Testing strong-field gravity with tidal Love numbers

Vitor Cardoso,1,2 Edgardo Franzin,3 Andrea Maselli,4 Paolo Pani,5,1 Guilherme Raposo1

arXiv:1701.01116v1 [gr-qc] 4 Jan 2017

1
CENTRA, Departamento de Fsica, Instituto Superior Tecnico IST,
Universidade de Lisboa UL, Avenida Rovisco Pais 1, 1049 Lisboa, Portugal
2
Perimeter Institute for Theoretical Physics, 31 Caroline Street North Waterloo, Ontario N2L 2Y5, Canada
3
Dipartimento di Fisica, Universit`
a di Cagliari & Sezione INFN Cagliari,
Cittadella Universitaria, 09042 Monserrato, Italy
4
Theoretical Astrophysics, Eberhard Karls University of Tuebingen, Tuebingen 72076, Germany and
Dipartimento di Fisica, Sapienza Universit`
a di Roma & Sezione INFN Roma1, Piazzale Aldo Moro 5, 00185, Roma, Italy

The tidal Love numbers (TLNs) encode the deformability of a self-gravitating object immersed
in a tidal environment and depend significantly both on the objects internal structure and on the
dynamics of the gravitational field. An intriguing result in classical general relativity is the vanishing
of the TLNs of black holes. We extend this result in three ways, aiming at testing the nature of
compact objects: (i) we compute the TLNs of exotic compact objects, including different families
of boson stars, gravastars, wormholes, and other toy models for quantum corrections at the horizon
scale. In the black-hole limit, we find a universal logarithmic dependence of the TLNs on the location
of the surface; (ii) we compute the TLNs of black holes beyond vacuum general relativity, including
Einstein-Maxwell, Brans-Dicke and Chern-Simons gravity; (iii) We assess the ability of present and
future gravitational-wave detectors to measure the TLNs of these objects, including the first analysis
of TLNs with LISA. Both LIGO, ET and LISA can impose interesting constraints on boson stars,
while LISA is able to probe even extremely compact objects. We argue that the TLNs provide a
smoking gun of new physics at the horizon scale, and that future gravitational-wave measurements
of the TLNs in a binary inspiral provide a novel way to test black holes and general relativity in the
strong-field regime.

CONTENTS

I. Introduction
A. The naturalness problem
B. Quantifying the existence of horizons
C. Executive summary

1
2
2
2

II. Setup: Tidal Love numbers of static objects

III. Tidal perturbations of exotic compact objects

A. Boson stars
B. Models of microscopic corrections at the
horizon scale

5
5
6

IV. Tidal perturbations of BHs beyond GR

A. Scalar-tensor theories
B. Einstein-Maxwell
C. Chern-Simons gravity

9
9
9
10

V. Detectability
A. Model-independent tests with GWs
B. Detectability of ECOs
C. Detectability of BSs
D. Testing GR

12
13
14
16
16

VI. Discussion and extensions

16

Acknowledgments

17

A. TLNs of neutron stars

17

B. Determination of TLNs

17

C. Tidal perturbations of boson stars

1. Background solutions
2. Perturbations and TLNs

19
19
20

D. Tidal perturbations of BH-like ECOs

1. Polar-type TLNs of BH-like ECOs
2. Axial-type TLNs of BH-like ECOs

21
22
23

References

23

I.

INTRODUCTION

Tidal interactions play a fundamental role in astrophysics across a broad range of scales, from stellar objects like ordinary stars and neutron stars (NSs) to large
celestial systems such as galaxies. Several astrophysical
structures (e.g., binaries and tidal tails [1, 2]) are consequences of tidal interactions. Tidal effects can be particularly strong and important in the regime that characterizes compact objects, giving rise to extreme phenomena
such as tidal disruptions.
The deformability of a self-gravitating object immersed
in an external tidal field is measured in terms of its tidal
Love numbers (TLNs) [3, 4]. These leave a detectable imprint in the gravitational-wave (GW) signal emitted by a
neutron-star binary in the late stages of its orbital evolution [57]. So far, a relativistic extension [6, 8, 9] of the
Newtonian theory of tidal deformability has been mostly
motivated by the prospect of measuring the TLNs of NSs
through GW detections and, in turn, understanding the
behavior of matter at supranuclear densities [1016]. The
scope of this paper is to show that tidal effects can also

2
be used to explore more fundamental questions related to
the nature of compact objects and the behavior of gravity
in the strong-field regime.1
An intriguing result in classical general relativity (GR)
is the fact that the TLNs of a black hole (BH) are precisely zero. This property has been originally demonstrated for small tidal deformations of a Schwarzschild
BH [8, 9, 18] and has been recently extended to arbitrarily strong tidal fields [19] and to the spinning case [2022],
at least in the axisymmetric case to quadratic order in the
spin [21] and generically to linear order in the spin [22].
A.

The naturalness problem

The precise cancellation of the TLNs of BHs within

Einsteins theory poses a problem of naturalness in
classical GR [23], one that can be argued to be as puzzling as the strong CP and the hierarchy problem in particle physics, or as the cosmological constant problem.
The resolution of this issue in BH physics could lead to
testable, since they would be encoded in GW data
smoking-gun effects of new physics.
This question can be solved in at least two (related)
ways, which we explore here. If new physics sets in,
for example through unexpectedly large quantum backreaction or changes in the equation of state, BHs might
simply not be formed, avoiding this and other problems (such as the information loss puzzle [24]) altogether. Instead, other objects might be the end product of gravitational collapse. These exotic compact objects (ECOs) include boson stars (BSs) [2528], gravastars [29, 30], wormholes [31], and various toy models describing quantum corrections at the horizon scale, like
superspinars [32], fuzzballs [33], 2-2 holes [34] and others [3537]. ECOs might be formed from the collapse of
exotic fields or by quantum effects at the horizon scale,
and represent the prototypical example of exotic GW
sources [3840] which might be searched for with groundor space-based detectors.
Alternatively, GR might not be a good description of
the geometry close to horizons. BHs other than Kerr
arise in theories beyond GR which are motivated by both
theoretical arguments and by alternative solutions to the
dark matter and the dark energy problems (for recent
reviews on strong-field tests of gravity in the context of
GW astronomy, see Refs. [41, 42]). Arguably, the simplest hairy BHs arise in Einstein-Maxwell theory and are
described by the Reissner-Nordstr
om solution. Although
astrophysical BHs are expected to be electrically neutral [43], Reissner-Nordstr
om BHs can be studied as a
proxy of BHs beyond vacuum GR and could also emerge
naturally in models of minicharged dark matter and dark

A related, independent work dealing with tidal effects for boson stars, conducted simultaneously to ours, is due to appear
soon [17].

photons [44]. In several scalar-tensor theories, BHs are

uniquely described by the Kerr solution, as in GR [45, 46].
However, these theories introduce a scalar degree of freedom (non-minimally) coupled to gravity and the response
of BHs to external perturbations is generically richer [47].
Finally, in quadratic theories of gravity the EinsteinHilbert action is considered as the first term of a possibly
infinite expansion containing all curvature invariants, as
predicted by some scenarios related to string theory and
to loop quantum gravity [42]. To leading order in the
curvature corrections, stationary BHs in these theories
belong to only two families [48, 49], usually dubbed the
Einstein-dilaton-Gauss-Bonnet solution [5052] and the
Chern-Simons solution [53, 54].

B.

Quantifying the existence of horizons

The observational determination of the tidal properties

of compact objects has also a bearing on another fundamental question: do event horizons exist, and how can
we quantify their existence in GW data?
It was recently shown that ultracompact horizonless
geometries are expected to mimic very well the last stages
of coalescence of two BHs, when they merge to form a
single distorted BH, ringing down to its final Kerr geometry [55, 56]. In this scenario, horizonless geometries
would show up as echoes in the gravitational waveforms
at very late times. The exclusion of echoes up to some
instant t after the merger rules out structure in the spacetime down to a region r/r+ 1 exp(t/r+ ), with r+
being the Schwarzschild radius of the spacetime. Thus,
more sensitive detectors will probe regions closer and
closer to the horizon.
The above picture refers to the final, post-merger state.
The understanding of the initial state can start by inferring from the inspiral signal the imprints of the structure
of the inspiralling objects. Putative structures will show
up in the way each of these objects reacts to the gravitational field created by the other, in other words, by their
TLNs. As we will show, the TLNs of all ECOs vanish in
the BH limit, logarithmically. Thus, observational upper
bounds on the TLNs can be converted into constraints
on the compactness of the inspiralling objects.
Throughout this work, we use G = c = 1 units and
denote the Planck length by `P 1.6 1033 cm.

C.

Executive summary

For the busy reader, in this section we summarize our

main results; possible extensions are discussed in Sec. VI.
We focus on spherically symmetric, static background
geometries, and compute the TLNs under the assumption
that the only surviving tide at large distances is gravitational. In this setting, the TLNs can be divided into two
classes according to their parity: an electric- or polartype, and a magnetic or axial-type, and each of these

3
sectors can in turn be expanded into a set of multipoles
labeled by an integer l.
Our main results are summarized in Table I and in
Fig. 1, and are discussed in detail in the rest of the paper.
Table I lists the lowest quadrupolar (l = 2) and octupolar (l = 3) polar and axial TLNs for various models of
ECOs in GR, and for some static BHs in other gravity
theories. Table I also compares the TLNs of these objects with the corresponding ones for a typical NS (cf.
also Table III in Appendix A).
One of our main results is that the TLNs of several
ECOs display a logarithmic dependence in the BH limit,
i.e. when the compactness of the object approaches that
of a BH,
C := M/r0 1/2 ,

(1)

where M and r0 are the mass and the radius of the object.
As shown in Table I, this property holds for wormholes,
thin-shell gravastars, and for a simple toy model of a
static object with a perfectly reflecting surface [35, 37].
It is natural to conjecture that this logarithmic behavior
is model independent and will hold for any ECO whose
exterior spacetime is arbitrarily close to that of a BH
in the r0 2M limit. This mild dependence implies
that even the TLNs of an object with r0 2M `P are
not extremely small, contrarily to what one could expect.
Indeed, we estimate that the dimensionless TLNs defined
in Eq. (3) below are
k2E,B O(103 ) ,

k3E,B O(104 ) ,

(2)

for an ECO with r0 2M `P and in the entire mass

range M [1, 100] M . Note that all TLNs of ultracompact exotic objects listed in Table I have the opposite
sign relative to the neutron-star case (cf. the discussion
in Refs. [58, 59] applied to a particular model).
Furthermore, we show that the TLNs of a charged BH
in Einstein-Maxwell theory and of an uncharged static
BH in Brans-Dicke theory vanish, as in GR, whereas the
TLNs of a BH in Chern-Simons gravity are non-zero, even
though the static BH solution to this theory is described
by the Schwarzschild metric. As expected, the TLNs are
proportional to the coupling constant of the theory so
that any constraint on them can be potentially converted
into a test of gravity.
The accuracy with which GW detectors can estimate
the TLNs of compact objects is shown in Fig. 1 for three
BS models, where the two inspiralling objects are assumed to be equal. In moderately optimistic scenarios,
a GW detection of a compact-binary coalescence with
LIGO can place an upper bound on the TLNs of the
two objects at the level of k2E 10, whereas the future Einstein Telescope (ET) [60] can potentially improve
this constraint by a factor of ten. Interestingly, the future space interferometer LISA [61] has the ability to
set much tighter constraints [cf. also Fig. 7] and to rule
out several candidates of supermassive ECOs. In essence,
both Earth and space-based detectors are able to discriminate even the most compact BSs, by imposing stringent

bounds on their TLNs. By contrast, as we show in Sec. V,

only LISA is able to probe the regime of very compact
ECOs, describing geometries which are microscopic corrections at the horizon scale, for which the compactness
C = 0.48 or higher.
II.

