TRANSACTIONS IEEE ANTENNAS ONAND PROPAGATION,

VOL. AP-30, NO.

4,JULY 1982

753

format. The coefficients of the Chebyshev polynomial were selected such that the sidelobes for each armby itself are -40 dB below the main lobe. Fig. 2(a) shows, as a reference,the pattern for a 40element linear array (a! = OD) with L = 2h/3 and a -40 dB sidelobes leveL Fig. 2(b) corresponds to a T shape array where the Xaxis arm has 20 elements with L = 2h and a complete Chebyshev loading The two arms on the 2 axis consists each of 20 elements with L = X. Both arms together form the elements of a Chebyshev array. In Fig. 2(c) each arm on the Z a i by xs itself has a complete Chebyshev loading. The main change from Fig. 2(b) is the additional high sidelobe at 0 = +7r/2. In Fig. 2(d) the antenna is similar to the one in Fig. 2(c) except the X-axis arm has ahalf Chebyshev loading. This results inthe gradual increase of the sidelobes nearthe main lobe (Le., along 0 = 0 and 0 = n). This is due to the fact that theX-axis arm is not optimallyloaded. Fig. 2(e) and 2(f) correspond to a Y shaped array with 20 elements on each arm and L = X. In Fig. 2(e) each arm has a complete Chebyshev loading. In Fig. 2(f) each arm has only a half-Chebyshev loading. This results in an increase ofthe sidelobes levels because of the nonoptimum loading. In Fig. 2(e) except for the main lobes, all the small sidelobes are at a level of -40 dB. In summary, in both the Y and T shaped arrays the main lobes are defined by the shape of the array, and the level of the sidelobes is mainly dependent on the Chebyshev loading (Le., complete orpartial). REFERENCES
[I]
Collin and Zucker, Antenna Theory, Part 1. New York: McGrawHill, New York, 1969, ch. 5 . [2] C. H. Papas, Theory o Electromagnetic Wave Propagation. f New York: McGraw-Hill, 1965, ch. 3.

expressions in the scattering problems [ 11-[ 51. The reaction in an anisotropic dispersive medium has been introduced into theharmonic field by using thecomplementarymedium whose permittivity tensor is defined by transposing the original one [ 2 ] , [ 3 ] . In this communication, the reaction of a radiation Source on field detectors is treated as a reaction field, which is expressed as a function of the time and the position of the radiation source. In other words, the reaction field is introduced as a measure of the amount of the reactions of a point radiation source on spatially distributed field detectors. It is found that the reaction field in a medium is subject to ,the equations adjoint to those of electromagnetic field. Here theadjoint means not only thetransposition of original permittivity tensor but also the space-time inversiona (r + -r, t + - t ) of original field equations. reciprocity A theorem having a clear-cut physical meaning is derived without using the complementary medium. The reciprocity theorem gives us a systematic way to derive the Huygenss principle in an anisotropic dissipative medium. Some variational formulations of the field equations are obtained with the use of the adjointfield.
11. REACTION FIELD IN DISPERSIVE MEDIA

Let us considerelectromagnetic field in ahomogeneous despersive medium. The Maxwell equationsofthe field are assumed to be

V X E(r,

t ) = - -PO Wr, 0 at

(1)

V X H(r, t ) = - D(r, t ) -I- J(r, t).

at

where the electric displacement D(r, t ) is defined by

D,(P, t ) = dr

dt

On the Concept of Reaction Field for Electromagnetic n Fields i Dispersive Media

YASAHIRO AGU
Absrruct-The concept of reaction field is introduced for electromagnetic fields in dispersive media as a measure of the amount of the reactions of a radiation source on spatially distributed field detectors. It is found that theequations of the reaction field in the medium take the adjoint forms to those of the electromagnetic field. A reciprocity theorem having a clear-cut physical meaning is derived with the use of the concept of the reaction field. Some variational expressions are derived with the use of the reaction field.
f f

f a a ( t t, r - r)E&, -

t).

(3)

Here E ( t - t, r - r) is a permittivity tensor of temporally @. and spatially dispersive medium, and the subscripts a, 0 run over x, y , z and the repeated indices are assumed to be summed. The electric field Wr, t ) at point r at time t which is produced by the current source J(r, t) at point r at time t is representedwith the use of the dyadic Greens function Gap ; E,(r, t ) = dr

dtG,p(r-I,

t - t)Jp(r, t).

