Degenerate Floquet Perturbation Theory

• Degenerate Floquet Perturbation Theory is a framework for periodically driven quantum systems that reorganizes perturbative expansions in Floquet space to handle resonant degeneracies.
• It constructs effective Hamiltonians within a resonance-adapted subspace using methods like Brillouin-Wigner, van Vleck, and extended degenerate schemes to accurately capture near-degeneracies.
• The theory has broad applications from two-level spin resonance to many-body systems, elucidating resonance transitions, Rabi oscillations, and nonlinear dynamical effects.

Searching arXiv for recent and foundational papers on degenerate Floquet perturbation theory and related formulations. Degenerate Floquet perturbation theory is a perturbative framework for periodically driven quantum systems in which the relevant small-denominator structure is organized in Floquet space rather than in the undriven Hilbert space. It is used when quasienergy levels become degenerate or nearly degenerate because energy differences are compensated by integer multiples of the drive frequency, so that standard non-degenerate perturbation theory or conventional high-frequency expansions fail or become unreliable. Across two-level, few-level, open-system, and many-body settings, the central operation is the construction of an effective Hamiltonian acting within a resonance-adapted subspace of the Floquet-Hilbert space, with different implementations based on Brillouin-Wigner, van Vleck, adiabatic-frequency, and extended degenerate perturbation schemes (Feng et al., 2024, Braver et al., 2024, Eckardt et al., 2015, Mikami et al., 2015).

1. Floquet-space formulation and the origin of degeneracy

Floquet theory reformulates the Schrödinger problem for a periodic Hamiltonian $H(t+T)=H(t)$ as a time-independent eigenvalue problem in an extended Hilbert space. In this representation, the Hamiltonian is decomposed into Fourier components, and the quasienergy operator acquires a block structure labeled by a Fourier or “photon” index. For a periodic Hamiltonian written as

$H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$

the associated Floquet Hamiltonian is block tridiagonal, with diagonal blocks $H_0+n\omega I$ and off-diagonal couplings $H_1$ and $H_{-1}$ that connect neighboring sectors (Feng et al., 2024). More generally, in the Fourier basis the quasienergy operator has matrix elements

$Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$

which makes explicit that the drive frequency separates otherwise identical copies of the spectrum into Floquet zones (Eckardt et al., 2015).

Degeneracy arises when an unperturbed level in one Floquet sector coincides with another level shifted by an integer multiple of the driving frequency. In the simplest two-level resonance problem this appears as

$E_a \approx E_\beta + \omega,$

which identifies a near-degenerate pair $|a,0\rangle$ and $|\beta,1\rangle$ (Feng et al., 2024). In driven nonlinear systems the same condition takes the form

$\hbar\omega_{\rm ex}(n'-n)=E_k-E_j,$

so that $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$0 defines an $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$1-photon resonance (Vierheilig et al., 2010). In weak-field radical-pair problems, near resonances occur when $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$2, so that only nearly degenerate Floquet subspaces need to be mixed (Hiscock et al., 2016). The common feature is the emergence of small or vanishing energy denominators, which invalidate regular perturbative expansions and require a degenerate treatment.

This Floquet-space degeneracy is not limited to isolated crossings. In the low-frequency regime, many Floquet channels can become nested together and are coupled by the laser field, producing a mathematically singular limit for standard Fourier-based perturbation theory (Martiskainen et al., 2014). In many-body systems or weakly disordered spectra, dense sets of resonances can proliferate, and denominators of the form $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$3 become small for many pairs, again driving perturbative breakdown (Pakrashi et al., 14 Jan 2026). This establishes degeneracy not as a peripheral complication but as a structural property of large classes of driven systems.

2. Construction of degenerate subspaces and effective Hamiltonians

The operational core of degenerate Floquet perturbation theory is the selection of a model subspace $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$4 containing degenerate or nearly degenerate Floquet states, together with its complement $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$5. In the spin-resonance formulation based on Floquet theory and the Brillouin-Wigner perturbation method, the resonance-adapted subspace is

$H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$6

and the effective Hamiltonian is written as

$H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$7

in recursive perturbative form (Feng et al., 2024). This construction is explicitly designed to handle near-degeneracies that are central to resonance phenomena.

A related strategy appears in van Vleck perturbation theory for the driven quantum Duffing oscillator. Near a one-photon resonance, the relevant manifold is spanned by $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$8 and $H(t)=H_0+H_1 e^{i\omega t}+H_{-1} e^{-i\omega t},$9, and the effective block Floquet Hamiltonian takes the form

$H_0+n\omega I$0

with first-order off-diagonal hybridization and second-order shifts evaluated within the resonant doublet (Vierheilig et al., 2010). The same logic extends to three-level systems with strong transverse driving, where a double-unitary-transformation brings the problem into a form suitable for generalized van Vleck nearly degenerate perturbation theory, reducing the infinite-dimensional Floquet Hamiltonian to a finite effective one (Han et al., 2020).

