Boundary Floquet Driving

• Boundary Floquet driving is a method where periodic modulation confined to edges or impurities generates distinct topological and dynamical boundary phenomena.
• It utilizes the full micromotion of time-periodic Hamiltonians to induce anomalous phases, creating chiral edge modes and unconventional transport properties not seen in static systems.
• The approach is demonstrated in varied experimental settings—from optical lattices to superconducting wires—enabling control over impurity-bound states, non-Hermitian effects, and synthetic frequency boundaries.

Boundary Floquet driving denotes a family of periodically driven quantum and wave systems in which the drive is applied at a physical boundary, confined to local boundary defects or impurities, or engineered so that its principal topological and dynamical consequences are boundary phenomena. In this literature, the term covers several distinct but related constructions: edge-only temporal modulation, local impurity driving, bulk drives whose full micromotion generates anomalous boundary states, spatially non-uniform drives that create artificial edges, and synthetic-frequency constructions in which a boundary appears in Floquet space rather than in ordinary real space (Quelle et al., 2017, Agarwala et al., 2017, Liu et al., 2022, Schindler et al., 2023, Hu et al., 23 Mar 2026).

1. Theoretical framework and scope

For a time-periodic Hamiltonian with period $T$, the central object is the one-period propagator

$U:=U(T,0),$

from which one defines a Floquet Hamiltonian

$H_F=-\,\frac{i}{T}\ln U.$

Because the logarithm requires a branch cut in quasienergy, Floquet topology is generally not exhausted by the stroboscopic bands of $H_F$. In the anomalous cases most relevant to boundary Floquet driving, the topology resides in the full time evolution during the drive cycle rather than in the endpoint $U(T,0)$ alone (Quelle et al., 2017).

A standard formulation uses a periodicized evolution operator $V_i(\mathbf{k},t)$ and the gap winding number

$W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$

which counts chiral edge modes crossing a quasienergy gap $\Delta_i$. The associated bulk Chern numbers satisfy

$C_i=W_i-W_{i-1}.$

This immediately permits anomalous phases with vanishing $C_i$ but nonzero equal $U:=U(T,0),$0, a possibility absent in static systems (Quelle et al., 2017).

A complementary formulation emphasizes the return map

$U:=U(T,0),$1

whose winding classifies anomalous Floquet topology even when the final Floquet bands are trivial. In this perspective, protected boundary states at $U:=U(T,0),$2 and $U:=U(T,0),$3 quasienergies arise from the winding of the full unitary evolution and from irreducible branch-cut crossings during the drive (Vu, 2021).

The literature represented here uses “boundary Floquet driving” in more than one sense. In the narrow sense, the drive is supported only near an edge or defect. In a broader sense, periodic driving is used as a boundary-engineering tool: it may generate chiral edge modes, corner charges, synthetic-frequency edge states, or effective non-Hermitian boundary chains even when the temporal modulation is not restricted to the physical edge (Zhang et al., 2020, Schindler et al., 2023).

Realization class	Drive support	Boundary outcome
Honeycomb shaking	Bulk lattice, stepwise	Chiral edge modes with $U:=U(T,0),$4
Kitaev honeycomb drive	Bulk spin model	Chiral Majorana/twist-defect pump
Local impurity or hopping drive	Site, bond, or end region	End modes, bound states, transport control
Boundary-only non-Hermitian drive	Edge-localized	Bulk spectral reconstruction via skin modes
Synthetic Floquet-lattice boundary	Frequency space	Edge states at selected frequencies

This variety is not a terminological accident. It reflects a recurring structural fact: in Floquet systems, the physically decisive object is often the drive-induced micromotion, and boundary responses are where that micromotion becomes spectrally or topologically visible.

