A&A 412, 603-613 (2003)
DOI: 10.1051/0004-6361:20031457
L. Rezzolla1,2 - O. Zanotti1,3 - J. A. Font3
1 - SISSA, International School for Advanced Studies,
via Beirut, 2-4 34014 Trieste, Italy
2 - INFN, Sezione di Trieste, via A. Valerio, 2 34127
Trieste, Italy
3 - Departamento de Astronomía y Astrofísica,
Universidad de Valencia, Dr. Moliner 50,
46100 Burjassot, Spain
Received 30 July 2003 / Accepted 12 September 2003
Abstract
We consider the effects of a cosmological constant on the
dynamics of constant angular momentum discs orbiting
Schwarzschild-de Sitter black holes. The motivation behind this
study is to investigate whether the presence of a radial force
contrasting the black hole's gravitational attraction can
influence the occurrence of the runaway instability, a robust
feature of the dynamics of constant angular momentum tori in
Schwarzschild and Kerr spacetimes. In addition to the inner cusp
near the black hole horizon through which matter can accrete onto
the black hole, in fact, a positive cosmological constant
introduces also an outer cusp through which matter can leave the
torus without accreting onto the black hole. To assess the impact
of this outflow on the development of the instability we have
performed time-dependent and axisymmetric hydrodynamical
simulations of equilibrium initial configurations in a sequence
of background spacetimes of Schwarzschild-de Sitter black holes
with increasing masses. The simulations have been performed with
an unrealistic value for the cosmological constant which,
however, yields sufficiently small discs to be resolved
accurately on numerical grids and thus provides a first
qualitative picture of the dynamics. The calculations, carried
out for a wide range of initial conditions, show that the
mass-loss from the outer cusp can have a considerable impact on
the instability, with the latter being rapidly suppressed if the
outflow is large enough.
Key words: accretion: accretion discs - black holes - relativity - hydrodynamics - cosmological constant
Discs around black holes may suffer from a number of instabilities produced either by axisymmetric or by non-axisymmetric perturbations and further triggered by the presence of magnetic fields. A type of instability that has been studied in a number of works and that could take place when the discs are geometrically thick and axisymmetric is the so-called runaway instability (see Font & Daigne 2002a; Zanotti et al. 2003 and references therein). To appreciate the mechanism leading to the development of this instability, consider an inviscid fluid torus with a vertical structure and internal pressure gradients orbiting around a black hole (either Schwarzschild or Kerr). If the fluid is non self-gravitating, it will be contained within isopotential surfaces which generically possess a cusp on the equatorial plane (Fishbone & Moncrief 1976; Kozlowski et al. 1978; Abramowicz et al. 1978). As a result, material from the disc can accrete onto the black hole through the cusp as the result of small deviations from hydrostatic equilibrium.
Any amount of matter lost by the disc and captured by the black hole will increase its mass (and angular momentum), resulting in a modification of the equipotential surfaces which may cause the cusp to move deeper inside the torus more rapidly than the inner edge of the torus. When this happens, additional disc material will be allowed to fall into the black hole in an increasingly accelerated manner leading to the runaway instability.
Although this instability was first studied in the 80's (Abramowicz et al. 1983; Wilson 1984), time-dependent hydrodynamical simulations have been performed only recently, either with SPH techniques and pseudo-Newtonian potentials (Masuda & Eriguchi 1997; Masuda et al. 1998), or with high-resolution shock-capturing (HRSC hereafter) techniques in general relativity (Font & Daigne 2002a; Zanotti et al. 2003). These investigations have shown that, under the (idealised) assumption of constant specific angular momentum distributions, relativistic tori around Schwarzschild and Kerr black holes are generically unstable to the runaway instability, if non self-gravitating. The inclusion of more generic initial conditions, however, can disfavour the occurrence of the instability. Recently, Font & Daigne (2002b) (see also Daigne & Font 2003) have shown through numerical simulations that the runaway instability is suppressed when a non-constant distribution of the angular momentum is assumed for the torus (increasing as a power-law of the radius), a result which is in agreement with studies based on a recent perturbative analysis (Rezzolla et al. 2003a,b). While a similar stabilizing effect has been shown to be provided by the black hole if this is rotating (Wilson 1984; Abramowicz et al. 1998), Masuda & Eriguchi (1997) were able to show that the inclusion of the self-gravity of the torus effectively favours the instability. Clearly, a final conclusion on the occurrence of this instability has not been reached yet and will have to wait for fully general relativistic simulations. However, the increasingly realistic investigations performed recently have addressed several important aspects and the prospects are that we may be close to reaching a detailed description of the dynamics of the instability.
