MSci 4261 ELECTROMAGNETISM
LECTURE NOTES VIII
Jh these notes we will formulate dassical electioxynamics in an explicitly Lorentz. covariant: manner.
8.1 The Charge-Current Density Four-Vector
(Consider an element, of chargp at rest in a frame KX, occupying the volume element dr! dr? dr, so that.
a= par! dae,
Inthefiame K’ this same dament of charge will occupy the volume element cir di! dbr = (hdr! dr? dr
and since the total amount of charge in the volume element is the same in both ffames,
q= 9 dt! de? de®,
and it follows that
=.
But note also that in K? the element of charge is moving, with a velocity uf
canrent, density in K’ given by
“Bc so that there is also a
fu! = —yBep.
So we have (p5] = 0) in K transforming to (pl = ys! = —7Bep) in K, which suggests waiting 79 = op,J=
(7,.P,P) since then the transformation of j# may be recognised a8 that of a fou-veotor. In fact our
dlanent of charge has a fouevelocity U7, and we have
#
where jp is the charge dersity in the frame in which the dament of charge is at rest. We may condude that
‘the chargo-omrent density j = (cp,J) is a contravariant 4-vector.
Note also that the continuity equation:
By.sa
arty J=0
* oP oF OP | OF
ah -
a+ art * am * as =°
or
Ou =0
where 8
n= a
‘The continuity equation is thus a manifestly covariant equation.
8.2. The Lorentz, Force
Having found the trarsformation rule for the souves ji" it is natural to tum to the question of the
transformation nules for the fiek’s themselves. As a step in this direction, recall that the fields are in fact
defined through the Lorentz force
F=qB+uxB)
1on a test particle of charge q moving with velocity u. Since F = 4, we have
at
Pa qeruxpe
=qB+uxByu
= 42GB +UxB),
and also the rate of change ofthe energy € of the test particle is given by
@ 1 dea
cad
i
21gbeu
{p.u.
é
E-U
EU" +c xB
GF Us
whee F° is a certain second rank tensor, the components of which are given in terns of the E and B fiddds.
It possible to obtain this result directly, but it is easier and also more useful to prooeed by introducing
the potential foursvector.
8.3 The Potential Four-Vector
Recall that: for any potentials A, with the definitions
B=VxA,
B=-vo-A,
twoof Maxvell’s equations, namely the pair
‘Thus
which may be written as
V-B=0,
VxE+B=0,
are automatically satisfied, Furthermore, the gauge transformations
ASA+W
B5S-%
to the vector and scalar potentials leave the fields E and B unaltered - the latter are are gauge invariant,
‘Under a gauge transformation, the quantity
1;
Vatae
transforms as
Lol=L-Oy
using the D’Alembertian 0 = aff — V2 (since
D=1fOa0s = 00a,
2it is a Lorentz, scalar), By suitable choice of y it is always passible to impose the Lorentz, gauge
a
L=V-At+se
This condition is indeed Lorentz invariant: if tre in one frame, it remains satisfied in any’ others
‘Whatever the choice of gauge, as we have seen, the pair of Maxwell's equations which do not involve
the souree terms are automatically satisfied, We now tum to the other pair of equations, namely
V-B= p09
Vx B= jd +cab).
Whitten in ters of the potentials, and making use of the vector identity
Vx (Vx A)=VV-A)— VA,
these became 1
DA+VV-A+ 6) = od
and
1
1a lg
O28 LV A+ SB) = pcp.
Recalling that: we have defined
and suppase that we define
what, we have is
OAM 1PM = pg.
And since we know that j# is a fourvector, this suggests that A! is likewise a four-vector, in which case
Lis indeed a Lorentz. scalar, and the fidd equations satisfied by the potentials AM are seen to be Lorentz,
covariant.
Note that if we impose the Lorentz gang condition, in which case L = 0, the equations simplify to
OA" = pois O,A
84 The Ficld-Strength Tensor
We have expressed the electric and magnetic fields in terms of the potentials, and we know how the
potentials change under a Lorentz. transformation,
AP AM = MAY,
From this follows the transformation laws for the electric and magnetic fidds. To obtain them mest simply,
note that the fidd components are all of them of the form
PO SPP,
whore 0 = 103 => P = a; 0! = -di, etc. For example, BY = GE - GY = PA — PA) and
ibe = 128-128 = 4-91.49), Thus the cmponents of the tensor F#° arrange as a matrix are
1-145 =
0 -EY/c -E/e -E/c
ve-(B oN al Be }.
Bie Be 0 BP
E:/e -BY Bt 0
3This second-rank contravariant antisymmettical tensor is the field-strength tensor; because of its an-
tisymmetry F°? = —P*, it has only six independent campanents. We can of course lower its indices to
obtain Fas = MoutavF™, with
0 Efe Be Efe
[Fase (3 s oS )
Bje Be 0 BF
-E/c -BY BF 0
For any choice of potentials Ay, the fields Fy = OyAy ~ 8,4, automatically satisfy
pF + AFrr + Pry = 0,
and expressed in terns of the electric and] magnetic fields, these equations are once again the homogeneous
pair of Maxwell's equations, ie., those which do noe involve the sources.
