole, chiral and time-reversal symmetries.

Within the unified framework, we can establish symmetry-protected selection rules that are robust against symmetry-conserved variations and unique for strong light-matter interactions. Among others, we investigate dark states, which are defined by the condition Vν,μ(m)=0V_{\nu,\mu}^{(m)}=0, such that the corresponding resonances in Eq. (4) vanish. This condition not only generalizes the dark state condition in the standard response theory to the strong-coupling regime for n=0n=0 but also introduces distinct dark states effects for n0n\neq 0. All selection rules are a consequence of destructive interference due to the synchronization of the system state with the periodic driving: (i) The dark state condition can be fulfilled by special combinations of parameters, which we denote as an aDS, or (ii) as a consequence of a symmetry, which we denote as a spDS. (iii) An entire Floquet band can vanish because χ~n(ωp)=0\tilde{\chi}_{n}(\omega_{p})=0 for specific nn, which we denote as a spDB. (iv) By analyzing the susceptibility in terms of Eq. (7), we establish the condition for the siT, which is due to a destructive interference of two transitions with Vν,μ(m)0V_{\nu,\mu}^{(m)}\neq 0.

Refer to caption
Figure 1: (a) Benzene driven by circularly polarized light propagating perpendicular to the ring plane. The probe field is polarized perpendicular to the plane so that it does not destroy the sixfold dynamical rotational symmetry. (b) Quasienergies of benzene for the tunneling constant J0=0.05ΩJ_{0}=0.05\Omega and on-site energy E0=0.45ΩE_{0}=0.45\Omega. (c) Susceptibility χ~0(ωp)\tilde{\chi}_{0}(\omega_{p}) (color gradient). Rotational spDSs are marked by dashed lines. One transition vanishes at the location of the aDS. The dephasing rates in all figures are γν,μ(m)=0.001Ω\gamma_{\nu,\mu}^{(m)}=0.001\Omega.

Rotational symmetry. With αS=βS=1\alpha_{S}=\beta_{S}=1, a unitary Σ^=R^\hat{\Sigma}=\hat{R}, and tS=tR=τNt_{S}=t_{R}=\frac{\tau}{N} with a positive integer NN, Eq. (6) defines a dynamical rotational symmetry [47] that gives rise to the eigenvalue equation |uμ(t)=πμ(R)R^|uμ(t+tR)\left|u_{\mu}(t)\right>=\pi_{\mu}^{(R)}\hat{R}\;\left|u_{\mu}(t+t_{R})\right> with eigenvalues πμ(R)=ei2πmμ/N\pi_{\mu}^{(R)}=e^{i2\pi m_{\mu}/N} and integer mμ={0,N1}m_{\mu}=\left\{0,N-1\right\}. As shown in detail in the Supplemental Material [48], for a dipole transition operator with R^V^R^=αV(R)V^\hat{R}^{\dagger}\hat{V}\hat{R}=\alpha_{V}^{(R)}\hat{V} and αV(R)=±1\alpha_{V}^{(R)}=\pm 1, the dynamical rotational symmetry establishes a sufficient condition for spDSs:

V^ν,μ(m){1ifei2πN(mμmν+m)αV(R)=1,0else.\hat{V}_{\nu,\mu}^{(m)}\propto\begin{cases}1&\text{if}\;e^{i\frac{2\pi}{N}\left(m_{\mu}-m_{\nu}+m\right)}\alpha_{V}^{(R)}=1,\\ 0&\text{else}.\end{cases} (8)

Applying Eq. (8) to evaluate the susceptibility in Eq. (4), we find

χ~n(ωp)={1ifei2πNn=1,0else,\displaystyle\tilde{\chi}_{n}(\omega_{p})=\begin{cases}1&\text{if}\;e^{i\frac{2\pi}{N}n}=1,\\ 0&\text{else},\end{cases} (9)

which is the condition for the complete disappearance of Floquet band nn, i.e., a spDB. Physically, this effect appears as the stationary state Eq. (3) synchronizes with the driving field such that the density matrix adopts the dynamical rotational symmetry, i.e., ρ(t+n/Nτ)=R^nρ(t)R^n\rho(t+n/N\tau)=\hat{R}^{n}\rho(t)\hat{R}^{\dagger n}.

As an example, we consider a benzene ring driven by circularly polarized light sketched in Fig. 1(a), which is described by a tight-binding Hamiltonian:

H^0(t)\displaystyle\hat{H}_{0}(t) =\displaystyle= j,j=16Jj,j|ejej|+j=16[ifj(t)|ejej+1|+H.c.],\displaystyle\sum_{j,j^{\prime}=1}^{6}J_{j,j^{\prime}}\left|e_{j}\right>\left<e_{j^{\prime}}\right|+\sum_{j=1}^{6}\left[if_{j}(t)\left|e_{j}\right>\left<e_{j+1}\right|+\text{H.c.}\right],

where |ej\left|e_{j}\right> denotes the excitation on site jj (defined modulo 66), Jj,j=E0J_{j,j}=E_{0} is the on-site energy, Jj,j=δj,j±1J0J_{j,j^{\prime}}=\delta_{j,j^{\prime}\pm 1}J_{0} is the tunneling constant, and fj(t)=fΩcos(Ωt+2πj/6)f_{j}(t)=f_{\Omega}\cos(\Omega t+2\pi j/6) is the time-dependent tunneling strength with the driving amplitude fΩf_{\Omega}. The driving terms are motivated by the Peierls substitution describing a vectorial current-gauge-field coupling 𝒋𝑨(t)\boldsymbol{j}\cdot\boldsymbol{A}(t) [49] with a circularly rotating vector potential 𝑨(t)\boldsymbol{A}(t). The dipole transition operator V^=j=1Nd0|ejg|\hat{V}=\sum_{j=1}^{N}d_{0}\left|e_{j}\right>\left<g\right| excites the ground state |g\left|g\right> to the single-excitation manifold, whose quasienergies are depicted in Fig. 1(b). The stationary state is ρs(t)=|gg|\rho_{s}(t)=\left|g\right>\left<g\right| in agreement with Eq. (3), i.e., a Floquet-Gibbs state for low temperatures. A rotational symmetry is fulfilled for N=6N=6 and R^=j=1n|ej+1ej|\hat{R}=\sum_{j=1}^{n}\left|e_{j+1}\right>\left<e_{j}\right|.

In Fig. 1(c), we depict the susceptibility χ~0(ωp)\tilde{\chi}_{0}(\omega_{p}) of the benzene model. The resonances of the dark states defined by Eq. (8) are marked by dashed lines (optically invisible), and only two transitions, V^0,1(0)V^1,0(0)\hat{V}_{0,1}^{(0)}\hat{V}_{1,0}^{(0)} and V^3,0(1)V^0,3(1)\hat{V}_{3,0}^{(1)}\hat{V}_{0,3}^{(-1)}, are visible. An aDS can be found for V^0,1(0)V^1,0(0)\hat{V}_{0,1}^{(0)}\hat{V}_{1,0}^{(0)} at fΩ=1.5Ωf_{\Omega}=1.5\Omega. As a consequence of the spDB in Eq. (9), only Floquet bands χ~n(ωp)\tilde{\chi}_{n}(\omega_{p}) with nmod6=0n\mod 6=0 appear.

Parity symmetry. A dynamical parity symmetry is a specification of the dynamical rotational symmetry with N=2N=2 and a Hermitian operator R=RR^{\dagger}=R such that the spDS condition Eq. (8) and the spDB condition Eq. (9) are equally valid. The spDSs will be illustrated for the two-level system (TLS) in Eq. (13) along with the siT discussed below.

Refer to caption
Figure 2: (a) Sketch of the dimer model Eq. (11) with h1(t)=fΩcos(Ωt)h_{1}(t)=f_{\Omega}\cos(\Omega t). (b) Quasienergy spectrum for J0/Ω=0.05J_{0}/\Omega=0.05, r=2,r=2, and Δ=0.2Ω\Delta=0.2\Omega. (c) The susceptibility |χ~0(ωp)|\left|\tilde{\chi}_{0}(\omega_{p})\right| is depicted as a color gradient. The spDSs (marked by dashed lines) are generated by a particle-hole symmetry.
Refer to caption
Figure 3: (a) Sketch of the ac-driven TLS. (b) Quasienergy spectrum for hx/Ω=0.05h_{x}/\Omega=0.05. (c) The spectrum of the susceptibility χ~0(ωp)\tilde{\chi}_{0}(\omega_{p}) exhibits a siT and spDSs. Here, p0=0.6p_{0}=0.6 and p1=0.4p_{1}=0.4 in Eq. (3) to highlight the siT.