A further physical process acting against the instability and
which has not been investigated so far, is provided by the existence of a
repulsive force pointing in the direction opposite to the black hole's
gravitational attraction. Such a force could disturb and even balance the
standard outflow of mass through the inner cusp, thus potentially
suppressing the runaway instability. As suggested recently by
Stuchlík et al. (2000), such conditions could arise naturally
in a black hole spacetime with a positive cosmological constant, i.e. in
a Schwarzschild-de Sitter spacetime. In such a spacetime, in fact, a
second cusp appears in the outer parts of the equilibrium tori, near the
so-called "static radius''. Assuming a value for the relict cosmological
constant
cm-2 as deduced from recent
cosmological observations of the vacuum energy density (Krauss 1998) and
compatible with a sample of observational estimates provided by the
analysis of a large number of high redshift supernovae (Perlmutter et al. 1999; Riess et al. 1998), Stuchlík et al.
(2000) find that the location of this outer cusp for the largest
stationary discs which can be built in a Schwarzschild-de Sitter
spacetime is at about 50-100 kpc for supermassive black holes with
masses in the range
.
As for the
inner one, a slight violation of the hydrostatic equilibrium at the outer
cusp would induce a mass outflow from the disc and away from the black
hole, which could affect the overall dynamics of the torus.
However, this is not the only way in which a cosmological constant could modify the dynamics of a disc orbiting around a Schwarzschild-de Sitter black hole. As argued by Stuchlík et al. (2000), in fact, a cosmological constant could produce a sensible modification in the accretion processes onto primordial black holes during the very early stages of expansion of the Universe, when phase transitions could take place, and the effective cosmological constant can have values in many orders exceeding its present value (Kolb & Turner 1990). Furthermore, a positive cosmological constant could also result into strong collimation effects on jets escaping along the rotation axis of the central black hole (Stuchlík et al. 2000).
The aim of this paper is to investigate one of these intriguing possibilities through numerical simulations. More precisely, we present a comprehensive study of the nonlinear hydrodynamics of constant angular momentum relativistic tori evolving in a sequence of background Schwarzschild-de Sitter spacetimes with increasing black holes masses. Our study clarifies the dynamical impact of a mass outflow on the occurrence of the runaway instability in such relativistic tori.
We note that our setup will not be an astrophysically realistic one. There are two important reasons for this. Firstly, given the present estimates for the value of the cosmological constant and for the masses of the supermassive black holes believed to exist in the centre of galaxies and active galactic nuclei, we are still lacking sufficient observational evidence that stationary thick accretion discs exist on scales of about 100 kpc. Secondly, even assuming that such objects are present within large galaxies, numerical calculations would have to face the present computational limitations which make it extremely hard to simulate accurately accretion discs with very low rest-mass densities and over such length scales. As a result, we will adopt a value for the cosmological constant that is unrealistically high. This yields discs with radial extents that are sufficiently small to be evolved numerically with satisfactory accuracy and provides a first qualitative description of the role that a cosmological constant could play on the dynamics of relativistic tori. In addition to this, the calculations reported here also offer a way of assessing how the self-gravity of the torus, which is basically contrasting the gravitational attraction of the black hole, could modify the overall inertial balance. This will provide a useful insight when fully relativistic calculations solving for the Einstein equations coupled to a self-gravitating matter source will be performed.