SETUP: TIDAL LOVE NUMBERS OF

STATIC OBJECTS

Let us consider a compact object immersed in a tidal

environment [4]. Following Ref. [8], we define the symmetric and trace-free polar and axial2 tidal multipole moments of order l as Ea1 ...al [(l 2)!]1 hC0a1 0a2 ;a3 ...al i
and Ba1 ...al [ 32 (l + 1)(l 2)!]1 ha1 bc Cabc2 0;a3 ...al i, where
Cabcd is the Weyl tensor, a semicolon denotes a covariant
derivative, abc is the permutation symbol, the angular
brackets denote symmetrization of the indices ai and all
traces are removed. The polar (respectively, axial) moments Ea1 ...al (respectively, Ba1 ...al ) can be decomposed
in a basis of even (respectively, odd) parity spherical harmonics. We denote by E lm and B lm the amplitudes of
the polar and axial components of the external tidal field
with harmonic indices (l, m), where m is the azimuthal
number (|m| l). The structure of the external tidal
field is entirely encoded in the coefficients E lm and B lm
(cf. Ref. [8] for details).
As a result of the external perturbation, the mass and
current multipole moments3 (Ml and Sl , respectively)
of the compact object will be deformed. In linear perturbation theory, these deformations are proportional to
the applied tidal field. In the non-rotating case, mass
(current) multipoles have even (odd) parity, and therefore they only depend on polar (axial) components of the
tidal field.4 Hence, we can define the (polar and axial)
TLNs as [6, 8]
r
1 l(l 1)
4 Ml
E
kl
,
2l+1
2 M
2l + 1 El0
r
(3)
4 Sl
3 l(l 1)
B
kl
,
2 (l + 1)M 2l+1 2l + 1 Bl0
where M is the mass of the object, whereas El0 (respectively, Bl0 ) is the amplitude of the axisymmetric5 component of the polar (respectively, axial) tidal field. The

It is slightly more common to use the distinction electric/magnetic components rather than polar/axial. Since we
shall discuss also electromagnetic fields, we prefer to use the former distinction.
We adopt the Geroch-Hansen definition of multipole moments [62, 63], equivalent [64] to the one by Thorne [65] in asymptotically mass-centered Cartesian coordinates.
This symmetry is broken if the compact object is spinning due
to spin-tidal couplings. In such case, there exists a series of
selection rules that allow to define a wider class of rotational
TLNs [21, 22, 66, 67]. In this paper, we neglect spin effects to
leading order.
We consider only non-spinning objects, hence the spacetime is

|/|[%]

1000

500

100
50

10

20

30

10
10

10

40

1


0.5

0.1

50

10

M [M]

20

30

40

50

|/|[%]

|/|[%]

0.10

0.01

5 10

50 100

M [10 M]

M [M]

FIG. 1. Relative percentage errors on the average tidal deformability for BS-BS binaries observed by AdLIGO (left panel),
ET (middle panel), and LISA (right panel), as a function of the BS mass and for different BS models considered in this work
(for each model, we considered the most compact configuration in the stable branch; see main text for details). For terrestrial
interferometers we assume a prototype binary at d = 100 Mpc, while for LISA the source is located at d = 500 Mpc. The
horizontal dashed line identifies the upper bound / = 1. Roughly speaking, a measurement of the TLNs for systems which
lie below the threshold line would be incompatible with zero and, therefore, the corresponding BSs can be distinguished from
BHs. Here is given by Eq. (72), the two inspiralling objects have the same mass, and / kE /k2E .
2

TABLE I. Tidal Love numbers (TLNs) of some exotic compact objects (ECOs) and BHs in Einstein-Maxwell theory and modified
theories of gravity; details are given in the main text. As a comparison, we provide the order of magnitude of the TLNs for static NSs
with compactness C 0.2 (the precise number depends on the neutron-star equation of state; see Table III for more precise fits). For BSs,
the table provides the lowest value of the corresponding TLNs among different models (cf. Sec. III A) and values of the compactness. In
the polar case, the lowest TLNs correspond to solitonic BSs with compactness C 0.18 or C 0.20 (when the radius is defined as that
containing 99% or 90% of the total mass, respectively). In the axial case, the lowest TLNs correspond to a massive BS with C 0.16 or
C 0.2 (again for the two definitions of the radius, respectively) and in the limit of large quartic coupling. For other ECOs, we provide
expressions for very compact configurations where the surface r0 sits at r0 2M and is parametrized by := r0 /(2M ) 1; the full results
are available online [57]. In the Chern-Simons case, the axial l = 3 TLN is affected by some ambiguity and is denoted by a question mark
[see Sec. IV C for more details].

Tidal Love numbers

k2B

k3B

1300

11

70

41.4

402.8

13.6

211.8

Gravastar

4
5(8+3 log )
8
5(7+3 log )
16
5(236 log 2+9 log )

8
105(7+2 log )
8
35(10+3 log )
16
35(316 log 2+9 log )

16
5(31+12 log )
32
5(25+12 log )
32
5(4312 log 2+18 log )

16
7(209+60 log )
32
7(197+60 log )
32
7(30760 log 2+90 log )

Einstein-Maxwell
Scalar-tensor

0
0

0
0

0
0

0
0

Chern-Simons

1.1 MCS
4

NSs

ECOs

BHs

Boson star
Wormhole
Perfect mirror

k2E

k3E

210

factor M 2l+1 was introduced to make the above quantities dimensionless. It is customary to normalize the
TLNs by powers of the objects radius R rather than
by powers of its mass M . Here we adopted the latter
non-standard choice, since the radius of some ECOs (e.g.

spherically symmetric and, without loss of generality, we can

define the TLNs in the axisymmetric (m = 0) case. Clearly, this
property does not hold when the object is spinning [20, 22, 66].

11.1 MCS
4 ?

BSs) is not a well defined quantity. Thus, our definition is

related to those used by Hinderer, Binnington and Poisson (HBP) [6, 8] through
klE,B
ours =

R
M

2l+1

klE,B
HBP .

(4)

Modified theories of gravity and ECOs typically require the presence of extra fields which are
(non)minimally coupled to the metric tensor. Here we
shall consider some representative example of both scalar

5
and vector fields. A full treatment of this problem would
require allowance for an extra degree of freedom, the external scalar and electromagnetic (EM) applied fields. It
is generically expected that, in astrophysical situations,
the ratio of a putative external (scalar or vector) field
to the ordinary gravitational tidal field should be small.
We will therefore focus only on situations where the only
surviving field at large distances is gravitational.
We expand the metric, the scalar field, and the
Maxwell field in spherical harmonics as presented in Appendix B. Since the background is spherically symmetric, perturbations with different parity and different harmonic index l decouple. In the following we discuss the
polar and axial sector separately; due to the spherical
symmetry of the background, the azimuthal number m
is degenerate and we drop it.
Finally, in order to extract the tidal field and the induced multipole moments from the solution, we have
adopted two (related) techniques. The first one relies on an expansion of the metric at large distances
[cf. Eqs. (B9) and (B10)] in terms of the multipole moments. The second technique relies on the evaluation of
the Riemann tensor in Schwarzschild coordinates, whose
tidal correction is related to the total tidal field in the
local asymptotic rest frame [18]. These two procedures
agree with each other and at least in the case of ECOs
the computation of the TLNs is equivalent to the case of
NSs [6, 8]. On the other hand, computing the TLNs of
BHs in extensions of GR presents some subtleties which
are discussed in Sec. IV.

III.

TIDAL PERTURBATIONS OF EXOTIC

COMPACT OBJECTS

In this section we describe some representative models

of ECOs and discuss their TLNs. Technical details are
given in the appendices.

A.

Potential
V (||2 )

Model

2 ||2

Minimal
Massive
Solitonic

Boson stars

4
||2 +
4 ||

2
2||2
2 ||2 1
2
0

Maximum mass
Mmax /M
 11 
8 10 m eV
 S
2

5 ~ 0.1mGeV
S
h 12 i2 

10
0

500 GeV
mS

TABLE II. Scalar potential and maximum mass for the BS

models considered in this work. In our units, the scalar field
is dimensionless and the potential V has dimensions of an
inverse length squared. The bare mass of the scalar field is
mS := ~. For minimal BSs, the scaling of the maximum
mass is exact. For massive BSs and solitonic BSs, the scaling
of the maximum mass is approximate and holds only when
 2 and when 0  1, respectively.

BSs are complex bosonic configurations held together

by gravity. In the simplest model they are solutions to
the Einstein-Klein-Gordon theory,


Z

R
S = d4 x g
g ab a b V ||2 . (5)
16
BSs have been extensively studied in the past and have
been proposed as BH mimickers and dark matter candidates, see e.g. Refs. [2527, 68, 69].
BSs are typically classified according to the scalar potential in the above action; here we investigate three of
the most common models: minimal BSs [70, 71], massive BSs [72] and solitonic BSs [73]. The corresponding
scalar potential for these models and the maximum mass
for non-spinning solutions are listed in Table II. A more
comprehensive list of BS models can be found in Ref. [26].
Depending on the model, compact BSs with masses
comparable to those of ordinary stars or BHs require a
certain range of the scalar mass mS := ~. For minimal BSs, the maximum mass in Table II is comparable
to the Chandrasekhar limit for NSs only for an ultralight
field with mS . 1011 eV. For massive BSs, the maximum mass is of the same order of the Chandrasekhar
limit if mS 0.1 GeV and the quartic coupling is large,
~ 1 [72]. Finally, solitonic BSs may reach massive
(M & M ) or supermassive (M & 106 M ) configurations even for heavy bosons with mS 500 GeV if the
coupling parameter in their potential is 0 . 1012 or
0 . 1015 , respectively [73]. For massive and solitonic
BSs, the scaling of the maximum mass in Table II is approximate and valid only when  2 and when 0  1,
respectively. In our numerical analysis, we have considered = 104 2 and 0 = 0.05, whereas the mass term
can be rescaled away (cf., e.g., discussion in Ref. [39]).
Even though BSs have a wide range of compactness,
which depends basically on their total mass (cf. Fig. 9
in Appendix C), interactions between BSs typically leads
to a net weight gain, clustering old BSs close to the mass
peak [27], which also coincides with the peak of compactness.
The details of the numerical procedure to compute the
TLNs of a BS are presented in Appendix C. Figure 2
shows the TLNs of the BS models presented above as a
function of the total mass M , the latter being normalized by the total mass Mmax of the corresponding model.
We only show static configurations in the stable branch,
i.e. with a mass smaller than Mmax (cf. discussion in Appendix C). For minimal BSs and for l = 2 polar case, our
results agree with those recently obtained in Ref. [74].
In addition, we also present the results for l = 2 and
l = 3, for both axial and polar TLNs, and for the three
BS models previously discussed.
The behavior of the TLNs of BSs is in qualitative
agreement with that of NSs. For a given BS model with
a given mass, the magnitude of the polar TLN is larger
than that of an axial TLN with the same l. Furthermore, in the Newtonian regime (M 0) the TLNs scale
as klE C (2l+1) and klB C 2l . This scaling is in

FIG. 2. Polar (top panels) and axial (bottom panels) TLNs for minimal, massive and solitonic BSs. Left and right panels refers
to l = 2 and l = 3, respectively. For massive and solitonic BSs we have considered = 104 2 and 0 = 0.05, respectively. With
these values, the maximum mass scales approximately as shown in Table II. Numerical data are available online [57]. These
plots include only stars in the stable branch.

agreement with the neutron-star case (cf. Ref. [8] and

Table III), whereas the sign of the axial TLNs is opposite. Finally, all TLNs are monotonic functions of the
compactness, so that more compact configurations have
smaller tidal deformability. The phenomenological implications of these results are discussed in Sec. V.

B.

Models of microscopic corrections at the

horizon scale

Several phenomenological models of quantum BHs introduce a Planck-scale modification near the horizon.
In this section, we consider three toy models for microscopic corrections at the horizon scale, namely a wormhole [31], a Schwarzschild geometry with a perfectly reflective surface near the horizon [35, 37], and a thinshell gravastar [29].6 These models have some common
features: (i) the exterior spacetime is described by the
Schwarzschild metric; (ii) the interior is either vacuum
or de Sitter and the tidal perturbation equations can be

solved for in closed form; (iii) simple junction or boundary conditions at the radius r0 of the object can be imposed to connect the perturbations in the interior with
those in the exterior. As a result of these properties, the
TLNs of these models can be computed in closed analytical form. As we show, the qualitative features are the
same and especially in the BH limit do not depend
strongly on the details of the models. Below, we present
explicit formulas for the BH limit, expressions for generic
compactness are provided online [57]. The details of the
computation are given in Appendix D.
1.

Wormholes

The simplest models of wormhole solutions consist in

taking two copies of the ordinary Schwarzschild solution
and remove from them the four-dimensional regions described by r1,2 r0 [31]. With this procedure, we obtain
two manifolds whose geodesics terminate at the timelike
hypersurfaces
1,2 {r1,2 = r0 | r0 > 2M } .

Many of these objects are unstable or require exotic matter distributions. We will not be concerned with these issues here.

