Sie sind auf Seite 1von 19

A Monte-Carlo Simulation of Radiative Processes

in Neutron Star Atmospheres


Nilankur Dutta
April 2015
We consider radiative transition of electron from landau states up to n = 500
to the ground state in high magnetic fields, under the framework of spinor
quantum electrodynamics. Attention is paid to the astrophysically relevant
case of fields around B 4 108 T.

Acknowledgements
I am infinitely grateful to Prof. D. Bhattacharya for providing me with an
opportunity to come and work at The Inter University Center for Astronomy
and Astrophysics, Pune. Without him, I would have missed out on working on
one of the most challenging projects of my career. He has been tolerant to my
many mistakes, and where I have succeeded credit lies entirely upon him.
I would also like to thank Prof A. Kshirsagar who acted as my internal guide
during this period. I found her advice most helpful.
But my most profound thanks goes to all the people who have worked on
this before me, whose names are mentioned in the References Section. Without
them there would be no wavefunctions, no approximations, and none of this at
all.

Contents
1 Introduction
2 Neutron Star Atmosphere
2.1 Landau Levels . . . . . . . .
2.1.1 Classical Mechanics .
2.1.2 Quantum Mechanics .
2.2 Cyclotron Emission . . . . . .
2.3 Cyclotron Absorption . . . .
2.4 Magneto-Compton Scattering

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

3 Assumptions

5
6
7
8
9
10
10
11

4 Details of the Code


4.1 Transition Rates . . . . . . . . .
4.2 Modelling . . . . . . . . . . . . .
4.3 RNG and Box-Mueller Transform
4.3.1 Box-Muller Transform . .
4.4 Laguerre Polynomials . . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

12
12
13
15
15
16

5 Results

16

6 Discussion

18

Introduction

Neutron stars are strongly magnetized objects, with surface field strengths in the
range 108 to 1015 G. The long-term evolution of these magnetic fields is a subject
of abiding interest, but with many open questions. One way to directly probe
the strength and the structure of the magnetic field of an accreting neutron
star is to observe the Cyclotron Resonance Scattering Features (CRSF), also
referred to as cyclotron lines, originating in the accretion region near the stellar
surface. The resonance energy of these features are determined by the local
field strength, and their profiles are shaped by the distribution of the field in
the emission region. Detailed observations and accurate modelling of cyclotron
lines can probe the field distortions caused by the accreting matter and yield
valuable information on whether or not screening can be responsible for the
secular reduction of the magnetic field of an accreting neutron star.
Cyclotron lines were first observed in the X-ray spectra of Her-X1 in 1978
and since then these are used as diagnostic tools to estimate the strength of the
magnetic fields of neutron stars and to probe the accretion physics in the polar
cap region.The cyclotron lines are generally found in hard X-rays (10-100 KeV).
In the spectra of nearly 20 X-ray pulsars cyclotron lines have been confirmed.
Building a theoretical model of cyclotron lines that can explain all observed
features is a formidable challenge. In this thesis we investigate how the cyclotron
line features may be affected by different emission region geometries and spatial
variations of density and magnetic field. We have developed a Monte-Carlo
radiative transfer code to accomplish this task.
The QED processes of Compton scattering and Cyclotron emission and absorption in strong magnetic fields, is commonly applied in atmospheric models of X-rays and Gamma rays. In space empty of external fields, energymomentum conservation requirements forbid first-order processes (Feynman diagrams with one vertex) in quantum electrodynamics. In space permeated by
macroscopically varying magnetic fields, however, first-order processes become
possible through a continuous exchange of kinetic momentum between electrons
and the external field. The first is just the well-known process of magnetic
bremsstrahlung or synchrotron radiation, which is observable at least in the
classical limit even for very weak fields (105 G in interstellar space, for example). This is in contrast to other one, field-induced conversion of a single photon
into an electron-positron pair. Intrinsically a quantum-mechanical effect, the
latter process requires extremely high field strengths and photon energies well
above the threshold value h
th = 2mc2 to be observable.
Previous works done on this subject include Daugherty and Ventura [4]
(1978), Arraya and Harding [1] (1998) and Sandeep Kumar, [7] (2012). But,
we have implemented the code independently, developing a simulation for these
three processes occurring in a slab geometry. We invoke Monte Carlo technique,
using the approximate relativistic cross sections developed by Gonthier, et al
[5] (2014) and transition rates by Harding and Preece [6] (1987).
The work in this thesis is that of predicting the nature of observable CRSF
given the physical properties of the emission region. We present here a versatile
4

