Anisotropic Stars: Exact Solutions
Anisotropic Stars: Exact Solutions
Anisotropic Stars: Exact Solutions
Abstract
arXiv:astro-ph/0012265v1 12 Dec 2000
relativistic treatment of this problem and obtain exact solutions for various
forms of the equation of state connecting the radial and tangential pressures.
It is shown that pressure anisotropy can have significant effects on the struc-
2M/R can approach unity (2M/R < 8/9 for isotropic objects) and the surface
I. INTRODUCTION
A common assumption in the study of stellar structure and evolution is that the interior of a
star can be modeled as a perfect fluid [1,2]. This perfect fluid model necessarily requires that
the pressure in the interior of a star to be isotropic. This approach has been used extensively
in the study of polytropes, including white dwarfs, and of compact objects such as neutron
stars [3]. However, theoretical advances in the last decades indicate that, in many systems,
deviations from local isotropy in the pressure, in particular at very high densities, may play
an important role in determining their properties [4,5].
The physical situations where anisotropic pressure may be relevant are very diverse. By
anisotropic pressure we mean that the radial component of the pressure, pr (r), differs from
the angular components, pθ (r) = pϕ (r) ≡ pt (r). (That pθ (r) = pϕ (r) is a direct consequence
1
of spherical symmetry.) Of course, spherical symmetry demands both to be strictly a func-
tion of the radial coordinate. A scalar field with non-zero spatial gradient is an example of
a physical system where the pressure is clearly anisotropic. [This anisotropic character of
a scalar field occurs already at the level of special relativity, where it is easy to show that
pr − pt = (dφ/dr)2]. Boson stars, hypothetical self-gravitating compact objects resulting
from the coupling of a complex scalar field to gravity, are systems where anisotropic pres-
sure occurs naturally [6]. Similarly, the energy-momentum tensor of both electromagnetic
and fermionic fields are naturally anisotropic. Isotropy appears as an extra assumption on
the behaviors of the fields or of the fluid modeling the stellar interior.
In the interior of neutron stars pions may condense. It has been shown that due to
the geometry of the π − modes, anisotropic distributions of pressure could be considered
to describe a pion condensed phase configuration [7]. The existence of solid cores and
type P superfluidity may also lead to departures from isotropy within the neutron star
interior [3]. Since we still do not have a detailed microscopic formulation of the possible
anisotropic stresses emerging in these and other contexts, we take the general approach of
finding several exact solutions representing different physical situations, modeled by ansatze
for the anisotropy factor, pt − pr . As a general rule, we find that the presence of anisotropy
affects the critical mass for stability, 2M/R, and the surface redshift, zs . These physical
consequences of pressure anisotropy are not new. Previous studies have found some exact
solutions, assuming certain relations for the anisotropy factor [8–14]. Our goal here is to
extend those results, offering a detailed analysis of the changes in the physical properties
of the stellar objects due to the presence of anisotropy. Hopefully, our results will be of
importance in the analysis of data from compact objects, as well as in the study of the
behavior of matter under strong gravitational fields.
This paper is organized as follows. In the next section, we set up the equations used
and the assumptions made in our study. We restrict our exact solutions to two classes,
investigated in sections III and IV respectively. In section III, after reviewing the results of
Bowers and Liang [8], we obtain several new exact solutions for stars of constant density.
2
In section IV, we examine solutions for the case ρ(r) ∝ 1/r 2, which has been used to model
ultradense neutron star interiors [15]. We conclude in section V with a brief summary of our
results and an outlook to future work. In Appendix A, we demonstrate the equivalence of
the Tolman and Schwarzschild masses for self-gravitating anisotropic spheres. In Appendix
B we present the general solution for stars featuring an energy density with a constant part
and a r −2 contribution, as discussed in Section IV, in terms of hypergeometric functions.
where all functions depend only on the radial coordinate r. The most general energy-
momentum tensor compatible with spherical symmetry is
We see that isotropy is not required by spherical symmetry; it is an added assumption. The
Einstein field equations for this spacetime geometry and matter distribution are
!
−λ ν′ 1 1
e + 2 − = 8πpr ; (3)
r r r2
!
−λ 1 ′′ 1 ′ ′ 1 ′ 2 (ν ′ − λ′ )
e ν − λ ν + (ν ) + = 8πpt ; (4)
2 4 4 2r
!
