
3
the location of the horizon. In the above, κis the sur-
face gravity. Such a Hamiltonian has been obtained in
Eddington- Finkelstein (EF) outgoing null coordinates
both for a general SSSBH as well as Kerr spacetime. In
the above, the classical path of the particle has been cho-
sen to be along the normal to null hypersurface, given by
Eddington null coordinate u= constant, where u=t−r∗
with r∗is the well known tortoise coordinate (see [28,29]
for details regarding the construction of (7)). The same
can also be obtained in Painleve coordinates for SSSBH
as well as Kerr BH, considering a particular trajectory
[30–32] (a generic null surface also provides such Hamilto-
nian [33]). We just mentioned that the form of the Hamil-
tonian (7) was obtained explicitly in two specific choices
of coordinates – one in EF and another in Painleve co-
ordinates. Therefore the time can be taken at this stage
as either EF time or Painleve time. Moreover, it must
be explicitly mentioned that such a Hamiltonian was ob-
tained for a specific choice of outgoing null path. For
instance in EF coordinates the tangent of the path is
normal to u= constant hypersurface, while in Painleve
we found this for radially outgoing null path. In this
sense the Hamiltonian is very specific to these choices of
null paths, and also it depends on a particular choice of
observer. We will see that the Hawking temperature can
be obtained by using the Hamiltonian. Since it is well
known that horizon thermodynamics is an observer de-
pendent concept, we can expect the underlying dynamics
may depend on choice of coordinates. However it would
be interesting to investigate the generality of such form of
Hamiltonian. Also note that here x= 0 is location of the
horizon. As the Hamiltonian has been obtained by ex-
panding the metric coefficients around x= 0 and keeping
only the leading order term, the structure of the Hamilto-
nian retains for both x > 0 (just outside the horizon) and
x < 0 (just inside the horizon). Therefore the applicabil-
ity of the Hamiltonian is valid in the range −ϵ≤x≤ϵ
with ϵ > 0, where ϵis a very small quantity. We will use
this in our latter analysis.
Nonetheless, it must be noted that the above Hamil-
tonian, very near the horizon, is unstable. First discuss
for x > 0. The solutions of the equations of motions are
given by x∼eκt and p∼e−κt. Since the near horizon
limit x→0 is equivalent to t→ −∞, the momentum di-
verges there. Moreover for a fixed energy (7) implies that
p→ ∞ as x→0. Both imply that the Hamiltonian has
radial instability in the near horizon regime. On the other
hand, (7) can be cast to that of an inverted harmonic os-
cillator (see [32]), which is unstable in nature. While for
x < 0, the momentum is given by p=−(H/κy), where
x=−ywith y > 0. This again diverges at the hori-
zon y= 0. In this case the classical solutions take the
forms as y∼e−κt and p∼eκt. Since here y→0 implies
t→ ∞, one observes that the classical value of palso
diverges at the horizon. Recently we explicitly showed
this “local instability” may cause the temperature to the
horizon at the semi-classical level [28,29,32,34].
Inspired by all these facts here, we like to investigate
whether the unstable Hamiltonian (7) can show a com-
plex path and thereby produce a correct form of Hawking
temperature. We will show by constructing an effective
non-relativistic Hamiltonian in the Schrodinger descrip-
tion that this is indeed the case. Since this Hamilto-
nian is related to both Kerr and SSSBH spacetimes, our
model is capable of presenting a very general situation,
rather than restricted to SSSBH as was done originally
in [24]. Moreover, we will see that the identified temper-
ature does not suffer from the factor of two ambiguity.
This shows that our choice of coordinates in which (7) has
been obtained are suitable ones. The same is also being
confirmed by obtaining the density of states directly from
original relativistic ∼xp Hamiltonian. Additionally, as
(7) is unstable (or unbounded) in nature, the outcome
from this again confirms the robustness of the corollary,
suggested in [24] – the necessary (but maybe not suffi-
cient) condition that the Hamiltonian must be unbounded
in order to obtain a complex path.
Let us now move forward toward the calculation in
favour of our claim. In this analysis the Hamiltonian (7)
will be treated as that of a quantum mechanical system
and calculation will be done using the standard prescrip-
tions of path integral formalism. Before this, we will first
calculate the density of states corresponding to (7) using
the path integral kernel and show that such is thermal
in nature. Later an effective Hamiltonian will be con-
structed under the Schrodinger description in which case
the path will be calculated.
II. DENSITY OF STATES AND THE GROUND
STATE
The density of states (DOS) corresponding to our
present model can be evaluated from the kernel or prop-
agator K(x2, t2;x1, t1). In path integral approach DOS
is given by [35,36]
ρ(E) = −1
πImZG(X, E)dX,(8)
where Gis the Laplace-Fourier transformation of K:
G(X, E) = Z∞
0
dt e−iEt
ℏK(x2=X, t2=t;x1=X, t1= 0) .
(9)
The propagator for the Hamiltonian (7) is given by [34]
K(x2, t2=t;x1, t1= 0) = e−κt
2δx1−x2e−κt,(10)
with t > 0.