Floquet-Informed Learning Algorithm

• Floquet-informed learning algorithms are computational approaches that use periodic quantum dynamics to structure learning via features such as quasienergies and Floquet modes.
• They integrate periodicity directly into target representations, cost functions, and control loops to enhance efficiency in tasks like Hamiltonian reconstruction and reinforcement control.
• Applications span from reinforcement learning for state preparation to neural operator surrogates, demonstrating improved fidelity and performance in periodically driven systems.

A Floquet-informed learning algorithm is a computational scheme for periodically driven quantum systems in which the learning component is constrained or structured by Floquet theory: periodicity, quasienergies, Floquet Hamiltonians, Floquet modes, micromotion, or Floquet-band structure enter the representation, target, cost function, similarity metric, or control loop. In current literature, the designation covers model-free reinforcement learning for Floquet state preparation, Hamiltonian learning of stroboscopic generators, Fourier-domain inverse problems for driven Hamiltonians, operator-learning surrogates, unsupervised phase-diagram inference, and hybrid quantum-classical eigensolvers (Bukov, 2018, Olsacher et al., 2022, Li, 2 Sep 2025, Qi et al., 8 Sep 2025, Ma et al., 2021, Kumar et al., 14 Mar 2025).

1. Floquet structure as an algorithmic prior

The common starting point is a Hamiltonian with temporal periodicity, $H(t)=H(t+T)$, together with the one-period propagator $U(T,0)$. Several papers use the standard Floquet relation

$e^{-iH_F T}=U(T,0),$

or its equivalent quasienergy formulation, as the mathematical object that organizes learning. In the kicked-top Hamiltonian-learning setting, the learned quantity is explicitly the stroboscopic generator $H_F(\tau)$, expanded perturbatively through the Floquet-Magnus series (Olsacher et al., 2022). In Floquet-ADAPT-VQE, the periodic problem is lifted into an extended Floquet-Hilbert space and converted into a time-independent eigenvalue problem for an enlarged Hamiltonian $\hhat H_\mathrm{F}$ (Kumar et al., 14 Mar 2025). In unsupervised Floquet topology, the informative object is not merely $U(T)$ but the periodic unitary

$$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$

which retains intra-period structure (Ma et al., 2021).

A central distinction in the literature is whether Floquet structure is used explicitly or implicitly. In explicit formulations, the algorithm reconstructs, approximates, or diagonalizes a Floquet Hamiltonian or a periodic unitary. In implicit formulations, Floquet theory defines the control target or the search landscape, but the learner itself does not manipulate an analytic $H_F$. The Kapitza-oscillator reinforcement-learning study is the clearest example: the target is a Floquet eigenstate localized near the inverted position, yet the agent is deliberately model-free and is not given the Hamiltonian or the Floquet solution (Bukov, 2018).

This separation is important because it prevents a common misreading: a Floquet-informed method need not “learn the Floquet Hamiltonian” in the narrow sense. The informing role of Floquet theory may instead reside in the target manifold, in the harmonic representation, in a symmetry-constrained similarity measure, or in an extended-space embedding.

2. Principal algorithmic families

The existing literature spans several distinct but related paradigms.

Family	Floquet-informed ingredient	Representative example
Reinforcement learning control	Floquet target state, micromotion, periodic drive	Quantum Kapitza oscillator (Bukov, 2018)
Reinforcement learning for bosonic codes	Engineered Floquet Hamiltonian and driven cavity controls	Cat-code preparation (Wu et al., 25 Oct 2025)
Hamiltonian learning	Floquet-Magnus-guided Ansatz for $H_F$	Kicked top (Olsacher et al., 2022)
Fourier inverse reconstruction	Truncated Fourier series and Floquet-band linear system	Periodically driven local Hamiltonians (Li, 2 Sep 2025)
Neural operator surrogate	Fourier-space operator learning on periodic time domain	FNO for $H_F$, observables, and operator growth (Qi et al., 8 Sep 2025)
Unsupervised phase discovery	Full intra-period unitary, adiabatic deformation, symmetry	Floquet topological boundaries (Ma et al., 2021)
Hybrid quantum-classical eigensolver	Extended Floquet-Hilbert-space Hamiltonian	Floquet-ADAPT-VQE (Kumar et al., 14 Mar 2025)