2. Anomalous and symmetry-protected boundary transport

A paradigmatic realization is the three-step shaking protocol in a honeycomb optical lattice. Each subcycle lasts $U:=U(T,0),$5, and the lattice is shaken along a different direction $U:=U(T,0),$6. In the high-frequency limit, the nearest-neighbor hopping $U:=U(T,0),$7 is renormalized as

$U:=U(T,0),$8

so that two of the three bond directions can be suppressed during a given subcycle by choosing the amplitude and frequency at the first zero $U:=U(T,0),$9 of $H_F=-\,\frac{i}{T}\ln U.$0. The resulting ordered product

$H_F=-\,\frac{i}{T}\ln U.$1

implements directed counter-clockwise motion around a hexagonal plaquette. In the anomalous phase $H_F=-\,\frac{i}{T}\ln U.$2, the winding numbers satisfy $H_F=-\,\frac{i}{T}\ln U.$3, while the two bulk-band Chern numbers vanish, $H_F=-\,\frac{i}{T}\ln U.$4. The quasienergy spectrum in a zigzag ribbon nonetheless contains protected chiral edge modes in both gaps $H_F=-\,\frac{i}{T}\ln U.$5 and $H_F=-\,\frac{i}{T}\ln U.$6 (Quelle et al., 2017).

A more intrinsically dynamical boundary phenomenon appears in the periodically driven Kitaev honeycomb model. There the three-step drive acts on $H_F=-\,\frac{i}{T}\ln U.$7, $H_F=-\,\frac{i}{T}\ln U.$8, and $H_F=-\,\frac{i}{T}\ln U.$9 bonds in succession,

$H_F$0

with

$H_F$1

At the strong-drive fixed point $H_F$2, bulk Majoranas circulate around plaquettes and return after two periods, while boundary Majoranas undergo a chiral translation around the perimeter. The edge transports “irrational fractional quantum information,” with chiral unitary index

$H_F$3

and more generally $H_F$4 in the $H_F$5 parafermionic generalization. The single-period anomaly is tied to bulk Floquet enriched topological order that exchanges $H_F$6 and $H_F$7 anyons each period; the corresponding bulk-boundary relation is expressed as

$H_F$8

This is a boundary pump of twist-defect degrees of freedom rather than of ordinary particles (Po et al., 2017).

Symmetry-protected variants can be implemented without long-range couplings. A six-step square-lattice protocol built from pairwise nearest-neighbor couplings realizes Floquet phases with time-reversal symmetry, chiral symmetry, or particle-hole symmetry, depending on parameter constraints. For fermionic time-reversal symmetry, the protocol supports counterpropagating boundary states protected by Kramers degeneracy; for particle-hole symmetry with $H_F$9, it supports copropagating chiral boundary states with even winding $U(T,0)$0 and vanishing band Chern numbers $U(T,0)$1 (Höckendorf et al., 2019).

The broader classification program makes these examples systematic. In anomalous Floquet topology for the real Altland–Zirnbauer classes BDI, D, DIII, and AII, the relevant invariant is carried by the full time-evolution unitary, and the descendant phases inherit boundary modes at the $U(T,0)$2gap through a dimensional hierarchy. The summary given in that framework includes end modes at the $U(T,0)$3gap in class BDI $U(T,0)$4, 1D chiral boundary modes in class D $U(T,0)$5, Majorana end modes in class D $U(T,0)$6, helical modes in class DIII $U(T,0)$7, Majorana Kramers pairs in class DIII $U(T,0)$8, and quantum spin Hall boundary modes in class AII $U(T,0)$9 (Vu, 2021).

A related but distinct open-boundary phenomenon is Floquet-band holonomy in a driven trimer chain. Under open boundary conditions, the spectrum contains anomalous chiral edge modes localized only at one edge, winding around the entire quasienergy Brillouin zone and exhibiting holonomy different from bulk states. These edge modes support quantized pumping $V_i(\mathbf{k},t)$0 for type-I modes and half-quantized pumping $V_i(\mathbf{k},t)$1 for type-II modes (Zhou et al., 2016).