The organization of the paper is as follows. In Sect. 2
we briefly review the main properties of relativistic tori in a
Schwarzschild-de Sitter spacetime. Next, in Sect. 3 we present
the hydrodynamics equations and the numerical methods implemented in our
axisymmetric evolution code. The material presented in
this section is
rather limited, since the details have previously been reported in a
number of papers. The last part of this section is devoted to a
discussion of the initial data we use for the simulations. The numerical
results are then described in Sect. 4 and, finally,
Sect. 5 contains our conclusions. Throughout the paper we use a
space-like signature (-,+,+,+) and a system of geometrized units in
which c = G = 1. The unit of length is chosen to be the gravitational
radius of the black hole,
,
where M is the
mass of the black hole. Greek indices run from 0 to 3 and Latin indices
from 1 to 3.
Building on a wide literature discussing equilibrium
configurations of perfect fluid relativistic tori orbiting around
Schwarzschild or Kerr black holes, Stuchlík et al. (2000) have
recently extended these results to the case of a Schwarzschild-de Sitter
black hole. In spherical coordinates (t, r,
the line element
of this spacetime reads
![]() |
(2) |
The perfect fluid with four-velocity
is described by
the usual stress-energy tensor
Once M and
have been prescribed, the explicit
expression for the potential
in the Schwarzschild-de Sitter
spacetime is simply given by (Stuchlík et al. 2000)
![]() |
Figure 1:
Schematic diagram for the Keplerian specific angular momentum
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
In other words, given a value for the dimensionless parameter
,
circular orbits can exist for
.
Clearly, these
orbits will be at radii smaller than the static radius
,
at which the angular momentum (6) is zero and where
the gravitational attraction of the black hole is exactly balanced by the
cosmological repulsion
.
Finally, it is worth reminding that for point-like particles, the
decreasing part of
(i.e. for
)
corresponds to unstable circular geodesics, while the
increasing part (i.e. for
)
corresponds
to stable circular geodesics. For a perfect fluid, however, internal
pressure gradients allow the torus to occupy a region which would be
geodesically unstable for point-like particles. A schematic diagram
showing the behaviour of the Keplerian specific angular momentum
and the corresponding effective potential W in a
Schwarzschild-de Sitter spacetime, as well as the locations of the
different radii that have been discussed so far is presented in
Fig. 1. The Schwarzshild case is also displayed for comparison
with a dashed line.
Once a value for the cosmological parameter and for the angular
momentum have been fixed, Eq. (4), supplemented by the
polytropic EOS, can be integrated analytically for any
,
to
yield the rest-mass density distribution inside the torus
Hereafter, we will focus on tori built in a parameter space that
is smaller than the one discussed so far. In particular, we will consider
tori with constant specific angular momentum in the range
,
where
is the angular
momentum of the unstable photon circular geodesic (Stuchlík et al. 2000). Furthermore, the hydrodynamical evolution of these tori will
be followed in Schwarzschild-de Sitter spacetimes with dimensionless
cosmological constant
,
where
corresponds to the value of y for which the
minimum of
is equal to the Keplerian angular momentum of
the outer marginally stable orbit (Stuchlík et al. 2000).
Suitable initial data for the torus (see Sect. 3.3 for
specific details) are evolved forward in time using a two-dimensional
(2D) general relativistic hydrodynamics code. The equations for a perfect
fluid are implemented using the formulation developed by Banyuls et al. (1997) in which, adopting a 3+1 decomposed spacetime, the
covariant conservation laws of the density current and of the
stress-energy tensor are recast into a first-order, flux-conservative
system of the type
![]() |
(11) |
![]() |
(12) |
![]() |
(13) |
![]() |
![]() |
![]() |
(14) |
![]() |
![]() |
![]() |
(15) |
![]() |
![]() |
![]() |
(16) |
![]() |
![]() |
![]() |
(17) |
![]() |
(18) | ||
![]() |
(19) |
The numerical code we use is the same employed by Zanotti et al. (2003) to study the dynamics of constant angular momentum relativistic discs around a Schwarzschild black hole. For the present investigation the code has been extended to account for the modifications introduced by the Schwarzschild-de Sitter geometry. The general relativistic hydrodynamics equations are solved by means of a HRSC scheme based on Marquina's flux formula (see e.g. Font 2000 for a review of these methods in numerical general relativistic hydrodynamics).
