1. Introduction
A self dual chiral medium is invariant under the duality transformations [1, 2]:
(1)
in which
E,
B,
D, and
H are the components of the electromagnetic field and
ε ·
μ ·
ξ constant permittivity, permeability, and chirality parameters. A simple illustration is supplied by the Post constitutive relations [
3] used in this work,
ε ·
μ are real and
ξ pure imaginary (
ξ2 < 0):
(2)
We analyze the behaviour of electromagnetic fields in this self-dual medium using successively the Gibbs conventional and the differential-form formulations of Maxwell’s equations. These last ones, in a isotropic homogeneous medium, have TE and TM waves as particular solutions when the fields do not depend on one coordinate
x,
y,
z. We prove here the existence of pseudo-TE/TM waves, pseudo meaning that only one of their components is null while two relations depending on
ε, μ, ξ exist between four of the five other components. An example is supplied by quasi-undistorted free fields of the Courant-Hilbert type [
4].
2. Gibbs Conventional Formulation
Taking (2) into account, the Maxwell equations with ∂τ = 1/c∂t
(3)
have the explicit form when
∂zE = 0,
∂zH = 0:
(5a)
(5b)
(5c)
(5d)
We are interested in the solutions of these equations when
Ez or
Hz = 0.
2.1. Pseudo-TE Waves (Ez = 0)
Let us write (4c)
(6)
Substituting (
5c) into the third term of (
6) with
Ez = 0 this equation becomes
(7)
while according to (
4a), (
4b),
Ez = 0 implies
(8)
so that (
4d) is satisfied. Now writing (
5a), (
5b)
,
, and substituting these expressions into (
7) gives the wave equation satisfied by
Hz (
ξ2 < 0):
(9)
Finally, taking into account (
8), (
5a) and (
5b) determine
Ex,y in terms of
Hz:
(10)
so that
∂xEx +
∂yEy = 0 which implies (
5d) according to (
8).
2.2. Pseudo-TM Waves (Hz = 0)
Proceeding similarly, we write (5c)
(11)
Substituting (
4c) into the third term of (
11) gives with
Hz = 0
(12)
while according to (
5a) and (
5b),
Hz = 0 implies in agreement with (
5d) that
(13)
Now, substituting (
4a) and (
4b) into (
12) written as
,
gives the wave equation satisfied by
Ez:
(14)
while according to (
13), (
4a) and (
4b) supplying
Hx,y in terms of
Ez become
(15)
so that
∂xHx +
∂yHy = 0 imply (4d).
2.3. Quasi Distortion Free Solutions
Harmonic plane waves are the most elementary solutions of the wave equations (9) and (14) but the Courant-Hilbert quasi-undistorted progressing waves [4] are of some interest. The form of these waves, defined on characteristic surfaces (wave fronts), does not change along their propagation but their amplitude is not constant. To prove that (9) and (14) have such solutions, we introduce the variables
(16)
and, these wave equations become
(17)
with quasi-undistorted solutions [
5,
6]:
(18)
and, it is easily noticed that the wave front
(19)
is a solution of the characteristic partial differential equation (
ξ2 <
0):
(20)
Substituting (
18) into (
16) gives {
Ex,
Ey} in terms of
Hz from which are obtained {
Hx,
Hy} according to (
8). A similar result follows from (
18) substituted into (
15), just exchanging {
μ,
ξ} into {
ε, −
ξ) (see (
13)) and the roles of
Ex,y,
Hx,y in the previous statement.
Now according to (16) ∂y = ∂u − ∂v so that (10) with , , and Hz = Ψ gives
(21)
Similar expressions are obtained from (
15) for the components
Hx,y of the pseudo-TM waves. The roles of
x,
y may be changed giving with
,
the quasi-undistorted waves:
(22)
The fields (
18) and (
22) represent 1D-modulated Gaussian beams [
7].
