Mode-Resolved Green's Function Method

• Mode-resolved Green's function method is a technique that decomposes a global Green's function into channel-specific contributions to analyze resonances, scattering, and transport.
• It employs diverse mathematical architectures—including Dyson iteration, reduced basis transformation, and recursive recursion—to enable efficient computation and physical interpretation.
• Applications span nuclear physics, atomistic transport, and open photonics, providing accurate extraction of state parameters and improved device modeling.

Searching arXiv for recent and foundational papers related to mode-resolved Green's function methods across photonics, transport, and phonon scattering. Mode-resolved Green's function method denotes a family of formulations in which a Green's function, or an observable derived from it, is decomposed into contributions associated with specific modes—partial waves, quasinormal modes, device eigenmodes, lattice momenta, phonon branches, or operator/site indices—so that spectral, transport, scattering, and field-conversion processes can be analyzed channel by channel. In the literature, this organizing idea appears in relativistic mean-field studies of nuclear resonances, mode-space nonequilibrium Green's function implementations for atomistic transport, eigenspectrum-based atomistic Green's functions for phonons, Liouvillian continued-fraction constructions for electronic Green's functions, and quasinormal-mode Dyson expansions for open photonic systems (Chen et al., 2020, Lemus et al., 2020, Sadasivam et al., 2017, Foley, 2024, Fuchs et al., 14 Oct 2025).

1. Conceptual scope and meaning of “mode resolution”

In the cited literature, “mode” is not a universal object but a problem-dependent basis label. In nuclear structure it is a partial wave labeled by $\kappa$; in open photonics it is a quasinormal mode indexed by $\mu$; in atomistic transport it is a reduced basis eigenmode of the device cross-section; in phonon transport it is a branch- and wavevector-resolved propagating channel; in strongly correlated many-body theory it can be a site, orbital, spin, or other operator label; and in lattice Green's function theory it is often a momentum or symmetry sector (Chen et al., 2020, Fuchs et al., 14 Oct 2025, Lemus et al., 2020, Sadasivam et al., 2017, Foley, 2024, Ray, 2014).

The practical consequence is that “mode resolution” may refer to different resolved quantities. Depending on the formulation, one obtains partial-wave densities of states, off-diagonal Green's function matrix elements, mode-resolved transmission matrices, polarization-resolved self-energies, exact scattering cross sections for an incident channel, or spatially resolved phase-coherence metrics. This suggests that the phrase is best understood as a methodological principle—decomposition into physically interpretable channels—rather than as a single algorithm.

Domain	Mode label	Resolved object
Nuclear resonances	$\kappa$	$n_\kappa(\varepsilon)$, pole structure
Atomistic electron transport	Reduced basis indices	$\Sigma_{i,j}$, $G_{k,l}$
Strongly correlated systems	Site/orbital/spin operators	$G_{BC}(\omega)$
Phonon interface transport	$\alpha,\beta$	$\mathcal{T}_{\alpha\beta}(\omega,q_t)$
Defect phonon scattering	$\nu,k$	$\mu$0
Open photonics	QNM index $\mu$1	Few-mode Green's function expansion

2. Core mathematical architectures

A common starting point is the operator resolvent. In the relativistic mean-field formulation for single-particle states, the Green's function satisfies

$\mu$2

and admits the spectral representation

$\mu$3

so that the single-particle energies $\mu$4 are the poles of the Green's function (Chen et al., 2020). For spherical systems, the radial $\mu$5 Green's-function matrix is constructed from two independent Dirac solutions, one regular at the origin and one outgoing at infinity, together with the Wronskian.

A second major architecture is Dyson iteration. For a multi-cavity electromagnetic system, the $\mu$6-cavity Green's function is constructed recursively from the $\mu$7-cavity Green's function by

$\mu$8

with termination inside each cavity provided by a single-cavity quasinormal-mode expansion

$\mu$9

The formulation factorizes multi-cavity scattering into products of two-cavity processes and avoids nested integrals (Fuchs et al., 14 Oct 2025).