In order to cover optimally the large spatial extent of the
equilibrium configurations and yet reach a satisfactory spatial
resolution in the regions closer to the two cusps where the fluid motion
needs to be calculated most accurately, we have introduced an important
technical modification in the handling of the radial-coordinate
grid. More precisely, we use a non-uniform radial grid with a logarithmic
spacing, which is double-varied in the vicinities of the inner and outer
cusps. The coordinate mapping used for this purpose is reminiscent of a
tortoise coordinate mapping but it has been extended to a
Schwarzschild-de Sitter metric as
![]() |
(20) |
As a result of this mapping, a radial grid of Nr=300 zones
allows to cover a spatial domain going from
to
with a minimum radial spacing of the innermost
part of the grid
and, correspondingly, a minimum
radial spacing
for the outermost part of the radial
grid. The two grids join smoothly at r=48.6, where the resolution is
.
The angular grid, on the other hand, is more
straightforward to build and consists of
equally spaced
zones extending from 0 to
(cf. Zanotti et al. 2003)
As in Zanotti et al. (2003), a low density "atmosphere" is introduced in those parts of the numerical domain outside the torus. The initial atmosphere model chosen corresponds to the spherically symmetric accretion solution of non-interacting test fluid particles. The maximum density of the atmosphere is typically 5 to 6 orders of magnitude smaller than the density at the centre of the torus. In all of the validating tests performed, the hydrodynamical evolution of the torus was found to be unaffected by the presence of this atmosphere, which is evolved as the bulk of the fluid.
Finally, the mass outflows at the innermost and outermost radial
points are computed respectively as
Table 1:
Main properties of the tori considered in the numerical
calculations. From left to right the columns report: the name of the
model, the specific angular momentum
(normalized to M), the
polytropic constant
,
the inner and outer cusps of the torus,
and
,
the radial position of the pressure
maximum
(all radii are in units of the gravitational
radius
), the potential gaps
and
,
where
is the potential at the
inner edge of the disc. The last column reports the orbital period at the
centre of the torus,
,
expressed in milliseconds. All of
the models share the same value of the cosmological parameter
y=10-6, the same mass for the black hole,
,
the same
adiabatic index
,
and the same torus-to-hole mass ratio
.
Note that the mass outflow given by Eq. (21)
corresponds effectively to the mass accretion rate onto the black hole
and is used to account for the instantaneous increase of the black hole
mass at every time step. This, in turn, provides information about the
changes in the background spacetime, fundamental for the appearance of
the runaway instability (Font & Daigne 2002a; Zanotti et al.
2003). As mentioned in the Introduction, since we neglect the
self-gravity of the torus, the hydrodynamics equations are solved in a
sequence of background Schwarzschild-de Sitter spacetimes with increasing
black hole masses. In practice, the spacetime evolution is achieved
through a remapping of the metric functions at each time level of the
type
![]() |
Figure 2:
Radial profiles of the potential at the equatorial plane for
models labelled A ( left), B ( centre) and C ( right),
respectively. The local maxima in each plot indicate the location of the
cusps. The horizontal lines fix the potential of the fluid element at the
edges of the torus for each of the models considered. Models A, B and
C differ among each other by the value of the angular momentum ![]() |
Open with DEXTER |
As shown by Font & Daigne (2002a) and Zanotti et al.
