Gap-Optimized Floquet Evolution

• Gap-Optimized Floquet Evolution is a framework that tunes periodic drives to control spectral gaps, enabling target dynamical or topological regimes in quantum systems.
• It leverages diverse control protocols—from photonic processors to multi-frequency modulation—to optimize both Hermitian and non-Hermitian gap structures.
• The approach addresses common misconceptions in gap control and employs quantum optimal control and geometric Floquet theory to tailor effective Floquet state preparation.

Searching arXiv for papers relevant to gap-optimized Floquet evolution. Gap-Optimized Floquet Evolution is an Editor's term for periodically driven evolution in which the drive is tuned to maximize, create, close, reopen, or relocate a relevant spectral gap so that the resulting quasienergy structure supports a target dynamical or topological regime. In the literature this optimization appears in several distinct but related senses: maximizing the minimal Floquet quasi-energy gap in discretized photonic drives; independently controlling the $0$ and $\pi/T$ gaps of anomalous Floquet walks; closing and reopening individual avoided crossings in optical lattices; and engineering point gaps in non-Hermitian systems, where the relevant spectrum is complex rather than real (Cataldo et al., 16 Sep 2025, Ning et al., 25 May 2026, Sandholzer et al., 2021, Dash et al., 2023, Ji et al., 21 Jan 2025). Taken collectively, these works suggest that the central object is not a single universal gap notion, but a family of gap structures whose optimization depends on symmetry class, drive protocol, and the observable being stabilized.

1. Spectral framework and meanings of “gap”

For a time-periodic Hamiltonian, the basic spectral object is the Floquet operator over one period. In non-Hermitian settings the non-unitary Floquet operator can be written as

$U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$

with eigenvalues $e^{-i\epsilon_\alpha \tau/\hbar}$ and generally complex Floquet quasienergies $\epsilon_\alpha$ (Dash et al., 2023). In lattice Floquet problems one also frequently writes the one-period operator directly in momentum space, as in the anomalous quantum walk

$$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$

or in discretized photonic evolution as a time-ordered product of substep unitaries, for example $U_F(T)=U_3U_2U_1$ for one drive order and $U_F(T)=U_1U_2U_3$ for the reversed order (Ning et al., 25 May 2026, Cataldo et al., 16 Sep 2025).

The phrase “gap” is not unique across Floquet literature. One distinction is between quasienergy gap phases, where the band structure plotted as quasienergy $\varepsilon$ versus quasimomentum $k$ is gapped in $\pi/T$0 for all $\pi/T$1, and quasimomentum gap phases, where the band structure plotted as $\pi/T$2 versus $\pi/T$3 is gapped in $\pi/T$4 for all $\pi/T$5 (Zhu, 2024). Another distinction is specific to non-Hermitian systems: a line gap means the spectrum does not cross a reference line in the complex energy plane, whereas a point gap means the spectrum does not encircle or cross a reference point in the complex plane (Dash et al., 2023, Ji et al., 21 Jan 2025).

A further refinement appears in continuous Floquet Hamiltonians with honeycomb potentials. There, a strict spectral gap of the full Floquet–Schrödinger operator is not expected because high-energy band folding can fill the unit circle, yet one can still define an effective quasi-energy gap: an interval of quasi-energies that does not support modes with large spectral projection onto band-limited Dirac wave-packets (Sagiv et al., 2021). This distinction is important because it separates a strict operator-theoretic gap from a physically relevant gap seen by a restricted set of probes and initial states.

2. Control protocols for gap optimization

Several control mechanisms recur across the literature. In non-Hermitian three-dimensional topological insulators driven by circularly polarized light, the effective Floquet Hamiltonian contains a photoinduced pseudo-magnetic field,

$\pi/T$6

with

$\pi/T$7

Here the drive shifts the effective mass parameter and changes the point-gap condition from $\pi/T$8 to

$\pi/T$9

so tuning light amplitude $U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$0, frequency $U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$1, and polarization angles $U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$2 directly controls whether the driven system lies in a point-gap or line-gap regime (Dash et al., 2023).

