IPHT-t18/002

General Relativity from Scattering Amplitudes

N. E. J. Bjerrum-Bohr,1 Poul H. Damgaard,1 Guido Festuccia,2 Ludovic Planté,3 and Pierre Vanhove4, 5
1
Niels Bohr International Academy and Discovery Center,
The Niels Bohr Institute, Blegdamsvej 17, DK-2100 Copenhagen Ø, Denmark
2
Department of Physics and Astronomy, Theoretical Physics,
Ångströmlaboratoriet, Lägerhyddsvägen 1, Box 516 751 20 Uppsala, Sweden
3
Ministère de l’économie et des finances, Direction générale des entreprises, France
4
Institut de Physique Théorique, Université Paris-Saclay,
CEA, CNRS, F-91191 Gif-sur-Yvette Cedex, France
5
National Research University Higher School of Economics, Russian Federation
(Dated: June 14, 2018)
We outline the program to apply modern quantum field theory methods to calculate observables
in classical general relativity through a truncation to classical terms of the multi-graviton two-
body on-shell scattering amplitudes between massive fields. Since only long-distance interactions
arXiv:1806.04920v1 [hep-th] 13 Jun 2018

corresponding to non-analytic pieces need to be included, unitarity cuts provide substantial simpli-
fications for both post-Newtonian and post-Minkowskian expansions. We illustrate this quantum
field theoretic approach to classical general relativity by computing the interaction potentials to
second order in the post-Newtonian expansion, as well as the scattering functions for two massive
objects to second order in the post-Minkowskian expansion. We also derive an all-order exact result
for gravitational light-by-light scattering.

PACS numbers: 04.60.-m, 04.62.+v, 04.80.Cc

Today it is universally accepted that classical general
relativity can be understood as the ~ → 0 limit of a quan-
tum mechanical path integral with an action that, min-
imally, includes the Einstein-Hilbert term. It describes
gravitational interactions in terms of exchanges and in-
teractions of spin-2 gravitons with themselves (and with
matter) [1, 2]. The language of effective field theory
encompasses this viewpoint, and it shows that a large-
distance quantum field theoretic description of gravity
is well defined order by order in a derivative expan-
sion [3, 4]. Quantum mechanics thus teaches us that
we should expect classical general relativity to be aug-
mented by higher-derivative terms. More remarkably,
what would ordinarily be a quantum mechanical loop ex-
pansion contains pieces at arbitrarily high order that are
entirely classical [5, 6]. A subtle cancellation of factors
of ~ is at work here. This leads to the radical conclusion
that one can define classical general relativity perturba-
tively through the loop expansion. Then ~ plays a role
only at intermediary steps, a dimensionful regulator that
is unrelated to the classical physics the path integral de-
scribes.
For the loop expansion, central tools have been the uni-
tarity methods [7] which reproduce those parts of loop
amplitudes that are “cut constructable”, i.e. all non-
analytic terms of the amplitudes. This amounts to an
enormous simplification, and most of today’s amplitude
computations for the Standard Model of particle physics
would not have been possible without this method. In
classical gravity, the long-distance terms we seek are pre-
cisely of such non-analytic
p kind, being functions of the di-
mensionless ratio m/ −q 2 , where m is a massive probe,
and q µ describes a suitably defined momentum trans-
fer [4]. This leads to the proposal that these modern
methods be used to compute post-Newtonian and post-
Minkowskian perturbation theory of general relativity for
astrophysical processes such as binary mergers. This has
acquired new urgency due to the recent observations of
gravitational waves emitted during such inspirals.
While the framework for classical general relativity as
described above would involve all possible interaction
terms in the Lagrangian, ordered according to a deriva-
tive expansion, one can always choose to retain only the
Einstein-Hilbert action. Quantum mechanically this is
inconsistent, but for the purpose of extracting only clas-
sical results from that action, it is a perfectly valid trun-
cation. This scheme relies on a separation of the long-
distance (infrared) and short-distance (ultraviolet) con-
tributions in the scattering amplitudes in quantum field
theory. We will follow that strategy here, but one may
apply the same amplitude methods to actions that con-
tain, already at the classical level, higher-derivative terms
as well. In the future, this may be used to put better ob-
servational bounds on such new couplings.
In [8] Damour proposed a new approach for convert-
ing classical scattering amplitudes into the effective-one-
body Hamiltonian of two gravitationally interacting bod-
ies. In this work we take a different route and we show
how scattering amplitude methods, which build on the
probabilistic nature of quantum mechanics, may be used
to derive classical results in gravity. We show how tree-
level massless emission from massive classical sources
arises from quantum multiloop amplitudes, thus provid-
ing an all-order argument extending the original observa-
tions in [6]. We apply this method to derive the scattering
angle between two masses to second post-Minkowskian
 2

