Sommerfeld Spectral Representation

• Sommerfeld spectral representation is an analytic framework that expresses physical operator spectra via integrals and series, unifying discrete and continuous analyses.
• It employs self-adjoint extensions, boundary conditions, and the Kreĭn–Višik–Birman formula to precisely quantify eigenvalues in both Dirac–Coulomb systems and electromagnetic radiation problems.
• The approach enables fast numerical evaluation and asymptotic solution construction, ensuring accurate computations across multi-band and surface-wave scenarios.

The Sommerfeld spectral representation designates a family of analytic approaches wherein the spectrum of a physical operator or field configuration is expressed, classified, and quantitatively determined via integrals or series expansions over elementary spectral functions. Core instances include (1) the discrete spectra and eigensolutions for the Dirac-Coulomb Hamiltonian at critical coupling, rigorously classified and generalized from Sommerfeld’s original fine-structure formula, and (2) the spectral-domain integral representation for radiation fields above lossy interfaces, notably in the Sommerfeld radiation problem of classical electromagnetics. These formulations serve as both analytic vehicles and computational frameworks for the extraction of discrete and continuous spectral data, boundary-induced spectral multiplicity, and the precise evaluation of multi-band contributions in wave propagation.

1. Operator-Theoretic Foundations: Dirac–Coulomb Hamiltonian

In the critical coupling regime $|\nu| \equiv |−Z\alpha_f| \in (\sqrt{3}/2, 1)$, the radial Dirac–Coulomb operator is realized as an unbounded, symmetric operator on $L^2(\mathbb{R}^+, \mathbb{C}^2, dr)$ with deficiency indices $(1,1)$, admitting a one-parameter family of self-adjoint extensions $h_\beta$, indexed by $\beta \in \mathbb{R} \cup \{\infty\}$ (Gallone et al., 2017). Each extension is specified by a boundary condition at $r \downarrow 0$, derived from the two-term expansion:

$$\psi(r) = g_0 r^{-B} + g_1 r^{+B} + o(r^{1/2}), \qquad B = \sqrt{1 - \nu^2} \in (0, 1/2).$$

The domain $D(h_\beta)$ then enforces $$(g_1^+) / (g_0^+) = c_{\nu,\kappa} \beta + d_{\nu,\kappa}$$, with explicit constants determined by angular channel $\kappa$, Coulomb coupling $\nu$, and spectral index. The Kreĭn–Višik–Birman formula decomposes the resolvent:

$$h_\beta^{-1} = h_D^{-1} + \frac{1}{\beta \|\Phi\|^2} |\Phi \rangle \langle \Phi|,$$

where $\Phi$ solves the homogeneous adjoint equation.

2. Sommerfeld Fine-Structure and Generalized Quantization

Sommerfeld's fine-structure formula in the distinguished extension ($\beta = \infty$) is obtained via truncation conditions in ODE (confluent hypergeometric) or supersymmetric factorization frameworks. The eigenvalues in the spectral gap $(-1,1)$ are:

$$E_{n,\kappa} = -\mathrm{sign}(\nu) \left[ 1 + \frac{\nu^2}{(n+\gamma)^2} \right]^{-1/2}, \quad \gamma = \sqrt{\kappa^2 - \nu^2},$$

which for $\kappa = \pm 1$ yields:

$$E_n = -\mathrm{sign}(\nu) \left[ 1 + \frac{\nu^2}{(n + \sqrt{1-\nu^2})^2} \right]^{-1/2}.$$

Extending to arbitrary $\beta$, the discrete eigenvalues $E_n^{(\beta)}$ are determined implicitly:

$$\mathcal{F}_{\nu,\kappa}(E) = c_{\nu,\kappa} \beta + d_{\nu,\kappa},$$

where $\mathcal{F}_{\nu,\kappa}(E)$ is a strictly monotone function—ensuring one solution per gap, and so a pure-point spectrum in $(-1,1)$. The family of extensions produces a fibre bundle of spectral curves over the gap, with each branch ($\beta$ fixed, $n$ varying) forming a monotonically varying spectrum from band edge to band edge.

3. Spectral Measure, Resolvent Poles, and Extension Dependence

The spectral measure $\mu_\beta$ associated with $h_\beta$ is atomic:

$$\mu_\beta(\{E_n^{(\beta)}\}) = \|\psi_n^{(\beta)}\|^{-2},$$

with each atom corresponding to a Riesz projection onto the eigenfunction. Resolvent poles occur precisely at $E_n^{(\beta)}$, with residues quantifying the spectral weight. The essential spectrum remains fixed for all extensions: $$\sigma_\mathrm{ess}(h_\beta) = (-\infty, -1] \cup [1, \infty)$$, while the union of discrete eigenvalues over all $\beta$ fibers densely covers the gap.