(6)

The two copies are now glued together by identifying

these two boundaries, 1 = 2 , such that the resulting
spacetime is geodesically complete and comprises of two

7
distinct regions connected by a wormhole with a throat
at r = r0 . Since the wormhole spacetime is composed by
two Schwarzschild metrics, the stress-energy tensor vanishes everywhere except on the throat of the wormhole.
The patching at the throat requires a thin-shell of matter
with surface density and surface pressure

120
0.050

100

0.010
0.005

80

0.001
5. 10-4

1
=
2r0

2M
1
,
r0

1
1 M/r0
p
p=
, (7)
4r0 1 2M/r0

which imply that the weak and the dominant energy conditions are violated, whereas the null and the strong energy conditions are satisfied when r0 < 3M [55]. To
cover the two patches of the spacetime, we use the radial
tortoise coordinate r , which is defined by


dr
2M
= 1
,
(8)
dr
r
where the upper and lower sign refer to the two sides
of the wormhole. Without loss of generality, we can
assume that the tortoise coordinate at the throat is
zero, r (r0 ) = 0, so that one side corresponds to r > 0
whereas the other side corresponds to r < 0.
In Fig. 3, we show the polar and axial TLNs with
l = 2, 3 as functions of := r0 /(2M ) 1. Interestingly,
in this case the TLNs have the opposite sign to those of a
NS. Furthermore, they vanish in the BH limit, i.e. when
r0 2M or 0. The behavior of the TLNs in the BH
limit reads
4
,
5(8 + 3 log )
8
,
k3E
105(7 + 2 log )
16
k2B
,
5(31 + 12 log )
16
k3B
,
7(209 + 60 log )
k2E

(9)
(10)
(11)
(12)


(log )2
get klE,B

where we have omitted subleading terms of O

On the other hand, in the Newtonian limit we

C (2l+1) . Interestingly, while the scaling for polar TLNs

agrees with that of NSs (cf. Table III), that for the axial
TLNs is different.
The logarithmic dependence of the TLNs is very interesting, because it implies that the deviations from zero
(i.e., from the BH case) are relatively large even when
the throat is located just a Planck length away from the
would-be horizon r0 2M `P 1.6 1033 cm. In
this case, the above results yield
k2E 3 103 ,

k2B 6 103 ,

k3E 4 104 ,

k3E 9 104 ,

(13)

for a wormhole in the entire mass range M [1, 100] M .

10-35

60

10-25

10-15

10-5

40

20

0
0.0

0.2

0.4

0.6

0.8

1.0

FIG. 3. The l = 2 and l = 3, axial- and polar-type TLNs for

a stiff wormhole constructed by patching two Schwarzschild
spacetimes at the throat radius r = r0 > 2M . The TLNs are
negative and all vanish in the BH limit, r0 2M . The latter
is better displayed in the inset.

2.

Perfectly-reflective mirror

Thermodynamical arguments suggest that any horizonless microscopic model of BH should act as a mirror,
at least for long wavelength perturbations [35, 37]. Motivated by this scenario, we consider a Schwarzschild geometry with a perfect mirror at r = r0 > 2M and impose
Dirichlet boundary conditions on the Regge-Wheeler and
Zerilli functions, for the axial and polar sector, respectively. Thus, our strategy is to consider the stationary limit of generically dynamical perturbations (in the
Fourier space, where is the frequency of the perturbation) of a Schwarzschild geometry.
The final result, in the 0 limit, reads (cf. Appendix D for details)
8
6 103 ,
5(7 + 3 log )
8
k3E
9 104 ,
35(10 + 3 log )
32
k2B
6 103 ,
5(25 + 12 log )
32
k3B
9 104 ,
7(197 + 60 log )
k2E

(14)
(15)
(16)
(17)

where the last step is evaluated at r0 2M `P 1.6

1033 cm and it is roughly valid in the entire mass range
M [1, 100] M due to the mild logarithmic dependence.
We note that also for this model all TLNs are negative
(i.e., they have the opposite sign relative to the neutronstar case) and that klE,B C (2l+1) in the Newtonian
limit. The TLNs klE,B for this model as functions of the
compactness are shown in Fig. 4.

8
Appendix D yields7
0.050

16
4 103 ,
(19)
5(23 6 log 2 + 9 log )
16
k3E
6 104 , (20)
35(31 6 log 2 + 9 log )
32
4 103 , (21)
k2B
5(43 12 log 2 + 18 log )
32
6 104 . (22)
k3B
7(307 60 log 2 + 90 log )

200

k2E

0.010
0.005

150

0.001
10-35

10-25

10-15

10-5

100

50

0
0.0

0.2

0.4

0.6

0.8

1.0

FIG. 4. TLNs for a toy model of Schwarzschild metric with

a perfectly reflective surface at r = r0 > 2M . The TLNs are
all negative and vanish in the BH limit, r0 2M . Close to
the BH limit, the polar- and axial-type Love numbers for the
same multipolar order are almost identical, as shown in the
inset.

3.

As already noted in Ref. [58], the Newtonian regime of a

gravastar is peculiar due to the de Sitter interior; consequently, the TLNs scale as klE,B C 2l . In this case,
the scaling of the polar TLNs is different from that of
an ordinary NS and of the other models of microscopic
corrections at the horizon scale, whereas the scaling of
the axial TLNs is the same as that for wormholes and for
the perfect-mirror model.
Interestingly, also in the gravastar case, the axial and
polar TLNs have a logarithmic behavior in the BH limit
and they are negative, as discussed in Refs. [58, 59] for the
polar case only. The behavior of klE,B as functions of the
compactness is shown in Fig. 5. The quadrupolar polartype TLNs for more generic thin-shell gravastar models
are presented in Ref. [59].

Thin-shell gravastars
40

0.010
0.005

For completeness, here we briefly consider the case of

another ECO, namely gravastars [29]. The interior of
these objects is described by a patch of de Sitter space,
which is smoothly connected to the Schwarzschild exterior through an intermediate region filled with a perfect
fluid. A particularly simple model is the so-called thinshell gravastar [30], in which the thickness of the intermediate region shrinks to zero. Remarkably, these models
are simple enough that the TLNs can be computed analytically [58, 59].

30
0.001
5. 10-4
10-35

10-25

10-15

10-5

20

10

0
0.0

0.2

0.4

0.6

0.8

1.0

Among thin-shell gravastars, we consider the simplest

case where the background metric is Eq. (B2) with,

e =e

1 2M
r
2
1 2C rr2
0

r > r0
.
r < r0

(18)

In this model, the thin shell is described by a fluid with

zero energy density and negative pressure [58]. By extending the formalism developed in Ref. [59] (cf. also
Refs. [58, 75]), it is easy to compute the TLNs of this
solution. In the BH limit, the computation derived in

FIG. 5. TLNs for a thin-shell gravastar with zero energy density as a function of the compactness. More generic gravastar
models are presented in Ref. [59]. The TLNs are all negative and vanish in the BH limit, r0 2M . Similar to the
perfectly-reflective mirror case, the polar- and axial-type Love
numbers for the same multipolar order coincide in the BH
limit, as shown in the inset.

This result corrects the computation performed in Ref. [58],

which is flawed due to the fact that it does not impose the correct
boundary conditions across the shell. For a stiff equation of state,
the correct boundary conditions read [[K]] = 0 = [[dK/dr ]] as
derived in Appendix D and in Ref. [59].

9
4.

On the universal BH limit

It is remarkable that the models described above display a very similar behavior in the BH limit, when the
radius r0 2M , cf. Table I. Indeed, although all TLNs
vanish in this limit, they have a mild logarithmic dependence. On the light of our results, it is natural to
conjecture that this logarithmic dependence is a generic
feature of ultracompact exotic objects, and will hold true
for any ECO whose exterior spacetime is arbitrarily close
to that of a BH in the r0 2M limit.
Due to this mild dependence, the TLNs are not extremely small, as one would have naively expected if the
scaling with were polynomial. Indeed, in the Planckian
case (r0 2M `P ) the order of magnitude of the TLNs
is the same for all models and it is given by Eq. (2). In
particular, the TLNs of Planckian ECOs are only five orders of magnitude smaller than those a typical NS. The
detectability of these deviations from the zero-Love
rule of BHs in GR is discussed in Sec. V.

IV.

TIDAL PERTURBATIONS OF BHS

BEYOND GR

In this section, we discuss the TLNs of BHs in other

theories of gravity. Technical details are given in Appendix B.

A.

Scalar-tensor theories

We start with scalar-tensor theories, which generically

give rise to stationary BH solutions which are identical to those of GR [42, 45, 46]. Therefore, the background solution which we deal with is still described by
the Schwarzschild geometry. In the Jordan frame, neglecting the matter Lagrangian, the simplest example of
scalar-tensor theory is described by the Brans-Dicke action (cf., e.g., Ref. [42])
Z


1
BD
S=
d4 x g R
,
(23)
16

where BD is a dimensionless coupling constant and

is a scalar field characteristic of the theory. Action (23)
yields the equations of motion,


BD
1
1

G = 2 g + ,

(24)
 = 0 .
(25)
As mentioned above, the background solution is
Schwarzschild with a vanishing scalar field.
Following the procedure described in Sec. II, we consider metric perturbations given by Eqs. (B3) and (B4)
for the polar and axial sector, respectively, and a scalar

field perturbation given by Eqs. (B6) and (B7). Since

the scalar perturbations are even-parity, axial gravitational perturbations do not couple to them, implying that
this sector is governed by equations identical to those of
vacuum GR. Therefore, all axial-type TLNs of a nonrotating BH in Brans-Dicke gravity are zero, klB = 0.
On the other hand, in the polar sector, scalar perturbations can be obtained from Eq. (25) by using the
decomposition in Eqs. (B6) and (B7),
00 +

2(r M )0 l(l + 1)
= 0,
r(r 2M )

(26)

The solution which is regular at the horizon is

 r

= Cl Pl
1 ,
(27)
M
where Pl is a Legendre polynomial, Cl is an integration
constant, and we have expanded the scalar field as in
(0)
Eq. (B6) with = 0. By comparing the above expression with the scalar-field expansion in Eq. (B13), we
conclude that Cl ElS and that the induced scalar multipoles l are zero. Therefore, Eq. (27) represents an
external scalar tidal field and the scalar TLN are identically zero.
Although we wish to focus on gravitational tidal fields,
it is instructive to investigate the role of a scalar tide in
scalar-tensor theory. By substituting Eq. (27) in Eq. (24)
we obtain an inhomogeneous differential equation for H0 .
For l = 2, we can identify C2 32 M 2 E2S and we get


2(r M ) 0 2 2M 2 6M r + 3r2
00
H
H0
H0 +
r(r 2M ) 0
r2 (r 2M )2


4M 2 2M 2 6M r + 3r2 S
=
E2 .
(28)
3r2 (r 2M )2
The above equation can be solved analytically. The solution which is regular at the horizon reads
2
(29)
H0 = r2 E2 + 2M rE2 M 2 E2S .
3
The induced quadrupolar moment is zero, and therefore
k2E = 0, just as in the GR case. It is straightforward
to show that this result generalizes to higher multipoles,
klE = 0. In conclusion, although in Brans-Dicke theory
the BH metric perturbations depend on scalar tides, all
TLNs of a static BH vanish, as in the case of GR.
B.

Einstein-Maxwell

We consider Reissner-Nordstrom BHs, which are the

unique static solution to Einstein-Maxwell theory, although our results are valid for any U (1) field minimally
coupled to gravity, as in the case of dark photons or
the hidden U (1) dark-matter sector [44]. The EinsteinMaxwell field equations read
G = 8 T ,
F = 0,

(30)
(31)

10
where F = A, A, is the Maxwell tensor and


1
1
g F F F F g ,
(32)
T =
4
4
is the stress-energy tensor of the EM field. The background spacetime is the well-known Reissner-Nordstrom
metric, whose line element reads as in Eq. (B2) with
e = eg = 1

2M
Q2
+ 2 f (r),
r
r

(33)

where M and Q denote the mass and the charge of the

BH, respectively. The background Maxwell 4-potential
reads
(0)

A = (Q/r, 0, 0, 0) .

Polar TLNs

The polar functions of the metric are coupled to the

EM function u1 through the field equations. In the
Lorenz gauge, we find the following coupled equations,
(1)
4Q
D u1 = 0 ,
r3 2M r2 + Q2 r 1
(2)
Q (1)
D2 u1 + D2 H0 = 0 ,
r
(2)

D1 H0 +

(35)
(36)

where we defined the operators,

(2)

d2
2(M r) d
1  2

+
Q r(4M ( 2)r)
dr2
r2 f
dr r6 f 2



r2 4M 2 2M r + r2 2Q4 ,


Q2 r2
d
=
+
,
dr r (r(r 2M ) + Q2 )
d2
4Q2 r2 d
= 2+
,
dr
r4 f
dr


2 M r Q2
d
,
=
+
dr
r3 f

D1 =

(1)

D1

(2)

D2

(1)

D2

with := l(l + 1). This system allows for a closed-form

solution. For simplicity, we impose the absence of electric
tidal fields, which requires that the function u1 does not
contain r3 -terms at large distance [cf. Eq. (B11)]. In this
case, the regular solution at the horizon for l = 2 reads
H0l=2 = E2 r2 f ,
E2 r2 Qf
ul=2
=
.
1
2

2.

(34)

Because the background is electrically charged, gravitational and EM perturbations are coupled to each other.
To compute the tidal deformations, we expand the metric as in Eqs. (B3) and (B4) and the Maxwell field as in
Eqs. (B5) and (B8). As before, we consider the polar and
the axial sectors separately.

1.

Due to the gravito-EM coupling, an external tidal field

induces a Maxwell perturbation which is proportional to
the BH charge Q. A simple comparison between the
above results and the expansions in Eqs. (B9) and (B11)
shows that the multipole moments are all vanishing.
Although the full solutions for l > 2 are cumbersome,
it can be shown that for any l > 2 the large-distance expansion of the solutions for H0 and u1 which are regular
at the horizon is truncated at the 1/r term for any l, and
it is an exact solution of the coupled system. Therefore,
the above result directly extends to any l, and we obtain that the TLNs of a charged BH are zero in the polar
sector, klE = 0, like in the Schwarzschild case.