(2003), the runaway instability is a robust feature of constant angular
momentum relativistic tori in Schwarzschild and Kerr spacetimes, if these
are non self-gravitating. This result does not depend on the way the
instability is triggered, i.e. by either artificially expanding the torus
over the potential barrier at the inner cusp (Font & Daigne 2002a), or
by introducing perturbations in an otherwise stable torus (Zanotti et al. 2003). As discussed above, however, the presence of an outer
cusp in a Schwarzschild-de Sitter spacetime is likely to affect the
robustness of this conclusion. In order to investigate to what extent an
outflow of mass can interfere with the development of the runaway
instability we have studied the behaviour of three different classes of
models, which we refer to as A, B and C. These models are
distinguished on whether the effective potential at the inner cusp,
,
is less than, equal to, or larger than the effective
potential
at the outer cusp, respectively. Furthermore, for
each of these classes of models we have considered three different
initial configurations, as shown in Fig. 2, with the potential
at the inner edge of the torus,
,
being different from the
potential barrier,
,
by the adjustable amount
.
Note that for all of the models of the class A, the hydrostatic
equilibrium is always violated at the inner cusp, i.e.
,
and a mass outflow will necessarily take place at the inner
edge of the disc once the initial data is evolved. Furthermore, a mass
loss will take place also at the outer cusp for models A2 and A3,
which have
.
For all of the
models of the class B, on the other hand, the hydrostatic equilibrium
is violated at both cusps and by the same amount, i.e.
and, again, the mass outflows are regulated by
the potential jump
.
Finally, for all of the models of
the class C, the hydrostatic equilibrium is always violated at the
outer cusp, i.e.
,
and also at the inner cusp
for models C2 and C3, for which
.
All of these different initial conditions are illustrated
in Fig. 2, whose different panels show the effective potential
curves and the values of
for the three classes of models.
Summarized in Table 1 are the potential jumps
and
,
as well as the most relevant
parameters of the various initial models considered here. Note that we
have used a polytropic index
and adjusted the polytropic
constant so as to have a small torus-to-hole mass ratio
,
thus minimizing the error introduced by neglecting the
self-gravity of the torus.
A final comment will be made on the value adopted for the
dimensionless cosmological constant y. As mentioned in the
Introduction, the primary goal of our investigation is to provide a first
qualitative analysis of the dynamics of thick discs in spacetimes with a
positive cosmological constant. Because of this, and in order to avoid
shortcomings with observations or computational resources, we have
performed numerical simulations with y=10-6 for a black hole of mass
.
These parameters, although unrealistic, allow us to
construct sufficiently compact discs and perform numerical simulations
with a satisfactory spatial resolution. We expect that all of the
qualitative conclusions we draw from our simulations will continue to
hold when considering more realistic values for both the cosmological
constant and the black hole masses.
As mentioned above, the development of the runaway instability
appears to be a robust feature of the dynamics of non self-gravitating
tori orbiting around Schwarzschild or Kerr black holes with constant
distributions of specific angular momentum. This conclusion has been
reached after numerous simulations have been performed for a large range
of torus-to-hole mass ratios
and under a number of
different initial conditions (Font & Daigne 2002a; Zanotti et al.
2003). In all of these simulations, the onset and full development of the
instability was observed when the spacetime geometry was suitably
modified to account for the black hole's mass-increase due to
accretion. As a result of the instability, the torus is very rapidly
accreted onto the black hole and this is most clearly signalled by the
exponential growth in time of the rest-mass accretion rate at the
innermost radial grid point. The growth-time for the instability is
inversely proportional to the ratio
and is comparable with
the dynamical (i.e. orbital) timescale when
.
![]() |
Figure 3:
Time evolution of the inner ( left panel) and of the outer ( right
panel) mass outflows for the models of class A. The data is shown in
units of solar masses per second, while the time is expressed in units
of the orbital period. Note that only models A1 (solid line) and A2(dotted line) are runaway unstable. The solid circles in the two panels
indicate the time at which
![]() |
Open with DEXTER |
![]() |
Figure 4: Velocity field and equally spaced isocontours of the logarithm of the rest-mass density for model A2 at an early time ( left panel) and at a later time ( right panel); the times reported are in units of the orbital period. Initially the outer mass flux dominates the dynamics of the torus. However, the gravitational attraction of the black hole eventually overcomes the effect of the cosmological constant and the runaway instability takes place. This leads to the large inward-directed fluxes and to the disappearance of the torus inside the black hole in a few orbital periods. |
Open with DEXTER |
In the case of a Schwarzschild-de Sitter black hole, however, the rest-mass of the torus can change not only because of losses through the inner cusp leading to accretion onto the black hole, but also because of outflows from the outer cusp and away from the black hole. While both allowed, the impact that these two mass outflows could have on the dynamics of the torus is very different. The first one, in fact, induces changes in the black hole mass and could therefore lead to the runaway instability. The second one, on the other hand, does not produce changes of the background spacetime and cannot therefore produce an instability. Nevertheless, it can affect the hydrodynamical evolution in a number of different ways: firstly, by reducing the amount of rest-mass in the torus available for accretion and, secondly, by producing significant alterations of the velocity field, especially in the outer regions of the torus.