( 4)

I. INTRODUCTION

of reaction in vacuum electromagnetic The concept theory has been introduced as a physical observable representing the amount of interaction between a field source and some other field source [ 11. The concept has been used to simplify the boundary value problems or to obtain variational
Manuscript received January 16, 1979; revised May 22, 1979, June 2,1981,andSeptember 22,1981. The author i with the Department of Electronics, Faculty of Engis neering, Ibaraki University, Hitachi, 360, Ibaraki, Japan.

Taking intoaccountthefactthatthereactionhas been introduced into the vacuum electromagnetic theory as a measure of the interactions between a field source and some other field source [ 11, we define reaction field E,*(r, t ) in the dispersive medium by the equation E,+(r, t ) = dr

I[

dtSp(r, t)Gga(r - r, t - t).

(5)

Here Sp(r, t) means the amount of the interactions at time t of the detector source located at r with the 0-componentof the electric field on that point. The reactionE,+(r, t ) defined

001 8-926X/82/0700-0753$00.75 0 1982 IEEE

..

_ .

754

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. A-30, NO. 4, JULY 1982

above represents thewhole amount of interactions in the semiinfinite time interval [t, a], a unit point current radiation of source along the unit vector & located at r, with the detector sources at some other points. The total amount of the reactions of the detector sources with the current radiation source J(r, t) is expressed as the superposition of the reaction field weighted by J(r, t ) ;

W =b r

I [

The electromagnetic field in the homogenousnondispersive medium given by the retarded solution of the field equations expands with a radial wave velocity as the spherical outgoing wave centered at its radiation source. Whereas the wave of the reaction field (1 3) contracts with theradial wave velocity such that it just converges on the detector point at the time of observation. The reaction field (1 0) and (1 1) should be solved in the backward time direction undersuitable f i n a l conditions. 111. RECIPROCITY THEOREM We consider the electromagnetic field in a temporary dispersive medium, Using theFouriertransforms of ( l ) , (21, (10), and (11) and following the usual procedure, we have the reciprocity theorem 161

dtE+(r, t ) J(r, t ) .

(6)

If the detector source S(i,t) represents a point charge distribution of the form e6(rrr- rs) X 6(t - T,) the Ep+(r, t ) means the 0-component of the force on the charge at time t, due to the electric field produced by the unit point current radiation source located at r at time t . In the othercase where the detector source S is composed of a current distribution then the E+(r, t ) represents the total work done on thedetector sourceby the point radiation source located at r at time t. Now we derive the equations of the reaction field E+(r, t ) . Eliminating H(r, t) from (1) and ( 2 ) , we have a wave equation for E(r, t ) :
X7 X

I,

dzn @ X fi+*+ E+ X

H> =

dr(S*

E-

E+*),
(14)

V X E(r, t) +?y,D(r,
at

a2

t)

+ go - J(r, t) = 0. at

(7)

Comparing the definition (5) of E,+(r, t) with (4), we find that the equation of E,+ is obtained by the space-time inversions r + - r, t --t of (7) foliowed by the transposition E,@ - epa of the permittivity tensor +
--f

a2 a V X V X E+@, t) + 7 D+(r, I) - p o - S(r, r) = 0,

at

at

where E (or means the Fourier component of E (or H), the asterisk means the complexconjugate of its attached quantity, o is the anguler frequency, Z denotes the surface of a given volume V and n is an outward unit vector normal to Z. A point worthy of note here is that in the reciprocity theorem already derived, the electromagnetic field in another dispersive medium having transposed the permittivity tensor (E@,) is introduced as a mathematical tooL Whereas, in the above reciprocity theorem, the physical observable called reaction field is used, which is definedin the same medium as the quantity representing the contribution of radiation source to detector source. An application of (18) isgiven as follows. Assuming the point detector source whose Fouriers component is specified by

where D+(r, t ) is defined by

m

S(r, w>= 6(r - r)b

(15)

for in (14), we have after its inverse Fourier transform,

dtepor(t - 1, r - rpp+(rr, t).

D,+(r, t) = \dr

Introducing themagnetic reaction @(r, t ) through

VXE+=--

a
+

at

P O P>

d ~ [ X (n

H(r, t - 7)) E+(r, - 7 )

(16)

into (8), we obtain

+ (E(r, t - T) X n)

H+(r, - T)]

VXH+=-D+-S.
at

where n is the inward unit vector defined by n = - . The n above formula represents the Hyugenss principle in anisotropic dissipative media. The first term on the right side represents the direct contributions of the radiation sources to the detector source and the second term the contributions from the electric surface current sheetsnX H and the magnetic current sheets E x n which are specified overthe surface Z.

