Representation of electromagnetic fields generated by dipoles in the presence of a perfectly conducting half-plane via Macdonald integrals
A. A. Tuzhilin
Submitted 1967-01-01 | RussiaRxiv: ru-196701.06481 | Translated from Russian

Abstract

A representation of Hertz vectors (electric or magnetic) and their corresponding electromagnetic fields $\mathbf E$ and $\mathbf H$, generated by arbitrarily oriented dipoles (electric or magnetic, respectively) in the presence of a perfectly conducting half-plane, has been obtained in terms of Macdonald integrals. The theory of Macdonald integrals, as special functions of two variables, was recently developed by the author to an extent that allows for the creation of programs to compute these functions. The use of Macdonald integrals in the diffraction problems under consideration enables numerical field calculations over a wide range of parameters. Bibliography: 15 items.

Full Text

Preamble

This section presents a mathematical analysis of electromagnetic fields in the presence of specific boundary conditions, building upon the foundational work of T. Senior \cite{1} and Yu. V. Vandakurov \cite{2}. The methods developed by these authors have been further refined in subsequent studies \cite{3, 4, 5, 6}, as well as in the works of G. D. Malyuzhinets \cite{7, 8, 9}.

We consider a source located at the point $r_0 = (\rho_0, \phi_0, z_0)$ in a cylindrical coordinate system. The primary field is generated by a dipole with moment $P$, which can be expressed as:
$$P = P (\cos \theta_0 e_1 + \sin \theta_0 e_2)$$
where $e_1$ and $e_2$ are unit vectors. Using the representation (1.12) for the function $G_\nu(k, R)$, we have:
$$G_{1/2}(k, R) = \sqrt{\frac{\pi}{2kR}} H_{1/2}^{(1)}(kR) = \frac{e^{ikR}}{R}$$ (1.13)
The potential $\Pi(\rho, \phi, z)$ can be expressed as:
$$\Pi(\rho, \phi, z) = G_{1/2}(k, R(\alpha)) e_\alpha + \dots$$ (1.14)
where $R(\alpha)$ is the distance function. The components of the electric and magnetic fields $E$ and $H$ are derived in Section 2 and Section 3, following the methodology established in \cite{14}.

2. Potential Representation

The potential $\Pi(\rho, \phi, z)$ is defined as:
$$\Pi(\rho, \phi, z) = \frac{iP}{8\pi} \cos \frac{\pi}{n} \int_C \left[ e(\phi/2 - \phi_0), e_j \right] H_0^{(1)}(kR) \, d\alpha$$
$$\Pi(\rho, \phi, z) = \frac{iP}{8\pi} \sin \frac{\pi}{n} \int_C \left[ e(\phi/2 - \phi_0) H_0^{(1)}(kR) + \dots \right]$$ (2.1)
For the case of a wedge, the potential $\Pi(\rho, \phi, z)$ is given by:
$$\Pi(\rho, \phi, z) = \frac{iP}{8\pi} \sin \frac{\pi}{n} \left[ e(\phi/2 - \phi_0) H_0^{(1)}(kR(\phi - \phi_0)) + \dots \right]$$ (2.2)
Using the results from \cite{8, 9}, we can express the integral in terms of the function:
$$H_0^{(1)}(z) = \frac{2}{i\pi} \int_0^\infty e^{iz \cosh t} \, dt$$ (2.3)
The function $\Pi(\rho, \delta)$ is then represented as:
$$\Pi(\rho, \delta) = \sum_{k} \frac{1}{2k+1} G_{1/2}(k, R(\phi + \delta)) \frac{\cos \frac{\pi}{n}}{\cosh t + \cos(\phi + \delta)}$$
$$\int_0^\infty \frac{G_{1/2}(k, R(\pi + it)) (\cosh t - 1) \cosh \frac{t}{n} \, dt}{\cosh t + \cos(\phi + \delta)}$$ (2.4)
According to (1.14), the potential $\Pi(\rho, \delta)$ can be decomposed as:
$$\Pi(\rho, \delta) = \Gamma(\rho, \delta) + \Delta(\rho, \delta)$$ (2.5)
where the geometric part $\Gamma(\rho, \delta)$ is:
$$\Gamma(\rho, \delta) = \sum G_{1/2}(k, R(\alpha)) e_{\alpha+\phi} \cot \frac{\alpha + \phi + \delta}{2n}$$ (2.6)
and the diffraction part $\Pi_d(\rho, \delta)$ is:
$$\Pi_d(\rho, \delta) = -\frac{1}{4\pi n} \int_{-\pi+i\infty}^{\pi+i\infty} G_{1/2}(k, R(\alpha)) e_{\alpha+\phi+\delta} \cot \frac{\alpha + \phi + \delta}{2n} \, d\alpha$$ (2.7)
As shown by A. A. Tuzhilin in (2.6), the poles of the cotangent occur at $\alpha = -\phi - \delta + 4\pi n$. For the range $-\pi < \text{Re } \alpha < \pi$, this condition is satisfied only for specific values of $n$.