order using the eikonal method. Below we show how this allows us to recover the first
We start with the Einstein-Hilbert action coupled to a post-Newtonian correction to the Schwarzschild metric
scalar field φ from quantum loops. We now explain how the classical
part emerges from higher-loop triangle graphs, starting
√ m2 2
Z  
1 1
S = d4 x −g R+ g µν ∂µ φ∂ν φ− φ . (1) with two-loop triangles
16πG 2 2
Here R is the curvature and gµν is the metric, de-
fined as the sum of a flat Minkowski √ component ηµν
and a perturbation κhµν with κ ≡ 32πG. It is cou- I..(1) (p1 , q) = (6)
pled to the scalar stress-energy tensor Tµν ≡ ∂µ φ ∂ν φ −
ηµν ρ 2 2


2 ∂ φ∂ ρ φ − m φ .
Scalar triangle integrals [9] are what reduces the one-
loop two-graviton scattering amplitude to classical gen-
eral relativity [4, 10] in four dimensions. For the long- + +
distance contributions these are the integrals that pro-
duce the tree-like structures one intuitively associates
with classical general relativity. To see this, consider
first the triangle integral of one massive and two mass- In the large mass limit |~`i |  m1 for i = 1, 2, 3 and
less propagators, approximating (`i + p1 )2 − m21 ' 2`i · p1 ' 2m1 `0i the
integral reduces in that limit to

I. (p1 , q) = 3
d4 `i
Z Y
1
I..(1) (p1 , q) = − 4 `2 + i
i=1
(2π) i
d4 `
Z
1 1 1 3 (3)
P3 ~
(2π) δ ( i=1 `i + ~q)
= , (2) × × 2πδ(`01 + `02 + `03 )
(2π) ` + i (` + q) + i (` + p1 ) − m21 + i
4 2 2 2
(`1 + q)2 + i
1 1 1 1
with p1 = (E, ~q/2),p p2 = (E, −~q/2) and q ≡ p1 − p2 =

× +
(0, ~q) and E = m21 + ~q2 /4, and we work with the 2m1 `01 + i 2m1 `02 − i 2m1 `03 + i 2m1 `01 − i
mostly negative metric (+ − −−). The curly lines are 1 1 
for massless fields and the left solid line is for a particle + . (7)
2m1 `03 + i 2m1 `02 − i
of incoming momentum p1 , outgoing momentum p2 and
mass p21 = p22 = m21 .
We note that
In the large mass approximation we focus on the region
|~`|  m1 we have (` + p1 )2 − m21 = `2 + 2` · p1 ' 2m1 `0
therefore the integral reduces in that limit to
 1 1
δ(`01 + `02 + `03 )
2m1 `01 + i 2m1 `02 − i
d4 `
Z
1 1 1 1
. (3) 1 1 1 1 
2m1 (2π)4 `2 + i (` + q)2 + i `0 + i + +
2m1 `03 + i 2m1 `01 − i 2m1 `03 + i 2m1 `02 − i
We perform the `0 integral by closing the contour of in- = 0 + O() , (8)
tegration in the upper half-plane to get

d3 ~` i 1 so that only the `0 residue at 2m1 `0 = ±i contributes,

Z
1 i
=− . (4)
~
|`|m (2π) 3 4m ~2 ~
` (` + q) 2 q|
32m|~ giving

This result can be obtained by performing the large mass Z 3~ 3~

expansion of the exact expression for the triangle integral i d `1 d `2 1 1 1 1
I..(1) (p1 , q) = .
as shown in the Appendix. 4m21 (2π)3(2π)3 ~`21 ~`22 (~`1 + ~`2 + ~q)2 (~`1 + ~q)2
In (4) we recognize the three-dimensional integral of (9)
two static sources localized at different positions, repre- We now consider the large mass expansion of the graph
sented as shaded blobs, and emitting massless fields

