Quasi-Normal Modes and Gravitational Wave Astronomy: The Date of Receipt and Acceptance Should Be Inserted Later
Quasi-Normal Modes and Gravitational Wave Astronomy: The Date of Receipt and Acceptance Should Be Inserted Later
Quasi-Normal Modes and Gravitational Wave Astronomy: The Date of Receipt and Acceptance Should Be Inserted Later
1 Introduction
In the past four years the sensitivity of the gravitational wave (GW) detectors
LIGO and Virgo has been improved at a formidable rate [1]. LIGO’s noise
curve has been lowered by about three orders of magnitude and now the
detectors are operating at the design sensitivity. Similar progresses have been
achieved by Virgo, although some effort remains to be done to reach the
planned sensitivity at low frequencies (∼ (10 − 40) Hz). In any event, these
detectors are now in a position to take data good enough to start making
science: a supernova exploding in the local group of nearby galaxies would
not be missed, as well as the coalescence of compact bodies, neutron stars or
black holes, with total mass smaller than about 100 M⊙ , out to a distance
of the orders of few Megaparsecs (these estimates are only indicative, since
they are continuously updated as the detector sensitivities are improved).
The detection of gravitational signals will allow to test the predictions of the
theoretical work that has been done over the years to construct waveforms
and energy spectra, and to extract distinctive features which could be traced
back to the nature and to the structure of the source. An important piece
Dipartimento di Fisica “G. Marconi”, Sapienza Università di Roma and INFN,
Sezione Roma 1, P.le A. Moro 2, 00185 Roma, Italy
2
In 1957 T. Regge and J.A. Wheeler [6] showed that the equations describing
the perturbations of a Schwarzschild black hole can be separated if the per-
turbed metric tensor is expanded in tensorial spherical harmonics. They also
showed that the relevant equations split into two decoupled sets belonging
to different parity — (−1)ℓ (even or polar) and (−1)ℓ+1 (odd or axial) —
and that, by a suitable choice of the gauge and by Fourier-expanding the per-
turbed functions, the equations for the radial part of the axial perturbations
of a Schwarzschild black hole can be reduced to a single Schroedinger-like
wave equation with a potential barrier, for a suitably defined function Zℓ− :
d2 Zℓ−
+ [ω 2 − Vℓ− (r)]Zℓ− = 0, (1)
dr∗2
where
1 2M r
Vℓ− (r) = 1− [ℓ(ℓ + 1)r − 6M ] , r∗ = r + 2M log( − 1). (2)
r3 r 2M
This equation is known as the Regge-Wheeler equation. A similar result
was obtained later in 1970 by F. Zerilli [7], who showed that also the polar
1
The study of stellar oscillations started at the beginning of the past century,
when Shapley [4] (1914) and Eddington [5] (1918) suggested that the variability
observed in some stars is due to periodic pulsations.
3
equations can be reduced to the wave equation (1) for a suitably defined
function Zℓ+ , and with a different potential barrier
2(r − 2M )
Vℓ+ (r) = [n2 (n + 1)r3 + 3M n2 r2 + 9M 2 nr + 9M 3 ] , (3)
r4 (nr + 3M )2
Z − (r, ω) X ℓm (θ, φ)
Z iω(t−r∗ ) X
1 e +
h+ (t, r, θ, φ) = Zℓm (r, ω)W ℓm (θ, φ) − ℓm dω
2π r iω sin θ
ℓm
Z iω(t−r∗ ) X −
X lm (θ, φ) Zℓm
1 e + (r, ω) ℓm
h× (t, r, θ, φ) = Zℓm (r, ω) + W (θ, φ) dω
2π r sin θ iω
ℓm
(4)
+ ×
where h , h are the two polarizations of the gravitational wave in the
transverse-traceless gauge (see [8], Chapter 35), and
ℓm 2 1
W (θ, φ) = ∂θ − cot θ∂θ − ∂φ Y ℓm (θ, φ)
2
sin2 θ
X ℓm (θ, φ) = 2 (∂θφ − cot θ∂φ ) Y ℓm (θ, φ) (5)
where Y ℓm are scalar spherical harmonics. Note that, since we are considering
±
perturbations of a spherically symmetric spacetime, Zℓm coincides with Zℓ±
for any value of m.
In 1970 Vishveshwara [9] pointed out that equation (1) for the functions
Zℓ− and Zℓ+ allows complex frequency solutions which satisfy the following
boundary conditions
Zℓ → eiωr∗ , r∗ → −∞,
Zℓ → e−iωr∗ , r∗ → +∞;
the former represents a pure ingoing wave, since nothing can escape from a
black hole horizon, the latter represents a pure outgoing wave at radial infin-
ity and corresponds to the requirement that no radiation is incoming from
infinity. This idea was confirmed by Press [10] who found, by integrating the
wave equation numerically, that an arbitrary initial perturbation decays as a
pure frequency mode. However, only in 1975 Chandrasekhar and Detweiler
4
[11] actually computed the discrete eigenfrequencies of these modes and clar-
ified their nature. Quoting Chandrasekhar from its book The Mathematical
Theory of Black Holes [12]:
“.. we may expect on general grounds that any initial perturbation will,
during its last stages, decay in a manner characteristic of the black hole
and independently of the original cause. In other words, we may expect that
during the very last stages, the black hole will emit gravitational waves with
frequencies and rates of damping, characteristic of itself, in the manner of
a bell sounding its last dying pure note. These considerations underlie the
formulation of the concept of the quasi-normal modes of a black hole.”
A Schwarzschild black hole is characterized by only one parameter, its
mass M ; consequently, the QNM frequencies depend only on M . In Table 1
we show the values of the complex characteristic frequencies of the first four
QNMs of a Schwarzschild black hole, respectively for ℓ = 2 and ℓ = 3.
