You are on page 1of 22

Semiclassical corrections to the photon orbits of a non-rotating black hole

Swayamsiddha Maharana∗ , Arundhati Dasgupta ,†


Physics and Astronomy,
Science and Academic Building
4401 University Drive,
University of Lethbridge,
arXiv:2011.00676v1 [gr-qc] 2 Nov 2020

Lethbridge T1K 3M4.

In this brief article we discuss the corrections to the photon orbits of a non-rotating black
hole due to semiclassical fluctuations of the metric. It is found that the photon orbit impact
parameter differences become of the order of the semiclassical fluctuations. We calculate
the effect of the semi-classical fluctuations on the photon orbits and we expect these to be
detected in future images of the event horizon obtained with greater resolution.

I. INTRODUCTION

The image of the black hole at the centre of the M87 Galaxy was obtained using VLBI and
announced in a series of papers [1]. A very remarkable achievement based on data from 8 infrared
interferometers, placed at various points on Earth, the image comprises of a central ‘shadow’,
surrounded by circular photon orbits. Whereas most galactic centre black holes have non-zero
rotation parameters, the shape of the image differs from the non-rotating one only by 4% [1]. In
this paper we begin the discussion of ‘quantum gravity’ corrections for this image by studying
the photon orbits which generate the non-rotating black hole image. We expect that our results
can be easily extended to the rotating example. Our calculations are valid for perturbations of
the metric originating from any existing quantum gravity theories, however, we compute the nu-
meric values of the corrections using the formulas found in [2]. In [2], semiclassical states in loop
quantum gravity (LQG) had been used to study corrections to the classical metric. Whereas, the
corrections are at the level of linear ‘perturbations’ of the metric, the form of the corrections are


MITACS summer student from IISER, Kolkata
† E-mail: arundhati.dasgupta@uleth.ca
2

non-polynomial in nature. Whether the predictions in [2] are true or not have to be verified us-
ing experiments. We predict the corrections using explicit numerical values, these are very small
∼ 10−9 M − 10−66 M (M being the mass scale of the black hole) and probably can be verified in
future images sensitive to distances between photon orbits and or interference fringes. The lower
limit of the range comes from the primordial blackholes whose radius is about 103 Planck length
and the upper limit of the range is for astrophysical solar mass black holes. In this article, we
follow the calculations of [3] for the metric of [2] and show how the orbits will change due to the
semi classical corrections. What we find interesting about the results is that the difference of the
photon orbit impact parameters with the critical impact parameter can be of the order of the semi-
classical fluctuations of the metric even at the third orbit for primordial back holes. We perform an
explicit numerical calculation using the corrections of [2] to see the physics of the semiclassical
fluctuations. We also try to obtain an analytic expression for the photon orbit corrections. We find
that for small black holes the effect on the orbits is rather drastic, but for astrophysical black holes
the nature of the correction is to slightly change the absorption cross-section. For small black
holes we identify an integer ‘n’ which characterizes the number of times the photons circle the
black hole, as a quantum number fixed by the semiclassical scale of the system. This number is
given as 2nπ ∝ − ln(t˜) and the classical limit is when n → ∞ and t˜ → 0. This n characterizes the
critical number of times the photon can rotate around a semiclassical black hole. We report on the
expected changes, but the photographic plate image construction is yet work in progress. Note this
calculation is highly restricted by the ‘semiclassical’ linear perturbation techniques. We expect
that non-perturbative quantum gravity will show the correct equations for the nature of photon or-
bits around quantum black holes. In the following section, we discuss the nature of the corrections
for generic perturbations of the metric, the section following that, we compute the coherent state
corrections and the exact numerical values. In the third section we conclude and discuss work for
the future.

II. THE CORRECTIONS TO THE GEODESIC AND THE PHOTON ORBITS

We take the ‘semi classically corrected’ Schwarzschild metric to be of the form


 
2
 rg 
2 1
+ t˜ hrr dr2 + r2 + t˜ hθθ dθ2

ds = − 1 − − t˜ htt dt + t˜ hrt dtdr +
r (1 − rg /r)
2 2
 2
+ r sin θ + t˜ hφφ dφ (1)
3

t˜ is a semiclassical parameter, and hi j (t, r, θ, φ) (i, j = t, r, θ, φ) are metric fluctuations which are
coordinate dependent. These corrections can arise due to quantum gravity, quantum energy mo-
mentum tensor fluctuations of matter fields etc. rg = 2GM is the Schwarzschild radius. The form
of the corrections, and the fact that there is only one cross term htr is motivated from the semi
classical corrections obtained in [2].
Using the calculations of [3] and [4], we take the geodesics in the θ = π/2 plane or the equa-
torial photon orbits and calculate their general behaviour. The geodesic equation up to O (t˜) semi
classical corrections is given as (where L is angular momentum, and E the energy of the system)
 2
1 L2 E 2
 
dr
+ + =0 (2)
ds g q̃ f
where the functions are appropriately defined as in [2].
 
 rg  1
q̃ = r2 sin2 θ + t˜ hφφ .

f = − 1 − − t˜ htt g= + t˜ hrr (3)
r (1 − rg /r)

Using the conservation of angular momentum equation i.e. q̃ dφ


ds = L gives from equation (2)

1 dr 2 1 1
   
1
+ + =0 (4)
q̃2 dφ g q̃ f b2
where ‘b’ is the impact parameter defined as
L2
b2 = (5)
E2
We can then write the above equation using the explicit forms of the functions as defined in (3)

