Preprint typeset in JHEP style.

- HYPER VERSION hep-th/0106111

Essential and inessential features of Hawking

radiation
arXiv:hep-th/0106111v1 13 Jun 2001

Matt Visser∗
Physics Department, Washington University in Saint Louis, MO 63130-4899, USA
E-mail: [email protected]

Abstract: There are numerous derivations of the Hawking effect available in the liter-
ature. They emphasise different features of the process, and sometimes make markedly
different physical assumptions. This article presents a “minimalist” argument, and
strips the derivation of as much excess baggage as possible. All that is really necessary
is quantum physics plus a slowly evolving future apparent horizon (not an event hori-
zon). In particular, neither the Einstein equations nor Bekenstein entropy are necessary
(nor even useful) in deriving Hawking radiation.

Keywords: Hawking radiation, Bekenstein entropy, apparent horizon.

13 June 2001; LATEX-ed May 28, 2018

∗
Research supported by the US DOE.
Contents
1. Introduction 1

2. Metric: Painlevé–Gullstrand form 4

3. Eikonal and WKB approximation (s wave) 5

4. Near-horizon modes: ingoing 7

5. Near-horizon modes: outgoing 7

6. Straddling the horizon: the Boltzmann factor 8

7. Beyond s wave 10

8. Essential features 12

9. Discussion 12

1. Introduction
Hawking radiation from black holes is a semiclassical quantum effect that has now
been with us for some 27 years [1], and whose theoretical importance is difficult to
exaggerate. Over the decades, the Hawking effect has accreted a quite considerable
mythology. Perhaps the two most pernicious myths attached to this effect are:

• “Hawking radiation has something to do with gravity”, and,

• “Hawking radiation automatically implies Bekenstein entropy”.

These myths were engendered by two historical accidents: (1) the Hawking effect was
first encountered within the context of general relativity, and (2) the fact that it was
discovered shortly after the notion of Bekenstein entropy (geometric entropy) had been
formulated [2, 3].

1
 Though the Hawking effect was at first partly motivated by the need for a consis-
tent thermodynamic interpretation for the notion of Bekenstein entropy, it was rapidly
appreciated that the Hawking effect is much more primitive and fundamental [4] —
in particular the Hawking effect continues to make sense even in situations where ge-
ometric entropy and even gravity itself are simply not relevant [5]. This observation
underlies much of the current interest in “analog models of/for general relativity” [6, 7];
there is now a realistic possibility for experimental detection of the Hawking effect in
condensed-matter analog systems using current or planned technology [8, 9, 10], with
“effective metrics” and “black holes” that have nothing to do with gravity itself [11].
In view of this situation, in this current article I will attempt to isolate an irreducible
minimum of physical assumptions needed for the Hawking effect to arise.
Since the literature is vast, I will not be able to do justice to all known deriva-
tions of the effect — some key derivations are those due to Hawking himself [1], the
Hartle–Hawking approach using analytic continuation of the propagator across the
event horizon of an eternal black hole [12], and the Gibbons–Hawking approach using
Euclideanization (Wick rotation) [13]. Extremely useful general surveys are provided
in [14] and [15].
A particularly relevant discussion is the early work of Damour and Ruffini [16], who
emphasise the behaviour of the “outgoing” modes as one crosses the horizon. Though
their presentation is given in terms of the Kerr–Newman geometry it is easy to verify
that the specifics of the geometry enter only in the evaluation of the surface gravity
of the future Killing horizon, and that the route from surface gravity to Hawking
radiation does not depend on either the Einstein equations or even the underlying
physical mechanism leading to the existence of the metric.
In a slightly different vein, the discussion of Parker [17] particularly emphasises
the relationship with particle production from a dynamical vacuum state, while Ger-
lach [18], Grove [19], Hu [20], and Brout and Parentani [14] emphasise in varying
degree the near-universal role of the exponential stretching associated with many types
of horizon.
More recently the contributions of Massar and Parentani [21], Parikh and Wilczek [22]
Padmanabhan et al [23], and Schutzhold [24] should be noted. They emphasise, in
slightly different forms, the analyticity properties of the modes and what is effectively
an imaginary contribution to the action localized at the horizon, an observation that
can be traced back to the work of Damour and Ruffini [16].
It is also important to realise that the Hawking radiation effect is independent
of whatever cutoff you introduce to the high-frequency physics — this is one of the
theoretical reasons for interest in analog models, because for acoustic black holes you
have an explicit model for the high frequency cutoff in terms of atomic physics [25, 26,

2
27]. I will not have anything specific to say about this particular issue in the present
paper.
Based on the results extracted from 27 years of research, as verified and made more
explicit by the recent interest in analog models, it is clear that the basic physics we
should be aiming for is:

• “Hawking radiation is kinematics”,

• “Bekenstein entropy is geometrodynamics”.