Monte-Carlo radiative transfer code that we have developed for this purpose.
The final aim of this project is to investigate the spectrum produced by photons
passing through multiple slabs with varying magnetic fields.

Figure 1: The centroid energy of the phase averaged cyclotron resonance line
feature in Her X-1 since its discovery. Data from before 1997 were originally
compiled by Gruber et al. 2001, where the original references can be found.
The data after 1997 are from observations by RXTE, INTEGRAL (Klochkov
et al. 2006, Staubert et al. 2007, Klochkov et al. 2008) and Suzaku (Enoto et
al. 2008), plus recent values as listed in Table 2 of Staubert et al. (2014). Here
only values measured at 35d phases<0.20 are shown.

Neutron Star Atmosphere

The theoretical concept of neutron stars was proposed by L. D. Landau in the


1930s. The first signal of a neutron star was observed by Hewish et al. in 1968.
Since then neutron stars have emerged as very fascinating celestial objects.
Very high magnetic fields and strong gravity of the neutron stars provide a
unique laboratory to test fundamental laws of physics. All subclasses of neutron
stars (isolated neutron stars and neutron stars in binary system) have very
high surface magnetic fields between 108 to 1015 G. The stellar magnetic fields
estimated from cyclotron lines 4.5 1012 G for Her X-1, 3.4 1012 G for Cen
5

Figure 2: Her X-1 pulse phase averaged cyclotron line energies Ecyc normalized
to a reference ASM count rate of 6.8 cts/s using a flux dependence of 0.44
keV/ASM-cts/s. The data are consistent with a linear decline of Ecyc with time
with a slope of 7.22 104 keVd1 (the dashed line).
X-3. At fields of this magnitude and higher relativistic quantum effects begin
to contribute. Quantum effects become important when the Larmor radius
rL = mv/eB approaches the De Broglie wavelength D = 1/(mv) [7] ie. for
B >= m2 2 2 /e = 2 2 Bc

(1)

Bc = m2 /e = 44.1 1012

(2)

with
in G. where m is the mass and e is the charge of the equation. In our work
we use the critical magnetic field Bc as a reference with which to measure the
magnetic field in a particular slab i.e B 0 = B/Bc 104 to 102 for the cases
under consideration

2.1

Landau Levels

Landau quantization in quantum mechanics is the quantization of the cyclotron


orbits of charged particles in magnetic fields. As a result, the charged particles
can only occupy orbits with discrete energy values, called Landau levels. The
Landau levels are degenerate, with the number of electrons per level directly
proportional to the strength of the applied magnetic field. Landau quantization
6

is directly responsible for oscillations in electronic properties of materials as a


function of the applied magnetic field.
2.1.1

Classical Mechanics

We are interested in a charged particle (electron) in a uniform magnetic field.


To simplify the setup, let us consider the electron to move only on xy plane
under a magnetic field pointing along the z direction. We can always bring
back the motion along the z direction, which is just a translational motion with
a constant momentum or velocity. In classical mechanics, we care about the
equation of motion. The Lorentz force causes the electron to spiral around
(Larmor motion or cyclotron motion). The centrifugal force must balance the
Lorentz force,
mv 2 /r = eBv

(3)

rl = mv/eB

(4)

and hence
which is the Larmor radius, therefore the classical expressions for angular
frequency and energy are given as,
= qB/m
2

E = mr /2

(5)
(6)

