Upload
trandang
View
249
Download
4
Embed Size (px)
Citation preview
1
Welcome to the
Lecture on stellar atmospheres
Stefan Dreizler
Universitäts-Sternwarte, University of Göttingen
Germany
Stellar Atmospheres
Outline:
• Introduction
• Radiation field
• Radiation transfer
• Emission and absorption
• Radiative equilibrium
• Hydrostatic equilibrium
• Stellar atmosphere models
1
Stellar Atmospheres: Motivation
1
Stellar Atmospheres: Literature
• Dimitri Mihalas– Stellar Atmospheres, W.H. Freeman, San Francisco
• Albrecht Unsöld– Physik der Sternatmosphären, Springer Verlag (in German)
• Rob Rutten– Lecture Notes Radiative Transfer in Stellar Atmospheres
http://www.fys.ruu.nl/~rutten/node20.html
Stellar Atmospheres: Motivation
2
Why physics of stellar atmospheres?
Physics
Stellar atmospheres as laboratories
Plasma-, atomic-, and molecularphysics, hydrodynamics, thermodynamics
Basic research
Technical application
Astronomy
Spectral analysis of stars
Structure and evolution of stars
Galaxy evolution
Evolution of the Universe
2
Stellar Atmospheres: Motivation
3
Magnetic fields in white dwarfs and neutron stars
Shift of spectral lines with increasing field strength
Stellar Atmospheres: Motivation
4
Hertzsprung Russell Diagram
L~R2 T4eff
100 R ¿¿¿¿
1 R ¿¿¿¿
=700000Km
0.01 R ¿¿¿¿
4 1026 W
5800K
3
Stellar Atmospheres: Motivation
5
Massive stars
Stellar Atmospheres: Motivation
6
Chemical evolution of the Galaxy
Carretta et al. 2002, AJ 124, 481
4
Stellar Atmospheres: Motivation
7
SN movie
Stellar Atmospheres: Motivation
8
SN Ia
5
Stellar Atmospheres: Motivation
9
SN Ia cosmology
ΩM, ΩΛ
0 , 1
0.5, 0.5
1 , 0
1.5, -.5
0 , 0
1 , 0
2 , 0
Redshift z
Stellar Atmospheres: Motivation
10
SN Ia Kosmologie
ΩΩ ΩΩΛΛ ΛΛ
6
Stellar Atmospheres: Motivation
11
Uranium-Thorium clock
Stellar Atmospheres: Motivation
12
Stellar atmosphere – definition
• From outside visible, observable layers of the star
• Layers from which radiation can escape into space– Dimension
• Not stellar interior (optically thick)
• No nebula, ISM, IGM, etc. (optically thin)
• But: chromospheres, coronae, stellar winds, accretiondisks and planetary atmospheres are closely related topics
7
Stellar Atmospheres: Motivation
13
Sonne
Stellar Atmospheres: Motivation
14
Fraunhofer lines
8
Stellar Atmospheres: Motivation
15
Spectrum - schematically
Inte
nsit
y
Wavelength / nm
Stellar Atmospheres: Motivation
16
Spectrum formation
9
Stellar Atmospheres: Motivation
17
Formation of absorption lines
Interior outerboundary
observer
continuum
line center
continuum
stellar atmosphere intensity
Stellar Atmospheres: Motivation
18
Line formation / stellar spectral types
spectral line temperaturestructure
interior
flux
wavelength
tempera
ture
depth / km
10
Stellar Atmospheres: Motivation
19
The spectral types on the main sequence
O B A F G K M
O5
B4
O7
B6
A1
A5
A8
A9
F8
G2
G5
G8
A7
F3
Stellar Atmospheres: Motivation
20
Die Spektraltypen der Hauptreihe
O B A F G K M F6
F8
G2
G5
G8
A7
F3
G6
G9
K4
K5
F8
G1
11
Stellar Atmospheres: Motivation
21
Classification scheme
M9
L3
L5
L8
Stellar Atmospheres: Motivation
22
Classification scheme
T dwarfs
12
Stellar Atmospheres: Motivation
23
Stellar atmosphere – definition
• From outside visible, observable layers of the star
• Layers from which radiation can escape into space– Dimension
• Not stellar interior (optically thick)
• No nebula, ISM, IGM, etc. (optically thin)
• But: chromospheres, coronae, stellar winds, accretiondisks and planetary atmospheres are closely related topics
Stellar Atmospheres: Motivation
24
Optical telescopes
Calar Alto (Spain)3.5m telescope
13
Stellar Atmospheres: Motivation
25
Optical telescopes
ESO/VLT
Stellar Atmospheres: Motivation
26
Why is it important?
UV / EUV observations
flux
flux
wavelength / Å
wavelength / Å
14
Stellar Atmospheres: Motivation
27
UV/opticaltelescopes
HST
Stellar Atmospheres: Motivation
28
X-ray telescopes
XMM
15
Stellar Atmospheres: Motivation
29
Gamma-ray telescopes
INTEGRAL
Stellar Atmospheres: Motivation
30
Infrared observatories
ISO
JWST
16
Stellar Atmospheres: Motivation
31
Sub-mm telescopes
Stellar Atmospheres: Motivation
32
Radio telescopes
100m dish at Effelsberg
17
Stellar Atmospheres: Motivation
33
Stellar atmosphere – definition
• From outside visible, observable layers of the star
• Layers from which radiation can escape into space– Dimension
• Not stellar interior (optically thick)
• No nebula, ISM, IGM, etc. (optically thin)
• But: chromospheres, coronae, stellar winds, accretiondisks and planetary atmospheres are closely related topics
Stellar Atmospheres: Motivation
34
PN – NGC6751 - HST
18
Stellar Atmospheres: Motivation
35
Planetary nebula spectrum
Stellar Atmospheres: Motivation
36
ISM spectrum
19
Stellar Atmospheres: Motivation
37
Quasar + IGM spectrum
Stellar Atmospheres: Motivation
38
Stellar atmosphere – definition
• From outside visible, observable layers of the star
• Layers from which radiation can escape into space– Dimension
• Not stellar interior (optically thick)
• No nebula, ISM, IGM, etc. (optically thin)
• But: chromospheres, coronae, stellar winds, accretiondisks and planetary atmospheres are closely related topics
20
Stellar Atmospheres: Motivation
39
Eta Carinae - HST
Stellar Atmospheres: Motivation
40
Stellar wind spectrum
21
Stellar Atmospheres: Motivation
41
Formation of wind spectrum (P Cygni line profiles)
Stellar Atmospheres: Motivation
42
Stellar winds – P Cyg profiles
22
Stellar Atmospheres: Motivation
43
Accretion disks
Stellar Atmospheres: Motivation
44
AM CVn disk spectrum
models
23
Stellar Atmospheres: Motivation
45
Temperature structure of an accretion disk
Distance from star [km]
Heig
ht
[km
]
Stellar Atmospheres: Motivation
46
Planetary atmospheres
24
Stellar Atmospheres: Motivation
47
Quantitative spectral analyses – what can we learn?
Shape of line profile:Temperature Film
Density Film
Abundance Film
Rotation
Turbulence
Magnetic field
Line position:Chemical composition
Velocities
Redshift
Temporal variation:Companion
Surface structure
Spots
Pulsation
Stellar Atmospheres: Motivation
48
Zeeman effect
25
Stellar Atmospheres: Motivation
49
Magnetic fieldsL
l
optical spectrum
circular polarization
position of line components
spectrum of a white dwarf (PG 1658+440) with fieldstrength of about 5 MG
Stellar Atmospheres: Motivation
50
Magnetic fieldsL
White dwarf Grw+70 8247B=300MG
optical spectrum
Circular polarization
positionof line components
26
Stellar Atmospheres: Motivation
51
Extrasolar planets
Stellar Atmospheres: Motivation
52
Velocity fields
Wavelength / Å
Dis
tance
/ 1
000 k
m
~ 0.01Å
Solar disk
time / min
dis
tance
/ 1
000km
27
Stellar Atmospheres: Motivation
53
Non-radial pulsation modes
Stellar Atmospheres: Motivation
54
Time resolved spectroscopy
28
Stellar Atmospheres: Motivation
55
Time dependent line profiles
Tim
e
Flu
x
Stellar Atmospheres: Motivation
56
Doppler tomography
29
Stellar Atmospheres: Motivation
57
Summary – stellar atmospheres theory
The atmosphere of a star contains less than one billionth of its total mass, so, why do we care at all?
• The atmosphere of a star is that what we can see, measure, and analyze.• The stellar atmosphere is therefore the source of information in order to put a
star from the color-magnitude diagram (e.g. B-V,mv) of the observer into the HRD (L,Teff) of the theoretician and, hence, to drive the theory of stellar evolution.
• Atmosphere analyses reveal element abundances and show us results of cosmo-chemistry, starting from the earliest moments of the formation of the Universe.
• Hence, working with stellar atmospheres enables a test for big-bang theory.• Stars are the building blocks of galaxies. Our understanding of the most
distant (hence most early emerged) galaxies, which cannot be resolved in single stars, is not possible without knowledge of processes in atmospheres of single stars.
• Work on stellar atmospheres is a big challenge. The atmosphere is that region, where the transition between the thermodynamic equilibrium of the stellar interior into the empty blackness of space occurs. It is a region of extreme non-equilibrium states.
Stellar Atmospheres: Motivation
58
Summary – stellar atmospheres theory
Important source of information for many disciplines in astrophysics– research for pure knowledge, contribution to our culture
– ambivalent applications (e.g. nuclear weapons)
Application of diverse disciplines– physics
– numerical methods
Still a very active field of research, many unsolved problems– e.g. dynamical processes
1
Stellar Atmospheres: The Radiation Field
1
The Radiation Field
Stellar Atmospheres: The Radiation Field
2
Description of the radiation field
Macroscopic description:
Specific intensity
as function of frequency, direction, location, and time; energy
of radiation field (no polarization)
•in frequency interval (ν,ν+dν)
•in time interval (t,t+dt)
•in solid angle dω around n
•through area element dσ at location r ⊥ n
( , , , )I n r tν νr r
σωννν
dddtd
EdtrnI
:),,,(
4
=rr
2
Stellar Atmospheres: The Radiation Field
3
The radiation field
dA dσσσσ
ϕϕϕϕ
ϑϑϑϑ
dωωωω
nr
Stellar Atmospheres: The Radiation Field
4
Relation
Energy in frequency interval (ν,ν+∆ν)
Energy in wavelength interval (λ,λ+∆λ)
i.e.