 2  
1 dr 1 − rg /r 1 1
+ − =0 (6)
(r2 + t˜hφφ )2 dφ 1 + t˜hrr (1 − rg /r) r2 + t˜hφφ b2 (1 − rg /r − t˜htt )
Using t˜ as a small parameter, one can do binomial expansion of the following (Note as r > 2M the
factor 1 − rg /r is not zero, and can be used the factor out of the binomials and as the semiclassical
parameter 10−9 > t˜ > 10−66 ).
" 
1 dr 2  hφφ −1
  
rg    rg −1 1
+ 1− 1 + t˜ hrr 1 − 1 + t˜ 2
r4 dφ r r r2 r
#
−1 
hφφ 2
 
1 htt
− 2 1 − t˜ 1 + t˜ 2 = 0. (7)
b (1 − rg /r) 1 − rg /r r

Keeping order t˜ terms in the binomial expansions one gets:

1 dr 2 t˜ hφφ
       
1 rg  1 hφφ htt
+ 2 1− − 2 1 + t˜ 2 − t˜hrr − 2 + = 0. (8)
r4 dφ r r b r b r2 1 − rg /r
4

In the above we have used Binomial expansion in powers of t˜. If we re-write the above in a
convenient form we get
 2    
1 dr 1 rg  hφφ 1 hφφ htt
+ 1 − 1 + ˜
t − t̃hrr = 1 + 2t˜ − t̃hrr + ˜
t . (9)
r4 dφ r2 r r2 b2 r2 1 − rg /r

The above equation is of the form

1 dr 2
 
1
4
= 2 H1 (r) −V1 (r) (10)
r dφ b
   
r  h h htt
where V1 (r) = r12 1 − rg 1 + t˜ rφφ2 − t˜hrr and H1 (r) = 1 + 2t˜ rφφ2 − t˜hrr + t˜ 1−r g /r
. The equa-
tion will have a solution iff
1
H1 (r) −V1 (r) ≥ 0. (11)
b2
From this we identify ‘the potential function’ as
1 V1 (r)
2
≥ ≥ V (r). (12)
b H1 (r)
where the potential function V (r) to order t˜ is identified as
!
1 rg  hφφ htt
V (r) ≡ 2 1 − 1 − t˜ 2 − t˜ r  . (13)
r r r 1 − rg

To find the extremum of the potential we take the derivative of the potential and set it to zero.
∂V
|r=rc = 0. (14)
∂r
That gives
  
1 3rg hφφ htt
−2 + 1 − t˜ 2 − t˜
r r r 1 − rg /r
hφφ ∂r hφφ
 
 rg  ∂r htt htt rg
+t˜ 1 − 2 3 − 2 − + = 0. (15)
r r r 1 − rg /r (1 − rg /r)2 r2
At the zeroeth order the above gives the critical radius to be r0 = (3/2)rg = 3M. This is an
‘unstable’ orbit, where the potential has a maximum. We next assume a correction to this critical
radius which is given by
rc = r0 + t˜ ξ.

The correction can be solved as


 
1 htt (r0 ) 3 ∂r htt (r0 )
ξ= 2hφφ (r0 ) − r0 ∂r hφφ + r0 rg − r0 . (16)
9rg (1 − rg /r0 )2 1 − rg /r0
5

A limit on the value of the ‘impact parameter’ can be found using the fact that (dr/dφ)2 ≥ 0;
as in (12) which is explicitly:

r
!!
1 1 − rg hφφ htt
≥ 1 − t˜ + r . (17)
b2 r2 r 2 1 − rg
Given that the potential has a maximum at rc this is same as

b ≤ (V (rc ))−1/2 . (18)

Which shows that there is a critical impact parameter bc = (V (rc))−1/2 . This is the critical impact
parameter, after which the photon is absorbed into the black hole, and cannot escape back to the
asymptotics. As the potential as well as the critical radius is corrected, we get a new ‘inner disk’
radius and hence a corrected absorption cross section for the black hole [6].
As stated above, the photons reaching the black hole with an impact ≤ bc are captured by the
black hole. Thus there is a ‘hole’ of radius bc a disc from within which light does not reach the
observer. If we inspect the corrections, then they are tiny. From the discussions in [2], one takes
the semiclassical parameter, in a certain range, depending on the ratio of the length scale of the
space-time and the Planck length. This range is 10−9 ≤ t˜ ≤ 10−66 , and therefore, the correction
to the impact parameter, the disc radius and the absorption cross section of the black hole is very
small. Given the resolution of the current image [1], it shall be difficult to discern the semiclassical
image corrections to the absorption cross section which is the area of the sphere with radius bc .
However, we make an observation based on [3] that the difference of impact parameters of photon
orbits which encircle the black hole with the critical radius is of the order of the semiclassical
parameter, and we try to interpret the physics. This stems from the fact that [3]

b − bc = 3.4823 M exp(−µ − 2nπ) (19)

where n represents the number of times a photon encircles the black hole horizon. We observe that
exp(−2nπ) ∼ t˜ for n = 3 for t˜ ∼ 10−9 a primordial black hole ([2]) and n=24 for t˜ ∼ 10−66 which
is a solar mass black hole. If this is true, then, the semiclassical fluctuations of the metric might
be dominant as t˜ sets the scale of the corrections to the above equation. To investigate this, we
solve for the geodesic equations of the semi-classically corrected metric. We follow the methods
of [3, 4] for the corrected metric (1). We take the equations for (8) and re-write in terms of u = 1/r
6

and separate the classical and order t˜ parts of the equation. Defining the following quantities as

u2 1
G0 (u) ≡ u3 − + (20)
2M 2Mb2
    1
 !
1 1 1 htt
H(u) ≡ u2 (1 − 2Mu)2hrr − u4 (1 − 2Mu)hφφ + 2 u
2M u u b (1 − 2Mu)
!
hφφ u1 u2 (1 − 2Mu)hrr u1

