SciELO - Scientific Electronic Library Online

vol.28 issue1Physical and Mathematical Modeling of non Transferred Plasma TorchesVector Supersymmetry of Three Dimensional Cohomological Field Theories author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Brazilian Journal of Physics

Print version ISSN 0103-9733On-line version ISSN 1678-4448

Braz. J. Phys. vol. 28 n. 1 São Paulo Mar. 1998 

One And The Same Route: Two Outstanding Electrodynamics


Antonio Accioly(1)* and  Hatsumi Mukai(2)
(1) Instituto de Física Teórica,
Universidade Estadual Paulista,
Rua Pamplona,145, 01405-900 São Paulo, SP, Brazil
(2) Departamento de Física,
Fundação Universidade Estadual de Maringá,
Av. Colombo 5790, 87020-900 Maringá, Pr, Brazil


Received August 27, 1997



We show that the same route that leads to Maxwell's electrodynamics leads also to Podolsky's electrodynamics, provided we start from Podolsky's electrostatic force law instead of the usual Coulomb's law.



I. Introduction

On the second edition of Jackson's seminal book on classical electrodynamics[1], there is a section named "On the Question of Obtaining the Magnetic Field, Magnetic Force, and the Maxwell Equations from Coulomb's Law and Special Relativity", where it is shown in detail that any attempt to derive Maxwell equations from Coulomb's law of electrostatics and the laws of special relativity ends in failure unless one makes use of additional assumptions. What hypotheses are these? In an ingenious paper, Kobe[2] gave the answer: all one needs to arrive at Maxwell equations is

(i) Coulomb's law;

(ii) the principle of superposition;

(iii) the assumption that electric charge is a conserved scalar (which amounts to assuming the independence of the observed charge of a particle on its speed[3]);

(iv) the requirement of form invariance of the electrostatic field equations under Lorentz transformations, i.e. the electrostatic field equations are thought as covariant space-space components of covariant field equations.

Soon after Kobe's paper, Neuenschwander and Turner[4] obtained Maxwell equations by generalizing the laws of magnetostatics, which follow from the Biot-Savart law and magnetostatics, to be consistent with special relativity.

The preceding considerations leads us to the interesting question: what would happen if we followed the same route as Kobe did, using an electrostatic force law other than the usual Coulomb's one? We shall show that if we start from the force law proposed by Podolsky[5], i.e.,

acfo1.gif (668 bytes)

where a is a positive parameter with dimension of length, Q and Q¢ are the charges at r and r = 0, respectively, and F(r) is the force on the particle with charge Q due to the particle with charge Q¢ - and if we follow the steps previously outlined, we arrive at the outstanding electrodynamics derived by Podolsky[5] in the early 40 s. In other words, we shall show that the same route that leads to Maxwell equations leads also to Podolsky equations. A notable feature of Podolsky's generalized electrodynamics is that it is free of those infinities which are usually associated with a point source. For instance, (1) approaches a finite value acfo36-1.gif (209 bytes) as r approaches zero. Thus, unlike Coulomb's law, Podolsky's electrostatic force law is finite in the whole space.

In Sec. II we obtain the equations that make up Podolsky's electrostatics. In Sec. III we arrive at Podolsky's field equations by generalizing the equations of Sec. II, so that they are form invariant under Lorentz transformations. For consistency, we show in Sec. IV that (1) is indeed the electrostatic force law related to Podolsky's theory. The conclusions are presented in Sec. V.

Natural units acfo36-2.gif (216 bytes)are used throughout. As far as the electromagnetic theories are concerned, we will use the Heaviside-Lorentz units with c = 1.


II. Podolsky's electrostatics

As is well-known, the force on a text charge is proportional to its charge, all other properties of the force being assigned to the electric field E (r), which is defined by

acfo36-3.gif (250 bytes)

where F is force on the charge Q situated at r and E(r) is the electric field at the position of this charge due to all other charges. (The source charge's coordinates will be distinguished from those of the field point, by a prime.) Accordingly, the electric field due to a point charge Q¢ situated at r¢ is given by

acfo2.gif (737 bytes)

where R = r - r¢. Note that this field is finite in the whole space. By the principle of superposition the electric field produced by the charge distribution r(r¢) is

acfo3.gif (830 bytes)

where the integration is carried out over all space. (r must of course vanish at sufficient large r, making the volume of integration less than infinity.)