Across these families, two organizing axes recur. The first is stroboscopic versus intra-period information. Some methods target $U(T,0)$0 or $U(T,0)$1 alone, whereas others require the full micromotion within a period. The second is physics-guided representation versus black-box optimization. Some methods build the model directly in Fourier or commutator bases; others leave the policy largely unconstrained and let Floquet structure enter only through rewards or targets.

3. Reinforcement learning for Floquet control and state preparation

In the quantum Kapitza oscillator, the control task is to steer the ground state of the undriven oscillator to a target state $U(T,0)$2, defined as the Floquet eigenstate localized near the inverted position $U(T,0)$3, while maximizing

$U(T,0)$4

The control field is a bang-bang sequence $U(T,0)$5. The agent uses a modified Watkins tabular Q-learning scheme with eligibility traces and replay; the RL state is the protocol history itself, the action set is $U(T,0)$6, and the reward is an estimated final fidelity from repeated binary measurements, $U(T,0)$7. The study reports that the agent, starting from zero knowledge of the system, learns protocols with fidelity saturating around $U(T,0)$8; it remains effective with noisy initial states and random control failures, yielding fidelities around $U(T,0)$9 and $e^{-iH_F T}=U(T,0),$0, respectively. It also shows that non-stroboscopic protocols with 8 bangs per cycle outperform stroboscopic protocols with 1 bang per period at moderate drive frequencies, establishing micromotion as a usable control resource rather than a nuisance (Bukov, 2018).

A more recent reinforcement-learning formulation targets bosonic code preparation in a periodically driven superconducting cavity. There the target state is $e^{-iH_F T}=U(T,0),$1, with $e^{-iH_F T}=U(T,0),$2 for the $e^{-iH_F T}=U(T,0),$3 rotational code, and the drive parameters $e^{-iH_F T}=U(T,0),$4 and $e^{-iH_F T}=U(T,0),$5 are tuned by TD3. The system state $e^{-iH_F T}=U(T,0),$6 is the density matrix $e^{-iH_F T}=U(T,0),$7, the action updates the driving amplitude and frequency, and the reward is shaped from the fidelity

$e^{-iH_F T}=U(T,0),$8

The dynamics include dissipation and dephasing through a Lindblad master equation with photon-loss rate $e^{-iH_F T}=U(T,0),$9 and dephasing rate $H_F(\tau)$0. For the four-fold cat code, the reported preparation time is about $H_F(\tau)$1, compared with roughly $H_F(\tau)$2 for an earlier adiabatic ramp, while the final fidelity remains around $H_F(\tau)$3. The paper further reports transfer of a network trained at $H_F(\tau)$4 to stronger noise such as $H_F(\tau)$5, and recovery of high fidelity by retraining at $H_F(\tau)$6 (Wu et al., 25 Oct 2025).

Taken together, these studies establish two concrete meanings of “Floquet-informed” in reinforcement learning. In one, the Floquet aspect lies in the target eigenstate and in the exploitable micromotion, while the learner is model-free. In the other, Floquet engineering supplies the target Hamiltonian structure and the learner optimizes continuous drive parameters inside an open-system control loop. This suggests that the phrase denotes a family of control strategies rather than a single update rule.