1
+ 2 − (21)
2M b2 b2

G(u) ≡ G0 (u) + t˜H(u) (22)

where r = u1 .
In the above, H(u) = G0 (u)[(1 − 2Mu)hrr − u2 hφφ ] + 1/b2 [hφφ u2 + htt /(1 − 2Mu)]. Substituting
1
r= u in the geodesic equation (8) gives
 2
du
= 2MG(u) (23)

G0 (u) is a cubic, and has three roots, in the classical limit; u01 , u02 , u03 . For which u01 < u02 < u03 and
u01 < 0. From [3] these are taken as

Q0 − P0 + 2M 1 Q0 + P0 − 2M
u01 = − u02 = u03 = (24)
4MP P0 4MP0

where Q20 = (P0 − 2M)(P0 + 6M) ; P0 being the location of the Periastron; and the impact
parameter is solved as by setting G0 (u) = 0

P03
b0 = . (25)
P0 − 2M

For the impact parameter at b = b0c , P0 = Q0 = 3M (unless stated otherwise x0 or x0 physical


quantities labelled with 0 represents a classical number) . In the G0 (u) function, without the
semiclassical fluctuations, at P0 = 1/3M there is a double root. When we try to solve Equation
p
(23) the integral of 1/ G0 (u) in the t˜ = 0 limit (or classical limit) can be approximated as ∼
R
du/(u − u2 ), which obviously has a logarithmic divergence at u = u02 = 1/3M. However, with
the introduction of a shift from this as u02 = 1/[3M(1 + δ)] the degeneracy of the roots is broken
and the integral is no longer divergent at u = u2 , but as expected the infinity is regulated as ln δ
which diverges as δ → 0. Due to the corrected form for G(u), we take the order t˜ corrections to the
above roots. Using the same derivation as in [3, 4] we take the Periastron distance P as a function
of the second root u2 = 1/P. The periastron distance in the corrected and uncorrected geodesic
7

are given as P and P0 respectively. Only O (t˜) corrections are considered to the periastron distance.
So,
1 1
u2 ≡ = + t˜ν (26)
P P0
1
and u02 ≡ P0 As P0 = r0 the correction to u2 is taken as ν = −ξ/r02 = −ξ/(3M)2 . If we deviate
away from the 3M then ν has a correction proportional to the deviation, at this level of ‘first order’
discussions we ignore that contribution to the discussions that follow next. Since P is the periastron
distance (the closest point or the turning point of the trajectory as the particle scatters off the black
hole)
!2
1 du
= 0 = G(u2 = u02 + t˜v). (27)
2M dφ u(φ)=u2
We can solve for the ν by observing that the corrected G(u) can be written as

G(u02 + t˜ν) = G0 (u02 ) + t˜νG′ (u02 ) + t˜H(u) = 0. (28)

If in addition we assume that the impact parameter is corrected upto t˜ as b = b0 + t˜b̃, one gets:

(u02 )2 1 t˜b̃
(u02 )3 − + 2
− 3
+ t˜νG′ (u02 ) + t˜H(u02 ) = 0. (29)
2M 2Mb0 Mb0

When u02 = 1/3M the system has a double root, and G′ (u02 ) is zero. The above equation can
however be used to solve for b̃ as
b̃ = Mb30 H(u02). (30)

When u02 6= 1/(3M) one gets


b̃ = Mb30 (G′ (u02 ) ν + H(u02 )). (31)

which also can be written as



− νG′ (u02 ) = H(u02 ) (32)
Mb30
It is difficult to compute the integral of the differential equation (23) analytically as the semiclassi-
cal function has a quintic. Numerical values of the integral for the semiclassical corrected equation
differ from the classical integral at the order of t˜ which is going to interfere with the classical cal-
culation of the impact parameter. We have used explicit expressions for the corrections to the
metric from [6]. The details of the expression can be found in the next section. Here we present
the numerical calculations to motivate the analytic calculations. Here we calculate
Z u2
1 du
φ∞ = √ p (33)
2M 0 G(u)
8

Where the photon traverses from r = ∞, u = 0 to r = 3M(1 + δ) + t˜ν; u = u2 the periastron, and
the angle φ changes from 0 to φ∞ . In the table, we have the description of columns (i) the value
of the semiclassical parameter t˜. (ii) The deviation from the maximum of the potential for the
Periastron distance δ (iii) The exact numerical value of the integral obtained using Mathematica
(iv) The value of the integral without the order t˜ terms in G(u) (v) Comparison of the classical
value with the semiclassical one labelled as ∆φ∞

t˜ δ Exact numerical integral Integral without semiclassical correction ∆φ∞


10−9 10−8 19.588629507213 19.588629510149 2.936 ×10−9
10−20 10−18 42.61448042675600592 42.61448042675600334 ∼ 10−15
10−12 10−11 26.496384771569947 26.496384775811023 ∼ 10−9
10−9 10−7 17.28604453548423 17.2860445371548 ∼ 10−8
10−12 10−10 24.193799682753037 24.193799682936977 ∼ 10−10
10−15 10−12 28.7989698687928421 28.7989698687930655 ∼ 10−13