Particle-hole symmetry. A particle-hole symmetry is defined for αS=βS=1-\alpha_{S}=\beta_{S}=1, tS=tP=τN1/2N2t_{S}=t_{P}=\tau N_{1}/2N_{2} with integers N1{0,1}N_{1}\in\left\{0,1\right\}, N21N_{2}\geq 1, and Σ^=P^κ^\hat{\Sigma}=\hat{P}\hat{\kappa} with a unitary operator P^\hat{P} and the complex conjugation operator κ^\hat{\kappa}, such that P^H^(t+tP)P^=H^(t).\hat{P}\hat{H}^{*}(t+t_{P})\hat{P}=-\hat{H}(t). The particle-hole symmetry establishes a symmetry between the excitation and deexcitation processes and has its origin in fermionic systems, where adding and removing quasiparticles results in physically equivalent behaviors. Here we use the particle-hole symmetry in a general context. Using the particle-hole symmetry in Eq. (2), we find that each Floquet state |uμ(t)\left|u_{\mu}(t)\right> with quasienergy ϵμ\epsilon_{\mu} has its symmetry related partner |uμ(t)=πμ(P)P^|uμ(t+tP)\left|u_{\mu^{\prime}}(t)\right>=\pi^{(P)}_{\mu}\hat{P}\left|u_{\mu}(t+t_{P})\right>^{*} with energy ϵμ=ϵμ\epsilon_{\mu^{\prime}}=-\epsilon_{\mu} and a gauge-dependent πμ(P)\pi^{(P)}_{\mu}. For tP=τ/(2N2)t_{P}=\tau/(2N_{2}), the particle-hole symmetry gives rise to a rotational symmetry defined by R^=P^P^\hat{R}=\hat{P}\hat{P} and tR=τ/N2t_{R}=\tau/N_{2} such that the dark state selection rules of the rotational symmetry apply. The particle-hole symmetry can give rise to a distinct dark state condition. For a dipole transition operator with P^V^P^=αV(P)V^\hat{P}^{\dagger}\hat{V}^{*}\hat{P}=\alpha_{V}^{(P)}\hat{V}, αV(P)=±1\alpha_{V}^{(P)}=\pm 1, tP=0,τ/2t_{P}=0,\tau/2, and P^P^=1\hat{P}^{*}\hat{P}=1, the particle-hole symmetry results in Vμ,μ(m)=αV(P)eimΩtPVμ,μ(m)V_{\mu,\mu^{\prime}}^{(m)}=\alpha_{V}^{(P)}e^{im\Omega t_{P}}V_{\mu,\mu^{\prime}}^{(m)} for the symmetry-related states μ,μ\mu,\mu^{\prime}, so that

V^μ,μ(m){0ifαV(P)eimΩtP=1;μ,μ sym. rel.1else,\hat{V}_{\mu,\mu^{\prime}}^{(m)}\propto\begin{cases}0&\text{if}\;\alpha_{V}^{(P)}e^{im\Omega t_{P}}=-1;\;\mu,\mu^{\prime}\text{ sym. rel.}\\ 1&\text{else},\end{cases} (10)

as shown in detail in the Supplementary Material [48]. In contrast to Eq. (8), where each transition can vanish for an appropriate mm, only transitions between symmetry-related states are affected by Eq. (10).

To illustrate Eq. (10), we use the dimer model sketched in Fig. 2(a), with the Hamiltonian given by

H0(t)=\displaystyle H_{0}(t)= Δ(A^f,fA^g,g)+J0A^e1,e2\displaystyle\Delta\left(\hat{A}_{f,f}-\hat{A}_{g,g}\right)+J_{0}\hat{A}_{e_{1},e_{2}} (11)
+\displaystyle+ h1(t)[A^e1,f+A^g,e1+rA^e1,e2],\displaystyle h_{1}(t)\left[\hat{A}_{e_{1},f}+\hat{A}_{g,e_{1}}+r\hat{A}_{e_{1},e_{2}}\right],

where A^α,β|αβ|+H.c.\hat{A}_{\alpha,\beta}\equiv\left|\alpha\right>\left<\beta\right|+\text{H.c.}, and g,e1,e2g,e_{1},e_{2} and ff label the ground state, two single-excitation states, and the double-excitation state, respectively. Δ\Delta is the excitation gap, J0J_{0} is the tunneling constant, and h1(t)=fΩcos(Ωt)h_{1}(t)=f_{\Omega}\cos(\Omega t) is the driving field. The rr term enhances higher-order dipole elements Vμ,μm0V^{m\neq 0}_{\mu,\mu^{\prime}}. The particle-hole symmetry is defined by P^=A^g,f+A^e1,e1A^e2,e2\hat{P}=\hat{A}_{g,f}+\hat{A}_{e_{1},e_{1}}-\hat{A}_{e_{2},e_{2}} and tP=0t_{P}=0. The quasienergy spectrum in Fig. 2(b) is symmetric with respect to E=0E=0. The dipole transition operator is V^=A^e1,f+A^g,e1\hat{V}=\hat{A}_{e_{1},f}+\hat{A}_{g,e_{1}}, such that P^V^P^=V^\hat{P}^{\dagger}\hat{V}^{*}\hat{P}=-\hat{V}. In Fig. 2(c), we depict the susceptibility in Eq. (4). According to the above considerations, the transitions between the particle-hole symmetry-related pairs vanish, i.e., V1,4(m)=V4,1(m)=V2,3(m)=V3,2(m)=0V_{1,4}^{(m)}=V_{4,1}^{(m)}=V_{2,3}^{(m)}=V_{3,2}^{(m)}=0 for all mm. These resonances are marked by dashed lines. The other transitions not affected by the symmetry constrain remain visible in Fig. 2(c).

Symmetry-induced transparency. The particle-hole symmetry can also give rise to a siT at the quasienergy crossing ϵμ=ϵμ=0\epsilon_{\mu}=\epsilon_{\mu^{\prime}}=0 of the symmetry-related Floquet states μ,μ\mu,\mu^{\prime}. While a spDS is generated by a vanishing dipole element, Vλ,μ(n)=0V_{\lambda,\mu}^{(n)}=0, the siT is generated by a destructive interference of two transitions with Vλ,μ(n)0V_{\lambda,\mu}^{(n)}\neq 0. As shown in the Supplementary Material [48] in detail, for two distinct particle-hole symmetries P^1±P^2\hat{P}_{1}\neq\pm\hat{P}_{2}, P^i2=𝟙\hat{P}_{i}^{2}=\mathbbm{1} and [P^1,P^2]=0\left[\hat{P}_{1},\hat{P}_{2}\right]=0, the siT condition reads

χ~n(mΩ){0ifeimΩ(tP1tP2)=1;ϵμ=ϵμ=01else,\tilde{\chi}_{n}(m\Omega)\propto\begin{cases}0&\text{if}\;e^{im\Omega(t_{P_{1}}-t_{P_{2}})}=1;\epsilon_{\mu}=\epsilon_{\mu^{\prime}}=0\\ 1&\text{else},\end{cases} (12)

where tPit_{P_{i}} denote the reference times related to P^i\hat{P}_{i}.

For illustration, we consider an ac-driven TLS sketched in Fig. 3(a) and described by the Hamiltonian

H^0(t)\displaystyle\hat{H}_{0}(t) =hx2σ^x+fΩ2cos(Ωt)σ^z,\displaystyle=\frac{h_{\rm x}}{2}\hat{\sigma}_{\rm x}+\frac{f_{\Omega}}{2}\cos\left(\Omega t\right)\hat{\sigma}_{\rm z}, (13)

where σ^x,σ^z\hat{\sigma}_{x},\hat{\sigma}_{\rm z} are the Pauli matrices, hxh_{x} is the tunneling amplitude, and fΩf_{\Omega} the driving strength. The TLS is weakly dissipative, as in the spin-boson model, such that it reaches the stationary state in Eq. (3). The dipole transition operator in Eq. (1) is V^=σ^x\hat{V}=\hat{\sigma}_{x}. For R^=σ^x\hat{R}=\hat{\sigma}_{\rm x} and tR=τ/2t_{R}=\tau/2, the TLS exhibits a dynamical parity symmetry defined above, which gives rise to the coherent destruction of tunneling effect at an exact quasienergy crossing, depicted in Fig. 3(b) at fΩ2.4Ωf_{\Omega}\approx 2.4\Omega [36, 50], and enables the siT in the current context. Additionally, the TLS exhibits spDSs and spDBs according to Eq. (8) and Eq. (9) as Vμ,ν(m)=0V_{\mu,\nu^{\prime}}^{(m)}=0 for even mm because of the dynamical parity symmetry.

For the TLS, a particle-hole symmetry is defined for P^1=σ^z\hat{P}_{1}=\hat{\sigma}_{\rm z} and tP1=τ/2t_{P_{1}}=\tau/2. For hx=0h_{\rm x}=0, a second particle-hole symmetry is given for P^2=𝟙\hat{P}_{2}=\mathbbm{1} and tP2=τ/2t_{P_{2}}=\tau/2. As in this case ϵμ=0\epsilon_{\mu}=0 and P^iσ^xP^i=(1)iσ^x\hat{P}_{i}\hat{\sigma}_{x}^{*}\hat{P}_{i}=(-1)^{i}\hat{\sigma}_{x}, siT with χ~n(mΩ)=0\tilde{\chi}_{n}(m\Omega)=0 appears according to Eq. (12), and the response χ~n(ωp)\tilde{\chi}_{n}(\omega_{\rm p}) is complete suppressed for all nn. In Fig. 3(c), we consider χ~0(ωp)\tilde{\chi}_{0}(\omega_{p}) for a finite but small hxΩh_{\rm x}\ll\Omega such that the quasienergy degeneracy is lifted except of the crossing, and the particle-hole symmetry P^2\hat{P}_{2} is slightly broken. As a consequence, the siT is not complete but scales as χ~n(mΩ)hx/Ω\tilde{\chi}_{n}(m\Omega)\propto h_{x}/\Omega at the crossing.