In order to find the true pulsation frequency, ν, and the damping time, τ ,
M ω0 + iM ωi M ω0 + iM ωi
ℓ=2 0.3737+i0.0890 ℓ=3 0.5994+i0.0927
0.3467+i0.2739 0.5826+i0.2813
0.3011+i0.4783 0.5517+i0.4791
0.2515+i0.7051 0.5120+i0.6903
from the values given in Table 1, we proceed as follows. Let us assume that
the black hole mass is M = nM⊙ , (M⊙ = 1.48 · 105 cm); converting to
physical unities we find
c 32.26 nM⊙ n · 0.4937 · 10−5
ν= = (M ω0 ) kHz, τ= = s.
2πn · M⊙ (M ω0 ) n (M ωi )c (M ωi )
(6)
Using these expressions we can check whether a gravitational signal emitted
by an oscillating black hole falls within the bandwidth of the ground based
interferometers Virgo/LIGO or within that of the space based interferometer
LISA. Virgo/LIGO bandwidth extends over a range of frequencies which
goes from about 10-40 Hz, up to few kHz. Thus, these detectors will be able
to detect the signal emitted by an oscillating black hole (if it is sufficiently
strong) with mass ranging within
3
∼M <
10 M⊙ < ∼ 10 M⊙ ,
corresponding to the frequency range ν ∈ [12 Hz, 1.2 kHz]; LISA will be
sensitive to the frequency region ν ∈ [10−4 , 10−1 ] Hz, and will see oscillating
black holes with mass
1.2 · 105 M⊙ <
∼M <
8
∼ 1.2 · 10 M⊙ .
5
For instance, LISA will be able to detect signals emitted by the oscillations
of the massive black hole at the center of our Galaxy SGR A*, the mass of
which is M = (3.7 ± 0.2) · 106 M⊙ [13].
While the frequencies of the lowest modes are rather easy to compute,
great care must be used to determine the entire spectrum. Many different
methods have been used to this purpose. For instance, a WKB approximation
and a higher order WKB approach have been used to find the lowest [14] and
the higher mode frequencies [15], respectively. In addition, new approaches
have been developed to study the QNM spectrum, as the continued frac-
tion method (developed for Kerr black holes in [16]) and the phase-integral
method [17]. Using these approaches, it has been found [18] that, for any
value of the harmonic index ℓ, as the order n of the mode increases, the real
part of a mode frequency approaches a non zero limiting value. Furthermore,
an analytical expression has been found for the imaginary part of the fre-
quency, valid in the limit n → ∞. We also mention that an exact, analytical
solution of the Regge-Wheeler equation has recently been found in terms of
the Heun functions. QNM eigenfrequencies can be computed in terms of this
solution, by solving numerically a boundary value problem [19].
After 1975 the study of black hole perturbations follows along two princi-
pal avenues. One studies directly the perturbations of the metric tensor via
Einstein’s equations linearized about a given background. The other studies
the perturbations of Weyl’s and Ricci’s scalars using the Newman-Penrose
formalism. Using this latter approach in 1972 Teukolsky [20] was able to
decouple and separate the equations governing the perturbations of a Kerr
black hole, and to reduce them to a single master equation for the radial part
of the perturbation Rℓm :
The angular part, Sℓm , satisfies the equations of the oblate spheroidal har-
monics
( h 2
i
[(1 − u2 )Sℓm,u ],u + a2 ω 2 u2 − 2amωsu + s + Aℓm − (m+su)
1−u2 Sℓm = 0, (8)
u = cos θ,
∞ ℓ
1
Z X X
ψs (t, r, θ, ϕ) = eiωt eimϕ Sℓm (u)Rℓm (r)dω. (9)
2π
ℓ=|s| m=−ℓ
6
source dates back to 1971: in [29] a Schwarzschild black hole was perturbed
by a radially infalling point-like body, with a mass much smaller than the
black hole mass. The energy spectrum of the emitted signal was computed
by solving the Zerilli equation with a source describing the infalling particle.
The waveform was explicitly computed in [30], and it was shown that, after
a transient, the signal exhibits a ringing tail, which can be fitted by a com-
bination of quasi-normal modes. In Figure 1, we show the gravitational wave
amplitude r · h(t − r∗ ) emitted in the considered process, and the analytical
fit with the first two ℓ = 2 quasi-normal modes belonging to the frequencies
M ω1 = 0.37 + i0.09 and M ω2 = 0.35 + i0.27. The fit becomes more accurate
if higher order modes are taken into account, but the main contribution is
due to ω1 and ω2 .
40
20
rh
-20
-40
the mode excitation. However, astrophysical phenomena are much more com-
plicate, and only recently major advances in numerical techniques allowed the
modelling of more realistic processes involving black holes.
Black hole coalescence is probably the most violent process occurring in
the universe (after the big bang), and it is expected to be the most power-
ful source of gravitational waves to be detected by interferometric detectors
Virgo and LIGO.
0.04
GSFC (R1)
UTB (s00)
Pretorius (d16)
0.02
r ψ4 Mf
0.004
0.002
-0.02
0
-0.002
-0.004
-250 -200 -150 -100
-0.04
-100 -50 0 50
t/Mf
in the poles of the Green’s functions associated to the Zerilli and Regge-
Wheeler equations. This information, combined with the knowledge of the
exciting source, allows to determine the quasi-normal mode content of the
gravitational signal.
2.5.1 Completeness
2.5.2 Stability
In recent years, it has been suggested that black hole QNM’s may play a role
in string theory and loop quantum gravity. In 2000 Horowitz and Hubeny [47]
proposed that the study of the black hole QNM’s in anti-de Sitter spacetime
could be useful to determine some properties of conformal field theories. Their
conjecture is deeply rooted in string theory and in the so-called “AdS-CFT
correspondence”. Stimulated by this work, many authors computed the QNM
eigenfrequencies in anti-de Sitter spacetime [48] (see also [49]). It is worth
reminding that anti-de Sitter solution of Einstein’s equations describes a
universe with a negative cosmological constant; therefore these black holes
should not be considered as astrophysical objects.
In 2003 Dreyer and Motl [50] suggested that, in the asymptotic limit
n → ∞, black hole quasi-normal modes would allow to fix the value of the
“Immirzi parameter”, which is a key parameter in loop quantum gravity.
