Lagrangian formulation of electromagnetic elds in nondispersive medium

by means of the extended EulerLagrange differential equation

Cem Civelek
a,
*
, Thomas Franz Bechteler
b
a
Dokuz Eyll University, Institute of Marine Sciences and Technology, Bak Bulvar No. 100, TR-35340 Inciralt, Izmir, Turkey
b
Yas ar University, Faculty of Engineering and Architecture, TR-35500 Bornova, Izmir, Turkey
a r t i c l e i n f o
Article history:
Received 27 March 2008
Received in revised form 22 May 2008
Accepted 13 June 2008
Available online 25 July 2008
Keywords:
Lagrangian density
EulerLagrange differential equation
Electromagnetic potentials
Maxwells equations
a b s t r a c t
This work is concerned with the Lagrangian formulation of electromagnetic elds. Here,
the extended EulerLagrange differential equation for continuous, nondispersive media
is employed. The Lagrangian density for electromagnetic elds is extended to derive all
four Maxwells equations by means of electric and magnetic potentials. For the rst time,
ohmic losses for time and space variant elds are included. Therefore, a dissipation density
function with time dependent and gradient dependent terms is developed. Both, the
Lagrangian density and the dissipation density functions obey the extended Euler
Lagrange differential equation. Finally, two examples demonstrate the advantage of
describing interacting physical systems by a single Lagrangian density.
2008 Elsevier Ltd. All rights reserved.
1. Introduction
In 1788, Lagrange elegantly reformulated classical mechanics for conservative systems [1]. He reduced a large number of
interacting forces in a system to only two forms of energy: the kinetic energy and the potential energy. Instead of dealing
with vectors, only a scalar function must be considered. Furthermore, Lagranges formalism holds in any coordinate system.
These advantages initiated many works in mechanics and electronics using Lagranges formalism.
For instance, a Lagrangian formulation for continuous systems in mechanics is given in [2]. Jos and Saletan [3] cover the
Lagrangian formulation of continuum dynamics, and Scheck [4] treats discrete and continuous systems, the transition to a
continuous system and the Hamilton variational principle for a continuous system.
In his works, Ssse [59] approaches theoretical foundations of electrical engineering using classical Lagrange and Ham-
ilton formalisms, including losses. In [10,11] electrical lumped devices and electromechanical systems are described using
Lagrange and Hamilton formalism with and without losses, whereas the generalized motion in Riemannian space, i.e.
non-Euclidian, is considered. Further investigations on Lagrangians for lumped RLC-circuits are presented, for instance, in
[1214], and nonlinear, lumped RLC networks are described in [15,16].
Lagrangian formulation in electrodynamics has to consider time and local variations. For example, Kosyakov [17] con-
structs Lagrangian densities for a lossless electromagnetic eld and for particles moving in such a eld, whereas he focuses
on developing EulerLagrange equations in tensor notation. Carroll [18] makes use of spacetime reversed elds to construct
a Lagrangian including ohmic loss. However, losses for stationary elds, i.e. for the case o=ot 0, are not considered. Ter Haar
[19] not only treats the Lagrangian density for continuous media without losses, but also derives two Maxwells equations
0020-7225/$ - see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijengsci.2008.06.007
* Corresponding author.
E-mail address: cem.civelek@deu.edu.tr (C. Civelek).
International Journal of Engineering Science 46 (2008) 12181227
Contents lists available at ScienceDirect
International Journal of Engineering Science
j our nal homepage: www. el sevi er . com/ l ocat e/ i j engsci
from the Lagrangian density for an electromagnetic eld. Simonyi [20] derives the same two Maxwells equations, namely
Gauss lawr D .
e
and Ampres lawrH
oD
ot
. However, a complete treatment with respect to deriving all of Maxwells
equations is lacking.
In the following, a Lagrangian density will be established for electromagnetic elds in order to derive all four Maxwells
equations. Additionally, ohmic losses are considered which leads to the extended EulerLagrange differential equation and a
dissipation density function. As a nal point, two examples are presented which demonstrate that two interacting physical
systems can be described by a single, scalar Lagrangian density.
2. EulerLagrange differential equation and Lagrangian density
Lagrange formalism in electromagnetics means, representing electromagnetic elds by energy densities and obtaining a
Lagrangian density which obeys the so-called EulerLagrange differential equation. This differential equation fulllls the
principle of least action. In electrodynamics, the elds are not only time dependent but also space dependent. Hence, the
EulerLagrange differential equation for a continuous, lossless medium must be written as