~ we can
To use the canonical formalism we will define the vector potential A,
do this in translationally and rotationally invariant gauges. Even though the
system is both translationally and rotationally (around the z-axis) invariant, it
is curious that we cannot find a vector potential that is invariant under both of
them. One common choice preserves the rotational invariance (the symmetric
gauge),
~ = (Ax , Ay ) = B (y, x)
A
2
another one preserves the translational invariance along the y ( or equivalently the x) axis (Landau gauge):
~ = (Ax , Ay ) = B(0, x)
A
The Hamiltonian can be then written as,
~ 2 /2m = (~
~ 2 /2m
H=
p eA/c)

(7)

giving us the Hamilton equation of motion,


H
~
~x =
= (~
p eA/c)/m
~
p
H
~ i /m
p~ =
= (pi eAi /c)A
~x
7

(8)
(9)

After some algebra, we arrive at the Newtons Equation with Lorentz Force
= e (~v B)
~
~x
mc

(10)

x
= vy

(11)

y = vx

(12)

x = X + r cos t

(13)

y = Y r sin t

(14)

In components,

The general solution to this is

2 2

The energy of the particle is E = m v /2, independent of the magnetic field if


expressed in terms of the velocity. Therefore the lowest energy solution is the
particle at rest. The center (X,Y) of Cyclotron motion can be given as:

2.1.2

X = x + vy /

(15)

Y = y vx /

(16)

Quantum Mechanics

In a magnetic field B z, the non-interacting eigenstates of the Hamiltonian of


charged particles are no longer plane waves, but rather Landau level states.
From Hamiltonian to effective Hamiltonian
We have
H = P~ 2 /2m

(17)

~ q A/c
~
P~ = i

(18)

The canonical momentum is

~ = Bx
Assuming Landau gauge A
y , with charge q = e, the Schrodinger
equation becomes
1
eB 2
h2
[2x + ( y
x) ](r) = E(r)
(19)
8m
i
hc
By our choice of gauge, the Hamiltonian operator is independent of y, so we
can separate the wave function (x, y) = eiky (x). Furthermore, the coefficient
of x is just 1/lb2 , which manifestly has the right dimensions. Factoring out eiky ,
the result is a one-dimensional Schrodinger equation Hx (x) = E(x), with the
effective Hamiltonian
H(r) =

Hx =

2
h
[2x + (x xk )2 ]
2m

where
8

(20)

xk = lb2 k

(21)

Obviously, this describes a one-dimensional harmonic oscillator with its well


centered at xk and a spring constant Kef f = k 2 /mlb2 .
Note that [H, py ] = 0
Imagine a system of dimensions Lx Ly . With periodic boundary conditions
in they direction, the allowed k-vectors are
k = 2my /Ly

(22)

for any integer m. Thus, the allowed values of xk are separated by x =


lb2 k = 2lb2 /Ly . So, the total number of independent states is (LxLy)/2lb2
and each Landau level has
k=

1
2lb2

(23)

states per unit area

2.2

Cyclotron Emission

In the magnetic field, quantum relativistic effects start rising and the motion of
the electron in transverse direction to the magnetic field is badly affected because of the quantization of transverse energy. The motion of charged particles
parallel to the field is still defined by the parallel momentum, a continuous variable. A quantum electrodynamical treatment is thus necessary in this regime
of high magnetic fields even for non relativistic particles. The transition rates
are used in computing the transition probabilities from upper Landau level to
lower Landau level and differential transition rates are used to determine the
angle of emission of transition photons. The energy of the emitted photon and
the parallel momentum of the electron in the final state is determined by the
kinematics of the transition. A particular transition can be denoted as
[(ni , si , pi )]e [(nf , sf , pf )e , (f , f , pf )p ]

(24)

where subscript e, p represents the electron and the photon and the subscript
i, f represents the initial and final state during transition. The state of the
electron is described by Landau level, spin, parallel momentum (n, s, p) and the
state of the photon is described by energy, angle and polarization (, , pf )p .
Expressions of energy in initial and final states can be written as [7]
q
(25)
Ei = (p2i + m2 + 2ni m2 (B/Bc ))
Ef =

q
(p2f + m2 + 2nf m2 (B/Bc ))

(26)

The energy of the emitted photon and the parallel momentum of the electron
in the final state can be obtained by using energy conservation and parallel
momentum conservation.
9