In atomistic nonequilibrium Green's function transport, mode resolution is introduced through a reduced basis transformation

$\kappa$0

followed by a mode-space representation of inelastic phonon scattering via the form-factor tensor

$\kappa$1

with self-energies of the form

$\kappa$2

The real part of the retarded scattering self-energy is obtained exactly in the reduced basis by the Kramers-Kronig relation

$\kappa$3

This construction preserves compatibility between spatially local scattering physics and a low-rank basis (Lemus et al., 2020).

A fourth architecture uses recursion in operator space. In the Liouvillian formalism,

$\kappa$4

and Lanczos tridiagonalization yields recursion coefficients $\kappa$5 and a continued-fraction representation,

$\kappa$6

Off-diagonal or mode-resolved matrix elements are then reconstructed by the LCCF or CFPH procedures, the latter using moments $\kappa$7 and primary and secondary polynomials (Foley, 2024).

3. Pole structure, densities of states, and resonant-state extraction

A particularly explicit mode-resolved Green's-function method appears in the study of single-particle resonant states within relativistic mean-field theory. Resonant states are identified either by searching for poles of the Green's function in the fourth quadrant of the complex energy plane, $\kappa$8, or by locating sharp extrema in the density of states and the immediate sign reversal after crossing the pole in the imaginary direction (Chen et al., 2020).

For a given partial wave $\kappa$9, the density of states is

$n_\kappa(\varepsilon)$0

Within this framework, the method treats both bound and continuum states equivalently and provides direct access to resonance parameters and density distributions in coordinate space. The paper emphasizes that this new approach is very effective for all kinds of resonant states, no matter it is broad or narrow, because it does not rely on peak shape in the real-energy density of states alone.

The application to $n_\kappa(\varepsilon)$1Sn illustrates the distinction between resolved partial-wave structure and unresolved continuum background. Four new broad resonant states, $n_\kappa(\varepsilon)$2, $n_\kappa(\varepsilon)$3, $n_\kappa(\varepsilon)$4 and $n_\kappa(\varepsilon)$5, are observed. For the very narrow resonant state $n_\kappa(\varepsilon)$6, the accuracy for the width is highly improved to be $n_\kappa(\varepsilon)$7 MeV. The dependence on the space size for the resonant energies, widths, and the density distributions in the coordinate space is found very stable. The resulting resonance parameters are very close to those by the complex momentum representation method and the complex scaling method. A common misconception is that Green's-function resonance extraction is reliable only for narrow, isolated states; this formulation was presented precisely to remove the ambiguity of broad or overlapping structures on the real axis.

4. Electronic transport, reduced bases, and superlattice recursion

In atomistically resolved electronic transport, mode resolution is chiefly a computational strategy for reducing dimensionality while retaining channel information. The mode-space-compatible inelastic scattering formulation extends atomistic NEGF to optical and acoustic phonons in silicon nanowires and includes the exact calculation of the real part of retarded scattering self-energies in the reduced basis representation using the Kramers-Kronig relations. Virtually perfect agreement with results of the original representation is achieved with matrix rank reductions of more than 97%. Time-to-solution improvements of more than $n_\kappa(\varepsilon)$8 and peak memory reductions of more than $n_\kappa(\varepsilon)$9 are shown, and this allows for the solution of electron transport scattered on phonons in atomically resolved nanowires with cross-sections larger than $\Sigma_{i,j}$0 nm $\Sigma_{i,j}$1 $\Sigma_{i,j}$2 nm (Lemus et al., 2020).