4. Analysis of $G_{1/2}(k, R)$

The function $G_{1/2}(k, R(\phi + \delta))$ is periodic. Since $R(-\phi - \delta + 4\pi n) = R(\phi + \delta)$, we have:
$$\Gamma(\rho, \delta) = \sum G_{1/2}(k, R(\phi + \delta)) e_{\alpha}$$ (2.8)
The diffraction term $\Delta(\rho, \delta)$ is evaluated along the contours $\alpha = \pm \pi + it$. Using the symmetry $R(\pm \pi \pm it) = R(\pi + it)$, we obtain:
$$\Delta(\rho, \delta) = -\frac{1}{4\pi n} \int_0^\infty G_{1/2}(k, R(\pi + it)) \left[ \dots \right] \, dt$$ (2.9)
where the bracketed term involves the cotangent functions evaluated at the shifted arguments. This leads to the expression:
$$\frac{\sin \frac{\pi}{n} \cosh \frac{t}{n}}{\cosh t + \cos(\phi + \delta)}$$
which matches the form in (2.4). For $m=0$ and $\nu > 0$, the integral (2.10) can be evaluated. As $a \to \infty$, the limit of the function $f(a)$ is zero. For $m > 0$ and $\cos a = -1$, we use the identity:
$$\int_0^\infty (\cosh t - 1)^m \cosh \frac{t}{n} \, dt = (1 + \cos a) \frac{2^{3/2m}}{\Gamma(m + 1/2)} \dots$$
The final evaluation of the integral $J(m)$ yields:
$$J(m) = \frac{(-1)^{m-1} b^{m-1} \Gamma(l + 1/2) \Gamma(m + 1/2)}{\Gamma(1/2) J(0)}$$ (2.11)
For $m=0$, the expression simplifies. The function $f(a)$ is then:
$$f(a) = \frac{(-1)^{m-1} 2^{m-1}}{\sqrt{\pi}} \int \dots + \dots$$ (2.10)
This allows for the determination of the field behavior near the edge.

For $\nu > 0$, the function $G_\nu(k, R(\pi + it))$ can be expanded. For $m=0$:
$$\int_0^\infty G_\nu(k, R(\pi + it)) (\cosh t - 1)^m \cosh \frac{t}{n} \, dt$$ (2.12)
where $G_\nu$ is defined by (1.13) and $R_0 = R(\pi)$. As shown by A. A. Tuzhilin, (2.12) holds for $\text{Im } k \ge 0$. For $k \neq 0$, the integral converges. If $\text{Im } k > 0$ and $m=0$, we use the substitution $k = i\kappa$, leading to:
$$G_\nu(i\kappa, R(\pi + it)) = \exp\left[-i\frac{\pi}{2}(\nu+1)\right] K_\nu(\kappa R)$$
Using the properties of the modified Bessel function $K_\nu(z)$ from \cite{15}, valid for $|\arg z| < \pi/4$:
$$K_\nu(z) = \sqrt{\frac{\pi}{2z}} e^{-z} \dots$$
The integral representation for $G_\nu$ becomes:
$$\int_0^\infty \exp\left[ -\kappa \sqrt{R^2(\pi) + t^2} \right] (\cosh t - 1)^m \cosh \frac{t}{n} \, dt$$
Substituting the asymptotic forms, we find the behavior for $\cos \frac{a}{2} > 0$ and $\cos \frac{a}{2} < 0$. The potential near the edge is dominated by terms of the form:
$$\frac{P \rho_0 \cos(a/2)}{\kappa R(a)^{v+3/2}} K_{v+1/2}(\kappa R(a))$$
By setting $\kappa = -ik$, we recover the Hankel function representation $H_\nu^{(1)}(kR)$ as seen in (2.3) and (2.12).

For the specific case $v=1/2$ and $\mu=0, 1$, the potential $\Pi$ simplifies to:
$$\Pi = \frac{e^{ikR(\phi+\delta)}}{R(\phi+\delta)} + \dots$$ (2.13)
This result is consistent with the primary field (1.12) and the boundary conditions (2.1), (2.2). Section 3 details the electric and magnetic field components $E$ and $H$. We define the position vectors:
$$r = \rho(\cos \phi i + \sin \phi j) + z k$$ (3.1)
$$r_0(\alpha) = \rho_0(\cos \alpha i + \sin \alpha j) + z_0 k$$ (3.2)
where $i, j, k$ are the Cartesian unit vectors. The distance function is $R(\alpha - \phi) = |r - r_0(\alpha)|$ as per (1.8) and (3.3).

Submission history

Representation of electromagnetic fields generated by dipoles in the presence of a perfectly conducting half-plane via Macdonald integrals