Floquet-Filtered States

• Floquet-filtered states are states of periodically driven systems that are selectively stabilized via additional filtering mechanisms applied to the Floquet spectrum.
• They enable clear separation and robust observation of distinct dynamical sectors, such as chiral edge channels and corner modes, through mechanisms like quasienergy-gap selection and temporal gating.
• Their applications span topological transport, many-body state control, and ultrafast state observability, offering a unifying framework for state selection in driven systems.

Floquet-filtered states are states of periodically driven systems that are selected, stabilized, or rendered observable by an additional filtering mechanism acting on the Floquet manifold. In the cited literature, that mechanism may be a quasienergy gap and bulk-edge correspondence, a structured reservoir density of states, decoherence and dissipation, finite pulse duration, coherent-phonon dressing, spatial disorder, defect-induced phase matching, or an explicit spectral filter applied to a Floquet operator. The resulting objects include chiral edge channels, time-periodic corner states, disclination states, bound states in the continuum, asymptotic periodic states of open systems, selectively populated Floquet eigenstates, and many-body filtered states whose local properties approach thermal behavior in a controlled way (Zhou et al., 2018, Diermann et al., 2019, Sato et al., 2019, Boîtier et al., 2021, Tai et al., 29 Jun 2026, Yang et al., 7 Jul 2026).

1. Conceptual scope and formal basis

For a time-periodic Hamiltonian, Floquet theory writes the dynamics in terms of quasienergies and periodic modes. A standard form used in the cited works is

$$|\Psi_{\alpha}(t)\rangle = e^{-i\epsilon_{\alpha}t/\hbar} |\phi_{\alpha}(t)\rangle, \qquad |\phi_{\alpha}(t+T)\rangle = |\phi_{\alpha}(t)\rangle .$$

This representation underlies both closed-system Floquet-Bloch physics and open-system generalizations based on quantum master equations (Tai et al., 29 Jun 2026, Mori, 2022).

Across the cited works, “filtering” is not restricted to one protocol. It refers to the selection of particular Floquet sectors, transition channels, or localized modes by additional structure beyond periodicity itself. In state-based topological formulations, this selection can also be expressed as a topological obstruction: a topological Floquet operator implies an obstruction to constructing a generalized Wannier function which is localized in real and frequency spaces, and for “Floquet-filtered” states topology can be written through the winding of the Aharonov-Anandan phase or hybrid Wannier-center connectivity (Nakagawa et al., 2019).

Filtering mechanism	Representative setting	Selected outcome
Quasienergy-gap selection	Driven topological lattices	Edge, corner, or defect states
Reservoir or dissipation selection	Driven-dissipative systems	Enhanced population of specific Floquet states
Temporal selection	Few-cycle or phonon-driven experiments	Observable sidebands and quantum beats
Explicit spectral filtering	Many-body Floquet operators	Narrow-band filtered states and time averages

This breadth is important because the phrase does not single out a unique phase of matter. Rather, it denotes a family of constructions in which periodic driving supplies the Floquet structure and a second ingredient determines which sectors survive, dominate transport, or acquire direct experimental visibility.

2. Quasienergy-gap filtering and topological transport

A central use of Floquet filtering appears in anomalous Floquet topology, where quasienergy gaps at $0$ and $\pi$ provide distinct channels for state selection. In a periodically quenched two-dimensional lattice model, the two Floquet bands can repeatedly exchange their Chern numbers via touching at quasienergies $0$ and $\pi$ under the change of a single system parameter. Each alternate gap closing increases the winding number by $2$, and two more pairs of chiral edge states appear in the appropriate gap. The number of edge states in a gap obeys

$n_{\rm edge}(E) = |W_E[\hat{U}]|,$

and the two-terminal conductance extracted from the Floquet scattering matrix,

$G(E) = \mathrm{Tr}\left[t^\dagger(E)t(E)\right],$

equals the number of chiral edge channels crossing the gap at quasienergy $E$ (Zhou et al., 2018). In this sense, the driven lattice acts as a controllable Floquet filter for robust edge transport.

A disorder-based version of the same idea was developed for a periodically driven honeycomb lattice. There, scattering-matrix invariants identify strong and weak topological phases, including an anomalous Floquet phase $\mathfrak{E}$ with $0$0 and $0$1, zero Chern numbers, and chiral edge states in both gaps. Because all bulk Floquet bands have zero Chern number in that phase, disorder can fully localize bulk states while leaving edge states delocalized. A strip of strong disorder therefore serves as a filter that allows the passage of only edge states; numerically, the bulk component of a propagated wave packet is strongly suppressed after the disordered region (Bhargava et al., 2021).