We have followed the hydrodynamical evolution of the models
described in Table 1 over a number of orbital periods
sufficiently large to reveal the impact of a positive cosmological
constant on the occurrence of the runaway instability. In
Figs. 3, 5 and 7 we show the two mass outflows
and
as a function of the orbital
period
at the centre of the torus,
and for the three classes of models listed in Table 1. The three
small insets shown in the panels for
offer a view of
the evolution of the rest-mass of the torus after this has been
normalized to its initial value. The description of the dynamics of the
tori is also completed with Figs. 4 and 6, which show
equally spaced isocontours of the logarithm of the rest-mass density and,
superimposed, the velocity field for models A2 and B2,
respectively, at two different times during the evolution.
As it is apparent from a rapid look at these figures, the runaway
instability is no longer the only possible evolution of the system, whose
dynamics is instead the result of the interplay between the inner and the
outer mass outflows. The occurrence of the runaway instability is clearly
visible in the left panel of Fig. 3 for model A1 (solid
line). Model A1, in fact, has initial conditions that resemble those
encountered for a Schwarzschild black hole, with the outer radius of the
torus located far from the outer cusp (cf. Figs. 1 and 2).
As a result, the right panel of Fig. 3, shows that the outer
mass outflow is in this case very small (indeed slightly negative as a
result of accretion onto the torus of the infalling atmosphere), while
the mass accretion rate onto the black hole (left panel) grows
exponentially and undisturbed until the full development of the runaway
instability at
.
The dynamical evolution is
different for models A2 (dotted line) and A3 (dashed line), where
the competition between the two outflows at the inner and outer edges of
the disc is closer to a balance and the initial outer mass-loss is
non-negligible. For model A2, in particular, this is clearly visible
in the left panel of Fig. 4, which shows that at early times
(
)
the largest fluid velocities (
)
are reached in the outer regions of the torus and are outwardly
directed. However, the outer mass-loss eventually becomes insufficient to
prevent the development of the runaway instability, which takes place
after
.
The corresponding velocity field (with
all vectors pointing towards the black hole) and the isocontours of the
logarithm of the rest-mass density displayed in the right panel of
Fig. 4 (at time
)
show the important
reduction in size undergone by the torus, which is about to disappear
entirely inside the black hole after a few more orbital periods. This
type of evolution does not take place for model A3, whose dynamics is
completely dominated by the mass outflow through the outer cusp and for
which the runaway instability does not develop (cf. Fig. 3).
![]() |
Figure 5: Same as Fig. 3 but for the models of class B. |
Open with DEXTER |
![]() |
Figure 6: Same as Fig. 4 but for model B2. The intense mass ouflow across the outer edge of the disc removes a large fraction of its mass, and suppresses the runaway instability. The final disc reaches a quasi-steady state. |
Open with DEXTER |
![]() |
Figure 7: Same as Fig. 3 but for the models of class C. |
Open with DEXTER |
To better interpret the dynamics behind these simulations it is
useful to compare the amount of rest-mass in the torus after the
first 10 orbital periods of the evolution for the three models of class
A (see inset in the left panel of Fig. 3). The residual
rest-mass of the torus is 99%, 90%, and 55% of the initial one
for the models A1, A2, and A3, respectively. This may appear
somewhat surprising given the fact that the mass outflow from the inner
cusp is smaller in the case of model A3 than it is for model A2,
despite A3 having a larger initial potential jump
.
