For a given trajectory of motion of a physical system the action,
S, is a suitable covariant scalar integral along the trajectory,
The variational principle is then formulated as follows: that trajectory,
on which the action has an extremum, is the solution. Or: the solution
is the stationary trajectory of the action. Or: the variation of the
action vanishes on the solution,
δS=0.
(1)
Example: a free body
The action of a free body with mass m is given by the following
covariant integral along the body's trajectory,
S=−m
⌠ ⌡
ds .
(2)
To calculate the variation of the action we
first vary the expression ds2=gabdxadxb,
2dsδds=δgabdxadxb+2gabδdxadxb
(3)
which gives the variation of ds,
δds=
1
2
δgabuaubds+dδxaua.
(4)
The second term should be integrated by parts using
dδxaua = d(δxaua) − δxcduc.
(5)
The full differential does not contribute to the variation, and we
finally arrive at1
δS=−m
⌠ ⌡
dsδxc
−
duc
ds
+
1
2
gab,cuaub
=0
(6)
Since the variation δx is arbitrary, it is the expression in
brackets that should be equal zero, which gives the equation of
motion
duc
ds
−
1
2
gab,cuaub=0.
(7)
which is equivalent to the geodesic equation (see the exercise).
Exercises
Derive the Newton's equation of motion
m
d
→
v
dt
=−
→
∇
V
of a non-relativistic body with mass m, coordinates
[(r)\vec], and velocity [(v)\vec] in a potential V([(r)\vec]) using the
action
S=
⌠ ⌡
m
→
v
2
2
−V(
→
r
)
dt.
In the Minkowski space derive the Lorentz force equation
m
dua
ds
=eFabub
for a relativistic body with
mass m and charge e moving in an electromagnetic field
Fab=∂aAb−∂bAa using the action
S=−mc
⌠ ⌡
ds − e
⌠ ⌡
Aadxa.
In the Minkowski space derive the second Maxwell
equation, ∂a∂aAb=4πjb, using the action
S=−
1
8π
⌠ ⌡
∂aAb∂aAbd4x−
⌠ ⌡
Aajad4x.
Show that with the Lorentz condition, ∂aAa=0, it is equivalent to
∂aFab=4πja.