(37)
(38)

Axial TLNs

The calculations for gravitational axial TLNs and magnetic TLNs are simpler. The r-component of Einsteins
equations leads to h1 = 0 which automatically satisfies
the -component. The final axial system reads




4Qu04
r2 f r2 h000
2Q2 + r (r 4M ) h0 = 0 ,

(39)
2Q
(Qu04 h0 ) = 0 .
r2 f u004 Qh00 + 2M u04 u4
r
(40)
Also in the axial sector, the coupled system admits an
analytic, closed-form solution. In the absence of EM tidal
fields, the solutions which are regular at the horizon read
r3
f B2 ,
3
r2
= QB2 (1 Q2 /r2 ) .
2

hl=2
=
0

(41)

ul=2
4

(42)

which proves that also the axial TLNs of a charged BH

are zero. It is straightforward to extend this result to
higher multipoles, finding klB = 0.
To conclude, we obtain the interesting result that all
TLNs or a static BH in Einstein-Maxwell theory are identically zero, as in the uncharged case.

C.

Chern-Simons gravity

In this section we compute the TLNs of a non-rotating

BH in Chern-Simons theory [54],


Z

1
CS
SCS = d4 x g R g ab a b +
RR ,
2
4
(43)
where CS is the coupling constant of the theory and

RR is the Pontryagin scalar,

RR =

1
Rabcd baef Rcd ef .
2

(44)

11
1
To O(CS
), the only correction is in the scalar-field equation, which reads

The field equations arising from action (43) are

1
a b CS Cab ,
2
CS
 =
RR,
4

Rab =

(45)

(2)

DS (1) =

(46)

12B2 M
,
r2 (r 2M )

(50)

where
cde(a

b)
e R d

dabc

where Cab = c 
+ d c R
. We will
focus on spherically symmetric background solutions to
Eqs. (45) and (46). In these conditions, the Pontryagin
scalar vanishes and the background is described by the
Schwarzschild metric with a vanishing scalar field [7678].

1.

Polar TLNs

In Chern-Simons gravity, the polar perturbations of a

Schwarzschild BH are equivalent to GR [78, 79]. Therefore, the analysis for the polar TLNs is identical to that
discussed in Ref. [8], and one can conclude that all polar TLNs of a non-rotating BH in Chern-Simons gravity
are zero, klE = 0. As discussed in Sec. V, the polar
TLNs are the dominant correction to the inspiral waveform [13]. Thus, the simple fact that in Chern-Simons
gravity these TLNs are vanishing already suggests that
it would be very difficult to constraint this theory with
GW measurements of the BH tidal deformability.

2.

Axial TLNs

On the other hand, the field transforms as a pseudoscalar and is therefore part of the axial sector. We can
thus expect that non-trivial axial-type TLNs in ChernSimons gravity may exist. In the stationary limit, we
find h1 = 0, and the field equations for the axial sector
reduce to a system of two coupled second-order differential equations for h0 and . This system can be solved
numerically for a generic coupling CS or perturbatively
when CS := CS /M 2  1. The latter case is consistent
with the action (43) being an effective field theory [54].
We have adopted both procedures, described below.
In the perturbative limit, we expand the metric and
scalar perturbations in powers of the coupling CS  1,
2
(2)
h = h(0)
+ CS h + ...
(1)

= CS

+ ...

(47)
(48)

and solve the perturbation equations order by order in

CS .
0
a. Quadrupolar TLNs. For l = 2 and to O(CS
), the
metric perturbation which is regular at the horizon reads


B2 3
2M
(0)
h0 =
r 1
,
(49)
3
r
as in the GR case. The advantage of the perturbative
approach is that the equations decouple from each other.

(l)

DS :=

2(M r) d
d2
l(l + 1)
2

.
dr2
r 2M r dr r2 2M r

(51)

Again for simplicity, we impose the absence of scalar tidal

fields, i.e. we require that does not contain a divergent
rl term at large distance [cf. Eq. (B13)]. In this case, the
solution which is regular at r = 2M reads

B2 M 2
(1) =
54 36y + 2 (2 + 3(y 2)y)
2
hyi h
h y ii
12(1 y) + (2 + 3(y 2)y) log
+ 3 log
2
2

h
yi
+ 6(2 + 3(y 2)y) polylog 2, 1
,
(52)
2
where we defined y = r/M . With the above solution
2
) equation for the axial perturbation
at hand, the O(CS
(2)
(2) (2)
reads DA h0 = SA with





3B2 M
(2)
SA (r) :=
(2 y) 2 3y 2 2 18(1 + 2y)
4
(y 2)y
hyi
h


yi
+ 3 log
6(y 2) 3y 2 2 polylog 2, 1
2
2
h
h y ii


4(3(y 1)y 4) (y 2) 3y 2 2 log
.
2
(53)
It is convenient to solve the above inhomogeneous
equation through the Greens function. The solution with
the correct boundary conditions is
Z
+ (r) r
(2)
(2)
dr0 SA (r0 ) (r0 )
h0 (r) =
W
2M
Z
(r) 0 (2) 0
+
dr SA (r )+ (r0 ) ,
(54)
W
r
where the two linearly independent solutions of the homogeneous problem read
A1 r2 (r 2M )
,
(55)
3
4M



A2
+ (r) =
2M 2M 3 + 2M 2 r + 3M r2 3r3
3
6M r

+ 3(2M r)r3 log(1 2M/r) ,
(56)

(r) =

and W = A1 A2 /M is the Wronskian, which is constant

by virtue of the field equations. The above solutions are
regular at the horizon and at infinity, respectively. At
(2)
large distances, SA B2 /r5 , and the first integral in
Eq. (54) is convergent, whereas the second integral does

12
not contribute to the current quadrupole moment S2 .
Interestingly, these integrals can be computed in closed
form although their final expression is cumbersome. We
(2)
report here only the large-distance behavior of h0 (r),
namely
 3
M
B2 M 5
9
(2)
,
(57)
h0 (r) [9 8(3)] 2 + O
5
r
r3
where (n) is the Riemann Zeta function. By comparing
the above result with Eq. (B10) and using Eq. (3), it is
straightforward to obtain
k2B =

9
2
2
[8(3) 9]CS
1.10962 CS
.
5

(58)

Interestingly, we find that the axial TLN is non-zero and

2
proportional to CS
, as expected. We have also confirmed this result by integrating numerically the field
equations for arbitrary values of CS and by extracting the quadratic correction in the small-coupling limit.
Figure 6 shows a comparison between the analytical result (58) and the numerical one.

0.25

0.20

0.15

0.10

0.05

0.00
0.0

0.1

0.2

0.3

0.4

0.5

FIG. 6. Axial TLNs of a BH in Chern-Simons gravity for l = 2

perturbations calculated for different values of CS . The dots
correspond to the values obtained directly from a numerical
integration, whereas the line is the analytical result shown in
Eq. (58).

b. Octupolar TLNs. Although the computation for

the l = 3 case proceeds as presented above for l = 2, there
0
are some notable differences. In this case, the O(CS
)
perturbation reads
(0)
h0

B3 M 2
=
r (r 2M )(3r 4M ) ,
27

(59)

whereas the scalar-field equation to O(CS ) is

(3)

DS (1) =

16B3 M 3 (3r 5M )
,
3(r 2M )r2

(60)

and its solution regular at the horizon can be written in

a form similar to Eq. (53). However, even after imposing

the absence of scalar tides, the large-distance behavior

of the scalar field is (1) 4B3 /3 + O(M/r), i.e. it
approaches a constant value. This has important conse2
quences for the metric perturbations to O(CS
), which are
(3) (2)
(3)
governed by the equation DA h0 = SA , where now the
(3)
source term SA 8B3 /r4 at large distances. Because
the divergent solution of the homogeneous problem reads
r4 at large distances, the first integral in Eq. (54)
(whose schematic form is valid for any values of l) yields
terms which grow linearly and logarithmically at large
distances. By repeating the same procedure as before,
we obtain the following large-distance behavior for l = 3,

4B3 M 7
4B3 M 6
(2)

75600(3) 84331
h0
3r2
3969r3

 4
 r 
M
7560 log
+O
.
(61)
2M
r4
We therefore obtain two terms (the 1/r2 and the log r/r3
term) which decay more slowly than the octupole term,
1/r3 . We believe that these terms arise as subleading
corrections to the external tidal field, which is not captured by the asymptotic expansion in Eq. (B10), because
the latter does not include the effects of a constant scalar
field at infinity. In other words, the multipolar structure
of a tidally deformed BH in Chern-Simons gravity is more
involved than in GR. We anticipate that this issue also
appears in other modified theories and we postpone a
more detailed analysis to the future [80]. In order to get
an estimate for the TLN, we simply consider the ordinary
octupolar correction. By proceeding in the usual way, we
obtain
1
2
2
k3B =
(75600(3) 84331) CS
11.13 CS
, (62)
588
which is the value reported in Table I although, for the
reasons mentioned above, should only be considered as
an estimate.
V.

DETECTABILITY

To estimate the detectability of the TLNs through GW

observations, we use a Fisher matrix approach. In the
following, we summarize the basic features of this formalism, referring the reader to Ref. [81] (and to the references therein) for a detailed discussion of the topic.
We consider the output of a generic interferometer,
~ + n(t) ,
s(t) = h(t, )

(63)

where n(t) is the detector noise (assumed stationary),

~ is the GW signal. Our goal is to recover the
and h(t, )
physical parameters ~ = {1 , . . . , n }, and their errors
~ = ~
~ with respect to the true values
~ . We there~
fore need to compute the probability distribution p(|s),
which can be written as
~
x)s)
~ p(0) ()e
~ 21 (h()s|h(~
p(|s)
,

(64)

13
~ prior on the parameters [82]. The bracket
with p(0) ()
(|) represents the inner product
Z
)
? (f )
h(f
g ? (f ) + h
g (f )
df
,
(65)
(g|h) = 2
S
(f
)
h

where Sh (f ) is the detectors noise spectral density. According to the principle of the maximum-likelihood estimator, the values of the source parameters can be estimated as those which maximize Eq. (64). In the limit
~
of large signal-to-noise ratio, such that p(|h)
is tightly
peaked around the true values of the source parameters,
~ around
a Taylor expansion of p(|s)
~ leads to
~ p(0) ()e
~ 12 ab a b ,
p(|s)

(66)

where

ab =



h  h
a  b =~
~

(67)

is the Fisher information matrix. Inversion of the latter

yields the covariance matrix,

ab
ab = 1
.
(68)
The error on the source parameters a are then given by

a = aa ,
(69)
and the correlation coefficients between a and b are
given by
cab =

h a b i
ab

=
.
aa bb
aa bb

(70)

The TLNs enter the GW signal as a fifth-order postNewtonian (PN) correction which adds linearly to the
phase of the waveform,
) = A(f )ei(PP +T ) ,
h(f

2 5 E
M k2 .
3

where q := m1 /m2 > 1 is the mass ratio. For nonspinning objects, the waveform depends on 6 parameters ~ = (ln A, c , tc , ln M, ln , ), i.e. the amplitude, the phase and time at the coalescence, the chirp
mass M = 3/5 (m1 + m2 ), the symmetric mass ratio
= m1 m2 /(m1 + m2 )2 and the average tidal deformability defined in Eq. (73). Nonetheless, ln A is completely
uncorrelated with the other variables, and therefore we
will restrict our analysis by performing derivatives only
with respect to the remaining parameters, leading to a
5 5 Fisher matrix.
The detector properties are encoded in the noise spectral density Sh (f ). We perform the analysis both for
terrestrial and space interferometers.
For the Earth-based detectors, we consider
(i) AdLIGO with its anticipated design sensitivity
curve ZERO DET high P [86] and (ii) the ET design
configuration, with noise described by the analytic fit
provided in Ref. [87]. As space-based detector, we
consider the most optimistic LISA configuration, namely
the N2A5 model defined in Ref. [88], with a 5 106 km
arm-length and an observing time of Tobs = 5 yr.
To compute the errors on the tidal deformability, we numerically integrate Eq. (67) within the freAdLIGO
quency range [fmin , fmax ], where fmin
= 20 Hz,
ET
LISA
fmin = 1 Hz, and fmin
= Max[105 , 4.149
3/8
105 (106 M)5/8 Tobs ] Hz [89]. For the upper freAdLIGO
quency we choose fmax
= (63/2 m)1 , while for LISA
LISA
3/2
fmax = Min[1 Hz, (6 m)1 ], being m = m1 + m2 the
total mass of the system.
A.