d3 ~` 1
Z
1
3
←→ (5) I..(2) (p1 , q) = (10)
~ ~
(2π) ` (` + q)2
2
 3

which to leading order reads where we used the result of the previous section to
evaluate the triangle integral. This reproduces the
1
Z Y3  4
d `i 1
 classical first post-Newtonian contribution to the 00-
I..(d) (p1 , q) = − component of the Schwarzschild metric evaluated in [13].
3 (2π)4 `2i + i
i=1 It also immediately shows how to relate a conventional
× (2π)3 δ (3) (`1 + `2 + `3 + q) Feynman-diagram evaluation with the computation of
 1 1 Duff [14] who derived such tree-like structures from clas-
× 2πδ(`01 + `02 + `03 ) sical sources.
2m1 `01 + i 2m1 `02 − i
1 1 1 1 
+ + , SCALAR INTERACTION POTENTIALS
2m1 `03 + i 2m1 `01 − i 2m1 `03 + i 2m1 `02 − i
(11)
For the classical terms we need only the graviton cuts,
and evaluates to and instead of computing classes of diagrams, we apply
the unitarity method directly to get the on-shell scatter-
d3 ~`1 d3 ~`2 1 1 ing amplitudes [10, 15]. We first consider the scattering
Z
1 1
I..(2) (p1 , q) = . of two scalars of masses m1 and m2 , respectively. At
12m21
(2π)3(2π)3 ~`21 ~`22 (~`1 + ~`2 + ~q)2
(12) one-loop order this entails a two-graviton cut of a mas-
The expressions (9) and (12) are precisely the coupling sive scalar four-point amplitude. We have shown that
of three static sources to a massless tree amplitude classical terms arise from topologies with loops solely en-
tering as triangles that include the massive states. When
we glue together the two on-shell scattering amplitudes,
we thus discard all terms that do not correspond to such
I..(1) (p1 , q), I..(2) (p1 , q) ↔ topologies. Rational terms are not needed, as they cor-
respond to ultra-local terms of no relevance for the long-
distance interaction potentials.
(13) We first recall the classical tree-level result from the
A generalisation of the identity (8) implies that the sum one-graviton exchange
of all the permutation of n massless propagators con-
nected to a massive scalar line results in the coupling of
classical sources to multileg tree amplitudes [11, 12]. The
M1 =
same conclusion applies to massive particles with spin as
we will demonstrate elsewhere.
This analysis applies directly to the computation of an 16πG 2 2
m1 m2 −2(p1 · p4 )2 −(p1 · p4 )q 2 , (16)


=− 2
off-shell quantity such as the metric itself. Consider the q
absorption of a graviton where incoming momenta are p1 and p4 and p21 = p22 =
m21 , p23 = p24 = m22 and the momentum transfer q =
−i d4 ` Pρσ,αβ
Z
Pκλ,γδ p1 − p2 = −p4 + p3 .
hp2 |Tµν |p1 i =
2m1 (2π)4 `2 + i (` + q)2 + i The two-graviton interaction is clearly a one-loop am-
τ1ρσ (p1 − `, p1 )τ1κλ (p1 − `, p2 )τ3 µν αβ,γδ (`, ` − q) plitude that can be constructed using the on-shell uni-
, (14) tarity method [10, 15]. The previous analysis shows that
(` + p1 )2 − m21 + i
the classical piece is contained in the triangle graphs
where τ1 is the vertex for the coupling of one graviton to
a scalar given in [13, eq. (72)], τ3 is the three graviton M2 = + (17)
vertex given in [13, eq. (73)] and Pµν,ρσ is the projection
operator given in [13, eq. (30)]. In the large m limit !
|q|/m  1 projects the integral on the 00-component of 2 c(m1 , m2 )I. (p1 , q) c(m2 , m1 )I. (p4 , −q)
= −i(8πG) 2 + 2 ,
the scalar vertex τ1µν (p1 − `, p1 ) ' iκm21 δ0µ δ0ν . Focusing (q 2 − 4m21) (q 2 − 4m22 )
on the 00 component we have in this limit [12] with for the interaction between two massive scalars

d4 ` c(m1 , m2 ) = (q 2 )5 + (q 2 )4 6p1 · p4 − 10m21

Z  

3 2 3~ 2
hp2 |T00 |p1 i ' 4iπGm31 ~q − `
(2π)4 8 2 + (q 2 )3 12(p1 · p4 )2 − 60m21 p1 · p4 − 2m21 m22 + 30m41


