1: % Converted from Microsoft Word to LaTeX

2: % by Chikrii Softlab Word2TeX converter (version 3.0)

3: % Copyright (C) 1999-2003 Chikrii Softlab. All rights reserved.

4: % http://www.chikrii.com

5: % mailto: info@chikrii.com

6: % License: CSL#0022AB

7: 

8: \documentclass{article}

9: \usepackage{latexsym}

10: \usepackage{graphicx}

11: \setlength{\oddsidemargin}{0cm}

12: \setlength{\evensidemargin}{0cm}

13: %\linespread{1.6}

14: \textwidth 154truemm    

15: \textheight 232truemm   

16: 

17: \begin{document}

18: 

19: \begin{center}

20: \textbf{\LARGE On the Solution of Maxwell's First \\ \vspace{.3cm} Order Equations}

21: \end{center}

22: \vspace{.1cm}

23: \begin{large}

24: 

25: \begin{center}

26: W. Engelhardt\footnote{Home address: Fasaneriestrasse 

27: 8, D-80636 M\"{u}nchen, Germany\par Electronic address: 

28: wolfgangw.engelhardt@t-online.de}, retired from:

29: \end{center}

30: 

31: \begin{center}

32: Max-Planck-Institut f\"{u}r Plasmaphysik, D-85741 Garching, Germany

33: \end{center}

34: 

35: \vspace{.6cm}

36: 

37: \noindent \textbf{Abstract }

38: 

39: \noindent In an attempt to solve Maxwell's first order system of equations, starting 

40: from a given initial state, it is found that a consistent solution depending 

41: on the temporal evolution of the sources cannot be calculated. The well 

42: known retarded solutions of the second order equations, which are based on 

43: the introduction of potentials, turn out to be in disagreement with a direct 

44: solution of the first order system.

45: \vspace{.3cm} \\

46: \noindent PACS number: 03.50.De

47: \vspace{.6cm}

48: 

49: \noindent \textbf{1. Introduction}

50: 

51: \noindent In recent papers [1, 2] it was shown that Maxwell's equations have different 

52: formal solutions depending on the chosen gauge. In [2] it was argued that 

53: the formalism of gauge invariance is based on the tacit assumption of 

54: Maxwell's equations having unique solutions which appeared, however, not to 

55: be guaranteed \textit{a priori. }In response to the publication of [2] it was pointed out in 

56: private communications [3] that uniqueness is a necessary consequence of the 

57: linear structure of the equations. These arguments are valid. If one finds, 

58: nevertheless, different solutions in Lorenz and in Coulomb gauge, it seems 

59: to indicate that a solution does not exist at all. Indeed, it was shown in 

60: [2] that the Li\'{e}nard-Wiechert fields based on the Lorenz gauge do not 

61: satisfy the equations in the source region, unless one postulates a velocity 

62: dependent ``deformation'' of point charges as in [1]. Furthermore, the 

63: formal solution for the vector potential in Coulomb gauge led to an 

64: undefined conditionally convergent integral which would even diverge upon 

65: differentiation. 

66: 

67: The reason for the difficulties encountered could have to do with the 

68: assumption of point sources which were exclusively considered in [2]. 

69: Therefore, it appears worthwhile to investigate the problem further, 

70: assuming smooth charge and current distributions as originally considered by 

71: Maxwell. In order to avoid any ambiguities arising from the introduction of 

72: potentials, it seems advisable to analyse directly the solvability of the 

73: first order system of Maxwell's equations (Sect. 2.). It turns out that the 

74: coupled first order system contains certain inconsistencies which prevent 

75: its solution when calculated by a numerical forward method proceeding in 

76: time.

77: 

78: The usual method of solution derives inhomogeneous wave equations from the 

79: first order system, and expresses the solutions as retarded integrals by 

80: application of Duhamel's principle. In [2] it was argued that this method is 

81: not plausible, since the wave equations obtained by differentiating the 

82: first order system connect the travelling fields with the stationary sources 

83: at the same time, while in the retarded solutions the differentiation of the 

84: sources is inconsistently dated back to an earlier time. In Sect. 3. we 

85: analyze the retarded solutions for smooth source distributions and find that 

86: these solutions do not satisfy the first order system. This is demonstrated 

87: in Sect. 4. by considering a specific example.