The same literature also identifies the principal approximation hierarchy inside mode-space transport. The full four-index form factor retains intra- and inter-mode scattering channels, whereas the diagonal approximation neglects the off-diagonal elements for indices $\Sigma_{i,j}$3 and $\Sigma_{i,j}$4, reducing complexity to $\Sigma_{i,j}$5. The paper states that this approximation yields minimal loss of accuracy for device-level quantities, though some mode-mixing processes are neglected. The real part of $\Sigma_{i,j}$6 is described as increasing the impact of scattering; omitting it leads to systematic errors such as underestimated off-state leakage and incorrect threshold voltages.

Recursive Green's function techniques play a complementary role for large superlattice-based devices. For systems such as twisted bilayer graphene with large supercells, the main bottleneck is the calculation of contact self-energies. The improved formulation treats only the surface region of the contact explicitly through small recursive slices and rewrites the problem so that contact self-energy computations no longer scale with the full supercell size. In the periodic case, the Green's function is written as

$\Sigma_{i,j}$7

which permits $\Sigma_{i,j}$8-resolved local density of states and band-structure extraction without constructing the full large Green's function. For twisted bilayer graphene with over $\Sigma_{i,j}$9 atoms, optimization by factors exceeding $G_{k,l}$0 in computation time is reported, and the paper cites small twist angles of $G_{k,l}$1, over $G_{k,l}$2 atoms, and superlattices with over a million atoms per device region as accessible regimes (Nguyen et al., 2024).

5. Phonon transmission, mode conversion, and defect scattering

For phonons, mode-resolved Green's-function methods are closely tied to the Caroli transmission formula and to the spectral structure of the leads. The eigenspectrum-based atomistic Green's function formulation rewrites the conventional AGF transmission

$G_{k,l}$3

as

$G_{k,l}$4

Here the contact surface Green's functions are constructed from the bulk phonon eigenspectrum via Dyson and Lippmann-Schwinger equations, creating a direct connection between the transmission function and the bulk phonon spectra of the materials forming the interface. Applied to Si-Ge interfaces with atomic intermixing, the method shows that intermixing of atoms near the interface increases the phase space available for phonon mode conversion and enhances thermal interface conductance at moderate levels of atomic mixing (Sadasivam et al., 2017).

A related development treats single-defect scattering rather than interface transmission. Using the extended AGF method, the numerically exact mode-resolved scattering cross section for a phonon scattered by a single lattice defect is computed as

$G_{k,l}$5

with the diagonal element of the transmission matrix furnishing the forward-scattering amplitude needed by the optical-theorem construction. The paper illustrates the approach with isotopic scattering in a monoatomic harmonic chain and with a carbon nanotube containing an encapsulated C$G_{k,l}$6 molecule. In the nanotube example, different phonon branches have dramatically different scattering cross sections; modes with atomic motion along the radial direction are strongly scattered, whereas modes with atomic motion along the axis are weakly scattered (Ong, 2024).

These formulations clarify a recurring theme in mode-resolved transport: the total conductance or total scattering rate can conceal qualitatively different channel behaviors. In the Si-Ge interface problem, the salient distinction is between specular and non-specular mode conversion. In the defect-scattering problem, it is between weakly and strongly coupled incident branches. In both cases, the Green's function is used not merely to sum over channels but to identify which channels dominate microscopic transport.

6. Open photonic systems and nonlinear guided-mode resonators

In open photonic systems, mode-resolved Green's functions are typically expressed in terms of quasinormal modes of lossy resonators. For multiple coupled optical resonators with finite retardation, the scattered electromagnetic Green's function is constructed from the quasinormal modes of the individual resonators within a few-mode approximation and a finite number of iteration steps, without requiring nested integrals. The method uses a Dyson scattering equation, converts volume integrals to surface integrals, and regularizes the QNMs outside the cavity so that explicit phase factors $G_{k,l}$7 capture finite propagation delays. It is reported to show excellent agreement with the full numerical Green's function for two coupled dipoles in the gaps of two metal dimers and to be easily extended to arbitrarily large separations and multiple cavities (Fuchs et al., 14 Oct 2025).