The exact integral differs from the one without the semiclassical parameter almost to order t˜.
How would these affect the image of the event horizon? This is a very interesting question. For
the purposes of this paper we see how the impact parameter as a function of δ and therefore the
scattering angle is modified due to the semiclassical corrections. In fact as we have seen in [3], the
b − bc for the photon orbits with different scattering angles differ by similar order of magnitude.
In [3] using φ∞ = π/2 + µ/2 one finds from the classical integral

b − bc = 3.4823M exp(−µ). (34)

As µ is the scattering angle, as shown in the figure, photons which emerge from the black hole after
encircling the horizon n number of times, also emerge at the same angle, but the impact parameter
differs. And thus we get
bn − bc = 3.4823M exp(−µ + 2nπ). (35)

As n → ∞ b → bc . However, our discussion is about let us say n = 3, then b3 − bc ≈ 2.39 ×


10−9 M for µ = 1 radians and therefore for a photon traversing back to the photographic plate
after encircling the black hole three times, the semiclassical fluctuations will be relevant. How
does the corrections to φ∞ observed in the table above due to inherent semiclassical fluctuations of
the metric affect the impact parameters? For that, we have to solve the equation analytically and
obtain a functional relation between φ∞ and the impact parameter b. However given the quintic
9

nature of the function G(u) analytical computations could not be obtained, neither did MAPLE
or MATHEMATICA give us analytic results. We therefore obtained an approximate value for the
integral analytically which we discuss next.
To estimate analytically what the new physics might be, we approximate the square root using a
linear order in t˜ expansion. From equation(23) if we observe the structure of the G(u), then it is of
the form H(u) = G0 (u)[(1 − 2Mu)hrr − u2 hφφ ] + 1/b2[hφφ u2 + htt /(1 − 2Mu)]. The first term has
a double root at 1/3M, however the second term has one root. Including the quantum correction,
we have about z = u − u2 ∼ 0 a form of G(u) = z2 + zX where X 6= 0 at u = 1/3M. If we complete
the square and evaluate the integral we have ‘schematically’
Z r !
dz X X
√ ∼ ln z + + z2 + z . (36)
z2 + zX 2 2

As long as z 6= 0, we can approximate the above as


r !
X X X
ln(z) + ln 1 + + 1+ ∼ ln(z) + . (37)
2z 2z 8z

Thus at order t˜ i.e. first order in X we get a correction of the form 1/z. Whereas we are taking X < z
to facilitate the above, obviously the functional dependence of z on φ∞ and the scattering angle are
different from that which is used to find the current image. To see the nature of the correction at
order t˜ we make a analytic calculation based on the following discussions and approximations:
At u = u2 = u02 + t˜ν we find that we can write the G(u) as G0 (u2 ) + t˜H(u2) = G0 (u02 ) +
t˜νG′ (u02) − t˜ Mb
b̃ 0
3 + t˜H(u2 ), the first term by itself is zero, and the combination of the t˜ terms cancel
0
each other. Thus we can use this split of the terms to obtain the function around u ≈ u2 , and in
general. (Note that this discussion is true if and only if u2 6= 1/3M.) This allows an approxi-
mation to the integral as a ‘binomial expansion’ in the small parameter t˜. We use the notation
H(u) + νG′ (u2 ) − Mb

3 = H̃(u).
0

Z u2  
1 1 t˜ H̃(u)
φ∞ = √ du p 1− (38)
2M 0 G0 (u) 2 G0 (u)
 1 Z u2 Z u0
P0 2 1 1 t˜ 2 H̃(u)
= 2 (K(k0 ) − F(ζ∞ , k0 )) + √ du p − √ du 3
Q0 2M u02 G0 (u) 2 2M 0 (G0 (u)) 2
(39)
10

where

Q20 ≡ (P0 − 2M)(P0 + 6M) (40)


Q0 − P0 + 6M
k02 ≡ (41)
2Q0
Q 0 P0 + 2M

sin2 ζ∞ ≡ (42)
Q0 − P0 + 6M
we put the integral in the third term as I(u2 ) which gives additional contribution to the solution for
φ∞ , apart from the contribution from the Elliptic terms. The second term almost remains constant
over the interval and we label that as E1 .
In the event that δ ≈ t 1/2 or smaller the I(u2 ) term contributes non-trivially and we investigate
this in a separate subsection. Setting φ∞ = π/2 + µ/2 where µ/2 is the angle of scattering, one
gets an equation for the impact parameter b. We try to solve the integral in the regime that the
periastron is very close to 3M.
The requirement that P0 > 3M comes from the restriction u1 < u2 < u3 which gives a condition
Q0 + P0 − 6M > 0 [4].
Let
P0 = 3M(1 + δ) (43)

where δ is a small number above the Mass of the blackhole. The equation which relates δ to
φ∞ also gets corrected. Hence,
Substituting equation (43) in equation (39) and then exponentiating both sides one gets
 π µ   
1 I(u2)
3.21 exp (−φ∞ ) = 3.21 exp − − = δ exp √ −E1 + t˜ . (44)
2 2 2M 2
The interesting aspect of this calculation is that for any µ as the number of cycles increases,
very soon the order of the corrections become comparable with the semiclassical corrections to
the Periastron. Notice in the above that the integrals E1 and I(u2 ) depend on the δ this is a tran-
scendental equation and cannot be solved if the exact form is not known. In the next section we
use the methods of [6] to calculate the exact form of these. The impact parameter is resolved as
follows:

htt (u02 )
 