88: \vspace{.6cm}

89: 

90: \noindent \textbf{2. The first order equations}

91: 

92: \noindent In \textit{vacuo} the first order system as devised by Hertz on the basis of Maxwell's 

93: equations is supposed to describe the electromagnetic field:

94: \begin{equation}

95: \label{eq1}

96: div\,\vec {E}_g =4\pi \,\rho 

97: \end{equation}

98: \begin{equation}

99: \label{eq2}

100: rot\,\vec {E}_r =-\frac{1}{c}\frac{\partial \vec {B}}{\partial t}

101: \end{equation}

102: \begin{equation}

103: \label{eq3}

104: div\,\vec {B}=0

105: \end{equation}

106: \begin{equation}

107: \label{eq4}

108: rot\,\vec {B}=\frac{4\pi }{c}\,\vec {j}+\frac{1}{c}\frac{\partial \left( 

109: {\vec {E}_g +\vec {E}_r } \right)}{\partial t}

110: \end{equation}

111: Here we have indicated that the electric field has two contributions of 

112: different structure. In (\ref{eq1}) only the irrotational part enters, whereas (\ref{eq2}) 

113: contains exclusively the rotational part of the field. Both parts enter 

114: equation (\ref{eq4}). One may separate out the instantaneous contribution of the 

115: magnetic field and write (\ref{eq4}) as two equations:

116: \begin{equation}

117: \label{eq5}

118: rot\,\vec {B}_0 =\frac{4\pi }{c}\,\vec {j}+\frac{1}{c}\frac{\partial \vec 

119: {E}_g }{\partial t}

120: \end{equation}

121: \begin{equation}

122: \label{eq6}

123: rot\,\vec {B}_1 =\frac{1}{c}\frac{\partial \vec {E}_r }{\partial t}

124: \end{equation}

125: The quasi-static solutions of (\ref{eq1}) and (\ref{eq5}) -- subject to the boundary 

126: condition that the fields vanish at infinity -- are represented by integrals 

127: over all space:

128: \begin{equation}

129: \label{eq7}

130: \vec {E}_g \left( {\vec {x},\,t} \right)=\int\!\!\!\int\!\!\!\int {\rho 

131: \left( {\vec {x}',\,t} \right)} \,\left( {\vec {x}-\vec {x}'} 

132: \right)\frac{d^3x'}{\left| {\vec {x}-\vec {x}'} \right|^3}

133: \end{equation}

134: \begin{equation}

135: \label{eq8}

136: \vec {B}_0 \left( {\vec {x},\,t} \right)=\frac{1}{c}\int\!\!\!\int\!\!\!\int 

137: {\left( {\vec {j}\left( {\vec {x}',\,t} \right)+\frac{1}{4\pi 

138: }\frac{\partial \vec {E}_g \left( {\vec {x}',\,t} \right)}{\partial t}} 

139: \right)} \times \left( {\vec {x}-\vec {x}'} \right)\,\frac{d^3x'}{\left| 

140: {\vec {x}-\vec {x}'} \right|^3}

141: \end{equation}

142: It remains then to determine the rotational part of the electric field and 

143: the contribution $\vec {B}_1 $. 

144: 

145: Applying a numerical forward method one obtains from (\ref{eq6}) the difference 

146: equation:

147: \begin{equation}

148: \label{eq9}

149: \vec {E}_r \left( {\Delta t} \right)=\vec {E}_r \left( 0 \right)+\Delta 

150: t\,c\,rot\,\vec {B}_1 \left( 0 \right)

151: \end{equation}

152: and from (\ref{eq2}):

153: \begin{equation}

154: \label{eq10}

155: \vec {B}\left( {\Delta t} \right)=\vec {B}\left( 0 \right)-\Delta 

156: t\,c\,rot\,\vec {E}_r \left( 0 \right)

157: \end{equation}

158: Assuming that the sources were constant for $t\le 0$ one has the initial 

159: conditions:

160: \begin{equation}

161: \label{eq11}

162: \vec {E}_r \left( 0 \right)=\vec {B}_1 \left( 0 \right)=0

163: \end{equation}

164: Substituting this into (\ref{eq9}) and (\ref{eq10}) one finds the curious result that 

165: neither $\vec {E}_r $ nor the total magnetic field $\vec {B}$ proceed after 

166: the first time step, and this will remain so forever, at least in the vacuum 

167: region outside the sources. If the current would linearly rise to a new 

168: stationary level, e.g., equation (\ref{eq10}) would predict that $\vec {B}$ stays 

169: constant at its initial value, in contrast to (\ref{eq8}) which predicts that $\vec 

170: {B}_0 $ rises simultaneously with the current and reaches a new stationary 

171: value as well. 