In a different setting, two-frequency phase modulation of a one-dimensional optical lattice asymmetrically hybridizes the lowest two bands. Using quasi-degenerate perturbation theory in the extended Floquet space, the driven system is reduced to an effective two-band model with gap

$U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$3

Experimentally, Landau–Zener transitions between Floquet-Bloch bands show that separate and simultaneous control over the closing and reopening of these band gaps is possible, with interference between single- and multi-photon couplings controlled by amplitude and relative phase (Sandholzer et al., 2021).

Discretized photonic Floquet drives implement a more operational notion of optimization. In a programmable photonic processor, each period is divided into substeps generated by non-commuting Hamiltonians, and robustness is obtained by maximizing the minimal quasi-energy gap

$U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$4

The reported consequence is robust directional flow stabilized by maximizing the minimal Floquet quasi-energy gap, together with chiral circulation that reverses under drive inversion and flux-controlled interference with high visibility (Cataldo et al., 16 Sep 2025).

Gap shaping can also be posed directly as a control problem. In graphene tight-binding models, quantum optimal control theory is combined with Floquet engineering to create a gap with opposing flat valence and conduction bands, create a gap with opposing concave symmetric valence and conduction bands, or close a pre-existing gap. The control variables are Fourier components and polarizations of a time-periodic drive with several frequency components, rather than a monochromatic field (Castro et al., 2022). This suggests that “optimization” need not mean monotonic gap enlargement; it can also mean tailoring the local geometry of Floquet pseudo-bands around a chosen avoided crossing.

3. Topological invariants, exceptional structures, and non-Hermitian gap engineering

In non-Hermitian Floquet systems, gap optimization is inseparable from topological classification. For one-dimensional momentum slices of a three-dimensional Floquet exceptional topological insulator, the spectral winding number is

$U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$5

where

$U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$6

A nonzero integer value indicates nontrivial spectral flow around the reference energy $U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$7 and signals a momentum-slice non-Hermitian skin effect. The same work defines a Floquet biorthogonal Chern number and a three-dimensional bulk invariant,

$U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$8

with $U(\tau,0)=\exp\left(-iH_F\tau/\hbar\right),$9, linking the optimized point-gap structure to surface states and the non-Hermitian skin effect (Dash et al., 2023).

A distinct exceptional-gap mechanism appears in Floquet $e^{-i\epsilon_\alpha \tau/\hbar}$0 exceptional points. These occur at quasienergy $e^{-i\epsilon_\alpha \tau/\hbar}$1 where the eigenvector rotates on the Bloch sphere and accumulates a $e^{-i\epsilon_\alpha \tau/\hbar}$2 geometric phase in one period. In the Floquet bipartite lattice studied in (Zhu, 2024), quasimomentum-gap phases host a pair of order-$e^{-i\epsilon_\alpha \tau/\hbar}$3 Floquet $e^{-i\epsilon_\alpha \tau/\hbar}$4 exceptional points at $e^{-i\epsilon_\alpha \tau/\hbar}$5, whereas the transition to a quasienergy-gap phase is marked by an order-$e^{-i\epsilon_\alpha \tau/\hbar}$6 Floquet $e^{-i\epsilon_\alpha \tau/\hbar}$7 exceptional point at $e^{-i\epsilon_\alpha \tau/\hbar}$8. Their merging is constrained by dynamical structure: exceptional points with the same dynamical structure can merge, while those with opposite dynamical structures cannot. The reported transition condition is

$e^{-i\epsilon_\alpha \tau/\hbar}$9

Point-gapped topological superconductors furnish another systematic construction. By combining Floquet theory with particle-hole symmetry, a point gap emerges at the overlap of Floquet bands with opposite winding numbers, and even weak non-Hermiticity opens a point gap from a gapless spectrum in the thermodynamic limit. The point-gap winding number is written as

$\epsilon_\alpha$0

while particle-hole symmetry imposes $\epsilon_\alpha$1. The same transition is accompanied by the appearance of the Floquet $\epsilon_\alpha$2 skin effect and by non-Bloch $\epsilon_\alpha$3-symmetry breaking (Ji et al., 21 Jan 2025). These constructions make clear that in non-Hermitian Floquet settings the optimized object is often a complex-spectral topology rather than a conventional Hermitian band gap.