3. Differential—Form Formulation
The three-dimensional differential-form formulation of Maxwell’s equations is [1], in absence of charge and current, with the exterior derivative operator d = dx∂x + dy∂y + dz∂z and τ = ct:
(23)
In these equations
,
are the 1-forms:
(24b)
Now, let
*h be the Hodge star operator, then
(25)
from which we get the permittivity, permeability, and chirality operators:
(26a)
so that in a self-dual chiral medium, the 2-forms
,
become
(27)
and the coefficients of the differentials in (
24b) are
(28a)
Substituting (
24a) and (
24b) into (
23), this set of Maxwell’s equations becomes
(29a)
(29b)
Finally, a simple calculation gives for the second set (
23) of Maxwell’s equations:
3.1. Wave Equations for Fields
We get according to (23) and (27)
(31)
and, eliminating
from (
31) supplies the differential-form formulation of the wave equation
We for the E-field. A simple calculation gives after multiplication by
(32)
The first two terms of this expression become using the Hodge star operators
ε, ξ and its inverse
μ−1:
(33)
In the last term of (
32), we have, still using the inverse Hodge star operator
μ−1,
(34)
so that Δ and ∇. are the Laplacian and the divergence operators, and we get
(35)
Then, substituting (
33) and (
35), multiplied by
μ into (
32) gives with
n2 =
εμ
(36)
and, substituting (
29a) into (
36), we get finally
(37)
Similarly, eliminating
from (
31) gives the wave equation
with
These equations will be used by imposing that one of the three terms in (37) and (38) is null, that is, one component of the E and H fields is solution of a wave equation, the other two components being obtained from Maxwell’s equations.
3.2. Pseudo-TE/TM Fields
To get the pseudo-TM fields we suppose that the coefficient of the dx⋀dy term in (37) is null which gives
(39)
In addition, these fields must satisfy the conditions
(40)
and then, (
39) reduces to the wave equation (
14) satisfied by
Ez. To get the other components, we further assume
Dx =
Dy = 0 so that
(41)
Then, the first two terms of the Maxwell equations (
29b) are null and, we are left with
(42a)
while we get from (
29a), still taking into accont (
40),(
41)
(42b)
The set (
42a), (
42b) supplies two equations to determine the components
Hx,
Hy in terms of the solutions
Ez of the wave equation (
14). For the pseudo-TE fields, making null the coefficient of
dx⋀
dy term in (
38) null and imposing (
40) with
Ez = 0 instead of
Hz = 0, supply the wave equation (
9) satisfied by
Hz. We further assume
Bx,y = 0 so that
(43)
Then the first two terms of (
29a) are null and one is left with
(44)
while we get from (
29b)
(45)
Thus, we obtain a set of two equations (
44), (
45) to determine
Ex,y once known the solution
Hz of the wave equation (
9).
The integration of these 2-forms has to be performed on 2D-manifolds with a suitable numerical technique such as finite elements [7] using a judicious choice of test functions among which the Whitney forms [8] have a particular interest. Many works have been devoted to this integration problem [8–10] where further references can be found.
4. Discussion
Two topics emerge from this work. The existence of pseudo-TE/TM waves in self media allows to make a comparison between the conventional and the differential-form formulations of Maxwell’s equations. Let us limit this discussion to TE waves (similar conclusions hold valid for pseudo-TM waves). In both formalisms, we start with the component Hz satisfying in some domain of R3 the wave equation (9) with proper boundary conditions. Then, since, according to (8), Hx,y are obtained at once from Ex,y, we are left to get these last two components in terms of Hz.
In the conventional formalism, this requirement leads, according to (21), to perform the numerical integration of the expressions:
(46)
which is a rather ordinary business.
On the other hand, in the differential-form formalism,we have to cope with the differential equations (44), (45) whose integration on a 2D-manifold M requires an important work as just mentioned at the end of Section 3. So, an interesting question is to investigate when a formalism outpaces the other one, taking into account accuracy and computation time.
The second topics concern the existence in self-dual media of quasi-undistorted progressing waves. These fields carry on an infinite energy but, using the finite aperture approximation for diffraction, we may obtain such fields with a finite energy, able, in principle, to be launched in the physical space [11]. Then, they could be used in communications [12] and to generate beams of directed energy (electromagnetic bullets) of interest in laser and radar technologies [13].