High-frequency formulations implement an analogous block-diagonalization, but with the zero-photon subspace as the model space. In the Floquet-space perspective of the high-frequency approximation, one treats the “photon” term as the unperturbed operator and all Fourier components of the Hamiltonian as perturbations, then constructs a unitary transformation that block diagonalizes the quasienergy operator with respect to photon number (Eckardt et al., 2015). Brillouin-Wigner theory gives a closely related projection formalism on the zero-photon subspace and yields a systematic $H_0+n\omega I$1 expansion for an effective Hamiltonian reproducing the quasienergies and eigenstates up to the desired order (Mikami et al., 2015).

A more recent development generalizes the definition of the degenerate manifold itself. Extended degenerate perturbation theory for the Floquet-Hilbert space proposes that the degenerate subspace include not only the degenerate levels of interest but rather all levels in a Floquet zone (Braver et al., 2024). In this construction one first reduces each unperturbed energy $H_0+n\omega I$2 into a chosen Floquet zone,

$H_0+n\omega I$3

then treats all intra-zone couplings exactly while couplings to other zones are handled perturbatively (Braver et al., 2024). This widens the usual scope of degenerate perturbation theory from isolated crossings to zone-wide mixing.

3. Resonance conditions, selection rules, and quasienergy structure

Within this framework, resonance is controlled by matrix elements that couple the degenerate manifold. In the two-level spin-resonance problem, the first-order effective Hamiltonian in the basis $H_0+n\omega I$4 is

$H_0+n\omega I$5

and the upper triangular element $H_0+n\omega I$6 determines whether the resonance happens (Feng et al., 2024). If $H_0+n\omega I$7, coupling allows Rabi oscillations between $H_0+n\omega I$8 and $H_0+n\omega I$9; if $H_1$0, no direct coupling exists, so, despite degeneracy, no resonance transition occurs (Feng et al., 2024). This provides a non-RWA-centric resonance selection rule formulated directly in terms of Floquet Fourier components.

For the driven Duffing oscillator, the resonance condition is expressed as

$H_1$1

and the one-photon resonance corresponds to the degeneracy of $H_1$2 and $H_1$3 (Vierheilig et al., 2010). In the periodically driven tilted Fermi-Hubbard chain, the relevant degeneracy is between Fock states under one-period evolution: $H_1$4 which, for the specific doublon-related process discussed in the model, leads to

$H_1$5

as the scar condition (Huang et al., 2 Apr 2025). There the resonant manifold is formed by degenerate Fock bases that can be connected by one hopping process, and the underlying physical mechanism is identified to be the Floquet resonances between these degenerate Fock bases that can be connected by one hopping process. It is the first-order hopping perturbation effect (Huang et al., 2 Apr 2025).

In open radical-pair systems subjected to weak radiofrequency fields, the same structural principle appears in a computationally selective form. Only nearly degenerate subspaces contribute appreciably to observables because the singlet-yield formula contains the factor

$H_1$6

which suppresses contributions from widely separated quasienergies (Hiscock et al., 2016). This allows the dynamics to be projected onto active near-resonant subspaces while bypassing the full infinite Floquet matrix.

A recurring misconception is that adiabaticity is guaranteed whenever the drive frequency is much smaller than the instantaneous spectral gap. A driven spin-$H_1$7 counterexample shows that when the parameters of the Hamiltonian lead to a quasi-degeneracy in the Floquet spectrum, the evolution is not adiabatic even if the frequency of the field is much smaller than the spectral gap of the Hamiltonian (Russomanno et al., 2017). In that setting the relevant condition is not the instantaneous gap alone but rather the Floquet resonance relation

$H_1$8

which induces long-lived beating and necessitates degenerate rather than non-degenerate treatment (Russomanno et al., 2017).

4. Perturbative orders, effective dynamics, and representative observables

Once the effective Hamiltonian in the degenerate manifold is constructed, its diagonalization yields resonance splittings, transition probabilities, and frequency shifts. In the spin-resonance problem, diagonalization gives

$H_1$9

so that the generalized Rabi frequency is

$H_{-1}$0

and the transition probability is

$H_{-1}$1

(Feng et al., 2024). In this example the generalized Rabi frequency is the first-order solution, while the Bloch-Siegert shift emerges as the second-order solution (Feng et al., 2024).