172: 

173: One may also split (\ref{eq2}) into two equations:

174: \begin{equation}

175: \label{eq12}

176: rot\,\vec {E}_{r0} =-\frac{1}{c}\frac{\partial \vec {B}_0 }{\partial t}

177: \end{equation}

178: \begin{equation}

179: \label{eq13}

180: rot\,\vec {E}_{r1} =-\frac{1}{c}\frac{\partial \vec {B}_1 }{\partial t}

181: \end{equation}

182: The quasi-static solution of (\ref{eq12}) is:

183: \begin{equation}

184: \label{eq14}

185: \vec {E}_{r0} =-\frac{1}{4\pi \,c}\int\!\!\!\int\!\!\!\int {\frac{\partial 

186: \vec {B}_0 }{\partial t}\times \frac{\vec {x}-\vec {x}'}{\left| {\vec 

187: {x}-\vec {x}'} \right|^3}} \,d^3x'

188: \end{equation}

189: and from (\ref{eq13}) follows:

190: \begin{equation}

191: \label{eq15}

192: \vec {B}_1 \left( {\Delta t} \right)=\vec {B}_1 \left( 0 \right)-\Delta 

193: t\,c\,rot\,\vec {E}_{r1} \left( 0 \right)=0

194: \end{equation}

195: If $\vec {B}_1 $ vanishes after the first time step as follows from (\ref{eq15}), 

196: and $\vec {B}$ stays also constant according to (\ref{eq10}), a clear contradiction 

197: with (\ref{eq8}) arises. Furthermore, equation (\ref{eq9}) predicts that the total 

198: rotational electric field stays constant, whereas the quasi-static part (\ref{eq14}) 

199: follows instantaneously all changes of $\vec {B}_0 \left( t \right)$ 

200: according to (\ref{eq14}).

201: 

202: We note that the quasi-static expressions (\ref{eq7}), (\ref{eq8}), (\ref{eq14}) can be seen as 

203: solutions of elliptic equations. On the other hand, one obtains from (\ref{eq2}) and 

204: (\ref{eq4}) by mutual elimination of the fields the inhomogeneous hyperbolic 

205: equations:

206: \begin{equation}

207: \label{eq16}

208: \Delta \vec {B}-\frac{1}{c^2}\frac{\partial ^2\vec {B}}{\partial 

209: t^2}=-\frac{4\pi }{c}\,rot\,\vec {j}

210: \end{equation}

211: \begin{equation}

212: \label{eq17}

213: \Delta \vec {E}_r -\frac{1}{c^2}\frac{\partial ^2\vec {E}_r }{\partial 

214: t^2}=-\frac{4\pi }{c}\frac{\partial \vec {j}}{\partial 

215: t}\,+\frac{1}{c^2}\frac{\partial ^2\vec {E}_g }{\partial t^2}\,

216: \end{equation}

217: As indicated in [2], the mixture of elliptic and hyperbolic equations 

218: inherent to Maxwell's system leads apparently to the inconsistencies which 

219: manifest themselves in the incongruities implied in (\ref{eq10}) as compared to (\ref{eq8}), 

220: and in (\ref{eq14}) as compared to (\ref{eq9}). The system (1 -- 4) does not permit a 

221: continuous temporal evolution from a given realistic initial state. In a 

222: region where the sources in (\ref{eq16}) and (\ref{eq17}) vanish the homogeneous hyperbolic 

223: equations describe correctly propagating electromagnetic fields, but their 

224: production mechanism in connection to the sources remains obscure. 