Following this proposal, studies of the asymptotic limit of QNM [18], [28]
have further been developed [51].
More generally, inspired by these consideration in the contexts of string
theory and loop quantum gravity, in recent years many authors have com-
puted the eigenfrequencies of black hole quasi-normal modes in various back-
10
ground spacetimes, both in four dimensions and for higher dimensional space-
times [52].
3 Stellar pulsations
Stellar pulsations are a very well known phenomenon in astronomy, since they
underlay a variety of astrophysical processes. For instance, they are observed
in the Sun, and a branch of solar sciences, named helioseismology, uses the
information encoded in the pulsation frequencies to investigate the internal
structure of our star and the physical processes that occur in the interior.
Non radial pulsations are associated to gravitational wave emission and, as
we shall see, the mode frequencies carry interesting information on the inner
structure of the emitting sources. Thus, if in the future GW-detectors will
be able to catch the gravitational signals emitted by pulsating stars, a new
branch of astrophysics will develop, the gravitational wave asteroseismology.
This will allow us to investigate the interior of neutron stars, where densities
and pressures are so extreme that they are unreachable by high energy exper-
iments on Earth. But before discussing how the equation of state (EOS) of
matter affects quasi-normal mode frequencies, let us preliminarily show the
equations we need to solve to determine these frequencies. We shall discuss
only pulsations of a non rotating star, i.e. of stars which are described by
static, spherically symmetric solutions of Einstein’s equations. The rotating
case is much more complicate, and an exhaustive description of the prob-
lems that arise when one is looking for the quasi-normal mode frequencies is
beyond the scope of this paper (see, for instance, [53] and references therein).
In eq. (12) there are seven unknown functions, i.e. one more than needed,
but the extra degree of freedom disappears when the boundary conditions of
the problem are fixed. As a consequence of a generic perturbation, the metric
functions, (ν, ψ, µ2 , µ3 , ω, q2 , q3 ), and the fluid variables, (ǫ, p, uα ), change by
a small amount with respect to their unperturbed values, which we assume
to be known; for instance ν −→ ν + δν, ǫ −→ ǫ + δǫ, and similarly for the
remaining variables. At the same time, each element of fluid undergoes an
infinitesimal displacement from its equilibrium position, which is described
by the lagrangian displacement ξ. All perturbed quantities are functions of
t, r and θ. If we now write Einstein’s equations supplemented by the hy-
drodynamical equations and the conservation of baryon number, expand all
tensors in tensorial spherical harmonics and Fourier-expand the time depen-
dent quantities, we find that, as for black holes, the perturbed equations
decouple into two sets, the polar and the axial, with a major difference: the
polar equations couple the thermodynamical variables to the metric variables.
Conversely the axial perturbations do not induce fluid motion except for a
stationary rotation; however, we shall see that the fluid plays a role, because
it shapes the potential barrier associated to the spacetime curvature.
The explicit expressions of the functions that describe the polar perturba-
tions, expanded in harmonics and Fourier-expanded are
δν = Nℓ (r)Pℓ (cos θ)eiωt δµ2 = Lℓ (r)Pℓ (cos θ)eiωt (13)
δµ3 = [Tℓ (r)Pℓ + Vℓ (r)Pℓ,θ,θ ]eiωt δψ = [Tℓ (r)Pℓ + Vℓ (r)Pℓ,θ cot θ]eiωt ,
δp = Πℓ (r)Pℓ (cos θ)eiωt 2(ǫ + p)eν+µ2 ξr (r, θ)eiωt = Uℓ (r)Pℓ eiωt
δǫ = Eℓ (r)Pℓ (cos θ)eiωt 2(ǫ + p)eν+µ3 ξθ (r, θ)eiωt = Wℓ (r)Pℓ,θ eiωt ,
where Pℓ (cos θ) are Legendre’s polynomials and ω is the frequency. After
separating the variables the relevant Einstein’s equations become
Xℓ,r,r + 2r + ν,r − µ2,r Xℓ,r + rn2 e2µ2 (Nℓ + Lℓ ) + ω 2 e2(µ2 −ν) Xℓ = 0,
2 n 2µ2
− 1)(Nℓ + Lℓ ) + r(ν,r − µ2,r )Xℓ,r + ω 2 e2(µ2 −ν) rXℓ ,
(r Gℓ ),r = nν,r (Nℓ − Lℓ ) + r (e
1 2µ2 2
−ν,r Nℓ,r = −Gℓ + ν,r [Xℓ,r + νn ,r (Nℓ − Lℓ )] + r 2 (e − 1)(Nℓ − rXℓ,r − ro Gℓ )
2
2µ2 1 2 2(µ2 −ν) r 1
−e (ǫ + p)Nℓ + 2 ω e Nℓ + Lℓ + n Gℓ + n [rXℓ,r + (2n + 1)Xℓ ] ,
Lℓ,r (1 − D) + Lℓ r − ν,r − r + hν,r D + Xℓ,r + Xℓ r1 − ν,r + DN
2 1
i ℓ,r +
+Nℓ Dν,r − D − F + 1 + Eν,r Nℓ − Lℓ + r2 Gℓ + 1 (rXℓ,r + Xℓ ) = 0,
r r n n
(14)
where
A = 21 ω 2 e−2ν , Q = (ǫ+p)γp ,
(ǫ+p) ∂p e−2µ2 ν,r
γ = p ∂ǫ entropy=const , B= 2(ǫ+p) (ǫ,r − Qp,r ),
A ω 2 e−2ν (ǫ+p) (15)
D = 1 − 2(A+B) =1− ω 2 e−2ν (ǫ+p)+e−2µ2 ν,r (ǫ,r −Qp,r )
,
E = D(Q − 1) − Q,
F = ǫ,r −Qp,r = 2[ǫ,r −Qp,r ](ǫ+p)
,
2(A+B) 2ω 2 e−2ν (ǫ+p)+e−2µ2 ν,r (ǫ,r −Qp,r )
12
The equations for the axial perturbations are much simpler than the polar
ones. Their radial behaviour is completely described by a function Zℓ− (r),
which satisfies the following Schroedinger-like equation
d2 Zℓ−
+ [ω 2 − Vℓ− (r)]Zℓ− = 0, (18)
dr∗2
Rr
where r∗ = 0 e−ν+µ2 dr (ν and µ2 are unperturbed metric functions), and
e2ν(r)
Vℓ− (r) = ℓ(ℓ + 1)r + r3 [ǫ(r) − p(r)] − 6m(r) .