Time-reversal and chiral symmetries. A time-reversal symmetry (chiral symmetry) is defined by Eq. (6) for αS=βS=1\alpha_{S}=-\beta_{S}=1 (αS=βS=1\alpha_{S}=\beta_{S}=-1), arbitrary tSt_{S}, and Σ^=T^κ^\hat{\Sigma}=\hat{T}\hat{\kappa}, (Σ^=C^\hat{\Sigma}=\hat{C}), where T^\hat{T} (C^\hat{C}) is a unitary operator. As shown in the Supplementary Material [48], neither time-reversal symmetry nor chiral symmetry alone implies spDSs. However, the combination of time-reversal symmetry and chiral symmetry defines a particle-hole symmetry with P^=C^T^\hat{P}=\hat{C}\hat{T}, and tP=tTtCt_{P}=t_{T}-t_{C}. When they further fulfill tTtC{0,τ/2}t_{T}-t_{C}\in\left\{0,\tau/2\right\}, C^C^=𝟙\hat{C}^{*}\hat{C}=\mathbbm{1}, T^T^=𝟙\hat{T}^{*}\hat{T}=\mathbbm{1}, and [C^,T^]=0\left[\hat{C},\hat{T}\right]=0, such that P^P^=1\hat{P}^{*}\hat{P}=1, spDSs appear because of the particle-hole symmetry. In general, the presence of any two symmetries out of particle-hole symmetry, chiral symmetry, and time-reversal symmetry implies the existence of the third one.

Conclusions. Using a unified conceptional framework based on Floquet response theory, we have predicted selection rules in periodically driven quantum systems, namely accidental dark states, symmetry-protected dark states, symmetry-protected dark bands, and symmetry-induced transparency. The latter three effects are protected by symmetries such that variations of symmetry preserving parameters do not destroy them. These symmetry-induced selection rules result from the destructive interference of a driven system synchronized to the periodic driving. The different effects have been illustrated in three example systems fulfilling different symmetries, demonstrating the flexibility and generality of our unified framework. The predicted selection rules are valid even for more complicated and realistic systems as long as the corresponding dynamical symmetries are fulfilled.

Our theoretical results are experimentally observable in systems that can reach the strong light-matter coupling regime such as cold-atom experiments [2] and superconducting circuits [51, 52, 53, 54]. For experiments with molecules, strong driving fields are necessary to generate high-order Floquet bands, but in cavity QED or plasmonic fields, the strong driving interaction condition can be relaxed for molecule ensembles interacting collectively with the light field [55, 56].

Acknowledgements. G. E. gratefully acknowledges financial support from the China Postdoc Science Foundation (Grant No. 2018M640054 ) and the Natural Science Foundation of China (Grant Nos. 11950410510 and U1930402), J. C. acknowledges support from the NSF (Grant Nos. CHE 1800301 and CHE1836913). The authors thank Tao Wang for helpful discussions.