That is: Hawking radiation is a purely kinematic effect that depends only on the exis-
tence of a Lorentzian metric (with no particular prejudice as to how this metric arises)
and some sort of horizon; Hawking radiation does not depend on the validity of the
Einstein equations (as may most quickly be verified by looking at Hawking’s original
derivation [1] and verifying that the Einstein equations are nowhere used nor needed).
In contrast, Bekenstein’s geometric entropy associated with the area of the event hori-
zon is an intrinsically geometrodynamic effect, wrapped up with the validity of the
Einstein equations: Entropy equals one quarter the area (plus perturbative correc-
tions) if and only if the Einstein equations are valid (plus perturbative corrections) [5].
The connection with the Einstein equations arises because when integrating the first
law to evaluate the Bekenstein entropy you need to use the relationship between total
mass of the black hole and its surface gravity (and hence Hawking temperature), and
it is in this relationship between surface gravity and mass-energy that the Einstein
equations enter.
I shall also distinguish the notion of “apparent horizon” from that of the “event
horizon” (absolute horizon) and demonstrate that the existence of a locally definable
apparent horizon is quite sufficient for obtaining the Hawking effect. (Remember that
to define the event horizon you need to know the entire history of the spacetime out
to the infinite future; you should be a little alarmed if the question of whether or not
a black hole is radiating now depends on what it is doing in the infinite future.)
The general theme of the analysis will be to do as much as possible with the
eikonal approximation (even WKB is mild overkill), and look for generic features in
the modes at/near the apparent horizon. Specifically, we will look for a Boltzmann
factor. Note that a good derivation should not depend on either grey-body factors or a
past horizon. Grey-body factors are not fundamental — they are simply transmission
coefficients giving the probability that modes which escape from the horizon make it
out to null infinity. Past horizons are specific to eternal black holes and simply not
relevant for astrophysical black holes. Similarly because the entropy-area law is tied to
the validity of the Einstein equations we shall seek to avoid any appeal to this property.

3
Since much of what I have to say is well-known to experts in the field (at least within
the general relativity community) I will place a premium on clarity and simplicity.

2. Metric: Painlevé–Gullstrand form

In general relativity any spherically symmetric geometry, static or not, can by suitable
choice of coordinates locally be put in the form

ds2 = −c(r, t)2 dt2 + (dr − v(r, t) dt)2 + r 2 [dθ2 + sin2 θ dφ2 ]. (2.1)

Equivalently

ds2 = −[c(r, t)2 − v(r, t)2 ] dt2 − 2v(r, t) dr dt + dr 2 + r 2 [dθ2 + sin2 θ dφ2 ]. (2.2)

These are so-called Painlevé–Gullstrand coordinates, a relatively obscure coordinate

system currently enjoying a resurgence. (See [28] for a geometric discussion.) A nice
feature of these coordinates is that the metric is nonsingular at the horizon. Though
the basic physics is coordinate independent, we shall see that these coordinates simplify
computations considerably.
In the acoustic geometries associated with sound propagation in a flowing fluid it
is often convenient to not set up coordinates this precise manner but rather to use
the natural coordinates inherited from the background Minkowski spacetime. Doing so
results in a minor modification of the Painlevé–Gullstrand metric
" #
2 ρ(r, t) n o
ds = −c(r, t)2 dt2 + (dr − v(r, t) dt)2 + r 2 [dθ2 + sin2 θ dφ2 ] , (2.3)
c(r, t)
where c(r, t) is now the speed of sound, v(r, t) is the radial velocity of the fluid, and
ρ(r, t) is its density. Phonons are then massless minimally coupled scalar fields in this
acoustic geometry [6, 7]. Fortunately the conformal factor does not affect the surface
gravity and does not affect the Hawking effect [29]; for simplicity I shall simply suppress
the conformal factor, it can easily be reinstated if desired.
In matrix form the Painlevé–Gullstrand metric takes the quasi–ADM form

2 2 .. 