(19)
r3
3
After these equations were derived, R.Ipser and R.H.Price showed that they
can be reduced to a fourth-order system [61].
13
The quasi-normal modes are solutions of the axial and polar equations that
satisfy the following boundary conditions. As for black holes, at radial infinity
the solution must behave as a pure outgoing wave
Zℓ± → e−iωt , r∗ → +∞. (21)
In addition, all perturbed functions must be regular at r = 0 and have to
match continuously the exterior perturbation on the stellar surface. For the
axial perturbations the matching condition is automatically satisfied, because
eq. (18) reduces to the Regge-Wheeler equation for r ≥ R, where R is the
stellar radius.
Stars possess many different classes of modes. The axial quasi-normal
modes are pure spacetime modes and do not exist in Newtonian gravity.
They are named w-modes and are highly damped, i.e. the imaginary part
of the frequency is comparable to the real part [62] and consequently the
damping times are small. If the star is extremely compact, the potential (19)
inside the star becomes a well, while in the exterior it remains a barrier. If the
well is deep enough, it allows for the existence of one or more slowly damped
quasi-normal modes, or s-modes; they are also said trapped modes because,
due to the slow damping, they are effectively trapped by the potential barrier,
and no much radiation can leak out of the star when these modes are excited
[63].
It is interesting to compare the eigenfrequencies of the axial modes of stars
and black holes, since they are both pure spacetime modes. As an example,
in table 2 we show the frequencies and the damping times of the first four
ℓ = 2, axial modes for a homogeneous star with mass M = 1.35M⊙ and
increasing compactness, and for a non rotating black hole with the same
mass. It should be mentioned that the w-mode frequencies depend upon the
equation of state of matter in the inner core of the star and, as shown in [64],
if detected they would allow to discriminate between the models underlying
different equations of state. Until very recently, the common belief was that
w-modes are unlikely to be excited in astrophysical processes. However in
2005 it has been shown that, in the collapse of a neutron star to a black hole,
w-modes are excited soon before the black hole forms. Thus, the gravitational
14
Table 2 The characteristic frequencies and damping times of the axial quasi-
normal modes of a homogeneous star of mass M = 1.35 M⊙ . The data are tabulated
for increasing values of the stellar compactness M
R
; they are compared to those of
a non rotating black hole with the same mass. We tabulate the first four values of
the frequency (in kHz) and of the damping time (in s) for ℓ = 2. ν s and τ s refer
to the trapped modes associated to the potential well inside the star (see text), ν w
and τ w refer to the axial w-modes, and ν BH and τ BH to the black hole
M
R
νs τs νw τw ν BH τ BH
0.4167 8.6293 1.52 · 10−3 11.1738 1.70 · 10−4 8.9300 7.49 · 10−5
– – 14.2757 8.03 · 10−5 8.2848 2.43 · 10−5
signal emitted in this process contains both the frequency of the w-modes of
the collapsing star, and those of the quasi-normal modes of the newly born
black hole [38].
The polar modes are classified following a scheme, introduced by Cowl-
ing in Newtonian gravity in 1942 [65], based on the restoring force which
prevails when the generic fluid element is displaced from the equilibrium
position. They are said g-modes, or gravity modes, if the restoring force is
due to buoyancy, and p-modes if it is due to pressure gradients. The mode
frequencies are ordered as follows
..ωgn < .. < ωg1 < ωf < ωp1 < .. < ωpn ..
and are separated by the frequency of the fundamental mode (f -mode), which
has an intermediate character between g− and p− modes. In addition, gen-
eral relativity predicts the existence of polar w-modes, that are very weakly
coupled to fluid motion and are similar to the axial w-modes [66]. Their
frequencies are typically higher than those of the fluid modes (g, f and p).
The relevance of the different modes to gravitational wave emission de-
pends on several factors, first of all on the amount of energy which can be
stored into a given mode. Moreover, it depends on the presence of other dis-
sipative processes that may compete with GW-emission in removing energy
from the star; for instance, neutrino diffusion or viscosity, if the oscillating
15
star is a newly born, hot neutron star. Last, but not least, of crucial impor-
tance are the values of the mode frequencies: for instance the w-modes of
neutron stars have typical frequencies of the order of 10 kHz or higher, far
too high to be detectable by ground based interferometers in their present
or advanced configuration.
Numerical simulations of the most energetic astrophysical processes, like
core collapse to a neutron star (NS) or binary coalescence leading to NS
formation, indicate that the mode which is most excited is the fundamental
mode. For this reason we will now focus on this mode, discussing the infor-
mation the f -mode frequency, νf , carries on the inner structure of neutron
stars.
It is worth mentioning that typical values of f -mode damping times are
of the order of a few tenths of seconds; consequently, the excitation of the
f -mode would appear, in the Fourier transform of the gravitational signal,
as a sharp peak. Therefore, the f -mode frequency could, in principle, be
extracted from the detector noise by an appropriate data analysis (provided
νf falls in the detection bandwidth of some GW-detector).
From Newtonian gravity we know that the νf scales with the average
M 1/2
density of the star νf ∼ R 3 ; as shown in [67,68], where νf has been
computed for a variety of equations of state proposed to describe matter in
a NS, a similar scaling law holds also in general relativity.
At densities exceeding the equilibrium density of nuclear matter, ρ0 =
2.67 × 1014 g/cm3 , the fluid in the inner core of a NS is basically a gas of
interacting nucleons. The equations of state proposed in the literature to de-
scribe this kind of matter are derived within two main, different approaches:
nonrelativistic nuclear many-body theory (NMBT) and relativistic mean field
theory (RMFT); we will now show how different ways of modeling hadronic
interactions affect the pulsation properties of the star. In what follows we
shall summarize the main results of a study we have done in [68], where we
have selected a restricted number of EOS obtained within the NMBT and
the RMFT approach. To describe the inner and outer crust of the NS, we
have used the Baym-Pethick-Sutherland EOS [69] and the Pethick-Ravenhall-
Lorenz EOS [70], respectively.