As for the inhomogenoous pair of Maxwells equations, these are obtained from.
are
Returning for a moment to the force equation, which we had written as
£(?)-«(nt Yan)
wwe now find that a
Sp aq FU! — FU? FU)
=G FU + FP + FMR)
FU,
and for example
Bop = g(-FU 4 UP) UF)
= FU + F702 + F803)
= oFYUs,
So as advertised,
Dip =a *Up.
‘To spell out what precisely és the Lorentz transformation law for the electromagnetic fidd, we recall
that the transformation law for any contravariant tensor such as F#” is given by
FY > Fr = NAY PO
‘where for the standard boost the coefficients AM, may be arranged as the matrix:
carh¢ —sith¢ 0 0
nye {—sinhG ch’ 0 0
Iara le | hs 5 .
1
0 o 0‘So in terms of matrices, we have
FoF =arn,
(AT denoting the matrix transpose of A), Straightforward matrix multiplication then yields
Bitar
EM = BY psh¢ = Bc sinh
BY = B* cohh( + BYesmhc;
BY = BF
BY = BY ccs + = sing
Be = BE cosh—S sinc,
‘These can also be written as 2
E’=(E+ Bx B)- => 66-5)
B o@-lexe)- app.
85 The Dual Fiek-Strength Tensor
Another useful tensor whose components are the electric and magnetic fied-strengths is the dual tensor
defined by 1
"Fag = 3605 F™
which makes use of the Levi-Civita alternating tensor exw:
ao = {21 app isa ve pomntation of 0123
am V9 otherwise
Thus *For
etc., 90 that
Homl* +anF®)=F2
—B*, etc.,and *Fi2 = $(.anF+e129F™) = FS
0 -Be -By =B
rea (& gave =.)
—E*/c,
Bic 0 =BR/c
-P/e Be 0
‘To show that * is in fact a tensor, it is simplest to show that cag is a tensor, and this is done by
consideration of the formula,
O27 Ox Ae? Ox
Soap > Come = Ops De BER HOM
Ox
= at (2B ate
‘The determinant is just [det A]-!, which from the result obtained earlier is +1. Its presence shows that
¢ is in fact a pseudo-tersor, it changes sign under a reflection or a change in the sense of time. Lorentz
transformations like these, which have det A -1 are called tmproper, and we will henceforth exdude them.
For the proper Lorentz transformations, the determinant factor is unity.
Since F and *F are both tensors, the contractions F! Fy, FF and *F!* Fy are Lorentz. scalars
It is easy to evaluate thom; they are rspectivdy 2(B? — E? /2),4B+B/c, and 2(8?/ —B?), In fact these
are the only independent, Lorentz scalars which can be constructed from the fiddestreneths. Note that they
allow a dassification of elctromagneticfidd configurations according as E? > B22, E? = BP2, E? < BPE
which is invariant under Lorentz transformations.86 The Energy-Momentum Tensor
Recall the definition of the Maxwell stress tensor, here denoted Tj in an carlior koture:
ry = Elo +CB.By— 4(B-E+CB-BYiw
‘The divergence of this expression was directly related to the conservation of momentum, with g = (Ex H)
the electromagnetic momentum density. Since Maxwells equations are invariant under the transformations
of special rdativity, and can be written in the manifestly covariant form O!Fjy, = 0, there nast exist a
covariant form of the stress tensor, which includes the expression above. Setting « = 1 for simplicity in the
following, itis easy to guess that this expression is
Br a P Pt Pe PRs.
‘The components of this tensor are
T?= 5 + CB"),
2
T®=-cExB),
AP = Fully + BB) 4E-E+CB-B)bar
Notice that the 7 component:is equal to the energy density £ and the vector defined by the: components
4 —c times the momentum density g. Finally, T°” is mins the Maxcwall tersor above.
It's straightforward to check that the condition
PT =O
follows using the equations of motion O!Fjy This equation expresses in a Lorentz, covariant: manner the
‘conservation of energy ane] momentum. Integrating it over a closed spacelike hypersurface yields the result.
that the energy flow across the surface is balanced by the change of enengy within, and similarity for the
momentum.
Jn the presence of currents due to charged particles, the Maxwell equations become O"Fz, = jlaji The
‘conservation law here becomes
OTw
(This is mostly easily derived using the Lagrangian formulation of dectrodynamics, as we will see later in
the course.) The components of this conservation law are
ae =
GtVP=T-B
Me opr =~ — (Ix Be
‘The first equation expresses the conservation of energy. The seoand expresses conservation of momentum,
‘The sources are described! by the four-vector j#* = (¢p,J) and P= Bx HL The four-vetor f? = F?jq is
the Lorentz force density, with components
p= a +E,pE+JxB).
Note that this discussion has assumed that the charged particles do not radiate or have selnteraction effects
due to the electromagnetic fie. In general (eg at relativistic velocities ar energies), this assumption cannot.
be made. A full relativistic treatment of the interaction of particks and fields requires quantum field thearys,
for dectrons and the electromagnetic fidd this is the theery of quantum electrodynamics, or QED for short.
‘Some discussion of this wil be given at the end of the couse,
6