3. Local and impurity-driven boundary states

Boundary Floquet driving also includes strictly local periodic modulation of a bond, site, or end region. In a Kitaev $V_i(\mathbf{k},t)$2-wave superconducting wire, periodic driving of the nearest-neighbor hopping,

$V_i(\mathbf{k},t)$3

generates end-localized Floquet modes in an open chain. Besides conventional Floquet Majorana end modes with eigenvalues $V_i(\mathbf{k},t)$4, the driven wire supports anomalous end modes with $V_i(\mathbf{k},t)$5, appearing in complex-conjugate pairs. All end modes have equal particle and hole probabilities. For small driving amplitude, the anomalous end-mode quasienergies and the peaks of their Fourier transforms lie close to extrema of the bulk Floquet spectrum, which furnishes a spectral rather than topological bulk-boundary correspondence (Saha et al., 2016).

A single periodically kicked site in a one-dimensional tight-binding chain provides an even more local boundary-driving mechanism. With

$V_i(\mathbf{k},t)$6

the kick acts as a tunable effective impurity. In the high-frequency regime,

$V_i(\mathbf{k},t)$7

so the transmission through the driven site becomes

$V_i(\mathbf{k},t)$8

The transmission can be tuned to exactly zero near $V_i(\mathbf{k},t)$9 when $W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$0 is small enough, a dynamical bond-cutting effect that the static problem does not realize. The same local drive generates Floquet bound states; when the frequency is lowered, these states delocalize and become resonances by mixing with Floquet bulk states (Agarwala et al., 2017).

With two periodically driven impurities at $W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$1 and $W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$2,

$W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$3

the region between them becomes a dynamical Fabry–Pérot cavity. After Floquet expansion, the problem maps to a multichannel scattering system with sideband energies $W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$4 and channel wave numbers

$W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$5

The total transmission is

$W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$6

The interplay between Fano interference and cavity resonances produces perfect transmission, perfect reflection, strong localization, bound states in the continuum, quasi-BICs, and giant Wigner time delays. In the cavity-enhanced regime discussed there, the width can reach $W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$7, with a corresponding delay $W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$8 (Bruno et al., 29 May 2026).

These impurity and end-driven constructions show that boundary Floquet driving is not confined to topological band engineering. It also functions as a local scattering control problem in which the drive tunes boundary localization, resonance widths, and coherent trapping.

4. Boundary-induced non-Hermiticity and dissipative responses

One major development is the realization that periodically driven Hermitian bulks can induce effective non-Hermitian boundary physics. In the Rudner–Lindner–Berg–Levin anomalous Floquet model, the Floquet operator at resonance $W_i=\frac{1}{8\pi^2}\int_{\mathrm{GBZ} dt\, dk_x\, dk_y\; \mathrm{Tr}\!\left( V_i^{-1}\partial_t V_i\, \bigl[V_i^{-1}\partial_{k_x}V_i,\, V_i^{-1}\partial_{k_y}V_i\bigr] \right),$9 and $\Delta_i$0 returns bulk states to themselves while moving boundary states unidirectionally. Interpreting the real-space Floquet operator as a static matrix $\Delta_i$1, the boundary becomes a one-way hopping chain identical to a maximally nonreciprocal Hatano–Nelson model with periodic boundary conditions. Cutting one boundary hopping in the Floquet operator converts this boundary chain from periodic to open, halting chiral propagation and producing a non-Hermitian skin effect. The effect is robust in the anomalous Floquet topological insulator but not in the Floquet Chern insulator, and it is diagnosed by a real-space spectral-localizer invariant $\Delta_i$2 (Liu et al., 2022).

A different mechanism generates non-Hermiticity only under open boundary conditions. In a two-step one-dimensional drive,

$\Delta_i$3

the shift operators commute under periodic boundary conditions but not under open boundary conditions. Consequently, the Floquet Hamiltonian is Hermitian under periodic boundary conditions, while the Baker–Campbell–Hausdorff expansion under open boundaries contains boundary-supported non-Hermitian terms such as

$\Delta_i$4

The resulting $\Delta_i$5-symmetry-breaking transition occurs when the quasienergy bandwidth under periodic boundary conditions covers the full frequency Brillouin zone, $\Delta_i$6, rather than at an ordinary static band touching. In the broken phase, the eigenstates exhibit scale-free localization with

$\Delta_i$7

This boundary-sensitive mechanism is specific to Floquet periodicity and OBC commutators (Li et al., 24 Mar 2026).