In the ultra-critical limit $\nu \to 1^-$, $B \to 0^+$ and spectral spacings collapse, generating a $Z_2$ band-edge accumulation. For large $\beta$, $E_n^{(\beta)} \to E_{n,\kappa} + O(1/\beta)$. For $n \to \infty$, all extensions recover the non-relativistic (Coulombic Rydberg) asymptotics:

$$E_n^{(\beta)} - (-1) \simeq -\frac{\nu^2}{2(n+|\kappa|)^2} + O(n^{-3}).$$

4. Sommerfeld Representation in Electromagnetic Radiation Problems

The Sommerfeld spectral integral formalism yields field solutions for radiators above lossy interfaces (Sommerfeld radiation problem) (Bourgiotis et al., 2019). For a vertical Hertzian dipole of moment $p \hat{e}_x$ at height $x_0$ above ground with complex permittivity $\varepsilon_2$, the total electric field $\mathbf{E}$ at cylindrical coordinates $(\rho, x)$ decomposes via:

$$\mathbf{E} = -\frac{i\,p}{8\pi\varepsilon_0\varepsilon_1} \left\{ \int_{-\infty}^{+\infty} \mathbf{f}_1(k_\rho) dk_\rho + \int_{-\infty}^{+\infty} \mathbf{f}_2(k_\rho) dk_\rho \right\},$$

with explicit forms for $\mathbf{f}_1 (k_\rho)$ and $\mathbf{f}_2 (k_\rho)$ involving the reflection coefficient

$$R_\parallel(k_\rho) = \frac{\varepsilon_2 \kappa_1 - \varepsilon_1 \kappa_2}{\varepsilon_2 \kappa_1 + \varepsilon_1 \kappa_2},$$

and spectral Hankel/Bessel factors. The Sommerfeld integrals span all propagation bands: space-wave and surface-wave contributions, with branch cut management yielding finite-interval quadratures in variable $\xi$.

5. Fast Numerical Evaluation and Asymptotic Solution Construction

The paper (Bourgiotis et al., 2019) presents reformulation schemes eliminating algebraic/logarithmic singularities via a three-band mapping:

• $|k_\rho|\le k_{01} : k_\rho = k_{01}\sin\xi$, $\xi \in [-\pi/2, \pi/2]$,
• $k_\rho \ge k_{01} : k_\rho = k_{01}\cosh\xi$, $\xi \in [0, \infty)$,
• $k_\rho \le -k_{01} : k_\rho = -k_{01}\cosh\xi$, $\xi \in [0, \infty)$.

Thus, the integrals for $\mathbf{E}^R$ reduce to rapidly-convergent Bessel quadratures amenable to adaptive Simpson or trapezoidal schemes, achieving millisecond-level computational efficiency and error $<10^{-6}$ in at-scale simulations.

6. Compact Asymptotic Formulae and Limits

In the high-conductivity regime ($\sigma \gg \omega \varepsilon_0$), the dominant field arises from a saddle point on the deformed Sommerfeld contour. The “etalon-based” expression derived in (Bourgiotis et al., 2019) yields:

$$\mathbf{E}^R = -\hat{e}_{\theta_2} \frac{p k_{01}^3}{2 \varepsilon_0 \varepsilon_1} \sqrt{\frac{-2i}{\pi k_{01} \rho}} e^{i k_{01} r_2 \cos \zeta_p} \sin^{3/2}\theta_2 \, \sin\frac{\zeta_p}{2}\, R_\parallel (\theta_2)\, X(k_{01} r_2, -\zeta_p),$$

with $X(k, \alpha)$ a Fresnel–error-function expression and $\zeta_p = \xi_p - \theta_2$ parameterizing the proximity to the pole. This single formula unifies the geometrical-optics (stationary phase) reflected wave regime and the pseudo–surface-wave near-grazing regime. The conditions and limiting formulas delineate applicability domains:

• Stationary Phase (GO) limit for $\sqrt{k_{01} r_2 \sin(\varphi/2)} \gg 1$.
• Pseudo–surface-wave limit for $\varphi \to 0$, $k_{01} \rho \delta^2 < 1$.

7. Comparative Evaluation and Unified Spectral Structure

Numerical results presented in (Bourgiotis et al., 2019) confirm agreement of revised Sommerfeld integrals with the classical Norton surface-wave results and stationary-phase theory in the high-frequency far-field regime. The etalon-based asymptotic solution extends validity into shallow grazing angles and intermediate frequencies, outperforming stationary phase limits for local ground-hugging and surface-wave-dominated links. Figures in the referenced literature illustrate magnitude and spatial dependence, with asymptotics tracking direct numerical integration across frequencies and geometries.

A plausible implication is that such spectral-domain representations, via integrals or quantization conditions parameterized by extension indices or propagation geometry, provide both exhaustive classification and efficient computation of physical spectra—yielding explicit analytic formulas and structurally unified, numerically tractable models across a range of physical contexts.