1
−(q 2 )2 120m21 (p1 · p4 )2 − 180m41 p1 · p4 − 20m41 m22 + 20m61


× 2
(` + i)((` + q)2 + i)((` + p1 )2 − m21 + i)
+q 2 360m41 (p1 · p4 )4 − 120m61 p1 · p4 − 4m61 (m21 + 15m22 )


3κ2 m3
= |~q| , (15) + 48m81 m22 − 240m61 (p1 · p4 )2 . (18)
128
 4

At leading order in q 2 , using the result (4) the derive the quantum mechanical cross section is not ap-
two gravitons exchange simplifies to just, in agreement propriate here, even if we keep only the classical part of
with [16, eq (3.26)] and [15, 17], the amplitude. That expression for the cross section is
based on incoming plane waves, and will not not match
6π 2 G2 the corresponding classical cross section beyond the lead-
M2 = (m1 + m2 )(5(p1 · p4 )2 − m21 m22 ) + O(|~q|) .
|~q| ing tree-level term. In fact, even the classical cross sec-
(19) tion is unlikely to be of any interest observationally. So
Note the systematics of this expansion. The Einstein a more meaningful approach is to use the classical scat-
metric is expanded perturbatively, and all physical mo- tering amplitude to compute the scattering angle of two
menta are provided at infinity. Contractions of momenta masses colliding with a given impact parameter b.
are performed with respect to the flat-space Minkowski We use the eikonal approach to derive the relationship
metric only, and no reference is made to space-time co- between small scattering angle θ and impact parameter
ordinates. This is a gauge invariant expression for the b. Generalizing the analysis of ref. [19] (see also [20]) to
classical scattering amplitude that is independent of co- the case of two scalars of masses m1 and m2 , we focus
ordinate choices (and gauge choices). To derive a classical on the high-energy regime s, t large and t/s small. Note
non-relativistic potential, we need to choose coordinates: that in addition to expanding in G, we are also expand-
We Fourier transform the gauge invariant momentum- ing the full result (17) in q 2 , and truncating already at
space scattering amplitude. This introduces coordinate next-to-leading order. We go to the center of mass frame
dependence even in theories such as Quantum Electrody- and define p2 ≡ |p~1 |2 = |p~4 |2 . The impact parameter
namics. Moreover, just as in Quantum Electrodynamics, is defined by a two-dimensional vector ~b orthogonal to
we must also be careful in keeping sub-leading terms of p~1 = −~p4 , with b ≡ |~b|. In the eikonal limit we find the
this Fourier transform and thus expand in q 0 consistently. exponentiated relationship between the scattering ampli-
This forces us to keep velocity-dependent terms in the tude
energy that are of the same order as the naively defined Z
~
static potential. One easily checks that the overall sign M (~b) ≡ d2 ~qe−i~q·b M (~q) , (21)
of the amplitudes in (16) and (19) are precisely the ones
required for an attractive force. and scattering function χ(b) to be
The result of this procedure has been well documented ~
elsewhere, starting with the pioneering observations of M (~b) = 4p(E1 + E2 )(eiχ(b) − 1) . (22)
Iwasaki [5], and later reproduced in different coordinates In order to compare with the first computation of post-
in [15, 18]. Although we are unable to reproduce the Minkowskian scattering to order G2 [21], we introduce
individual contributions in [5, eqs. (A.1.4)–(A.1.6)] our new kinematical variables M 2 ≡ s, M̂ 2 ≡ M 2 − m21 −
final result for the interaction energy is to this order m22 . We go to the center of mass frame where p2 =
(M̂ 4 − 4m21 m22 )/4M 2 . In terms of the scattering angle
p~21 p~2 p~4 p~4 θ we have t ≡ q 2q = [(M̂ 4 − 4m21 m22 ) sin2 (θ/2)]/M 2 , and
H= + 4 − 13 − 43 (20)
2m1 2m2 8m1 8m2
4p(E1 + E2 ) = 2 M̂ 4 − 4m21 m22 . Keeping, consistently,
Gm1 m2 G2 m1 m2 (m1 + m2 )
− − only the leading order in q 2 of the one-loop amplitude
r 2r3
 2 (21), we obtain
p24

