Multispectral Floquet Scattering Matrix

• Multispectral Floquet scattering matrix is the operator that transforms a single-frequency input into a ladder of harmonics via periodic modulation.
• It encapsulates methods to analyze sideband conversion, pseudounitarity, and inelastic processes across electron, atomic, and photonic platforms.
• The framework impacts quantum optics, ultracold-atom scattering, and metasurfaces, offering insights into resonant frequency conversion and Floquet topology.

Searching arXiv for recent and foundational papers on Floquet scattering matrices, pseudounitarity, and multispectral sideband scattering. A multispectral Floquet scattering matrix is the scattering operator of a periodically driven system written in the basis of Floquet sidebands, so that an input at quasifrequency $\omega$ or energy $E$ is mapped to outputs at $\omega+n\Omega$ or $E+n\hbar\Omega$. In this representation, periodic modulation converts a nominally monochromatic or monoenergetic input into a ladder of harmonically related channels, and the scattering problem becomes intrinsically multi-frequency, multi-energy, or, in space-time modulated settings, multi-angle. The concept appears in mesoscopic electron quantum optics, ultracold-atom scattering, layered photonics, time-modulated metasurfaces, photonic time crystals, and Floquet topological transport, with the matrix structure encoding elastic processes on the diagonal and inelastic sideband conversion off the diagonal (Moskalets, 2013, Globosits et al., 2024, Gupta et al., 2017).

1. Definition and formal structure

For a periodically driven mesoscopic conductor, the outgoing annihilation operator is related to incoming operators by the Floquet scattering matrix

$$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$

so an electron entering with a well-defined energy leaves with a superposition of sidebands $E+n\hbar\Omega$. In matrix notation one often writes $S_{mn}(E)\equiv S_F(E_m,E_n)$, with $E_m=E+m\hbar\Omega$; the resulting object has a block Toeplitz structure in the sideband indices (Moskalets, 2013).

In ultracold-atom scattering, the same structure appears as an infinite matrix $S_{n'n}^l(E)$ in Floquet-channel space, where the asymptotic energies are $E_n=E_{\text{in}}+n\Omega$. The partial-wave amplitudes satisfy

$E$0

and the differential cross section into Floquet channel $E$1 is

$E$2

This makes the multispectral character explicit: the discrete label $E$3 is a channel index exactly analogous to a multichannel scattering problem, except that the channels differ by energy offsets induced by the drive (Smith, 2016).

For a finite photonic Floquet medium, the asymptotic expansion is written in channel amplitudes $E$4 over all sidebands $E$5, and the Floquet scattering matrix is the linear map

$E$6

Its matrix elements $E$7 describe conversion from an incoming wave in lead $E$8 and frequency channel $E$9 to an outgoing wave in lead $\omega+n\Omega$0 and channel $\omega+n\Omega$1. In this setting, diagonal channel blocks represent elastic scattering, whereas off-diagonal blocks represent inelastic Floquet frequency conversion (Globosits et al., 2024).

Space-time modulated Huygens’ metasurfaces add a second discrete index through the modulation wavevector. A monochromatic input at $\omega+n\Omega$2 generates harmonics

$\omega+n\Omega$3

so each temporal sideband emerges with its own transverse momentum and refraction angle. The corresponding field equations become a banded matrix system in the Floquet index $\omega+n\Omega$4, with couplings to $\omega+n\Omega$5 and $\omega+n\Omega$6 induced by the modulated Lorentz susceptibilities (Gupta et al., 2017).

2. Correlation functions, joint spectra, and many-particle structure

In electron quantum optics, the multispectral Floquet scattering matrix directly generates first- and second-order correlation functions. For emitted excitations, the first-order coherence is

$\omega+n\Omega$7

and the two-particle part satisfies the Slater-determinant relation

$\omega+n\Omega$8

The corresponding single-particle spectral distribution is

$\omega+n\Omega$9

and the joint two-particle distribution $E+n\hbar\Omega$0 contains the irreducible term

$E+n\hbar\Omega$1

At $E+n\hbar\Omega$2, $E+n\hbar\Omega$3, as required by the Pauli principle (Moskalets, 2013).