thus
with
ωλϑλ dddtdAIEd cos 4 =
νI→
λI→
|||| λν λν dIdI =
222 νλλν λνλ
νλ Ic
IIc
Ic
dȜdȞ
c ==−=⇒=
I I
energy energyDimension
area time freq. solid angle area time wavelength solid angle
Unit
ν λ
22
erg erg
cm s Hz sterad cm s A sterado
λν II ↔
3
Stellar Atmospheres: The Radiation Field
5
Invariance of specific intensity
2
2
2
Irradiated energy:
( , ) cos
as seen from dA subtends solid angle
cos /
cos cos , )
now, as seen from
cos cos , )
if no
=
′
′ ′=
′ ′→ =
′
′ ′′ ′=
Ȟ
Ȟ
dE I d dA d
dA d
d dA d
dA dAdE I ( dȞ
d
dA dA
dA dAdE I ( dȞ
d
ν ν ϑ ν ϑ ω
ω
ω ϑ
ϑ ϑν ϑ
ϑ ϑν ϑ
sources or sinks along d:
′ ′= ⇔ =dE dE I Iν ν
The specific intensity is distance independent if no
sources or sinks are present. G
Stellar Atmospheres: The Radiation Field
6
Irradiance of two area elements
dA dA´
dϑϑϑϑ ϑϑϑϑ´
4
Stellar Atmospheres: The Radiation Field
7
Specific Intensity
Specific intensity can only be measured from extended
objects, e.g. Sun, nebulae, planets
Detector measures energy per time and frequency interval
2
cos
e.g. is the detector area
~ (1 ) is the seeing disk
dE I d A
A
d
ν ϑ ω
ω
=
′′
Stellar Atmospheres: The Radiation Field
8
Special symmetries
• Time dependence unimportant for most problems
• In most cases the stellar atmosphere can be described in
plane-parallel geometry
Sun: atmosphere 200 km 1
1radius 700000 km 3500
: cos ( , , )ȞI I zνµ ϑ ν µ
= = <<
= =
z
ϑϑϑϑ
5
Stellar Atmospheres: The Radiation Field
9
• For extended objects, e.g. giant stars (expanding
atmospheres) spherical symmetry can be assumed
spherical coordinates: Cartesian coordinates:
( , , ) ( , , )I r I p zν νν µ ν
r=R outer boundary
p r
z ϑϑϑϑ
ϑϑϑϑ
Stellar Atmospheres: The Radiation Field
10
Integrals over angle, moments of intensity
• The 0-th moment, mean intensity
• In case of plane-parallel or spherical geometry
4
2 /2
0 / 2
2 1
0 1
1( ) with spherical coordinates
4
1sin with : cos
41
4
J I n d
I d d
I d d
ν ν
π
π π
νπ
π
ν
ωπ
ϑ ϑ φ µ ϑπ
µ φπ
−
−
=
= =
=
∫
∫ ∫
∫ ∫
r
1
1
2
1
2
energy erg
area time frequency cm s Hz
J I dν ν µ−
=
//
∫
6
Stellar Atmospheres: The Radiation Field
11
Jν is related to the energy density uν
4
radiated energy through area element during time :
hence, the energy contained in volume element per frequency interval is
4
Ȟ
dA dt
dE I d dt d dA
l c dt dV l dA c dt dA
dV
u dV d d t dAI d dνπν ω π
ν ω
ν ν
=
= ⇒ = =
= =∫4
3
430 0
energy erg
volume frequency cm Hz
total radiation energy in volume element:
energy erg
volume cm
/
c
c
dȞ
u J
u u d J d
dJ V c
πν ν
π
ν
ν νν ν∞ ∞
=
= =∫ ∫
l
dV
dA
dωωωω
Stellar Atmospheres: The Radiation Field
12
The 1st moment: radiation flux
4
,
, , ,
( )
propagation vector in
spherical coordinates:
sin cos
sin sin
cos
( , )sin cos sin
in plane-parallel or spherical geometry:
0, 2 ( )
x
x y z
F I n n d
n
F I d d
F F F F I d
ν νπ
ν
ν ν ν ν ν
ω
ϑ φ
ϑ φ
ϑ
ϑ φ ϑ φ ϑ ϑ φ
π µ µ
=
=
⇒ =
= = = =
∫
∫∫
r r r
r
1
1
2
energy erg
area time frequency cm s Hz
µ−
//
∫
ϕϕϕϕ
ϑϑϑϑ
x
y
z
7
Stellar Atmospheres: The Radiation Field
13
Meaning of flux:
Radiation flux = netto energy going through area ⊥ z-axis
Decomposition into two half-spaces:
Special case: isotropic radiation field:
Other definitions:
1
1
1 0
0 1
1 1
0 0
2 ( )
2 ( ) 2 ( )
2 ( ) 2 ( )
netto outwards - inwards
F I d
I d I d
I d I d
F F
π µ µ µ
π µ µ µ π µ µ µ
π µ µ µ π µ µ µ
−
−
+ −
=
= +
= − −
= −
=
∫∫ ∫∫ ∫
0=F
F astrophysical flux
Eddington flux
F 4
H
F H
ν
ν
ν ν νπ π= =
Stellar Atmospheres: The Radiation Field
14
Idea behind definition of Eddington flux
In 1-dimensional geometry the n-th moments of intensity are
1
1
1
1
12
1
1n
1
10-th moment: ( )
21
1st moment: ( ) 21
2nd moment: ( ) 21
n-th moment: ( ) 2
J I d
H I d
K I d
I d
ν
ν
ν
µ µ
µ µ µ
µ µ µ
µ µ µ
−
−
−
−
=
=
=
=
∫
∫
∫
∫
8
Stellar Atmospheres: The Radiation Field
15
Idea behind definition of astrophysical flux
Intensity averaged over stellar disk = astrophysical flux
2 2 2
2
2
sin
(1 )
2 2
p R
p R
dpp R
d
pdp R d
ϑ
µ
µµ
µ µ
=
= −
= −
= −
R
p=Rsinϑϑϑϑ
ϑϑϑϑnr
Stellar Atmospheres: The Radiation Field
16
Idea behind definition of astrophysical flux
Intensity averaged over stellar disk = astrophysical flux
2 0
1 2
2 0
1( )2
1
( )2
/ F
F 0 no inward flux at stellar surface
F
R
I I p pdpR
I R dR
F
I
ν ν
ν
ν ν
ν
ν ν
ππ
µ π µ µπ
π+ +
−
=
=
= =
=
=
∫
∫
dp
p
9
Stellar Atmospheres: The Radiation Field
17
Flux at location of observer
4
2 22 2 * *
2 2
( )
Flux at distant observer's detector normal to the line of sight:
/ Fv v
F I n n d
R Rf I d I R d F
d d
ν νπ
ν ν ν
ω
ω π π∗
=
= = = =
∫r r r
νI
R∗d
νf
dω
G
Stellar Atmospheres: The Radiation Field
18
Total energy radiated away by the star, luminosity
Integral over frequency at outer boundary:
Multiplied by stellar surface area yields the luminosity
0 0F F d F dν νν ν
∞ ∞+= =∫ ∫
2 2 2 2 2* * *4 4 F 16
energy erg
time s
L R F R R Hπ π π= = =
G
10
Stellar Atmospheres: The Radiation Field
19
The photon gas pressure
Photon momentum:
Force:
Pressure:
Isotropic radiation field:
cEp /νν =
ϑνν cos1
dt
dE
cdt
dpF == ⊥
νν
π
ν
ν
νν
πµµ
πωϑν
νωϑ
ϑ
Kc
dIc
dIc
P
ddIc
dAdt
dE
cdA
FdP
42 cos
1)(
cos1
cos1
1
1
2
4
2
2
===
=
==
∫∫−
I ( ) I J
I4 4 1P( ) u I P( ) u J 3K
c 3 c 3
ν ν ν
νν ν ν ν ν
µ = =
π πν = = ⇒ ν = =
G
Stellar Atmospheres: The Radiation Field
20
Special case: black body radiation (Hohlraumstrahlung)
Radiation field in
Thermodynamic
Equilibrium with matter
of temperature T13
2
1
3
12
5
5
( , , , ) ( )
( , ) bzw. ( , )
in cavity: 0
2( , ) exp 1
2( , ) exp 1
2( , ) exp 1
2( , )
−
−
−
=
= =
= = =
= −
= −
= −
=
r r
r
I n r t I
I B T I B T
F J I B
h hB T
c kT
hc hcB T
kT
hc hcB T
kT
hB T
c
ν ν
ν ν λ λ
ν ν ν
ν
ν
λ
λ
ν ν
ν ν
ν νν
λλ λ
λλ λ
νν
1
3exp 1
−
−
h
kT
ν
11
Stellar Atmospheres: The Radiation Field
21
Hohlraumstrahlung
Stellar Atmospheres: The Radiation Field
22
Asymptotic behaviour
• In the „red“ Rayleigh-
Jeans domain
• In the „blue“ Wien domain
4
2
2
2 ),(
2 ),(
1exp 1
λλ
νν
ννν
ν
ν
ckTTB
c
TkTB
kT
h
kT
h
kT
h
=
=
+≈
<<
−=
−=
≈−
>>
kT
hchcTB
kT
h
c
hTB
kT
h
kT
h
kT
h
λλλ
ννν
ννν
ν
ν
exp2
),(
exp2
),(
exp1exp 1
3
2
3
12
Stellar Atmospheres: The Radiation Field
23
Wien´s law
( )
( )
max max
max
13
2
x
x
x xmax
xmax
maxmax
2( , ) exp 1
3 1 e
e 1
0 3 e / e 1 0
3 1 e 0
numerical solution: 2.821
−
−
= −
− = + −
= → − − =
→ − − =
= =
d d h hB T
d d c kT
xB
dB x
d
x
hx
kT
ν
ν
ν
ν νν
ν ν
ν ν
ν
νλ
( )max
max
xmax
max max
max
0.5100cm deg
0 5 1 e 0
numerical solution: 4.965 0.2897 cm deg
−
=
= → − − =
= = =
T
dB x
d
hcx T
kT
λλ
λλ
x:=hv/kT
Stellar Atmospheres: The Radiation Field
24
Integration over frequencies
Stefan-Boltzmann law
Energy density of blackbody radiation:
13
20 0
4 3 4 44 4
2 3 2 30
4
5 44 5 -2 -1 -4
2 3
2( ) ( ) exp 1
2 2
1 15
15
2 with 5.669 10 erg cm s deg
15
x
h hB T B T d d
c kT
k x kT dx T
c h e c h
kT ı
c h
ν
ν νν ν
π
π
σ π
π
−∞ ∞
∞
−
= = −
= =−
=
= = = ⋅
∫ ∫
∫
4
0
4)(
4)(
4T
cTB
cdJ
cu
σπνν
πν === ∫
∞
13
Stellar Atmospheres: The Radiation Field
25
Stars as black bodies – effective temperature
Surface as „open“ cavity (... physically nonsense)
Attention: definition dependent on stellar radius!