1 1 2
 0 2 0 1 2
 0

= 1 − 3δ 1 − ˜
t u 2 h φφ (u 2 ) − ˜
t = 1 − 3δ 1 − ˜
t a(u 2 ) (45)
b2 27M 2 1 − rg u02 27M 2
h (u02 )
where we have introduced ã(u2 ) = u02 2 hφφ (u02 ) + 1−r
tt
0 to make the calculation easier.
g u2
11

√ t˜
  
3 2 0
b = 3 3M 1 + δ 1 + a(u2 ) . (46)
2 2
Next we use the form of the δ obtained from (16) to find an equation which relates the impact
parameter to the scattering angle if we ignore the contributions from E1 and I(u2).
√ t˜
 
0
b = 3 3M (1 + 15.48 exp(−π − µ)) 1 + a(u2 ) . (47)
2

for δ ∼ t˜1/2 , we find that the orbits differ from the critical one by an amount equal to the quantum
fluctuation, as that being t˜ dominates over δ. If the images are eventually sensitive to be able
to differentiate the bn th orbit from the bn+1 th orbit, then the presence of quantum fluctuations
will ‘wash’ out the differences after a certain orbit. As we see this can happen even at the third
orbit onwards depending on the semiclassical parameter. In the next section we calculate the
a(u2 ), I(u2), E1 as from LQG coherent states and try to see the nature of the above equation in the
case that the semi-classical fluctuations of the metric are dominant over the classical difference in
the impact parameter from the critical values.
In fact if we compute explicitly to order t˜

 
b − b0c 0
= 3.48 M exp(−µ) 1 − a(u2 ) . (48)
2

As is obvious from the above, for t˜ ≈ 10−9 a semiclassical fluctuation will have the same impact
parameter as the one encircling the black hole thrice and reaching asymptotic region. However,
t˜ ≈ 10−9 only for primordial black holes with horizon of the order of 103 Planck lengths. For solar
mass black holes this t˜ ≈ 10−66 here the semi-classically corrected impact parameter would be of
the order of the impact parameter of the 24th orbit.
To find an explicit numerical values of the corrections, we use the form of the semiclassical
metric as obtained in [2]. The functional form of the solution of the equation (44) can only be
obtained after we have found the integrals E1 and I(u2).

A. Nature of the potential obtained from LQG Coherent States

The semiclassical corrections to the non-rotating black hole were predicted in [5], computed
in [2] and then discussed in details regarding their usefulness in making observations in a gauge
invariant context in [6]. Here we briefly give explicit expressions for the corrections to the metric
as predicted in [2]. The Schwarzschild metric is well known in spherical coordinates, but the time
12

Semi-classical trajectory

Source

Periastron π−μ

observer
ϕ(u)

FIG. 1: The impact parameter increases due to semiclassical physics, but φ∞ changes in a complicated way
as a function of δ

slicing is not very convenient for the LQG graph embedding. If one takes the Lemaitre coordinates
as in [5], then the time slices are flat. The slices also extend into the horizon up to the singularity
such that one can address questions about singularity resolution as in [7]. The coherent states are
obtained in this slicing, and semiclassical corrections to the metric are computed in these using
techniques of LQG. The corrections to the flat metric in the three slices are as shown in [2]
  
ab ab Per
qcorr = q 1 + 2 t˜ f (49)
Sea
where  
1 1
f (x) = − coth(x) (50)
x x
and qrr = 1, qθθ = 1/r2 , qφφ = 1/(r2 sin2 θ) and Pea is the appropriate LQG momenta for edges
ea of a three dimensional graph, embedded in the time slices. In LQG the phase space is defined
on a discretization of the three manifold in which the canonical degrees of freedom are defined.
The three metric qab is re-written in soldering forms eIa (a=1,2,3 is the index on the three manifold
13

world volume and I=1,2,3 is the tangent space index), such that eIa eIb = qab a set of densitized

triads EIa = qeaI when smeared over two dimensions surfaces which comprise the discretization
of the three manifold are the above momenta. In the following the ea is the labelling of the edges
in the ‘a’ direction in three dimensions. The densitized triads are smeared over two dimensional
surfaces Sea (the details of this can be found in [5, 7]). These surfaces form the ‘dual’ to the graph
embedded in the three spatial slicing of the Lemaitre metric.
Z
PeIa = ∗E I (51)
Sea

and one can use a ‘gauge invariant’ combination (inner product in the tangent space index I)
q
Pea = PeIa PeIa . (52)

For the ‘momenta’ induced on the graphs, one obtains (Sea → 0):

Per r2 sin θ
= (53)
Ser rg2
Peθ r sin θ
= (54)
Seθ rg
Peφ r
= . (55)
Seφ rg

If one sees what these are, they are the gauge invariant densitized triads of the LQG multiplied
by 1/rg the Schwarzschild radius to make the quantity dimensionless.
From the Lemaitre coordinates we make a transformation back to the Schwarzschild coordi-
nates to obtain a ‘semi-classically corrected’ metric. Details of this can be found in [2],
and the transformed semi classically corrected perturbations are:
  !!
rg Per 1
htt = −2 f = −2rg3 u3 rg2 u2 − coth 2 2
r Ser rg u
!!
rg2 u2
 
1 Per 1
hrr = −2 f = −2 r2 u2 − coth
(1 − rg /r)2 Ser (1 − rg u)2 g r g u2
2
    
2 Peθ 2rg 1
hθθ = −2r f =− rg u − coth
Seθ u rg u
!
Peφ
  
2 2 rg 1
hφφ = −2r sin θ f = −2 rg u − coth (56)
Seφ u rg u

where we have taken 1/r = u and set sin θ = 1 in the formulas.