It is shown that the magnetic reaction @(r, t ) introduced above is interpreted as the amount of the interactions of the detector source S ( i , t) with the electric field E(r, t) excited by the pointmagnetic current source located atr at time t. In order to get the physical picture of the reactionAfield, let assume the point observation along unit vector p specified by

IV. VARIATIONAL EXPRESSIONS

First we consider the quantityL ( l ) defined by ([ 71)

S(r, t) = $(r - r,p(t - ts),

(12)

thenthe reaction field E,+(r, t ) is reduced to the dyadic Greens function itself:

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. AP-30, NO. 4, JULY 1982

755

The first variation of L ( l ) for the small variations in the field E is given by .6EL()=/drL; d t [ S + Q X H * - - -aD * t

where E,+ is the adjointfield to E,

*6E

(18)

k ,

= \dr

dtjp(f, t)Gpa(r - r, t - t ) .

(23)

where the partial integration and the changes of the order of the integration with respect to t and t, and r and r were done under usual fixed boundary conditions. The stationary condition = 0 gives rise to ( 1 1). Equation (10) is also obtained from the stationary property of the first variation of L() withrespect to H. Similarly the small variations of E+ and H in the quantity L ( l ) bring about the electromagnetic field equations (1) and (2). These facts show thatthe quantity L ( l ) plays the role of the Lagrangian for the electromagnetic field equations ( 1) and (2) and for the reaction field equations (1 0) and ( 1 1). The value of L ( ) under the stationary condition becomes

The quantityL(3) takes thefamiliar form knownas variational expression. The maximal condition ?f L(3) gives rise to the ( 4 ) and (5), under the condition that and E, are specified. The equality sign of (22) hplds when J, coincides with the correct curfent distributionJ , corresponding to the prescribed fields E,, E ,. The stationary value of L(3) given by the right side of the inequality (22) represents the total amount of the dissipated field energy.

J!?,

REFERENCES
v . H. Rumsey,Reactionconcept in electromagnetictheory, P h w R e v . , vol. 94, no. 6, pp. 1483-1491 vol. 95, p. 1705,1954. R. F. Haraington and A. T. Vileneuve, ..Rec,iprocity relationships for gyrotropic media, IRE Trans. Microwave Theory Tech, vol. MTT-6, no. 3, p. 308, 1958. J. A . Kong, Theory of Electromagnetic Waves. New York: Wiley, 1975, p. 269, p. 215. R. F. Harrington, Time-HarmonicElectromagneticFields. New York: McGraw-Hill, 1961, p. 340. J. A. Kong, Theorems of bianisotropic media, proc. VO~. 60, no. 9, p. 1036,1972.
I

LSJ1) f \r d

dtS

E E dr

[l

dtJ

E+,

(19)

which represents the total amount of the interactions of the radiation source with the detector source S which is given by (6). The second equality of (1 9) i shown with the use of (I), s (2), (lo), and ( 1 1). We note that the Lagrangian (1 7) may be modified into the normalized form
r

/dr

I-

dt[S.E+J*E+]

Now we consider electromagnetic field in dissipative medium. The energy loss Q of the field in dissipative medium is expressed as

Q=

\m -m

dt

161 M. Agu, On the concept of reaction for electromagnetic fields in dispersive media, Summuries of Papers, 1978 Int. Symp. Antenna and Propagation, c-2-2, Japan, pp. 367-370. 171 G. I. Marchuk and V. V. Orlov, The theory of adjoint functions, Neutron Phy. Soviet, pp. 24-34, 1961.

-m

dtfdr/drJ,(r, t )

The Equivalent Radiusof a Narrow Conducting strip

which is reasonably assumed to have positive difinite proferty toranycurrent density 7,. Letthecurrentdensiq J , be J, qJ, where 17 is an arbitrary real number ad, n! is the current density corresponding to the prescribed field E&, t ) :

CHALMERS hl. BUTLER, SENIOR MEMBER, IEEE

Absrrucr-By means of a simple and direct analytical procedure, is it confiimed that the equivalent radins of a narrow conducting strip is one-fourth its width. The analysis proceeds from the integral equation for the surface current density induced on a conducting cylinder of general cross section and its incorporates knowledge of the current density induced on a narrow strip. Explicit expressions are given for the total axial current induced on the narrow strip and for its echo width.

INTRODUCTION The purpose of this communication is to present a simple analytical procedure for determining the equivalent radius of a narrow strip which is based upon the dynamic solution [ 1 ]
Manuscript received April 6,1981; revised June 9,1981. The author is with the Department of Electrical Engineeiing, University of Mississippi, University, MS 38677.

001 8-926X/82/070C-0755%00.75 0 1982 IEEE