( )
4
2 2 2 4* *
1/ 4 1/ 4 1/ 2eff *
, 0
for 0
0 for 0
with F and F ( ) T
luminosity: 4 F 4
hence, eff. temperature: 4
+ −
− − −
= =
>=
≤
= = =
= =
=
I B I
BI
B B T
L R R T
T L R
ν ν ν
νν
ν ν
µ
µ
σ
π
π σπ
σπ
F
G
Stellar Atmospheres: The Radiation Field
26
Stars as black bodies – effective temperature
14
Stellar Atmospheres: The Radiation Field
27
Examples and applications
• Solar constant, effective temperature of the Sun
• Sun‘s center
2 -1 -2
0
213 10
2*
10 -1 -2
4eff eff
( ) 1.36 kW/m 1.36 erg s cm
F with 1.5 10 cm 6.69 10 cm
F 2.01 10 erg s cm flux at solar surface
F 5780 K
f d f
df d R
R
T T
ν ν ν
π
π
σ
∞
= = =
= = ⋅ = ⋅
= ⋅
= ⇒ =
∫
7
max
max
1.4 10 K
Planck maximum at 3.4Å ( )
or 2.1Å ( )
with 1Å 12.4 keV maximum 4 keV
cT
B
B
ν
λ
λ
λ
= ⋅
⇒ =
=
≈
F
Stellar Atmospheres: The Radiation Field
28
Examples and applications
• Main sequence star, spectral type O
• Interstellar dust
• 3K background radiation
** eff
4 2*6eff* *
*
eff
max max
10 , 60000 K
10
882Å ( ) or 501Å ( )
R R T
TL RL L
L T R
B Bν λλ λ
= =
= ⇒ =
= =
)( mm3.0 ,K 20 max νλ BT ==
)( mm9.1 ,K 7.2 max νλ BT ==
15
Stellar Atmospheres: The Radiation Field
29
Radiation temperature
... is the temperature, at which the corresponding blackbody
would have equal intensity
Comfortable quantity with Kelvin as unit
Often used in radio astronomy
1
3rad
1
rad3
12
ln1exp2
)(
−−
+=⇒
−
=
νν
λλλλ
I
hc
k
hcT
ȜkT
hchcI
Stellar Atmospheres: The Radiation Field
30
The Radiation Field
- Summary -
16
Stellar Atmospheres: The Radiation Field
31
Summary: Definition of specific intensity
dA dσσσσ
ϕϕϕϕ
ϑϑϑϑ
dωωωω
nr
4d EI ( ,n, r, t) :
d dt d dν ν =ν ω σ
r r
Stellar Atmospheres: The Radiation Field
32
Summary: Moments of radiation field
In 1-dim geometry (plane-parallel or spherically symmetric):
1
1
1
1
12
1
10-th moment: ( )
21
1st moment: ( ) 21
2nd moment: ( ) 2
−
−
−
=
=
=
∫
∫
∫
J I d
H I d
K I d
ν
ν
ν
µ µ
µ µ µ
µ µ µ
Mean intensity
Eddington flux
K-integral
F =astrophysical flux =Eddington flux flux F 4 = = =H F F Hν ν ν ν ν νπ π
4
0 0energy density
∞ ∞
= =∫ ∫cu u d J dπ
ν νν ν
0 0total flux at stellar surface
∞ ∞+= =∫ ∫F F d F dν νν ν
2 2 2 2 2* * *stellar luminosity 4 4 F 16= = =L R F R R Hπ π π
17
Stellar Atmospheres: The Radiation Field
33
Summary: Moments of radiation field
4pressure of photon gas P( ) K
cν
πν =
13
2
2blackbody radiation ( , ) exp 1
−
= −
h hB T
c kTν
ν νν
44energy density of blackbody radiation u T
c
σ=
4
0
Stefan-Boltzmann law ( ) ( )∞
= =∫B T B T d Tν
σν
π
2 2 2 2 2 4* * * effeffective temperature 4 F 4 4= = =L R R B R Tπ π σπ
maxWien´s law constant=Tλ
1
Stellar Atmospheres: Radiation Transfer
1
Radiation Transfer
Stellar Atmospheres: Radiation Transfer
2
Interaction radiation – matter
Energy can be removed from, or delivered to, the radiation field
Classification by physical processes:
True absorption: photon is destroyed, energy is transferred
into kinetic energy of gas; photon is
thermalized
True emission: photon is generated, extracts kinetic energy
from the gas
Scattering: photon interacts with scatterer
→ direction changed, energy slightly
changed
→ no energy exchange with gas
2
Stellar Atmospheres: Radiation Transfer
3
Examples: true absorption and emission
• photoionization (bound-free) excess energy is transferred
into kinetic energy of the released electron → effect on
local temperature
• photoexcitation (bound-bound) followed by electron
collisional de-excitation; excitation energy is transferred to
the electron → effect on local temperature
• photoexcitation (bound-bound) followed by collisional
ionization
• reverse processes are examples for true emission
Stellar Atmospheres: Radiation Transfer
4
Examples: scattering processes
• 2-level atom absorbs photon
with frequency ν1, re-emits photon
with frequency ν2; frequencies not
exactly equal, because
– levels a and b have non-vanishing energy width
– Doppler effect because atom moves
• Scattering of photons by free electrons: Compton- or
Thomson scattering, (anelastic or elastic) collision of a
photon with a free electron
ν1 ν2
b
a
3
Stellar Atmospheres: Radiation Transfer
5
Fluorescence
Neither scattering nor true absorption process
c-b: collisional de-excitation
b-a: radiative
c
b
a
Stellar Atmospheres: Radiation Transfer
6
Change of intensity along path element
generally:
plane-parallel geometry:
spherical geometry:
ds
dIν
with dI dI
dt µdsds dt
ν νµ= − = −
r
rds
d
d
d
ds
d
ds
rd)d(
ds
drdsdr
/)1(
)sin(1
sin
sinsin
cos
2µ
ϑϑϑ
ϑ
µµ
ϑϑϑϑ
µϑ
−=
−−==⇒
−=≈+
=⇒=
r
I
r
I
ds
dI
ds
dI
ds
dr
r
I
ds
dI
21
µ
µµ
µ
µ
ννν
ννν
−
∂
∂+
∂
∂=⇒
∂
∂+
∂
∂=
4
Stellar Atmospheres: Radiation Transfer
7
Plane-parallel geometry
outer boundary
geom
etr
icaldepth
t
dt
ds
ϑ
withdI dI
dt µdsds dt
ν νµ= − = −
Stellar Atmospheres: Radiation Transfer
8
Spherical geometry
dr
ds
ϑϑϑϑ
ϑϑϑϑ+dϑϑϑϑ
-dϑϑϑϑ
r
I
r
I
ds
dI
ds
dI
ds
dr
r
I
ds
dI
21
µ
µµ
µ
µ
ννν
ννν
−
∂
∂+
∂
∂=⇒
∂
∂+
∂
∂=
r
rds
d
d
d
ds
d
ds
rd)d(
ds
drdsdr
/)1(
)1
sin(sin
sinsin
cos
2µ
ϑϑϑ
ϑ
µµ
ϑϑϑϑ
µϑ
−=
−−==⇒
−=≈+
=⇒=
-rdϑϑϑϑ
5
Stellar Atmospheres: Radiation Transfer
9
Change of intensity along path element
generally:
plane-parallel geometry:
spherical geometry:
ds
dIν
with dI dI
dt µdsds dt
ν νµ= − = −
r
rds
d
d
d
ds
d
ds
rd)d(
ds
drdsdr
/)1(
)sin(1
sin
sinsin
cos
2µ
ϑϑϑ
ϑ
µµ
ϑϑϑϑ
µϑ
−=
−−==⇒
−=≈+
=⇒=
r
I
r
I
ds
dI
ds
dI
ds
dr
r
I
ds
dI
21
µ
µµ
µ
µ
ννν
ννν
−
∂
∂+
∂
∂=⇒
∂
∂+
∂
∂=
Stellar Atmospheres: Radiation Transfer
10
Right-hand side of transfer equation
• No absorption (vacuum)
• Absorption only, no emission
.const0 =⇒= νν I
ds
dIinvariance of intensity
energy removed from ray:
is proportional to energy content in ray:
and to the path element:
ννν dIIdsI +
σωνν dddtddIdE −=
σωνν dddtdI
ds
F
6
Stellar Atmospheres: Radiation Transfer
11
Absorption coefficient
• thus:
κ absorption coefficient, opacity
• dimension: 1/length unit: cm-1
• but also often used: mass absorption coefficient, e.g., per
gram matter
κ in general complicated function of physical quantities T, P,
and frequency, direction, time...
• often there is a coordinate system in which κ isotropic, e.g.
co-moving frame in moving atmospheres
• counter-example: magnetic fields (Zeeman effect)
dsIdI νν κ−=
( )tnr ,,, νκκrr
=
( )νκκ ,rr
=
G
Stellar Atmospheres: Radiation Transfer
12
only absorption, plane-parallel geometry
outer boundary
geom
etr
icaldepth
t
dt ds
ϑ
0
/
optical
( , )( , ) ( , ) ( , ) ( , ) ( , )
with : ( , ) ( , ) with 0 at 0
( , ) 1( , )
( , ) integration constant, fixed by
depth
boun
=
= − ⇒ − = −
′ ′= → = = =
⇒ =
⇒ = ⋅
∫t
t
dI t dIt I t µ t t I t
ds dt
d dt t t dt t
dII
d
I c e c
ν νν ν
νν
τ µν
µκ ν µ µ κ ν µ
τ κ τ ν κ ν τ
µ τµ τ
τ µ
µ τ dary values
7
Stellar Atmospheres: Radiation Transfer
13
Schuster boundary-value problem
τ=0 outer boundary
µττνν
µτν
µτνν
µν
ττµτµ
ττµµ
τµτµ
τµµ
/)(
max
/
max
/
/0
max
max
),(),(
),(:0
)0,(),(
)0,(:0
−−++
+
−−−
−
==
⋅==>
==
=⋅==<
eII
ecI
eII
cecI
τ=τmax inner boundary
−νI
+νI
G
G
Stellar Atmospheres: Radiation Transfer
14
Example: homogeneous medium
e.g. glass filter
max
( 0) /
max
/
max
( , ) = thickness of filter
( , 0) ( , )
( , ) ( , 0)
d
d
t t d d
I I e
I I e
κ µν ν
κ µν ν
κ µ κ τ κ τ κ
µ τ µ τ τ
µ τ τ µ τ
+ + − −
−− −
= → = → =
= = = ⋅
= = = ⋅
s
I+
d/µs
I -
d/|µ|
8
Stellar Atmospheres: Radiation Transfer
15
Half-width thickness
2/1 : 2/1
2/1 =− sesκ
200000Solar atmosphere
6600Glas fiber
330City air
0.066Window glass
0.033River water
S1/2 / meterMaterial
Stellar Atmospheres: Radiation Transfer
16
Physical interpretation of optical depth
What is the mean penetration depth of photons into medium?