The state-based characterization of such filtering can be phrased in Wannier language. For “Floquet-filtered” states realized under block-diagonal Floquet operators, the winding number may be written as

$0$2

so that a nonzero $0$3 indicates winding of the frequency-domain hybrid Wannier centers across the unit cell and an obstruction to localized real/frequency-space Wannier functions (Nakagawa et al., 2019). This formulation makes explicit that filtering by quasienergy gap is simultaneously a transport phenomenon and a state-space topological obstruction.

3. Higher-order, defect, and bound-state realizations

Floquet filtering also appears in higher-order topology. In a two-dimensional Floquet higher-order topological insulator realized in a three-dimensional acoustic lattice, direct acoustic measurements reveal Floquet corner states with time-periodic evolution. Corner modes can occur in the zero gap or in the $0$4 gap, and $0$5 modes exhibit oscillation with period twice the driving period. The system can also host chiral edge states in one gap and corner states in the other, so first-order and higher-order boundary responses coexist under topological protection (Zhu et al., 2020). Here the filter is the quasienergy gap itself: the drive separates zero modes and $0$6 modes into dynamically distinct sectors.

An even more localized form occurs in non-reciprocal unitary scattering networks. Floquet disclination states nucleate in the anomalous phase from a resonant rotation-symmetric phase matching condition,

$0$7

For a $0$8-cut disclination the spectral charge is $0$9 per disclination site, while for a $\pi$0-cut it is $\pi$1. The inverse participation ratio scales as $\pi$2, identifying these states as zero-dimensional. Once coupled to the radiation continuum feeding the anomalous chiral edge state, they can induce topological disclination bound states in the continuum with extreme confinement and lifetime (Qin et al., 2023). The filtering mechanism is simultaneously topological and local: only states satisfying the Floquet condition and the defect resonance survive as long-lived localized modes.

The relation to bound states in the continuum is broader than the scattering-network setting. Floquet BIC were introduced as normalizable Floquet states of a time-periodic Hamiltonian whose quasienergy is embedded in the spectrum of Floquet scattered states. In a driven tight-binding lattice with a defect, such states arise near the dynamic localization regime and can be interpreted, in the high-frequency limit, through selective destruction of tunneling; at lower frequency, they reflect interference among multiple Floquet channels (Longhi et al., 2013). This establishes a localized, spectrally embedded endpoint of Floquet filtering: a state remains confined even though it lies inside an extended continuum.

4. Reservoir, dissipation, and asymptotic open-system filtering

In open systems, Floquet filtering commonly refers to the selective population of particular Floquet transitions by a structured environment. For a periodically driven system weakly coupled to a thermal reservoir, transitions between Floquet states occur at

$\pi$3

with rates

$\pi$4

If the reservoir density of states is peaked at a specific frequency, only a narrow subset of sideband transitions is efficient. In the single-selected-sideband limit,

$\pi$5

so a particular Floquet state can acquire population far above what the undriven equilibrium ground state would have at the same bath temperature. This is the mechanism of Floquet-state cooling (Diermann et al., 2019).

A related but more global construction is the effective Floquet-Gibbs state. For weak but finite system-bath coupling, the asymptotic density matrix can be approximated by

$\pi$6

Its emergence requires, in the formulation given in the cited work, that the heating time be much longer than the relaxation time, that the time-dependent parts of the Hamiltonian commute at all times, and that either the bath cutoff frequency be much less than the drive frequency or the system-bath coupling commute with the driving part (Shirai et al., 2015). The general review of Floquet states in open quantum systems emphasizes the same point in quantum-master-equation language: equilibrium statistical mechanics applies only under restricted conditions, and otherwise the driven-dissipative steady state is a nonthermal Floquet-selected state shaped by the sideband structure of the dissipator (Mori, 2022).

The filtering effect of the environment can also be destructive. In a driven two-level system described by the Maxwell-Bloch equation, transverse relaxation $\pi$7 destroys Floquet-state formation whereas longitudinal relaxation $\pi$8 does not directly affect it. Rabi splitting persists only when coherence survives for a sufficient fraction of the Rabi cycle, although strong driving can recover Floquet features even when $\pi$9 is shorter than the driving period (Sato et al., 2019). In the language used there, the open system acts as a filter on the Floquet manifold, leaving visible only those dressed states that are robust against dissipation.

For non-Markovian systems, asymptotic periodic states can still be obtained by embedding the dynamics into a minimally enlarged auxiliary space where the evolution becomes time-local and $0$0-periodic. Standard Floquet theory is then applied in the enlarged space, and projection back to the physical subspace yields the asymptotic periodic state. The spectral gap

$0$1

of the Floquet propagator estimates the relaxation time toward this limit cycle (Magazzù et al., 2017). In this setting, “Floquet-filtered states” are the physical periodic states selected jointly by memory effects and periodic modulation.