Second-order corrections typically renormalize the resonance condition through virtual excursions خارج the model subspace. In the same two-level formulation, inclusion of the second-order term yields an upward resonance shift

$H_{-1}$2

identified as the Bloch-Siegert shift (Feng et al., 2024). In the Duffing problem, second-order terms $H_{-1}$3 and $H_{-1}$4 correct the resonant doublet energies and determine the quasienergy splitting and mixing angle near resonance (Vierheilig et al., 2010).

Low-frequency perturbation theory can organize corrections differently. In the adiabatic-frequency expansion for quasienergy Floquet solutions, the Floquet operator is written as

$H_{-1}$5

and the quasienergy expansion is

$H_{-1}$6

with all odd-order terms vanishing (Martiskainen et al., 2014). Here the zero-order quasienergy is the period average

$H_{-1}$7

and the zero-order Floquet functions can be improved by a phase factor

$H_{-1}$8

to match the phase of the exact Floquet solution more accurately (Martiskainen et al., 2014). This is not the same perturbative regime as high-frequency expansions, but it addresses a closely related degeneracy problem created by nested Floquet channels at $H_{-1}$9.

At a more non-perturbative level, exact-WKB analysis of two-level Floquet systems shows that perturbative low-frequency expansions miss exponentially small gap openings at resonant points. There the quasi-energy and Floquet effective Hamiltonian are expressed in terms of cycle integrals, and resonant oscillations reveal non-perturbative features that cannot be captured by the perturbative expansion (Fujimori et al., 17 Apr 2025). This suggests that degenerate Floquet perturbation theory captures the leading resonant structure, but exponentially weak tunneling effects may require methods beyond any finite perturbative order.

5. Variants of the theory and their domains of applicability

Several technically distinct versions of degenerate Floquet perturbation theory coexist, each matched to a different parameter regime.

Formulation	Core idea	Typical regime
Brillouin-Wigner Floquet resonance treatment	Projection onto a resonance-adapted model subspace with recursive effective Hamiltonian	Two-level or few-level near-resonance problems (Feng et al., 2024)
van Vleck quasi-degenerate Floquet theory	Block-diagonal effective Hamiltonian for nearly degenerate manifolds	Near $Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$0-photon resonances, e.g. Duffing oscillator (Vierheilig et al., 2010)
Floquet-space high-frequency block diagonalization	Degenerate perturbation theory across photon sectors, projected to zero-photon block	Off-resonant high-frequency engineering (Eckardt et al., 2015, Mikami et al., 2015)
Adiabatic-frequency perturbation theory	Expansion in $Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$1 using adiabatic states as zero order	Low-frequency nested-channel regime (Martiskainen et al., 2014)
Extended DPT (EDPT)	Exact treatment of all intra-zone couplings within one Floquet zone	Resonant interacting systems with broad intra-zone mixing (Braver et al., 2024)

Brillouin-Wigner theory for high-frequency expansion constructs an effective Hamiltonian on the projected zero-photon subspace and can write down the whole infinite series expansion, as compared to the van Vleck degenerate perturbation theory (Mikami et al., 2015). It also avoids spurious dependence on the driving phase that can appear in Floquet-Magnus truncations (Mikami et al., 2015). By contrast, the Floquet-space block-diagonalization approach emphasizes the joint derivation of the effective Hamiltonian and the micromotion operator, again through degenerate perturbation theory in the extended Floquet Hilbert space (Eckardt et al., 2015).

Van Vleck and Brillouin-Wigner methods differ less in physical content than in technical organization. The Duffing analysis shows that standard Floquet perturbation works well only when there is no near degeneracy, whereas the quasi-degenerate van Vleck approach is required at and near resonance to capture the splitting of the doubly-degenerate level and the associated hybridization (Vierheilig et al., 2010). The spin-resonance analysis similarly emphasizes that the Brillouin-Wigner method enables going to higher orders in a transparent way and gives a direct, physically transparent construction of the relevant model subspace (Feng et al., 2024).

EDPT modifies the usual notion of what counts as “degenerate.” Rather than isolating only the resonant pair or manifold, it includes all states within a chosen Floquet zone as forming the degenerate subspace, so that all intra-zone couplings are treated exactly (Braver et al., 2024). The resulting approach is shown to resemble a high-frequency expansion, provided the quasienergy matrix is constructed such that each $Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$2th diagonal block contains energies reduced to the $Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$3th Floquet zone (Braver et al., 2024). This makes EDPT a bridge between narrow-manifold DPT and global Floquet-zone effective-Hamiltonian methods.

6. Breakdown, convergence, and non-perturbative limits

Degenerate Floquet perturbation theory is designed to repair the failure of non-degenerate expansions near resonances, but it also has its own domains of validity. In the low-frequency adiabatic-based expansion, convergence is controlled by the nearest branch point in the complex $Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$4 plane. For a two-level adiabatic basis, the branch point occurs at

$Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$5

and the radius of convergence is

$Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$6

(Martiskainen et al., 2014). As the field amplitude increases, the radius of convergence shrinks (Martiskainen et al., 2014).