Ef = Ei f

(27)

pf = pi f cos()

(28)

where f is the angle between magnetic field and photon emission direction.
Eliminating the pf and Ef from the above equations one can get the energy of
the emitted photon as
f = (Ei pi cos f ) [(pi cos f Ef )2 2m2 (B/Bc )(ni nf ) sin2 f ]/ sin2 f
(29)

2.3

Cyclotron Absorption

In cyclotron emission the electron is de-excited from a higher state to a lower


state and emits a photon. The cyclotron absorption process is the reverse of
it. In this, the electron is excited from a lower state (ni , si , pi ) to a higher
state (nf , sf , pf ) and the photon (with energy i , angle i , momentum ki and
polarization pi ) is absorbed. However a cyclotron absorption process is always
followed by cyclotron emission in the regime of low density, high field where
collisional de-excitations can be ignored. The absorption process can depicted
as
[(ni = 0, si , pi )]e , (i , i , pi )p ] [(nf , sf , pf )e ]

(30)

the energy and parallel momentum of the electron in the final state nf can be
obtained by using energy and parallel momentum conservation

2.4

Ef = Ei + f

(31)

pf = pi + f cos()

(32)

Magneto-Compton Scattering

In the magneto-Compton scattering an electron in initial state (ni , si , pi ) absorbs


a photon (with energy i , angle i , momentum ki and polarization p i) is excited
to an intermediate virtual state (nn , sn , pn ). It then de-excites to the lower state
(nf , sf , pf ) by emitting a photon (with energy f , angle f , momentum kf and
polarization pf ) . This process can be denoted as
[(ni = 0, si , pi )]e , (i , i , pi )p ] [(nf , sf , pf )e , (f , f , pf )p ]

(33)

the energy of the scattered photon f and the parallel momentum of the electron
pf in the final state can be obtained by using the energy and parallel momentum
conservation
Ef + f = Ei + i
(34)
10

pf + f cos f = pi + f cos i

(35)

after eliminating pf in the above equations the expression of energy of scattered


photon f can be represented in the form [7]

f =

C +

C 2 sin2 f [(E + i )2 m2 (pi + i cos i )2 2nf m2 B]


sin2 f
(36)

where
C = (Ei + i ) (pi + i cos i ) cos f

(37)

In this equation only the negative sign is physically meaningful.


The magneto-Compton scattering cross-section is a second order process
represented by two Feynman diagrams. In the first diagram the electron first
absorbs and then emits the photon while in the second the electron first emits
and then absorbs the photon.

Assumptions

We intend to perform the radiative transfer in a sea of electrons in cylindrical


or slab geometry with accreted matter in a static state. The photon-electron
interaction is considered only through the following processes: cyclotron absorption, cyclotron emission and magneto-Compton scattering. Photons from
a prescribed continuum are injected at the source plane and the modification
in the continuum is assumed to occur only via the above processes. We ignore
the free free process for the generation and absorption of continuum and also
collisional de-excitations. The electrons are assumed to be in their ground state
ni = 0 before any scattering or absorption. The Landau level population is determined by the excitation and de-excitation rates. The radiative de-excitation
rates as given in Latal (1986) [8] are:
Tradi = m(B/Bc ) 2 1027

(38)

in G s1 , with (.05, 2). The collisional de-excitation rates, on the other


hand, are Bonazzola et al. (1979) [2]
Tcol = 5 108 (ne /1021 cm3 ) 1018
1

12

(39)

in G s . For high fields B 10 G and low electron density ne 1021


cm the radiative de-excitation far exceeds the collisional de-excitation rates
(Tradi /Tcol 4 106 ) and hence the collisional de-excitation can be ignored.
The mean collision time being large compared to the radiative de-excitation
time, the electrons can be safely assumed to be in their ground state n = 0.
Since the magnetic field varies in the simulation region, the computation of
the free path is not very straight forward. For this reason we have divided the
simulation volume into cuboid cells within each of which the local density and
3

11

magnetic field are assumed constant at the value evaluated at the centre of the
cell.
The mound and the column are assumed to be in static state in which flow
of matter is not considered and matter is assumed to be in thermodynamic
equilibrium with one single temperature over the entire simulation volume.