m
k1
o
ox
k
oL
o
og
i
ox
k
_ _
_
_
_
_

o
ot
oL
o
og
i
ot
_ _
_
_
_
_

oL
og
i
F
i
; 1
where there is one equation for each value of i, and where Lis the Lagrangian density, F
i
is the external force corresponding
to g
i
, which is a time and space dependent potential, x
k
are the system coordinates, and t is the time. The Lagrangian density,
a scalar function, is of the form
L L g
i
; rg
i
;
og
i
ot
_ _
i 1; 2; . . . ; n: 2
Since the generalized forces F
i
act only on g
i
, the forces can be successively decoupled and are therefore included in the
Lagrangian density as negative potential densities g
i
F
i
:

m
k1
o
ox
k
oL
o
og
i
ox
k
_ _
_
_
_
_

o
ot
oL
o
og
i
ot
_ _
_
_
_
_

oL
og
i
0; 3
with
L L g
i
; g
i
F
i
; rg
i
;
og
i
ot
_ _
i 1; 2; . . . ; n: 4
The task is to setup the Lagrangian density (4) for the electromagnetic eld obeying the EulerLagrange differential Eq. (3).
2.1. Lagrangian density for the electromagnetic eld
The Lagrangian density consists of two parts, the kinetic part and the potential part. For an electromagnetic eld, the ki-
netic part is the energy density stored in the electric eld and the potential part is the energy density stored in the magnetic
eld. Energy densities in a linear electric eld and a linear magnetic eld are
w
e

1
2
D E and w
m

1
2
B H; 5
respectively, where D eE is the electric displacement, E is the electric eld, B lH is the magnetic ux density, and H is
the magnetic eld. Electrical properties of the medium are described by the permittivity e and permeability l. For an elec-
tromagnetic eld, the Lagrangian density in its contravariant form was discovered by Larmor [21] and is written as
L w
e
w
m

1
2
eE
2

1
2
lH
2
: 6
The electric eld E and magnetic eld H can be expressed by potential functions u and A [22]. Therefore, the Lagrangian
density is conveniently written as
L
1
2
e ru
oA
ot
_ _
2

1
2l
rA
2
; 7
with
E ru
oA
ot
8
lH B rA; 9
where u and A are the electric scalar potential and the magnetic vector potential, respectively.
C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227 1219
We can identify the potentials g
i
in Eq. (3) with g
1
u, g
2
A
x
, g
3
A
y
and g
4
A
z
, where x, y, and z are the Cartesian
coordinates. The sources for an electromagnetic eld are the electric volume charge density .
e
F
1
and the electric cur-
rent density vector J
e0
with J
e0;x
F
2
, J
e0;y
F
3
, and J
e0;z
F
4
. With these sources, the Lagrangian density in Eq. (7) can be
extended to
L
1
2
e ru
oA
ot
_ _
2

1
2l
rA
2
u.
e
J
e0
A: 10
Now, it must be shown that the Lagrangian density given by Eq. (10) results in Maxwells equations. Substituting g
1
u in
Eq. (3) and plugging L into the EulerLagrange differential Eq. (3) results in
oL
o
ou
ox
_ _ e
ou
ox

oA
x
ot
_ _
D
x
oL
o
ou
oy
_ _ e
ou
oy

oA
y
ot
_ _
D
y
oL
o
ou
oz
_ _ e
ou
oz

oA
z
ot
_ _
D
z
11
)