(72)

The contribution of higher multipoles and of the axialtype TLNs is subleading and will be neglected in the
following.
In our analysis, we use the so-called TaylorF2 approximant of the GW template in the frequency domain [83],
which is 3.5PN accurate in the point-particle phase and
2PN accurate in the tidal term [84, 85].8 For binary systems for which i=1,2 6= 0, finite-size effects are described

(73)

Model-independent tests with GWs

(71)

where PP (f ) is the point-particle contribution, while

T (f ) describes the tidal effects and to the leading
order depends on the l = 2 polar TLNs through the
constant [5, 6]
:=

in terms of the average deformability,



1
q + 12
=
(1 + 12q)1 +
2 ,
26
q

For the purpose of this paper, we consider the amplitude at the

leading order.

Before discussing the detectability for different families

of ECOs, it is instructive to analyze the impact of the
TLNs on the GW signal in a more general framework.
Figure 7 shows the relative uncertainty / for equalmass binaries at d = 100 Mpc (for AdLIGO and ET) and
at d = 500 Mpc (for LISA), as a function of the mass of
the objects and for different values of the TLNs k2E (top
panels) and of the average tidal deformability (bottom
panels). In the panels of Fig. 7, the dashed horizontal
line denotes the upper bound = . Therefore, each
point above that line is indistinguishable from a BH-BH
binary in GR ( = k2E = 0) within the errors, whereas
a measurement of the TLNs for systems which lie below
the threshold line would be incompatible to zero and,
therefore, the ECOs can be distinguished from BHs in
this case.
It is worth remarking that motivated by the prospect
of measuring the TLNs of NSs through GW detections
several efforts have been devoted to investigate the detectability of for objects with M . 2M . The latter

14

kE2 = 10

10

00
kE 2 = 10

1000

10

100

10

= 100

1
0.1

107
105

10

15

20

25

30

106
105
104
1000
100
10
1
0.1

100

10

15

20

10
1
0.100
0.010

0.001

1011

M [M]

|/|[%]

|/|[%]

25

30

|/|[%]

100

k E2 = 1

|/|[%]

|/|[%]

1000

|/|[%]

109
107
105
1000
10
0.1

kE

kE

=1

=1
0
kE
2 =
100
2

kE

=1
000

10

k
10 2
=
2
10
= 24
10
= 26
10
=
28
10
=
20

50

100

M [104 M]

M [M]

FIG. 7. Relative percentage errors on the average tidal deformability for equal-mass binaries at 100 Mpc (for AdLIGO and
ET, left and middle panel, respectively) and at 500 Mpc (for LISA, right panel) as functions of the mass of the single object
and for different values of the TLN k2E (top panels) and of (bottom panels) of the two objects. The horizontal dashed line
identifies the upper bound / = 1.

represents the mass range in which terrestrial interferometers will provide new information on matter at supranuclear densities from neutron-star binaries. On the other
hand, our results shown in Fig. 7 do not assume any specific model and extend the analysis of the detectability
of the TLNs to a regime unexplored so far, where more
massive ECOs can contribute to the GW signal through
finite-size effects. Likewise, to the best of our knowledge,
this work presents the first analysis on the detectability
of tidal effects with LISA.
From the bottom panels of Fig. 7 we note that, for a
fixed , the detectability is favored for low-mass systems,
as the tidal phase scales with the inverse of the total mass
T m10/3 (1 + q)3 /q. Moreover, for 2M . M .
5M , AdLIGO will constrain the TLNs for small compactness only (i.e., for large ). This picture improves
for ET, which leads to an upper bound / = 1 up
to M ' 15M . Therefore, as far as terrestrial interferometers are considered, the high-compactness regime for
ECOs seems to be available only for the third generation
of detectors. This result is also evident from the top panels of the left and middle plots in Fig. 7, which show that
AdLIGO will not be able to set any significant constraint
below k2E ' 10, regardless the ECO mass.
On the other hand, space interferometers open a completely new window onto finite-size effects. The top-right
panel of Fig. 7 shows that LISA is capable to bound the
Love numbers with a relative accuracy / . 10% in
almost the entire mass range M [104 , 106 ] M . In other
words, binary systems made of intermediate-mass compact objects will provide interesting constraints on the
TLNs, with k2E ' 1 and below, and therefore also on
the nature of these objects. The exquisite precision of
LISA can be traced back on the dependence of the tidal
deformability (which is the physical parameter entering
the waveform) on the ECOs mass, i.e. M 5 , which
amplifies the tidal effect on the GW signal. This is confirmed by the right-bottom panel of Fig. 7, where the

upper bound on = 1024 km5 for M = 5 104 M corresponds to k2E 0.7.

B.

Detectability of ECOs

Let us now turn our attention to some specific models and, in particular, to the models of ECOs investigated in the previous sections. Based on the previous discussion, as a general setup we consider equalmass binaries at distances d = 100 Mpc with M
[2, 30]M for AdLIGO/ET, and at d = 500 Mpc with
M [104 , 106 ]M for LISA. We note that the GW signal is proportional to 1/d, and therefore the covariance
matrix (68) (i.e., the error on ) scales linearly with the
distance.
In Fig. 8 we show the percentage relative errors /
for models of wormholes, perfect mirrors, and gravastars,
as a function of the mass of the object and for different
values of its compactness. Some qualitative results are independent of the nature of the ECO: the left panels confirm that AdLIGO would be able to constrain the tidal
deformability only for small values of the compactness,
namely C . 0.2. As the errors scale with the distance,
an upper bound / 1 for C = 0.3, would require a
source located at a distance 10 Mpc.
Furthermore, the relative errors decrease for larger
masses, reach a minimum, and then increase again. This
behavior can be explained by looking at the functional
form of Eq. (72). For a fixed compactness (i.e., for
fixed k2E ), the average tidal deformability grows with the
ECO mass, thus making the tidal part of the gravitational waveform more easy to be detected. However,
since the template is truncated at the last stable orbit, fmax m1 , increasing the mass also reduces the
number of effective cycles spent into the detectors bandwidth. This is particular penalizing for tidal effects,
which enter the GW signal as high-PN/high-frequency

15

50

10
5

1000
500
100
50

100
10

10
Gravastar

5 10 15 20 25 30 35 40

100
50
10
5

Perfect Mirror

1
0.5

5 10 15 20 25 30 35 40

M [M]

M [M]

Wormhole

=0
.49
=
0.4

=0
.3

=0
.2

=0

.1

Perfect Mirror

1000

100
10
1
0.10
0.01

Gravastar

1000
100
10
1
0.10
0.01

Perfect Mirror

1
0.1

= 0.1

Wormhole

1000

500

100

= 0.2

1
0.1

= 0.3

|/|[%]

10
5

10

|/|[%]

Wormhole

|/|[%]

100

50
0.1

100

|/|[%]

0.2
=

|/|[%]

= 0.3

|/|[%]

1000
500

|/|[%]

|/|[%]

|/|[%]

Gravastar

1000
100
10
1
0.10
0.01
1

5 10

[104

50 100

M ]

FIG. 8. Relative percentage errors on the tidal deformability for binaries observed by AdLIGO (left panels), ET (middle
panels), and LISA (right panels), as functions of the ECO mass and for different values of the compactness. For terrestrial
interferometers we consider prototype binaries at d = 100 Mpc, while for LISA we set the source at d = 500 Mpc. Top, middle
and bottom panels refer to wormholes, perfect-mirror models, and gravastars, respectively.

corrections.

sole detection of the TLNs) up to compactness C 0.3.

It is worth noticing that a network of advanced interferometers would improve these results, even though it
will not drastically change the upper bound on the compactness of these objects. Indeed, if we consider that
the experiments are all independent, the Fisher matrices computed for each detector simply sum up, and the
overall error on is given by the inverse of the total
ab . Assuming five detectors with the same sensitivity
of AdLIGO,
the relative error would decrease roughly
by a factor 5 which, from our results in Fig. 8, is still
not enough to constrain objects much more compact than
C 0.2.

This scenario improves drastically for space-based detectors such as LISA (right panels of Fig. 8). Within the
considered mass range, tidal effects may be measured for
ECOs with C . 0.3 up to 1% of accuracy. Moreover,
LISA will be able to put strong constraints even for more
compact objects: for M & 105 M it would be possible
to set an upper bound / = 1 in the entire parameter space. As discussed in the previous section, these
results rely on the magnitude of the ECOs mass, which
strengthens the effect of tidal interactions in the waveform. The right panels of Fig. 8 show indeed that for all
the considered ECO models, LISA leads the analysis to
nearly explore the BH limit C 1/2.

Third-generation ground-based detectors, like ET

(middle panels of Fig. 8), hold more promising results.
In this case, the relative errors / decrease roughly
by one order of magnitude relative to AdLIGO. A GW
detection of an ECO binary at d = 100 Mpc would allow
to distinguish the system from a BH-BH binary (by the

It is worth remarking that, as finite-size effects develop

during the late inspiral, eventually leading to complex
phenomena like the excitations of modes [90], a more
accurate template which extends the frequency domain
of the waveform up to the merger phase, would improve

16
this analysis.

C.

Detectability of BSs

For each model of BSs listed in Table II, we focus on the

most compact configuration in the stable branch. The
mass of this configuration depends on the parameters
of the potential. In Fig. 1, we show the results of the
Fisher matrix analysis for an equal-mass BS-BS binary
as a function of the BS mass, obtained by considering the
most compact configuration and by varying the parameters of the potential. Since, for each model, we consider
the maximum compactness allowed in the non-spinning
case, our results can be seen as conservative, since less
compact configurations are easier to discriminate.
The forecast for detecting BS-BS binaries by using GW
tidal effects are more promising than for other ECOs, because the compactness of BSs is at most C . 0.3 and,
in turn, their TLNs are larger. Advanced LIGO will be
able to discriminate between minimal BSs and BHs in all
the mass range. These errors worsen for massive objects.
However, for all the models analyzed, second-generation
detectors will set a strong upper bound on the tidal deformability, unless solitonic BS are considered. Results
in Fig. 1 suggest that models of minimal and massive
BSs can be distinguished from BHs through AdLIGO detections, whereas discriminating between BHs and more
compact BSs such as solitonic models will require future
detectors like ET. Again, the uncertainties significantly
decrease with LISA, especially in the high-mass regime,
where relative errors are below 1% for every binary configuration with M & 5 104 M .
D.

Testing GR

Our results suggest that, in generic theories of gravity, the GW signal from a BH coalescence contains a
5PN term which depends on the TLNs. Although the
inclusion of tidal corrections in BH binaries is important
for a correct modeling of the waveform, in most theories this term is subleading relative to other corrections
coming both from dissipative effects (for example due
to scalar- or vector-wave emission) and from corrections
to the Hamiltonian of the binary [41, 42]. Nonetheless,
there are cases in which the tidal deformability found
in this work is the dominant correction relative to the
GR waveform. This is the case of Chern-Simons gravity,
where corrections to the GW phase enter at 2PN order
if the components of the binary are spinning, but only
at 7PN order in the absence of spin [91, 92]. Therefore,
for non-spinning binaries the tidal correction will be the
dominant one and it is interesting to estimate to which
level the Chern-Simons parameter can be constrained by
GW observations of BH binaries.
Unfortunately, the polar TLNs of Schwarzschild BHs
in Chern-Simons gravity are zero as in GR and the axial

TLNs enter the waveform only to subleading order. Be2

cause k2B CS
[cf. Eq. (58)], in order to place an upper
bound on the Chern-Simons coupling at the level of (say)
CS . 0.2, one should be able to constrain the quadrupolar axial TLN at the level of k2B . 0.05. The results
shown in Fig. 8 suggest that this will be very challenging also with ET. The constraints on the dimensionless
coupling CS will significantly improve with LISA, but
they will translate in poor bounds on the dimensionful
coupling of the theory, CS := CS M 2 , due to the large
mass of LISA sources.
Finally, for spinning BH binaries in Chern-Simons
gravity, the leading correction enters at 2PN order and it
is quadratic in the spin [91]. Therefore, our results should
provide a reliable estimate for binaries with dimensionless
3
spin s  vorb
, where vorb is the typical orbital velocity
corresponding to the peak of sensitivity in the detector
band. Interestingly, for spinning BHs in Chern-Simons
gravity, we expect that also the polar TLNs will be nonzero due to the spin-tidal effects discussed in Refs. [20
22, 66]. In this case, the tidal effects on the waveform
will be stronger, albeit subleading relative to dissipative
effects.
VI.

DISCUSSION AND EXTENSIONS

The theory of the tidal deformability of compact objects has attracted considerable attention over the last
few years. So far, applications of this theory have been
mostly limited to astrophysics and to the possibility of
constraining the equation of state of NSs with GW observations. In this paper, we argued that tidal effects can
also be used to explore fundamental questions related to
the nature of event horizons, the existence of ECOs, and
the behavior of gravity in the strong-field regime.
Our main results can be summarized as follows:
In the framework of GR, the TLNs of ECOs are
generically non-zero. In the limit that the ECO
compactness C 1/2, all TLNs vanish but only
logarithmically. This result holds for all models of
ECO we have considered. It is therefore natural
to conjecture that this logarithmic dependence is a
generic feature of ultracompact exotic objects.9
The TLNs of a charged BH in Einstein-Maxwell
theory and of an uncharged static BH in BransDicke theory vanish, as in GR. These are both compelling extensions of GR, but our results indicate
that the TLNs of BHs are non-zero in other interesting extensions. In particular, we have explicitly shown that the axial TLNs of a Schwarzschild

Of course, we are implicitly assuming that the radial coordinate

r, in terms of which the compactness is defined, is special. Nevertheless, some models do suggest corrections at the Planck scale
in this same radial coordinate, and it arises naturally also when
expressing the results in terms of proper length [34].