The explanation for this comes from looking at the right panel of
Fig. 3 which shows that the mass outflow from the outer cusp is
however larger for model A3 than it is for model A2. As a result,
the torus is emptied more efficiently, and this happens mostly through
the outer cusp.
It is also very instructive to compare the mass outflows at the
inner and outer edges of the disc for the three models of class
A. Doing so leads to the important result that
at all times for model A1 and that
at all times for model A3. In other words, the
unstable and the stable models seem to differ from each other on whether
the mass outflow from the inner edge is larger or smaller than the
corresponding mass-loss from the outer edge. In the case of model A2,
on the other hand, the two mass outflows are closer to a balance and
only for
,
after which the runaway instability clearly develops; the time
when this happens is shown with the filled circles in the two panels of
Fig. 3. It appears, therefore, that an increasing potential gap
at the outer edge of the torus favours the outflow of mass from the
outer cusp but it also disfavours (and to a larger extent)
the mass outflow from the inner cusp, with the corresponding
accretion onto the black hole being severely suppressed. As a result, the
feed-back of the black hole spacetime onto the overall dynamics of the
system is considerably reduced and the runaway instability
suppressed. Stated differently, the development of the runaway
instability appears to be related to the efficiency of the mass-loss
through the edges of the disc and, in particular, the instability is
effectively "extinguished'' whenever the mass outflow from the outer edge of the disc and away from the black hole is larger than
the mass outflow from the inner edge of the disc and onto the black
hole. The condition
can thus be used
as a simple sufficient condition for the suppression of the runaway
instability in a thick disc orbiting around a Schwarzschild-de Sitter
black hole.
The role played by a positive cosmological constant on the
dynamics of the discs and described so far for the models of class A is
present also for the models of class B and C, although with some
slight differences. As discussed in Sect. 3.3 and illustrated in
Fig. 2, the models of class B are built with the outer and
inner edges having the same effective potential. One would therefore
expect that this would yield to very similar mass outflows at the two
boundaries of the disc. However, the two panels of Fig. 5 show
that the mass fluxes through the outer edges of the discs to infinity are
always larger than the ones towards the black hole (i.e.
). As a result, the models of class B are all stable
to the runaway instability. This is particularly apparent in models B2and B3, for which the outer mass outflows are at least a couple of
orders of magnitude larger than the corresponding mass outflows onto the
black hole, and which become negligibly small (i.e.
)
well before 10 orbital periods. As a result,
a large amount of the matter in those discs is not accreted onto the
black hole, but escapes to infinity. This is illustrated in the small
inset of the left panel of Fig. 5 which shows that after about
20 orbital periods more than 60% of the torus rest-mass is lost for
model B2 and more than 90% for model B3.
Once the outflows die-off in the tori of class B, the remaining
matter reaches a quasi-stable equilibrium, accreting onto the black hole
on a timescale which is essentially controlled by the rate of mass-loss
through the inner cusp. The importance of the mass outflow at the outer
edges of the discs of class B is also apparent from Fig. 6,
which shows the velocity field and isocontours of the logarithm of the
rest-mass density of model B2 at an early and a later stage of the
evolution. Note how the left panels of Figs. 4 and 6
have velocity fields that differ mostly in modulus but are equally
oriented, while the right panels are substantially different with
velocity fields that have opposite orientations leading to the
disappearence of the torus into the black hole and to infinity,
respectively. We also note that while the difference between the inner
and outer mass outflows remains large also in the case of model B1,
the dynamics is in this case much closer to an equilibrium, with the
torus being still progressively emptied to infinity, but on a much larger
timescale. No runaway instability was observed for this model over the
time for which the calculations were carried out (
).