λ λ′ 1 1
e − 2 + = 8πρ . (5)
r r r2
3
It is often useful to transform the above equations into a form where the hydrodynamical
properties of the system are more evident. For systems with isotropic pressure, this formu-
lation results in the Tolman-Oppenheimer-Volkov (TOV) equation. The generalized TOV
equation, including anisotropy, is
dpr ν′ 2
= −(ρ + pr ) + (pt − pr ) , (6)
dr 2 r
with
1 ′ m(r) + 4πr 3 pr
ν = , (7)
2 r(r − 2m)
and
Z r
m(r) = 4πr 2 ρdr . (8)
0
Taking r = R in the above expression gives us the Schwarzschild mass, M. [This implicitly
assumes that ρ = 0 for r > R.]
A more general formula is the Tolman mass formula [16]:
Z R 1
M= (2T00 − Tµµ )(−g) 2 r 2 dr. (9)
0
The equivalence of the Tolman and Schwarzschild masses for systems with anisotropic pres-
sure is demonstrated in Appendix A.
In order to solve the above equations we must impose appropriate boundary conditions.
We require that the solution be regular at the origin. This imposes the condition that
m(r) → 0 as r → 0. If pr is finite at the origin then ν ′ → 0 as r → 0. The gradient dpr /dr
will be finite at r = 0 if (pt − pr ) vanishes at least as rapidly as r when r → 0. This will be
the case in all scenarios examined here.
The radius of the star is determined by the condition pr (R) = 0. It is not necessary for
pt (R) to vanish at the surface. But it is reasonable to assume that all physically interesting
solutions will have pr , pt ≥ 0 for r ≤ R.
4
III. EXACT SOLUTIONS FOR ρ = CONSTANT
Bowers and Liang [8] solved the generalized TOV equation by considering the following
equation of state,
ρ0 2
r 2 (pr 2 + 43 pr ρ0 + 3
)
pt − pr = C , (11)
1 − 38 πρ0 r 2
that is, they considered the term related to the anisotropy to be simply proportional to the
usual hydrodynamic term on the right hand side. The constant parameter C, which we will
also use, measures the amount of anisotropy.
They found that the radial pressure is given by
(1 − 2m/r)Q − (1 − 2M/R)Q
" #
pr = ρ0 , (12)
3(1 − 2M/R)Q − (1 − 2m/r)Q
1 3C
where the total mass M ≡ m(R), R is the radius of the system, and Q = 2
− 4π
. The
central pressure is given by
1 − (1 − 2M/R)Q
" #
pc = ρ0 . (13)
3(1 − 2M/R)Q − 1
An equilibrium configuration exists for all values of 2M /R such that pc is finite. The critical
model results for that value of 2M /R such that pc becomes infinite. From eq. 13, this occurs
when the denominator vanishes, that is, when
1 2
(2M/R)crit = 1 − ( ) 1−ξ , (14)
3
where ξ ≡ 3C/2. We see that for 0 ≤ ξ < 1, (2M/R)crit can be greater than 8/9, the
maximum value for an isotropic configuration [15]. For a given ρ0 and C, the critical mass
is
3
3 1 2
1 2
Mcrit =( ) 2 1 − ( ) 1−ξ . (15)
32πρ0 3
5
Equation 15 shows that the critical mass is less than the isotropic value when C < 0. When
C > 0 the critical mass exceeds the isotropic limit. For a given ρ0 , the maximum value of
the ratio of the critical mass to the isotropic mass approaches
1
zs = (1 − 2M /R)− 2 − 1 . (17)
1
zc = 3 1−ξ − 1. (18)
where C measures the amount of anisotropy and the function F (pr , ρ0 ) includes 6 separate
cases,
F = (pr 2 ; pr ρ0 ; ρ0 2 ; pr 2 + pr ρ0 ; pr 2 + ρ0 2 ; pr ρ0 + ρ0 2 ) . (20)
For the sake of brevity, we will only discuss 2 of the 6 possible cases. It should be
straightforward to obtain solutions for the other cases.
6
Case I.1 F (pr , ρ0 ) = ρ0 2
Thus we find
pr dpr r dr
Z Z
h i = −4π 8 . (22)
0 pr 2 + 4
pρ + ( 13 − C
)ρ0 2 R 1− πρ0 r 2
3 r 0 2π 3
The solutions naturally divide into 3 subcases, according to the value of the anisotropy
parameter, which we write as k ≡ C/(2π).