EDPT gives a corresponding convergence criterion,

$Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$7

for all relevant states and Fourier indices, except as excluded by the theory (Braver et al., 2024). This makes explicit that even after reorganizing the perturbation problem around a degenerate Floquet zone, inter-zone resonances can still destroy perturbativity.

A broader limitation appears when resonances become dense rather than isolated. In periodically driven quasiperiodic lattices, exact Floquet dynamics, Floquet perturbation theory, and optimal-order van Vleck analysis reveal that the van Vleck expansion achieves superasymptotic accuracy up to an optimal order; it ultimately breaks down due to resonant hybridization at a weak quasiperiodic potential (Pakrashi et al., 14 Jan 2026). There the effective Hamiltonian contains resonant denominators

$Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$8

and at weak disorder a proliferation of resonances causes the expansion to fail everywhere (Pakrashi et al., 14 Jan 2026). The observed localization is therefore nonperturbative in origin (Pakrashi et al., 14 Jan 2026).

The low-frequency driven spin-$Q_{m'm}=H_{m'-m}+\delta_{m'm}m\hbar\omega,$9 problem points to the same conclusion from another direction. Because quasi-degeneracies accumulate as $E_a \approx E_\beta + \omega,$0, the zero-frequency limit is singular, and an explanation based on a perturbation theory in $E_a \approx E_\beta + \omega,$1 cannot be given (Russomanno et al., 2017). The system can instead be described by a mapping to an extended Hilbert space in terms of resonances of an effective two-band Wannier-Stark ladder, where near-degenerate levels undergo Rabi-like oscillations (Russomanno et al., 2017). This suggests that degenerate Floquet perturbation theory is often locally correct around isolated resonances but may require non-perturbative completion in dense-resonance regimes.

7. Applications and physical significance across quantum platforms

The practical reach of degenerate Floquet perturbation theory spans single-particle spectroscopy, nonlinear oscillators, open quantum systems, and interacting many-body matter. In two-level spin resonance, the method yields a direct criterion for whether resonance happens, and solves the generalized Rabi frequency and the Bloch-Siegert shift straightforwardly as first-order and second-order solutions (Feng et al., 2024). In the dissipative quantum Duffing oscillator, the quasi-degenerate van Vleck approach is required near resonance and correctly captures splitting, hybridization, and antiresonant lineshapes in the dissipative response (Vierheilig et al., 2010).

In radical-pair chemistry, a modified Floquet theory treating the time-dependent magnetic field as a perturbation exploits the slow radical-pair recombination and approximates the product yield by considering only nearly-degenerate sub-spaces of the Floquet space (Hiscock et al., 2016). The resulting method is found to give product yields in good agreement with exact quantum mechanical results for a variety of simple model radical pairs and can be applied to radical pairs containing significantly more nuclear spins (Hiscock et al., 2016). A plausible implication is that degenerate Floquet perturbation theory can be used not only to describe resonance physics but also as a complexity-reduction principle for observables dominated by narrow quasienergy differences.

In the periodically driven tilted Fermi-Hubbard chain, degenerate Floquet perturbation theory derives the exact conditions under which Floquet scarring states emerge (Huang et al., 2 Apr 2025). The mechanism is resonance between degenerate Fock bases connected by one hopping process, and phenomena such as quantum revivals and subharmonic responses are studied within this framework (Huang et al., 2 Apr 2025). This places Floquet many-body scars within the same conceptual lineage as few-level resonance doublets: a projected effective Hamiltonian on a resonant manifold organizes the emergent nonergodic dynamics.

For resonantly driven interacting systems more generally, EDPT provides a practical compromise between accuracy and efficiency. Applied to a driven Bose-Hubbard model, it yields more accurate quasienergy spectra than the conventional DPT, while its computational complexity is intermediate between DPT and the numerically exact approach (Braver et al., 2024). This suggests that zone-wide degenerate constructions may be especially useful in interacting problems where narrow-manifold projections miss important intra-zone mixing.

Taken together, these developments show that degenerate Floquet perturbation theory is not a single algorithm but a family of effective-Hamiltonian constructions adapted to the resonance structure of periodically driven systems. Its unifying principle is the same in each case: identify the Floquet states for which the quasienergy denominator structure becomes singular or nearly singular, treat their mixing non-perturbatively within an effective subspace, and organize all remaining couplings perturbatively. Where resonances are isolated, this program yields compact analytical control; where resonances become dense, it also marks the boundary beyond which intrinsically nonperturbative Floquet phenomena must be expected (Feng et al., 2024, Braver et al., 2024, Pakrashi et al., 14 Jan 2026).