The EOS we choose to describe NS matter at ρ > ρ0 , are the following.
For the NMBT approach we select two groups of EOS: Group I, named
(APR1 , APR2 , APRB200 , APRB120 ), and Group II, named (BBS1 ,
BBS2 ), respectively. In both cases matter is composed of neutrons, protons,
electrons and muons in weak equilibrium, and the dynamics is described by
a non-relativistic Hamiltonian which includes phenomenological potentials
that describe two- and three-nucleon interactions. The potential are obtained
from fits of existing scattering data. For all EOS the two-body potential is
v18 , whereas the three-body potential is Urbana IX for Group I, and Urbana
VII for Group II.
A first major difference between the two groups is that in Group I
the ground state energy is calculated using variational techniques [71,72],
whereas in Group II is calculated using G-matrix perturbation theory [73].
There are also differences among the EOS in each group:
16
– Group I
APR2 is an improved version of the APR1 model. In APR1 nucleon-
nucleon potentials describe interactions between nucleons in their center
of mass frame, in which the total momentum P vanishes. In the APR2 the
two-nucleon potential is modified including relativistic corrections which
arise from the boost to a frame in which P 6= 0, up to order P2 /m2 .
These corrections are necessary to use the nucleon-nucleon potential in
a locally inertial frame associated to the star. As a consequence of this
change, the three-body potential also needs to be modified in a consistent
fashion.
The EOS APRB200 and APRB120 are the same as APR2 up to ∼ 4ρ0 ,
but at higher density there is a phase of deconfined quark matter de-
scribed within the MIT bag model. The mass of the strange quark is
assumed to be ms = 150 MeV, the coupling constant describing quarks
interaction is set to αs = 0.5, and the value of the bag constant is 200
MeV/fm3 for APRB200 and 120 MeV/fm3 for APRB120 . We will dis-
cuss in some more detail quark matter in the next section. The phase
transition from nuclear matter to quark matter is described requiring the
fulfillment of Gibbs conditions, leading to the formation of a mixed phase,
and neglecting surface and Coulomb effects [71,74]. Thus, these stars are
hybrid stars.
– Group II
The main difference between the equations of state BBS1 and BBS2 is
that in BBS2 strange heavy baryons (Σ − and Λ0 ) are allowed to form in
the core. Neither BBS1 nor BBS2 include relativistic corrections.
As representative of the RMFT, we choose the EOS named G240 . Matter
composition includes leptons and the complete octet of baryons (nucleons,
Σ 0,± , Λ0 and Ξ ± ). Hadron dynamics is described in terms of exchange of
one scalar and two vector mesons. It should be reminded that in this case,
the EOS is obtained within the mean field approximation [75].
For any of the above EOS we have solved the TOV equations for different
values of the central density, finding the equilibrium configurations. Then, for
each EOS and for each equilibrium model, we have solved the equations of
stellar perturbations finding the f -mode frequency, νf . The results are shown
in Fig. 3 where we plot νf as a function of the mass, up to the maximum mass
allowed by each EOS. From this picture we learn the following. Comparing
the values of νf for APR1 and APR2 we immediately see that the relativistic
corrections and the associated redefinition of the three-body potential, which
improve the Hamiltonian of APR2 with respect to APR1 , play a relevant
role, leading to a systematic difference of about 150 Hz in the mode fre-
quency. Conversely, the presence of quark matter in the star inner core (EOS
APRB200 and APRB120 ) does not seem to significantly affect the pulsation
properties of the star. We also see that the frequencies corresponding to the
BBS1 and APR1 models, which are very close at M < ∼ 1.4 M⊙ , diverge for
larger masses. This behavior can be traced back to the different treatments
of three-nucleon interactions, whose role in shaping the EOS becomes more
and more important as the star mass (and central density) increases: while
the variational approach of ref. [71] used to derive the EOS APR1 naturally
17
f-mode frequency
2.6
2.5
2.4
2.3
2.2
νf ( KHz )
2.1
1.9
APR1
1.8 APR2
APRB200
1.7 APRB120
BBS1
1.6 BBS2
G240
1.5
1.2 1.4 1.6 1.8 2 2.2 2.4
M/Mo
Fig. 3 The frequency of the fundamental mode is plotted as a function of the mass
of the star for the selected EOS (see text).
3.2
APR2
3
Neutron stars APRB200
APRB120
+ BBS1
2.8 BBS2
Strange stars G240
2.6
νf ( kHz )
2.4
2.2
1.8
1.6
1.4
0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4
M/Mo
Fig. 4 The frequency of the fundamental mode is plotted as a function of the
mass of the star, for neutron/hybrid stars (continuous lines) and for strange stars
modeled using the MIT bag model, spanning the set of parameters indicated in
(22) (dashed region).
mass in the range we consider. Note that 1.8 M⊙ is the maximum mass above
which no stable strange star can exist. There is a small range of frequency
where neutron/hybrid stars are indistinguishable from strange stars; however,
there is a large frequency region where only strange stars can emit. For
instance if M = 1.4 M⊙ , a signal with νf & 2 kHz would belong to a strange
star. Even if we do not know the mass of the star (as it is often the case for
isolated pulsars) the knowledge of νf allows to gain information about the
source nature; indeed, if νf & 2.2 kHz, apart from a very narrow region of
masses where stars with hyperons would emit (EOS BBS2 and G240 ), we
can reasonably exclude that the signal is emitted by a neutron star.