3
k1
o
ox
k
oL
o
ou
ox
k
_ _ r D
o
ot
oL
o
ou
ot
_ _ 0
oL
ou
.
e
:
By inspection, this is Maxwells rst equation, also called Gauss law:
r D .
e
: 12
Further, substituting g
2
A
x
in Eq. (3) and plugging L into the EulerLagrange differential equation results in
o
ox
oL
o
oAx
ox
_ _ 0
o
oy
oL
o
oAx
oy
_ _
o
oy
1
l
oA
y
ox

oA
x
oy
_ _

o
oy
H
z
o
oz
oL
o
oAx
oz
_ _
o
oz
1
l
oA
x
oz

oA
z
ox
_ _

o
oz
H
y
o
ot
oL
o
oAx
ot
_ _
o
ot
e
ou
ox

oA
x
ot
_ _

o
ot
D
x
oL
oA
x
J
e0;x
13
same for g
3
A
y
:
o
ox
oL
o
oAy
ox
_ _
o
ox
1
l
oA
y
ox

oA
x
oy
_ _

o
ox
H
z
o
oy
oL
o
oAy
oy
_ _ 0
o
oz
oL
o
oAy
oz
_ _
o
oz
1
l
oA
z
oy

oA
y
oz
_ _

o
oz
H
x
o
ot
oL
o
oAy
ot
_ _
o
ot
e
ou
oy

oA
y
ot
_ _

o
ot
D
y
oL
oA
y
J
e0;y
14
1220 C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227
And likewise for g
4
A
z
:
o
ox
oL
o
oAz
ox

o
ox
1
l
oA
x
oz

oA
z
ox
_ _

o
ox
H
y
o
oy
oL
o
oAz
oy

o
oy
1
l
oA
y
oz

oA
z
oy
_ _

o
oy
H
x
o
oz
oL
o
oAz
oz
_ _ 0
o
ot
oL
o
oAz
ot
_ _
o
ot
e
ou
oz

oA
z
ot
_ _

o
ot
D
z
oL
oA
z
J
e0;z
15
Bringing all sets of Eqs. (13)(15) into vector form and rearranging the terms, yields Maxwells second equation, also called
Ampres law:
rH J
e0

oD
ot
: 16
As shown, only two of the four Maxwells equations can be derived by means of the Lagrangian density dened in Eq. (10). To
obtain the other two Maxwells equations the dual Lagrangian density is introduced.
2.2. Dual Lagrangian density for the electromagnetic eld
Dirac [23] argued that there are magnetic monopoles, analogous to electric charges. Accepting the existence of magnetic
monopoles, would yield a symmetric set of Maxwells equations as presented, for instance, in [22]. As in [24,25], using the
duality theorem, the magnetic eld and the electric displacement are now dened as
H ru
m

oA
e
ot
; 17
and
D rA
e
; 18
where u
m
is the magnetic scalar potential and A
e
is the electric vector potential. This leads to the dual Lagrangian density
L
d

1
2
lH
2

1
2
eE
2

1
2
l ru
m

oA
e
ot
_ _
2

1
2e
rA
e

2
19
Assuming a magnetic volume charge density .
m
and a magnetic current density J
m0
, the dual Lagrangian density can be ex-
tended to
L
d

1
2
l ru
m

oA
e
ot
_ _
2

1
2e
rA
e

2
u
m
.
m
J
m0
A
e
: 20
In line with Section 2.1, it can be shown that the dual Lagrange density leads to the other two equations of Maxwell. Substi-
tuting g
5
u
m
in Eq. (3) and plugging L
d
into the EulerLagrange differential Eq. (3), results in Maxwells third equation,
also called Gauss law for magnetization:
r B .
m
: 21
Since magnetic monopoles have not been detected yet, the external force .
m
can be set to zero, and the conventional form
r B 0 is obtained.
Furthermore, substituting g
6
A
e;x
, g
7
A
e;y
, and g
8
A
e;z
in Eq. (3) and plugging L
d
into the EulerLagrange differential
equation, leads to Maxwells fourth equation in vector form, also called Faradays law of induction:
rE J
m0

oB
ot
: 22
Again, when there is no magnetic current J
m0
, the conventional form rE
oB
ot
is obtained.
C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227 1221
2.3. Total Lagrangian density
The sum of the Lagrangian density and the dual Lagrangian density is the total Lagrangian density L
T
. All four of Max-
wells equations can be derived from this single, total Lagrangian density (23).
L
T