In genuinely non-Hermitian chains, boundary-only driving can reconstruct the bulk quasienergy spectrum even in the thermodynamic limit. For

$\Delta_i$8

with $\Delta_i$9 supported only near the two ends, a Floquet non-Bloch band theory applies when

$C_i=W_i-W_{i-1}.$0

Then the non-Bloch Floquet Hamiltonian is the time average,

$C_i=W_i-W_{i-1}.$1

while the difference between the full Floquet operator and that of the averaged Hamiltonian acts as an effective boundary perturbation. At large $C_i=W_i-W_{i-1}.$2, Floquet-zone folding allows this perturbation to resonantly couple skin modes with different localization lengths, producing cusps in the Floquet generalized Brillouin zone and boundary-driven $C_i=W_i-W_{i-1}.$3-symmetry breaking (Hu et al., 23 Mar 2026).

Open dissipative systems add a Liouvillian version of the same theme. In the Floquet–Lindblad treatment of the Creutz ladder, periodic step modulation of the damping matrix alternates two half-period configurations and defines a Floquet damping matrix $C_i=W_i-W_{i-1}.$4. Under open boundaries, the Liouvillian skin effect undergoes a unipolar-to-bipolar transition, accompanied by a chiral-to-helical damping crossover. The work introduces a drive-renormalized localization length and a dynamical polarization

$C_i=W_i-W_{i-1}.$5

together with a polarization drift $C_i=W_i-W_{i-1}.$6 that is finite in the topological phase and zero in the trivial phase. The driven Liouvillian spectrum develops a hierarchy of rapidities in which rapidly decaying bulk skin modes leave a long-lived regime dominated by boundary modes (Roy et al., 28 Nov 2025).

5. Engineered, synthetic, and higher-order boundaries

Not all boundary Floquet driving is realized by driving a physical edge. Spatially non-uniform Floquet engineering enlarges the concept by allowing the drive itself to vary across the lattice,

$C_i=W_i-W_{i-1}.$7

with the effective static Hamiltonian obtained perturbatively from the Magnus expansion. For a monochromatic drive $C_i=W_i-W_{i-1}.$8,

$C_i=W_i-W_{i-1}.$9

This inverse-design framework can remove disorder, undo Anderson localization, induce dynamic localization in irregular lattices, and, most directly for the present topic, sever selected links to “create an artificial sharp edge in a topological system.” The paper is explicit that it is not about boundary driving in the narrow sense, but it makes boundary creation itself a Floquet design variable (Schindler et al., 2023).

A synthetic version occurs in commensurate multi-frequency drives. For a Hamiltonian with $C_i$0, Fourier transformation leads to a Floquet-lattice equation with an effective quasi-electric field $C_i$1. In the two-frequency case with coprime integers $C_i$2, the strong-frequency limit reduces the problem to a quasi-one-dimensional Rice–Mele/SSH-like chain,

$C_i$3

Because the Floquet lattice is periodically identified under $C_i$4, the effective dimerization can generate a boundary in frequency space. The resulting edge-localized Floquet states are localized at particular frequencies rather than at real-space edges, and their position can be tuned by changing $C_i$5 (Park et al., 2022).

Periodic driving also enables higher-order and fragile boundary signatures. In a Floquet honeycomb model and a Floquet $C_i$6-flux square-lattice model driven by time-dependent gauge fields via the Peierls substitution,

$C_i$7

the drive-induced renormalized nearest-neighbor hopping and next-nearest-neighbor terms move the system between fragile topological, Chern-insulating, and atomic phases. The boundary manifestations are distinct: chiral edge modes in the Chern phase, fractional corner charges $C_i$8 in the honeycomb fragile phases, and $C_i$9 or $U:=U(T,0),$00 corner charges in the square-lattice fragile phases depending on symmetry (Zhang et al., 2020).