References

  • Fang et al. [2017] K. Fang, J. Luo, A. Metelmann, M. H. Matheny, F. Marquardt, A. A. Clerk, and O. Painter, Generalized nonreciprocity in an optomechanical circuit via synthetic magnetism and reservoir engineering, Nat. Phys. 13, 465 (2017).
  • Yin et al. [2020] M.-J. Yin, T. Wang, X.-T. Lu, T. Li, Y.-B. Wang, X.-F. Zhang, W.-D. Li, A. Smerzi, and H. Chang, Rabi spectroscopy and sensitivity of a Floquet engineered optical lattice clock, arXiv:2007.00851 (2020).
  • Li et al. [2020a] X. Li, D. Dreon, P. Zupancic, A. Baumgärtner, A. Morales, W. Zheng, N. R. Cooper, T. Donner, and T. Esslinger, Measuring the dynamics of a first order structural phase transition between two configurations of a superradiant crystal (2020a), arXiv:2004.08398 [cond-mat.quant-gas] .
  • Pino et al. [2015] J. D. Pino, J. Feist, and F. J. Garciavidal, Signatures of Vibrational Strong Coupling in Raman Scattering, J. Phys. Chem. C 119, 29132 (2015).
  • Shalabney et al. [2015] A. Shalabney, J. George, J. A. Hutchison, G. Pupillo, C. Genet, and T. W. Ebbesen, Coherent coupling of molecular resonators with a microcavity mode, Nat. Commun. 6, 5981 (2015).
  • Sukharev and Nitzan [2017] M. Sukharev and A. Nitzan, Optics of exciton-plasmon nanomaterials, J. Phys. Condens. Matter 29, 443003 (2017).
  • Pietikäinen et al. [2017] I. Pietikäinen, S. Danilin, K. S. Kumar, A. Vepsäläinen, D. S. Golubev, J. Tuorila, and G. S. Paraoanu, Observation of the Bloch-Siegert shift in a driven quantum-to-classical transition, Phys. Rev. B 96, 020501 (2017).
  • Forn-Díaz et al. [2019] P. Forn-Díaz, L. Lamata, E. Rico, J. Kono, and E. Solano, Ultrastrong coupling regimes of light-matter interaction, Rev. Mod. Phys. 91, 025005 (2019).
  • Kockum et al. [2019] A. F. Kockum, A. Miranowicz, S. De Liberato, S. Savasta, and F. Nori, Ultrastrong coupling between light and matter, Nat. Rev. Phys. 1, 19 (2019).
  • Mukamel [1995] S. Mukamel, Principles of nonlinear optical spectroscopy (Oxford University Press, New York, 1995).
  • Kohler [2018] S. Kohler, Dispersive readout: Universal theory beyond the rotating-wave approximation, Phys. Rev. A 98, 023849 (2018).
  • Gu and Franco [2018] B. Gu and I. Franco, Optical absorption properties of laser-driven matter, Phys. Rev. A 98, 063412 (2018).
  • Cabra et al. [2020] G. Cabra, I. Franco, and M. Y. Galperin, Optical properties of periodically driven open nonequilibrium quantum systems, J. Chem. Phys. 152, 094101 (2020).
  • Engelhardt and Cao [tion] G. Engelhardt and J. Cao, Signatures of dynamical symmetries in Floquet response theory and spontaneous emission processes (in preparation).
  • Kumar et al. [2020] A. Kumar, M. Rodriguez-Vega, T. Pereg-Barnea, and B. Seradjeh, Linear response theory and optical conductivity of Floquet topological insulators, Phys. Rev. B 101, 174314 (2020).
  • Altland and Zirnbauer [1997] A. Altland and M. R. Zirnbauer, Nonstandard symmetry classes in mesoscopic normal-superconducting hybrid structures, Phys. Rev. B 55, 1142 (1997).
  • Langbehn et al. [2017] J. Langbehn, Y. Peng, L. Trifunovic, F. von Oppen, and P. W. Brouwer, Reflection-symmetric second-order topological insulators and superconductors, Phys. Rev. Lett. 119, 246401 (2017).
  • Kawabata et al. [2019] K. Kawabata, T. Bessho, and M. Sato, Classification of exceptional points and non-Hermitian topological semimetals, Phys. Rev. Lett. 123, 066405 (2019).
  • Duan et al. [2020] C. Duan, C.-Y. Hsieh, J. Liu, J. Wu, and J. Cao, Unusual transport properties with noncommutative system–bath coupling operators, J. Phys. Chem. Lett. 11, 4080 (2020).
  • Thingna et al. [2020] J. Thingna, D. Manzano, and J. Cao, Magnetic field induced symmetry breaking in nonequilibrium quantum networks, New J. Phys. 22, 083026 (2020).
  • Wu et al. [2013] J. Wu, R. J. Silbey, and J. Cao, Generic Mechanism of Optimal Energy Transfer Efficiency: A Scaling Theory of the Mean First-Passage Time in Exciton Systems, Phys. Rev. Lett. 110, 200402 (2013).
  • Chuang et al. [2016] C. Chuang, C. K. Lee, J. M. Moix, J. Knoester, and J. Cao, Quantum diffusion on molecular tubes: Universal scaling of the 1D to 2D transition, Phys. Rev. Lett. 116, 196803 (2016).
  • Thingna et al. [2016] J. Thingna, D. Manzano, and J. Cao, Dynamical signatures of molecular symmetries in nonequilibrium quantum transport, Sci. Rep. 6, 28027 (2016).
  • Engelhardt and Cao [2019] G. Engelhardt and J. Cao, Tuning the Aharonov-Bohm effect with dephasing in nonequilibrium transport, Phys. Rev. B 99, 075436 (2019).
  • Fleming [2009] I. Fleming, Molecular orbitals and organic chemical reactions. Student edition (Wiley-VCH, Weinheim, Germany, 2009).
  • Roy and Harper [2017] R. Roy and F. Harper, Periodic table for Floquet topological insulators, Phys. Rev. B 96, 155118 (2017).
  • Peng and Refael [2019] Y. Peng and G. Refael, Floquet second-order topological insulators from nonsymmorphic space-time symmetries, Phys. Rev. Lett. 123, 016806 (2019).
  • Engelhardt et al. [2016] G. Engelhardt, M. Benito, G. Platero, and T. Brandes, Topological instabilities in ac-driven bosonic systems, Phys. Rev. Lett. 117, 045302 (2016).
  • Engelhardt et al. [2017] G. Engelhardt, M. Benito, G. Platero, and T. Brandes, Topologically enforced bifurcations in superconducting circuits, Phys. Rev. Lett. 118, 197702 (2017).
  • Gómez-León and Platero [2012] A. Gómez-León and G. Platero, Transport blocking and topological phases using ac magnetic fields, Phys. Rev. B 85, 245319 (2012).
  • Yuen-Zhou et al. [2016] J. Yuen-Zhou, S. K. Saikin, T. Zhu, M. C. Onbasli, C. A. Ross, V. Bulovic, and M. A. Baldo, Plexciton Dirac points and topological modes, Nat. Commun. 7, 11783 (2016).
  • Yan et al. [2019] Y. Yan, Z. Lü, J. Y. Luo, and H. Zheng, Role of generalized parity in the symmetry of the fluorescence spectrum from two-level systems under periodic frequency modulation, Phys. Rev. A 100, 013823 (2019).
  • Yan et al. [2018] Y. Yan, Z. Lü, J. Y. Luo, and H. Zheng, Multiphoton-resonance-induced fluorescence of a strongly driven two-level system under frequency modulation, Phys. Rev. A 97, 033817 (2018).
  • Chinzei and Ikeda [2020] K. Chinzei and T. N. Ikeda, Time crystals protected by Floquet dynamical symmetry in Hubbard models, Phys. Rev. Lett. 125, 060601 (2020).
  • Buča et al. [2019] B. Buča, J. Tindall, and D. Jaksch, Non-stationary coherent quantum many-body dynamics through dissipation, Nat. Commun. 10, 1730 (2019).
  • Grossmann et al. [1991] F. Grossmann, T. Dittrich, P. Jung, and P. Hänggi, Coherent destruction of tunneling, Phys. Rev. Lett. 67, 516 (1991).
  • Bavli and Metiu [1993] R. Bavli and H. Metiu, Properties of an electron in a quantum double well driven by a strong laser: Localization, low-frequency, and even-harmonic generation, Phys. Rev. A 47, 3299 (1993).
  • Alon et al. [1998] O. E. Alon, V. Averbukh, and N. Moiseyev, Selection rules for the high harmonic generation spectra, Phys. Rev. Lett. 80, 3743 (1998).
  • Alon et al. [2000] O. E. Alon, V. Averbukh, and N. Moiseyev, High harmonic generation of soft x-rays by carbon nanotubes, Phys. Rev. Lett. 85, 5218 (2000).
  • Alon [2002] O. E. Alon, Dynamical symmetries of time-periodic hamiltonians, Phys. Rev. A 66, 013414 (2002).
  • [41] G. Engelhardt and J. Cao, Signatures of dynamical symmetries in floquet response theory and spontaneous emission processes, (in preparation) .
  • Kohler et al. [1998] S. Kohler, R. Utermann, P. Hänggi, and T. Dittrich, Coherent and incoherent chaotic tunneling near singlet-doublet crossings, Phys. Rev. E 58, 7219 (1998).
  • Blattmann et al. [2015] R. Blattmann, P. Hänggi, and S. Kohler, Qubit interference at avoided crossings: The role of driving shape and bath coupling, Phys. Rev. A 91, 042109 (2015).
  • Shirai et al. [2015] T. Shirai, T. Mori, and S. Miyashita, Condition for emergence of the Floquet-Gibbs state in periodically driven open systems, Phys. Rev. E 91, 030101 (2015).
  • Engelhardt et al. [2019] G. Engelhardt, G. Platero, and J. Cao, Discontinuities in Driven Spin-Boson Systems due to Coherent Destruction of Tunneling: Breakdown of the Floquet-Gibbs Distribution, Phys. Rev. Lett. 123, 120602 (2019).
  • Xu and Cao [2016] D. Xu and J. Cao, Non-canonical distribution and non-equilibrium transport beyond weak system-bath coupling regime: A polaron transformation approach, Front. Phys. 11, 110308 (2016).
  • [47] One can construct a Hamiltonian fulfilling a time-rotational symmetry defined by H^0(t)=jmfj(tmτN)A^j(m),\hat{H}_{0}(t)=\sum_{j}\sum_{m}f_{j}\left(t-m\frac{\tau}{N}\right)\hat{A}_{j}^{(m)}, where A^j(m)=R^mA^j(0)R^m\hat{A}_{j}^{(m)}=\hat{R}^{m}\hat{A}_{j}^{(0)}\hat{R}^{m} for arbitrary Hermitian operators A^j(0)\hat{A}_{j}^{(0)}, and the arbitrary functions fj(t+τ)=fj(t)f_{j}\left(t+\tau\right)=f_{j}\left(t\right) shall be τ\tau periodic.
  • [48] See Supplemental Materials for a detailed derivation of the symmetry-protected selection rules.
  • Bernevig and Taylor [2013] B. A. Bernevig and L. H. Taylor, Topological Insulators and Topological Superconductors (Princeton University Press, 2013).
  • Gong et al. [2009] J. Gong, L. Morales-Molina, and P. Hänggi, Many-body coherent destruction of tunneling, Phys. Rev. Lett. 103, 133002 (2009).
  • Wang et al. [2019] T. Wang, Z. Zhang, L. Xiang, Z. Jia, P. Duan, Z. Zong, Z. Sun, Z. Dong, J. Wu, Y. Yin, and G. Guo, Experimental Realization of a Fast Controlled-Z Gate via a Shortcut to Adiabaticity, Phys. Rev. Applied 11, 034030 (2019).
  • Zha et al. [2020] C. Zha, V. M. Bastidas, M. Gong, Y. Wu, H. Rong, R. Yang, Y. Ye, S. Li, Q. Zhu, S. Wang, Y. Zhao, F. Liang, J. Lin, Y. Xu, C.-Z. Peng, J. Schmiedmayer, K. Nemoto, H. Deng, W. J. Munro, X. Zhu, and J.-W. Pan, Ergodic-localized junctions in a periodically-driven spin chain, arXiv:2001.09169  (2020).
  • Magazzù et al. [2018] L. Magazzù, P. Forndiaz, R. Belyansky, J. L. Orgiazzi, M. A. Yurtalan, M. Otto, A. Lupascu, C. Wilson, and M. Grifoni, Probing the strongly driven spin-boson model in a superconducting quantum circuit, Nat. Commun. 9, 1403 (2018).
  • Chen et al. [2020] M.-B. Chen, B.-C. Wang, S. Kohler, Y. Kang, T. Lin, S.-S. Gu, H.-O. Li, G.-C. Guo, X. Hu, H.-W. Jiang, G. Cao, and G.-P. Guo, Double resonance landau-zener-stückelburg-majorana interference in circuit qed (2020), arXiv:2011.03697.
  • Herrera and Spano [2016] F. Herrera and F. C. Spano, Cavity-controlled chemistry in molecular ensembles, Phys. Rev. Lett. 116, 238301 (2016).
  • Li et al. [2020b] T. E. Li, A. Nitzan, and J. E. Subotnik, On the origin of ground-state vacuum-field catalysis: Equilibrium consideration, J. Chem. Phys. 152, 234107 (2020b).

Supplementary Material

I Floquet equation and dynamical symmetries

In this supplemental information, we provide details to establish the symmetry-protected dark state conditions, which are sketched in the main text. The starting point is the Floquet equation

[H^0(t)iddt]|uμ(t)=ϵμ|uμ(t),\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right]\left|u_{\mu}(t)\right>=\epsilon_{\mu}\left|u_{\mu}(t)\right>, (14)

and the time-spatial symmetry relation for a Hamiltonian in the Floquet space

Σ^[H^0(ts+βSt)iddt]Σ^1=αS[H^0(t)iddt],\hat{\Sigma}\left[\hat{H}_{0}(t_{s}+\beta_{S}t)-i\frac{d}{dt}\right]\hat{\Sigma}^{-1}=\alpha_{S}\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right], (15)

where the symmetry is specified by the spatial operator Σ^\hat{\Sigma}, the time-shift tSt_{S}, αS=±1\alpha_{S}=\pm 1 and βS=±1\beta_{S}=\pm 1.