Details of the Code

The momentum distribution of the electrons is described by the one dimensional


relativistic Maxwellian [7].
p
1 (pi /m)2 1
dpi
(40)
fe (pi )dpi = N
kTe
radiative transfer is performed in the following steps.
1. A photon is injected at a location (xinj , yinj , zinj ) at the source plane
with energy inj in the direction (inj , inj )
2. After injection, the propagation length of the photon before any scattering
or absorption is determined by the free path (B 0 , i , i ). The photon either
escapes, or is absorbed or scattered after propagating the distance . The
probability of scattering and absorption are computed and one from absorption
or scattering is selected.
3. For scattering or absorption the parallel momentum pi of the electron is
selected.
4. The electron is excited to a higher state nf either through scattering
or through absorption. If absorption occurs then the photon is removed, if
scattering occurs then the angle of scattering f is selected.
5. The electron de-excites via radiative transitions to a lower Landau levelnt
and generates a transition photon of energy t and emitted in the direction
(t , t ). The electron continues to de-excite and emit transition photons until
it reaches the ground state. All these transition photons are again injected into
the Monte-Carlo scheme. This process of photon propagation continues until
photons escape from the boundary of the simulation region.

4.1

Transition Rates

In a uniform, homogeneous magnetic field B 0 = B/Bc electrons must occupy


discrete energy states
p
En = p2 + m2 + 2nB 0 m2
(41)
characterized by principal quantum number n = l + (s + 1)/2 = 0, 1, 2, ...,
with (l = 0, 1, 2 ...) and momentum component parallel to the field, p (we
will use natural units, where h = c = 1 throughout). In each Landau state,
the electron may have spin-up (s = +1) or spin-down (s = 1) along the field
direction, except in the ground state (n = 0), where only the spin-down state

12

is allowed. The state of the electron is therefore completely described by the


three quantities (n, s, p). Transitions between the states (n, s, p) and (n0 , s0 , p0 ),
where n0 < n, result in the emission of photons at angle 0 to the magnetic eld
with energy co given from the kinematics. The differential transition rate from
state (n, s, p) to state (n0 , s0 , p0 ) is
Z
e2
s,s0
(par + perp )(E E 0 )d
(42)
Rn,n
0 () =
2
where par , perp are the rst-order matrix elements for photons polarized
parallel and perpendicular to the plane dened by k and B. We use the expressions for par , perp given by Sokolov and Ternov [9] for the case where the
electron spin states are described by their projections along the B field. The
total transition rate between two electron Landau states is obtained through
integration over all possible photon energies and angles. The function is used
to perform the integral over , where the dependence of E 0 on through the
nal momentum p0 in the argument of the -function must be taken into account.
The physical signicance of this step is that for given values of p0 and there
is a unique value of which conserves energy and momentum. The resulting
expression for the total transition rate is, [6]
0

s,s
Rn,n
0 () =

(E )(par + perp )
2 E p cos sin2

(43)

where now has the value given by equation (2) and is the fine-structure
constant. The integration over photon angle must be done numerically.

4.2

Modelling

we calculate the total rate of transitions from a state (n, s) to a state (n0 , s0 )
for the case where the electron has initial parallel momentum p = 0. These
are then stored in tables of synchrotron rates for each field strength. The rates
stored in the tables are computed in the electron circular motion frame where
p = 0; rates in any other frame where the electron has a nonzero parallel momentum may then be obtained from these by a Lorentz transformation along
the magnetic-eld direction. Each test electron in state (n, s, p) in the lab frame
is transformed to state (n, s, p = 0) in the circular frame which moves with
velocity
1/2
vpar = cos (E 2 m2 ) /E
(44)
with respect to the lab frame, where
sin = [2nB 0 m2 /(E 2 m2 )]1/2 .