When k > −1/9 the radial pressure is given by
A A
(1 − 3k)[(1 − 2m/r) 2 − (1 − 2M/R) 2 ]
pr (r) = ρ0 A A
, (23)
(2 + A)(1 − 2M/R) 2 − (2 − A)(1 − 2m/r) 2
1
where A ≡ (1 + 9k) 2 . Clearly, when k = 0 we recover the isotropic limit. In figure 1.1, we
show the radial pressure as a function of the radius parameterized in terms of fractions of the
critical configuration, (2M/R)crit . The solutions are normalized to pr (R) = 0. Alternatively,
to obtain a solution we can fix ρ0 and pc , the central pressure, and compute the value of r
for which pr (R) = 0.
Fig. 1.1: Radial pressure as a function of radius for the case k = 1/2π, ρ = const and
(pt − pr ) ∝ ρ2 parameterized in terms of fractions of the critical configuration (2M/R)crit.
The central pressure is given by
7
A
(1 − 3k)[1 − (1 − 2M/R) 2 ]
pc = ρ0 A
. (24)
(2 + A)(1 − 2M/R) 2 − (2 − A)
For |k| ≪ 1 we find that, correcting slightly Bowers and Liang [8],
2M 8 4
≃ + − ln 3 k + O(k 2 ) . (26)
R crit 9 3
We see that a positive anisotropy, k > 0, leads to a violation of the isotropic limit
(2M/R)crit = 8/9. However, as k → 1/3, (2M/R)crit → 1 and pr < 0: there is a maxi-
mum allowed anisotropy in this case. When k < 0, (2M/R)crit is always less than 8/9.
The maximum surface redshift for k > −1/9 is
A1
2+A
zc = −1 . (27)
2−A
ρ0 2 B
pc = B tan arctan − ln(1 − 2M/R) − 2 , (30)
3 B 4
2M 4 2 π
= 1 − exp arctan − . (31)
R crit B B 2
In figure 1.2, we plot the critical values of 2M/R as a function of the anisotropic parameter
C.
8
The maximum surface redshift for k < −1/9 is
π 2 2
zc = exp − arctan( ) − 1 . (32)
4 B B
In figure 1.3, we plot the maximum surface redshift as a function of the anisotropic
parameter C. We find that for A → 2, i.e., k → 1/3, the surface redshift becomes infinite.
Thus, in this model, positive anisotropies generate arbitrarily large surface redshifts.
Fig. 1.2 Critical values of 2M/R as a function of the anisotropic parameter C for the
case ρ = const and (pt − pr ) ∝ ρ2 .
Fig 1.3: Maximum surface redshift as a function of the anisotropic parameter C for the
case ρ = const and (pt − pr ) ∝ ρ2 .
Case I.2: F (pr , ρ0 ) = p2r
This solution also separates into 3 subcases, depending on the value of the anisotropy
parameter, k = C/(2π). When k > −1/3, the radial pressure is given by
A A
(1 − 3k)[(1 − 2m/r) 2 − (1 − 2M/R) 2 ]
pr (r) = ρ0 A A
, (33)
(2 + A)(1 − 2M/R) 2 − (2 − A)(1 − 2m/r) 2
9
1
where A ≡ (1 + 3k) 2 . In figure 2.1, we show the radial pressure as a function of the radial
coordinate for fractions of (2M/R)crit and k = 1/π.
Fig 2.1: Radial pressure as a function of radius for the case k = 1/π, ρ = const and
(pt − pr ) ∝ p2r parameterized in terms of fractions of the critical configuration (2M/R)crit .
The central pressure is given by
A
(1 − 3k)[1 − (1 − 2M/R) ] 2
pc = ρ0 A
. (34)
(2 + A)(1 − 2M/R) 2 − (2 − A)
The critical configurations for the anisotropy parameter k > −1/3 are given by
" #2
2M (2 − A) A
=1− , (35)
R crit (2 + A)
2M
= 1 − e−2 . (39)
R crit
10
The solution for k < −1/3 is
( " !# )
ρ0 2 B 1 − 2m/r
pr (r) = B tan arctan + ln −2 , (40)
3(1 − k) B 4 1 − 2M/R
1
with B ≡ (3|k| − 1) 2 . The central pressure is
ρ0 2 B
pc = B tan arctan − ln(1 − 2M/R) − 2 . (41)
3(1 − k) B 4
Now,
2M 4 2 π
= 1 − exp arctan( ) − , (42)
R crit B B 2
π 2 2
zc = exp − arctan −1 . (43)
4 B B
Note that the critical values of 2M/R and surface redshifts for cases I.1 and I.2 are identical,
up to a shift in k → k/3. In figures 2.2 and 2.3 we plot the critical values of 2M/R and the
maximum surface redshift as a function of the anisotropic parameter C.