II Dynamical Dipole elements

The susceptibility, introduced in the main text,

χ~n(ω)=iλ2ν,μ,mVν,μ(nm)Vμ,ν(m)(pνpμ)ϵμϵν+mΩωiγν,μ(m),\tilde{\chi}_{n}(\omega)=i\lambda^{2}\sum_{\nu,\mu,m}\frac{V_{\nu,\mu}^{(-n-m)}V_{\mu,\nu}^{(m)}\left(p_{\nu}-p_{\mu}\right)}{\epsilon_{\mu}-\epsilon_{\nu}+m\Omega-\omega-i\gamma_{\nu,\mu}^{(m)}}, (16)

is expressed in terms of the dynamical dipole elements

Vμ,ν(n)=1τ0τuμ|V^(t)|uνeinΩt𝑑t,V_{\mu,\nu}^{(n)}=\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}\right|\hat{V}(t)\left|u_{\nu}\right>e^{-in\Omega t}dt, (17)

which fulfill

Vμ,ν(n)\displaystyle V_{\mu,\nu}^{(n)} =\displaystyle= 1τ0τuμ(t)|V^|uν(t)einΩt𝑑t,\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t)\right|\hat{V}\left|u_{\nu}(t)\right>e^{-in\Omega t}dt, (18)
=\displaystyle= 1τ0τ(uν(t)|V^|uμ(t))einΩt𝑑t,\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left(\left<u_{\nu}(t)\right|\hat{V}\left|u_{\mu}(t)\right>\right)^{*}e^{-in\Omega t}dt,
=\displaystyle= (1τ0τuν(t)|V^|uμ(t)einΩtdt,)\displaystyle\left(\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\nu}(t)\right|\hat{V}\left|u_{\mu}(t)\right>e^{in\Omega t}dt,\right)
=\displaystyle= [Vν,μ(n)].\displaystyle\left[V_{\nu,\mu}^{(-n)}\right]^{*}.

This relation will be used to prove the symmetry-protected dark state condition based on a dynamical particle-hole symmetry in Sec. IV.

III Rotational symmetry

Here we present the derivation of the dark state condition imposed by a dynamical rotational symmetry. For a unitary operator Σ^=R^\hat{\Sigma}=\hat{R}, tS=tP=τ/Nt_{S}=t_{P}=\tau/N with a positive integer NN, and αS=βS=1\alpha_{S}=\beta_{S}=1, the general dynamical symmetry condition Eq. (15) specifies to a rotational symmetry

R^H^(t+τN)R^=H^(t),\hat{R}\hat{H}\left(t+\frac{\tau}{N}\right)\hat{R}^{\dagger}=\hat{H}(t), (19)

Applying Eq. (19) to Eq. (14), we find

[H^0(t)iddt]R^|uμ(t+tR)=ϵμR^|uμ(t+tR).\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right]\hat{R}\left|u_{\mu}(t+t_{R})\right>=\epsilon_{\mu}\hat{R}\left|u_{\mu}(t+t_{R})\right>. (20)

This implies that every Floquet state is also an eigenstate of the rotational symmetry operator, such that

R^|uμ(t+tR)=πμ(R)|uμ(t),\hat{R}\;\left|u_{\mu}\left(t+t_{R}\right)\right>=\pi_{\mu}^{(R)}\left|u_{\mu}(t)\right>, (21)

with eigenvalues πμ(R)=ei2πmμ/N\pi_{\mu}^{(R)}=e^{i2\pi m_{\mu}/N} and integers mμ={0,N1}m_{\mu}=\left\{0,N-1\right\}. We require that the transition dipole operator obeys R^V^R^=αV(R)V^\hat{R}^{\dagger}\hat{V}\hat{R}=\alpha_{V}^{(R)}\hat{V} with αV(R)=±1\alpha_{V}^{(R)}=\pm 1. Combining with Eq. (14), the dynamical dipole element fulfills

Vν,μ(n)\displaystyle V_{\nu,\mu}^{(n)} =\displaystyle= 1τ0τuν(t)|V^|uμ(t)einΩt𝑑t,\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\nu}(t)\right|\hat{V}\left|u_{\mu}(t)\right>e^{-in\Omega t}dt,
=\displaystyle= 1τm=0N1mτ/N(m+1)τ/Nuν(t)|V^|uμ(t)einΩt𝑑t,\displaystyle\frac{1}{\tau}\sum_{m=0}^{N-1}\int_{m\tau/N}^{(m+1)\tau/N}\left<u_{\nu}(t)\right|\hat{V}\left|u_{\mu}(t)\right>e^{-in\Omega}tdt,
=\displaystyle= 1τm=0N10τ/Nuν(t+mτN)|V^|uμ(t+mτN)einΩ(t+mτN)𝑑t,\displaystyle\frac{1}{\tau}\sum_{m=0}^{N-1}\int_{0}^{\tau/N}\left<u_{\nu}\left(t+m\frac{\tau}{N}\right)\right|\hat{V}\left|u_{\mu}\left(t+m\frac{\tau}{N}\right)\right>e^{-in\Omega\left(t+m\frac{\tau}{N}\right)}dt,
=\displaystyle= 1τm=0N1ei2πmn/N0τ/N[(πμ(R))πν(R)]muν(t)|R^mV^R^m|uμ(t)einΩt𝑑t,\displaystyle\frac{1}{\tau}\sum_{m=0}^{N-1}e^{-i2\pi mn/N}\int_{0}^{\tau/N}\left[\left(\pi_{\mu}^{(R)}\right)^{*}\pi_{\nu}^{(R)}\right]^{m}\left<u_{\nu}\left(t\right)\right|\hat{R}^{\dagger m}\hat{V}\hat{R}^{m}\left|u_{\mu}\left(t\right)\right>e^{-in\Omega t}dt,
=\displaystyle= Vν,μ(n)Nm=0N1ei2πnm/N(αV(R))meim(mνmμ)2π/N.\displaystyle\frac{V_{\nu,\mu}^{(n)}}{N}\sum_{m=0}^{N-1}e^{-i2\pi nm/N}\left(\alpha_{V}^{(R)}\right)^{m}e^{im\left(m_{\nu}-m_{\mu}\right)2\pi/N}.

Recalling that αV(R)=±1\alpha_{V}^{(R)}=\pm 1, the last line establishes the dark state condition for a dynamical rotational symmetry

Vν,μ(n){1ifei2πN(mμmν+n)αV(R)=1,0else,V_{\nu,\mu}^{(n)}\propto\begin{cases}1&\text{if}\;e^{i\frac{2\pi}{N}\left(m_{\mu}-m_{\nu}+n\right)}\alpha_{V}^{(R)}=1,\\ 0&\text{else},\end{cases} (22)

which is presented in the main text.

IV Pariticle-hole symmetry

Here, we provide details to establish the dark state condition induced by a dynamical particle-hole symmetry

P^H^(tP+t)P^=H^(t),\hat{P}\hat{H}^{*}(t_{P}+t)\hat{P}^{\dagger}=-\hat{H}(t), (23)

where tPt_{P} assumes the values tP=N1τ/2N2t_{P}=N_{1}\tau/2N_{2} for integers N1=0,1N_{1}=0,1 and N21N_{2}\geq 1 and depends on the specific system. We obtain Eq. (23) from the general definition of the dynamical symmetries Eq. (15) for Σ^=P^κ^\hat{\Sigma}=\hat{P}\hat{\kappa}, with the complex conjugation operator κ^\hat{\kappa} and the unitary operator P^\hat{P}, αS=βS=1\alpha_{S}=-\beta_{S}=-1 and tS=tPt_{S}=t_{P}. Applying the definition in Eq. (23) to the Floquet equation Eq. (14), we find

[H^0(t)iddt]P^|uμ(tP+t)=ϵμP^|uμ(tP+t),\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right]\hat{P}\left|u_{\mu}(t_{P}+t)\right>^{*}=-\epsilon_{\mu}\hat{P}\left|u_{\mu}(t_{P}+t)\right>^{*}, (24)

which indicates that for every Floquet state |uμ(t)\left|u_{\mu}(t)\right> with quasienergy ϵμ\epsilon_{\mu}, there is a symmetry-related partner

|uμ(t)=πμ(P)P^|uμ(tP+t)\left|u_{\mu^{\prime}}(t)\right>=\pi_{\mu}^{(P)}\hat{P}\left|u_{\mu}(t_{P}+t)\right>^{*} (25)

with quasienergy ϵμ=ϵμ\epsilon_{\mu^{\prime}}=-\epsilon_{\mu}, and a gauge-dependent phase factor πμ(P)\pi_{\mu}^{(P)}. The phase factor cannot be removed by a simple gauge transformation as the two Floquet states μ\mu and μ\mu^{\prime} are coupled. However, we can apply Eq. (25) twice and obtain

|uμ(t)=πμ(P)πμ(P)P^P^|uμ(2tP+t).\left|u_{\mu}(t)\right>=\pi_{\mu}^{(P)}\pi_{\mu^{\prime}}^{(P)*}\hat{P}\hat{P}^{*}\left|u_{\mu}(2t_{P}+t)\right>. (26)