[ ]
(mathematically: expectation value of probability
/
function ( ) )
0
( ) / ( 0)
( ) : probability for absorption in interval ,
(
note nor
)
m
1alizati
p
I I
p d
p
d
e
d
d
ν ν
τ
τ τ
µ
ττ ττ τ
τ τ τ τ τ
τµ− −
−
∞
=
=
= +
= ⋅
∫
/
00 0
/
0 0
f
on: ( ) 1
1 e 1
mean penetration depth
1 1s mean free path
x
x
f
dp d e e
d xe dx
t
τ µ
τ µ
ττ τ
µ
τ τ τ µ µµ
τ µ
τ
µ µ κ κ
∞ ∞∞− −
∞ ∞− −
= = − =
= ⋅ = = ⋅
=
= = =
∫ ∫
∫ ∫
τ
0
1
9
Stellar Atmospheres: Radiation Transfer
17
The right-hand side of the transfer equation
• transfer equation including emission
emission coefficient ην
• dimension: intensity / length unit: erg cm-3 sterad -1
Energy added to the ray:
is proportional to path element:
ννν dIIdsI +
σωνν dddtddIdE +=
dsdI νν η=
ds
Stellar Atmospheres: Radiation Transfer
18
• Transfer equation including emission
η in general a complicated function of physical quantities
T,P,..., and frequency
η is not isotropic even in static atmospheres, but is usually
assumed to isotropic (complete redistribution)
if constant with time:
The right-hand side of the transfer equation
( )tnr ,,, νηηrr
=
( )νηη ,rr
=
10
Stellar Atmospheres: Radiation Transfer
19
The complete transfer equation
Definition of source function:
• Plane-parallel geometry
• Spherical geometry
ννν νκη I
ds
dI)(−=
( )ννν νκ IS
ds
dI−= )(
)(νκ
ηνν =S
( )),,(),,(),(),,(
tItStdt
tdIµνµννκ
µνµ νν
ν −=−
( )),,(),,(),(),,(1),,( 2
rIrSrrI
rr
rIµνµννκ
µ
µνµµνµ νν
νν −=∂
∂−+
∂
∂
Stellar Atmospheres: Radiation Transfer
20
Solution with given source function:
Formal solution
• Plane-parallel case
linear 1st-order differential equation of form
has the integrating factor
und thus the solution
(proof by insertion)
in our case:
( )ννν
µτSI
d
dI−=
1
)()( xgyxfy =+′
= ∫
x
x
dxxfxM
0
)(exp)(
0
x
0
x
1y(x) g(x)M(x)dx C C=y(x )
M(x)
= +
∫
)()(
)( /1)(
/1)(
νν
νν
ν
τ
τµ
µ
τ
Ixy
Sxg
xf
x
→
−→
−→
→
dI 1 1I S
d
νν ν+ =
τ −µ −µor:
11
Stellar Atmospheres: Radiation Transfer
21
Formal solution for I+
Reference point x0 : τ=τmax for I+ (µ> 0) outgoing radiation
−−+
′
−′−′=
+′
′−′
−−=
−=
′=
++
++
∫
∫
∫
µ
τττ
µ
τ
µ
ττττ
ττµ
τττ
µµ
τττ
µ
τττ
µτ
ν
τ
τ
νν
ν
τ
τ
νν
τ
τ
maxmax
maxmaxmax
max
exp)(exp)()(
)(exp)(1
-exp)(
exp1
-exp)(
max
max
max
Id
SI
IdSI
dM
weighted mean over
source function
exponentially absorbed ingoing radiation
from inner boundary
I+#
τ´τ τ+∆τ=τ+µ
pin point
e−∆τ/µHence, as rough approximation:
)()( µττ νν +≈+ SI
F
S
Stellar Atmospheres: Radiation Transfer
22
Formal solution for I -
Reference point x0 : τ=0 for I - (µ< 0) ingoing radiation
−+
′
′−−′=
+′
′′
−=
=
′=
−−
−−
∫
∫
∫
µ
τ
µ
τ
µ
ττττ
τµ
ττ
µµ
ττ
µ
ττ
µτ
ν
τ
νν
ν
τ
νν
τ
exp)0(exp)()(
)0(exp)(1
-exp)(
exp1
-exp)(
0
0
0
Id
SI
IdSI
dM
weighted mean over source
function
exponentially absorbed ingoing radiation
from outer boundary
F
12
Stellar Atmospheres: Radiation Transfer
23
Emergent intensity
Eddington-Barbier-Relation
max
max
maxmax
0
max
0
(0) ( ) exp ( ) exp
for semi-infinite atmospheres: :
(0) ( ) exp
hence, approximately: (0) ( )
dI S I
dI S
I S
τ
ν ν ν
τ
ν ν
ν ν
ττ ττ τ
µ µ µ
τ
τ ττ
µ µ
τ µ
+ +
+
+
′ ′ ′= − + −
→∞
′ ′ ′= −
≈ =
∫
∫
0 1
0 1 0 1
0 0
Relation is exactly valid if source function is linear in :
i.e. with ) and : / we have:
(0) ( )
∞ ∞+ − −
′= + ⋅ =
= + = + ⋅ =∫ ∫
Ȟ Ȟ Ȟ
x x
Ȟ Ȟ Ȟ Ȟ
S (IJ S S IJ x
I S e dx S xe dx S S Sν ν
τ
τ µ
µ µ µ
Stellar Atmospheres: Radiation Transfer
24
The source function
In thermodynamic equilibrium (TE): for any volume element it is:
absorbed energy = emitted energy
per second per second
The local thermodynamic equilibrium (LTE): we assume that
In stellar atmospheres TE is not fulfilled, because– System is open for radiation
– T(r) ≠const (temperature gradient)
νν
ν
νν
νν
κ
η
ηκ
νωσηνωσκ
BS
B
dddsddddsdI
==
=
=
))(,(),( rTBrSrr
νν νν =
Local temperature, unfortunately unknown at the outset
µ
τ
µ
ττ
τ
νν
′
′−′= ∫+ d
TBImax
0
exp))(()0( z.B.
Kirchhoff´s law
13
Stellar Atmospheres: Radiation Transfer
25
Source function with scatteringExample: thermal absorption + continuum scattering
(Thomson scattering of free electrons)
Inserting into formal solution:
( ) ( )
( ) ( )
0
( ) ( ) ( ) , ; , ,4
, ; , ,
(1 ) with /( )
dB R n n I n d
R n n
B J
B JS J B
I
ν ν ν
ν ν ν
ν νν ν ν
ωκ ν χ ν σ ν η χ σ ν ν ν ν
π
ν ν δ ν ν
η χ σ
χ σρ ρ ρ σ σ χ
χ σ
∞
′ ′ ′ ′ ′= + = +
′ ′ ′=
= +
+= = + − = +
+
∫ ∫r r r
r r
0 0
(0) (1 ) exp exp
( , )4
d dB J
dI
ν ν ν
ν
τ τ τ τρ ρ
µ µ µ µ
ωτ µ
π
∞ ∞+ ′ ′ ′ ′
= − − + −
∫ ∫
∫
true absorption
scattering
redistribution function
isotropic, coherent:
integral equation for Iν
Stellar Atmospheres: Radiation Transfer
26
The Schwarzschild-Milne equations
Expressions for moments of radiation field obtained by
integration of formal solution over angles µ
0-th moment1
1
1 0
0 1 0
1( ) ( , )
2
1( ) ( )exp ( ) exp
2
(written for semi-infinite atmosphere without irradiation from outside)
exchange integrals ( ,
−
∞
−
=
′ ′ ′ ′ − −′ ′= − + −
∫
∫ ∫ ∫ ∫
J I d
d dJ d S d S
S
ν ν
τ
ν ν ν
τ
τ τ µ µ
τ τ τ τ τ ττ µ τ µ τ
µ µ µ µ
( )( ) ( )( )
( )( ) ( )( )
2 2
1
2 2
0 1
1 1
1 1 independent of ) ,
1( ) ( ) exp ( ) exp
2
1( ) ( ) exp ( ) exp
2
∞ ∞
∞
∞ ∞ ∞
= = → =
′ ′ ′ ′ ′ ′= − − − + − −
′ ′ ′ ′ ′ ′= − − + − −
∫ ∫ ∫ ∫
∫ ∫
m mdw dw
w dd w
dw dwJ d S w w d S w w
w w
dw dwJ d S w d S w
w w
τ
ν ν ν
τ
ν ν ν
τ
τ µ µµ µ µ
τ τ τ τ τ τ τ τ τ
τ τ τ τ τ τ τ τ τ0
∫ ∫τ
14
Stellar Atmospheres: Radiation Transfer
27
The Schwarzschild-Milne equations
0-th moment
Karl Schwarzschild (1914)
( )( ) ( )( )
( ) ( )
1 0 1
1 1
0
-1
1
1
1( ) ( ) exp ( ) exp
2
1( ) ( ) ( )
2
with (x): t exponential integral of 1st order
( )
xt
dw dwJ d S w d S w
w w
J S E d S E d
E e dt
J
τ
ν ν ν
τ
τ
ν ν ν
τ
ν
τ τ τ τ τ τ τ τ τ
τ τ τ τ τ τ τ τ τ
τ
∞ ∞ ∞
∞
∞−
′ ′ ′ ′ ′ ′= − − + − −
′ ′ ′ ′ ′ ′= − + −
=
=
∫ ∫ ∫ ∫
∫ ∫
∫
( )1
0
1( )
2S E dν τ τ τ τ
∞
′ ′ ′−∫ G
Stellar Atmospheres: Radiation Transfer
28
The Lambda operator
Definition
In analogy, we obtain the Milne equations for the
1st moment
2nd moment
[ ] ( )
( )
1
0
1( ) ( )
2
( )
f t f t E t dt
J Sν ν
τ
τ
∞
Λ = −
⇒ = Λ
∫
( ) ( ) ( )2 2
0
1 1 1( ) ( ) ( )
2 2 4H S t E t dt S t E t dt S
τ
ν ν ν ν
τ
τ τ τ∞
= − − − = Φ∫ ∫
( ) ( )3
0
1
1 1( ) ( )
2 4
with ( )xt
n n
K S t E t dt S
eE x dt
t
ν ν ντ τ∞
∞ −
= − = Χ
=
∫
∫
GG
15
Stellar Atmospheres: Radiation Transfer
29
LTE
Strict LTE
Including scattering
Integral equation for
Solve
( ))()( ττ νν TBJ Λ=
( )( ))()1()(
)()1(
τρρτ
τρρ
ννν
ννν
TBJJ
TBJS
−Λ+Λ=
−+=
)(τνJ
)(41)(
)(41)(
)()()(
ττ
ττ
τττ
νν
νν
ννν
SK
SH
ISJ
Χ=→
Φ=→
→→
Stellar Atmospheres: Radiation Transfer
30
Excursion: exponential integral function
see Chandrasekhar: Radiative Transfer III.18
• For classical LTE atmosphere models, >50% of
computation time is needed to calculate En(x)
• In non-LTE models, En(x) is needed to calculate electron
collisional rates
• Recursion formula
1
( 1) ( 1)
1 1
( 1)( 1)
1
1
1
integration by parts ( )
1 1with product rule ( ) ( )
1 1
1 10
1 1
1( ) ( ) for 1
1
( ) ( )
n xt
n
n xt n xt
n
nx n xt
x
n n
n
E x t e dt
E x t e t x e dtn n
e x t e dtn n
E x e xE x nn
E x E x
∞− −
∞ ∞− − − − − −
∞− −− − − −
−−
=
= − − − −− −
= + −− −
= − > −
→
∫
∫
∫
G
16
Stellar Atmospheres: Radiation Transfer
31
Excursion: exponential integral function
• differentiation
( 1)
1 1 1
1
1 1
1
11 1
1
( ) ( ) ( )
( ) ( ) 1
1 ( ) ( ) ( )
( )
∞ ∞ ∞− − − − − − −
−
∞∞ ∞ −− − − − −
−
= = − = −
= − >
= = − = = −
= −
∫ ∫ ∫
∫ ∫
n xt n xt n xt
n
n n
xxt xt xt
x
d dE x t e dt t t e dt t e dt
dx dx
dE x E x n
dx
d d eE x t e dt t t e dt e
dx dx x x
d eE x
dx x
Stellar Atmospheres: Radiation Transfer
32
Excursion: exponential integral function
• integrals0
1 1
0 00
1 1
1 1
00
1 2
1
repeated integration by parts
1 1
d etc. until ( )
1 1 dx
1 ( 1)( 2) ( 1
s
l
n
ss sl ll
n n n
s sl l x
n n
l l l n
n n
x E (x)dx
x xx E (x)dx E (x) E (x)dx
l l
x x eE (x) E (x)dx E x
l l x
s s xE (s) E (s)
l l l l
+ +
+ + −
−
+ + +
−
′= −+ +
= + = −+ +
= + + ++ + + +
∫
∫ ∫
∫
L 1
1
0
1
0 0
( ))( 2) ( )
1
( 1)( 2) ( )
for
1 ( 1)! ( 1)! ! !
( 1)( 2) ( ) ( 1)( 2) ( ) ( )!
s
l n x
l l n x
n
E sl l n
x e dxl l l n
s
l n l n l lx E (x)dx x e dx
l l l n l l l n l n l n
+ − −
∞ ∞+ − −
+ +
++ + +
→∞
+ − + −= = = =
+ + + + + + + +
∫
∫ ∫
L
L
L L
17
Stellar Atmospheres: Radiation Transfer
33
Excursion: exponential integral function
• asymptotic behaviour
1
1
1 1
1 0 0
1
(1 ) (1 )
( ) ln (1
∞ ∞− − −
∞− − −
−
= +
= − − + + −
= − − + −
∫ ∫ ∫
∫ ∫ ∫ ∫
u u u
x x
x
u u u
x
du du due e e
u u u
du du du due e e
u u u u
E x x eγ0
1
1
1
0
1 2 3
)
0.5772156 Euler s constant
series expansion for the integral:
( ) ln ( 1)!