17
order. A natural extension of our work is to include
tidal corrections also when they are subleading relative to other, point-particle, beyond-GR terms.
A related point is the fact that, when the TLNs
are small, their effect might be smaller than other
point-particle terms entering the waveform at 5PN
order. In this case, the full 5PN waveform might
be needed in order to extract the TLNs properly.10

BH in Chern-Simons gravity are non-zero. This

result also extends to the TLNs of static BHs in
Einstein-dilaton-Gauss-Bonnet gravity [42], but a
full analysis is left for future work [80]. In this
case, the presence of extra charges makes the multipolar expansion more involved. Preliminary investigation indicates that both the polar and the
axial TLNs of static BHs in Einstein-dilaton-GaussBonnet gravity are proportional to the coupling
constant squared, similarly to the Chern-Simons
case [cf. Eq. (58)].
We have explored the detectability of these tidal
effects in some details, both for ground- and for
space-based detectors. Ground-based detectors
such as AdLIGO and ET can constrain ECO models with compactness C . 0.2 and C . 0.3, respectively, whereas a LISA-like mission can constrain
supermassive ECOs up to C . 0.49. Interestingly,
Advanced LIGO can set stringent constraints on
various BS models, and both ET and LISA will be
able to discriminate a BS binary from a BH binary
just by measuring the TLNs of the binary components.
The prospects for testing deviations from GR are
less promising. While the TLNs of BHs beyond GR
are different from zero, their effect in the GW signal
is small and typically subleading relative to other,
point-particle, beyond-GR effects such as dipolar
emission. Nevertheless, the non-vanishing of the
BH TLNs remains a smoking gun of deviations from
GR and its phenomenological implications are under investigation [80].
At the same time, our work is intended to be only a first
step in understanding the tidal deformability of ECOs
and of BHs beyond GR; as such, it can be extended in
several interesting ways:
We neglected the presence of tidal fields of different nature, for example EM tidal fields in EinsteinMaxwell theory or scalar fields in Brans-Dicke theory or Chern-Simon gravity. We anticipate that
the presence of extra tidal fields will give rise
to new families of TLNs, which are related to
the mass/current multipole moments induced by
a (scalar or vector) extra tidal field [80].
We focused on non-rotating objects. The spin of
the individual components of a neutron-star binary
are typically small, but this might not be the case
for ECOs and BHs. In general, subleading spin effects might be included by applying the formalism
developed in Refs. [2022, 66] to the systems studied in this work.
We have estimated the detectability of tidal corrections in the GW signal from BH binaries beyond
GR only in the simple case in which the leadingorder correction to the GR waveform enters at 5PN

ACKNOWLEDGMENTS

We thank Leonardo Gualtieri for interesting discussions. V.C. acknowledges financial support provided
under the European Unions H2020 ERC Consolidator
Grant Matter and strong-field gravity: New frontiers
in Einsteins theory grant agreement no. MaGRaTh
646597. Research at Perimeter Institute is supported
by the Government of Canada through Industry Canada
and by the Province of Ontario through the Ministry
of Economic Development & Innovation. This project
has received funding from the European Unions Horizon 2020 research and innovation programme under the
Marie Sklodowska-Curie grant agreement No 690904, the
NewCompstar COST action MP1304, and from FCTPortugal through the projects IF/00293/2013. The authors thankfully acknowledge the computer resources,
technical expertise and assistance provided by CENTRA/IST. Computations were performed at the clusters
Baltasar-Sete-Sois and Marenostrum, and supported
by the MaGRaTh646597 ERC Consolidator Grant.

Appendix A: TLNs of neutron stars

It is interesting to compare the TLNs of an ECO with

those a NS, in order to investigate whether GW measurements of the tidal deformability can be used to distinguish an ECO from an ordinary compact star. For
completeness, in Table III we report some fitting formulas for the polar and axial TLNs of a NS with two different
equations of state. The order of magnitude of the TLNs
for compact NSs is given in Table I.

Appendix B: Determination of TLNs

In order to compute the TLNs we need to calculate the

expressions for the induced mass and current multipole
moments as a function of the external tidal field. We use
a linear perturbation theory approach to disturb slightly

10

For NSs, the TLNs are enhanced by a factor (R/M )5 105 ,

alleviating this issue; we are indebted to Leonardo Gualtieri for
highlighting this point.

18
P
ai (1/2 C)i and klB C 2l 4i=0 ai (1/2 C)i for the TLNs of a static neutron
star with a stiff (MS1 [93]) and relatively softer (SLy4 [94]) equation of state. Data are taken from Ref. [21] and agree with the results in
Ref. [8] after using the conversion in Eq. (4).

TABLE III. Coefficients of the fit klE C (2l+1)

EOS

P4

i=0

a0

a1

a2

a3

a4

MS1

k2E
k3E
k2M
k3M

-0.581
-0.207
-0.096
-0.034

8.721
3.230
1.369
0.514

-48.69
-18.66
-7.099
-2.910

123.4
47.82
17.53
7.511

-112.8
-43.87
-16.11
-7.009

SLy4

k2E
k3E
k2M
k3M

-0.414
-0.150
-0.063
-0.023

6.227
2.326
0.876
0.346

-35.35
-13.57
-4.483
-1.978

92.70
35.56
11.57
5.288

-87.70
-33.38
-11.24
-5.115

the spacetime metric,

symmetric background metrics which are described by,

(0)

g = g + h ,

(B1)

(0)

where g is the background spacetime metric and h

is a small perturbation. We focus on static, spherically

(0)
g = diag e , eg , r2 , r2 sin2 .

We decompose h in spherical harmonics and separate the perturbation in even and odd parts, h =
odd
heven
+ h , according to parity. In the Regge-Wheeler
gauge [95], h can be decomposed as

e H0lm (t, r)Y lm H1lm (t, r)Y lm

0
0
H lm (t, r)Y lm eg H lm (t, r)Y lm
0
0

1
2
=
2
lm
lm

0
0
r K (t, r)Y
0
2
2
lm
0
0
0
r sin K (t, r)Y lm

heven

lm
lm
hlm
0
0
hlm
0 (t, r)S
0 (t, r)S

lm
lm
0
0
hlm
hlm

1 (t, r)S
1 (t, r)S
= lm
lm
lm
lm
h0 (t, r)S h1 (t, r)S
0
0
lm
lm
lm
lm
0
0
h0 (t, r)S h1 (t, r)S

hodd




lm
with Slm , Slm Y,
/ sin , sin Y,lm .
In the presence of scalars or vectors, spacetime fluctuations are accompanied by the corresponding fluctuations
in these fields,
(0)

A = A + A ,
(0)

=
(0)

+ ,

(B5)
(B6)

(0)

where A and are background quantities while A

and are small perturbations. We expand the scalar
perturbation and 4-potential A as [96, 97],
= lm Y lm ,
(B7)
 lm

lm
lm lm
lm lm
u1
u e
u Y + u4 Sb
A =
Y lm , 2
Y lm , 3 b
,
r
r

(B8)

(B2)

(B3)

(B4)



lm
and = l(l + 1). Hereafter we
with Yblm Y,lm , Y,
shall drop the (lm) superscripts on all quantities with the
exception of multipole moments.
Expressions for the metric functions in (B3) and (B4),
the electromagnetic functions in (B8), and the scalar
fields can be obtained by solving the linearized field equations for a given model. The remaining task consists in
extracting the multipole moments and tidal fields from
the spacetime metric. Thorne developed a method to
define the multipole coefficients of any spacetime metric given in coordinates which are asymptotically Cartesian and mass centered (ACMC) [65]. Another definition of the multipole moments of an axisymmetric and
asymptotically flat spacetime was given by Geroch and
Hansen [62, 63]. These two distinct definitions of moments were shown to be equivalent [64].

19
The multipole moments can be extracted from the
asymptotic behavior of the spacetime metric and fields,

#
!


2
4
Ml Y l0 + (l0 < l pole)
rl El Y l0 + (l0 < l pole) ,
rl+1
2l + 1
l(l 1)
l2
"r
!
#
l+1
X


2J
2
4
S
2r
l
gt =
sin2 +
S l0 + (l0 < l pole) +
Bl Sl0 + (l0 < l pole) ,
r
rl
2l + 1 l
3l (l 1)
l2
"r
#
!
X


2
4
2
Q
l0
0
l
l0
0
At = +
Ql Y + (l < l pole)
r El Y + (l < l pole) ,
r
rl+1
2l + 1
l(l 1)
l1
"r
#
!
X 2

4 Jl l0
2rl+1 
0
l0
0
A =
S + (l < l pole) +
Bl S + (l < l pole) ,
rl
2l + 1 l
3l (l 1)
l1

X 1 

 S

l0
0
l
0
= 0 +
l Y + (l < l pole) r El + (l < l pole) .
rl+1
2M X
+
gtt = 1 +
r

"r

(B9)

(B10)

(B11)

(B12)
(B13)

l1

The decomposition of the scalar field and the vector

potentials were chosen so they could be easily compared
with the metric decomposition. An appropriate comparison between the solution of the field equations and the expansions (B9) to (B13) allows us to extract the multipole
moments and, in turn, the TLNs as defined in Eq. (3).
Appendix C: Tidal perturbations of boson stars
1.

Background solutions

We consider spherically symmetric BSs, with background metric given by Eq. (B2), and the following
ansatz for the background scalar field,
(0)

(t, r) = 0 (r) eit .

(C1)

Despite the time dependence of the scalar field, the

Einstein-Klein-Gordon background equations are static,


1 eg
g
+ 8r 2 eg 20 + 02
0 + e V , (C2)
r


eg 1
g
0 =
+ 8r 2 eg 20 + 02
0 e V , (C3)
r


 0

g 0
dV
2
00
g
2
0 = e
e
0 +

00 ,
d|2 |
2
r
(C4)

0g =

The equilibrium spherically symmetric BS solutions

are found by integrating numerically Eqs. (C2) to (C4)
along with suitable boundary conditions. Namely, we impose regularity at the origin, = c , g = 0, = c ,
0 = 0, whereas at infinity we impose the metric to be
Minkowski and the scalar field to vanish. For a given

value c of the scalar at the center of the star, the problem is then reduced to an eigenvalue problem for the frequency , which we solve through a standard shooting
method. The value of c is arbitrary and can be tuned
in order to have (r ) = 0.
The total mass of the solution is M = m(r ),
where m(r) is defined by
eg (r) 1

2m(r)
.
r

(C5)

Contrary to the minimal and massive case in which

the scalar profile decays exponentially in the solitonic
model the scalar profile has a very steep profile which
makes the numerical integration of the background equations very challenging, requiring very fine-tuned shooting
parameters [39].
BSs do not have a hard surface, as the scalar is spread
out all over the radial direction. However, the configuration is highly localized in a radius 1/ and it is
customary to define the effective radius R as the radius
within which the 99% of the total mass is contained, i.e.
m(R) = 0.99M . This is slightly different for the solitonic model, in this case the steepness of the scalar profile
makes the definition of the radius more natural.
By using the above procedure, we can compute a
sequence of background solutions characterized by the
value of the scalar at the center of the star, c . In Fig. 9,
we show the gravitational mass M against R (left panel)
and the compactness C = M/R as a function of the mass
(right panel). For minimal and interacting BSs, we observe a maximum which separates stable (on the right)
from unstable (on the left) configurations [98100]. On
the other hand, solitonic BSs display a completely different behaviour: there exists a stable branch for high
values of R and small masses (bottom-right part of the

20

FIG. 9. Left panel: ADM mass as a function of the effective radius R for different models of BSs, including some unstable
configurations (to the left of maximum mass). Right panel: Compactness of the models in the stable branch as a function of
the mass. For massive and solitonic BSs we have considered = 104 2 and 0 = 0.05, respectively.

left panel in Fig. 9), an unstable branch that starts after the first maximum roughly at R 50, and then a
second stable branch which starts at R 10 up to the
maximum on the top-right part of the plot.

2.

Perturbations and TLNs

a.

Polar perturbations

We consider perturbations of the equilibrium configuration, sourced by an external static tidal field. The
metric perturbation is given by (B3) with H1 = 0. We
write the scalar perturbation as
X
=
eit 1 (r)Y lm (, ).
(C6)
m

As shown in the right panel of Fig. 9, for minimal

BSs the compactness is typically of O(0.01), and slightly
larger for massive BSs. On the other hand, solitonic BSs
can be almost as compact as BH (i.e., C 1/2), meaning that their radius can be of order of the Schwarzschild
light ring [55, 101].