1
2
e ru
oA
ot
_ _
2
l ru
m

oA
e
ot
_ _
2
_ _

1
2
rA
2
l

rA
e

2
e
_ _
u.
e
J
e0
A u
m
.
m
J
m0
A
e
23
3. Extended EulerLagrange differential equation: losses
Whenever a medium is present, losses are introduced. Therefore, a dissipation density function D has to be considered,
similar to the Lagrangian density. The Lagrangian density is an energy density whereas the dissipation density function is a
power density. The dissipation density function is the time average rate of energy dissipation per unit volume ow
h
=ot, with
w
h
as the energy density dissipated in heat, and the energy dissipation due to EM-eld variation in space. In the same way as
a potential w must satisfy the well-known three-dimensional wave equation

3
k1
c
2
o
2
w
ox
2
k

ow
2
ot
2
0; 24
with c 1=

el
p
as the speed of light, the dissipation density function D has to satisfy

m
k1
oD
o c
og
i
ox
k
_ _
oD
o
og
i
ot
_ _ 0; 25
when there are no losses. However, considering losses, the right hand side of Eq. (25) is not equal to zero. Therefore, the Eu-
lerLagrange differential equation has to be extended with a generalized velocity dependent term oD=o
og
i
ot
_ _
and a gradient
dependent term
1
c

m
k1
oD
oog
i
=ox
k

. Then, the extended EulerLagrange differential equation is written as

m
k1
o
ox
k
oL
o
og
i
ox
k
_ _
_
_
_
_

o
ot
oL
o
og
i
ot
_ _
_
_
_
_

1
c

m
k1
oD
o
og
i
ox
k
_ _
oD
o
og
i
ot
_ _
oL
og
i
0: 26
It has to be mentioned that the external forces F
i
can be included either as negative potential densities in the Lagrangian
density or as negative loss densities in the dissipation density function. The rst case has been preferred in this work, as
shown in Sections 2.1 and 2.2.
3.1. Losses due to electric conductivity the dissipation density function
In case of electric conductivity the losses are known as ohmic losses. In electromagnetics, Ohms law is dened as
J r
e
E; 27
where J is the current density vector, r
e
is the electric conductivity of the medium and E is the electric eld causing charge
transport. Losses due to electric conductivity depend, as the electric eld, on ruand on oA=ot. Hence, the dissipation density
function, plugged into the Extended EulerLagrange differential Eq. (26), should yield the additive term given by (27) in Max-
wells second equation. The ansatz to nd D is
1
c

3
k1
oD
o
oAx
ox
k
_ _
oD
o
oAx
ot
_ _ r
e
ou
ox

oA
x
ot
_ _
28
1
c

3
k1
oD
o
oAy
ox
k
_ _
oD
o
oAy
ot
_ _ r
e
ou
oy

oA
y
ot
_ _
29
1
c

3
k1
oD
o
oAz
ox
k
_ _
oD
o
oAz
ot
_ _ r
e
ou
oz

oA
z
ot
_ _
30
1
c

3
k1
oD
o
ou
ox
k
_ _
oD
o
ou
ot
_ _ 0: 31
1222 C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227
However, these partial differential equations are coupled with respect to the spatial derivatives ou=ox
k
. To decouple these
equations, the well-known Lorenz gauge
r A
1
c
2
ou
ot
0; 32
is used. Hence, Eq. (31) can be rewritten as
1
c