Fig 2.2: Critical values of 2M/R as a function of the anisotropic parameter C for the case
ρ = const and (pt − pr ) ∝ p2r .
11
Fig 2.3: Maximum surface redshift as a function of the anisotropic parameter C for the case
ρ = const and (pt − pr ) ∝ p2r .
CASE II
The second class of exact solutions with constant density follows the ansatze,
rn ρ0 2
C Rn
exp(− Rr )(pr 2 + 43 pr ρ0 + 3
)
pt − pr = , (44)
ρ0 1 − 83 πρ0 r 2
and
rn 2 ρ0 2
C Rn
exp(− Rr 2 )(pr 2 + 34 pr ρ0 + 3
)
pt − pr = , (45)
ρ0 1 − 83 πρ0 r 2
where n ≥ 2 is an integer. The motivation for this choice of anisotropy comes from boson
stars [6], where it is found that the anisotropy factor vanishes at the origin and outside the
star, reaching a maximum somewhere around the approximate radius of the configuration.
(Boson stars do not have a sharp boundary between the inside and the outside, as the scalar
field vanishes exponentially for r > R. One may think of it as a diffuse “atmosphere” around
the denser stellar core.)
For all cases considered, we found that there are values of C > 0 for which (2M/R)crit
can be greater than 8/9 and zc can approach arbitrarily large values. Let us focus on the
case where the anisotropy falls exponentially with distance, as in eq. 44. The case for
the “Gaussian” anisotropy (eq. 45) can be solved by following the same procedure. After
integration we obtain,
12
Z(r) − 1
pr (r) = ρ0 , (46)
3 − Z(r)
where
!1
1 − 8πρ0 r 2 /3 2
4C r
Z
Z(r) ≡ exp r ′(n−1) exp(−r ′ /R)dr ′ . (47)
1 − 8πρ0 R2 /3 3Rn R
For a given n the integral can be easily performed and we obtain an expression for the radial
pressure. Note that for physically acceptable solutions (i .e., with pr (r) > 0) the function
Z(r) must satisfy 1 ≤ Z(r) ≤ 3. In figure 3.1 we show the radial pressure as a function
of radial distance for various fractions of (2M/R)crit and n = 2. In figure 3.2, we plot the
anisotropy factor for the same fractions of (2M/R)crit and n = 2.
Fig. 3.1: Radial pressure as a function of radius for the case ρ = const and pt − pr ∝
( Rr )2 exp(− Rr ) parameterized in terms of fractions of the critical configuration (2M/R)crit .
Fig. 3.2: Anisotropy factor as a function of the radius for the same case as in fig. 3.1.
Of more interest is the behavior at the origin. From the above solution, the central
pressure can be written as,
13
Z(0) − 1
pc = ρ0 , (48)
3 − Z(0)
where
4C 0
−1/2 Z
Z(0) = 1 − 8πρ0 R2 exp r ′(n−1) exp(−r ′ /R)dr ′ . (49)
3Rn R
The critical configuration is obtained for Z(0) = 3. For example, for n = 2 we obtain,
2M 1 8C
−1
= 1 − exp − 1 − 2e . (50)
R crit 9 3
4C
−1
zc = 3 exp 1 − 2e −1 . (51)
3
Fig. 3.3: Critical values of 2M/R as a function of anisotropy for the case ρ = const and
pt − pr ∝ ( Rr )n exp −( Rr ) with n = 1, 2, 3, 4.
14
Fig. 3.4: Maximum surface redshift as a function of anisotropy for the case ρ = const and
pt − pr ∝ ( Rr )n exp −( Rr ), with n = 1, 2, 3, 4.
We will now consider anisotropic stellar configurations with the following expression for the
energy density
1 a
ρ= + 3b , (52)
8π r2
where both a and b are constant. The choice of the values for a and b is dictated by the
to a relativistic Fermi gas. If we take a = 3/7 and b 6= 0 then we have a relativistic Fermi
gas core immersed in a constant density background. For large r the constant density term
dominates (rc2 ≫ a/3b), and can be thought of as modeling a shell surrounding the core.