225: 

226: Since in all textbooks it is claimed that Maxwell's equations do have 

227: solutions which are uniquely determined when the behaviour of the sources is 

228: given as a function of space and time, we must discuss the usual procedure 

229: to obtain these solutions which -- according to our analysis -- cannot 

230: satisfy the first order system. 

231: \vspace{.6cm}

232: 

233: \noindent \textbf{3. The retarded solutions}

234: 

235: \noindent The normal method of solution expresses the fields by potentials:

236: \begin{equation}

237: \label{eq18}

238: \vec {B}=rot\,\vec {A}\;,\quad \vec {E}=-\nabla \phi 

239: -\frac{1}{c}\frac{\partial \vec {A}}{\partial t}

240: \end{equation}

241: which leads to inhomogeneous wave equations in Lorenz gauge:

242: \begin{equation}

243: \label{eq19}

244: \Delta \phi -\frac{1}{c^2}\frac{\partial ^2\phi }{\partial t^2}=-4\pi \,\rho 

245: \end{equation}

246: \begin{equation}

247: \label{eq20}

248: \Delta \vec {A}-\frac{1}{c^2}\frac{\partial ^2\vec {A}}{\partial 

249: t^2}=-\frac{4\pi }{c}\,\,\vec {j}

250: \end{equation}

251: They are solved by application of Duhamel's principle to yield the retarded 

252: solutions, e.g.:

253: \begin{equation}

254: \label{eq21}

255: \vec {A}\left( {\vec {x}{\kern 1pt},\,t} 

256: \right)=\frac{1}{c}\int\!\!\!\int\!\!\!\int {\vec {j}} \left( {\vec 

257: {x}'{\kern 1pt},\,t-{\left| {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} 

258: \right|} \mathord{\left/ {\vphantom {{\left| {\vec {x}{\kern 1pt}-\vec 

259: {x}'{\kern 1pt}} \right|} c}} \right. \kern-\nulldelimiterspace} c} 

260: \right)\,\frac{d^3x'}{\left| {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} 

261: \right|}\,\,

262: \end{equation}

263: 

264: Instead of introducing potentials one may solve the wave equations for the 

265: fields directly. The magnetic field, for example, can be expressed as the 

266: sum:

267: \begin{equation}

268: \label{eq22}

269: \vec {B}=\vec {B}_0 +\vec {B}_1 

270: \end{equation}

271: where $\vec {B}_0 $ is the instantaneous part (\ref{eq8}), and $\vec {B}_1 $ 

272: satisfies according to (\ref{eq16}) the equation:

273: \begin{equation}

274: \label{eq23}

275: \Delta \vec {B}_1 -\frac{1}{c^2}\frac{\partial ^2\vec {B}_1 }{\partial 

276: t^2}=\frac{1}{c^2}\frac{\partial ^2\vec {B}_0 }{\partial t^2}

277: \end{equation}

278: In analogy to (\ref{eq21}) this equation has the retarded solution: 

279: \begin{equation}

280: \label{eq24}

281: \vec {B}_1 \left( {\vec {x}{\kern 1pt},\,t} \right)=-\frac{1}{4\pi 

282: \,c^2}\int\!\!\!\int\!\!\!\int {\left( {\frac{\partial ^2\vec {B}_0 \left( 

283: {\vec {x}',\,t'} \right)}{\partial t'^2}} \right)} \,\frac{d^3x'}{\left| 

284: {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|}\;,\quad t'=t-{\left| 

285: {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|} \mathord{\left/ 

286: {\vphantom {{\left| {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|} c}} 

287: \right. \kern-\nulldelimiterspace} c

288: \end{equation}

289: Similarly, one may write:

290: \begin{equation}

291: \label{eq25}

292: \vec {E}=\vec {E}_0 +\vec {E}_1 

293: \end{equation}

294: and obtain from (\ref{eq17}) a second order differential equation for $\vec {E}_1 $:

295: \begin{equation}

296: \label{eq26}

297: \Delta \vec {E}_1 -\frac{1}{c^2}\frac{\partial ^2\vec {E}_1 }{\partial 

298: t^2}=\frac{1}{c^2}\frac{\partial ^2\vec {E}_0 }{\partial t^2}\,

299: \end{equation}

300: where $\vec {E}_0 $ is the instantaneous part of the electric field 

301: resulting from (\ref{eq5}) and (\ref{eq12}):

