Field
Equations & Equations of Motion
(General Relativity)
Velocity
is a vector (tensor) or vector (tensor) field. In familiar notation,
the velocity v is represented by
v
= viei
where
vi represent the components of the velocity, and ei
represent basis (unit) vectors in the selected coordinate system. (As
usual in tensor notation, summation is assumed over all repeated indices
unless otherwise specified.)
Acceleration
is the first time-derivative of velocity, and involves derivatives of
both the vi and the ei :
a
= dv/dt = (dvi/dt)ei + vi(dei/dt)
.
The
second term may be further expanded as
vi(dei/dt)
= vi(ijk
dxj/dt ek )
where
ijk
are the appropriate Christoffel symbols. Substituting, the expression
for acceleration becomes
a
= (dvi/dt + vssij
dxj/dt )ei
(with
suitable change of indices on vs sij).
If,
in a Euclidean space, the components of velocity, vi , are
referred to an inertial (non-accelerated) Cartesian (geodesic) coordinate
system, then the jik
all vanish (i.e., jik
= 0
values of i, j, & k) and the expression for acceleration has the
form
a
= (dvi/dt)ei .
If
a non-Cartesian inertial coordinate system is used, say a polar or a
spherical coordinate system, then the jik
do not all necessarily vanish, and the expression for acceleration may
involve non-zero values of some of the vssijdxj/dt
ei.
[eg.:
In the case of an inertial polar coordinate system, the non-zero values
of vssijdxj/dt
ei simply reflect the fact that the base unit vectors,
and
depend for their direction on their location in the space. Specifically,
with i and j being unit vectors in the Cartesian coordinate
system, the familiar transformations are:
= i sin
+ j cos
, and =
-i cos
+ j sin
,where, for a moving object,
= (t)
and
= (t)
.]
If
the coordinate system to which the vi are referred is non-inertial (i.e.,
accelerating: say it is rotating or linearly accelerating (or both)),
then the jik
do not all vanish, and the expression for acceleration again involves
non-zero values of the terms vssijdxj/dt
ei . In this case, these non-zero values are associated
with the so-called inertial accelerations, i.e., "g's", and the Coriolis
and centrifugal accelerations. These accelerations are independent of
any applied forces, and are due only to the accelerated motion of the
coordinate system.
In
a non-inertial system, the total force, ma, is the vector sum of
1.
The applied force(s), m dvi/dt ei , and
2.
The inertial force(s) m vssijdxj/dt
ei.
Even
if the applied force is zero, we still have the inertial acceleration(s):
a
= vssijdxj/dt
ei .
These
accelerations have the characteristic that if several different test
masses are sequentially placed at a point in the system, they will all
experience the same inertial acceleration (i.e., the inertial
force on the various test masses will be proportional to the masses
only, with the acceleration being a constant). Gravitational acceleration
exhibits identical behavior in this regard; i.e., in classical mechanics,
the gravitational force on a body is proportional to its mass only,
the acceleration being a constant at every point in the field. This
observation leads to the identity of gravitational and inertial mass,
noted by Newton, and used as a motivation toward General Relativity
by Einstein. Let me now present a heuristic approach to the equations
of General Relativity.
One
method of setting up the equations of motion for bodies in classical
circular orbits is to set the gravitational force equal to the centrifugal
force in a coordinate system which is revolving with the body:
mg
= - (mv2/r)u.
(where
u is a unit vector). This expression is equivalent to setting
the total force on the orbiting body equal to zero, and results in the
usual equations of motion for the orbiting body:
f
= mg + (mv2/r)u = 0 .
These
equations may be solved if a field law is given for the gravitational
field g. In classical mechanics, this law is
g
= - (Gm/r2) u .
(where
m is the field-generating mass).
The
same reasoning may be applied to the tensor equations developed above.
We first set the total force equal to zero everywhere in the gravitational
field so that
dvi/dt
= - vssijdxj/dt
.
Using
the relationship vi = dxi/dt , substituting, and
rearranging terms, we then obtain
0
= d2xi/dt2 + sij(dxs/dt)
(dxj/dt) ei .
This
expression is the differential equation for a straight line in Euclidean
space, or a geodesic in a non-Euclidean space. If the classical requirement
that physical space be Euclidean is relaxed, and non-Euclidean spaces
are introduced, the motion of bodies in the gravitational field may
be described by this equation (equation of motion) without recourse
to any gravitational 'force'; i.e., the law of motion becomes: The paths
followed by bodies in a gravitational field are geodesics in a [suitable]
non-Euclidean space [space-time]. The problem becomes one of
properly selecting the values of the sij
(components of the gravitational field). As before, this problem may
be solved by specifying a field law. Einstein chose the expression
Rij
- (1/2)gijR = Tij
where
Rij (= h(ihj
) - j(hhi
) + hhl
ilj
- ilh
jhl
, with m
= /xm
defined for notational convenience) is the contracted Riemann-Christoffel
curvature tensor (Riss j , a.k.a. the
Ricci tensor), R is the associated scalar gi jRi j
, gi j is the fundamental tensor, and Ti j is
the stress-energy tensor. (The expression on the l.h.s. has a vanishing
divergence, satisfying the conservation of mass-energy in the gravitational
field). When these equations are used with the equations of motion18
0
= d2xi/dt2 + sij
(dxs/dt) (dxj/dt) .
the
orbits of bodies and beams of light are accurately described.
To
a first order of approximation, for speeds that are small compared with
the speed of light, and mass densities which are comparable to those
observed in our solar system, General Relativity gives results in agreement
with the equations of Newton. When the field equations Ti j
= 0 (Schwartzchild19) are used (specifying
zero mass-energy density in the space surrounding the sun or any star
- a good approximation for our solar system), the "classical tests"
of the General Theory result: i.e.,
-
The
rotation of Mercury's perihelion at a rate of 42.9 arc-sec per century
(not predictable by classical mechanics unless an intra-Mercurial
planet is assumed [Vulcan]);
-
The
1.75 arc-sec deflection of starlight grazing the sun's surface (measured
by Eddington; originally predicted to be only half as big by classical
mechanics in conjunction with Special Relativity); and
-
The
red shift of starlight traveling outward in the gravitational field
of a star (in agreement with classical mechanics in conjunction
with Special Relativity).
Finally,
for cases of very high velocities (approaching the speed of light) and/or
very large mass-energy densities, the predictions of General Relativity
significantly diverge from those of Newton, but are confirmable by astronomical
observations.
18
In General Relativity, the differential time dt is no longer used because
it is not an invariant (i.e., a scalar; it is the component of
a four-vector and subject to the Lorentz transformation); rather, the
invariant interval ds2= gij dxi dxj
is used instead.
19
Tij=0 Rij
- (1/2)gij R=0. Multiplying the left hand side by gmi
and summing yields Rjm-(1/2)jmR=0.
Next, setting j = m and summing yields R - 2R = 0 or R = 0. But R =
0
Rjm = 0
Rij = 0 ; therefore, Tij = 0
Rij = 0. The last expression is the set solved by Schwartzchild
and is known as Schwartzchild's equation.