The preceding expression for the electric field arising from a charge distribution may be easily expressed as a gradient of a scalar integral as follows

acfo4.gif (708 bytes)

Since Ñ acts only on the unprimed coordinates, we may take it outside the integral (4) to obtain

acfo5.gif (733 bytes)

Taking into account that the curl of any gradient vanishes, we have immediately

acfo6.gif (279 bytes)

which shows that Podolsky's electrostatic field, like Maxwell's one, is conservative. If the divergence of (3) is taken, the result is

acfo7.gif (1413 bytes)

where use has been made of the identity

acfo36-4.gif (401 bytes)

Now, taking the Laplacian of (7), and using the identity (see appendix)

acfo36-5.gif (496 bytes)


acfo37-1.gif (1820 bytes)


acfo8.gif (561 bytes)

Equations (6) and (8) are the fundamental laws of Podolsky's electrostatics. We will digress slightly at this stage to analyze an interesting feature of Podolsky's electrostatics. In this vein, we compute the flux of the electrostatic field across a spherical surface of radius R with a charge Q at its center. Using (2) we arrive at the result

acfo9.gif (688 bytes)

which tells us that

acfo37-2.gif (570 bytes)

Therefore, a sphere of radius R << a, unlike what happens in Maxwell's theory, shields its exterior from the field due to a charge placed at its center. We remark that in Maxwell's electrostatics a closed hollow conductor shields its interior from fields due to charges outside, but does not shield its interior from the field due to charges placed inside it[6]. Note, however, that in order not to conflict with well established results of quantum electrodynamics, the parameter a must be small. Incidentally, it was shown recently that this parameter is of the order of magnitude of the Compton wavelength of the neutral vector boson z, l(z) » 2.16×10-16 cm, which mediates the unified and electromagnetic interactions[7]. After this parenthesis, let us return to our main subject. Equations (6) and (8) are now ready to be generalized using special relativity and the hypotheses that electric charge is a conserved scalar. We shall do that in the next section.


III. Podolsky's field equations

To begin with let us establish some conventions and notations to be used from now on. We use the metric tensor

acfo37-3.gif (969 bytes)

with Greek indices running over 0,1,2,3. Roman indices - i, j, etc , - denote only the three spatial components. Repeated indices are summed in all cases. The space-time four vectors (contravariant vectors) are xm = ( t , x ) , and the covariant vectors, as a consequences are xm = ( t, - x ) . The four-velocities are found, according to

acfo37-4.gif (581 bytes)

where t is the proper time (dt2 = dt2 - dx2 ), and g denotes dt/dt = (1- v2)-1/2.

Let us then generalize (6) so that it satisfies the requirement of form invariance under Lorentz transformations. To do that, we write the mentioned equation in terms of the Levi-Civita density enml, which equals +1 (-1) if n, m, l  is an even(odd) permutation of 1,2,3, and vanishes if two indices are equal. The curl equation becomes

acfo10.gif (346 bytes)

It we define the quantities

acfo11.gif (381 bytes)

(10) can be rewritten as

acfo38-1.gif (287 bytes)

We imagine now the curl law to be the space-space components of a manifestly covariant field equation (invariance under Lorentz transformations). As a result, we get

acfo12.gif (395 bytes)

where emnab is a completely antisymmetric tensor of rank four with e0123 = +1 .

Of course, this generalization introduces the components F00, Fl0, and Flk, for which at this point we lack a physical interpretation. Note that the F0i are not necessarily static anymore.

On the other hand, as is well-known, the charge density r is defined as the charge per unit of volume, which has as a consequence that the charge dq in an element of volume d3 x is dq = rd3x. Since dq is a Lorentz scalar[3], r transforms as the time-component of a four-vector, namely, the time-component of the charge-current four-vector jm = (r, j ). The electric charge, in turn, is conserved locally[3], which implies that it obeys a continuity equation

acfo13.gif (289 bytes)

(8) can now be rewritten as

acfo38-2.gif (450 bytes)

where i º / xi and i º / xi . Note that i = - i. Using (11), yields

acfo38-3.gif (447 bytes)

In order that the left-hand side of the preceding equation transforms as the time-component of a four-vector, we must write it as

acfo38-4.gif (423 bytes)


acfo38-5.gif (618 bytes)

The requirement of form invariance of this equation under Lorentz transformations leads then to the following result

acfo14.gif (482 bytes)

Imagine now a particle of mass m and charge Q at rest in a lab frame where there is an electrostatic field E. Newton's second law allows us to write

acfo15.gif (322 bytes)

In terms of the proper time this becomes

acfo38-6.gif (418 bytes)

where u0 is the time part of the velocity four-vector um. For the component along de xi direction, we have

acfo38-7.gif (323 bytes)

In order that the right-hand side of this equation transforms like a space-component of a four-vector, it must be rewritten as

acfo38-8.gif (329 bytes)

whose covariant generalization is

acfo16.gif (404 bytes)

If (16) is multiplied by pm = mum, where m is the rest mass, the result is

acfo38-9.gif (549 bytes)


acfo38-10.gif (623 bytes)

Therefore, we come to the conclusion that

acfo39-1.gif (272 bytes)

Using this result Kobe[2] and Neuenschwander and Turner[3] showed that Fmn is an antisymmetric tensor (Fmn = -Fnm). Since Fmn is an antisymmetric tensor of second rank, it has only six independent components, three of which have already been specified. We name therefore the remaining components

acfo17.gif (366 bytes)

Note that Fkl = -ekljBj. Writing out the components of (17) explicitly,

acfo39-2.gif (798 bytes)

Hence, a clever physicist who were only familiar with Podolsky's electrostatics and special relativity could predict the existence of the magnetic field B, which naturally still lacks physical interpretation.