5. Ultrafast observability and temporal filtering

Experimental work on ultrafast driving has expanded the meaning of Floquet filtering from state preparation to state observability. In graphene-covered Ir(111), laser-excited coherent phonons drive Floquet-Bloch states that are detected by time-resolved multiphoton photoemission and quantum beat spectroscopy. The beat frequency matches the coherent-phonon energy $0$2 meV, corresponding to $0$3 THz, and the observed phonon-driven Floquet-Bloch states persist for one to two orders of magnitude longer than conventional light-driven Floquet-Bloch states. Quantum beat spectroscopy resolves sideband splittings in the time domain even when they are smaller than the analyzer energy resolution, and the cited work explicitly interprets these sidebands as Floquet-filtered states selected by the coherent lattice modulation (Tai et al., 29 Jun 2026).

For few-cycle optical driving, pulse duration itself acts as the filter. In photoionization of Ne atoms dressed by few-femtosecond infrared pulses, clear Floquet sidebands are established for pulses as short as about $0$4 cycles, while an analytical model based on Floquet theory remains quantitatively accurate down to $0$5 cycles. The sideband amplitudes depend on duration, intensity, and frequency, and shorter pulses suppress higher-order sidebands, acting as a temporal gate on which Floquet orders populate (Lucchini et al., 2022). Filtering here is not environmental or topological; it is imposed directly by the finite time support of the drive.

Under stronger optical fields, filtering can proceed through non-adiabatic pathways among instantaneous Floquet states. On Cu(111), interferometrically time-resolved multi-photon photoemission reveals a transition from perturbative dressing to a regime in which Floquet states undergo non-adiabatic Landau-Zener tunneling at avoided crossings. The cited work describes this as a regime where observable access to the final continuum is governed by whether the instantaneous field drives the system through a Floquet avoided crossing, creating a filter for population transfer based on Floquet-Landau-Zener dynamics (Yen et al., 6 Mar 2025).

Periodic optical pumping of magnetization in ferromagnets provides a further example. In an iron-garnet film driven by a train of circularly polarized femtosecond laser pulses at $0$6 GHz repetition rate, tuning the external magnetic field to the center of the Brillouin zone increases the precession amplitude by one order of magnitude when the precession frequency becomes a multiple of the pulse repetition rate. The cited summary characterizes this selective amplification as a form of Floquet filtering, in which only those magnetization modes compatible with the periodic drive are strongly excited (Savochkin et al., 2018).

6. Many-body filtering, thermality, and state following

A mathematically explicit notion of Floquet-filtered state appears in many-body theory. A truncated cosine filter applied to a Floquet operator $0$7 produces a narrow-band state centered at quasi-energy $0$8, and in the Floquet regime such filtered states are equivalent to weighted time averages of the initial state. Under the Floquet eigenstate thermalization hypothesis, the trace distance between Floquet-filtered states and thermal states is bounded by the square root of the filter width, and for local observables the deviation scales as $0$9. The same work derives distinct Rényi-$\pi$0 entropy scalings, including

$\pi$1

with corresponding $\pi$2 and $\pi$3 growth for $\pi$4 and $\pi$5 in the cosine-filter constructions summarized there (Yang et al., 7 Jul 2026). This use of “Floquet-filtered states” is literal: the state is generated by a spectral filter and then related to thermality and entanglement structure.

Computational and control-oriented work complements this picture. An iterative algorithm for high-dimensional Hilbert spaces computes individual Floquet states without full diagonalization and allows one to follow a specific Floquet state under small parameter changes. Applied to a periodically driven Bose-Hubbard chain, it demonstrates pseudoadiabatic Floquet state following and shows that a driving-induced Mott insulatorlike target Floquet state can be populated with high efficiency if the driving amplitude is turned on smoothly but not too slowly (Krüger et al., 2021). Although this is not phrased as spectral filtering in the same sense as the previous result, it is a state-selection protocol on the Floquet manifold.

At the single-particle topological level, periodically driven systems can also isolate relevant subspaces beyond standard adiabatic elimination. In microwave Floquet simulators built from periodically bending coupled corrugated waveguides, quasi-adiabatic elimination and high-frequency-limited elimination isolate edge states through quasi-energy separation and high-frequency averaging, despite the lack of conventional adiabaticity. The cited work presents these as driven-induced eliminations that reveal a spectral decomposition for tailoring and manipulating Floquet states with quasi-energies (Pan et al., 2018). A plausible implication is that Floquet filtering, in this control-theoretic sense, can be viewed as a broader strategy for engineering reduced dynamical subspaces in driven systems.

Taken together, these results show that Floquet-filtered states are best understood as a unifying operational category rather than a single invariant-defined phase. The common structure is periodic driving plus a mechanism that restricts which Floquet sectors remain populated, localized, transmitted, or experimentally visible. The differences among the cited realizations—topological, dissipative, ultrafast, defect-based, and many-body—are substantial, but they share the same organizing principle: Floquet theory supplies the quasienergy architecture, and filtering determines the physically realized subset of states.