1 1 1Values at 0 : (0) 0 (0) 1 (0) 1, (0)
1 1 2
∞−
=
−−
=
= − − + −
= = − ⋅ = > = = − −
∫
∑
L
x
u
nn
n
n n
du
u
xE x x
n n
x E e E n E En n
γ
γ
1 2 2 3
1 1
1
1
1 1 1 2 6: ( ) 1
10 : ( )
∞ ∞− − −−
∞−
→∞ = = + = = − + − +
→ = =
∫ ∫
∫
L Lx xt x
xt
xt
e e ex E x e dt dt
t x x t x x x x
x E x e dtt
Stellar Atmospheres: Radiation Transfer
34
Example: linear source function
... one can show this
Conclusions:
The mean intensity approaches the local source function
The flux only depends on the gradient of the source function
( )
( ) ( )
[ ]
[ ]
1
0
1 1
0 0
3 2
3 4
S( ) a b
1J( ) S (a b )E d
2
1 1a E d b E d
2 2
1J( ) a b bE ( ) aE ( )
2
1 1H( ) b aE ( ) bE ( )
3 2
∞
∞ ∞
τ = + τ
′ ′ ′τ = Λ = + τ τ − τ τ
′ ′ ′ ′ ′= τ − τ τ + τ τ − τ τ
τ = + τ+ τ − τ
τ = + τ − τ
∫
∫ ∫
νν ττ SbaJxeE x
n =+→→≈>> − 0/:1
3/bH →ν
G….
18
Stellar Atmospheres: Radiation Transfer
35
Moments of transfer equation
• Plane-parallel geometry
• 0-th moment
• 1st moment
ννν
τµ SI
d
dI−=
∫−
1
12
1µdL
(I)
2
1)(
2
1)(
2
11
1
1
1
1
1
ννν
ννν
τ
µµµµµµτ
SJHd
d
dSdIdId
d
−=
−=
∫∫∫−−−
∫−
1
12
1µµdL
(II)
2
1)(
2
1)(
2
11
1
1
1
1
1
2
νν
ννν
τ
µµµµµµµµτ
HKd
d
dSdIdId
d
=
−=
∫∫∫−−−
G
Stellar Atmospheres: Radiation Transfer
36
( )
( ) ( ) (I) 1
20
2
2
11
2
1
2
1
2
11
2
1
2
1
2
2
1
1
1
1
2
1
1
1
1
1
1
21
1
ννν
νννν
ννννν
ννν
ν
κ
κ
κκµµµ
µκµκµµ
µµµ
JSHrrr
JSHr
Hr
JSdIr
Ir
Hr
dIdSdI
rdI
r
−=∂
∂
−=++∂
∂
−=−
−
−+
∂
∂
−=∂
∂−+
∂
∂
∫
∫∫∫∫
−−
−−−−
Moments of transfer equation
• Spherical geometry
• 0-th moment∫−
1
12
1µdL
( )),,(),,(),(),,(1),,( 2
rIrSrrI
rr
rIµνµννκ
µ
µνµµνµ νν
νν −=∂
∂−+
∂
∂
19
Stellar Atmospheres: Radiation Transfer
37
( )
(II) 3
31
0
031
2
1
2
1
2
1
2
1
2
1
2
1
1
1
21
1
3
1
1
1
1
1
1
31
1
2
ννν
ν
νννν
νννν
ννν
ν
κ
κ
κµµµµ
µµκµµκµµ
µµµµ
Hr
J
r
KK
r
HKJr
Kr
HdIr
Ir
Kr
dIdSdI
rdI
r
−=−+∂
∂
−=−−+∂
∂
−=−
−
−+
∂
∂
−=∂
∂−+
∂
∂
∫
∫∫∫∫
−−
−−−−
Moments of transfer equation
• 1st moment ∫−
1
12
1µµdL
Stellar Atmospheres: Radiation Transfer
38
Solution of moment equations
Problem: n-th momentum equation contains (n+1)-st moment
→always one more unknowns than differential equations
→ to close the system, another equation has to be found
Closure by introduction of variable Eddington factors
Eddington factor, is found by iteration
ννν JfK ⋅=
νf
new
starting estimate for ( ) ( ), solve
/
f I II K
f K J
ν ν
ν ν ν
→ + →
→ =
20
Stellar Atmospheres: Radiation Transfer
39
Solution of moment equations
2 differential eqs. for
Start: approximation for , assumption: anisotropy small, i.e.
substitute by (Eddington approximation)
ννν
ννν
τ
τ
Hd
JfdII
SJd
dHI
=
−=
)( )(
)(
νν HJ ,
νf
νI νJ
3
1
3
1
3
1
3
1
2
1
2
1
2
1)(
1
1
3
1
1
2
1
1
2
=→
=→
=
=≈=−−−
∫∫
ν
νν
ννννν µµµµµτ
f
JK
JJdJdIK
Stellar Atmospheres: Radiation Transfer
40
Eddington approximation
Is exact, if Iν linear in µ
(one can show by Taylor expansion of S in terms of B that this linear
relation is very good at large optical depths)
0 1
1
0
1
11 3
1 1
1 1
11 32
0 0
1 1
I ( ) I I
1J I ( )d I
2
1 1 1H I ( ) d I I
2 2 3 3
1 1 1K I ( ) d I I
2 2 3 3
1K J
3
ν ν ν
ν ν ν
−
ν ν ν ν
− −
ν ν ν ν
− −
ν ν
µ = +µ
= µ µ =
µ= µ µ µ = =
µ= µ µ µ = =
⇒ =
∫
∫
∫
G
21
Stellar Atmospheres: Radiation Transfer
41
Summary: Radiation Transfer
Stellar Atmospheres: Radiation Transfer
42
ννν νκη I
ds
dI)(−=Transfer equation
, ( )νη κ νEmission and absorption coefficients
Definitions: source function
optical depth
/ ( )Sν νη κ ν=
d dsτ κ= ⋅
Formal solution of transfer equation
max
maxmax( ) ( ) exp ( ) exp
dI S I
τ
ν ν ν
τ
τ ττ τ ττ τ τ
µ µ µ+ +′ ′ −−
′= − + −
∫
(0) ( )+ ≈ =I Sν ν τ µEddington-Barbier relation
LTE Local Thermodynamic Equilibrium
))(,(),( rTBrSrr
νν νν = ( )T rr
local temperature
Including scattering: (1 ) with /( )= + − = +S J Bν ν νρ ρ ρ σ σ χ
22
Stellar Atmospheres: Radiation Transfer
43
Schwarzschild-Milne equations
Moment equations of formal solution
( )( ) = ΛJ Sν ντ ( )1
( )4
= ΦH Sν ντ integral operatorsΛ, Φ
Moments of transfer equation (plane-parallel)( )
dH d K
J S Hd d
ν νν ν ν
τ τ= − =
Differential equation system (for J,H,K),
closed by variable Eddington factor : /=f K Jν ν ν
Stellar Atmospheres: Radiation Transfer
44
Summary: How to calculate I and the moments J,H,K
(with given source function S)?
( )1dI
I Sdτ µ= −Solve transfer equation
(no irradiation from outside, semi-infinite
atmosphere, drop frequency index)
Formal solution: ( ) ( )exp ( , analogous d
I S Iτ
τ τ ττ τ µ
µ µ
∞+ −′ ′ −
′= − > 0)
∫
How to calculate the higher moments? Two possibilities:
1. Insert formal solution into definitions of J,H,K:1
n
1
1
2 −∫ I dµ µ
ĺ ( )( ) = ΛJ Sτ ( )1
( )4
= ΦH Sτ ( )1
( )4
= ΧK Sτ Schwarzschild-Milne equations
2. Angular integration of transfer equation, i.e. 0-th & 1st moment1
n
1
1...
2 −∫ dµ µ
ĺ( )
dH d K
J S Hd dτ τ= − = 2 moment equations for 3 quantities J,H,K
Eliminate K by Eddington factor f: K f J= ⋅
ĺ ( )
dH d f JJ S H
d dτ τ
⋅= − = solve: J,H,K F new f (=K/J) iteration
1
Stellar Atmospheres: Emission and Absorption
1
Emission and Absorption
F
Stellar Atmospheres: Emission and Absorption
2
Chemical composition
Stellar atmosphere = mixture, composed of many chemicalelements, present as atoms, ions, or molecules
Abundances, e.g., given as mass fractions βk
• Solar abundances
M
M
001.0
009.0
001.0
004.0
28.0
71.0
=
=
=
=
=
=
Fe
O
N
C
He
H
β
β
β
β
β
β
Universal abundance for Population I stars
2
Stellar Atmospheres: Emission and Absorption
3
Chemical composition
• Population II stars
• Chemically peculiar stars,
e.g., helium stars
• Chemically peculiar stars,
e.g., PG1159 stars
0.1 0.00001
H H
He He
Z Z
β β
β β
β β
=
=
= L
¿
¿
¿
0.002
0.964
0.029
0.003
0.002
H H
He He
C C
N N
O O
β β
β β
β β
β β
β β
≤ <<
= >>
= >>
= ≈
= <
¿
¿
¿
¿
¿
0.05
0.25
0.55
0.02
0.15
H H
He He
C C
N
O O
β β
β β
β β
β
β β
≤ <<
= >>
= >>
<
= >>
¿
¿
¿
¿
Stellar Atmospheres: Emission and Absorption
4
Other definitions
• Particle number density Nk = number of atoms/ions of element k per unit volume
relation to mass density:
with Ak = mean mass of element k in atomic mass units (AMU)
mH = mass of hydrogen atom
• Particle number fraction
• logarithmic
• Number of atoms per 106 Si atoms (meteorites)
kHkk NmA=ρβ
∑ ′k
k
N
N
00.12)/log( += Hkk NNε
3
Stellar Atmospheres: Emission and Absorption
5
The model atom
The population numbers (=occupation numbers)
ni = number density of atoms/ions of an element, which are in the level i
Ei = energy levels, quantized
E1 = E(ground state) = 0
Eion = ionization energy
bound states, „levels“
free states
ionization limit
1
65432
Energ
y
Eion
Stellar Atmospheres: Emission and Absorption
6
Transitions in atoms/ions
1. bound-bound transitions = lines
2. bound-free transitions = ionization and
recombination processes
3. free-free transitions = Bremsstrahlung
We look for a relation between macroscopic quantities and microscopic (quantum mechanical) quantities, whichdescribe the state transitions within an atom
Energ
ie
Eion
Photon absorption cross-sections
1 2
3
)(),( νηνκ ν
4
Stellar Atmospheres: Emission and Absorption
7
Line transitions:Bound-free transitions: thermal average of electron velocities v(Maxwell distribution, i.e., electrons in thermodynamic equilibrium)
Free-free transition: free electron in Coulomb field of an ion, Bremsstrahlung, classically: jump into other hyperbolic orbit, arbitrary
For all processes holds: can only be supplied or removed by:– Inelastic collisions with other particles (mostly electrons), collisional
processes
– By absorption/emission of a photon, radiative processes