−(c − v ) . −v r̂ j 
gµν (t, ~x) ≡ 
············ · ······.
 (2.4)
..
−v r̂i . δij
The inverse metric is
 .. j

−1 . −v r̂
1 
g µν (t, ~x) ≡ 2 

c
· · · · · · · · · · · .
········  (2.5)
..

i 2 ij 2 i j
−v r̂ . (c δ − v r̂ r̂ )

4
Because of spherical symmetry finding the apparent horizon is particularly easy: it
is located at c(r, t) = |v(r, t)|. It is easy to see that the metric is nonsingular at
the apparent horizon, with det(gµν ) = −c(r, t)2 . To get a future apparent horizon,
corresponding to an astrophysical black hole, we need v < 0. I make no claims as to
the location or even existence of any event horizon.
[Since for static black holes the event horizon is coincident with the apparent hori-
zon, the general feeling is that for quasi-stationary black holes the event horizon is
likely to be “near” the apparent horizon. If accretion dominates over evaporation the
event horizon is likely to be just outside the apparent horizon, while if evaporation
dominates over accretion it is likely to be just inside the apparent horizon. There is an
unpopular minority opinion that takes the view that because of Hawking evaporation
a black hole will never form a true event horizon (absolute horizon). For the derivation
presented herein this entire issue is simply not relevant.]
Define a quantity:

1 d[c(r, t)2 − v(r, t)2 ] 

 
d[c(r, t) − |v(r, t)|] 
gH (t) =  = cH  . (2.6)
2 dr 
H
dr 
H

If the geometry is static, this reduces to the ordinary definition of surface gravity [7]:
gH
κ= . (2.7)
cH
If the geometry is not static this is a natural definition of the “surface gravity” of the
apparent horizon.

3. Eikonal and WKB approximation (s wave)

Consider a quantum field φ(r, t) on this Painlevé–Gullstrand background and take the
eikonal (geometric optics/acoustics) approximation for the s wave
  Z r 
′ ′
φ(r, t) = A(r, t) exp[∓iϕ(r, t)] = A(r, t) exp ∓i ω t − k(r ) dr (3.1)

whereby the field is written as a rapidly varying phase times a slowly varying envelope.
The second equality above, where we have written the time dependence of the phase
as ωt, is valid provided the geometry is slowly evolving on the timescale of the wave,
that is, provided ω ≫ max{|ċ/c|, |v̇/v|}. (I take ω to be positive.)
In the eikonal approximation the d’Alembertian equation of motion becomes

g µν ∂µ ϕ ∂ν ϕ + iǫ = 0. (3.2)

5
Note in particular that I have used the eikonal approximation to immediately impose
Feynman’s “iǫ-prescription” on the field (ǫ is real, positive, and infinitesimal). The met-
ric signature is the general relativity standard −+ ++, which is why the iǫ-prescription
appears reversed relative to the particle physics standard. Also note that in invoking
the prescription I have implicitly used the fact that the spacetime geometry is smooth,
even at the horizon, so that it makes sense to both adopt an eikonal approximation
and then within this framework use ordinary flat-space results. The use of the iǫ pre-
scription in this way can be traced back, at least, to the early paper of Damour and
Ruffini [16]. If one prefers, the iǫ-prescription can be rephrased in terms of analyticity
of the fields on the complexified past light cone, Z(x) = {z = x + iy}, where x is
any point in spacetime and y lies in the past light cone. This is the Lorentz-invariant
generalization of analyticity in the lower half-plane, but for all practical purposes can
be replaced by the iǫ prescription.
From the preceding equation

ω 2 − 2v(r, t) ωk − [c(r, t)2 − v(r, t)2 ]k 2 + iǫ = 0. (3.3)

So that
(ω − vk)2 = c2 k 2 + iǫ. (3.4)
Whence
ω − vk = σ (1 + iǫ) ck; σ = ±1. (3.5)
For specific real frequency ω this gives us the wave-vector k(r, t) as

ω σω σ (1 + iǫ) c − v
k= = = ω. (3.6)
σ (1 + iǫ) c + v (1 + iǫ) c + σv (1 + iǫ)2 c2 − v 2

Note:
σ = +1 ⇒ outgoing mode. (3.7)
σ = −1 ⇒ ingoing mode. (3.8)
Keeping track of the iǫ is important only near the apparent horizon, where it is critical;
everywhere else it can safely be set to zero.
While we do not really need to use the WKB approximation (physical acous-
tics/optics) it can be invoked at very little additional cost. The (approximate) con-
served current is
Jµ = |A(r, t)|2 (ω, k, 0, 0). (3.9)
Then
1
∇µ J µ = 0 ⇒ |A(r, t)| ∝ . (3.10)
r

6
So we can write
N Z r
  
φ(r, t) ≈ √ exp ∓i ω t − k(r ′ ) dr ′ , (3.11)
2ω r
where N is some conveniently chosen normalization, to be discussed more fully below.