The same multispectral logic persists in quantum-optical $E+n\hbar\Omega$4-photon scattering. For excitation-number conserving systems with periodic $E+n\hbar\Omega$5, the $E+n\hbar\Omega$6-photon scattering matrix has the general form

$E+n\hbar\Omega$7

Hence total photon frequency is conserved only modulo $E+n\hbar\Omega$8, and the connected amplitudes inherit poles from the Floquet eigenvalues of the non-Hermitian effective Hamiltonian $E+n\hbar\Omega$9. For single photons, the sideband amplitudes are controlled by Fourier components of Floquet modes and resolvents of the form

$$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$0

while the two-photon connected part also involves the two-excitation Floquet spectrum $$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$1 (Trivedi et al., 2020).

A common simplification is to regard the Floquet matrix as merely a bookkeeping device for sideband intensities. The correlation-function formulations show that it is more restrictive: it is the kernel from which antisymmetry, joint spectral densities, and multiphoton pole structure are built.

3. Electromagnetic implementations and matrix constructions

For space-time modulated Huygens’ metasurfaces, Generalized Sheet Transition Conditions and Lorentzian surface susceptibilities lead, after Floquet expansion, to a finite block-matrix equation

$$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$2

where $$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$3 contains reflected and transmitted harmonic amplitudes and $$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$4 is assembled from banded Toeplitz matrices $$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$5 and a diagonal $$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$6. Solving

$$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$7

yields the scattered spectra, with the reflected and transmitted Floquet amplitudes contained in the subblocks $$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$8 and $$\hat b(E)=\sum_{n=-\infty}^{\infty} S_F(E,E_n)\hat a(E_n),\qquad E_n=E+n\hbar\Omega,$$9. The off-diagonal bands encode first- and second-order harmonic coupling and thereby determine the strength of higher harmonics as modulation depth and frequency are varied (Gupta et al., 2017).

Layered optomagnonic structures admit a different but closely related construction. The periodic spin-wave modulation of the Bi:YIG permittivity tensor generates optical sidebands $E+n\hbar\Omega$0. In each homogeneous layer, the field envelopes are expanded as

$E+n\hbar\Omega$1

and the layer modes are obtained from a Floquet eigenproblem for $E+n\hbar\Omega$2. Interface continuity conditions then define interface scattering matrices, which are rephased into $E+n\hbar\Omega$3 blocks and recursively composed through the multilayer. The total transmitted and reflected intensities are sums over sideband-resolved contributions $E+n\hbar\Omega$4 and $E+n\hbar\Omega$5, making the structure explicitly multispectral (Pantazopoulos et al., 2019).

A single time-modulated conducting sheet supports an especially compact pole-expansion representation. For a Floquet-sheet resonator with a resonant surface mode in the $E+n\hbar\Omega$6 Floquet channel and 0th-order propagating external channels, temporal coupled-mode theory gives

$E+n\hbar\Omega$7

where $E+n\hbar\Omega$8 is the background 0th-order scattering matrix, $E+n\hbar\Omega$9 is the surface-mode frequency, and the radiative decay rate obeys

$S_{mn}(E)\equiv S_F(E_m,E_n)$0

The factor $S_{mn}(E)\equiv S_F(E_m,E_n)$1 is a direct consequence of photon-number conservation during frequency conversion and distinguishes the Floquet-sheet TCMT from static TCMT (Wang et al., 22 Feb 2026).

These implementations differ in spatial basis—plane waves, multipoles, defect modes, or sheet resonances—but they share the same organizing principle: periodic temporal modulation turns scattering from a single-frequency boundary-value problem into a coupled linear system over a ladder of sidebands.