– In addition: scattering processes = (in)elastic collisions of photons with electrons or atoms- scattering off free electrons: Thomson or Compton scattering- scattering off bound electrons: Rayleigh scattering
Photon absorption cross-sections
ffE∆E∆
( )2e
bf th ion low
unbound state ion free electron 1/ 2 m v
E E E E
= +
∆ > = −
( )lowupbb EEE −±=∆
+
Stellar Atmospheres: Emission and Absorption
8
The line absorption cross-section
Classical description (H.A. Lorentz)
Harmonic oscillator in electromagnetic field
• Damped oscillations (1-dim), eigen-frequency ω0
Damping constant γ
• Periodic excitation with frequency ω by E-field
Equation of motion:
inertia + damping + restoring force = excitation
Usual Ansatz for solution:
tieeExmxmxm
ωωγ 020 =++ &&&
tiextx ω0)( =
( ) tiem
eExi ωωωγω 02
02 =++−
5
Stellar Atmospheres: Emission and Absorption
9
The line absorption cross-section
( ) ( )
22
3
2 22 20 0
2 2 2 2 2 2 2 2 2 20 0
22 2 4 2 2 52 2 60 02 2 20
2 2 2
Electrodynamics:
2( )
3
( ) cos ( )sin( ) ( )
2cos co
radiated p
s sin s
ower
in
ep(t) x
c
eEx(t) t t
m
eE(x(t)) t t t t
m N N N
ω ω γωω ω ω ω
ω ω ω γ ω ω ω γ
ω ω ω γ ω ω ω γ ωω ω ω ω
=
−= − + −
− + − + − − = + +
&&
&&
&&
( )
( )( )
2 2 00
0
2 20
2 200
2 2 2 2 20
2 20 0
2 2 2 2 2 2 2 2 2 20 0
1
expand ( )
real part Re cos sin( ) ( )
− + + =
= ⋅− +
− −= ⋅
− +
−= +
− + − +
i t
t
ti
i
eEi x(t) e
m
eEx(t) e
m i
eEx(t) e
m
eE(x(t)) t t
i
m
ω
ω
ω
ω ωγ ω
ω ω ωγ
ω ω ωγ
ω ω ω γ
ω ω γωω ω
ω ω ω γ ω ω ω γ
Stellar Atmospheres: Emission and Absorption
10
The line absorption coss-section
( )
( )( )
( )
2 2
22 2 2 2 2202 0
222 2 2 20
2 42 0
22 2 2
2
0
3
2
average over one period
power, averaged ove
cos sin 1/ 2,
r one
cos sin 0
1(
2
1
perio
(2
3
d
2(
= = =
− +
= − +
=
=
− +
&
&
&
&&
&e
p x
t t t t
eEx)
E)
m
c
m
ex
ω ω ω ω
ω ω γ ωω
ω ω γ ω
ω
ω ω γ ω
( )
( )
4 20
2
4 22
04
22 2 2 2
3
4
22 2 2
3
0
0
2
2
C=normalization constant ( )3
( ) profi
)3
le functi
( ) /
/
n( 2
o )
= = /2
=−
− +
+
=
C
e
e Ep
m c
C
E
m c
ω
ω ω γ
ν ω π
νϕ ν
ν ν γ π ν
ω
ϕ ν
G
6
Stellar Atmospheres: Emission and Absorption
11
The line absorption cross-section
0
0
0 0 0
2 2 2 2 2 20 0 0 0 0
2 20 0
2 2 2 20 0
0
since - , :
( ) (( )( )) 4 ( )
( )4( ) ( / 2 ) 4 ( ) ( / 4 )
now: calculating the normalization constant
( ) 1
4substitution: : ( )
(
+∞
−∞
∆ = << ≈
− = + − ≈ −
= =− + − +
=
= −
∫
C C
d
x
ν
ν
ν ν ν ν ν ν ν
ν ν ν ν ν ν ν ν ν
ν νϕ ν
ν ν γ π ν ν γ π
ϕ ν ν
πν ν
γ
ϕ0
0
2 22 00 2 2 2
0
4)
4 1
=
+∞ +∞
−∞ −∞
= = ⇒ =+∫ ∫
C dxd C C
x
ν
ν
ν ππ γν ν ν
γ γ ν π
π
Stellar Atmospheres: Emission and Absorption
12
Profile function, Lorentz profile
properties:
• Symmetry:
• Asymptotically:
• FWHM:
The line absorption cross-section
220
2
)4/()(
4/)(
πγνν
πγνϕ
+−=
γ/4Max =
Max2
1
0ν ν
)(νϕ
FWHM
))(())(( 00 ννϕννϕ −−=−+
220 1)(1)( ννννϕ ∆=−→
π
γ
π
γν
πγν
πγ
γ 24
2
)4/()2(
4/2FWHM22
FWHM
2
==∆⇒+∆
=
7
Stellar Atmospheres: Emission and Absorption
13
The damping constant
• Radiation damping, classically (other damping mechanisms later)
• Damping force (“friction“)
power=force ⋅velocity
electrodynamics
• Hence, Ansatz for frictional force is not correct
• Help: define γ such, that the power is correct, when time-averaged over one period:
classical radiation damping constant
)(txmF &γ=
( )2)()( txmtp &γ=
( )2
3
2
)(3
2)( tx
c
etp &&=
22 4
03
2 (where we used ( ) )
3= = i te
m x t x ec
ωγ ω ω
3
20
2
0 3
2
mc
eȦȦ ωγ =⇒≈
G
Stellar Atmospheres: Emission and Absorption
14
Half-width
Insert into expression for FWHM:2 2
0FWHM 3
24FWHM FWHM
FWHM FWHM2 2
4
2 3
41.18 10 Å
3
e
mc
c e
mc
π νγν
π
ν λ πλ ν
ν λ ν−
∆ = =
∆ ∆= ⇒ ∆ = ∆ = = ⋅
8
Stellar Atmospheres: Emission and Absorption
15
The absorption cross-section
Definition absorption coefficient κ
with nlow = number density of absorbers:
absorption cross-section (definition), dimension: area
Separating off frequency dependence:
Dimension : area ⋅ frequency
Now: calculate absorption cross-section of classical harmonicoscillator for plane electromagnetic wave:
dsIdI νν νκ )(−=
low)()( nνσνκ =)(νσ
)()( 0 νϕσνσ =
0σ
)1()(8
),( 20
0
−′−=′
=
µδννδπ
µνν
ω
Ec
I
eEEti
x
Stellar Atmospheres: Emission and Absorption
16
Power, averaged over one period, extracted from the radiation field:
On the other hand:
Equating:
Classically: independent of particular transition
Quantum mechanically: correction factor, oscillator strength
4 2 2 2 2 20 0 0
class.2 3 3
4 2 2 2 3 2 20 0 0
2 3 2 2 20
2( ) with
3 3
3( ) ( )
3 2 4 8
= = =
= =
e E ep
m c mc
e E mc e Ep
m c e m
π ν ωϕ ν γ γ
γ
π νϕ ν ϕ ν
π ν
20
2 22 00
22
0
( ) ( , ) ( )8
( ) ( )8 8
( ) ( ) 0.026537 cm Hz
cp I d d E
e EcE
m
e
mc
ν µ
σ ν ν µ ν µ σ νπ
σ ν ϕ νπ
πσ ν ϕ ν σ
′
′ ′= =
=
= ⇒ =
∫ ∫
)()( lu
2
lowlu
2
lu νϕπ
νκπ
σ fmc
enf
mc
e==
F
index “lu” stands for transition lowerĺupper level
9
Stellar Atmospheres: Emission and Absorption
17
Oscillator strengths
Oscillator strengths flu are obtained by:
• Laboratory measurements
• Solar spectrum
• Quantum mechanical computations (Opacity Project etc.)
• Allowed lines: flu≈1,
• Forbidden: <<1 e.g. He I 1s2 1S→1s2s 3S flu=2 10-14
3280.12H β4861.3
gupglowfluLineλ/Å
820.41Ly α1215.7
5080.04H γ4340.5
1880.64H α6562.8
3220.03Ly γ972.5
1820.07Ly β1025.7
Stellar Atmospheres: Emission and Absorption
18
Opacity status report
Connecting the (macroscopic) opacity with (microscopic) atomic physics
View atoms as harmonic oscillator– Eigenfrequency: transition energy
– Profile function: reaction of an oscillator to extrenal driving (EM wave)
– Classical crossection: radiated power = damping
,
2
low low,up( ) ( )
low up
en f
mcσ
πκ ν ϕ ν=
1442443
Population number of lower level
QM correction factorProfile function
Classical crossection
10
Stellar Atmospheres: Emission and Absorption
19
Extension to emission coefficient
Alternative formulation by defining Einstein coefficients:
Similar definition for emission processes:
profile function, complete redistribution:
induced 0up ul
spontaneous 0up ul
( )4
( )4
hn B I
hn A
ν ν
ν
νη ψ ν
πν
η ψ νπ
=
=
0low lu
20
lu lu
hv( ) n B ( )
4
hv e i.e. B f
4 mc
κ ν = ϕ νπ
π=
π
)(νψ )()( νψνϕ =
G
Stellar Atmospheres: Emission and Absorption
20
Relations between Einstein coefficients
Derivation in TE; since they are atomic constants, theserelations are valid independent of thermodynamic state
In TE, each process is in equilibrium with its inverse, i.e., within one line there is no netto destruction or creation of photons (detailed balance)
( )
( )
0 0 0ul up ul up lu low
ul ul up lu low
low lu up ul up ul
1
ul low lu
ul up ul
TE: ( )4 4 4
( ) ( )
(
emitted intensity absorbed inte
)
nsity
( ) 1
h h hB I n A n B I n I B T
B B T A n B B T n
B T n B n B n A
A n BB T
B n B
ν ν ν ν
ν ν
ν
ν
ν ν ν
π π π
−
+ = =
+ =
− =
= −
=
11
Stellar Atmospheres: Emission and Absorption
21
Relations between Einstein coefficients
( )
0
0
0
1
up upul low lu
ul up ul low low
1
ul low lu
ul up ul
31
02
( ) 1 with Boltzmann equation:
( ) 1 comparison with Planck blackbody radiation:
2( ) 1
h kT
h kT
h kT
n gA n BB T e
B n B n g
A g BB T e
B g B
hB T e
c
νν
νν
νν
ν
−
−
−
−
= − =
= −
= −
⇒3
ul 02
ul
low lulow lu up ul
up ul
2
1
A h
B c
g Bg B g B
g B
ν=
⇒ = ⇒ =
G
Stellar Atmospheres: Emission and Absorption
22
Relation to oscillator strength
dimension
Interpretation of as lifetime of the excited state
order of magnitude:
at 5000 Å:
lifetime:
2 2
lu lu
0
2 2up up
ul lu lu
low low 0
3 2 2 2up up0 0
ul ul lu ul lu2 3low low
4
4
2 83
eB f
mchv
g g eB B f
g g mchv
g ghv e vA B f f
c g mc g
π
π
πγ
=
= =
= = = ulA 1 time
ul1 A
ulul γ≈A18 s10 −
s10 8−
F
12
Stellar Atmospheres: Emission and Absorption
23
Comparison induced/spontaneous emission
When is spontaneous or induced emission stronger?