4. Platform-specific realizations

In graphene antidot lattices, circularly polarized electromagnetic driving produces a sequence of gap reorganizations computed non-perturbatively within the Floquet formalism. As drive amplitude is varied, the quasienergy gap at $\epsilon_\alpha$4 can close and restore a Dirac-like dispersion in real time relative to the gapped equilibrium state; at larger amplitude the main quasienergy gap reopens at the $\epsilon_\alpha$5 point; bands can flatten near $\epsilon_\alpha$6; and a Floquet semi-Dirac material appears when the gap at $\epsilon_\alpha$7 closes with quadratic and linear dispersions in orthogonal directions (Cupo et al., 2021). The same study associates band flattening with selective dynamical localization and notes that shifting the gap between high-symmetry points can change which crystal momenta dominate scattering processes relevant to transport and optical emission.

Two recent photoemission studies provide direct spectroscopic realizations of Floquet-induced avoided-crossing gaps. In monolayer graphene under resonant driving, time- and angle-resolved photoemission spectroscopy reveals gap opening at Floquet band crossings, accompanied by coherent Floquet sidebands, with a measured gap $\epsilon_\alpha$8 meV. The gap exhibits pronounced momentum anisotropy, featuring two Dirac nodes protected by the spatiotemporal symmetry and tunable by light polarization, and both the gap and sidebands are present only during pump-probe overlap (Wang et al., 30 Mar 2026). In bulk graphite, intense mid-infrared pumping reveals Floquet-induced gaps at resonance points in both valence and conduction bands, accompanied by coherent Floquet sidebands; the measured hybridization gaps range from approximately $\epsilon_\alpha$9 meV to $$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$0 meV, and the distinct timescales of gap formation and carrier relaxation disentangle coherent dressing from photo-excitation (Wang et al., 30 Mar 2026).

Few-layer black phosphorus realizes a gate-tunable Floquet gap in a Dirac semimetal phase. Starting from a continuum Hamiltonian,

$$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$1

circularly polarized light generates an effective Floquet mass near each Dirac node,

$$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$2

opening a Floquet gap

$$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$3

The accompanying Berry curvature produces a photoinduced DC Hall current that is tunable by both periodic driving and electrostatic gating (Kim et al., 21 Apr 2025).

Synthetic platforms show equally explicit gap control. A one-dimensional flux-controlled anomalous Floquet quantum walk has chiral symmetry, independent topological information in the $$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$4 and $$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$5 gaps, and phase sectors organized in the $$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$6 plane. In the coexistence sector, a $$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$7 mode and a $$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$8 mode on the same edge span a natural boundary logical subspace and produce a clear $$U(k)=e^{-i\theta_2(k)\sigma_x}e^{-i\theta_1(k)\sigma_z},$$9 response in local boundary observables (Ning et al., 25 May 2026). In programmable photonics, discretized evolution of synthetic magnetic fields yields chiral circulation, flux-controlled interference, and robust directional flow when the minimal quasi-energy gap is optimized (Cataldo et al., 16 Sep 2025).

5. Geometry, computation, and state preparation

One conceptual difficulty of Floquet optimization is that quasienergies are defined modulo $U_F(T)=U_3U_2U_1$0, so the ordering of the spectrum is ambiguous. Geometric Floquet theory resolves this by fixing the gauge freedom using the parallel-transport gauge and decomposing evolution into a purely geometric and a purely dynamical part. Schindler and Bukov formulate the evolution as

$U_F(T)=U_3U_2U_1$1

where the dynamical average-energy operator $U_F(T)=U_3U_2U_1$2 provides an unambiguous sorting of the quasienergy spectrum, identifies a Floquet ground state, and suggests a way to define the filling of Floquet-Bloch bands (Schindler et al., 2024). Within this framework, $U_F(T)=U_3U_2U_1$3-quasienergy splitting in discrete time crystals and $U_F(T)=U_3U_2U_1$4-edge modes in anomalous Floquet topological insulators are traced to the geometric sector rather than to the dynamical average-energy spectrum.