4. Conservation laws, current metrics, and generalized unitarity

In static lossless scattering, flux-normalized scattering matrices are unitary. In periodically driven photonic media, ordinary unitarity generally fails because energy is exchanged with the modulation. The correct invariant is wave action, and in photon-flux normalization the Floquet scattering matrix satisfies the pseudounitary relation

$S_{mn}(E)\equiv S_F(E_m,E_n)$2

with $S_{mn}(E)\equiv S_F(E_m,E_n)$3 an indefinite metric that assigns $S_{mn}(E)\equiv S_F(E_m,E_n)$4 to positive-frequency channels and $S_{mn}(E)\equiv S_F(E_m,E_n)$5 to negative-frequency channels. In this basis, deviation from $S_{mn}(E)\equiv S_F(E_m,E_n)$6 is below numerical precision $S_{mn}(E)\equiv S_F(E_m,E_n)$7, whereas $S_{mn}(E)\equiv S_F(E_m,E_n)$8 can deviate strongly near channels where positive and negative frequencies couple (Globosits et al., 2024).

The same idea appears in one-dimensional Floquet lattices in the language of a conjugate symplectic transfer matrix. Writing the spatial evolution as $S_{mn}(E)\equiv S_F(E_m,E_n)$9, current conservation is equivalent to

$E_m=E+m\hbar\Omega$0

with

$E_m=E+m\hbar\Omega$1

This implies reciprocal eigenvalue pairing, biorthogonality in the symplectic metric, and a flux-conserving scattering problem in the propagating subspace (Zhang et al., 27 Jan 2026).

A recurrent misconception is that negative-frequency channels are a numerical nuisance that can always be discarded. The photonic Floquet formulation shows that this is true only in restricted weak-coupling limits where positive and negative frequencies do not mix. In general, these channels are propagating, carry photon flux, and are precisely where pseudounitarity departs from ordinary unitarity (Globosits et al., 2024).

5. Resonances, singularities, exceptional points, and CPA-lasing

The analytic structure of the multispectral Floquet scattering matrix is governed by poles associated with quasi-bound or Floquet resonant states. In driven short-range quantum scattering, poles of the Floquet S-matrix can cross the real energy axis as a function of drive amplitude. At the corresponding singular point, the S-matrix becomes nonanalytic, and an incoming low-energy s-wave can be totally absorbed and converted into high-energy outgoing waves that are mostly p-wave. The same analysis shows discontinuous jumps in the angular momentum and energy of emitted or absorbed waves as the singular point is crossed (Landa, 2017).

In periodically time-varying photonic media, pseudounitarity implies that the eigenvalues of the Floquet scattering matrix either lie on the unit circle or occur as inverse complex-conjugate pairs. With increasing driving strength, eigenvalues on the unit circle meet at exceptional points and then split into the broken-symmetry regime. In time-symmetric systems, the associated symmetry operator corresponds to the time-reversal operator. At the parametric resonance condition, one eigenvalue vanishes while its partner diverges, signifying simultaneous coherent perfect absorption and lasing (Globosits et al., 3 Oct 2025).

For slabs, spheres, and metasurfaces, this CPA-lasing transition is not tied to a single geometry. A slab admits a two-band approximation in which the exceptional-point threshold scales with static Fabry–Pérot resonance conditions; a time-modulated sphere exhibits the same zero/divergent eigenvalue structure in a Floquet–Mie basis; and a metasurface sustaining quasi-bound states in the continuum can reach CPA and lasing for minimal modulation strength (Globosits et al., 3 Oct 2025).

This suggests that pole-zero collisions, exceptional points, and perfect frequency-converting absorption are not pathologies of a particular basis choice. They are structural consequences of the indefinite metric and the frequency-mixing nature of periodically modulated scattering.

6. Wavefront shaping, effective-Hamiltonian reductions, and optimal pulses

Pseudounitary Floquet scattering matrices support a direct generalization of the Wigner–Smith operator,

$E_m=E+m\hbar\Omega$2

which is Hermitian in an energy-flux normalized basis. Its eigenstates are multispectral light pulses that are optimally shaped in both spatial and temporal degrees of freedom for controlling time-varying media. The associated eigenvalue obeys the near-field relation

$E_m=E+m\hbar\Omega$3

so the eigenstates maximize or minimize a period-averaged overlap between field intensity and the parameter-sensitivity profile (Globosits et al., 2024).