302: \begin{equation}

303: \label{eq27}

304: \vec {E}_0 \left( {\vec {x}{\kern 1pt},\,t} 

305: \right)=-\frac{1}{c^2}\int\!\!\!\int\!\!\!\int {\left( {\frac{\partial \vec 

306: {j}\left( {\vec {x}',\,t} \right)}{\partial t}} \right)} 

307: \,\frac{d^3x'}{\left| {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|}

308: \end{equation}

309: The retarded solution of (\ref{eq26}) is then:

310: \begin{equation}

311: \label{eq28}

312: \vec {E}_1 \left( {\vec {x}{\kern 1pt},\,t} \right)=-\frac{1}{4\pi 

313: \,c^2}\int\!\!\!\int\!\!\!\int {\left( {\frac{\partial ^2\vec {E}_0 \left( 

314: {\vec {x}',\,t'} \right)}{\partial t'^2}} \right)} \,\frac{d^3x'}{\left| 

315: {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|}\;,\quad t'=t-{\left| 

316: {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|} \mathord{\left/ 

317: {\vphantom {{\left| {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|} c}} 

318: \right. \kern-\nulldelimiterspace} c

319: \end{equation}

320: It turns out that the fields as obtained from (\ref{eq21}) and (\ref{eq18}) are not the same 

321: fields as that calculated from (\ref{eq22}), (\ref{eq8}), and (\ref{eq24}), and from (\ref{eq25}), (\ref{eq27}), and 

322: (\ref{eq28}). This will be demonstrated in the next Section by choosing a specific 

323: example. Hence, we must conclude that the retarded solutions cannot be 

324: considered as true solutions of the first order equations.

325: 

326: The reason for this failure must be sought in the inconsistency which lies 

327: in the fact that equations (\ref{eq20}), (\ref{eq23}), (\ref{eq26}) connect the sources $\vec 

328: {j}\left( {\vec {x}{\kern 1pt},t} \right)$, $\vec {B}_0 \left( {\vec 

329: {x}{\kern 1pt},t} \right)$, $\vec {E}_0 \left( {\vec {x}{\kern 1pt},t} 

330: \right)$, respectively, with the travelling wave fields $\vec {A}\left( 

331: {\vec {x}{\kern 1pt},t} \right)$, $\vec {B}_1 \left( {\vec {x}{\kern 1pt},t} 

332: \right)$, $\vec {E}_1 \left( {\vec {x}{\kern 1pt},t} \right)$ at the \textit{same }time 

333: $t$, whereas in the retarded solutions (\ref{eq21}), (\ref{eq24}), (\ref{eq28}) the differentiation 

334: of the source is dated back to the earlier time $t'=t-{\left| {\vec 

335: {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|} \mathord{\left/ {\vphantom 

336: {{\left| {\vec {x}{\kern 1pt}-\vec {x}'{\kern 1pt}} \right|} c}} \right. 

337: \kern-\nulldelimiterspace} c$. As pointed out in [2], the source may be very 

338: far away from the observation point, and may not even exist anymore when the 

339: fields $\vec {B}_1 \left( {\vec {x}{\kern 1pt},t} \right)$, $\vec {E}_1 

340: \left( {\vec {x}{\kern 1pt},t} \right)$ are measured at time $t$. It makes 

341: little sense to differentiate non-existent instantaneous fields at time $t$, 

342: but this was necessary to derive equations (\ref{eq16}), (\ref{eq17}) from the system (1 -- 

343: 4). Obviously, it constitutes a \textit{contradictio in adjecto }connecting the travelling fields predicted 

344: by (\ref{eq16}) and (\ref{eq17}) with the stationary sources in the first order system at 

345: the same time.

346: \vspace{.6cm}

347: 

348: \noindent \textbf{4. A specific example}

349: 

350: \noindent In order to facilitate the calculations we choose an example where we have 

351: $div\,\vec {j}=0$. In this case the scalar potential vanishes because of 

352: $\rho =0$ which makes Lorenz and Coulomb gauge identical: $div\,\vec {A}=0$. 

353: The chosen example is a hollow cylinder which carries a closed oscillating 

354: current driven by an rf-generator through a resistor R, as sketched in Fig. 