In general, πμ(P)πμ(P)\pi_{\mu}^{(P)}\neq\pi_{\mu^{\prime}}^{(P)}, and we cannot find a gauge transformation such that πμ(P)πμ(P)=1\pi_{\mu}^{(P)}\pi_{\mu^{\prime}}^{(P)*}=1. Only if P^P^=𝟙\hat{P}\hat{P}^{*}=\mathbbm{1} and tP{0,τ/2}t_{P}\in\left\{0,\tau/2\right\}, we have πμ(P)πμ(P)=1\pi_{\mu^{\prime}}^{(P)*}\pi_{\mu}^{(P)}=1, which will be used in the evaluation of the dynamical dipole elements. Furthermore, we require P^V^P^=αV(P)V^\hat{P}^{\dagger}\hat{V}\hat{P}=\alpha_{V}^{(P)}\hat{V}^{*} with αV(P)=±1\alpha_{V}^{(P)}=\pm 1. Using Eq. (25) and Eq. (26), we find that the dynamical dipole element obeys

Vμ,μ(n)\displaystyle V_{\mu,\mu^{\prime}}^{(n)} =\displaystyle= 1τ0τuμ(t)|V^|uμ(t)einΩt𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t)\right|\hat{V}\left|u_{\mu^{\prime}}(t)\right>e^{-in\Omega t}dt (27)
=\displaystyle= 1τ0τπμ(P)πμ(P)uμ(t+tP)|P^V^P^|uμ(t+tP)einΩt𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\pi_{\mu^{\prime}}^{(P)*}\pi_{\mu}^{(P)}\left<u_{\mu^{\prime}}(t+t_{P})\right|^{*}\hat{P}^{\dagger}\hat{V}\hat{P}\left|u_{\mu}(t+t_{P})\right>^{*}e^{-in\Omega t}dt
=\displaystyle= 1τ0ταV(P)uμ(t)|V^|uμ(t)einΩ(ttP)𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\alpha_{V}^{(P)}\left<u_{\mu^{\prime}}(t)\right|^{*}\hat{V}^{*}\left|u_{\mu}(t)\right>^{*}e^{-in\Omega(t-t_{P})}dt
=\displaystyle= 1ταV(P)einΩtPdt0τ(uμ(t)|V^|uμ(t))einΩt𝑑t\displaystyle\frac{1}{\tau}\alpha_{V}^{(P)}e^{-in\Omega t_{P}}dt\int_{0}^{\tau}\left(\left<u_{\mu^{\prime}}(t)\right|\hat{V}\left|u_{\mu}(t)\right>\right)^{*}e^{in\Omega t}dt
=\displaystyle= αV(P)einΩtPVμ,μ(n)=αV(P)einΩtPVμ,μ(n),\displaystyle\alpha_{V}^{(P)}e^{-in\Omega t_{P}}V_{\mu^{\prime},\mu}^{(-n)*}=\alpha_{V}^{(P)}e^{-in\Omega t_{P}}V_{\mu,\mu^{\prime}}^{(n)},

where we have used Eq. (18) in the last line. From line two to line three we have used the τ\tau-periodicity of the integrand. This proves the particle-hole symmetry induced condition for dark states

V^μ,μ(m){0ifαV(P)einΩtP=1,μ,μ sym. rel.1else,\hat{V}_{\mu,\mu^{\prime}}^{(m)}\propto\begin{cases}0&\text{if}\;\alpha_{V}^{(P)}e^{-in\Omega t_{P}}=-1,\quad\mu,\mu^{\prime}\text{ sym. rel.}\\ 1&\text{else},\end{cases} (28)

which is presented in the main text.

V Chiral symmetry

Here we derive a constrain for the dynamical dipole elements under a dynamical chiral symmetry, and explain why the chiral symmetry on its own does not imply a symmetry-protected dark state condition. For a unitary Σ^=C^\hat{\Sigma}=\hat{C}, αS=βS=1\alpha_{S}=\beta_{S}=-1 and an arbitrary tS=tCt_{S}=t_{C}, the general dynamical symmetry relation Eq. (15) defines a chiral symmetry

C^H^(tCt)C^=H^(t).\hat{C}\hat{H}(t_{C}-t)\hat{C}^{\dagger}=-\hat{H}(t). (29)

Applying this definition to the Floquet equation Eq. (14), we find

[H^0(t)iddt]C^|uμ(tCt)=ϵμC^|uμ(tCt).\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right]\hat{C}\left|u_{\mu}(t_{C}-t)\right>=-\epsilon_{\mu}\hat{C}\left|u_{\mu}(t_{C}-t)\right>. (30)

Similar to the particle-hole symmetry, Eq. (30) implies that for every Floquet state |uμ(t)\left|u_{\mu}(t)\right> with quasienergy ϵμ\epsilon_{\mu}, there is a symmetry-related partner

|uμ(t)=πμ(C)C^|uμ(tCt),\left|u_{\mu^{\prime}}(t)\right>=\pi_{\mu}^{(C)}\hat{C}\left|u_{\mu}(t_{C}-t)\right>, (31)

with quasienergy ϵμ=ϵμ\epsilon_{\mu^{\prime}}=-\epsilon_{\mu} and a gauge-dependent phase factor πμ(C)\pi_{\mu}^{(C)}. The πμ(C),πμ(C)\pi_{\mu}^{(C)},\pi_{\mu^{\prime}}^{(C)} can not be arbitrarily changed by a gauge transformation. Chiral symmetry is not sufficient to determine a constrain for the dynamical dipole elements. To see this we conjugate Eq. (31) and insert it into itself, and obtain

|uμ(t)=πμ(C)πμ(C)C^C^|uμ(t).\left|u_{\mu}(t)\right>=\pi_{\mu}^{(C)}\pi_{\mu^{\prime}}^{(C)*}\hat{C}\hat{C}^{*}\left|u_{\mu}(t)\right>^{*}. (32)

When requiring C^C^=𝟙\hat{C}^{*}\hat{C}=\mathbbm{1}, we can find a gauge transformation, for which πμ(C)πμ(C)=1\pi_{\mu}^{(C)}\pi_{\mu^{\prime}}^{(C)*}=1. Using this and C^VC^=αV(C)V^\hat{C}^{\dagger}V\hat{C}=\alpha_{V}^{(C)}\hat{V} with αV(C)=±1\alpha_{V}^{(C)}=\pm 1 to evaluate the dynamical dipole elements, we find

Vμ,μ(n)\displaystyle V_{\mu,\mu^{\prime}}^{(n)} =\displaystyle= 1τ0τuμ(t)|V^|uμ(t)einΩt𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t)\right|\hat{V}\left|u_{\mu^{\prime}}(t)\right>e^{-in\Omega t}dt (33)
=\displaystyle= 1τ0τπμ(C)πμ(C)uμ(tCt)|C^V^C^|uμ(tCt)einΩt𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\pi_{\mu^{\prime}}^{(C)*}\pi{\mu^{\prime}}^{(C)}\left<u_{\mu^{\prime}}(t_{C}-t)\right|\hat{C}^{\dagger}\hat{V}\hat{C}\left|u_{\mu}(t_{C}-t)\right>e^{-in\Omega t}dt
=\displaystyle= 1τ0ταV(C)uμ(t)|V^|uμ(t)einΩ(tCt)𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\alpha_{V}^{(C)}\left<u_{\mu^{\prime}}(t)\right|\hat{V}\left|u_{\mu}(t)\right>e^{-in\Omega(t_{C}-t)}dt
=\displaystyle= einΩtC1ταV(C)0τuμ(t)|V^|uμ(t)einΩt𝑑t\displaystyle e^{-in\Omega t_{C}}\frac{1}{\tau}\alpha_{V}^{(C)}\int_{0}^{\tau}\left<u_{\mu^{\prime}}(t)\right|\hat{V}\left|u_{\mu}(t)\right>e^{in\Omega t}dt
=\displaystyle= einΩtCαV(C)Vμ,μ(n)\displaystyle e^{-in\Omega t_{C}}\alpha_{V}^{(C)}V_{\mu^{\prime},\mu}^{(-n)}
=\displaystyle= einΩtCαV(C)Vμ,μ(n),\displaystyle e^{-in\Omega t_{C}}\alpha_{V}^{(C)}V_{\mu,\mu^{\prime}}^{(n)*},

where we have used Eq. (18) in the last line, and the τ\tau periodicity of the integrand in line three. This relation alone does not imply a dark-state condition for the dynamical dipole elements. If Vμ,μ(n)V_{\mu,\mu^{\prime}}^{(n)*} was real (e.g., because of another symmetry relation), Eq. (33) would indeed imply a dark state condition, but chiral symmetry alone is not sufficient to guarantee this.