(45)

Its final state (n0 , s0 , p0c ) in the circular motion frame is then determined by
choosing three random numbers. The first, r1 , is used to fix the final spin state.
If

13

0
Rns,s

n
X

s,s
Rn,

(46)

=1

is the total rate for transitions out of state (n, s), then
0

0
Pns,s =1

Rns,s =1
0

(Rns,s =1 + Rns,s = 1)

(47)

is the probability of a final spin s0 = +1, regardless of the final n0 value. The
final spin is then determined by
0

s0 = +1, r1 < Pns,s =1


0

s0 = 1, r1 > Pns,s =1

(48)
(49)

The second random number, r2 , fixes the final Landau state n0 by choosing
harmonic number :

P
s,s0
Rn,j
=1
r2 =
(50)
0
Rns,s
The third random number, r3 , determines the photon emission angle. Here
we make the approximation that photon emission angles are distributed uniformly between min , and min centered at = /2, where min , is estimated
from the behavior of the asymptotic expression for the transition rate.
cos min = tanh z,

(51)

z = (3m2 B 0 /E)1/3

(52)

Once the photon emission angle , is chosen, the photon energy , and
final electron momentum, p0c in the circular frame are determined from and the
transition is completely specified from equations (28) and (29). The photon
energy and angle as well as the electron final momentum are then transformed
back to the lab frame by the formulae:
cos =

cos c + vpar
,
1 + cos c vpar

(53)

= par (1 + cos c vpar )

(54)

p0 = par (p0c + vpar Ec0 )

(55)

14

4.3

RNG and Box-Mueller Transform

A random number generator (RNG) is a computational or physical device designed to generate a sequence of numbers or symbols that lack any pattern, i.e.
appear random. Random number generators have applications in gambling, statistical sampling, computer simulation, cryptography, completely randomized
design, and other areas where producing an unpredictable result is desirable.
Random number generators are very useful in developing Monte Carlo-method
simulations, as debugging is facilitated by the ability to run the same sequence
of random numbers again by starting from the same random seed. We discuss
the linear congruential generator - the one most commonly used for generating
random integers
Xn+1 = aXn + c( mod m)
(56)
where Xn = nth random integer
Xn+1 = n + 1th random integer
a = the integer multiplier
c = the integer additive constant
m = the modulus for modulo arithmetic
The next random integer Xn+1 is generated using the previous random integer Xn , the integer constants a and c, and the integer modulus m. After
the integer (aXn + c)is generated, modulo arithmetic using the modulus m is
performed, to yield the new random integer Xn+1 . The method requires an
initial seed, X0 , to get started.
4.3.1

Box-Muller Transform

Box-Muller transformation allows us to transform uniformly distributed random variables, to a new set of random variables with a Gaussian (or Normal)
distribution.
The most basic form of the transformation looks like:
p
(57)
y1 = (2 ln(x1 )) cos 2x2
p
y2 = (2 ln(x1 )) sin 2x2
(58)
We start with two independent random numbers, x1 and x2, which come
from a uniform distribution (in the range from 0 to 1). Then apply the above
transformations to get two new independent random numbers which have a
Gaussian distribution with zero mean and a standard deviation of one. The
polar form of the Box-Muller transformation is fast and numerically robust.
The algorithmic description of it is:

float x1, x2, w, y1,y2;


do
x1 = 2.0 * ranf() - 1.0;
x2 = 2.0 * ranf() - 1.0;
15

w = x1 * x1 + x2 * x2;
while ( w 1.0 );
w = sqrt( (-2.0 * log( w ) ) / w );
y1 = x1 * w;
y2 = x2 * w;

4.4

Laguerre Polynomials

Associated Laguerre polynomials are solutions of


xy 00 + ( + 1 x)y 0 + ny = 0

(59)

One can also define the generalized Laguerre polynomials recursively, defining the first two polynomials as
()
L0 (x) = 1
()
L1 (x) = 1 + x
and then using the following recurrence relation for any k 1:
()

()

Lk+1 (x) =

()

(2k + 1 + x)Lk (x) (k + )Lk1 (x)


.
k+1

(60)

Associated Laguerre polynomials crop up in the code in the formula for the
totally relativistic transition rates. A recursive subroutine is written to calculate
the same.