355: 1. It is assumed that the current was switched on at time $t=-\infty $ and 

356: oscillates with a sinusoidal time dependence: $I\,\exp \left( {-i\,\omega 

357: \,t} \right)$. The current flows in a thin central filament, and returns 

358: symmetrically on the cylindrical surface. This can be achieved to an 

359: arbitrary degree of accuracy, if the inverse wave vector $k=\omega 

360: \mathord{\left/ {\vphantom {\omega c}} \right. \kern-\nulldelimiterspace} c$ 

361: is large compared to the dimensions of the device.

362: 

363: \begin{figure}[htbp]

364: \centerline{\includegraphics[width=3.in,height=4.in]{fig1.eps}}

365: \label{fig1}

366: \caption{\large Oscillating current flowing in a closed circuit of cylindrical 

367: geometry}

368: \end{figure}

369: \newpage

370: \noindent The instantaneous magnetic field component (\ref{eq8}) produced in this configuration is:

371: \begin{eqnarray}

372: \label{eq29}

373: B_0 &=&\frac{2\,I}{c\,r}\exp \left( {-i{\kern 1pt}\omega {\kern 1pt}t} 

374: \right)\;,\quad r\le a\;,\quad -b\le z\le b \\ 

375:  B_0 &=&0\;,\quad r>a\;,\quad z<-b\;,\quad z>b \nonumber

376:  \end{eqnarray}

377: and the instantaneous electric field (\ref{eq27}) becomes: 

378: \begin{eqnarray}

379: \label{eq30}

380: \vec {E}_0 =\frac{i\,k\,I}{2\pi \,c}\,\exp \left( {-i{\kern 1pt}\omega 

381: {\kern 1pt}t} \right)\int\limits_0^{2\pi } {\left\{ {\int\limits_0^a {\left[ 

382: {\frac{\cos \varphi '\,dr'}{R}} \right]_{z'=-b}^{z'=+b} \,\vec {e}_r 

383: +\int\limits_{-b}^b {\left[ {\frac{dz'}{R}} \right]_{r'=a}^{r'=0} \vec {e}_z 

384: } } } \right\}} \,d\varphi ' \\ 

385:  R=\sqrt 

386: {r^2+r'^2-2\,r\,r'\cos \varphi '+\left( {z-z'} \right)^2} \nonumber 

387:  \end{eqnarray}

388: The retarded solution of the vector potential as obtained from (\ref{eq21}) is:

389: \begin{equation}

390: \label{eq31}

391: \vec {A}=\frac{I\,e^{-i{\kern 1pt}\omega {\kern 1pt}t}}{2\pi 

392: \,c}\int\limits_0^{2\pi } {\left\{ {\int\limits_0^a {\left[ {\frac{\exp 

393: \left( {i\,k\,R} \right)}{R}} \right]} _{z'=-b}^{z'=+b} \cos \varphi 

394: '\,dr'\,\vec {e}_r +\int\limits_{-b}^{+b} {\left[ {\frac{\exp \left( 

395: {i\,k\,R} \right)}{R}} \right]_{r'=a}^{r'=0} dz'\,\vec {e}_z } } \right\}} 

396: \,d\varphi '

397: \end{equation}

398: It may be substituted into (\ref{eq18}) to yield the fields as given by Jackson for 

399: a localized oscillating source [4]:

400: \begin{eqnarray}

401: \label{eq32} 

402: B\!\!&=&\!\!\frac{I\,e^{-i{\kern 1pt}\omega {\kern 1pt}t}}{2\pi 

403: \,c} \int\limits_{0}^{2\pi}\left\{\int\limits_0^a {\left[ 

404: {\frac{e^{i\,k\,R}\left( {1-i\,k\,R} \right) \,s\,\cos \varphi ' dr'}{R^3}} 

405: \right]_{s=z-b}^{s=z+b} } \right. \nonumber \\ &\quad& \quad \quad \quad 

406: \left.-\int\limits_{z-b}^{z+b} {\left[ 

407: {\frac{e^{i\,k\,R}\left( {1-i\,k\,R} \right)\,\left( {r-r'\cos \varphi '} 

408: \right)\,dz'}{R^3}} \right]_{r'=0}^{r'=a} } \right\} \,d\varphi '