In this section we will consider cases where the pressure anisotropy closely resembles the
1 c
pt − pr = +d , (53)
8π r2
with c and d constant. The motivation for these ansatze comes from similar approaches
15
We have found it convenient to seek solutions for the metric function ν(r) directly, rather
than solving the generalized TOV equation. We will then use the known functions λ(r) and
ν(r) to find the radial and tangential pressures. From eqs. 3, 4, and 5, we find
ν ′′ (ν ′ )2 −λ
! ! !
λ′ 1 1 λ′ −λ 1
+ e − ν′ + e−λ − 2 + e + 2 = 8π(pt − pr ). (54)
2 4 4 2r r 2r r
ν
Introducing a new variable y = e 2 , eq. 54 becomes,
! " #
λ′ 1 −λ 1 λ′ 1
(y ′′ )e−λ − y ′ + e − y ( 2 + )e−λ − 2 = 8πy(pt − pr ). (55)
2 r r 2r r
In this section we will define the functionIb2 (x) ≡ 1 − a − bx2 to simplify our expressions.
When b = 0, we will write I02 ≡ 1−a. Using the expression for e−λ in eq. 55 and substituting
We give the full solution of eq. 57 with a, b, c, d 6= 0 in Appendix B. Below, we will consider
a
ρ= . (58)
8πr 2
e−λ = 1 − a. (59)
have a < 1. (Also, the metric coefficient grr becomes infinite when a = 1). A density profile
with this spatial dependence on the radial coordinate was found to be an exact isotropic
16
solution of the TOV equation for the interior of ultra high-density neutron stars by Misner
and Zapolsky [18]. Assuming that the neutron star core can be modeled as a relativistic
Fermi gas, i .e., pr = ρ(r)/3, they found the density to be given by eq. 58, with a = 3/7.
We note that the Misner-Zapolsky solution cannot be used to construct a complete star,
since this would require the radius of the star to be infinite. Here, we want to construct
stars with finite radii and density given by eq. 58 in the context of anisotropic pressure.
Thus, we impose boundary conditions such that pr (R) = 0. We also note that Herrera
investigated anisotropic solutions with similar energy density, in the context of “cracking,”
when perturbations in the fluid induce anisotropic stresses in the star [13,14]. However, we
follow a different approach, focusing on the physical properties of static anisotropic solutions.
The solutions of this equation divide into three classes, depending on the value of
1
(1 + c − 2a) 2
q≡ 1 (61)
(1 − a) 2
with the constants A+ and A− fixed by boundary conditions. For the case under considera-
dν a
e−λ(R) = eν(R) = I02 , and eν(R) |R = . (63)
dr R
17
I0 1 − 3I02
A+ = + and A− = A+ (q → −q). (64)
2 4qI0
The radial pressure for this case, after substituting the expressions for A+ and A− , is
We note that the boundary conditions automatically guarantee that pr (R) = 0. The radial
pressure is always greater than zero provided a < 2/3 and a2 > 4c(1 −a). Since by definition
a > 0, the second condition implies c > 0. Thus, this model does not allow for negative
anisotropy. Further, since we are considering the case q > 0, we must impose the condition
1 + c < 2a. Combining the two inequalities for a and c, we obtain, 2a − 1 < c < a2 /4(1 − a).
Since we have 0 < a < 2/3 we find that 0 < c < 1/3. We note that for the anisotropic case
the maximum value of a is 2/3, corresponding to a 33% increase when compared with the
isotropic case (a = 3/7). In figure 4 we plot the radial pressure, pr , as a function of the
radial coordinate r, for a = 3/7 and several values of c. Note that for this choice of a, the
inequality c < a2 /4(1 − a) imposes that c < 0.08 for positive pressure solutions. This can
be seen in the figure. For larger anisotropies, no static self-gravitating stable configuration
is possible.
18
Choosing a = 3/7 we recover, in the limit c → 0, the Misner-Zapolsky solution [18], with
pr = 1/(56πr 2) = ρ/3.