4. Near-horizon modes: ingoing

The ingoing wavenumber (σ = −1) is
ω
kin = − . (4.1)
(1 + iǫ) c − v
Thus in the vicinity of the future apparent r ≈ rH (with v ≈ −c) the ingoing wavevector
is approximately
ω
kin → − . (4.2)
2cH
Thus the ingoing modes are approximately
Nin r − rH
  
φ(r, t)in ≈ √ exp ∓iω t + . (4.3)
2ω rH 2cH
This means the phase velocity of the ingoing mode as it crosses the horizon (in co-
ordinate distance per coordinate time) is 2cH . (Phase velocity equals group velocity
because there is no dispersion.) We see that the ingoing modes contain no real surprises
and are relatively uninteresting.

5. Near-horizon modes: outgoing

Now consider the outgoing mode σ = +1
ω
kout = . (5.1)
(1 + iǫ) c + v
In the vicinity of the future apparent r ≈ rH (with v ≈ −c) the outgoing wavevector
is approximately
ω
kout ≈ . (5.2)
[gH /cH ](r − rH ) + iǫ cH
But because ǫ is infinitesimal, and both gH and cH are by hypothesis positive, we
can for all practical purposes rewrite this in terms of the “principal part” and a delta
function contribution. That is, near the apparent horizon
cH ω 1
   
kout ≈ ℘ − iπ δ(r − rH ) . (5.3)
gH r − rH

7
Thus we can ignore the iǫ unless we are actually intending to cross the apparent horizon.
In particular, just outside the apparent horizon
Z r Z r
dr ′ ω Z r
dr ′ cH ω cH ω
k= ≈ = ln[r − rH ]. (5.4)
c(r ) − |v(r )|
′ ′ gH (r − rH )
′ gH
Therefore (for r > rH )
 h i 
ωcH
exp ±i gH
ln[r − rH ]
φ(r, t)out ≈ Nout √ exp {∓iωt}
2ω rH
[r − rH ]±iωcH /gH
≈ Nout √ exp {∓iωt} .
2ω rH
The phase velocity of the outgoing mode as it crosses the horizon (in coordinate distance
per coordinate time) is zero.
The fact that these outgoing modes have the surface gravity, κ = gH /cH , show up
in such a fundamental and characteristic way is already strongly suggestive; and this is
really all there is to Hawking radiation. The logarithmic phase pile-up at the horizon
is characteristic of many derivations of Hawking radiation, in particular [1], and for
many readers this will be sufficient to convince them that Hawking radiation is present
under the current circumstances (slowly evolving apparent horizon without prejudice
as to where the metric comes from). In fact, this calculation is the easiest and fastest
way I know of to deduce the existence of the phase pileup using completely elementary
methods.
To properly describe the modes that escape to infinity we should normalize on the
half-line r > rH using the standard Klein-Gordon norm. This results in replacing Nout
with some specific normalization constant Nescape whose precise value we do not need
to know.

6. Straddling the horizon: the Boltzmann factor

In addition to looking at outgoing modes that escape to infinity, it is also useful to
consider what happens to “outgoing” modes that straddle the apparent horizon. Inside
the apparent horizon the real part of kout goes negative, indicating that while the
outgoing mode is trying to escape “upstream” it is actually being overcome by the
flow/shift vector and swept back “downstream”. Additionally the phase picks up an
imaginary contribution, due ultimately to the iǫ prescription,
Z r+ Z r+
cH ω 1
   
kout ≈ dr ′
℘ ′ − iπ δ(r ′ − rH )
r− r− gH r − rH
( )
cH ω |r+ − rH |
= ln − iπ . (6.1)
gH |r− − rH |