In addition, it is possible to show that if a signal emitted by an oscillating
strange star would be detected, since νf is an increasing function of the bag
constant B it would be possible to set constraints on B much more stringent
than those provided by the available experimental data [78].
high sensitivity at frequencies in the range ∼ 1.5 − 3 kHz, typical for νf . Fea-
sibility studies of interferometric, high frequency detectors have been consid-
ered in recent years [83], and high frequency, wide-band, resonant detectors
are under study [84]; however, if we restrict to Virgo or LIGO in their present
configuration, to detect a signal emitted by a NS pulsating in the f -mode in
our Galaxy, with a signal-to-noise ratio of 5, the energy stored into the mode
should be Ef −mode ∼ 6 · 10−7 M⊙ c2 . In order to understand whether it is
plausible that the fundamental mode is excited to such an extent, we can refer
either to the results of numerical simulations, or to astrophysical data. Nu-
merical simulations of gravitational collapse show that the amount of energy
released in gravitational waves is in the range EGW tot ∼ [10−9 − 10−6 ] M⊙ c2
[85]. Although computed waveforms show that the f -mode is excited, at
present there is no conclusive indication on the fraction of EGW tot which
may go into that mode, since it depends on the initial conditions and on the
physical assumptions that are made in modeling the collapse. Just to men-
tion one, usually numerical simulations assume axisymmetric collapse, but
in the non axisymmetric case energy released in GWs may be higher.
Thus, we can only say that Ef −mode ∼ 6 · 10−7 M⊙ c2 is not totally
unreasonable. Unfortunately, gravitational collapse is a rare event (about 3
events every hundred years, per galaxy), and if we restrict to our Galaxy
chances to detect one in our lifetime are not too high.
The f -mode may also be excited in a cold, old neutron star as a conse-
quence of a glitch. Glitches are sudden changes in the rotation frequency of
the neutron star crust. They are observed in many pulsars and are thought
to be related to quakes occurring in the solid structures such as the crust,
the superfluid vortices and, perhaps, the lattice of quark matter in the stellar
core [86,87,88]. The rotational energy released in a glitch is ∆E ≈ IΩ∆Ω,
where I is the moment of inertia of the star, and typical spin variations are
∆Ω/Ω ≈ 10−6 −10−8 . For the glitches observed in the Crab and Vela pulsars
observations give ∆E ≃ 2 · 10−13 M⊙ c2 and ∆E ≃ 3 · 10−12 M⊙ c2 , respec-
tively. As before, we do not know which fraction of ∆E goes in the f -mode
excitation; in any event being ∆E so small, we can conclude that there is
no hope to detect anything like this with the detectors that are actually in
operation.
The second reason why we are far from being able to infer the EOS of
matter in the inner core of a NS using gravitational waves, is that the EOS
proposed in the literature only loosely constrain the dynamics of nuclear
matter. This statement can be made more clear by the following example.
Since from Newtonian gravity we know that νf scales as the square root
of the average density, we expect a similar relationqto hold also in general
M
relativity. Indeed, a linear relation between νf and R 3 has been obtained
in [67], fitting the data referring to stars modeled with a large set of EOS. In
[68] a similar fit has been found using the same EOS considered in previous
sections; since the two fits are similar, in what follows we shall explicitly use
our fit:
r
M
νf = a + b , a = 0.79 ± 0.09, b = 33 ± 2, (23)
R3
21
q
M
where a is in kHz and b in km·kHz. The fit is plotted in figure 5 versus R 3,
f-mode frequency
3
2.8
2.6
2.4
νf ( KHz )
2.2
2 Fit
APR1
APR2
1.8 APRB120
APRB200
1.6 BBS1
BBBS2
G240
1.4
0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.06 0.065
3 1/2 -1
(M/R ) (Km )
Fig. 5 The frequency of the fundamental mode is plotted as a function of the
square root of the average density for the EOS considered in this paper. We also
plot the fit given in eq. (23)
where a and b are in km·kHz. In these fits, frequencies are expressed in kHz,
masses and radii in km.
Let us now consider a star belonging to the EOS APR2 , with mass
M = 1.4 M⊙ and radius R = 11.58 km. Let us assume that, as a consequence
of some astrophysical process, both the fundamental mode and the first p-
mode are excited and that the emitted gravitational wave has been detected.
With the detected values of νf and νp1 (which we know to be νf = 1.983 kHz,
νp = 6.164 kHz) we could plot the fits (23) and (24) in the (R, M )-plane,
and we would find what is shown in figure 6: the two curves intersect in a
point which corresponds to M = 1.30 M⊙ , R = 11.36 km; consequently, we
would be able to estimate the values of the mass and of the radius with an
error of 7% and 2%, respectively. This would be great, but unfortunately,
the fit comes with error bars. If, for instance we plot the curve referring to
the f -mode, and we show the entire region where the parameters a and b
can vary (the dashed region in figure 7), we see that the error bar induces a
22
15
14
13
12
●
R (km)
11
10
9 −− νf
8 −− νp1
7
0.8 1 1.2 1.4 1.6 1.8 2 2.2
M/Mo
Fig. 6 The fits (23) and (24) are plotted in the R − M -plane, assuming the mode
frequencies νf = 1.983 kHz and νp = 6.164 kHz have been identified in a detected
signal. The black dot corresponds to the true values of the mass and radius of the
emitting star.
very large error on R; so large indeed that, even knowing the mass, we would
estimate R with an error of the order of 18%.
Therefore, for the time being, we can only say that gravitational wave
asteroseismology will become possible when GW-detectors will become more
sensitive to the high frequency region, and when nuclear matter studies will
put tighter constraints on the parameters that characterize the equation of
state of superdense matter.
Since science always looks forward for expanding the horizon of knowl-
edge, we are confident that one day this will be possible.