Gm1 m2 3~ p1 3~ 7~p1 · p~4 p1 · ~r)(~
(~ p4 · ~r)
− + 2 − − , −2M ∂
2r m21 m2 m1 m2 m1 m2 r2 2 sin(θ/2) = q (χ1 (b) + χ2 (b)) , (23)
∂b
M̂ 4 − 4m21 m22
which precisely leads to the celebrated Einstein-Infeld-
Hoffmann equations of motion. It is crucial to correctly where χ1 (b) and χ2 (b) are the tree-level and one-loop
perform the subtraction of the iterated tree-level Born scattering functions given respectively by the Fourier
term in order to achieve this. transform of the scattering amplitudes
d2 ~q −i~q·~b
Z
1
χi (b) = q e Mi (~q) . (24)
THE POST-MINKOWSKIAN EXPANSION (2π)2
2 M̂ 4 − 4m21 m22

The scattering problem of general relativity can be At the leading order in q 2 the tree-level and one-loop
treated in a fully relativistic manner, without a truncated amplitudes in (16) and (19) read
expansion in velocities. To this end, we consider here the 8πG
full relativistic scattering amplitude and expand in New- M1 (~q) = (M̂ 4 − 2m21 m22 ) ,
~q2
ton’s constant G only. For the scalar-scalar case we thus
return to the complete classical one-loop result (17). The 3π 2 G2
M2 (~q) = (m1 + m2 )(5M̂ 4 − 4m21 m22 ) . (25)
conventional Born-expansion expression that is used to 2|~q|
 5

Only the triangle contribution contribute to the one-loop

loop scattering function because the contributions from
the boxes and cross-boxes contributed to the exponenti-
ation of the tree-level amplitude [19, 22]. The scattering Mγγ = + +
functions then read
(28)
M̂ 4 − 2m21 m22

1
 
b
 2tr(f1 f2 f3 f4 ) + 2tr(f1 f3 f4 f2 ) − tr(f1 f2 )tr(f3 f4 )
χ1 (b) = 2G q − log − γE , − 8πG
d−2 2 (p1 − p2 )2
M̂ 4 − 4m21 m22
2tr(f1 f4 f3 f2 ) + 2tr(f1 f3 f2 f4 ) − tr(f1 f4 )tr(f2 f3 )
− 8πG
3πG2 m1 + m2 (p1 + p4 )2
χ2 (b) = q (5M̂ 4 − 4m21 m22 ) ,
4 2 2 b 2tr(f1 f3 f4 f2 ) + 2tr(f1 f3 f2 f4 ) − tr(f1 f3 )tr(f2 f4 )
8 M̂ − 4m1 m2 − 8πG ,
(26) (p1 − p3 )2

where the traces are evaluated over the Lorentz indices

with the scattering angle to this order and fiµν = µi pνi −νi pµi are the field-strength of the photon
fields. When considering polarised photons, it is imme-
diate to check that this amplitude is non-vanishing only
4GM  M̂ 4 − 2m21 m22
 
θ for scattering of photons of opposite polarisation (helic-
2 sin =
2 b M̂ 4 − 4m21 m22 ity) as no force is expected between photons of the same
polarisation (heliticity) [24]. This is a result that one
3π G(m1 + m2 ) 5M̂ 4 − 4m21 m22 
+ . (27) can also derive by solving to leading order the Einstein
16 b M̂ 4 − 4m21 m22 equations of classical general relativity [25, §114].