3
k1
oD
o
ou
ox
k
_ _
oD
o
ou
ot
_ _ r
e
r A
1
c
2
ou
ot
_ _
: 33
Integrating both sides with respect to spatial and time derivatives
1
c
oD
o
ou
ox
_ _ r
e
oA
x
ox
;
1
c
oD
o
ou
oy
_ _ r
e
oA
y
oy
;
1
c
oD
o
ou
oz
_ _ r
e
oA
z
oz
;
oD
o
ou
ot
_ _
r
e
c
2
ou
ot
;
correspondingly, leads to
D
u
cr
e
ou
ox
oA
x
ox

ou
oy
oA
y
oy

ou
oz
oA
z
oz
_ _

r
e
2c
2
ou
ot
_ _
2
: 34
The dissipation density function given by Eq. (34) satises Eqs. (28)(30) except for the terms r
e
oA
x
=ot, r
e
oA
y
=ot and
r
e
oA
z
=ot. Plugging D
u
into these equations yields only r
e
ru. Integrating
oD
ooA
i
=ot
r
e
oA
i
ot
on both sides, with i x; y; z, leads
to the time dependent term D
A

1
2
r
e
oA
x
=ot
2
oA
y
=ot
2
oA
z
=ot
2
_ _
. The complete dissipation density function is then
the sum of D
u
and D
A
:
D cr
e
ou
ox
oA
x
ox

ou
oy
oA
y
oy

ou
oz
oA
z
oz
_ _

1
2
r
e
c
2
ou
ot
_ _
2

1
2
r
e
oA
x
ot
_ _
2

oA
y
ot
_ _
2

oA
z
ot
_ _
2
_ _
;
35
which satises Eqs. (28)(31). The rst three terms of Eq. (35) are gradient dependent terms and describe losses due to local
eld variations. The last four terms of Eq. (35) are time dependent terms and are analogous to the generalized velocity
dependent (Rayleigh) dissipation function
oD
o
_
q
R
_
q ) D
1
2
R
_
q
2
for lumped elements, as in [1012], where R and
_
q are resistance and current, respectively.
3.2. Losses due to magnetic conductivity the dual dissipation density function
Just as there is a dual Lagrangian density, there is a dual dissipation density function D
d
. Similar to electric conductivity, a
magnetic conductivity r
m
can be dened [26]. Due to the magnetic eld the magnetic conductivity causes a magnetic cur-
rent density
J
m
r
m
H: 36
The dual dissipation density function can be established in the same manner as the dissipation density function in Section
3.1. It is given as
D
d
cr
m
ou
m
ox
oA
e;x
ox

ou
m
oy
oA
e;y
oy

ou
m
oz
oA
e;z
oz
_ _

1
2
r
m
c
2
ou
m
ot
_ _
2

1
2
r
m
oA
e;x
ot
_ _
2

oA
e;y
ot
_ _
2

oA
e;z
ot
_ _
2
_ _
; 37
with the Lorenz gauge
r A
e

1
c
2
ou
m
ot
0: 38
C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227 1223
3.3. Total dissipation density function
Analogous to the total Lagrangian density, the total dissipation density function is the sum of Eqs. (35) and (37):
D
T
DD
d
: 39
Plugging L
T
and D
T
into the extended EulerLagrange differential equation yields all four of Maxwells equations, including
electric and magnetic losses in Maxwells second and fourth equations:
rE J
m0
r
m
H
oB
ot
40
rH J
e0
r
e
E
oD
ot
: 41
4. Examples
4.1. Lagrangian density for a non-relativistic, charged particle interacting with an EM-eld
The Lagrangian density for interacting systems is the sum of the Lagrangian density for the rst system, the Lagrangian
density for the second system and a Lagrangian density L
i
accounting for the interacting part [3,27]. In this example, a
charged particle of mass m and electric charge q moves in an electromagnetic eld. By means of the Lagrangian density
and EulerLagrange differential equation, the equations of motions for the particle are obtained. The particle is described
by the Lagrangian density L
mass
, and the electromagnetic eld is described by the Lagrangian density L
EM
.
L L
mass
L
EM
L
i
: 42
In [5], for instance, this system has been described by a Lagrangian. To obtain the Lagrangian density, the Lagrangian in [5]
has to be derived with respect to volume V:
L
oL
oV