8
So just inside the apparent horizon
|r − rH |±iωcH /gH
( )
π ω cH
φ(r, t)straddle(r<rH ) ≈ Nstraddle √ exp + exp [∓iωt] . (6.2)
2ω rH gH
Compare with the situation just outside the apparent horizon
|r − rH |±iωcH /gH
√
φ(r, t)straddle(r>rH ) ≈ Nstraddle exp [∓iωt] . (6.3)
2ω rH
Note that this straddling mode has associated with it a normalization factor Nstraddle ,
which is distinct from that of the escaping mode Nescape ; the straddling mode is to be
normalized on the entire half-line r > 0. In terms of the Heaviside step function
" ( ) #
π ω cH
φ(r, t)straddle ≈ Nstraddle Θ(rH − r) exp + + Θ(r − rH )
gH
|r − rH |±iωcH /gH
√× exp [∓iωt] . (6.4)
2ω rH
See Damour and Ruffini [16], equation (5b), for an early occurrence of a very similar
statement; see also Massar and Parentani [21], equation (13). Note note in particular
the presence of a Boltzmann-like factor
( )
πωcH
exp + (6.5)
gH
which (we shall soon see) corresponds to the Hawking temperature
h̄ gH
k TH =
. (6.6)
2π cH
For many physicists the presence of this Boltzmann-like factor will be enough: The
occurrence of Boltzmann factors of this type was the key to the Hartle–Hawking deriva-
tion [12], though they were working with the full propagator and dealing with past and
future horizons of a maximally extended Kruskal–Szekeres eternal black hole. Damour
and Ruffini [16] demonstrated the existence of similar Boltzmann factors for mode func-
tions evaluated at the future Killing horizon of a Kerr–Newman black hole (dispensing
with the past horizon entirely). In the present situation, the same Boltzmann factor is
seen to arise for any slowly evolving apparent horizon.
The imaginary contribution to the integrated wavenumber is, in slightly disguised
form, equivalent to the imaginary contribution to the action that occurs in the Parikh–
Wilczek approach [22],1 and is also related but not identical to the imaginary contri-
bution arising from complex paths in the approach of Padmanabhan et al [23].
1
Note that Parikh and Wilczek have subsequently and implicitly made use of the Einstein equations
at the stage when they then relate the emission process to the entropy change. This comment also
applies to the Massar–Parentani approach.

9
 But it is possible to quite easily do a lot more: While I have so far carefully not
specified specific values for Nescape and Nstraddle , the relationship between them is very
simple. Since the straddling mode is to be normalized on (0, +∞), [which we actually
approximate by the full line (−∞, +∞)], while the escaping mode is only normalized
on the half line (rH , ∞), we have
" ( ) #
2 2π ω cH
|Nstraddle | exp + − 1 = |Nescape |2
gH

So the relative normalization is

 2
N  1
 straddle 
= n o .
 Nescape  + 2π;ω cH
 
exp gH
−1

It is this Planckian form of the normalization ratio that then leads to a Planckian dis-
tribution for the outgoing flux. (The straddling mode contains a Planckian distribution
of escaping modes.)
Note the physics assumption hidden here: one is assuming that the quantum vac-
uum state is that corresponding to φstraddle . That is, freely falling observers (who get
to see both sides of the horizon) should not see any peculiarities as one crosses the
horizon. Picking the quantum vacuum corresponding to φstraddle implies choosing the
Unruh vacuum — and when we look at this vacuum state far from the horizon we see
the Planckian flux of outgoing particles.
(If for whatever reason you don’t like this normalization calculation or the related
thermodynamic arguments you can alternatively use the phase pileup property directly
to perform a Bogolubov coefficient calculation in the style of [1]; all roads lead to
Rome.)

7. Beyond s wave

What happens if we go beyond the s wave? There is now some momentum transverse
to the apparent horizon so that

∂µ ϕ = (ω, −k, −k⊥ ). (7.1)

If we resolve the field in terms of partial waves

2 ℓ(ℓ + 1)
k⊥ = . (7.2)
r2

10
Then in the eikonal approximation (I now suppress the iǫ, it has done its job and would
now only serve to clutter the formulae)

−ω 2 + 2v(r, t) ωk + [c(r, t)2 − v(r, t)2 ] k 2 + c(r, t)2 k⊥

2
= 0. (7.3)

That is
(ω − vk)2 = c2 k 2 + c2 k⊥
2
. (7.4)
This is a quadratic for k as a function of ω and k⊥ :
q
2
σ c2 ω 2 − (c2 − v 2 )c2 k⊥ −v ω
k= . (7.5)
c2 − v 2
For ingoing modes near the apparent horizon one must evaluate using L’Hôpital’s rule:

ω 2 − c2 k ⊥
2
kin → − . (7.6)
2 cH ω
So the ingoing modes do depend on k⊥ .