This result agrees with the expression found by Westp-
fahl [21] who explicitly solved the Einstein equations to
this order in G and in the same limit of small scattering
angle. We find the present approach to be superior in
efficiency, and very easily generalizable to higher orders
in G.
Taking the massless limit m2 = 0 and approximating
2 sin(θ/2) ' θ, we recover the classical bending angle of
G2 m21
light θ = 4Gmb
1
+ 15π
4 b2 , including its first non-trivial
correction in G, in agreement with [23, §101]. We have
additionally computed the full expression for the classical
part of the scalar-fermion (spin 1/2) amplitude up to and
including one-loop order, but do not display the results
here for lack of space. We stress that the small-angle
scattering formula derived above is based on only a small
amount of the information contained in the full one-loop
scattering amplitude (17).
LIGHT-BY-LIGHT SCATTERING IN GENERAL
RELATIVITY
Photon-photon scattering is particularly interesting, as
our analysis will show how to derive an exact result in
general relativity. As explained above, classical contri-
butions from loop diagrams require the presence of mas-
sive triangles in the loops. For photon-photon scattering
there are no such contributions to any order in the ex-
pansion, and we conclude that photon-photon scattering
in general relativity is tree-level exact
CONCLUSION
We have explicitly shown how loops of the Feynman di-
agram expansion become equal to the tree-like structures
coupled to classical sources thus demystifying the appear-
ance of loop diagrams in classical gravity, and, at the
same time, linking the source-based method directly to
conventional Feynman diagrams. Interestingly, the man-
ner in which the `0 -integrations conspire to leave tree-like
structures from loops of triangle graphs also forms the
precise bridge to classical general relativity computations
based on the world-line formulation (see, e.g., [26–31]).
Enormous simplifications occur when computing what
corresponds to on-shell quantities, based on the unitar-
ity method [7]. Non-analytic terms [4] involving powers
p
of m/ −q 2 produce the long-distance classical contribu-
tions from the loops. By the rules of unitarity cuts, we
can reconstruct these non-analytic pieces by gluing tree-
level amplitudes together while summing over physical
states of the graviton legs only [10, 15–17, 22].
Scalar interaction potentials form the backbone of
gravitational wave computations for binary mergers. The
fact that the unitarity method provides these results
straightforwardly provides hope that this is the begin-
ning of a new approach to both post-Newtonian and
post-Minkowskian calculations in general relativity, in-
cluding those relevant for the physics of gravitational
waves. Since the method applies to the general effective
field theory of gravity, this opens up a way to constrain
terms beyond the Einstein-Hilbert action that may affect
the observational signal of gravitational waves.
 6

ACKNOWLEDGMENTS [4] J. F. Donoghue, Phys. Rev. Lett. 72, 2996 (1994)