VI Time-reversal symmetry

Similar to the chiral symmetry, the time-reversal symmetry alone does not impose a dark state condition. A time-reversal symmetry in Eq. (15) is defined for Σ^=T^κ^\hat{\Sigma}=\hat{T}\hat{\kappa}, with a unitary operator T^\hat{T}, the complex conjugation operator κ^\hat{\kappa}, αS=βS=1\alpha_{S}=-\beta_{S}=1, and an arbitrary tS=tTt_{S}=t_{T} , so that

T^H^(tTt)T^=H^(t).\hat{T}\hat{H}^{*}(t_{T}-t)\hat{T}^{\dagger}=\hat{H}(t). (34)

Applying this definition to the Floquet equation, we find

[H^0(t)iddt]T^|uμ(tTt)=ϵμT^|uμ(tTt),\left[\hat{H}_{0}(t)-i\frac{d}{dt}\right]\hat{T}\left|u_{\mu}(t_{T}-t)\right>^{*}=\epsilon_{\mu}\hat{T}\left|u_{\mu}(t_{T}-t)\right>^{*}, (35)

Thus, all Floquet states fulfill

|uμ(t)=πμ(T)T^|uμ(tTt)\left|u_{\mu}(t)\right>=\pi_{\mu}^{(T)}\hat{T}\left|u_{\mu}(t_{T}-t)\right>^{*} (36)

with a phase factor πμ(T)\pi_{\mu}^{(T)}, which can be transformed away by a gauge transformation. Using this to evaluate the dipole elements, we find

Vμ,ν(n)\displaystyle V_{\mu,\nu}^{(n)} =\displaystyle= 1τ0τuμ(t)|V^|uν(t)einΩt𝑑t,\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t)\right|\hat{V}\left|u_{\nu}(t)\right>e^{-in\Omega t}dt,
=\displaystyle= 1τ0τuμ(tTt)|T^V^T^|uν(tTt)einΩt𝑑t,\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t_{T}-t)\right|^{*}\hat{T}^{\dagger}\hat{V}\hat{T}\left|u_{\nu}(t_{T}-t)\right>^{*}e^{-in\Omega t}dt,
=\displaystyle= 1τ0ταV(T)uμ(t)|V^|uν(t)einΩ(tTt)𝑑t,\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\alpha_{V}^{(T)}\left<u_{\mu}(t)\right|^{*}\hat{V}^{*}\left|u_{\nu}(t)\right>^{*}e^{-in\Omega(t_{T}-t)}dt,
=\displaystyle= einΩtTαV(T)(1τ0τuμ(t)|V^|uν(t)einΩt𝑑t),\displaystyle e^{-in\Omega t_{T}}\alpha_{V}^{(T)}\left(\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t)\right|\hat{V}\left|u_{\nu}(t)\right>e^{-in\Omega t}dt\right)^{*},
=\displaystyle= einΩtTαV(T)Vμ,ν(n),\displaystyle e^{-in\Omega t_{T}}\alpha_{V}^{(T)}V_{\mu,\nu}^{(n)*},

Similar to the chiral symmetry, this does not establish a dark state condition.

VII Combined chiral symmetry and time reversal symmetry

Importantly, the chiral-symmetry relation Eq. (33) and the time-reversal symmetry relation Eq. (LABEL:eq:dynamicalDipolElementsTRSymmetrySI) together do provide a sufficient condition for a symmetry-protected dark state. In the derivations in Sec. V, and Sec. VI we could choose the gauge fields independently, such that πμ(C)πμ(C)=1\pi_{\mu}^{(C)}\pi_{\mu}^{(C)*}=1 and πμ(T)πμ(T)=1\pi_{\mu}^{(T)}\pi_{\mu}^{(T)*}=1 . Yet, these gauges conditions can not be simultaneous fulfilled in general. However, the combination of a time-reversal symmetry and a chiral symmetry defines a particle-hole symmetry with P^=C^T^\hat{P}=\hat{C}\hat{T} and tP=tTtCt_{P}=t_{T}-t_{C}. When we additionally require that C^C^=𝟙\hat{C}\hat{C}^{*}=\mathbbm{1}, T^T^=𝟙\hat{T}\hat{T}^{*}=\mathbbm{1}, [C^,T^]=0\left[\hat{C},\hat{T}\right]=0, such that P^P^=𝟙\hat{P}^{*}\hat{P}=\mathbbm{1}, and tTtC=0,τ/2t_{T}-t_{C}=0,\tau/2, all requirements for symmetry-protected dark states in Sec. IV are fulfilled.

VIII Symmetry-induced transparency

Refer to caption
Figure 4: (a) and (b) depict the quasienergies for hx/Ω=0h_{\rm x}/\Omega=0 and hx/Ω=0.05h_{\rm x}/\Omega=0.05, respectively. (c) and (d) show the susceptibility χ~0(ωp)\tilde{\chi}_{0}(\omega_{\rm p}) depicted by a color gradient for hx/Ω=0h_{\rm x}/\Omega=0 and hx/Ω=0.05h_{\rm x}/\Omega=0.05, respectively. Dynamical parity spDS are marked by dashed lines, while transitions suppressed by the symmetry-protected transparency are marked by dashed-dotted lines. In (d), one particle-hole symmetry is slightly broken. Consequently, the symmetry-protected transparency becomes a symmetry-induced transparency (siT) appearing at a quasienergy crossing, which is magnified in panel (e).

The symmetry-protected dark states and symmetry-protected dark bands introduced before are defined by vanishing dynamical dipole elements in Eq. (17). Here we want to introduce a distinct transparency effect, which is also protected by dynamical symmetries. In order to distinguish it from the symmetry-protected dark states and symmetry-protected dark bands, we coin it symmetry-protected transparency. We also explain a weaker form of it, i.e., symmetry-induced transparency.

While a symmetry-protected dark state is generated by a vanishing dipole element, i.e., Vμ,ν(m)=0V^{(m)}_{\mu,\nu}=0, the symmetry-protected transparency is generated by a destructive interference of two transitions with non-vanishing dipole elements. More precisely, the transparency effect appears when for a given resonance ωmΩ\omega\approx m\Omega the corresponding two terms in the susceptibility in Eq. (16) cancel each other, which requires

Vμ,μ(nm)Vμ,μ(m)=Vμ,μ(nm)Vμ,μ(m),V^{(-n-m)}_{\mu^{\prime},\mu}V^{(m)}_{\mu,\mu^{\prime}}=V^{(-n-m)}_{\mu,\mu^{\prime}}V^{(m)}_{\mu^{\prime},\mu}, (38)

which will be investigated in the following.

The transparency effect appears as a consequence of two distinct particle-hole symmetries, if two symmetry-related resonances corresponding to the quasienergies ϵμ=ϵμ\epsilon_{\mu}=-\epsilon_{\mu^{\prime}} become degenerate, i.e., ϵμ=ϵμ=0\epsilon_{\mu^{\prime}}=\epsilon_{\mu}=0. We denote the two particle-hole symmetries with P^1=P^1\hat{P}_{1}=\hat{P}_{1}^{\dagger}, P^2=P^2\hat{P}_{2}=\hat{P}_{2}^{\dagger} with corresponding time shifts tP1,tP2t_{P_{1}},t_{P_{2}}, and require [P^1,P^2]=0\left[\hat{P}_{1},\hat{P}_{2}\right]=0 as well as P^1±P^2\hat{P}_{1}\neq\pm\hat{P}_{2} (i.e., they are distinct). The Floquet states μ\mu with ϵμ=0\epsilon_{\mu}=0 form a subspace of dimension two. As P^1\hat{P}_{1} and P^2\hat{P}_{2} commute, and P^i=P^i\hat{P}_{i}=\hat{P}_{i}^{\dagger}, there is a basis of this subspace such that

P^1|vμ(tP1+t)\displaystyle\hat{P}_{1}\left|v_{\mu}(t_{P_{1}}+t)\right>^{*} =\displaystyle= |vμ(t),\displaystyle\left|v_{\mu}(t)\right>,
P^1|vμ(tP1+t)\displaystyle\hat{P}_{1}\left|v_{\mu^{\prime}}(t_{P_{1}}+t)\right>^{*} =\displaystyle= |vμ(t),\displaystyle\left|v_{\mu^{\prime}}(t)\right>,
P^2|vμ(tP2+t)\displaystyle\hat{P}_{2}\left|v_{\mu}(t_{P_{2}}+t)\right>^{*} =\displaystyle= |vμ(t),\displaystyle\left|v_{\mu}(t)\right>,
P^2|vμ(tP2+t)\displaystyle\hat{P}_{2}\left|v_{\mu^{\prime}}(t_{P_{2}}+t)\right>^{*} =\displaystyle= |vμ(t).\displaystyle-\left|v_{\mu^{\prime}}(t)\right>. (39)