This broader picture motivates a useful distinction. In one usage, boundary Floquet driving means “drive the boundary.” In another, it means “use Floquet engineering to create, move, or diagnose a boundary phenomenon.” The latter usage encompasses artificial sharp edges, corner anomalies, and synthetic-frequency edge states, all of which are explicit in the cited literature.

6. Diagnostics, critical dynamics, and experimental relevance

The experimental observables depend strongly on the realization. In the anomalous honeycomb optical-lattice protocol, two diagnostic routes were proposed: measuring the vanishing Chern number and Hall conductivity, and directly detecting edge states through boundary-localized wave-packet dynamics. Because the anomalous phase can have nearly flat bulk bands, a wave packet injected near a sharp edge separates into a strongly localized boundary component and a weakly dispersive bulk component. The same work also points to state tomography of driven Bloch states as a route to reconstructing the full time evolution and measuring the winding number $U:=U(T,0),$01 rather than only stroboscopic Chern data (Quelle et al., 2017).

In boundary-driven critical systems, the observables are different. For the transverse-field Ising chain on the half-line with periodic boundary magnetic field $U:=U(T,0),$02, the universal diagnostics are the entanglement entropy $U:=U(T,0),$03 of a boundary interval and the Loschmidt echo

$U:=U(T,0),$04

The dynamics separate into three regimes. In the slow-driving limit, the evolution is interpreted as a sequence of local quenches and the Loschmidt echo is governed by boundary-condition changing operators. In the fast-driving limit, the boundary effectively experiences a single quench with the averaged field. In the intermediate regime, the system heats, the Loschmidt echo decays exponentially, and the entanglement entropy can become volume-law-like. Smooth protocols crossing the critical boundary value are described by Kibble–Zurek-renormalized exponents (Berdanier et al., 2017).

Open non-Hermitian and dissipative settings introduce spectral diagnostics. In boundary-driven non-Hermitian chains, the onset of boundary-driven bulk reconstruction is associated with the appearance of higher-zone auxiliary generalized Brillouin-zone components and, beyond threshold, with complex quasienergies and cusp singularities in the Floquet GBZ (Hu et al., 23 Mar 2026). In Floquet–Lindblad dynamics, the least damped Liouvillian rapidities isolate the long-lived boundary sector, while quantities such as dIPR, MIPR, $U:=U(T,0),$05, and $U:=U(T,0),$06 distinguish skin-localized bulk modes from topological boundary modes (Roy et al., 28 Nov 2025).

The proposed platforms are comparably diverse. They include ultracold atoms in shaken optical lattices, curved optical waveguides, photonics, ring resonators, acoustic metamaterials, synthetic-dimension platforms, and mesoscopic one-dimensional channels with driven impurities (Quelle et al., 2017, Schindler et al., 2023, Liu et al., 2022, Zhou et al., 2016, Bruno et al., 29 May 2026). This suggests a unifying practical lesson: boundary Floquet driving is less a single protocol than a control principle. The drive period, amplitude, phase structure, spatial support, and commensurability relations can be used to choose between chiral boundary transport, end-mode generation, bound-state formation, synthetic-frequency localization, non-Hermitian skin accumulation, and dissipation-assisted boundary-state isolation.

A recurrent misconception is that Floquet boundary phenomena should be classifiable solely from the effective Floquet Hamiltonian. The cited works consistently show otherwise. In anomalous phases, the decisive invariant is the winding of the full micromotion; in non-Hermitian settings, boundary-localized Floquet perturbations can reshape bulk spectra through skin-mode sensitivity; in synthetic and higher-order settings, the “boundary” may itself be engineered by the drive. Boundary Floquet driving is therefore best understood as the study of how periodic time dependence converts boundary conditions, boundary defects, or boundary observables into the primary carriers of nonequilibrium topology and dynamics.