Case I.2: q = 0
The solution for y with q = 0 is
r r r
y(r) = As + Bs ln , (67)
R R R
where the constants As and Bs are determined from the boundary conditions. We find
1 − 3I02
As = I0 and Bs = . (68)
2I0
(1 − 3I02 )
8πr 2 pr = 3I02 − 1 + (1−3I02 )
. (69)
r
1+ 2I02
ln R
The radial pressure is positive provided a < 2/3. Since we are considering the case q = 0,
we must require c = 2a − 1 and thus −1 < c < 1/3.
r r r
y= As cos u ln + Bs sin u ln (70)
R R R
1 − 3I02
As = I0 , and Bs = . (71)
2uI0
An analysis of eq. 72 shows that pr > 0 provided that a < 2/3 for any c. Also, since we
are considering the case q imaginary here, we require that c < 2a − 1.
19
Case II: Including a Crust (a, b, c 6= 0, d = 0)
We will now derive solutions of the field equations with the density profile given by eq.
52. This density profile is essentially a combination of the two profiles (ρ = const and
ρ ∝ 1/r 2 ) that we have studied so far. We may think of this situation as modelling an
ultradense core immersed in a background of constant density; at large distances from the
core (rc2 ≫ a/3b), the constant density shell dominates the energy density. As in Case I, we
impose boundary conditions such that pr (R) = 0.
For this density profile,
ν
and the equation for y = e 2 is
(1 − a)r 2 − br 4 y ′′ − (1 − a)ry ′ + ay = 8πr 2 y(pt − pr ). (74)
1 − Ib2 (R)
y(R) = Ib2 (R) and y ′(R) = . (75)
2RIb (R)
Here, we have chosen the ansatz for the anisotropic pressure to be the same as the case with
b = 0, that is, we do not include a constant contribution to the anisotropy (d = 0 in eq. 53).
The general solution for d 6= 0 is given in Appendix B. As with the case above with b = 0,
there are three classes of solutions depending on the value of
1
(1 − 2a + c) 2
q= . (76)
(1 − a)
20
(3Ib2 (R) − 1) Ib (R)
" #
A+ = + and A− = A+ (q → −q). (78)
4qI0 2
We note that when b = 0, we recover the solution with no crust. Also, if we take a = 0
and b 6= 0, we have a new class of anisotropic solutions with constant density, and anisotropy
proportional to r −2 .
Case II.2: q = 0
When q = 0, the solution is
( )
r r [I0 + Ib (r)][I0 − Ib (R)]
ν(r)/2
y(r) = e = As + Bs ln (80)
R R [I0 − Ib (r)][I0 + Ib (R)]
with
(3Ib2 (R) − 1)
As = Ib (R), and Bs = . (81)
4I0
ν
8πr 2 pr = 3Ib2 (r)2 − 1 − 4Ib (r)I0 Bs e− 2 . (82)
with
(1 − 3Ib2 (R))
As = 2Ib (R) tan(F (R)) tan(2F (R)) − tan(2F ), Bs = Ib (R) − As , (84)
sI0
21
and
1 1
(1 − a) 2 + (1 − a − bx2 ) 2
F (x) = s ln 1 . (85)
(1 − a) 2
We intend to investigate numerically the allowed range of the parameters a, b, c, and d,
as well as the stability of these solutions, in a forthcoming publication.
V. CONCLUSIONS
We have presented two broad classes of general relativistic exact solutions of spherically
symmetric stellar configurations exhibiting anisotropic pressure. Our motivation was to
explore the changes in the general properties of the stars induced by varying amounts of
anisotropy. In particular, to each of the solutions we have demonstrated that anisotropy
may indeed change the critical mass and surface redshift of the equilibrium configurations,
results which we believe are of interest to the astrophysical community and will stimulate
further investigation. It is an open question if, indeed, anisotropy is revelant for compact
objects, as, for example, ultradense neutron stars. We have motivated our results based
on past work on this subject, where isotropic equations of state are at best a reasonable
hypothesis. Given that we do not have as of yet a complete understanding of the physical
processes controlled by strong interactions in ultradense matter [3], it is wise to keep an
open mind to the possibility that anisotropic stresses do occur. Furthermore, hypothetical
compact objects made of gravitational bound states of bosonic fields, the so-called boson
stars, are naturally anisotropic [19]. These objects have been extensively studied in the
literature, as they represent an interesting new class of compact objects whose stability
against gravitational collapse comes from a combination of Heisenberg’s uncertainty principle
and model-dependent self-interactions. They may also have some connection with dark
matter, if perturbations on a fundamental scalar field lead to instabilities which trigger the
gravitational collapse of overdense regions.