The content of (12) and (14) can now be seen. For m = 0, (12) gives

acfo18.gif (277 bytes)

showing that there are no magnetic monopoles in Podolsky's electrodynamics, while for m = i we obtain

acfo19.gif (335 bytes)

which says that time-varying magnetic fields can be produced be E fields with circulation.

The components m = 0 and m = i of (14) give, respectively,

acfo20.gif (415 bytes)

acfo21.gif (564 bytes)

which are nothing but a generalization of Gauss and Ampère- Maxwell laws in this order.

For n = i, (16) becomes

acfo22.gif (440 bytes)

containing the Lorentz force. For n = 0, (15) assumes the form

acfo23.gif (355 bytes)

where U = p0 is the particle's energy. Accordingly, our smart physicist, who was able to predict the B field only from its knowledge of electrostatics and special relativity, can now-by making judicious use of (22) and (23) - observe, measure and distinguish the B field from the E field of (15). The new field couples to moving electric charge, does not act on a static charged particle, and, unlike the electrostatic field, is capable only of changing the particle's momentum direction.

Equations (18-21) make up Podolsky's higher-order field equations. Of course, in the limit a = 0, all the preceding arguments apply equally well to Maxwell's theory.

Two comments fit in here:

(1) equation (14) is consistent with the continuity equation (13). In fact, if the divergence of (14) is taken, we obtain

acfo39-3.gif (490 bytes)

Since Fmn is an antisymmetric tensor m nFmn is identically zero. On the other hand, according to (14), m jm = 0. Thus, the equation in hand is identically zero;

(2) as was recently shown[8], it is not necessary to introduce a formula for the force density acfo39-4.gif (136 bytes) representing the action of the field on a text particle. We have only to assume that acfo39-5.gif (178 bytes) is the simplest contravariant vector constructed with the current jm and a suitable derivative of the field Fmn. Applying this simplicity criterion to Podolsky's electrodynamics, we promptly obtain

acfo39-6.gif (260 bytes)

where, as we have already mentioned, jm = (r,j). Therefore,

acfo39-7.gif (320 bytes)


acfo40-1.gif (896 bytes)

Thus, the force density for Podolsky's electrodynamics is the same as that for Maxwell's electrodynamics, namely, the well-known Lorentz force density.


IV. Finding the force law for Podolsky's electrostatics

We show now that (1) is indeed the force law for Podolsky's electrostatics. ¿From (5) it follows that

acfo24.gif (279 bytes)


acfo40-2.gif (506 bytes)

Eq. (8) can then be rewritten as

acfo40-3.gif (514 bytes)

For a charge Q at the origin of the radius vector this equation reduces to

acfo25.gif (635 bytes)

We solve this equation using the Fourier transform method. First we define acfo40-4.gif (175 bytes) as follows:

acfo26.gif (627 bytes)

acfo27.gif (617 bytes)

where d3 k and d3 r, respectively, stands for volumes in the three-dimensional k-space and the coordinate space. If we substitute (26) into (25) and take into account that

acfo40-5.gif (490 bytes)

we promptly obtain

acfo40-6.gif (564 bytes)

where M2 º 1/a2. So,

acfo40-7.gif (663 bytes)

Since the orientation of our coordinate system is arbitrary, we may choose the z-axis along r and obtain

acfo40-8.gif (1049 bytes)

where r = |r|, k = |k|, and (q, f) are conventional spherical polar coordinates. As a consequence,

acfo40-9.gif (1031 bytes)


acfo40-10.gif (687 bytes)


acfo40-11.gif (470 bytes)


acfo29.gif (795 bytes)

Integral (28) may be found in any textbook on the theory of functions of a complex variable[9]. It can also be carried out by means of a trivial trick[10]. Indeed, let

acfo41-1.gif (1357 bytes)

But I(¥) = 0. Therefore C = p/2. I(a) = [(p)/ 2] -tan-1 a and I1 = 2 I(0) = p.

Integral (29) can be easily evaluated by the method of contour integration[11]. Consider in this vein acfo41-2.gif (304 bytes) , where the contour of integration g was chosen to be the real axis and a semicircle of infinite radius in the upper half plane. Along the real axis the integral is I2. In the large semicircle in the upper half plane we get zero, since exp(iz) ® 0 for z ® i¥.