4. Hamiltonian learning and reconstruction of driven generators

Hamiltonian learning provides a more explicit use of Floquet structure. In the kicked top, one period of Trotterized evolution is

$H_F(\tau)$7

and for sufficiently small $H_F(\tau)$8, $H_F(\tau)$9 admits a Floquet-Magnus expansion. The learning protocol prepares several initial states, evolves them under the black-box Trotter block, and constructs a constraint matrix

$\hhat H_\mathrm{F}$0

Assuming an Ansatz $\hhat H_\mathrm{F}$1, the coefficients are reconstructed by

$\hhat H_\mathrm{F}$2

with the smallest singular value

$\hhat H_\mathrm{F}$3

serving as the central diagnostic. Below the Trotter threshold, if the Ansatz includes terms up to order $\hhat H_\mathrm{F}$4, the residual scales as $\hhat H_\mathrm{F}$5. Near and beyond the threshold, $\hhat H_\mathrm{F}$6 saturates to a $\hhat H_\mathrm{F}$7-independent plateau, the reconstructed coefficients become effectively random, and the same regime coincides with spectral statistics approaching the circular unitary ensemble, with $\hhat H_\mathrm{F}$8 rather than the Poisson value $\hhat H_\mathrm{F}$9 (Olsacher et al., 2022).

A different reconstruction strategy exploits periodicity directly in Fourier space. The Hamiltonian is written as

$U(T)$0

with total coefficient count $U(T)$1. Substituting the Fourier expansions into Schrödinger’s equation gives the Floquet-band relation

$U(T)$2

which is converted into a linear inverse problem $U(T)$3. The paper states that each correlator requires $U(T)$4 time-domain measurements, that the total sample complexity is

$U(T)$5

often summarized as $U(T)$6, and that the classical least-squares post-processing runs in $U(T)$7. It also introduces an adaptive cutoff rule that increments $U(T)$8 and stops when successive reconstructions differ by less than a threshold (Li, 2 Sep 2025).

These two lines of work share a common logic: the learning problem is made tractable by imposing the structural constraints already implied by periodic driving. In one case, the structure is the commutator hierarchy of the Floquet-Magnus expansion; in the other, it is the harmonic sparsity of a truncated Fourier representation.

5. Neural surrogates, unsupervised phase discovery, and periodic eigenproblems

Fourier Neural Operators extend Floquet-informed learning from reconstruction to surrogate modeling. In this formulation, the neural operator learns continuous maps between time-dependent functions, with Fourier layers

$U(T)$9

The paper studies three tasks: $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$0, $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$1, and $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$2 for operator growth. Hamiltonians are encoded in a local Pauli basis, the output Floquet Hamiltonian is truncated to a 3-local basis, and the network is trained with $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$3 Fourier layers and $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$4 retained modes. Reported results include relative error around $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$5 for Floquet Hamiltonian learning, relative error below about $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$6 across the period in the benchmarked setting, good magnetization dynamics for hundreds of periods, and CPU inference times such as about $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$7 s exact versus $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$8 s FNO for learning $$U_\varepsilon(k,t)=U(k,t)\exp\!\big[iH^{\rm eff}_\varepsilon(k)t\big],$$9 at $H_F$0 (Qi et al., 8 Sep 2025).

Unsupervised Floquet phase identification uses a different representation of periodic structure. Samples are compared through the full intra-period unitary by the similarity

$H_F$1

then improved by symmetry-preserving adiabatic deformations, embedded by a diffusion map, and clustered with $H_F$2-means. In the Floquet SSH model, the method recovers 4 phases when the 0 and $H_F$3 gaps are considered together; in the 2D bipartite model, it recovers 8 total Floquet topological phases, including anomalous Floquet insulators and an anomalous Floquet higher-order topological insulator (Ma et al., 2021). A related unsupervised framework, designed for a periodically driven non-Hermitian lattice, uses amortized clustering and an algorithm selector over K-means++, hDBSCAN, Agglomerative clustering, and Mean-shift. Its input tasks are the $H_F$4 eigenvectors of the finite lattice at each gain/loss value $H_F$5, and it classifies corner, edge, bulk, and hybrid corner-edge or edge-bulk Floquet modes in different non-Hermitian transition regimes (Xia et al., 1 Aug 2025).