At wavelengths shorter than λ∗ spontaneous emission isdominant
( )**
**
spontaneous 3 2ul * up ul *
induced * * 2 3ul * up ul *
**
*
*
with
( ) 4 211
( ) ( ) 4 ( ) 2
: 1 2 ln 2
e.g. 10000K : 20000 A
50000K : 4160 A
v v
h kTv
v v
h kT
*
*
I B
A h n A h ce
B T B h n B B T c h
e h kT
T
T
ν
ν
ν
ν ψ ν πη ν
η ν ψ ν π ν
ν
λ
λ
=
= = = −
= ⇒ = ⇒ =
= =
= =
o
o
Stellar Atmospheres: Emission and Absorption
24
Induced emission as negative absorption
Radiation transfer equation:
Useful definition: κ corrected for induced emission:
spontaneous induced
spontaneous induced
induced0 0lu lu low lu ul up
with
( ), ( )4 4
= − = +
= + −
= = v
dII
ds
dII
ds
h hB n B n I
νν ν ν ν ν
νν ν ν
η κ η η η
η η κ
ν νκ ϕ ν η ϕ ν
π π
( )spontaneous 0ul up lu low
2low
lu lu low up
up
3 2spontaneous 0 lowlu lu up2
up
( ) 4
( )
2 ( )
dI hB n B n I
ds
gef n n
mc g
h gef n
c mc g
νν ν
νη ϕ ν
π
πκ ϕ ν
ν πη ϕ ν
= + −
= −
=
transition lowĺup
So we get (formulated withoscillator strength insteadof Einstein coefficients):
13
Stellar Atmospheres: Emission and Absorption
25
The line source function
General source function:
Special case: emission and absorption by one line transition:
• Not dependent on frequency
• Only a function of population numbers
• In LTE:
κηvvS =
( )
1
up
low
low
up
2
30lu
uplowlow
up
up
2
30
0upullowlu
0upul
lu
lulu
12
-
2
)(4
)(4
−
−=
=
−
==
n
n
g
g
c
hvS
nng
g
n
c
hv
vhv
nBnB
vhv
nA
S
v
vv
ϕπ
ϕπ
κ
η
[ ] ),(12
0
1
2
30lu 0 TvBe
c
hvS v
kThv
v =−=−
Stellar Atmospheres: Emission and Absorption
26
Every energy level has a finite lifetime τ against radiative decay (except ground level)
Heisenberg uncertainty principle:
Energy level not infinitely sharp
q.m. ⇒ profile function = Lorentz profile
Simple case: resonance lines (transitions to ground state)example Lyα (transition 2→1):
example Hα (3→2):
Line broadening: Radiation damping
∑<
=ul
ul1 Aτ
h=⋅∆ τE
∑∑<<
+=+=l
l
l
AAj
juk
uku
11
ττγ
21 cl 1 2 12A 3 g g fγ = = γ cl cl3 2 8 0.41 0.31= γ ⋅ = γ
1 2 1cl 12 23 13 cl cl
2 3 3
g g g 2 8 23 f f f 3 0.41 0.64 0.07 1.18
g g g 8 18 18
γ = γ + + = γ + + = γ
14
Stellar Atmospheres: Emission and Absorption
27
Line broadening: Pressure broadening
Reason: collision of radiating atom with other particles
⇒Phase changes, disturbed oscillation
t0 = time between two collisions
0( ) ~ i tE t e ω
Stellar Atmospheres: Emission and Absorption
28
Line broadening: Pressure broadening
Reason: collision of radiating atom with other particles
⇒Phase changes, disturbed oscillation
Intensity spectrum (=power spectrum) of the cut wave train:
t0 = time between two collisions
0
0
0
2
1
2
02/ 2
10/ 2
~ Fourier transform
sin2
( ) ~
2−
− =
−
∫t
i t i t
t
I
t
I e e dtω ω
ω ω
ωω ω
0( ) ~ i tE t e ω
15
Stellar Atmospheres: Emission and Absorption
29
Line broadening: Pressure broadening
Probability distribution for t0
Averaging over all t0 gives
Performing integration and normalization gives profile function of intensity spectrum:
i.e. profile function for collisional broadening is a Lorentz profile with
∫∞
−
−
−
⋅=0
0/
2
00 /22
sinconst)( 0 τωωωω
ω τdtetI
t
v
( )0 /0 0 0( ) average time between two collisionst
W t dt e dtτ τ τ−= =
( ) ( )220 1
1)(
τωω
πτωϕ
+−=
12 , ~ = particle density of colliders
approximately constant
-N N
N Ȗ Ȗγ τ τ
γ
=
′ ′= ⋅
(to calculate γ´: calculation of τ necessary; for that: assumption about phase shift needed, e.g., given by semi-classical theory)
Stellar Atmospheres: Emission and Absorption
30
Line broadening: Pressure broadening
• Semi-classical theory (Weisskopf, Lindholm), „Impact Theory“
Phase shifts ∆ω:
find constants Cp by laboratory measurements, or calculate
• Good results for p=2 (H, He II): „Unified Theory“– H Vidal, Cooper, Smith 1973
– He II Schöning, Butler 1989
• For p=4 (He I)– Barnard, Cooper, Shamey; Barnard, Cooper, Smith; Beauchamp et al.
ppAnsatz: C r , p 2,3, 4,6 , r(t) distance to colliding particle∆ω = = =
hydrogen-like ions
neutral atoms with each other, H+H
ions
metals + H
linear Stark effect
resonance broadening
quadratic Stark effect
van der Waals broadening
2
3
4
6
dominant atnamep=
Film logg
16
Stellar Atmospheres: Emission and Absorption
31
Thermal broadening
Thermal motion of atoms (Doppler effect)
Velocities distributed according to Maxwell, i.e.
for one spatial direction x (line-of-sight)
Thermal (most probable) velocity vth:
kTm
xxxew2
A v21~)v( −
( )
2 2th
2 2 2th
1/ 24th A
th
v vx x
0
v vth th
0 0 th
x
v 2 12.85 10 A km/s
example: T 6000K, A 56 (iron): v 1.33 km/s
i.e. (v ) , with (v ) v 1 we obtain:
1C v C v x 1 v 1
v
(v )
x
x
x x x
x
x
x
kT m T
w C e w d
e d e d C C
w
ππ
∞−
∞ ∞− −
= =
= = =
= ⋅ =
⋅ = ⋅ = ⇒ = ⇒ =
=
∫
∫ ∫2 2
thv v
th
1
vxe
π−
Stellar Atmospheres: Emission and Absorption
32
Line profile
Doppler effect:profile function:
Line profile = Gauss curve– Symmetric about ν0
– Maximum:– Half width:– Temperature dependency:
cc
th
0
th
0
v ,
v=
∆=
∆
ν
ν
ν
ν
02 2
th
0
2 20 th
01
th
( )
th
(v ) ( ) , with 1 we obtain:
1( )
Ȟ
x x
Ȟ
Cw e ( )dȞ
cȞ
eȞ
ν ν
ν ν ν
νϕ ν ϕ ν
π
ϕ νπ
+∞
−∆ ∆
−∞
− − ∆
⇒ = =∆
=∆
∫πth
1Max
v∆=
Max2
1
0ν ν
)(νϕ
FWHM
πth1 v∆
ththFWHM 67.12ln2 vvv ∆=∆=∆
Tv ~th∆
17
Stellar Atmospheres: Emission and Absorption
33
Examples
At λ0=5000Å:
T=6000K, A=56 (Fe): ∆ λth=0.02Å
T=50000K, A=1 (H): ∆ λth=0.5Å
Compare with radiation damping: ∆ λFWHM=1.18 10-4Å
But: decline of Gauss profile in wings is much steeper thanfor Lorentz profile:
In the line wings the Lorentz profile is dominant
210 43th
2 6rad
Gauss (10 ) : e 10
Lorentz (1000 ) : 1 1000 10
− −
−
∆λ ≈
≈
∆λ ≈
Stellar Atmospheres: Emission and Absorption
34
Line broadening: Microturbulence
Reason: chaotic motion (turbulent flows) with length scales smaller than photon mean free path
Phenomenological description:
Velocity distribution:
i.e., in analogy to thermal broadening
vmicro is a free parameter, to be determined empirically
Solar photosphere: vmicro =1.3 km/s
2micro
2 vv
microx v
1)v( xew x
−=π
18
Stellar Atmospheres: Emission and Absorption
35
Joint effect of different broadening mechanisms
Mathematically: convolution
commutative:
multiplication of areas:
Fourier transformation:
y y xx
xprofile A + profile B = joint effect
dxxfdxxfdxxff BABA ∫∫∫∞
∞−
∞
∞−
∞
∞−
⋅=∗ )()())((
ABBA ffff ∗=∗
∫∞
∞−
−=∗ dyyxfyfxff BABA )()())((
BABA ffff ⋅∗ =~~
2~
πi.e.: in Fourier space the convolution is a multiplication
Stellar Atmospheres: Emission and Absorption
36
Application to profile functions
Convolution of two Gauss profiles (thermal broadening + microturbulence)
Result: Gauss profile with quadratic summation of half-widths; proof by Fourier transformation, multiplication, and back-transformation
Convolution of two Lorentz profiles (radiation + collisional damping)
Result: Lorentz profile with sum of half-widths; proof as above
2 2 2 2
2 2 2 2 2C
( ) 1 ( ) 1
G ( ) ( ) ( ) 1 with C
x A x B
A B
x C
A B
G x A e G x B e
x G x G x C e A B
π π
π
− −
−
= =
= ∗ = = +
2 2 2 2
2 2
/ /( ) ( )
/( ) ( ) ( ) with
A B
C A B
A BL x L x
x A x B
CL x L x L x C A B
x C
π π
π
= =+ +
= ∗ = = ++
19
Stellar Atmospheres: Emission and Absorption
37
Application to profile functions
Convolving Gauss and Lorentz profile (thermal broadening + damping)
( )
2 20
2
2( )
220
0 0
2
1 / 4( )
( ) / 4
depends on , , : ( ´) ´ ´
Transformation: v: : 4 : ´
/1( )
− − ∆
∞
−∞
−
= =∆ − +
= ∗ , ∆ = −
= − = = −
∆= =
+∆
∫
D
D
D
D D D
y D
D
G e L( )
V G L ǻ V( ) G L(v )d
( ) ǻ a /( ʌǻ ) y ( ) ǻ
aG y e L(y)
y
ν ν ν γ πν ν
ν π ν ν γ π
ν ν γ ν ν ν ν ν
ν ν ν γ ν ν ν ν
ν π
ν π
2
2
2 2 2
2 2
Voigt fu
1
(v )
1Def: ( , v) with ( , v)
(v )
, no analytical representation possible.
(approximate formulae or numerical evaluation)
Norm
nc
a
o
l
ti n
∞ −
−∞
∞ −
−∞
=− +∆
= =− +∆
∫
∫
y
D
y
D
a eV dy
a y a
a eV H a H a dy
y a
πν π
πν π
ization: ( , v) v∞
−∞
=∫ H a d π
Stellar Atmospheres: Emission and Absorption
38
Voigt profile, line wings
20
Stellar Atmospheres: Emission and Absorption
39
Treatment of very large number of lines
Example: bound-bound opacity for 50Å interval in the UV:
Direct computation would require very much frequency points
• Opacity Sampling
• Opacity Distribution Functions ODF (Kurucz 1979)
MöllerDiploma thesisKiel University 1990
Stellar Atmospheres: Emission and Absorption
40
Bound-free absorption and emission
Einstein-Milne relations, Milne 1924: Generalization of Einstein relations to continuum processes: photoionizationand recombination
Recombination spontaneous + induced
Transition probabilities:
I) number of photoionizations
II) number of recombinations
Photon energy
In TE, detailed balancing: I) = II)
[ ][ ]
: probability for photoionization in
(v) : spontaneous recapture probability of electron in v, v v
(v) : corresponding induced probability v=electron velocity
P d
F d
G
ν ν ν ν, +
+
[ ] dtdIGFnn v vv)v()v()v(eup +
dtdIPn vv νlow
v vv21 2ion dhmdvmEhv =→+=
G
21
Stellar Atmospheres: Emission and Absorption
41
Einstein-Milne relations
[ ][ ]
low up e
low up e
13
1low
2up e
3
2
low
up e
low up
(v) (v) (v) with
(v) (v) (v)
(v) 21 1
(v) (v) (v)
(v) 2
(v)
(v) (v)
f
v
v
h kTv
h kTv
n P I dvdt n n F G I h m dvdt I B
n P B n n F G B h m
n P mF hB e
G n n hG c
F h
G c
n P me
n n hG
n n
ν ν ν ν
ν ν
νν
ν
ν
ν
−
−
= + =
= +
= − = −
⇒ =
⇒ =
• ion
2
3/ 2uplow
2up e low
3/ 2
v 2 2e e e
2 2rom Saha equation:
(v) : Maxwell distribution: (v) v 4 v v2
E kT
m kT
gn mkTe
n n h g
mn n d n e d
kT
π
ππ
−
−
=
• =
G
Stellar Atmospheres: Emission and Absorption
42
Einstein-Milne relations
Einstein-Milne relations, continuum analogs to Aji, Bji, Bij
2ion
3/ 2up 3/ 2
up
low
3
2
2low
2up 2
3lo
/ 2up
2e low
e
3/ 2
v 2 2e
w
(v)
2 4 v
8
(v)
4 v2
v(v
2 2
)
hv kT
m kTE kT
v
hv kT
v
n
mn e
kT
P he
G m
he
m
gh mm
n
n
gmkT
m h g
gP m
G h g
en h g
π
ππ
π
π
− −
=
=
=
=
F
22
Stellar Atmospheres: Emission and Absorption
43
Absorption and emission coefficients
absorption coefficient (opacity)
emission coefficient (emissivity)
And again: induced emission as negative absorption
and (using Einstein-Milne relations)
LTE:
vv nhvPnv σκ lowlow)( ==
[ ] mvhIGFnnv vv /)v()v()v()( 2eup +=η
2low up e
*
up up /low e low up
low low
( ) (v) (v) /
(v)1 (v) ... h kT
n P h n n G h m
n nG hn P h n n n e
n P m n
ν
νν ν
ν
κ ν ν ν
ν σ −
= −
= − = = −
[ ]
vv
v
kThv
v
Bvv
mhFnnv
ehvPnv
)()(
/v)v()v()(
1)(2
eup
low
κη
η
κ
=
=
−= −
M
*
up /up3
low
2( ) ...