Finally, for models C (see Fig. 7), the dynamics of the discs is particularly simple and the final result is rather clear to interpret. In this case, in fact, all the discs are built with an effective potential which is larger at the outer edge (cf. Fig. 2) and represent, therefore, initial conditions that are conceptually the opposite of those in models A. Because of the high potential barrier at the inner edge of the disc, the inner mass outflow is always rather minute and several orders of magnitude smaller than the corresponding mass outflow from the outer edge. As a result, the mass in the torus is lost very rapidly to infinity and very little is accreted onto the black hole. In particular, in the most dramatic case of model C3, the residual rest-mass in the torus is less than 20% after only 6 orbital periods.
We have investigated the effect of a positive cosmological constant on the dynamics of non self-gravitating thick accretion discs orbiting Schwarzschild-de Sitter black holes with constant distributions of specific angular momentum. The motivation behind this investigation has been that of assessing the role played by an effective repulsive force in the onset and development of the runaway instability, which represents a robust feature in the dynamics of constant angular momentum tori. In addition to the inner cusp near the black hole horizon, through which matter can accrete onto the black hole when small deviations from the hydrostatic equilibrium are present, thick discs in a Schwarzschild de-Sitter spacetime also possess an outer cusp through which matter can leave the torus without accreting onto the black hole. As a result of this mass-loss to infinity, the changes in the background metric (which are responsible for the development of the runaway instability) may be altered considerably and the instability thus suppressed.
As a simple way to evaluate this effect we have considered a sequence of Schwarzschild-de Sitter spacetimes differing only in their total mass and have performed time-dependent general relativistic hydrodynamical simulations in these background metrics of thick discs which are initially slightly out of hydrostatic equilibrium. In doing this we have adopted an unrealistically high value for the cosmological constant which however yields sufficiently small discs (extending up to about a few hundred gravitational radii) to be accurately resolved with fine enough axisymmetric numerical grids.
We have performed a number of simulations involving initial
configurations of constant specific angular momentum discs differing both
for the relative amplitude of the peaks in the effective potential and
for the potential jump at the inner and outer cusps. The results obtained
indicate that the runaway instability is no longer the only possible
evolution of these systems but that their dynamics is rather the
end-result of the interplay between the inner and the outer mass
outflows. On the one hand, in fact, we have evolved initial models for
which the cosmological constant has a weak influence; these models have
negligible mass outflows to infinity while maintaining large mass
outflows onto the black hole, which then lead to the development of the
runaway instability. On the other hand, we have evolved initial models
which are significantly influenced by the cosmological constant; these
models develop mass outflows through the outer cusp which are much larger
than those appearing at the inner cusp and, hence, do not develop the
runaway instability. Placed somewhere between these two classes of
initial configurations there exist initial models for which the mass
outflows from the inner and outer cusps are more closely balanced. In
these cases the runaway instability may or may not develop and we have
noticed that a simple comparison between the mass outflows can be used to
deduce the fate of the accreting disc. More specifically, we have found
that the condition
provides a simple
sufficient condition for the suppression of the runaway instability in a
thick disc orbiting around a Schwarzschild-de Sitter black hole.
In spite of the idealized setup used, the simulations performed here provide a first qualitative description of the complex nonlinear dynamics of thick discs in Schwarzschild-de Sitter spacetimes and we expect that most of the results obtained will continue to hold also when more realistic values for the cosmological constant are used. Aa a final comment we note that besides providing a qualitative description of the role that a cosmological constant could play on the dynamics of relativistic tori, these calculations also offer a way of assessing, at least qualitatively, the inertial role that the self-gravity of the torus plays in the development of the runaway instability. This will be very useful when studying the dynamics of relativistic tori with numerical codes solving also the full Einstein equations.
Acknowledgements
It a pleasure to thank Z. Stuchlík for useful comments. Financial support for this research has been provided by the Italian MIUR, by the Spanish Ministerio de Ciencia y Tecnología (grant AYA 2001-3490-C02-01) and and by the EU Network Programme (Research Training Network Contract HPRN-CT-2000-00137). L.R. also acknowledges the kind hospitality at the KITP in Santa Barbara, where part of this research was carried out (NSF grant PHY99-07949). The computations were performed on the Beowulf Cluster for numerical relativity "Albert100'', at the University of Parma.