(r − rH )[ω 2 − c2 k⊥
2
" ( )#
Nin ]
φ(r, t)in ≈ √ exp ∓iω t + . (7.7)
2ω rH 2 cH ω

But that does not matter: The ingoing modes are not the relevant ones. For the
outgoing modes, near the apparent horizon we see
cH ω
kout → . (7.8)
gH (r − rH )

That is, for the outgoing modes:

—The near-horizon asymptotic behaviour is independent of k⊥ .
—The phase pile-up is independent of k⊥ .
—Continuation of the outgoing modes across horizon is independent of k⊥ .
—The Hawking temperature independent of k⊥ .
—This behaviour is universal for all partial waves.
—Adding a mass term corresponds to:

c2 k ⊥
2
→ c2 k ⊥
2
+ (m c2 /h̄)2 . (7.9)

In view of the above we see that the behaviour of the outgoing modes near the horizon
is also completely independent of the mass and transverse momentum. Consequently
the same universal Hawking temperature applies to all masses and all partial waves.
(Again, this is the easiest way I know of to convince oneself by elementary means that

11
restricting attention to the s-wave captures almost all the essential physics of Hawking
radiation.)
Note that while the Hawking temperature is completely independent of both an-
gular momentum and mass, the grey-body factors are another matter: they do depend
on both angular momentum and mass and are responsible for effectively cutting off the
higher angular momentum modes.

8. Essential features
The only real physics input has been basic quantum physics plus the existence of a
Lorentzian metric with:
— an apparent horizon;
— non-zero gH ;
— slow evolution.
The need for slow evolution of the geometry is hidden back in the approximation used
to write the modes as exp(±iωt) times a position-dependent factor. This makes sense
only if the geometry is quasi-static on the timescale set by ω. So for consistency we
should only trust the Boltzmann factor, and the Planckian nature of the Hawking
radiation, for frequencies greater than max{|ċ/c|, |v̇/v|}. In particular, in order for the
peak in the Planck spectrum to be meaningful we require
kTH
≈ ωpeak ≫ max{|ċ/c|, |v̇/v|}. (8.1)
h̄
In particular we require 
d[c(r, t) − |v(r, t)|]  ċH
 ≫ . (8.2)
dr 
H
cH
Near the horizon spatial gradients should dominate over temporal gradients. In par-
ticular the closer the black hole is to extremality the slower it is permitted to evolve if
there is to be any hope for even a small segment of quasi-thermal spectrum.
That’s it. It is truly remarkable how basic and primitive the Hawking radiation
phenomenon is, and how few physical assumptions are really necessary.

9. Discussion
Can the essential conditions for the Hawking effect be further relaxed?
One obvious question is the use of spherical symmetry, which precludes direct
application of the current approach to Kerr and other rotating black holes. This is a
technical problem, not a fundamental problem, and working in axisymmetric geometries

12
will be do-able but somewhat messier. (For Kerr–Newman black holes the Damour–
Ruffini analysis can be adapted to this end [16].) A tricky point for general analog model
geometries is that without the Einstein equations, and something like the dominant
energy condition, there is no longer any reason to believe in the zero’th law: the
surface gravity and Hawking temperature can then in principle vary over different parts
of the apparent horizon; these complications were suppressed in the current article via
the simple expedient of enforcing spherical symmetry. (For Killing horizons there are
derivations of the zero’th law that do not depend on the Einstein equations [30], but
such considerations lose their force once the horizon becomes time dependent.)
Secondly, there are simple linguistic issues of definition: How far can we push the
Hawking effect before we should give it another name? As argued in this article, based
on the physics there is a very good case for keeping the name the same for the effect
in arbitrary “effective geometries”, no matter how derived. Some would even argue
that the Hawking effect and Unruh effects are fundamentally identical; I prefer to view
then as distinct, possibly as two sides of the same coin — the response of the quantum
vacuum to externally imposed conditions.
Thirdly: What is the energy source for the Hawking radiation? The infalling
particles have negative energy as seen from outside the apparent horizon. In the case
of general relativity black holes the infalling particles serve to reduce the total mass-
energy of the central object, and it is ultimately the total mass-energy of the black
hole that provides the energy emitted in the hawking flux. In the case of an acoustic
black hole the infalling negative energy phonon steals kinetic energy from the fluid
flow used to generate the acoustic geometry. For effective geometries associated with
“slow light” the electromagnetic control field, used to generate EIT (electromagnetically
induced transparency) and so reduce the group velocity, provides energy to the system
which is available to ultimately be converted to Hawking-like photons. The general
message is that the Hawking effect steals energy from whatever process is used to set
up the effective geometry in question.
Finally, since this point still seems to cause much confusion, I should make the
explicit comment that
• Hawking radiation is not a test of quantum gravity.
Instead, searching for Hawking radiation is a test of the general principles of quantum
field theory in curved spacetimes. As such it is an ingredient useful for testing semi-
classical quantum gravity, though it does not necessarily probe quantum gravity itself.
In particular, all the proposed experimental tests of Hawking radiation via “analog
models” will only probe kinematic aspects of black hole physics. To start to address
the dynamics of general relativity black holes one needs the Einstein equations (or some