Plugging (B3) and (C6) into the linearized Einstein

equations, we find H0 = H2 H, whereas K can be
written as a function of H, 1 and of the background
functions. We are then left with a radial equation for the
perturbation function H, which is coupled to the perturbed Klein-Gordon equation for the scalar perturbation 1 ,





 0
g + 0
2
2
2 2 g
0
02
0
00
0
2
g
H +
8r 0 e
+ 8r0 H + 32 0

0 0 e
1
r
2
r


20
eg (l2 + l + 2) 2
48eg 2 20 + 1602
+
H=0
+ 02
0
r
r2



 0


g + 0
2
2
0
00
0
2
g
001 +
8r 2 20 eg + 0 8r02

e
H
0
0
1
0
0
r
2
r
 00  0


0g

2 00
l(l + 1)eg
dV g
0 +
+
+ 3202
+
1
e 0 = 0
0
0
2
r 0
r2
d|2 |
00

We now solve the perturbation system supplied by regular boundary conditions at the origin,




(l)
(l)
H0 H0 rl + O rl+2 , 1 1 rl + O rl+2 . (C9)
(l)

Since the system is linear, the value of H0 can be set

(C7)

(C8)

to 1, and the correct value can be recovered a posteriori.

(l)
The value of 1 is determined by requiring that 1 0
as r using a shooting method.
At distances Rext much larger than the BS effective

21
radius, Eq. (C7) reduces to
H 00 +

2(r M ) 0 4M 2 2M r + r2
H = 0 , (C10)
H
r(r 2M )
r2 (r 2M )2

with := l(l + 1). Eq. (C10) has a general solution in

terms of the associate Legendre functions Pl2 and Q2l . Using their asymptotic behavior and comparing with (B9)
we find [6]


8
2
2
(1 2C) [2C(y 1) y + 2] 3(1 2C) [2C(y 1) y + 2] log (1 2C)
5

1
+ 2C 4C 4 (y + 1) + 2C 3 (3y 3) + 2C 2 (13 11y) + 3C(5y 8) 3y + 6
,
(C11)






6
2
2
k3E =
15(1 2C) 2C 2 (y 1) 3C(y 2) + y 3 log(1 2C)
(1 2C) 2C 2 (y 1) 3C(y 2) + y 3
7

1
+ 2C 4C 5 (y + 1) + 2C 4 (9y 2) 20C 3 (7y 9) + 5C 2 (37y 72) 45C(2y 5) + 15(y 3)
.
(C12)
k2E =

for l = 2 and l = 3, respectively, and where C = M/Rext

and y = rH 0 /H evaluated at Rext . The values of klE are
independent of the extraction radius Rext if the latter is
sufficiently large.

that we solve along with regular boundary conditions at

the origin,
(l+1) l+1

h0 h0

+ O rl+3 .

(C14)

(l+1)

b.

Axial perturbations

In this case, the metric perturbation is given by (B4),

while the scalar perturbation is given again by (C6). The
perturbed Einstein equations require h1 = 1 = 0, and
we are the left with a single radial equation for the perturbed function h0 ,
h000

0g + 0 0 r(0g + 0 ) (l2 + l 2)eg 2

h0 +
h0 = 0 ,
2
r2
(C13)

k2B =

k3B =

Notice that the value of h0

is not given but it can be
fixed arbitrarily to 1 and corrected a posteriori once the
intensity of the tidal field is known.
Outside the star, the equation for the odd perturbation (C13) reduces to the simple differential equation
h000 +

4M l(l + 1)r
h0 = 0,
r2 (r 2M )

whose solution can be written in terms of elementary

functions once l is fixed. By matching its asymptotic
behaviour to (B10), we find (for l = 2 and l = 3)

8
2C(y 2) y + 3
,
5 2C [2C 3 (y + 1) + 2C 2 y + 3C(y 1) 3y + 9] + 3[2C(y 2) y + 3] log(1 2C)

 

8 2
8C (y 2) 10C(y 3) + 3(y 4) 15 8C 2 (y 2) 10C(y 3) + 3(y 4) log(1 2C)
7

1
+ 2C 4C 4 (y + 1) + 10C 3 y + 30C 2 (y 1) 15C(7y 18) + 45(y 4)
,

where again C = M/Rext but now y = rh00 /h0 evaluated at Rext . Even in this case, the values of klB are
independent of the extraction radius Rext if the latter is
sufficiently large.

Appendix D: Tidal perturbations of BH-like ECOs

The exterior spacetime of all models considered in

Sec. III B is commonly described by the Schwarzschild
metric. Therefore, the perturbation formalism is iden-

(C15)

(C16)

(C17)

tical to the one developed for a non-rotating uncharged

BH, where the metric is perturbed according to Eq. (B1)
and the even and odd sector perturbations can be described as Eqs. (B3) and (B4). On the other hand, the
interior and the junction/boundary conditions at the radius r0 are model dependent. In this Appendix we discuss
the procedure to compute the TLNs for these objects.

22
1.

Polar-type TLNs of BH-like ECOs

a.

Exterior spacetime

Let us first consider the exterior spacetime. Einsteins equations for static polar-type perturbations of
the Schwarzschild metric lead to (the notation follows
Appendix B)


2f dH0
l(l + 1) 4M 2
d 2 H0
+
+ 4
H0 = 0 ,(D1)
f
dr2
r dr
r2
r
dK
f 1
=
H0 + H00 ,
(D2)
dr
rf
[1 + f (l2 + l 2 f )]H0 rf (f 1)H00
. (D3)
K=
f (l2 + l 2)
Here f = 1 2M/r and primes stand for derivatives with
respect to r. The above equations can be solved for [6, 8]
H0ext = C1 Pl2 (r/M 1) + C2 Q2l (r/M 1) ,

(D4)

for any value of l, and where C1 and C2 are two integration constants. The term proportional to C1 diverges at large distances and is identified with the external tidal field, whereas the term proportional to C2
is the bodys response. The metric function K follows
straightforwardly from Eq. (D3).

b.

Interior spacetime

The interior spacetime depends on the model under

consideration. In the wormhole model, we consider that
the other universe is an exact copy of exterior metric, so
that polar perturbations are described by Eq. (D4) with
two independent constants,
H0int = C3 Pl2 (r/M 1) + C4 Q2l (r/M 1) ,

(D5)

On the other side of the wormhole, we require that

there are no tidal fields, i.e. C3 = 0.
In the perfect-mirror model, perturbations do not penetrate the surface and the interior solution is irrelevant.
On the other hand, in the gravastar model the interior
solution which is regular at the origin reads




r02 rl
l1 l
3 2Cr2
H0int
F
,
,
;
l
+
;
2 1
r02 2Cr2
2 2
2 r02
(D6)
for any value of l.

c.

an overall amplitude is irrelevant so we need to impose

two junction conditions at r = r0 in the wormhole and
gravastar cases, and one boundary condition at r = r0 in
the perfect-mirror case.
In the former cases, we can impose the DarmoisIsrael junction conditions [102], which relate the discontinuity of the extrinsic curvature across the radius
with the properties of a thin shell of matter located
at r = r0 . By adapting the formalism developed in
Ref. [38], for polar perturbations,
we find that [[K]] =
p
0 and [[dK/dr ]] = 8 f (r0 ), where is the
perturbation of the surface energy density of the thin
shell and the symbol [[. . .]] denotes the jump of a
given quantity across the spherical shell, i.e. [[A]]
lim0 A(r r0 + ) A(r r0 ). For simplicity,
we assume that the thin-shell material is stiff, so that
0. Therefore, in the polar sector we impose the
following conditions

Matching conditions and TLNs

In the wormhole and in the gravastar case the interior

and the exterior solutions are overall described by three
independent constants, whereas the perfect-mirror model
is described by two constants. Since the problem is linear,

[[K]] = 0 ,

[[dK/dr ]] = 0 .

(D7)

These two conditions completely specify the matching between the interior and the exterior solution in the wormhole and gravastar cases. In the latter case, the junction
conditions (D7) agree with those derived in Ref. [59].
In the perfect-mirror case, we shall impose a Z2 symmetry on the surface and, therefore, the wavefunction
vanishes at r = r0 . In this case, one can solve for the
Zerilli function Z in the static limit and then reconstruct the metric function H0 through [103]
H0 =




1
l r2 6M 2 3l M r 0Z (r)
r(6M + l r)
Z (r)
36M 3 /r + 18l M 2 + 3l 2 M r
+
6M + l r


+l 2 (l /2 + 1)r2 ,
(D8)

where l = (l 1)(l + 2). In the static limit, H1 = 0

and K can be also written in terms of Z and of its
derivative. The Zerilli function satisfies a second-order
differential equation [103], which can be solved for analytically in the static limit. By imposing the Dirichlet
boundary condition Z = 0, the ratio C1 /C2 in Eq. (D4)
is completely specified.
Finally, once the perturbation equations are completely specified (modulo an overall amplitude) through
the junction/boundary conditions, it is straightforward
to compare the analytical expression for H0 at large distance with Eq. (B9), extract the multipole moments and
the tidal-field amplitudes, and finally compute the polar TLNs by using Eq. (3). Interestingly, this procedure yields exact results in closed form, although the
general expression for the polar TLNs is cumbersome.
The full result is provided online [57], whereas the highcompactness regime is discussed in the main text.

23
2.

regular near the origin reads



l1 l+2
3 2Cr2
l+1
,
,
;l + ; 2
h0 (r) r
2 F1
2
2
2 r0

Axial-type TLNs of BH-like ECOs

a.

Exterior spacetime

Einsteins equations for static axial-type perturbations

of the Schwarzschild metric imply h1 = 0 and
(l)

DA h0 h000 +

4M l(l + 1)r
h0 = 0 ,
r2 (r 2M )

for any value of l. Finally, as in the polar case, the interior

solution of the perfect-mirror model is irrelevant for the
purposes of computing the TLNs.

(D9)
c.

which can be solved for h0 in terms of hypergeometric

functions. The general solution reads [8]

r 
2
hext
0 = c1 r 2 F1 1 l, l + 2; 4;
2M

!

r  1 l, l + 2
2,0
,
+ c2 G2,2

2M  1, 2

(D10)

where G2,0
2,2 is the Meijer function and 2 F1 is one of the
hypergeometric functions. The terms proportional to c1
and c2 are identified with the external tidal field and
with the body response, respectively. The above solution
reduces to simple expressions for integer values of l, which
can be written in terms of polynomial and logarithmic
functions.

Interior spacetime

Also in the axial case, the interior spacetime is model

dependent. In the wormhole case we consider the same
solution as in Eq. (D10) but with zero tidal field, namely

hint
0

c2 G2,0
2,2

r
2M


!
 1 l, l + 2

.

 1, 2

(D11)

In the gravastar case, the interior solution which is

[1] A. Toomre and J. Toomre, Astrophys. J. 178, 623

(1972).
[2] W.H. Press and S.A. Teukolsky, Astrophys. J. 213, 183
(1977).
[3] C. Murray and S. Dermott, Solar System Dynamics
(Cambridge University Press, Cambridge, UK, 2000).

11

Note that our definition differs from the standard one by a factor
, which has been included so that h1 0 in the static limit,
whereas h0 remains finite, as expected.

Matching conditions and TLNs

The junction conditions for axial perturbations are easier because they do not couple to the matter of a putative thin shell. In the dynamical case, they simply read
[[h0 ]] = 0 = [[h1 ]] [38]. In the static case, h1 vanishes
identically and one is left with a single second-order differential equation for h0 . Therefore, regularity of the
axial perturbations across the shell imposes that h0 and
its derivative with respect to r be smooth. Thus, for
the wormhole and gravastar cases in the axial sector we
impose
[[h0 ]] = 0 ,

[[dh0 /dr ]] = 0 .

(D13)

For the perfect-mirror model, we follow the same procedure outlined above, namely we impose a Dirichlet condition on the Regge-Wheeler function RW evaluated at
r = r0 . This function is defined as11
h0 =

b.

(D12)

d(rRW )
,
dr

h1 =

ir
RW , (D14)
1 2M/r

and satisfies the Regge-Wheeler equation [95]. The latter can be solved analytically in the static limit, = 0.
Again, the ratio of the two integration constants in
Eq. (D10) is fixed by imposing RW = 0.
After the perturbations are fully specified through the
junction/boundary conditions, the axial TLNs can be
computed by comparing the large-distance behavior of
h0 with Eq. (B10), extracting the multipole moments,
and finally using the definition (3). As in the polar case,
we find a closed-form, cumbersome expression for the axial TLN [57], whose high-compactness limit is provided
in the main text for the various models.