o
3
ozoyox
1
2
mr

2
quq r

A
_ _
; 43
with m, q, and r

x; y; z as the mass, charge and velocity vector of the particle, respectively. The potentials u and A de-
scribe the electromagnetic eld. This results in the three terms of the Lagrangian density. The mass related part is
L
mass

1
2
o
3
m
ozoyox
x

2
y

2
z

2
_ _

1
2
q
M
r

2
; 44
where q
M
is the specic mass of the particle. The part describing the electromagnetic eld is
L
EM

o
3
q
ozoyox
u
o
2
q
oyox
ou
oz

o
2
q
ozox
ou
oy

o
2
q
ozoy
ou
ox

oq
ox
o
2
u
ozoy

oq
oy
o
2
u
ozox

oq
oz
o
2
u
oyox
q
o
3
u
ozoyox
; 45
where o
3
q=ozoyox is the electric volume charge density .
e
. The part of the Lagrangian density which accounts for the inter-
action is derived as
L
i
.
e
r

A
_ _

o
2
q
oyox
r
oA
oz
_ _

o
2
q
ozox
r
oA
oy
_ _

o
2
q
ozoy
r
oA
ox
_ _

oq
ox
r
o
2
A
ozoy
_ _

oq
oy
r
o
2
A
ozox
_ _

oq
oz
r
o
2
A
oyox
_ _
q r
o
3
A
ozoyox
_ _
:
46
Additionally assuming losses of the moving particle leads to the velocity dependent (Rayleigh) dissipation density
function
oD
o r
i

o
3
B
ozoyox
r
i

) D
1
2
br

2
; 47
with B and b as friction and friction volume density, respectively, and r
i

as the vector components of the velocity vector of

the particle. Substituting the generalized coordinates g
i
x; y; z; q; u; A
x
; A
y
; A
z
in the extended EulerLagrange Eq. (26)
yields the equations of motion for the particle in density form:
r
o
3
A
ozoyox

o
3
u
ozoyox
0; .
e
0; .
e
r

0; 48
and
q
M
r
i

o
3
ozoyox
q
oA
i
ot
_ _

o
3
ozoyox
q

3
k1
r
k
oA
k
or
i
_ _

o
3
ozoyox
q
ou
or
i
_ _
b r
i

; 49
1224 C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227
with i 1; 2; 3. Except for the losses b r
i

, Eqs. (48) and (49) are analogous to the equations of motion as given, for instance,
in [5]. Integrating Eqs. (48) and (49) with respect to the volume o
3
=ozoyox yields
r

A u 0; q 0; q r

0; 50
and
m r

q
oA
ot
ru
_ _
q r rA B r

; 51
including losses due to friction, here called the dissipative Lorentz force.
4.2. Lagrangian density for an EM-eld interacting with an elastic wave
In this second example, a stationary EM-eld shall interact with an acoustic wave. The acoustic wave, which is an elastic
wave, causes spatial strain variations in the medium which in turn generates spatial permittivity variations [28]. Hence, the
EM-eld is affected by this acoustic wave. The electromagnetic eld is described by L
EM
and the acoustic eld is described
by L
acoustic
L L
EM
L
acoustic
L
i
: 52
A stationary EM-eld with no x-component of the electric eld is assumed, i.e. o=ot 0 and E
x
0. Therefore, its Lagrangian
density is
L
EM

1
2
eru
2

1
2l
rA
2
u.
e
: 53
The acoustic wave is a longitudinal wave propagating in x-direction. Potential and kinetic energies for an acoustic wave
are given in [29] and, hence, the corresponding Lagrangian density is given as
L
acoustic