13
approximation thereto), in which case one can begin to discuss Bekenstein entropy (or
some approximation thereto). This requires a whole extra layer of theoretical super-
structure, and a key point of this paper is that it is important and useful to keep these
notions logically distinct.

Acknowledgments

I acknowledge comments and feedback from the participants at the Black Holes III
conference in Kananaskis, Canada. I particularly wish to thank Bei-Lok Hu, Ted
Jacobson, Renaud Parentani, and Amanda Peet for their comments.

References
[1] S. W. Hawking, “Black Hole Explosions”, Nature 248 (1974) 30.
S. W. Hawking, “Particle Creation By Black Holes”, Commun. Math. Phys. 43 (1975)
199.

[2] J. D. Bekenstein, “Black Holes And The Second Law”, Lett. Nuovo Cim. 4 (1972) 737.
J. D. Bekenstein, “Black Holes And Entropy”, Phys. Rev. D 7 (1973) 2333.
J. D. Bekenstein, “Generalized Second Law Of Thermodynamics In Black Hole Physics”,
Phys. Rev. D 9 (1974) 3292.

[3] J. M. Bardeen, B. Carter and S. W. Hawking, “The Four Laws Of Black Hole Mechan-
ics”, Commun. Math. Phys. 31 (1973) 161.

[4] W. G. Unruh, “Notes On Black Hole Evaporation”, Phys. Rev. D 14 (1976) 870.

[5] M. Visser, “Hawking radiation without black hole entropy”, Phys. Rev. Lett. 80 (1998)
3436 [gr-qc/9712016].

[6] W. G. Unruh, “Experimental black hole evaporation?”, Phys. Rev. Lett, 46, 1351
(1981).

[7] M. Visser, “Acoustic propagation in fluids: An Unexpected example of Lorentzian

geometry”, gr-qc/9311028.
M. Visser, “Acoustic black holes: Horizons, ergospheres, and Hawking radiation”, Class.
Quant. Grav. 15 (1998) 1767 [gr-qc/9712010].
M. Visser, “Acoustic black holes”, gr-qc/9901047.
S. Liberati, S. Sonego and M. Visser, “Unexpectedly large surface gravities for acoustic
horizons?”, Class. Quant. Grav. 17 (2000) 2903 [gr-qc/0003105].

14
 [8] L. J. Garay, J. R. Anglin, J. I. Cirac and P. Zoller, “Black holes in Bose-Einstein
condensates”, Phys. Rev. Lett. 85 (2000) 4643 (2000) gr-qc/0002015.
L. J. Garay, J. R. Anglin, J. I. Cirac and P. Zoller, “Sonic black holes in dilute Bose-
Einstein condensates”, gr-qc/0005131.

[9] U. Leonhardt and P. Piwnicki, “Relativistic effects of light in moving media with ex-
tremely low group velocity”, Phys. Rev. Lett. 84, 822 (2000) [cond-mat/9906332].
M. Visser, “Comment on Relativistic effects of light in moving media with extremely
low group velocity”, Phys. Rev. Lett. 85 (2000) 5252, [gr-qc/0002011].
U. Leonhardt and P. Piwnicki, “Reply to the Comment on Relativistic Effects of Light
in Moving Media with Extremely Low Group Velocity”, Phys. Rev. Lett. 85 (2000)
5253, [gr-qc/0003016].

[10] R. Schutzhold, G. Plunien and G. Soff, “Dielectric black hole analogues,” quant-
ph/0104121.

[11] C. Barceló, S. Liberati and M. Visser, “Analog gravity from field theory normal
modes?”, gr-qc/0104001.
C. Barceló, M. Visser and S. Liberati, “Einstein gravity as an emergent phenomenon?”,
gr-qc/0106002.