The residue of

acfo41-3.gif (608 bytes)

at z = Mri (which is the only pole that lies inside the contour of integration) is acfo41-4.gif (167 bytes). Hence,

acfo41-5.gif (453 bytes)

As a result,

acfo41-6.gif (691 bytes)

Accordingly, the electric field due to a charge Q at the origin is given by

acfo41-7.gif (1006 bytes)

It follows then the force law for Podolsky's electrostatics is

acfo41-8.gif (586 bytes)

which is nothing but the force law for which we were looking (see Eq. (1)).

Recently an algorithm was devised which allows one to obtain the energy and momentum related to a given field in a simple way[8]. Using this prescription we can show that in the framework of Podolsky's electrostatics the energy is given by

acfo41-9.gif (620 bytes)

Making use of the expression for the electrostatic field we have just found, we promptly obtain

acfo41-10.gif (313 bytes)

which tells us that the energy for the field of a point charge has a finite value in the whole space. This is indeed a nice feature of Podolsky's generalized electrodynamics.


V. Final remarks

Despite the simplicity of its fundamental assumptions, Podolsky's model has been little noticed. Currently some of its aspects have been further studied in the literature[7,8,12,13]. In particular, the classical self-force acting on a point charge in Podolsky's model was evaluated and it was shown that in this model, unlike what happens in Maxwell's electrodynamics, the electromagnetic mass is finite and enters the particle's equation of motion in a form consistent with special relativity.

To conclude we call attention to the fact the same assumptions that lead to Maxwell's equations lead also to Podolsky's equations, provided we start from a generalization of the Coulomb's law instead of the usual Coulomb's law. Yet, in spite of the great similarity between the two theories, Podolsky's generalized electrodynamics leads to results that are free of those infinities which are usually associated with a point source.


Appendix: An important identity involving d functions

An useful identity involving d functions is

acfo41-11.gif (525 bytes)

where R = | r - r¢ |.

To prove this let us consider

acfo41-12.gif (532 bytes)

If r is not in the region V over which we are integrating, it never coincides with r¢, and it is easily verified by direct differentiation that acfo42-1.gif (344 bytes). Any region V that contains r¢ may be subdivided into a small sphere of radius r centered on r surrounded by surface S and the remaining volume where acfo42-2.gif (334 bytes). Hence,

acfo42-3.gif (1467 bytes)

Now, according to the mean value theorem,

acfo42-4.gif (1063 bytes)

where x is some point in the sphere. Applying the divergence theorem to the last integral, we obtain

acfo42-5.gif (1031 bytes)

Note that Ñr = - Ñr '. On the other hand,

acfo42-6.gif (956 bytes)

As r tends to 0, x must approach r, so that

acfo42-7.gif (730 bytes)


acfo42-8.gif (1416 bytes)



1. J. D. Jackson, Classical Electrodynamics (Wiley, New York, 1975), 2nd. ed., pp. 578-581.         [ Links ]

2. D. H. Kobe, Am. J. Phys 54, 631 (1986).         [ Links ]

3. See Ref. 1, pp. 547-552.

4. D. E. Neuenschwander and B. N. Turner, Am. J. Phys. 60, 35 (1992).         [ Links ]

5. B. Podolsky, Phys. Rev. 62, 66 (1942);         [ Links ]B.Podolsky and P. Schwed, Rev. Mod. Phys, 20, 40 (1948).

6. See Ref. 1, p.49.

7. J. Frenkel and R. B. Santos, "The self-force on point charged particles in Podolsky's generalized electrodynamics," (to be published).         [ Links ]

8. Antonio Accioly, Am. J. Phys. 65, 882 (1997).         [ Links ]

9. F. W. Byron, Jr, and R. W. Fuller, Mathematics of Classical and Quantum Physics (Addison-Wesley Publishing Company, New York, 1970), volume two, pp. 366-367.         [ Links ]

10. Jon Mathews and R.L. Walker, Mathematical Methods of Physics (W.A. Benjamin, Inc., New York, 1965) p. 58.         [ Links ]

11. See Ref. 9, pp. 362-365.

12. C. A. P. Galvão and B. M. Pimentel, Can. J. Phys. 66, 460 (1988);         [ Links ]J. Barcelos-Neto, C. A. P. Galvão, C. P. Natividade, Z. Phys. C52, 559 (1991); L. V. Belvedere, C. P. Natividade, C. A. P. Galvão, Z. Phys. C56, 609 (1992).         [ Links ]

13. A. J. Accioly and H. Mukai, Z.Phys. C 75, 187 (1997).         [ Links ]



* e-mail:

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License