References
1. www.ligo.caltech.edu; www.virgo.infn.it
2. Kokkotas K.D. and Schmidt B.G. 1999 Living Rev. Rel. 2, 2
3. Nollert H.P. 1999 Class. Quant. Grav. 16 R159
4. Shapley H. 1914 Astrophys. J. 40 448
5. Eddington A.S. 1918 Mon Not Roy. Astron. Soc. 79, 2
6. Regge T. and Wheeler J.A. 1957 Phys. Rev. 108, 1063
7. Zerilli J.F. 1970 Phys. Rev. D 2, 2141; Zerilli J.F. 1970 Phys. Rev. Lett. 24,
737
8. Misner C.W., Thorne K S and Wheeler J A 1973, Gravitation, W.H. Freeman
& C, New York
9. Vishveshwara C.V. 1970 Phys. Rev. D 1, 2870
10. Press W.H. 1971 Astrophys. J. 170, L105
11. Chandrasekhar S. and Detweiler S.L. 1975 Proc. R. Soc. Lond. A 344, 441
12. Chandrasekhar S. 1984 The mathematical theory of black hole, Claredon Press,
Oxford
23
15
14
13 R= 12.75 km
12
●
R (km)
11
R= 10.65 km
10
8 ∆R/R ≈ 18 %
7
1 1.2 1.4 1.6 1.8 2 2.2
M/Mo
Fig. 7 The fit (23) is plotted in the R − M -plane including the error bars on the
fit parameters (dashed region).
13. Ghez A.M., Salim S., Hornstein S.D., Tanner A., Morris M., Becklin E.E. and
Duchene G. 2005, Astrophys. J. 620, 744
14. Schutz B.F. and Will C.M. 1985 Astrophys. J. Lett. 291, L33; Iyer S. and Will
C.M. 1987 Phys. Rev., D 35, 3621; Iyer S. 1987 Phys. Rev. D 35, 3632 (1987);
Kokkotas K.D. and Schutz B.F. 1988 Phys. Rev. D 37, 12
15. Guinn J.W., Will C.M., Kojima Y. and Schutz B.F. 1990 Class. Quantum
Grav. 7, L47
16. Leaver E.W. 1985 Proc. R. Soc. Lond. A 402, 285
17. Andersson N.,1992 Proc. R. Soc. Lond. A 439, 47
18. Nollert H.-P. 1993 Phys. Rev. D 47, 5253; Andersson N. 1993 Class. Quantum
Grav. 10, L61; Barreto A.S. and Zworski M. 1997 Math. Res. Lett. 4, 103;
Padmanabhan T. 2004 Class. Quantum Grav. 21, L1; Motl L. 2003 Adv.
Theor. Math. Phys. 6, 1135
19. Fiziev P.P. 2006 Class. Quantum Grav. 23, 2447
20. Teukolsky S. 1972 Phys. Rev. Lett. 29, 1114; Teukolsky S. 1973 Astrophys. J.
185, 635
21. Starobinski A.A. and Churilov S.M. 1973 Soviet JEPT 38, 1
22. Press W.H. and Teukolsky S. 1973 Astrophys. J. 185, 649
23. Detweiler S.L. 1977 Proc. R. Soc. Lond. A 352, 381; Detweiler S.L. 1979 in
Sources of Gravitational Radiation, edited by L. Smarr, Cambridge, England,
211; Detweiler S.L. 1978 Astrophys. J. 225, 687; Detweiler S.L. 1980 Astrophys.
J. 239, 292
24. Seidel E., Iyer S. 1990 Phys. Rev D 41, 374
25. Kokkotas K.D. 1991 Class.Quantum Grav. 8, 2217
26. Onozawa H. 1997 Phys. Rev. D 55, 3593
27. Ferrari V. and Mashoon B. 1984 Phys. Rev. Lett. 52, 1361; Ferrari V. and
Mashoon B. 1984 Phys. Rev. D 30, 295
28. Berti E., Cardoso V., Kokkotas K.D. and Onozawa H. 2003 Phys. Rev. D 68,
124018; Berti E., Cardoso V. and Yoshida S. 2004 Phys. Rev. D 69, 124018
29. Davis M., Ruffini R., Press W.H. and Price R.H. 1971 Phys. Rev. Lett 27, 1466
30. Davis M., Ruffini R. and Tiommo J. 1971 Phys. Rev. D 5, 2932; Ferrari V. and
Ruffini R. 1981 Phys. Lett. B 98, 381
24
31. Haugan M.P., Shapiro S.L. and Wasserman I. 1982 Astrophys. J. 257 283;
Shapiro S.L. and Wasserman I. 1982 Astrophys. J. 260 838; Pertich L.I.,
Shapiro S.L. and Wasserman I. 1985 Astrophys. J. Suppl. S 58 297
32. Oohara K. and Nakamura T. 1983 Phys. Lett. A 98 403
33. Berti E., Cardoso V. and Will C.M. 2006 AIP Conf. Proc. 848, 687
34. Nakamura T., Oohara K. and Kojima Y. 1987 Prog. Theor. Phys. Suppl. 90
218
35. Pretorius F. 2005 Phys. Rev. Lett 95 121101; Campanelli M., Lousto C.O.,
Marronetti P. and Zlochower Y., 2006 Phys. Rev. Lett 96 111101; Baker J.G.,
Centrella J., Choi D.-I., Koppitz M. and van Meter J. 2006 Phys. Rev. Lett 96
111102
36. Baker J.G., Campanelli M., Pretorius F. and Zlochower Y. 2007 Class. Quant.
Grav. 24, S25
37. Campanelli M., Lousto C.O. and Zlochower Y. 2006 Phys. Rev. D 74, 041501
38. Baiotti L., Hawke I., Rezzolla L. and Schnetter E. 2005 Phys. Rev. Lett. 94,
131101
39. Zanotti O., Font J.A., Rezzolla L. and Montero P.J. 2005 Mon Not Roy. Astron.
Soc. 356, 1371; Nagar A., Zanotti O., Font J.A. and Rezzolla L. 2007 Phys.