[4] E. Poisson and C. Will, Gravity: Newtonian, PostNewtonian, Relativistic (Cambridge University Press,
Cambridge, UK, 2014).
[5] E.E. Flanagan and T. Hinderer, Phys. Rev. D77,
021502 (2008), arXiv:0709.1915.
[6] T. Hinderer, Astrophys. J. 677, 1216 (2008), Erratum:
ibid. 697, 964 (2009), arXiv:0711.2420.
[7] T. Hinderer et al., Phys. Rev. Lett. 116, 181101 (2016),
arXiv:1602.00599.
[8] T. Binnington and E. Poisson, Phys. Rev. D80, 084018
(2009), arXiv:0906.1366.

24
[9] T. Damour and A. Nagar, Phys. Rev. D80, 084035
(2009), arXiv:0906.0096.
[10] J. Lattimer and M. Prakash, Science 304, 536 (2004),
arXiv:astro-ph/0405262.
[11] T. Hinderer, B.D. Lackey, R.N. Lang, and J.S. Read,
Phys. Rev. D81, 123016 (2010), arXiv:0911.3535.
[12] S. Postnikov, M. Prakash, and J.M. Lattimer, Phys.
Rev. D82, 024016 (2010), arXiv:1004.5098.
[13] J. Vines, E.E. Flanagan, and T. Hinderer, Phys. Rev.
D83, 084051 (2011), arXiv:1101.1673.
[14] T. Damour, A. Nagar, and L. Villain, Phys. Rev. D85,
123007 (2012), arXiv:1203.4352.
[15] W. Del Pozzo, T.G.F. Li, M. Agathos, C. Van
Den Broeck, and S. Vitale, Phys. Rev. Lett. 111, 071101
(2013), arXiv:1307.8338.
[16] A. Maselli, L. Gualtieri, and V. Ferrari, Phys. Rev. D88,
104040 (2013), arXiv:1310.5381.
[17] N. Sennett et al., (in preparation, 2017).
[18] H. Fang and G. Lovelace, Phys. Rev. D72, 124016
(2005), arXiv:gr-qc/0505156.
[19] N. G
urlebeck, Phys. Rev. Lett. 114, 151102 (2015),
arXiv:1503.03240.
[20] E. Poisson, Phys. Rev. D91, 044004 (2015),
arXiv:1411.4711.
[21] P. Pani, L. Gualtieri, A. Maselli, and V. Ferrari, Phys.
Rev. D92, 024010 (2015), arXiv:1503.07365.
[22] P. Landry and E. Poisson, Phys. Rev. D91, 104018
(2015), arXiv:1503.07366.
[23] R.A. Porto, Fortsch. Phys. 64, 723 (2016),
arXiv:1606.08895.
[24] S.D. Mathur, Proceedings, CERN Winter School on
Strings, Supergravity and Gauge Theories. Geneva,
Switzerland, February 913 2009, Class. Quantum
Grav. 26, 224001 (2009), arXiv:0909.1038.
[25] F.E. Schunck and E.W. Mielke, Class. Quantum Grav.
20, R301 (2003), arXiv:0801.0307.
[26] S.L. Liebling and C. Palenzuela, Living Rev. Relat. 15,
6 (2012), arXiv:1202.5809.
[27] R. Brito, V. Cardoso, C.F.B. Macedo, H. Okawa,
and C. Palenzuela, Phys. Rev. D93, 044045 (2016),
arXiv:1512.00466.
[28] P. Grandclement, (2016), arXiv:1612.07507.
[29] P.O. Mazur and E. Mottola,
(2001), arXiv:grqc/0109035.
[30] M. Visser and D.L. Wiltshire, Class. Quantum Grav.
21, 1135 (2004), arXiv:gr-qc/0310107.
[31] M. Visser, Lorentzian wormholes: from Einstein to
Hawking (AIP, Woodbury, NY, USA, 1996).
[32] E.G. Gimon and P. Horava, Phys. Lett. B672, 299
(2009), arXiv:0706.2873.
[33] K. Skenderis and M. Taylor, Phys. Rept. 467, 117
(2008), arXiv:0804.0552.
[34] B. Holdom and J. Ren, (2016), arXiv:1612.04889.
[35] M. Saravani, N. Afshordi, and R.B. Mann, Int. J. Mod.
Phys. D23, 1443007 (2015), arXiv:1212.4176.
[36] S.B. Giddings, Phys. Rev. D90, 124033 (2014),
arXiv:1406.7001.
[37] J. Abedi, H. Dykaar, and N. Afshordi,
(2016),
arXiv:1612.00266.
[38] P. Pani, E. Berti, V. Cardoso, Y. Chen, and R. Norte,
Phys. Rev. D80, 124047 (2009), arXiv:0909.0287.
[39] C.F.B. Macedo, P. Pani, V. Cardoso, and
L.C.B. Crispino, Phys. Rev. D88, 064046 (2013),
arXiv:1307.4812.

[40] G.F. Giudice, M. McCullough, and A. Urbano,

J. Cosmol. Astropart. Phys. 1610, 001 (2016),
arXiv:1605.01209.
[41] N. Yunes and X. Siemens, Living Rev. Relat. 16, 9
(2013), arXiv:1304.3473.
[42] E. Berti et al., Class. Quantum Grav. 32, 243001 (2015),
arXiv:1501.07274.
[43] E. Barausse, V. Cardoso, and P. Pani, Phys. Rev. D89,
104059 (2014), arXiv:1404.7149.
[44] V. Cardoso, C.F.B. Macedo, P. Pani, and V. Ferrari, J. Cosmol. Astropart. Phys. 1605, 054 (2016),
arXiv:1604.07845.
[45] S.W. Hawking, Comm. Math. Phys. 25, 167 (1972).
[46] T.P. Sotiriou and V. Faraoni, Phys. Rev. Lett. 108,
081103 (2012), arXiv:1109.6324.
[47] E. Barausse and T.P. Sotiriou, Phys. Rev. Lett. 101,
099001 (2008), arXiv:0803.3433.
[48] N. Yunes and L.C. Stein, Phys. Rev. D83, 104002
(2011), arXiv:1101.2921.
[49] P. Pani, C.F.B. Macedo, L.C.B. Crispino, and V. Cardoso, Phys. Rev. D84, 087501 (2011), arXiv:1109.3996.
[50] S. Mignemi and N.R. Stewart, Phys. Rev. D47, 5259
(1993), arXiv:hep-th/9212146.
[51] P. Kanti, N.E. Mavromatos, J. Rizos, K. Tamvakis, and
E. Winstanley, Phys. Rev. D54, 5049 (1996), arXiv:hepth/9511071.
[52] P. Pani and V. Cardoso, Phys. Rev. D79, 084031 (2009),
arXiv:0902.1569.
[53] N. Yunes and F. Pretorius, Phys. Rev. D79, 084043
(2009), arXiv:0902.4669.
[54] S. Alexander and N. Yunes, Phys. Rept. 480, 1 (2009),
arXiv:0907.2562.
[55] V. Cardoso, S. Hopper, C.F.B. Macedo, C. Palenzuela, and P. Pani, Phys. Rev. D94, 084031 (2016),
arXiv:1608.08637.
[56] V. Cardoso, E. Franzin, and P. Pani, Phys. Rev. Lett.
116, 171101 (2016), Erratum: ibid. 117, 089902 (2016),
arXiv:1602.07309.
[57] http://blackholes.ist.utl.pt/?page=Files.
[58] P. Pani,
Phys. Rev. D92,
124030 (2015),
arXiv:1506.06050.
[59] N. Uchikata, S. Yoshida, and P. Pani, Phys. Rev. D94,
064015 (2016), arXiv:1607.03593.
[60] M. Punturo et al., Proceedings, 14th Workshop on Gravitational wave data analysis. Rome, Italy, January 26
29, 2010, Class. Quantum Grav. 27, 194002 (2010).
[61] P. Amaro-Seoane, S. Aoudia, S. Babak, P. Binetruy,
E. Berti, et al., Class. Quantum Grav. 29, 124016
(2012), arXiv:1202.0839.
[62] R.P. Geroch, J. Math. Phys. 11, 2580 (1970).
[63] R.O. Hansen, J. Math. Phys. 15, 46 (1974).
[64] Y. G
ursel, Gen. Rel. Gravit. 15, 737 (1983).
[65] K.S. Thorne, Rev. Mod. Phys. 52, 299 (1980).
[66] P. Pani, L. Gualtieri, and V. Ferrari, Phys. Rev. D92,
124003 (2015), arXiv:1509.02171.
[67] P. Landry and E. Poisson, Phys. Rev. D92, 124041
(2015), arXiv:1510.09170.
[68] P. Jetzer, Phys. Rept. 220, 163 (1992).
[69] C.F.B. Macedo, P. Pani, V. Cardoso, and
L.C.B. Crispino, Astrophys. J. 774, 48 (2013),
arXiv:1302.2646.
[70] D.J. Kaup, Phys. Rev. 172, 1331 (1968).
[71] R. Ruffini and S. Bonazzola, Phys. Rev. 187, 1767
(1969).

25
[72] M. Colpi, S.L. Shapiro, and I. Wasserman, Phys. Rev.
Lett. 57, 2485 (1986).
[73] R. Friedberg, T.D. Lee, and Y. Pang, Phys. Rev. D35,
3658 (1987).
[74] R.F.P. Mendes and H. Yang, (2016), arXiv:1606.03035.
[75] N. Uchikata and S. Yoshida, Class. Quantum Grav. 33,
025005 (2016), arXiv:1506.06485.
[76] R. Jackiw and S.Y. Pi, Phys. Rev. D68, 104012 (2003),
arXiv:gr-qc/0308071.
[77] N. Yunes and C.F. Sopuerta, Phys. Rev. D77, 064007
(2008), arXiv:0712.1028.
[78] C. Molina, P. Pani, V. Cardoso, and L. Gualtieri, Phys.
Rev. D81, 124021 (2010), arXiv:1004.4007.
[79] V. Cardoso and L. Gualtieri, Phys. Rev. D80,
064008 (2009), Erratum: ibid. D81, 089903 (2010),
arXiv:0907.5008.
[80] V. Cardoso, E. Franzin, A. Maselli, P. Pani, and G. Raposo, In preparation (2017).
[81] M. Vallisneri, Phys. Rev. D77, 042001 (2008), arXiv:grqc/0703086.
[82] C. Cutler and E.E. Flanagan, Phys. Rev. D49, 2658
(1994).
[83] T. Damour, B.R. Iyer, and B.S. Sathyaprakash, Phys.
Rev. D62, 084036 (2000).
[84] J. Vines, E.E. Flanagan, and T. Hinderer, Phys. Rev.
D83, 084051 (2011).
[85] D. Bini, T. Damour, and G. Faye, Phys. Rev. D85,
124034 (2012).
[86] D. Shoemaker (LIGO), Advanced LIGO anticipated sensitivity curves, Tech. Rep. T0900288-v3 (2010).
[87] B.S. Sathyaprakash and B.F. Schutz, Living Rev. Relat.
12, 2 (2009), arXiv:0903.0338.
[88] A. Klein, E. Barausse, A. Sesana, A. Petiteau, E. Berti,

[89]
[90]

[91]

[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]

[102]
[103]

S. Babak, J. Gair, S. Aoudia, I. Hinder, F. Ohme, and

B. Wardell, Phys. Rev. D93, 024003 (2016).
E. Berti, A. Buonanno, and C.M. Will, Phys. Rev. D71,
084025 (2005).
J. Steinhoff, T. Hinderer, A. Buonanno, and
A. Taracchini, Phys. Rev. D94, 104028 (2016),
arXiv:1608.01907.
K. Yagi, L.C. Stein, N. Yunes, and T. Tanaka, Phys.
Rev. D85, 064022 (2012), Erratum: ibid. D93, 029902
(2016), arXiv:1110.5950.
P. Pani, V. Cardoso, and L. Gualtieri, Phys. Rev. D83,
104048 (2011), arXiv:1104.1183.
H. Mueller and B.D. Serot, Nucl. Phys. A606, 508
(1996), arXiv:nucl-th/9603037.
F. Douchin and P. Haensel, Astron. Astrophys. 380, 151
(2001), arXiv:astro-ph/0111092.
T. Regge and J.A. Wheeler, Phys. Rev. 108, 1063
(1957).
J.G. Rosa and S.R. Dolan, Phys. Rev. D85, 044043
(2012), arXiv:1110.4494.
P. Pani, E. Berti, and L. Gualtieri, Phys. Rev. D88,
064048 (2013), arXiv:1307.7315.
M. Gleiser, Phys. Rev. D38, 2376 (1988), Erratum: ibid.
D39, 1257 (1988).
M. Gleiser and R. Watkins, Nucl. Phys. B319, 733
(1989).
S.H. Hawley and M.W. Choptuik, Phys. Rev. D62,
104024 (2000), arXiv:gr-qc/0007039.
V. Cardoso, L.C.B. Crispino, C.F.B. Macedo,
H. Okawa, and P. Pani, Phys. Rev. D90, 044069
(2014), arXiv:1406.5510.
W. Israel, Nuovo Cim. S10, B44, 1 (1966), Erratum:
ibid. B48, 463 (1967).
F.J. Zerilli, Phys. Rev. Lett. 24, 737 (1970).