Results

The first challenge was to arrange the momenta of the electrons in a Maxwellian
distribution. We generate this from two uniform variates using the Box-Mueller
algorithm, rescaling then all momenta to achieve zero average momenta and
correct temperature.
We also successfully initialized photons of energy varying uniformly between two values min and max . We use the method of spectral weights where
the photons are distributed uniformly between energy bins. Later one associates
a weight to each of these bins to simulate non-uniform spectra. These photons
are correctly initialized to approach the slab from from the bottom at positions (xinj , yinj , 0) entering with angles distributed uniformly over the upper
hemisphere (inj (0, /2), inj (0, 2))
Then,we write a subroutine on the calculation of photon local mean free
path, travelled by a photon before encountering an absorption or scattering,
may then be from the total cross section, using the approximate forms of the
fully relativistic scattering and absorption cross-sections from Gonthier et al [5].
We also calculate the total rates of transition from equation (48), and the
kinematics for cyclotron emission and absorption described in sec [2.3],[2.4],
thus modelling the transition process. To do this we had to write a recursive
subroutine to generate associated Laguerre polynomials.
16

We also wrote subroutines for Lorentz transform for electrons from the circular frame to the lab frame (equations (53)-(55)) and vice versa (equations
(44) (45)).
Thus, so far rates of transitions from level nf = 500 and 100 have been
generated, but not scaled to natural units.

70
x=0
x=0
x=1
x=2
x=2
x=3
x=3

60
50

L_xk(y)

40

k=1
k=0
k=1
k=1
k=2
k=2
k=3

30
20
10
0
-10
-20
-2

10

12

y
(x)

Figure 3: Graph of the first few associated laguerre polynomials, Lk (y) vs y


generated by the code

17

50
45

Energy Levels

40
35
30
25
20
15
1

Figure 4: Cyclotron Emissions, followed by Cyclotron Absorptions shown in an


energy level scheme. In each scheme, an electron in energy level ni =20,40,50
emits a photon falling to an intermediate state. The same photon is then absorbed bringing the electron back to nf =20,40,50

Discussion

The work on this simulation is not complete yet. In the next month, the motivation would be to calculate the the cross-sections of scattering, and hence
fully simulate the photon spectrum. Furthermore, this has to be subjected to a
battery of tests, eg: escape probability for a photon, or Pencil Beam Injection
Test [7]. The ultimate test for any simulation is agreement with observational
data and so far we have not subjected this work to that test.
The effects of gravitational shifts and light bending should also be incorporated. So far, the work on this project has been restricted to a slab geometry,
however other configurations are possible eg: cylindrical. In the future, we plan
to extend our investigations to more complex emission geometries including
hollow and asymmetric columns, combination of accretion mound and overlying
column with magnetic distortions in the mound propagating into the column
and other such cases. Density distribution in the accretion column in different
accretion regimes will be taken into account.

18

References
[1] Araya, R.A., Harding A. K.,1999, THE ASTROPHYSICAL JOURNAL
517, 334
[2] Bonazzola S., Heyvaerts J., Puget J. L., 1979, ASTRONOMY ASTROPHYSICS, 78, 53
[3] Daugherty J. K., Harding A. K., 1986, THE ASTROPHYSICAL JOURNAL, 309, 362
[4] Daugherty J. K., Ventura J., 1978, PHYSICAL REVIEW D 90, 18, 1053
[5] Gonthier P., Baring M. et al, 2014, PHYSICAL REVIEW D 90, 043014
[6] Harding A. K., Preece R., 1987, THE ASTROPHYSICAL JOURNAL, 319,
939
[7] Kumar S., 2012, PhD THESIS, University of Pune, Accretion Induced
Evolution of Neutron Star Magnetic Fields
[8] Latal H. G., 1986, THE ASTROPHYSICAL JOURNAL, 309, 372
[9] Sokolov A. A., Ternov I. M., 1986, AMERICAN INSTITUTE OF
PHYSICS, Radiation from Relativistic Electrons

19

Das könnte Ihnen auch gefallen