Rev. D 75, 044016
40. Ferrari V., Gualtieri L. and Rezzolla L. 2006 Phys. Rev. D 73, 124028
41. Leaver H.W. 1986 Phys. Rev. D 34, 384
42. Sun Y., and Price R. H. 1988 Phys. Rev. D 38, 1040; Andersson, N. 1995
Phys. Rev. D 51, 353; Nollert P. and Price R.H. 1999 J. Math. Phys. 40, 980;
Glampedakis K. and Andersson N. 2001 Phys. Rev. D 64, 104021; Berti E. and
Cardoso V. 2006 Phys. Rev. D 74, 104020
43. Price R.H. 1972 Phys. Rev. D 5 2419
44. Vishveshwara C.V. 1970 Phys. Rev. D 1, 2870
45. Wald R.M. 1979 J. Math. Phys. 20, 1056; Kay B.S. and Wald R.M. 1987 Class.
Quantum Grav. 4, 893
46. Whiting B.F. 1989 J. Math. Phys. 30, 1301; Beyer H.R. 2001 Comm. Math.
Phys. 221, 659
47. Horowitz G.T. and Hubeny V.E. 2000 Phys. Rev. D 62, 024027
48. Wang B., Lin C.Y. and Abdalla E. 2000 Phys. Lett. B 481, 79; Wang B., Molina
C. and Abdalla E. 2001 Phys. Rev. D 63, 084001; Cardoso V. and Lemos J.P.S.
2001 Phys. Rev. D 63, 124015; Cardoso V. and Lemos J.P.S. 2001 —IT Phys.
Rev. D 64, 084017; Berti E. and Kokkotas K.D. 2003 Phys. Rev. D 67, 064020;
Cardoso V., Konoplya R. and Lemos J.P.S. 2003 Phys. Rev. D 68, 044024
49. Chan J.S.F. and Mann R.B. 1996 Phys. Rev. D 55, 7546; 1999 Phys. Rev. D
59, 064025.
50. Dreyer O. 2003 Phys. Rev. Lett. 90, 081301; Motl L. 2003 Adv. Theor. Math.
Phys. 6, 1135
51. Motl L. and Neitzke A. 2003 Adv. Theor. Math. Phys. 7, 307; Cardoso V.,
Natario J. and Schiappa R. 2004 J. Math. Phys. 45, 4698; Natario J. and
Schiappa R. 2004 Adv. Theor. Math. Phys. 8, 1001.
52. Cardoso V. and Lemos J.P.S. 2003 Phys. Rev. D 67, 084020; Konoplya R.A.
2003 Phys. Rev. D 68, 024018; Cardoso V., Lemos J.P.S. and Yoshida S. 2004
Phys. Rev. D 69, 044004
53. Stergioulas N. 2003 Living Rev. Rel. 6, 3
54. Thorne K.S. and Campolattaro A. 1968 Astrophys. J. 149, 591; 1970 Astrophys.
J. 159, 847
55. Chandrasekhar S. and Ferrari V. 1991 Proc. R. Soc. Lond. A 432, 247;
56. Ferrari V. 1992 Phil. Trans. R. Soc. Lond. A 340, 423
57. Detweiler S.L. 1975 Astrophys. J. 201, 440
58. Lindblom L. and Detweiler S.L. 1983 Astrophys. J. Suppl. 53, 73
59. Lindblom L. and Splinter R.S. 1989 Astrophys. J. 345, 925
60. Detweiler S.L. and Lindblom L. 1985 Astrophys. J. 292, 12
61. Ipser J.R. and Price R.H. 1992 Phys. Rev. D 43 1768
62. Kokkotas K.D. 1994 Mon. Not. R. Ast. Soc. 268 1015
63. Chandrasekhar S. and Ferrari V. 1991 Proc. R. Soc. Lond. 434, 449
64. Benhar O., Berti E. and Ferrari V. 1999 Mon. Not. Roy. Astron. Soc. 310, 797
25
65. Cowling T.G. 1942 Mon. Not. R. Ast. Soc. 101, 367
66. Kokkotas K.D. and Schutz B.F. 1992 Mon. Not. R. Ast. Soc. 255 119
67. Andersson N. and Kokkotas K.D. 1998 Mon. Not. R. Ast. Soc. 299 1059
68. Benhar O., Ferrari V. and Gualtieri L. 2004 Phys. Rev. D 70, 124015
69. Baym G., Pethick C.J. and Sutherland P. 1971 Astrophys. J. 170, 299
70. Pethick C.J., Ravenhall B.G. and Lorenz C.P. 1995 Nucl. Phys. A 584, 675
71. Akmal A. and Pandharipande V.R. 1997 Phys. Rev. C 56, 2261
72. Akmal A., Pandharipande V.R. and Ravenhall D.G. 1998 Phys. Rev. C 58,
1804
73. Baldo M., Burgio G.F. and Schulze H.J. 2000 Phys. Rev. C, 61, 055801
74. Rubino R., thesis, Università “La Sapienza”, Roma (unpublished); Benhar O.
and Rubino R., to be published.
75. Glendenning N.K. 2000 Compact Stars (Springer, New York)
76. Lejeune A., Grangé P., Martzoff P. and Cugnon J. 1986 Nucl. Phys. A 453,
189
77. Witten E. 1984 Phys. Rev. D 30, 272
78. Benhar O., Ferrari V., Gualtieri L. and Marassi S. 2007 Gen. Rel. Grav. 39,
1323
79. Chodos A., Jaffe R.L., Johnson K., Thorne C.B. and Weiskopf W.F. 1974 Phys.
Rev. D 9,3471
80. De Grand T., Jaffe R.L., Johnsson K. and Kiskis J. 1975 Phys. Rev. D 12,
2060
81. Carlson C.E., Hansson T.H. and Peterson C. 1983 Phys. Rev. D 27, 1556
82. Farhi E. and Jaffe R.L. 1984 Phys. Rev. D 30, 2379
83. www.astro.cf.ac.uk/geo/euro
84. www.dual.lnl.infn.it
85. Dimmelmeier H., Font J.A. and Müller E. 2002 Astron. Astrophys. 393, 523;
Müller E., Rampp M., Buras R., Janka H.-T. and Shoemaker D.H. 2004 As-
trophys. J. 603, 221;
Ott C.D., Burrows A., Dessart L. and Livne E. 2006 Phys. Rev. Lett. 96,
201102.
86. Anderson P.W. and Itoh N. 1975, Nature 256, 25
87. Ruderman M. 1976, Nature 203, 213
88. Pines D. and Alpar M.A. 1985, Nature 316, 27