409: \end{eqnarray} 

410: 

411: \begin{equation}

412: \label{eq33}

413: \vec {E}=\frac{i\,k\,I\,e^{-i{\kern 1pt}\omega {\kern 1pt}t}}{2\pi 

414: \,c}\int\limits_0^{2\pi } {\left\{ {\int\limits_0^a {\left[ 

415: {\frac{e^{i\,k\,R}\cos \varphi '\,dr'}{R}} \right]_{z'=-b}^{z'=+b} \,\vec 

416: {e}_r +\int\limits_{-b}^b {\left[ {\frac{e^{i\,k\,R}\,dz'}{R}} 

417: \right]_{r'=a}^{r'=0} \vec {e}_z } } } \right\}} \,d\varphi '

418: \end{equation}

419: where $s=z-z'$. It is doubtful whether these solutions satisfy also the 

420: differential equations (\ref{eq23}) and (\ref{eq26}). In order to check on this we consider, 

421: e.g., equation (\ref{eq23}) adapted to our case:

422: \begin{equation}

423: \label{eq34}

424: r^2\frac{\partial ^2B_1 }{\partial r^2}+r\,\frac{\partial B_1 }{\partial 

425: r}-B_1 \left( {1-r^2k^2} \right)+r^2\frac{\partial ^2B_1 }{\partial 

426: z^2}=-\frac{2\,k^2I\,r}{c}\,e^{-i\,\omega \,t}

427: \end{equation}

428: where the right-hand-side must be set to zero outside the cylinder of Fig. 

429: 1. We integrate this equation with respect to $r$ and obtain:

430: \begin{equation}

431: \label{eq35}

432: \frac{\partial B_1 }{\partial r}-\frac{B_1 }{r}+\frac{1}{r^2}\int\limits_0^r 

433: {r^2} \left( {\frac{\partial ^2B_1 }{\partial z^2}+k^2B_1 } 

434: \right)\,dr=-\frac{\,k^2I}{c}\,e^{-i\,\omega \,t}

435: \end{equation}

436: The contribution $B_1 $ may be calculated from (32) by expansion of the 

437: exponential function for $k\,R<1$. In zero order one obtains the 

438: instantaneous field (\ref{eq29}), and in second order one has:

439: \begin{equation}

440: \label{eq36}

441: B_1 =-\frac{I\,k^2e^{-i{\kern 1pt}\omega {\kern 1pt}t}}{4\pi 

442: \,c}\int\limits_0^{2\pi } {\left\{ {\int\limits_0^a {\left[ {\frac{s\cos 

443: \varphi 'dr'}{R}} \right]_{s=z-b}^{s=z+b}\!\!\!\! -\int\limits_{z-b}^{z+b} {\left[ 

444: {\frac{\left( {r-r'\cos \varphi '} \right)\,dz'}{R}} \right]_{r'=0}^{r'=a} } 

445: } } \right\}d\varphi ' }+ O\left( {k^{n>2}} \right)

446: \end{equation}

447: The integration over $r'$ and $z'$ may be carried out analytically to yield:

448: \begin{equation}

449: \label{eq37}

450: B_1 =\frac{I\,k^2e^{-i{\kern 1pt}\omega {\kern 1pt}t}}{4\pi 

451: \,c}\int\limits_0^{2\pi } {\left[ {s\cos \varphi '\ln \left( {r'-r\cos 

452: \varphi '+R} \right)+\left( {r-r'\cos \varphi '} \right)\ln \left( {s+R} 

453: \right)} \right]_{s=z-b,\,r'=0}^{s=z+b,\,r'=a} d\varphi '} 

454: \end{equation}

455: Expanding this expression in a power series of $r$, and inserting it into 

456: the left-hand-side of (\ref{eq34}) we find for $z=0$:

457: \begin{equation}

458: \label{eq38}

459: \frac{I\,k^2e^{-i{\kern 1pt}\omega {\kern 1pt}t}}{2\,c}\,\left[ {\left( 

460: {\frac{b}{\left( {a^2+b^2} \right)^{\frac{3}{2}}}-\frac{1}{b^2}} 

461: \right)\,r^2+\left( {\frac{1}{2\,b^4}-\frac{b\left( {2b^2-3a^2} 

462: \right)}{4\left( {a^2+b^2} \right)^{\frac{7}{2}}}} \right)\,r^4} 

463: \right]\;+O\left( {r^{n>4}} \right)

464: \end{equation}

465: which is obviously at variance with the right-hand-side of (\ref{eq35}). A similar 

466: conclusion is reached, if (\ref{eq33}) is substituted into (\ref{eq26}). This can only be 

467: checked numerically, since the instantaneous field $\vec {E}_0 $ does not 

468: vanish outside the cylinder, in contrast to $\vec {B}_0 $.