1
2
B
on
ox
_ _
2

1
2
q
M
on
ot
_ _
2
; 54
where
n is the longitudinal displacement of particles from their equilibrium position,
q
M
is the specic mass of particles, and
B is the adiabatic bulk modulus.
The strain S on=ox causes a spatial variation De of the mediums permittivity. The permittivity is then e De. For small
variations in permittivity [28], the relation between De and S on=ox (tensor notation omitted) is
De e
2
P
on
ox
; 55
with P as a dimensionless constant depending on the medium. This leads to the part of the Lagrangian density which ac-
counts for the interaction, written as
L
i

1
2
e
2
P
on
ox
ru
2
: 56
The Lagrangian density L L
EM
L
acoustic
L
i
obeys the EulerLagrange differential Eq. (3) with g
1
u, g
2
A
x
,
g
3
A
y
, g
4
A
z
and g
5
n. Substituting g
1
u and plugging L into Eq. (3) results in
o
ox
oL
o
ou
ox
_ _ 0 since E
x
0 )
ou
ox
0
o
oy
oL
o
ou
oy
_ _
o
oy
e
ou
oy
e
2
P
on
ox
ou
oy
_ _
e De
o
oy
E
y
o
oz
oL
o
ou
oz
_ _
o
oz
e
ou
oz
e
2
P
on
ox
ou
oz
_ _
e De
o
oz
E
z
o
ot
oL
o
ou
ot
_ _ 0
oL
ou
.
e
)r D r e DeE .
e
57
C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227 1225
It is seen that the acoustic wave affects Maxwells rst equation, i.e. Gauss law. A modulation of strain in the medium results
in a modulation of the electric eld. Since L
i
does not include the vector potential A, the substitutions g
2
A
x
, g
3
A
y
and
g
4
A
z
simply yield rH 0, i.e. Ampres law for a stationary EM-eld. Substituting g
5
n and plugging L into Eq. (3)
results in
o
ox
oL
o
on
ox

o
ox
B
on
ox

1
2
e
2
Pru
2
_ _

o
ox
B
on
ox
o
oy
oL
o
on
oy

0
o
oz
oL
o
on
oz

0
o
ot
oL
o
on
ot

o
ot
q
M
on
ot
oL
on
0
)
o
2
n
ox
2

q
M
B
o
2
n
ot
2
0; 58
which is the wave equation for the acoustic wave, whereas the speed of this acoustic wave is dened as v