The two-cavity Green's function between emitters in different cavities is written in the separable form

$G_{k,l}$8

with the coupling coefficient $G_{k,l}$9 obtained from a surface integral involving the regularized field of one cavity evaluated on the surface of the other. For three cavities, the iterative method expresses the three-cavity Green's function in terms of products of two-cavity terms, exemplified by

$G_{BC}(\omega)$0

This factorization is one of the central structural advantages of the method.

A nonlinear extension appears in the Green's function integral method for second-harmonic generation in nonlinear guided-mode resonators. The second-harmonic field is represented as

$G_{BC}(\omega)$1

and the phase-matching factor is defined as a ratio between the modulus of the summed complex phasor field and the sum of the phasor magnitudes. The pointwise form is

$G_{BC}(\omega)$2

with an analogous port-integrated overall factor $G_{BC}(\omega)$3 (Liang et al., 22 Dec 2025).

In this formulation, the Green's function is mode-resolved in a spatially distributed sense: each nonlinear source point contributes a complex phasor whose phase and amplitude are set by the harmonic Green's function and the fundamental field. The paper states that conventional guided-mode resonators can exhibit severe phase mismatch even when local field enhancement is large. Guided by PMF profiles, the authors introduce a high-index waveguide layer that confines the fundamental field in the nonlinear material to regions where the harmonic Green's function varies slowly. The resulting configuration achieves a PMF exceeding $G_{BC}(\omega)$4, approaching the ideal value of unity, and yields an SHG efficiency of $G_{BC}(\omega)$5 at a low pump intensity of $G_{BC}(\omega)$6 kW/$G_{BC}(\omega)$7.

7. Exact lattice formulations, methodological boundaries, and recurrent misconceptions

An exact analytic branch of the subject appears in lattice Green's function theory. There the Green's function for displacement $G_{BC}(\omega)$8 is written in Fourier form as

$G_{BC}(\omega)$9

or, with a resolvent parameter $\alpha,\beta$0, as a multidimensional contour integral

$\alpha,\beta$1

Residue evaluation then yields representations in terms of generalized hypergeometric functions, with different lattices leading to $\alpha,\beta$2, $\alpha,\beta$3, Lauricella, or higher generalized hypergeometric forms. The method is also shown to be useful for computing Green's functions with next-nearest neighbor hopping (Ray, 2014). Here, mode resolution is effectively momentum or symmetry resolution rather than a reduced finite set of channels.

Several misconceptions recur across the literature. The first is that mode-resolved Green's-function methods constitute a single formalism. The papers instead show distinct constructions—Dyson equations, Lippmann-Schwinger equations, recursive block elimination, mode-space projections, operator-space Lanczos recursion, and contour-integral resolvents—linked by a common aim of channel decomposition. The second is that mode reduction is automatically exact. In the NEGF and photonic QNM settings, diagonal form-factor approximations and few-mode truncations are deliberate approximations, justified empirically in the cited works but not universal (Lemus et al., 2020, Fuchs et al., 14 Oct 2025).

A third misconception is that mode resolution removes all scaling barriers. The Liouvillian recursion method demonstrates otherwise. Although it extends recursion techniques to non-Hermitian operators and arbitrary temperature, its exact application to larger clusters is described as severely limited by exponential scaling, $\alpha,\beta$4 in orbitals for the exact method, because the Liouvillian acts on operators rather than states. The paper identifies tensor network methods and other variational techniques as promising routes for larger and more realistic systems (Foley, 2024). Similar scaling pressures motivate the reduced-basis and recursive-slicing strategies used in atomistic transport and superlattice modeling.

Taken together, these works define the mode-resolved Green's function method as a broad research program rather than a narrow technique. Its unifying function is to convert a global Green's-function object into a decomposition aligned with physically meaningful channels, thereby exposing pole structure, transmission pathways, mode conversion, scattering selectivity, spatial phase coherence, and local spectral content in forms that are often inaccessible to channel-summed observables alone.