14

Hence, the critical radius is given as (16), we find the explicit values of the functions at r0 = 3M,
as

−.20734
htt = 0.343 ∂r htt = hφφ = 5.2574M 2 ∂r hφφ = 3.096M (57)
M
and
ξ = 2.030 M. (58)

Given the critical radius, one can compute the critical impact parameter, beyond which the
light rays get absorbed by the black hole. The light rays which fall on the black hole at this radius,
follow an unstable radian geodesic. Light rays which are incident on the black hole with impact
parameter bigger than the critical impact parameter escape back to the asymptotic, but they can
encircle the black hole n-number of times, before emerging as discussed in the previous section.
Here the critical impact parameter is given as

bc = 3 3M (1 + 0.8066 t˜) . (59)

For the semiclassical corrections mentioned in this section we finally compute the impact pa-
rameter for the n-th orbit as : Given (16);

 
1
ã = −.2389 + 0.2004(3δ) (60)
3M(1 + δ)
Using equation (46)

√ 3
b=3 3 M+ (9Mδ2 ) + 4.1895 t˜ M − 3.509 t˜ 3Mδ (61)
2

where we have ignored the integrals E1 and I(u2). Note that the order t˜ term is the contribution to
the impact parameter.
This discussion is relevant for δ > t˜, but as we found before, about the 24th orbit and higher
for t˜ ∼ 10−66 , the difference of the impact parameter and the critical radius is of the order of the
semiclassical parameter for solar mass black holes. Thus the question is what happens 25th orbit
onwards? Next we try to compute the integral in Equation (70), with an explicit formula to see the
corrections. Note that in the Equation for φ∞ as the zeroeth order term is ln(δ) at order t˜ we keep
only the terms which can compete with the same i.e. t˜/δ; t˜ ln δ and ignore terms which contribute
as t˜δ and higher. We compute:
15

Z u0
2 H̃(u)
I(u2 ) = du 3 . (62)
0 (G0 (u)) 2
This using the definitions in Equations (22, 39) we find the above as
Z u0
"  !
    1
2 du 1 1 1 h tt
u2 (1 − 2Mu)2 hrr − u4 (1 − 2Mu)hφφ + 2 u
0 G0 (u)3/2 2M u u b (1 − 2Mu)
!# Z 0
hφφ 1u u2 (1 − 2Mu)hrr 1u b̃/Mb30 − νG′0 (u02 )

1 u2
+ 2 − − du (63)
2M b2 b2 0 (G0 (u))3/2

We concentrate on the first integral; by collecting the terms proportional to hφφ and hrr one can
factorize one power of G(u). One gets the integrals as
Z u0
"
(1 − 2Mu)hrr 1u

1 2 u2 hφφ u2 hφφ
p −p + 2
2M 0 G0 (u) G0 (u) b (G0 (u))3/2
#
1 htt
+ 2 du (64)
b (1 − 2Mu)(G0(u))3/2

The first two terms in the integral are not divergent as functions of δ where P = 3M(1 + δ) , but
the second two terms have singular behaviour with δ. In our analysis we keep terms which are
divergent as functions of δ to solve the equation analytically. We compute them explicitly in the
next section, but the divergence comes from the δ dependence in the Elliptic integral, as well as
the dependence on u − u2 ∼ e from the power of (G0 (u))3/2 . However as we show in the appendix
this potential divergence is cancelled in the total I(u2) due to contributions from the second term
of Equation(63).
From Equation(64) the second two terms are (2M factored out)
Z u0 2

2 1 u hφφ + htt /(1 − rg u) du
2
. (65)
0 b (G0 (u))3/2

Using the explicit formulas for the hφφ and htt from the equations (56), one gets the integral as
Z 0
rg u − 2rg2 u2 + 2rg3 u3 − rg5 u5 du

2 u2
. (66)
b2 0 (1 − rg u)(u01 − u)3/2 (u02 − u)3/2 (u − u03 )3/2

We have approximated the coth(1/rgu) and coth(1/rg2 u2 ) as 1 in the above as the value of the
function varies from 1.00 to 1.023 in the domain of definition of u. The corrections will be pro-
portional to exp(−1/rg u) and exp(−1/rg2 u2 ) in the integrand and can be ignored at this level of
the approximation. The explicit form of the integral is given in the Appendix, with the explicit
16

form of the u01 u02 and u03 obtained upto quadratic powers of δ when P0 = 3M(1 + δ) plugged into
the formula. Here δ is a dimensionless number to facilitate the calculation. From the terms in the
appendix, we find that the terms proportional to 1/e as well as inverse powers of δ dominate. We
find the equation to be from the Appendix Equations (102) for I(u2 ) and (103) for E1 in Equation
(44)

3.21 exp(−φ∞ ) = 3.21 exp(−π/2 − µ/2) (67)


!
0.47t˜1/2 t˜
= δ1−0.13t˜ exp + − 0.017t˜ + 0.73 (68)
(0.67δ + 0.225t˜)1/2 δ
= w(δ) (69)