The periodic-eigenproblem perspective also appears in a physics-informed neural solver for the Floquet-Bloch eigenproblem in 2D periodic potentials. There the networks $H_F$6 and $H_F$7 learn Bloch functions and band energies under the composite loss

$H_F$8

with $H_F$9 and $H_F$0. For the weak honeycomb potential $H_F$1, the reported final losses are total loss $H_F$2, PDE residual $H_F$3, normalization error $H_F$4, and boundary-condition error $H_F$5 (Mian, 20 Dec 2025). That work concerns Floquet-Bloch periodic operators rather than temporal driving in the narrow sense, but it shows the same general pattern: periodic symmetry is built into the learning architecture rather than recovered afterward.

6. Quantum algorithms, mixed quantum–classical extensions, and recurring limitations

Floquet-ADAPT-VQE imports the same logic into hybrid quantum-classical simulation. The periodically driven problem is embedded into a time-independent eigenproblem for $H_F$6, and the optimization target is the squared shifted Hamiltonian

$H_F$7

The ADAPT loop appends the operator with largest gradient and reoptimizes all parameters. A key design choice is the auxiliary initial state $H_F$8, for which the initial squared Floquet energy is independent of both the number of auxiliary qubits and the drive frequency. The method also provides a fixed-depth circuit for time-dependent observables through single-qubit $H_F$9-rotations on the auxiliary register. The paper emphasizes, however, that finite auxiliary truncation makes the method approximate and that the number of Pauli terms in auxiliary operators grows as $H_F$0 (Kumar et al., 14 Mar 2025).

In open-system nonadiabatic dynamics, the Floquet-informed algorithmic role is played by a master-equation reduction. The Floquet surface-hopping framework begins from a driven Anderson–Holstein impurity, derives a Floquet quantum master equation, takes a Wigner transform to obtain the Floquet classical master equation, and then solves it with three trajectory schemes: FSH, FaSH, and FaSH-density. The transition kernels are modified by the Floquet-dressed Fermi function $H_F$1, and the benchmark conclusion is that FaSH-density works best because it captures both fast oscillations due to the driving and correct steady-state observables (Wang et al., 2023).

A neighboring but non-ML computational development is the iterative algorithm for following individual Floquet states in high-dimensional Hilbert spaces. It uses repeated one-period propagation rather than diagonalization, supports maximum-likeness branch following under parameter changes, and was demonstrated for a periodically driven Bose-Hubbard chain with $H_F$2 sites, $H_F$3 bosons, unit filling, and Hilbert-space dimension $H_F$4. The drive-induced Mott-insulator-like target Floquet state can be populated with high efficiency when the amplitude is turned on smoothly but not too slowly, producing a window of pseudoadiabatic following (Krüger et al., 2021).

Several recurrent limitations emerge across the literature. Some are representation-specific: low-order Floquet-Magnus truncations fail beyond a threshold, auxiliary-space truncations can require more resources at lower frequencies or larger amplitudes, and Fourier reconstructions depend on band limitation or sufficiently decaying harmonic content (Olsacher et al., 2022, Kumar et al., 14 Mar 2025, Li, 2 Sep 2025). Others are methodological: unsupervised phase identification depends on dense enough sampling and optimization parameters, clustering-based mode classification is indirect and tied to the algorithm library, and neural solvers for periodic eigenproblems remain sensitive to loss weighting and degeneracies (Ma et al., 2021, Xia et al., 1 Aug 2025, Mian, 20 Dec 2025). The cumulative picture nevertheless is coherent. A Floquet-informed learning algorithm is best understood not as a single method, but as a design principle: when temporal periodicity organizes the physics, the algorithm gains efficiency or robustness by representing that periodicity directly, whether through Floquet eigenstates, Floquet generators, Fourier harmonics, symmetry-constrained intra-period unitaries, or extended Floquet-Hilbert spaces.