469: 

470: Result (\ref{eq38}) proves that the standard solutions (\ref{eq32}) and (\ref{eq33}) do not satisfy 

471: the first order system from which equations (\ref{eq16}) and (\ref{eq17}) were derived. 

472: Hence, our conclusion in Sect. 2., namely that the first order system does 

473: not permit a solution, cannot be refuted by referring to the retarded 

474: solutions as taught in the textbooks such as [4]. 

475: 

476: There is also a physical reason to reject Jackson's solution (\ref{eq31}) for the 

477: considered case. If one calculates the fields with (\ref{eq18}) from (\ref{eq31}) and 

478: evaluates the Poynting vector $\vec {E}\times \vec {B}$ at large distance, 

479: one can integrate the total radiation power emitted by the closed circuit of 

480: Fig. 1:

481: \begin{equation}

482: \label{eq39}

483: P_{tot} =\int\!\!\!\int {\frac{c}{4\pi }} \left( {\vec {E}\times \vec {B}} 

484: \right)\cdot \vec {d}^2x=\frac{I^2\,a^4\,b^2\,k^6}{6\,c}

485: \end{equation}

486: This result is obviously not physical. The device in question may be seen as 

487: a short-circuited cable which should not continuously loose energy to the 

488: outside world; in particular not when the enclosing shell would be made out 

489: of superconducting material. The predicted power loss (\ref{eq39}) could certainly 

490: not be confirmed experimentally.

491: 

492: \vspace{.6cm}

493: 

494: \noindent \textbf{5. Conclusions}

495: 

496: \noindent It has been shown that an attempt to calculate numerically the temporal 

497: evolution of the electromagnetic field from the full set of Maxwell's first 

498: order equations will fail due to the internal inconsistencies built into the 

499: coupled system of equations. As noted earlier [2], the reason lies in the 

500: fact that the travelling wave fields are connected with the stationary 

501: sources at the same time.

502: 

503: Maxwell's equations describe correctly the production of the instantaneous 

504: electromagnetic field, and also the propagation of wave fields in empty 

505: space. The production mechanism of electromagnetic waves by time varying 

506: sources, however, does not find an explanation in the framework of Maxwell's 

507: theory. Contrary to what is commonly believed, the retarded solutions for 

508: the electromagnetic potentials do not lead to fields which are in agreement 

509: with a direct solution of the second order differential equations for the 

510: fields.

511: 

512: \vspace{.6cm}

513: \newpage

514: \noindent \textbf{Acknowledgment}

515: 

516: \noindent The author is indebted to V. Onoochin for initiating this work. Vladimir 

517: contributed significantly in the early discussions, but he modestly felt 

518: that he should not be a co-author of the paper. 

519: 

520: \vspace{.6cm}

521: 

522: 

523: \begin{thebibliography}{99}

524: \bibitem{ Onoochin }

525: { Onoochin V V 2002 \textit{Annales de la Fondation Louis de Broglie} \textbf{27} 163 }

526: \bibitem { Engelhardt } { Engelhardt W 2005 \textit{Annales de la Fondation Louis de Broglie} \textbf{30} 157 }

527: \bibitem{ Funaro } { Prof. Daniele Funaro (University of Modena), Prof. Michel de Haan (Free 

528: University of Brussels), private communications }

529: \bibitem{ Jackson} { Jackson J D 1975 \textit{Classical Electrodynamics, }Second Edition (New York: John Wiley {\&} Sons, Inc), 

530: Sect. 9.1 }

531: 

532: \end{thebibliography}

533: 

534: 

535: \end{large}

536: 

537: \end{document}

538: 

539: