MATHEMATICS OF GRAVITATION

PART II, GRAVITATIONAL WAVE DETECTION

BANACH CENTER PUBLICATIONS, VOLUME 41
INSTITUTE OF MATHEMATICS
POLISH ACADEMY OF SCIENCES
WARSZAWA 1997

POST-NEWTONIAN APPROXIMATIONS AND

EQUATIONS OF MOTION OF GENERAL RELATIVITY

G E R H A R D S C H Ä F E R
Max-Planck-Arbeitsgruppe Gravitationstheorie
Friedrich-Schiller-Universität, 07743 Jena, Germany
E-mail: [email protected]

Abstract. A post-Newtonian approximation scheme for general relativity is defined using

the Arnowitt-Deser-Misner formalism. The scheme is applied to perfect fluids and point-mass
systems. The two-body point-mass Hamiltonian is given explicitly up to the post2.5 -Newtonian
order.

1. Introduction. The Newtonian theory of gravity describes the motion of celestial

bodies with remarkably good precision. Only very precise measurements show deviations
from the Newtonian description. The Newtonian dynamics is therefore strongly expected
to define an excellent starting point for an approximation scheme of general relativity,
in situations for which the velocities of the bodies are small compared to the velocity of
light ((v/c)2 << 1). On the other, theoretical side, (3+1)-formalisms of general relativity
with elliptic constraint and hyperbolic evolution equations are closest to the elliptic-type
Newtonian theory. Therefore it seems natural to define post-Newtonian approximation
schemes in those formalisms. In the following we shall choose the Arnowitt-Deser-Misner
(ADM) Hamiltonian formalism of general relativity, [3], as one of the best-known and
fully developed (3+1)-formalisms and define a post-Newtonian approximation scheme in
that context. For post-Newtonian approximation schemes in other settings, e.g. see [8],
[9]; for a critical investigation of the Newtonian limit in the theory of general relativity,
see [15].

2. The ADM Hamiltonian of general relativity. In the ADM canonical formal-

ism of general relativity the Hamiltonian of the total system, H, is the most important
quantity. The Hamiltonian is defined on space-like three-dimensional, asymptotically flat
Cauchy hypersurfaces which in their asymptotic regions are covered by cartesian coordi-
nate systems. The Hamiltonian is conserved in time and it contains the full information

1991 Mathematics Subject Classification: 83C10, 83C25.

The paper is in final form and no version of it will be published elsewhere.

[43]
44 G. SCHÄFER

about the field-plus-matter dynamics. H is given by, [3], [24],

c4
Z I
3 i
H = d x(N h + N hi ) + d2 si (γij,j − γjj,i ), (1)
16πG ∞
where
(16πG)2 ij 16πG 1/2
h= π Gijkl π kl − γ 1/2 (3) R + γ nµ T µν nν (2)
c6 c4
denotes the so-called super Hamiltonian and
hi = −2γik Dj π kj − πi (3)
the so-called super momentum.
The independent gravitational field variables in the Hamiltonian are the metric com-
ponents of the space-like hypersurfaces, γij := gij = gji (xi , with i = 1, 2, 3, are spatial
coordinates in the surfaces and x0 = ct is labelling the surfaces, c denotes the velocity of
light and t is a time parameter), and their canonical conjugate momenta, π ij . The lapse
and shift functions, N := (−g 00 )−1/2 and N i := γ ij g0j , respectively, are Lagrangian mul-
tipliers. γ, (3) R, γ ij , and Dj denote, respectively, the determinant of the 3-metric, the
curvature scalar of a hypersurface, the inverse 3-metric, and the 3-dimensional covariant
derivative. Gijkl := 12 γ −1/2 (γik γjl + γil γjk − γij γkl ) can be regarded as a contravariant
metric in a 6-dimensional pseudo-Riemannian manifold with signature +4, [14]. G is
the Newtonian gravitational constant. The future-oriented unit normal vector of a hy-
persurface is denoted by nµ . T µν is the symmetric stress-energy tensor of the matter.
The momentum density of the matter is given by πi = −c−1 γ 1/2 nµ Tiµ = c−1 (−g)1/2 Ti0 .
The surface integral in equation (1) is taken over the 2-sphere at infinity in a Cauchy
hypersurface.
If we take as model for the matter a barotropic perfect fluid, the stress-energy tensor
takes the simple form
T µν = (e + p)uµ uν + pg µν , (4)
where e denotes the energy density in the rest frame of the fluid and p its pressure. The
time-like 4-velocity of the fluid, uµ , is normalized according to uµ uν gµν = −1. gµν is the
space-time metric (µ, ν = 0, 1, 2, 3; metric signature: +2). The conservation law for the
baryonic mass number has the form Dµ (%uµ ) = 0, where % denotes the baryonic mass
density and Dµ the 4-dimensional
R 3 ∗ covariant derivative. ∗The conserved baryonic mass
∗ ∗
number, m , reads m = d x% , where, by definition, % = −γ %nµ u = (−g)1/2 %u0
1/2 µ

holds. %∗ fulfils the continuity equation ∂t %∗ + ∂i (%∗ v i ) = 0, where the velocity v i is

defined by v i = cui /u0 . Introducing the specific energy per unit baryonic mass, Π, which
depends on % only, the energy density and the pressure are given by e = %c2 (1+Π/c2 ), p =
%2 dΠ/d%.
If %∗ and πi are chosen as dynamical variables for the fluid, the equations of motion
read,
∂%∗ δH ∗
= −∂i ( % ), (5)
∂t δπi
∂πi δH δH δH
= −∂s ( πi ) − ∂i ( )πs − ∂i ( ∗ )%∗ , (6)
∂t δπs δπs δ%
 POST-NEWTONIAN APPROXIMATIONS 45

δH
where (δ...)/(δ...) denotes the Fréchet derivative and ∂i := ∂/∂xi . δπ i
= v i defines a
∗ i
Legendre transformation to a Lagrangian description with % and v as independent fluid
variables, L = d3 x(πi v i + π ij ∂t γij ) − H.
R

The equations of motion, (5) and (6), can be written also in terms of Poisson brackets,
so-called ‘Lie-Poisson’ (-‘Kirillov-Kostant’) brackets which are intimately related with the
3-dimensional diffeomorphism group of the Cauchy hypersurfaces, e.g. see [1], [17].
The Einstein field equations split into constraint equations,
h = 0 (Hamiltonian constraint), (7)
hi = 0 (momentum constraint), (8)
and evolution equations,
∂γij δH
= ij , (9)
∂t δπ
∂π ij δH
=− . (10)
∂t δγij
The Hamiltonian constraint is a generalization of the theorema egregium by Gauss and the
momentum constraint generalizes the Codazzi theorem of vanishing covariant divergence
of the extrinsic curvature. It is an important property of the constraint equations that
they are conserved in time, i.e. they commute with the Hamiltonian (1).
Within the full theory we may choose the following four coordinate conditions, [2],
corresponding to the Hamiltonian and momentum constraint, respectively, (i.e. the co-
ordinate conditions are, respectively, not invariant against the transformations generated
by the super Hamiltonian and super momentum, e.g. see [30]),
π ij δij = 0, (11)
 4
2πG 16πG T T
γij = 1 + 2 φ δij + hij . (12)
c c4
hTijT , apart from a factor, denotes the part of the 3-metric which is transverse and traceless
in the flat-space metric δij (Kronecker’s δ). The canonical conjugate to (16πG/c4 )hTijT
will be denoted by c−2 pT T ij , i.e. pT T ij = c2 π T T ij , where π T T ij is the transverse-traceless
part of π ij . The fields hT T (:= hTijT ) and pT T (:= pT T ij ) represent the “true” degrees of
freedom of the gravitational field. The fields hT T and pT T are scaled in such a way that,
in the limit of vanishing 1/c, they remain finite if generated by a slowly moving source
(v/c → 0).
The application of the coordinate conditions, (11) and (12), to the field equations, (9)
and (10), result in elliptic-type equations for the lapse and shift functions (notice that
the coordinate conditions (12) are equivalent to 3∂j γij − ∂i γjj = 0). These equations
guarantee that the coordinate conditions hold at any instant of time during the dynamical
evolution, i.e. they commute with the Hamiltonian (1). Therefore, like the constraint
equations which determine through elliptic-type equations φ and the vector part of the
flat-space-traceless π ij (π ij = π T T ij + ∂i τ j + ∂j τ i − δij ∂l τ l + ∇−2 ∂i ∂j ∂l τ l ), the equations
for the lapse and shift functions can be solved for all instants of time.
46 G. SCHÄFER

The evolution equations for hT T and pT T read,

∂hT T c6 δHred
= , (13)
∂t 16πG δpT T
∂pT T c6 δHred
=− . (14)
∂t 16πG δhT T
Hred denotes H if restricted to the space of functions which fulfil the constraint equations
and the coordinate conditions, i.e.
Z I
Hred = −c2 d3 x ∆φ = −c2 d2 si ∂i φ = H[%∗ , π, hT T , pT T ], (15)
∞

(π := πi ).
Now, in the equations of motion for the perfect fluid,
R 3 (5) and (6), HT Tgets Tsubstituted by
T 6
Hred . Notice also the reduced Lagrangian: Lred = d x(πv +16πGp ∂t h /c )−Hred ,
(v := v i ).
For completeness we add the expressions for the
R total linear momentum and angular
momentum of the matter system, d3 xπi and d3 xijk xj πk , respectively, where ijk
R

denotes the Levi-Civita anti-symmetric tensor.

After solving the constraint equations, the metric components gµν are functionals of
the independent variables %∗ , π, hT T , and pT T .

3. Post-Newtonian approximations in the ADM formalism. A series expansion

in powers of 1/c of the metric components and of the equations of motion of the matter
is called a post-Newtonian expansion of the theory.
In terms of the independent field and matter variables we are able to define a post-
Newtonian expansion of the metric tensor, [28],
∞  n
X 1 [n]
c2 g00 (t, x) = −c2 + 2
h00 (x; [%∗ (t), π(t), hT T (t), pT T (t)]), (16)
n=0
c
∞  n
X 1 [n]
c g0i (t, x) = 2
h0i (x; [%∗ (t), π(t), hT T (t), pT T (t)]), (17)
n=1
c
∞  n
X 1 [n]
gij (t, x) = δij + 2
hij (x; [%∗ (t), π(t), hT T (t), pT T (t)]), (18)
n=1
c
[n]
(x := xi ). Here, the hµν are the metric components at the postn -Newtonian order, i.e. n-th
post-Newtonian approximation. The components are determined by solely elliptic-type
[0]
equations in flat space. h00 is minus two times the usual Newtonian potential. The field
variables hT T and pT T enter the metric components at the post2 - and post3 -Newtonian
approximations, respectively. As hT T and pT T vanish for spherically symmetric processes,
the post-Newtonian power expansions will exist in those situations to all orders.
The reduced Hamiltonian takes now the form,
Z ∞  n
X 1 [n]
Hred = c2 d3 x%∗ + 2
Hred [%∗ , π, hT T , pT T ]. (19)
n=0
c
 POST-NEWTONIAN APPROXIMATIONS 47

In the following we shall always suppress the baryonic mass-energy c2 d3 x%∗ in the
R

reduced Hamiltonian as it drops out completely from the dynamics.

[0]
Hred is the Hamiltonian of the Newtonian theory. For our matter model, it simply
reads, [26],
π2 %∗ (t, x)%∗ (t, y)
Z Z Z Z
[0] 1 1
Hred = d3 x%∗ Π(%∗ ) + d3 x ∗ − d3 xd3 y , (20)
2 % 2 |x − y|
where |x − y| denotes the usual Euclidean distance between x and y.
[0] [1]
Hred + c12 Hred is the post1 -Newtonian Hamiltonian. Up to this order no field variables
are involved. At the post2 -Newtonian order, hT T is entering; at the post3 -Newtonian level,
pT T is coming in, i.e. at this level the gravitational field starts evolving in time, [27]. It can
be shown that at the post2 -Newtonian order the field variable hT T may be substituted
in the Hamiltonian completely by matter variables; see the post2 -Newtonian two-body
Hamiltonian below.
A post-Newtonian expansion of the equations of motion, (5) and (6), is immediately
achieved by use of the Hamiltonian (19).
To further discuss the time evolution of the field variable hT T , we develop its field
equation into a second (differential-)order form wave equation,
∞  n
∂2 X 1
(∇2 − 2 2 )hT T (t, x) = 2
TT
D[n] (x; [%∗ (t), π(t), hT T (t), ∂t hT T (t)]). (21)
c ∂t n=0
c
The functional D[0]TT
depends on %∗ and π only. It contains the full information about
the leading order quadrupole radiation. The whole information of the post1 -Newtonian
wave generation is incorporated in D[0] TT
+ c12 D[1]
TT TT
. D[1] depends on %∗ and π, and on
TT
h linearly.
The formal solution of equation (21), assuming the condition of no incoming radiation
from light-like past infinity (t + r/c = b, t → −∞, b = constant), uniquely reads,
∞  n Z
1 X 1 d3
TT
h (t, x) = − DT T (y; [%∗ (u), π(u), hT T (u), ∂u hT T (u)]), (22)
4π n=0 c2 |x − y| [n]
where u = t − |x − y|/c is the time at the source point y, represented in terms of the
retarded time t at the field point x. Obviously, a uniform expansion of equation (22) in
powers of 1/c is not feasible in general.
The solution of pT T is obtained by solving the equation (13) for pT T (Legendre trans-
formation for the gravitational field).
The solution (22) exists if its energy (numerical value of the Hamiltonian) is finite.
For this it is sufficient to check that the numerical value of the free-field part of the
Hamiltonian, H rad , is finite. H rad is given by
Z
rad 16πG 1 1
Hred = 4
d3 x[ (∇hT T )2 + 2 (pT T )2 ]. (23)
c 4 c
An approximate treatment of equation (22) has shown that the gravitational field gener-
ated from freely infalling bodies has finite energy, to leading order at least, [25].
The explicit solution (22) can be constructed order by order in the ordering in n.
At any finite order a well-defined approximate solution can be expected if the starting
48 G. SCHÄFER

approximate solution has finite energy. Nothing is known, however, about the convergence
of the series of approximate solutions if the independent variables depend on time. If they
do not depend on time, a convergent series can be expected, e.g. for stationary rotating
bodies. In stationary situations the solution for hT T is fully expressible in powers of
(1/c)2 as the independent variables do not depend on time. In non-stationary situations
one can try to expand the arguments of the independent variables in powers of 1/c,
either in the near zone, if one is interested in the equations of motion for the matter,
or in the far zone, if one is interested in the radiation. In the near zone, for fixed t,
this expansion is an expansion in powers of |x − y|/c, in the far zone, for fixed t − r/c,
it is an expansion in powers of y · n/c (notice: u = t − r/c + y · n/c + O(r−1 )), where
n = x/r, r = |x|). However, already at the post4 -Newtonian order, i.e. expansion of
TT
D[0] to the order c−4 , those power-series expansions break down. If one is interested
in information beyond post3.5 -Newtonian order one has to treat the independent field
varibles in their non-1/c-expanded form. Nevertheless, as the non-analytic terms in 1/c
belong more to the “phases” than to the “amplitudes” of the independent field variables,
it may happen that some final expressions are fully expandable in powers of 1/c if one
chooses the phases correctly, i.e. if the flat-space time parameter of the source, t − rc , is
shifted to the true-space time parameter, t − rc − 2M G r
c3 ln( cα ), where M denotes the total
mass of the system and where the parameter α fixes the coordinate system in the far
zone, e.g. see [7]. Furthermore, to really know that the radiation comes from the source
only, the no-incoming radiation condition has to be checked with the true light-like past
infinity (t + rc + 2M G r
c3 ln( cα ) = b, t → −∞).

4. Energy balance considerations. Let us decompose the Hamiltonian into a part

mat
which depends on matter variables only, but without the baryonic mass-energy, Hred ,a
f ield
part which depends on field variables only, Hred , and an interaction part which vanishes
int
if either the matter or the field variables vanish, Hred . Then we get
mat ∗ int ∗ f ield T T
Hred = Hred [% , π] + Hred [% , π, hT T , pT T ] + Hred [h , pT T ]. (24)
By construction, Hred is conserved in time.
The power supplied by the matter system into the field, or vice versa, is given by (the
mat int
Hamiltonian which evolves the matter system is Hred + Hred ),
mat
d(Hred int
(t) + Hred (t)) dH f ield (t)
= − red . (25)
dt dt
f ield int ∗
(Notice, dt Hred = −∂t Hred [% , π, hT T (t), pT T (t)], which shows the importance of Hred
int

for the exchange of energy.)

The approximation of the balance equation (25) to first post-Newtonian order is given
by, considering dissipative terms only,
1pN
dHred (t) dH rad (t)
= − red , (26)
dt dt
1pN mat(1pN )
where the post1 -Newtonian Hamiltonian Hred is identical with Hred and where
rad rad f ield(1pN )
Hred is given by equation (23). In some sense, Hred could be denoted by Hred .
 POST-NEWTONIAN APPROXIMATIONS 49

int
The dropped Hred is of the order (v/c)4 ; its leading order term reads,
Z
int[4] πi πj 1
Hred = −8πG d3 xhTijT [ ∗ + ∂i U ∗ ∂j U ∗ ], (27)
% 4πG
where U ∗ = d3 y%∗ /|x − y|.
R

If we average the equation (26) over several periods of the emitted wave it turns out
that only the wave zone contributes to the right side of the equation (26) if we postulate
a quasi stationary process which lasts longer than it takes time for the wave to reach the
wave zone. In the wave zone we may then use the homogeneous field equations which
rad
result from Hred , i.e.
∂hT T ∂pT T c2
= 2pT T , = ∇2 hT T . (28)
∂t ∂t 2
Assuming no-incoming radiation conditions, it is easily shown that the following re-
lation holds,
rad I
dHred (t) 8πG
< >t = 5 < (∂t hT T (t))2 >t r2 dΩ, (28)
dt c w.z.
where < ... >t denotes the time average over several periods of the wave. As a result
we find that the enery loss in the matter system appears as radiation power in the wave
zone (w.z.). Notice that through our assumption of quasi-stationarity we were able to
treat the whole radiation emission problem in a small sandwich (with a width of a few
periods of the wave) of space-like hypersurfaces. As in the wave zone the Hamiltonian of
rad
the gravitational field is always given by Hred the generalization of our energy balance
treatment to the full dynamics is straightforward.
For the losses of linear momentum and angular momentum similar considerations hold.
Here, corresponding to equation (25), the time derivative of the momentum constraint
(8) comes in. The explicit wave zone expressions can be found, e.g. in [31].

5. Post2.5 -Newtonian two-body point-mass Hamiltonian. The simplest model

for an astrophysical two-body system are two non-spinning point-like masses interacting
gravitationally. Although point-like objects do not exist in general relativity – black holes
are the most compact objects, and they are extended – the model can be used in ap-
proximate calculations as long as no inconsistencies occur. Calculations have shown that
the property of being “point-like” might already be invalidaded at the post3 -Newtonian
order, [21], latest, however, at post5 -Newtonian order, [8]. It is a remarkable property of
the Einstein theory that up to the post2.5 -Newtonian approximation, as well as at the
post3.5 -Newtonian order, well-separated compact objects can be treated as “point-like”
with only mass parameters entering the equations of their external motion. This property
is sometimes called the “effacement” of the internal structure in the external problem,
[8], [9].
The point-mass model results from the fluid-matter model in putting Π = p = 0 and
setting %∗ (t, x) = Σa ma δ 3 (x−xa (t)) and πi (t, x) = Σa pai δ 3 (x−xa (t)), where ma , xia , and
pai are mass parameter, position vector, and linear momentum of body a, respectively.
After regularization of the gravitational potentials and of the Hamiltonian, respectively
the equations of motion, it turns out that ma are the rest masses of the bodies. Powerful
50 G. SCHÄFER

methods to achieve the regularization are Hadamard’s “partie finie” and Riesz’s analytic
continuation procedure; for applications, e.g. see [25] and [8], respectively; for details,
see the contribution by P. Jaranowski, in these proceedings.
For point masses the equation of motion (5) and (6) translate, respectively, into the
equations
dxia ∂H
= , (30)
dt ∂pai
dpai ∂H
=− i. (31)
dt ∂xa
mat int
In the coordinate system introduced above, the Hamiltonian Hred + Hred , up to the
post -Newtonian approximation, appears as follows, the additive constant (m1 + m2 )c2
2

being dropped, [12],

1 p21 p22
 4
p42
  
2pN Gm1 m2 1 p1
Hred (x, p1 , p2 ) = + − − 2 + 3
2 m1 m2 r 8c m31 m2
G2 (m1 + m2 )m1 m2 p21 p2
  
Gm1 m2 (p1 · p2 )
+ 2 2
+ 2
−12 2 + 22 + 28
2c r 8c r m1 m2 m1 m2
 6 6
G m1 m2 (m1 + m22 + 5m1 m2 )
3 2
 
(n · p1 )(n · p2 ) 1 p1 p2
+4 + + −
m1 m2 16c4 m51 m52 4c4 r3
  4
p4 p2 p2 (p1 · p2 )2

Gm1 m2 p1
+ 5 4 + 24 − 11 12 22 − 2
4
8c r m1 m2 m1 m2 m21 m22
p21 (n · p2 )2 + p22 (n · p1 )2 (p1 · p2 )(n · p1 )(n · p2 )
+5 2 2 − 12
m1 m2 m21 m22
(n · p1 )2 (n · p2 )2 G2 m1 m2 p21 p22
   
−3 + m2 10 2 + 19 2
m21 m22 4c4 r2 m1 m2
p2 p2
  
27(p1 · p2 ) + 6(n · p1 )(n · p2 )
+m1 10 22 + 19 12 − (m1 + m2 ) , (32)
m2 m1 m1 m2
where x = x1 − x2 , r = |x|, and n = x/r hold. For the obtention of the Hamiltonian (32)
the solution of the equation for the field variable hT T was needed in leading order, [22].
We remark that in harmonic coordinates, at the post2 -Newtonian approximation,
an ordinary Hamiltonian is not achieveable. In harmonic coordinates the two-body dy-
namics is most easily formulated on the equations of motion level ([8], [21]). At the
post2 -Newtonian level of approximation the corresponding Lagrangian contains higher
time derivatives of the position variables [8], [11], [16], [10]. This leads to a constraint
Hamiltonian system, e.g. see [13].
The near-zone expansion of hT T to order 1/c yields the part of the Hamiltonian which,
to leading order, is responsible for the back-reaction of the emitted gravitational waves.
It takes the form, [28],
xi xj
 
reac 16πG T T (reac) p1i p1j p2i p2j
Hred (x, pa , t) = − 4 hij (t) + − Gm1 m2 3 , (33)
c 2m1 2m2 2r
 POST-NEWTONIAN APPROXIMATIONS 51

where a = 1, 2 and where hT T (reac) reads

T T (reac) 1 d3
hij (t) = − Qij (t). (34)
20πc dt3
Qij is the usual Newtonian mass-quadrupole tensor for the matter system, Qij = m1 (xi1 xj1
−(1/3)δij |x1 |2 ) + m2 (xi2 xj2 − (1/3)δij |x2 |2 ) . The dissipation property of the reaction
Hamiltonian, equation (33), has its origin in the time-asymmetry of hT T (reac) .
The reaction Hamiltonian adds up with the post2 -Newtonian Hamiltonian, equation
(32), to give the post2.5 -Newtonian two-body Hamiltonian,
2.5pN 2pN reac
Hred (x, pa , t) = Hred (x, pa ) + Hred (x, pa , t), (35)
2.5pN
where a = 1, 2. The Hamiltonian H describes the motion of a two-body point-mass
system in general relativity up to the first radiation reaction level.
Explicit solutions for the dynamics defined by the conservative Hamiltonian (32) are
given in [12], [29], and [32]. The energy and angular momentum losses related with the
dissipative Hamiltonian (33) have been treated in [23] already many years ago. Energy
and angular momentum losses in general binary systems are known up to the order 1/c7 ,
[6], [20]. Recently, for circular orbits, energy losses at the orders 1/c9 , [5], and 1/c10 ,
[4], have been obtained. The two-body equations of motion are known in the dissipative
part up to the post3.5 -Newtonian level, [18]; the dissipative n-body Hamiltonian at the
post3.5 -Newtonian level has been derived only quite recently, [19].

References

[1] V. I. A r n o l d, Sur la géometrie differentielle des groupes de Lie de dimension infinie et

ses applications a l’hydrodynamique des fluids parfaits, Ann. Inst. Fourier (Grenoble) 16
(1966), 319–361.
[2] R. A r n o w i t t, S. D e s e r, and C. M. M i s n e r, Gravitational-electromagnetic coupling and
the classical self-energy problem, Phys. Rev. 120 (1960), 313–320.
[3] R. A r n o w i t t, S. D e s e r, and C. M. M i s n e r, The dynamics of general relativity, in:
Gravitation: An Introduction to Current Research, L. W i t t e n (ed.), Wiley, New York
1962, 227–265.
[4] L. B l a n c h e t, Energy losses by gravitational radiation in inspiralling compact binaries to
five halves post-Newtonian order, Phys. Rev. D 54 (1996), 1417–1438.
[5] L. B l a n c h e t, T. D a m o u r, B. R. I y e r, C. M. W i l l, and A. G. W i s e m a n, Gravitational-
radiation damping of compact systems to second post-Newtonian order, Phys. Rev. Lett. 74
(1995), 3515–3518.
[6] L. B l a n c h e t and G. S c h ä f e r, Higher–order gravitational radiation losses in binary sys-
tems, Mon. Not. R. astr. Soc. 239 (1989), 845–867; and Erratum: Mon. Not. R. astr. Soc.
242 (1990), 704.
[7] L. B l a n c h e t and G. S c h ä f e r, Gravitational wave tails and binary star systems, Class.
Quantum Grav. 10 (1993), 2699–2721.
[8] T. D a m o u r, Gravitational radiation and the motion of compact objects, in: Gravitational
Radiation, N. Deruelle and T. Piran (eds.), North-Holland Publishing, Amsterdam 1983,
59–144.
52 G. SCHÄFER

[9] T. D a m o u r, The problem of motion in Newtonian and Einsteinian gravity, in: 300 Years
of Gravitation, S. W. Hawking and W. Israel (eds.), Cambridge University Press, Cam-
bridge 1987, 128–198.
[10] T. D a m o u r, L. P. G r i s h c h u k, S. M. K o p e j k i n, and G. S c h ä f e r, Higher-order rela-
tivistic dynamics of binary systems, in: Proc. 5th Marcel Grossmann Meeting on General
Relativity, The University of Western Australia 1988, D. G. Blair and M. J. Buckingham
(eds.), World Scientific, Singapore 1989, 451–459.
[11] T. D a m o u r and G. S c h ä f e r, Lagrangians for n point masses at the second post-
Newtonian approximation to general relativity, General Relativity and Gravitation 17
(1985), 879–905.
[12] T. D a m o u r and G. S c h ä f e r, Higher-order relativistic periastron advances and binary
pulsars, Nuovo Cimento B 101 (1988), 127–176.
[13] T. D a m o u r and G. S c h ä f e r, Redefinition of position variables and the reduction of higher
order lagrangians, Journ. Math. Phys. 32 (1991), 127–134.
[14] B. S. D e W i t t, Quantum theory of gravity. I. The canonical theory, Phys. Rev. 160 (1967),
1113–1148.
[15] J. E h l e r s, Über den Newtonschen Grenzwert der Einsteinschen Gravitationstheorie, in:
Grundlagenprobleme der modernen Physik, J. Nitsch, J. Pfarr, and E.-W. Stachow (eds.),
Bibliographisches Institut, Mannheim 1981, 65–84.
[16] L. P. G r i s h c h u k and S. M. K o p e j k i n, Equations of motion for isolated bodies with
relativistic corrections including the radiation reaction force, in: Relativity in Celestial
Mechanics and Astrometry, J. Kovalevsky and V. A. Brumberg (eds.), Reidel, Dordrecht
1986, 19–34.
[17] D. D. H o l m, Hamiltonian formalism for general-relativistic adiabatic fluids, Physica 17D
(1985), 1–36.
[18] B. R. I y e r and C. M. W i l l, Post-Newtonian gravitational radiation reaction for two-body
systems: Nonspinning bodies, Phys. Rev. D 52 (1995), 6882–6893.
[19] P. J a r a n o w s k i and G. S c h ä f e r, Radiative 3.5 post-Newtonian ADM Hamiltonian for
many-body point-mass systems, Phys. Rev. D (1996), submitted.
[20] W. J u n k e r, G. S c h ä f e r, Binary systems: higher order gravitational radiation damping
and wave emission, Mon. Not. R. astr. Soc. 254 (1992), 146–164.
[21] S. M. K o p e j k i n, General-relativistic equations of binary motion for extended bodies, with
conservative corrections and radiation damping, Sov. Astron. 29 (1985), 516–524.
[22] T. O h t a, H. O k a m u r a, T. K i m u r a, and K. H i i d a, Coordinate condition and higher
order gravitational potential in canonical formalism, Progress of Theoretical Physics 51
(1974), 1598–1612.
[23] P. C. P e t e r s, Gravitational radiation and the motion of two point masses, Phys. Rev.
136 (1964), B1224–B1232.
[24] T. R e g g e and T. T e i t e l b o i m, Role of surface integrals in the Hamiltonian formulation
of general relativity, Annals of Physics 88 (1974), 286–318.
[25] G. S c h ä f e r, The gravitational quadrupole radiation-reaction force and the canonical for-
malism of ADM, Ann. Phys. (N.Y.) 161 (1985), 81–100.
[26] G. S c h ä f e r, Higher-order post-Newtonian hydrodynamics, in: Proc. 5th Marcel Gross-
mann Meeting on General Relativity, The University of Western Australia 1988, D. G.
Blair and M. J. Buckingham (eds.) World Scientific, Singapore, 1986, 467–470.
[27] G. S c h ä f e r, Reduced Hamiltonian formalism for general-relativistic adiabatic fluids and
applications, Astron. Nachrichten 311 (1990), 213–217.
 POST-NEWTONIAN APPROXIMATIONS 53

[28] G. S c h ä f e r, The general relativistic two-body problem. Theory and experiment, in: Sym-
posia Gaussiana, Proc. 2nd Gauss Symposium, Conf. A: Mathematical and Theoretical
Physics, Munich 1993, M. Behara, R. Fritsch, and R. G. Lintz (eds.), Walter de Gruyter,
Berlin 1995, 667–679.
[29] G. S c h ä f e r and N. W e x, Second post-Newtonian motion of compact binaries, Phys. Lett.
A 174 (1993), 196–205; and Erratum: Phys. Lett. A 177 (1993), 461.
[30] K. S u n d e r m e y e r, Constraint Dynamics, Lecture Notes in Physics 169, Springer-Verlag,
Berlin 1982.
[31] K. S. T h o r n e, Multipole expansion of gravitational radiation, Rev. Mod. Phys. 52 (1980),
299–339.
[32] N. W e x and R. R i e t h, The solution of the second post-Newtonian two-body problem,
in: Symposia Gaussiana, Proc. 2nd Gauss Symposium, Conf. A: Mathematical and
Theoretical Physics, Munich 1993, M. Behara, R. Fritsch, and R. G. Lintz (eds.), Walter
de Gruyter, Berlin 1995, 681–693.