In this basis, there are no pairs of symmetry-related partners as introduced in Sec. IV. From Eq. (39) we obtain a basis with a symmetry-related pair by

|uμ(t)\displaystyle\left|u_{\mu}(t)\right> =\displaystyle= 12(|vμ(t)+|vμ(t)),\displaystyle\frac{1}{\sqrt{2}}\left(\left|v_{\mu}(t)\right>+\left|v_{\mu^{\prime}}(t)\right>\right),
|uμ(t)\displaystyle\left|u_{\mu^{\prime}}(t)\right> =\displaystyle= 12(|vμ(t)|vμ(t)),\displaystyle\frac{1}{\sqrt{2}}\left(\left|v_{\mu}(t)\right>-\left|v_{\mu^{\prime}}(t)\right>\right), (40)

which are related by the symmetry operations as

P^1|uμ(tP1+t)\displaystyle\hat{P}_{1}\left|u_{\mu^{\prime}}(t_{P_{1}}+t)\right>^{*} =\displaystyle= |uμ(t),\displaystyle\left|u_{\mu}(t)\right>,
P^1|uμ(tP1+t)\displaystyle\hat{P}_{1}\left|u_{\mu}(t_{P_{1}}+t)\right>^{*} =\displaystyle= |uμ(t),\displaystyle\left|u_{\mu^{\prime}}(t)\right>,
P^2|uμ(tP2+t)\displaystyle\hat{P}_{2}\left|u_{\mu}(t_{P_{2}}+t)\right>^{*} =\displaystyle= |uμ(t),\displaystyle\left|u_{\mu}(t)\right>,
P^2|uμ(tP2+t)\displaystyle\hat{P}_{2}\left|u_{\mu^{\prime}}(t_{P_{2}}+t)\right>^{*} =\displaystyle= |uμ(t).\displaystyle\left|u_{\mu^{\prime}}(t)\right>. (41)

Thus, we have established symmetry-related pairs with respect to P^1\hat{P}_{1}. Please note that the gauges phases are fixed by the procedure, such that we do not consider them in the following calculations. Using the symmetry P^2\hat{P}_{2} to evaluate the dynamical dipole elements, we find

Vμ,μ(n)\displaystyle V_{\mu,\mu^{\prime}}^{(n)} =\displaystyle= 1τ0τuμ(t)|V^|uμ(t)einΩt𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t)\right|\hat{V}\left|u_{\mu^{\prime}}(t)\right>e^{-in\Omega t}dt
=\displaystyle= 1τ0τuμ(tP2+t)|P^2V^P^2|uμ(tP2+t)einΩt𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t_{P_{2}}+t)\right|^{*}\hat{P}_{2}^{\dagger}\hat{V}\hat{P}_{2}\left|u_{\mu^{\prime}}(t_{P_{2}}+t)\right>^{*}e^{-in\Omega t}dt
=\displaystyle= 1τ0ταV(P)uμ(t)|V^|uμ(t)einΩ(tP1+t)𝑑t\displaystyle\frac{1}{\tau}\int_{0}^{\tau}\alpha_{V}^{(P)}\left<u_{\mu}(t)\right|^{*}\hat{V}^{*}\left|u_{\mu^{\prime}}(t)\right>^{*}e^{-in\Omega(t_{P_{1}}+t)}dt
=\displaystyle= αV(P1)einΩtP1(1τ0τuμ(t)|V^|uμ(t)einΩt𝑑t)\displaystyle\alpha_{V}^{(P_{1})}e^{-in\Omega t_{P_{1}}}\left(\frac{1}{\tau}\int_{0}^{\tau}\left<u_{\mu}(t)\right|\hat{V}\left|u_{\mu^{\prime}}(t)\right>e^{in\Omega t}dt\right)^{*}
=\displaystyle= αV(P1)einΩtP1Vμ,μ(n).\displaystyle\alpha_{V}^{(P_{1})}e^{-in\Omega t_{P_{1}}}V_{\mu,\mu^{\prime}}^{(-n)*}.

Using P^1\hat{P}_{1} to evaluate the dipole elements based on Eq. (27), we find Vμ,μ(n)=αV(P2)einΩtP2Vμ,μ(n)V_{\mu,\mu^{\prime}}^{(n)}=\alpha_{V}^{(P_{2})}e^{-in\Omega t_{P_{2}}}V_{\mu^{\prime},\mu}^{(n)*}. Combining both results, we obtain

Vμ,μ(n)=Vμ,μ(n)αV(P2)αV(P1)einΩ(tP1tP2).V_{\mu^{\prime},\mu}^{(n)}=V_{\mu,\mu^{\prime}}^{(n)}\alpha_{V}^{(P_{2})}\alpha_{V}^{(P_{1})}e^{-in\Omega(t_{P_{1}}-t_{P_{2}})}. (43)

Inserting Eq. (43) into Eq. (38), we find the transparency condition

χ~n(mΩ){0ifeinΩ(tP1tP2)=1;ϵμ=ϵμ=01else,\tilde{\chi}_{n}(m\Omega)\propto\begin{cases}0&\text{if}\;e^{-in\Omega(t_{P_{1}}-t_{P_{2}})}=1;\epsilon_{\mu}=\epsilon_{\mu^{\prime}}=0\\ 1&\text{else},\end{cases} (44)

which is the condition presented in the main text. Noteworthy, for einΩ(tP1tP2)=1e^{-in\Omega(t_{P_{1}}-t_{P_{2}})}=-1 the two resonances can also add up constructively.

For illustration, we consider the ac-driven two-level system (TLS) which is introduced in the main text in Eq. (12). Its Hamiltonian reads

H^0(t)\displaystyle\hat{H}_{0}(t) =hx2σ^x+fΩ2cos(Ωt)σ^z,\displaystyle=\frac{h_{\rm x}}{2}\hat{\sigma}_{\rm x}+\frac{f_{\Omega}}{2}\cos\left(\Omega t\right)\hat{\sigma}_{\rm z}, (45)

where σ^x,σ^z\hat{\sigma}_{x},\hat{\sigma}_{\rm z} are the Pauli matrices, hxh_{x} is the tunneling amplitude, and fΩf_{\Omega} the driving strength. The dipole transition operator is V^=σ^x\hat{V}=\hat{\sigma}_{x}.

For hx=0h_{\rm x}=0, a particle-hole symmetry is defined for P^1=σ^z\hat{P}_{1}=\hat{\sigma}_{\rm z} and tP1=τ/2t_{P_{1}}=\tau/2, and a second particle-hole symmetry is given for P^2=𝟙\hat{P}_{2}=\mathbbm{1} and tP2=τ/2t_{P_{2}}=\tau/2. The quasienergy spectrum is flat, i.e., ϵμ=0\epsilon_{\mu}=0, as depicted in Fig. 4(a). As P^iσ^xP^i=(1)iσ^x\hat{P}_{i}\hat{\sigma}_{x}^{*}\hat{P}_{i}=(-1)^{i}\hat{\sigma}_{x}, a symmetry-protected transparency with χ~n(mΩ)=0\tilde{\chi}_{n}(m\Omega)=0 appears for all nn according to the condition Eq. (44). We note that the transitions for even mm, e.g. V^ν,μ(2)V^μ,ν(2)\hat{V}_{\nu,\mu}^{(-2)}\hat{V}_{\mu,\nu}^{(2)}, are additionally suppressed because of a dark state selection rule imposed by the dynamical parity symmetry introduced in the letter.

For a finite but small hx0h_{\rm x}\neq 0, the symmetry P^2\hat{P}_{2} is slightly broken. The quasienergy spectrum is not flat anymore as can be seen in Fig. 4(b). There is still a quasienergy crossing close to fΩ2.4Ωf_{\Omega}\approx 2.4\Omega, giving rise to a coherent destruction of tunneling. In Fig. 4(d) we depict χ~0(ωp)\tilde{\chi}_{0}(\omega_{\rm p}). Because of the lifted quasienergy degeneracy, the transitions V^ν,μ(1)V^μ,ν(1)\hat{V}_{\nu,\mu}^{(-1)}\hat{V}_{\mu,\nu}^{(1)} are now visible. However, the system is still transparent at the quasienergy crossing. This is a consequence of the symmetry-protected transparency, which remains approximately valid as the symmetry P^2\hat{P}_{2} is only slightly broken. In order to emphasize the relation to the symmetry-protected transparency, we coin this weaker effect symmetry-induced transparency. The symmetry-induced transparency scales as χ~n(mΩ)hx/Ω\tilde{\chi}_{n}(m\Omega)\propto h_{x}/\Omega at the crossing. It is related to a small asymmetry of the response peaks of χ~n(ωp)\tilde{\chi}_{n}(\omega_{p}). Instead of the Floquet-Gibbs distribution, we choose p0=0.6p_{0}=0.6 and p1=0.4p_{1}=0.4 as the stationary state distribution in Fig. 4(c) to highlight the symmetry-induced transparency.

Please note that there is additionally a dynamical parity symmetry of the TLS in Eq. (45) for R^=σx\hat{R}=\sigma_{\rm x} and tR=τ/2t_{R}=\tau/2. Consequently, the quantum number of the Floquet states related to the parity symmetry operation are m0=0m_{0}=0 and m1=1m_{1}=1. According to the spDS condition Eq. (22) for N=2N=2 and αV(R)=1\alpha_{V}^{(R)}=1, we find that V0,1(n)=V1,0(n)=0V_{0,1}^{(n)}=V_{1,0}^{(n)}=0 for even nn.