We have divided our work into two classes of solutions, those with a constant energy
density and those with an energy density falling as r −2 . Within each of these classes we
22
presented several possible cases, which we hope approximate to some extent possible realistic
objects, including a combined situation where the star’s energy density has an ultradense
interior (ρ ∝ r −2 ) immersed in a shell of constant density. We intend to study the stability of
these solutions against radial perturbations, as well as provide a detailed numerical analysis
of the allowed parameter space, in a future work. We also added two appendices, one
proving the equality between the Tolman and the Schwarzschild mass formulas and the
other providing the general solution for the ρ ∝ r −2 + const case in terms of hypergeometric
functions.
VI. ACKNOWLEDGEMENTS
We would like to thank Joseph Harris, Paul Haines, and especially Vincent Moncrief for
their interest in our work and many useful suggestions. KD thanks Dartmouth College for a
Dartmouth Fellowship. MG was supported in part by National Science Foundation Grants
PHY-0070554 and PHYS-9453431.
VII. APPENDIX A
Here we will establish the equivalence of the Tolman and Schwarzschild mass formulas for
static spherically symmetric spacetimes with anisotropic pressure. The general expression
for the Tolman mass formula is
Z 1
MT = 2T00 − Tνν (−g) 2 d3 x . (86)
Here we have
Let
23
Z R ν+λ
Z R λ+ν
2
I1 = 4πr ρe 2 dr and I2 = 4πr 2 (pr + 2pt )e 2 dr ; (89)
0 0
then,
MT = I1 + I2 . (90)
Defining
Z r
m(r) = 4πr 2 ρdr , (91)
0
MS = m(R) . (92)
The first term evaluates to MS , since m(0) = 0 and, at r = R, λ + ν = 0. From the field
equations it follows that
(λ + ν)′ 4πr 2 (ρ + pr )
= . (94)
2 (r − 2m)
R 4πmr 2 (ρ + pr ) λ+ν
Z
I1 = MS − e 2 dr. (95)
0 (r − 2m)
Next we consider the integral I2 . Using the generalized TOV equation we can substitute for
pt and write
with
24
Z R λ+ν
I2a = 4πr 2(3pr )e 2 dr , (98)
0
and
The first term of this equation is equal to zero, since pr (R) = 0, and substituting for ν ′ + λ′
from equation 94 we find that
MT = MS . (103)
This demonstrates the equivalence of the Tolman and Schwarzschild mass formulas for static
spherically symmetric spacetimes with anisotropic pressures.
VIII. APPENDIX B
The solution for eq. 57 with a, b, c, d 6= 0, a 6= 1 and boundary conditions given by eq.
75 is
1+q 1−q
r r
y = A+ F+ (r̃) + A− F− (r̃) (104)
R R
br
Here r̃ = 1−a
and F is the hypergeometric function:
∞
X (α)k (β)k xk
F = F (α, β, γ, x) = 1 + , (α)k = α(α + 1)...(α + k − 1) , (105)
k=1 (γ)k k
25
d αβ
F (α, β, γ, x) = F ′ = F (α + 1, β + 1, γ + 1, x) (106)
dx γ
and
with
1 b − 4d
α+ = [1 + 2q − ( )] ,
4 b
1 b − 4d
β+ = [1 + 2q + ( )] ,
4 b
1
γ+ = (1 + q) ,
4
1
1 − 2a + c 2
q= . (108)
1−a
26
REFERENCES
[3] N. K. Glendenning, Compact Stars: Nuclear Physics, Particle Physics and General
Relativity, (Springer-Verlag, Berlin, 1997); H. Heiselberg and M. H.-Jensen, Phys. Rep.
328, 237 (2000).
[6] For comprehensive reviews see, A. R. Liddle and M. S. Marsden, Int. J. Mod. Phys. D1
101, (1992); P. Jetzer, Phys. Rep. 220, 163 (1992); E. W. Mielke and F. E. Schunck, in
Proceedings of 8th M. Grossmann Meeting, T. Piran (ed.), (World Scientific, Singapore,
1998).
27
[16] R. C. Tolman, Phys. Rev. 35, 875 (1930).
[18] C. Misner and H. Zalopsky, Phys. Rev. Lett., 12, 635 (1964).
[19] M. Gleiser, Phys. Rev. D38, 2376 (1988); M. Gleiser and R. Watkins, Nucl. Phys.
B319, 733 (1989).
28