The computational treatment of gap-optimized Floquet evolution often proceeds through Sambe space. For a periodic Hamiltonian expanded in Fourier components, the infinite-dimensional Floquet eigenvalue problem is truncated by a rigorously justified cutoff,

$U_F(T)=U_3U_2U_1$5

which guarantees exponentially small truncation error in quasienergy estimation (Mizuta, 2024). The same work organizes quantum algorithms for quasienergy estimation and Floquet eigenstate preparation with nearly optimal query complexity, and shows that a preferred gapped Floquet eigenstate can be deterministically implemented with nearly optimal query complexity in the gap.

Continuous systems motivate a different caveat. For honeycomb Schrödinger equations with slow time-periodic forcing, Dirac wave-packets are approximated on long times by an effective time-periodic Dirac equation with a gap in its quasienergy spectrum, yet the full Floquet–Schrödinger operator is believed not to possess a strict gap because of band folding. The introduced effective quasi-energy gap therefore functions as a controlled relaxation of the strict notion of spectral gap (Sagiv et al., 2021). This is particularly relevant when optimization is assessed by probes confined to momenta and energies near a Dirac point.

6. Consequences, diagnostics, and recurrent misconceptions

The physical consequences of gap-optimized Floquet evolution vary with platform, but several patterns recur. In photonic processors, maximizing the minimal quasi-energy gap suppresses sensitivity to perturbations and stabilizes oscillatory, directionally robust transport, quantified by the per-period winding number extracted from the phase of the first spatial Fourier harmonic (Cataldo et al., 16 Sep 2025). In quantum walks, dual gap opening protects a boundary $U_F(T)=U_3U_2U_1$6 logical subspace, while frame-resolved mean chiral displacements approach the two winding numbers in the clean pre-reflection window of the symmetric time frames (Ning et al., 25 May 2026). In non-Hermitian settings, optimized point gaps control the existence of surface states, modulate skin localization, and allow photo-tunable control over the position, number, and order of exceptional points (Dash et al., 2023, Zhu, 2024, Ji et al., 21 Jan 2025).

Several misconceptions are addressed directly by the cited works. First, a larger or better-controlled Floquet gap does not always mean a strict global spectral gap of the full driven operator; continuous honeycomb models instead support an effective quasi-energy gap for band-limited Dirac wave-packets (Sagiv et al., 2021). Second, quasienergy alone does not necessarily provide a unique notion of spectral order or ground-state filling because of quasienergy folding; the average-energy operator in geometric Floquet theory supplies that missing ordering principle (Schindler et al., 2024). Third, in non-Hermitian systems the relevant optimized structure may be a point gap rather than a line gap, and a system may show momentum-slice non-Hermitian skin effect even though the system as a whole does not (Dash et al., 2023).

Taken together, these studies establish gap optimization as a unifying design principle of Floquet engineering rather than a single protocol. Depending on context, the optimized object may be an avoided-crossing gap at a Floquet band crossing, a dual pair of chiral-symmetry-protected gaps at $U_F(T)=U_3U_2U_1$7 and $U_F(T)=U_3U_2U_1$8, a quasimomentum gap phase boundary marked by a Floquet $U_F(T)=U_3U_2U_1$9 exceptional point, or a complex-spectral point gap protected by non-Bloch topology. The common theme is that carefully structured periodic evolution reorganizes spectral separation so that transport, localization, topology, and state preparation become controllable on stroboscopic timescales (Sandholzer et al., 2021, Wang et al., 30 Mar 2026, Wang et al., 30 Mar 2026, Castro et al., 2022).