[12] J. B. Hartle and S. W. Hawking, “Path Integral Derivation Of Black Hole Radiance”,
Phys. Rev. D 13 (1976) 2188.

[13] G. W. Gibbons and S. W. Hawking, “Action Integrals And Partition Functions In

Quantum Gravity”, Phys. Rev. D 15 (1977) 2752.
G. W. Gibbons and S. W. Hawking, “Cosmological Event Horizons, Thermodynamics,
And Particle Creation”, Phys. Rev. D 15 (1977) 2738.

[14] R. Parentani and R. Brout, “Physical interpretation of black hole evaporation as a

vacuum instability”, Int. J. Mod. Phys. D 1 (1992) 169.

[15] R. Brout, S. Massar, R. Parentani and P. Spindel, “A Primer for black hole quantum
physics”, Phys. Rept. 260 (1995) 329.

[16] T. Damour and R. Ruffini, “Black Hole Evaporation In The Klein-Sauter-Heisenberg-

Euler Formalism”, Phys. Rev. D 14 (1976) 332.

[17] L. Parker, “The Production Of Elementary Particles By Strong Gravitational Fields”,

In Cincinnati 1976, Proceedings, Asymptotic Structure Of Space-time*, New York 1977,
107-226.

[18] U. H. Gerlach, “The Mechanism Of Blackbody Radiation From An Incipient Black

Hole”, Phys. Rev. D 14 (1976) 1479.

15
[19] P. G. Grove, “On An Inertial Observer’s Interpretation Of The Detection Of Radiation
By Linearly Accelerated Particle Detectors”, Class. Quant. Grav. 3 (1986) 801.

[20] B. L. Hu and A. Raval, “Thermal radiance from black hole and cosmological space-
times: A Unified view”, Mod. Phys. Lett. A 11 (1996) 2625.
D. Koks, B. L. Hu, A. Matacz and A. Raval, “Thermal particle creation in cosmological
spacetimes: A stochastic approach”, Phys. Rev. D 56 (1997) 4905 [Erratum-ibid. D 57
(1997) 1317] [gr-qc/9704074].
B. L. Hu, “Hawking-Unruh thermal radiance as relativistic exponential scaling of quan-
tum noise”, gr-qc/9606073.
A. Raval, B. L. Hu and D. Koks, “Near-thermal radiation in detectors, mirrors and
black holes: A stochastic approach”, Phys. Rev. D 55 (1997) 4795 [gr-qc/9606074].

[21] S. Massar and R. Parentani, “How the change in horizon area drives black hole evapo-
ration”, Nucl. Phys. B 575 (2000) 333 [gr-qc/9903027].

[22] M. K. Parikh and F. Wilczek, “Hawking radiation as tunneling”, Phys. Rev. Lett. 85
(2000) 5042 [hep-th/9907001].

[23] S. Shankaranarayanan, K. Srinivasan and T. Padmanabhan, “Method of complex paths

and general covariance of Hawking radiation”, Mod. Phys. Lett. A 16 (2001) 571 [gr-
qc/0007022].
S. Shankaranarayanan, T. Padmanabhan and K. Srinivasan, “Hawking radiation in
different coordinate settings: Complex paths approach”, gr-qc/0010042.

[24] R. Schutzhold, “On the Hawking effect”, gr-qc/0011047.

[25] W.G. Unruh, “Dumb holes and the effects of high frequencies on black hole evapora-
tion”, Phys. Rev. D 51, 2827 (1995) [gr-qc/9409008]. (Title changed in journal: “Sonic
analog of black holes and...”)

[26] T. Jacobson, “Black hole radiation in the presence of a short distance cutoff”, Phys.
Rev. D 48, 728 (1993) [hep-th/9303103].

[27] R. Brout, S. Massar, R. Parentani and P. Spindel, “Hawking radiation without trans-
Planckian frequencies”, Phys. Rev. D 52 (1995) 4559 [hep-th/9506121].

[28] K. Martel and E. Poisson, “Regular coordinate systems for Schwarzschild and other
spherical spacetimes”, Am. J. Phys. 69 (2001) 476 [gr-qc/0001069].

[29] T. Jacobson and G. Kang, “Conformal invariance of black hole temperature”, Class.
Quant. Grav. 10 (1993) L201 [gr-qc/9307002].

[30] I. Racz and R. M. Wald, “Global extensions of space-times describing asymptotic final
states of black holes”, gr-qc/9507055.

16