We find that the plot of the function w(δ) is mostly linear for a range of values t˜ < 10−20 . For
10−9 > t˜ > 10−20 , the graph shows a ‘turning’ behaviour at order t˜. We could interpret this as
a breakdown of the classical approximation. This is justified in hindsight as the bc is corrected
to order t˜ and therefore the angle of scattering stops at a finite value. After this the metric’s
behaviour for the solution of geodesic changes. We refrain from commenting on the interpretation
of the results, but we define a quantum integer n such that exp(−2nπ) = t˜ or 2nπ = − ln t˜ at which
the photon orbits circulation of the horizon ceases, and the critical radius is reached. As δ > t˜,
the linear behaviour is retained, the straight line has a slightly different slope and an intercept than
the classical graph, however as δ ∼ t˜, the graph starts deviating. Note, we require a more rigorous
calculation than this to identify the quantum behaviour of photons. This is a calculation to show
that semiclassical fluctuations are important for photon trajectories with high ‘n’ number of circles
around the horizon.

III. CONCLUSION

Due to the fact that the distance between the photon orbit impact parameters diminish to tiny
scale times the black hole mass; we find that the classical behaviour is almost of the order of a
semi-classical parameter for coherent states used in [2]. This motivates us to find if the corrections
to the trajectories studied in [2] will be relevant in the observation of the EHT image, i.e. if there
are any further physical implications. We find a finite orbit number ‘n’ for the photons after which
the impact parameter deviates again from the critical value. We show that this might be a sign
of quantization 2nπ = − ln(t˜) = ln(1/t˜) = ln(10m ) = m ln(10) the exponent of the semiclassical
parameter. There is a finite integer n or number of times a photon can encircle a horizon. The
17

FIG. 2: t˜ ∼ 10−9

physical implications of this on a photographic plate are under investigation. As the results are
dependent on the semiclassical states, the actual image observed might be different, depending on
the theory used to find (46). We are currently investigating the details of the image formation.

IV. APPENDIX

The Integral is
Z u0
2 2 (rg u − 2rg2 u2 + 2rg3 u3 − rg5 u5 ) du
I˜ = 2 (70)
b0 0 (1 − rg u)(u01 − u)3/2 (u02 − u)3/2 (u − u03 )3/2

Using a factorization, the integral becomes


Z u0
2 2 rg u[1 − rg u + rg2 u2 + rg3 u3 ]
= 2 du (71)
b0 0 (u01 − u)3/2 (u02 − u)3/2 (u − u03 )3/2

The integral requires Elliptic functions, and we use MAPLE to compute. However, to enable
MAPLE to give the answer correctly, we had to redefine the variables. We use

u02 − u u01 − u03


x2 = k0 k0 = (72)
u01 − u u02 − u03
18

FIG. 3: t˜ ∼ 10−20 in two different ranges of the x-axis and in the third t˜ ∼ 10−66

Using this substitution we get the integral to be of the form


s
Z u02
k0
u01 ax8 + bx6 + cx4 + dx2 + f
Λrg dx (73)
0 x2 (1 − x2 )3/2 (h − x2 )3/2

where Λ, a, b, , c, d, f , h are functions of u01 , u02 , u03 . We also have to introduce a parameter e as there
is a new divergence at x = 0. We thus take the limit as x = e to regulate the divergence. The result
of the integral using MAPLE is
h n    q 
(h−1)
1
ah3 + − a2 + 2b h2 + 2c + d2 + f h − 2f F x, 1h
 
− (h−1) 2 2 − h3/2 (74)
 q 
+E x, 1h ah4 + 2b − a h3 + a + b2 + c + d2 + f h2 + d2 − f h + f }
   
(75)
n   
1 √ 4 1 ax4 − 1 x2 a + bx4 + − f − a − b − c − d x2 + f h3

+ x h2 √−x2 +1 −x2 +h
h 2 2 2 2 2 2 2 (76)
    2
oi
+ 2a + b2 + c + d2 + f x4 − 2c + 2f x2 − f h2 + (x−2)((d−2 f 2)x − f )(x+1)h + f x4 − f x2

(77)
19

Upon computing the integral and writing the explicit values of u1 , u2 , u3 the answer is obtained
using MAPLE. The coefficients are individually
29 59 97 2
a = + δ− δ (78)
81M 81M 243M
116 818 1636 2
b = − − δ+ δ (79)
81M 243M 729M
58 464 250 2
c = + δ− δ (80)
27M 81M 81M
116 346 28 2
d = − − δ− δ (81)
81M 81M 27M
29 287 149 2
f = + δ+ δ (82)
81M 243M 729M
4 8
h = 1 + δ − δ2 (83)
3 9
√ M 7/2  
4
Λ = −9 2 2 1+ δ (84)
δ 3
The lower limit of the integral is 0 and the upper limit is 1 − (1/3)δ + (1/2)δ2.
From this we subtract   Z u0
b̃ ′ 0 2 du
3
− νG (u 2 ) (85)
Mb0 0 G(u)3/2
This can be transformed using the same variables and Equation(32) as above to a integral of the
form