B=q
M
_
. For
g
5
n, the interacting part of the Lagrangian density L
i
does not contribute to Eq. (58), since E
x
0 and therefore
o
ox
1
2
e
2
Pru
2
_ _
0.
In general, it is very hard to nd the interacting part of the Lagrangian density. However, with this simple example it has
been demonstrated that it is possible to formulate elds of mixed type with a single Lagrangian density.
5. Conclusion
In this work, a Lagrangian density comprising electric and magnetic potentials has been developed. Furthermore, electric
(ohmic) and magnetic losses are described by a dissipation density function.
It is now possible to derive all four Maxwells equations in a straightforward way from a single, scalar Lagrangian density
function and a single, scalar dissipation density function. Both functions satisfy the extended EulerLagrange differential
equation.
The advantage of a Lagrangian formulation of time and space variant elds is not only its compactness but also the ability
to formulate elds of mixed type. For instance, a single Lagrangian can describe acousto-optic systems where EM-elds
interact with elastic waves.
References
[1] J.-L. Lagrange, Mcanique Analytique, Vvc Desaint, Paris, 1788.
[2] H. Goldstein, Classical Mechanics, Addison-Wesley, Massachusetts, 1980.
[3] J.V. Jos, E.J. Saletan, Classical Dynamics, Cambridge University Press, Cambridge, 1998.
[4] F. Scheck, Mechanics, Springer, Berlin, 2003.
[5] R. Ssse, B. Marx, Theoretische Elektrotechnik I. B.I. Wissenschaftsverlag, Mannheim, 1994.
[6] R. Ssse, U. Diemar, G. Michel, Theoretische Elektrotechnik II, VDI-Verlag, Dsseldorf, 1996.
[7] R. Ssse, E. Kallenbach, T. Strhla, Theoretische Elektrotechnik III. Wissenschaftsverlag Ilmenau, Ilmenau, 1997.
[8] R. Ssse, U. Dietmar, Theoretische Elektrotechnik IV, Wissenschaftsverlag Ilmenau, Ilmenau, 2000.
[9] R. Ssse, B. Marx, Theoretische Elektrotechnik V, Wissenschaftsverlag Ilmenau, Ilmenau, 2004.
[10] C. Civelek, Berechnung von elektrotechnischen Systemen mit Wandlern mittels erweitertemLagrange- bzw. Hamilton-Formalismus. Wissenschaftsverlag
Ilmenau, Ilmenau, 2001.
[11] R. Suesse, C. Civelek, Analysis of engineering systems by means of Lagrange and Hamilton formalisms depending on contravariant, covariant tensorial
variables, Engineering Research 68 (2003) 6674. July.
[12] H. Sira-Ramirez, R. Ortega, G. Escobar, Lagrangian modeling of switch regulated DC-to-DC power converters, in: Proceedings of 35th IEEE Conference
on Decision and Control, Kobe, Japan, December 1996, pp. 44924497.
[13] J. Clemente-Gallardo, J.M.A. Scherpen, Relating Lagrangian and Hamiltonian formalisms of LC circuits, IEEE Trans. Circuits Syst. I 50 (10) (2003) 1359
1363.
[14] F. Albertini, D. DAlessandro, Lagrangian formulation and geometric control of switching LC electrical networks, in: Proceedings of 45th IEEE
Conference on Decision and Control, San Diego, USA, December 2006, pp. 52045209.
[15] L.O. Chua, J.D. McPherson, Explicit topological formulation of Lagrangian and Hamiltonian equations for nonlinear networks, IEEE Trans. Circuits Syst.
21 (2) (1974) 277286.
[16] H.G. Kwatny, F.M. Massimo, L.Y. Bahar, The generalized Lagrange formulation for nonlinear RLC networks, IEEE Trans. Circuits Syst. 29 (4) (1982) 220
233.
[17] B.P. Kosyakov, Introduction to the Classical Theory of Particles and Fields, Springer, Berlin, 2007.
[18] J.E. Carroll, A lagrangian and hamiltonian formulation of electromagnetic theory with ohmic loss, Int. J. Electron. 54 (1) (1983) 139145.
[19] D. Ter Haar, Elements of Hamiltonian Mechanics, Pergamon Press, Oxford, 1961.
[20] K. Simonyi, Theoretische Elektrotechnik, VEB, Berlin, 1979.
[21] J. Larmor, Ether and Matter, Cambridge University Press, Cambridge, 1900.
[22] J.D. Jackson, Classical Electrodynamics, John Wiley, New York, 1975.
1226 C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227
[23] P.A.M. Dirac, The theory of magnetic poles, Phys. Rev. 74 (7) (1948) 817830.
[24] G. Lehner, Elektromagnetische Feldtheorie, Springer, Berlin, 2006.
[25] C.G. Someda, Electromagnetic Waves, CRC Press, Boca Raton FL, 2006.
[26] H.F. Harmuth, Nonsinusoidal waves, modied Maxwell equations, dogma of the continuum, Electromagnetic Phenomena 6 (1) (2007) 1935.
[27] E.J. Saletan, A.H. Cromer, Theoretical Mechanics, Wiley, New York, 1971.
[28] E. Dieulesaint, D. Royer, Elastic Waves in Solids, Wiley, New York, 1980.
[29] L.E. Kinsler, A.R. Frey, A.B. Coppens, J.V. Sanders, Fundamentals of Acoustics, Wiley, New York, 1982.
C. Civelek, T.F. Bechteler / International Journal of Engineering Science 46 (2008) 12181227 1227