Examples include pulses that maximize intensity inside a modulated target, focus at the target boundaries, avoid the target, or exert maximal optical force on the target center. For step-like modulation, the same framework selects pulses localized in prescribed temporal windows or focused at the jump times $E_m=E+m\hbar\Omega$4 and $E_m=E+m\hbar\Omega$5 (Globosits et al., 2024).

A complementary high-frequency viewpoint replaces the full multispectral problem by an effective Hamiltonian plus micromotion-renormalized couplings. In the single-propagating-channel limit, the Floquet scattering matrix reduces to

$E_m=E+m\hbar\Omega$6

showing that micromotion modifies not only the effective Hamiltonian $E_m=E+m\hbar\Omega$7 but also the couplings to the leads. In the two-mode example studied there, the leading non-reciprocal transport term is $E_m=E+m\hbar\Omega$8 and is controlled by the imaginary off-diagonal component of the micromotion correction (Li et al., 2018).

7. Topological and experimental perspectives

Scattering formulations of Floquet topology use the reflection block of a unitary or pseudounitary scattering matrix rather than bulk bands alone. For periodically driven systems with absorbing terminals, the Floquet scattering matrix

$E_m=E+m\hbar\Omega$9

yields strong invariants from the winding of $S_{n'n}^l(E)$0 and weak invariants from the signs of $S_{n'n}^l(E)$1 at special twists. These invariants remain valid even when all bulk Floquet bands are trivial and can describe weak topological Floquet phases that are destroyed by breaking translational symmetry in time rather than in space (Fulga et al., 2015).

A static higher-order topological insulator can simulate a lower-dimensional Floquet operator through its reflection matrix $S_{n'n}^l(E)$2. In that construction, $S_{n'n}^l(E)$3 acts as an effective Floquet operator, and a nested scattering matrix

$S_{n'n}^l(E)$4

produces $S_{n'n}^l(E)$5 invariants

$S_{n'n}^l(E)$6

This establishes a direct equivalence between static HOTI boundary scattering and lower-dimensional Floquet topology (Franca et al., 2020).

In one-dimensional Floquet lattices with spatially adiabatic boundaries, the multispectral scattering matrix simplifies enough to expose a direct bulk–boundary correspondence: the integrated imbalance of left and right transmittances equals $S_{n'n}^l(E)$7, where $S_{n'n}^l(E)$8 is the bulk winding number. In that setting, the bulk invariant appears as a rigid shift in the transmission energy windows (Zhang et al., 27 Jan 2026).

Experimental access depends on platform. In mesoscopic electron optics, energy filters and zero-frequency shot-noise measurements reconstruct the single-particle spectral function $S_{n'n}^l(E)$9 and the joint spectral density $E_n=E_{\text{in}}+n\Omega$0 from dc currents and cross-correlators (Moskalets, 2013). In space-time metasurfaces, Floquet predictions have been validated against GSTC-FDTD calculations and Fourier propagation of refracted harmonics (Gupta et al., 2017). In Floquet-sheet resonators, the TCMT reduction is validated by COMSOL simulations, including the lossy case (Wang et al., 22 Feb 2026). In layered optomagnonic cavities, the time-Floquet scattering-matrix approach identifies the regime where adiabatic approximations fail and strong inelastic light scattering emerges under triple resonance (Pantazopoulos et al., 2019).

Taken together, these developments define the multispectral Floquet scattering matrix as a unifying object for periodically driven scattering: it is simultaneously a sideband-coupling kernel, a generator of many-body and multiphoton correlation spectra, a pseudounitary operator constrained by conserved action or current metrics, and a practical design tool for resonant conversion, wavefront shaping, and topological diagnosis across electronic, atomic, and photonic platforms.