Z
(x2 − h)5/2
H(u02 )Λ dx (86)
x2 (1 − x2 )3/2
where  
58 118 2 250 2
Mb20 (H(u02 )) = − δ + δ (87)
81 81 81
The result of the integrals is function of Elliptical Integrals as shown by MAPLE.
r ! r !!
2 1 1 1
(h − )(h − 1)F x, + h(h2 − h + 1)E x,
h 2 h h
2 p 
1

h 2
2 2
− √ h − x2 h −h+ x − (88)
x 1 − x2 2 2
(The worksheet is available on request).
In the integral there are also terms of the form E(e, k) and F(e, k) which we use the small e
expansion of the Elliptic functions [8].


x2m+1
F(x, k) = ∑ F1 (−m, 1/2; 1; 1 − k2) (89)
m=0 2m + 1 2
20

we keep the m = 0 term, which gives for small x

F(x, k) ≈ x (90)

For the Elliptic E(φ, k) integral, for the x ≈ 1, k ≈ 1 we use the asymptotic forms as given in
[9]:

 
2 ′ ′ ′ ′ −1 1
+(1−k′2 sin2 φ) cot φ+O(1+tan2 φ)1/2 cot2 φ(d0′ −..

E(x, k) = E(k)− K (k ) − E (k ) sinh
π k′ tan φ
(91)
where x = sin φ. The constants d0′ are functions of k′ and tend to zero as k′ → 0.
where K(k), E(k) is the complete Elliptic functions of the first and second kind, and F(a, b; c, z)
is the Hypergeometric function, a series on positive powers of z.
For the Elliptic F(φ, k) integral we have
 
2 ′ −1 1
F(φ, k) = K − K sinh ′
+ (1 + k′2 tan2 φ)1/2 cot2 φ(c′0 − ..) (92)
π k tan φ
and c′0 is dependent on k′ and tends to zero as k′ → 0. The first term in Equation(92) has singular-
ities proportional to ln (4/k′ ) as k → 1.
We find the that the incomplete integrals can be approximated using the above as
 
4 −1 1
F(φ, k) ≈ ln ′ − sinh (93)
k k′ tan φ
√ 
    
1 2 1  δ 1
≈ ln 4 − ln δ − ln(1 + δ) − ln √ − ln √ 1 + 3 + √ 1 − √
2 3 2 2 2 3
and √ !
2δ 2 3 1
E(φ, k) ≈ 1 + (1 − δ) ln √ − ln(1 − δ) − (94)
3 δ 2
The terms from the Integral which contribute to the equation as δ → 0 are the following
(MAPLE worksheet is available on request):
−113280δ2 E(φ, k) 92160δ2 F(φ, k)
√ +√ (95)
9 + 12δ − 8δ2 η(δ) 9 + 12δ − 8δ2 η(δ)
√ √
4864 δ2 3 24δ − 40δ2 101952δ E(φ, k)
− √ − (96)
24δ − 24δ2 η(δ) η(δ)

where η(δ) = (3δ − 2)(3δ2 − 2δ + 12)(9 + 12δ − 8δ2)2 (6 − 2δ + 3δ2 )(e2 − 1). where
1 − 1 δ + 21 δ2
φ= q 3 (97)
1 + 43 δ − 98 δ2
21

and r
4 8
k= 1 + δ − δ2 (98)
3 9
Note that the computer algorithm keeps all powers of δ as generated from the integral and the
upper and lower limits. These are then approximated using the Elliptic integral approximations as
     
2 2 1 2 2 2δ 1
= −3.23δ + 2.64δ 0.58 − ln δ − 0.722δ − 8.74δ 1 − δ 1+ 0.74 − ln δ
2 3 3 2
(99)
The contribution from I(u2) is therefore

  
1 1 2 2 1
√ Λ −3.23δ + 2.64δ 0.58 − ln δ (100)
2 2M (2M)3/2 b20 2
   
2 2 2δ 1
−0.722δ − 8.74δ 1 − δ 1+ 0.74 − ln δ
3 3 2
1
0.20δ2 − 8.74δ + 1.59δ2 ln δ

= − (101)
12δ2
Thus in total the contribution to φ∞ is
1 0.73
√ I(u2) ≈ −0.017 + − 0.13 ln δ (102)
2 2M δ
For the E1 integral, we simply take that as
q
u02 − u2 √
r
0
(u2 − u2 ) 0.225t˜
= −q ≈− 2M (103)
δ ˜
p
G0 (u2 ) 0 0
(u1 − u2 )(u2 − u3 ) 0.67 + 0.225 t

This gives a rather strange t˜1/2 dependence, but we keep it for the calculation of the φ∞ as a
function of δ.

[1] The Event Horizon Collaboration, Astrophysical Journal Letters 875 (2019) 3.
[2] A. Dasgupta, Journal of Cosmology and Astroparticle Physics, 05 011 (2010).
[3] J. P. Luminet, Astronomy and Astrophysics, 75 228 (1979).
[4] S. Chandrasekhar, The Mathematical Theory of Black Holes, Clarendon Press, Oxford University
(1991).
[5] A. Dasgupta, Journal of Cosmology and Astroparticle Physics 0308 004 1 (2003).
[6] A. Dasgupta, Canadian Journal of Physics 96 (4) 366 (2018).
22

[7] A. Dasgupta, Classical and Quantum Gravity 23 635-671 (2006).


[8] D. Karp et al arXiv:math/0410009
[9] H. Van de Vel ‘On the series Expansion Method for computing Incomplete Elliptic Integrals of the
first and second kind’ https://www.ams.org/journals/mcom/1969-23-105/S0025-5718-1969-0239732-
8/S0025-5718-1969-0239732-8.pdf

You might also like