We thank John-Joseph Carrasco and Thibault Damour [gr-qc/9310024]. Phys. Rev. D 50, 3874 (1994) [gr-
for useful comments on the manuscript. N.E.J.B and qc/9405057].
P.H.D. are supported in part by the Danish National [5] Y. Iwasaki, Prog. Theor. Phys. 46, 1587 (1971).
[6] B. R. Holstein and J. F. Donoghue, Phys. Rev. Lett. 93,
Research Foundation (DNRF91). P.V. is partially sup- 201602 (2004).
ported by “Amplitudes” ANR-17- CE31-0001-01, and [7] Z. Bern, L. J. Dixon, D. C. Dunbar and D. A. Kosower,
Laboratory of Mirror Symmetry NRU HSE, RF Govern- Nucl. Phys. B 425, 217 (1994) [hep-ph/9403226]. Nucl.
ment grant, ag. N◦ 14.641.31.0001. G.F. are supported Phys. B 435, 59 (1995) [hep-ph/9409265].
by the ERC STG grant 639220. [8] T. Damour, Phys. Rev. D 94, no. 10, 104015 (2016)
[arXiv:1609.00354 [gr-qc]]. Phys. Rev. D 97, no. 4,
Appendix: The finite massive triangle integral 044038 (2018) [arXiv:1710.10599 [gr-qc]].
[9] Z. Bern, L. J. Dixon and D. A. Kosower, Phys. Lett. B
We evaluate the ultraviolet and infrared finite triangle 302 (1993) 299 Erratum: [Phys. Lett. B 318 (1993) 649]
integral for p2 > 0 and q 2 < 0 [hep-ph/9212308].
Z 4 [10] N. E. J. Bjerrum-Bohr, J. F. Donoghue and P. Vanhove,
d ` 1 JHEP 1402, 111 (2014) [arXiv:1309.0804 [hep-th]].
I. (p, q) = .
(2π)4 (`2 +i)((`+q)2 +i)((`+p)2 −p2 +i) [11] L. Planté, PhD thesis (2016), Université Paris-Saclay
(29) [12] J. Galusha, MSc thesis, Niels Bohr Institute, Copenhagen
The parametric representation of this integral is [32] University (2015).
Z [13] N. E. J. Bjerrum-Bohr, J. F. Donoghue and B. R. Hol-
i dx1 dx2 stein, Phys. Rev. D 68 (2003) 084005.
I. (p1 , q) = . (30) [14] M. J. Duff, Phys. Rev. D 7, 2317 (1973).
16π 2 xx1 ≥0 (1 + x 1 + x2 )(q 2 x1 x2 − p2 )
≥0 2 [15] D. Neill and I. Z. Rothstein, Nucl. Phys. B 877, 177
(2013) [arXiv:1304.7263 [hep-th]].
Integrating over x1 leads to [16] A. Guevara, arXiv:1706.02314 [hep-th].
 2 [17] F. Cachazo and A. Guevara, arXiv:1705.10262 [hep-th];
Z ∞ log(1 + x2 ) + log(x2 ) + log − pq 2 [18] B. R. Holstein and A. Ross, arXiv:0802.0716 [hep-ph].
I. (p1 , q) = i dx2 . [19] R. Akhoury, R. Saotome and G. Sterman,
0 16π 2 (p2 + q 2 x2 (1 + x2 ))
arXiv:1308.5204 [hep-th].
(31)
2 2 [20] D. N. Kabat and M. Ortiz, Nucl. Phys. B 388, 570 (1992)
We are interested
p in the case q /p < 0, and
p we setting [hep-th/9203082].
ρ+ ≡ − 21 + 21 1 − 4p2 /q 2 and ρ− ≡ − 12 − 12 1 − 4p2 /q 2 . [21] K. Westpfahl, Fortschr. Phys., 33, 417 (1985)
It is immediate to evaluate the integration over x2 [22] N. E. J. Bjerrum-Bohr, J. F. Donoghue, B. R. Holstein,
  L. Planté and P. Vanhove, Phys. Rev. Lett. 114, no. 6,
6ζ(2) + log (−ρ+ ρ− ) log − ρρ−
+ 061301 (2015) [arXiv:1410.7590 [hep-th]]; JHEP 1611,
I. (p1 , q) = i p 117 (2016) [arXiv:1609.07477 [hep-th]].
16π 2 q 2 1 − 4p2 /q 2 [23] L. D. Landau and E. M. Lifschits, “The Classical Theory
    of Fields : Course of Theoretical Physics, Volume 2”,
Li2 2, − ρ1+ − Li2 2, − ρ1− Butterworth-Heinemann; 4 edition (December 31, 1975)
+i p , (32) [24] D. Boccaletti, V. De Sabbata, P. Fortini and C. Gualdi,
16π 2 q 2 1 − 4p2 /q 2 Nuovo Cim. B 60 (1969) 320.
[25] R. C. Tolman, “Relativity Thermodynanics and Cosmol-
where we have introduced the dilogarithm function ogy”, New York : Dover Publications, 1987.
(see [33] for a definition and properties) [26] I. J. Muzinich and S. Vokos, Phys. Rev. D 52, 3472 (1995)
Z z [hep-th/9501083].
dt [27] W. D. Goldberger and I. Z. Rothstein, Phys. Rev. D 73,
Li2 (z) ≡ − log(1 − t) , for z ∈ C . (33)
0 t 104029 (2006) [hep-th/0409156].
[28] J. B. Gilmore and A. Ross, Phys. Rev. D 78, 124021
For small |q 2 /p2 |  1 we find (2008) [arXiv:0810.1328 [gr-qc]].
s [29] S. Foffa and R. Sturani, Phys. Rev. D 84, 044031 (2011)
i q2 [arXiv:1104.1122 [gr-qc]].
I. (p1 , q) = − + O(q 2 /p2 ) . (34) [30] L. Blanchet, Living Rev. Rel. 17, 2 (2014)
32q 2 p2 [arXiv:1310.1528 [gr-qc]].
[31] S. Foffa, P. Mastrolia, R. Sturani and C. Sturm, Phys.
Rev. D 95, no. 10, 104009 (2017) [arXiv:1612.00482 [gr-
qc]].
[32] P. Vanhove, Proc. Symp. Pure Math. 88 (2014) 161
[1] R. P. Feynman, Acta Phys. Polon. 24, 697 (1963). [arXiv:1401.6438 [hep-th]].
[2] B. S. DeWitt, Phys. Rev. 160 (1967), 1113; 162 (1967) [33] D. Zagier, “The dilogarithm function”, In Frontiers in
1195; 162 (1967) 1239. Number Theory, Physics and Geometry II, P. Cartier,
[3] S. Weinberg, in ”General Relativity - An Einstein Cente- B. Julia, P. Moussa, P. Vanhove (eds.), Springer-Verlag,
nary Survey”, S.W. Hawking and W. Israel, Eds., Cam- Berlin-Heidelberg-New York (2006), 3-65
bridge University Press (1979).