ch kT
nhn e
n
νν ν
νη ν σ −
= =
definition. of cross-section σ
Stellar Atmospheres: Emission and Absorption
44
Continuum absorption cross-sections
H-like ions: semi-classical Kramers formula
Quantum mechanical calculations yield correction factors
Adding up of bound-free absorptions from all atomic levels: example hydrogen
( )3
th th th
318 2
th th 2 22 2 2
for ( )
0 else
8 threshold frequency, 7.906 10 cm
3 3
principal quantum number, Z nuclear charge
h n n
Z Zm ce
n
σ ν ν ν νσ ν
ν σπ
−
>=
= = = ⋅
( )3
th th bf bf( ) ( , ) , ( , ) Gaunt factorg n g nσ ν σ ν ν ν ν=
∑=
=max
1bf
totbf )()(
n
n
n
nnvv σκ
23
Stellar Atmospheres: Emission and Absorption
45
Continuum absorption cross-sections
Optical continuum dominatedby Paschen continuum
Stellar Atmospheres: Emission and Absorption
46
The solar continuum spectrum and the H-ion
H- ion has one bound state, ionization energy 0.75 eVAbsorption edge near 17000Å,hence, can potentially contribute to opacity in optical band
H almost exclusively neutral, but in the optical Paschen-continuum, i.e. population of H(n=3) decisive:
Bound-free cross-sections for H- and H0 are of similar orderH- bound-free opacity therefore dominates the visual continuum
spectrum of the Sun
0 0
4 7.5eSun: 6000K, log 13.6 Saha equation: 10 , 10H H
H H
n nT n
n n
+ −− −= = = =
500106
103
)3(
)1(
)1()3(
1062
18
)1(
)3(
10
8
104.23/eV1.12
1
3
0
0
00
0
0
=⋅
⋅=
=
=
==
=
⋅====
=
−
−
−−−
−−
nn
nn
nn
n
nn
n
eeg
g
nn
nn
H
H
H
H
H
H
kT
H
H
24
Stellar Atmospheres: Emission and Absorption
47
The solar continuum spectrum and the H-ion
Ionized metals deliver free electrons to build H
-
Stellar Atmospheres: Emission and Absorption
48
The solar continuum spectrum and the H-ion
G
25
Stellar Atmospheres: Emission and Absorption
49
The solar continuum spectrum and the H-ion
Stellar Atmospheres: Emission and Absorption
50
Scattering processes
Thomson scattering at free electrons
Absorption coefficient follows from power of harmonic oscillator ( Thomson cross-section)
Thomson cross-section is wavelength-independent
eeσκ n=
eσ
( ) ( )
4 2 40
22 3 22 2 20
0
4 220
e 02 3
425 2
e 2 4
3 2
free electrons: no resonance frequency, no friction: 0; 0
, on the other hand we had 3 8
86.65 10 cm
3
e Ep
m c
e E cp p E
m c
e
m c
ν
ν ν γ π ν
ν γ
σπ
πσ −
= − + = =
→ = =
= = ⋅
26
Stellar Atmospheres: Emission and Absorption
51
Scattering processes
Rayleigh scattering of photons on electrons bound in atomsor molecules
Rayleigh scattering on Lyα important for stellar spectral typesG and K
( ) ( )
4 2 40
22 3 22 2 20
0
4 2 420
R 02 3 4
4 4 4
R lu e lu2 4 4 4
4
R e lu 4
3 2
semi-classical:
on the other hand we had 3 8
8
3
( )
lu
lu
lu lu
l
l lu
e Ep
m c
e E cp p E
m c
ef f
m c
n f
ν
ν ν γ π ν
ν ν ν
νσ
ν π
π ν νσ σ
ν ν
νκ ν σ
ν
= − + << =
→ = =
= =
→ =∑
(here we have included the oscillator strengthas the quantum mechanical correction)
Stellar Atmospheres: Emission and Absorption
52
Raman scattering
Discovered in symbiotic nova RR Tel
Raman scattering of O VI resonance line (Schmid 1987)
n=1
n=2
n=3virtuallevel
1215Å1026Å 1032/38Å
Raman-scattered line 6825/7082Å
ĮLy OVI
111
λλλ−=
Schmid 1989,
Espey et al. 1995
27
Stellar Atmospheres: Emission and Absorption
53
Two-photon processes
Stellar Atmospheres: Emission and Absorption
54
Free-free absorption and emission
Assumption (also valid in non-LTE case):
Electron velocity distribution in TE, i.e. Maxwell distribution
Free-free processes always in TE
Similar to bound-free process we get:
generalized Kramers formula, with Gauntfaktor from q.m.
• Free-free opacity important at higher energies, becauseless and less bound-free processes present
• Free-free opacity important at high temperatures
),()(/)()( ffffff TvBvvvS vvv == κη
( )ff h / kTff e k
2 2 6
ff ff3/ 2 3
( ) ( )n n 1 e
16 Z e( ) g (n, ,T)
hc(2 T
1
m)3 3
1
− νκ ν = σ ν −
π
νσ ν = ⋅ ⋅ ν
π
1/ 2 3/ 2ffff bf bf~ T , but ~ T (Saha), therefore: / T − −σ σ κ κ
28
Stellar Atmospheres: Emission and Absorption
55
Computation of population numbers
General case, non-LTE:
In LTE, just
In LTE completely given by:
• Boltzmann equation (excitation within an ion)
• Saha equation (ionization)
( , , )i i v
n n T Iρ=
( , )i i
n n Tρ=
Stellar Atmospheres: Emission and Absorption
56
Boltzmann equationDerivation in textbooks
Other formulations:
• Related to ground state (E1=0)
• Related to total number density N of respective ion
( ) / statistical weight
excitation energy i jE E kT ii i
ij j
gn ge
En g
− −=
kTEii ieg
g
n
n /
11
−=
1 1
1 1 1
1
1
1
/
/partition function
(
1
, with ( ) :)
−
−
= = =
= =→
∑
∑
∑ ∑ ∑ j
jE kT
j
i i i i
jj j
i i
E kT
j
n n n nn g
nn n n n n
n
g e
U T
n n gU g
nT
Ne
G
29
Stellar Atmospheres: Emission and Absorption
57
Divergence of partition function
e.g. hydrogen:
Normalization can be reached only if number of levels is finite.
Very highly excited levels cannot exist because of interaction with neighbouring particles, radius H atom:
At density 1015 atoms/cm3 → mean distance about 10-5 cm
r(nmax) = 10-5 cm → nmax ~43
Levels are “dissolved“; description by concept of occupation probabilities pi (Mihalas, Hummer, Däppen 1991)
i
2i i i Ion
E / kTi
g 2n g , E E
i.e. g e
i i
i
lim lim
lim−
= → = ∞ =
= ∞
→ ∞ → ∞
→ ∞
Nni
i =∑
20)( nanr =
with w0 hen → ∞→ →i i ii
g pg p i
Stellar Atmospheres: Emission and Absorption
58
Hummer-Mihalas occupation probabilities
30
Stellar Atmospheres: Emission and Absorption
59
Saha equation
Derivation with Boltzmann formula, but upper state is now a 2-particle state (ion plus free electron)
Energy:
Statistical weight:
Insert into Boltzmann formula
Statistical weight of free electron =number of available states in interval[p,p+dp] (Pauli principle):
2ion eE E p 2m p=electron momentum)= +
)(up pGgg ⋅=
2ion e low
2up low e
up up ( 2 ) /
low low
( ) /up up 2
low low 0
( ) ( )
( )
E p m E kT
E E kT p m kT
n p g G pe
n g
n ge G p e dp
n g
− + −
∞− − −
=
→ = ∫
3
2 2 2 3e e
phase space volume( )( ) 2 2 spins
phase space cell
( ) 4 1 4 ( ) 8x y z
d pG p dp
h
d p dxdydz dp dp dp dV p dp n p dp G p p h nπ π π
Ω=
Ω = ⋅ = ⋅ = ⋅ → =
F
weight of ion * weight of free electron
Summarize over all final statesBy integration over p
Stellar Atmospheres: Emission and Absorption
60
Saha equation
Insertion into Boltzmann formula gives:
Saha equation for two levels in adjacent ionization stages
Alternative:
( )
( )
2up low e
2up low
up low
up
( ) /up up 22e3
low low e0
3/ 2( ) /up 2e3
low e 0
3/ 2( ) /upe3
low e
3/ 2(up upe
3low e low
8 with / 2
82
82
4
22
E E kT p mkT
E E kT x
E E kT
E
n ge p e dp x p m kT
n g h n
ge m kT x e dx
g h n
ge m kT
g h n
n gm kTe
n n h g
π
π
π π
π
∞− − −
∞− − −
− −
−
= =
=
=
=
∫
∫
low ) /E kT−
33/216/)(
low
up2/3
low
eup cm K 1007.2 )( lowup −−−⋅=== Ce
g
g
C
TTf
n
nn kTEE
31
Stellar Atmospheres: Emission and Absorption
61
Example: hydrogen
Model atom with only one bound state:
5
low I I
up II II
3/ 21.5810 K/II
I
II I
II
22
(H I ground state) 2
(H II ) 1
1( )
2
pure hydrogen: ,
ionization degree:
(( )
1
Te
e II
e
n n n g
n n n g
n n Te f T
n C
n n N n n
n n x
N N
x N f Tf T x
x
− ⋅
= = =
= = =
= =
= = +
= =
⇒ = ⇒ +−
2
) ( )0
( ) ( ) ( )
2 2
( , )
f Tx
N N
f T f T f Tx
N N N
x x T N
− =
⇒ = − + +
⇒ =
G
Stellar Atmospheres: Emission and Absorption
62
Hydrogen ionization
Ioniz
atio
n d
egre
e x
Temperature / 1000 K
32
Stellar Atmospheres: Emission and Absorption
63
More complex model atoms
j=1,...,J ionization stages
i=1,...,I(j) levels per ionization stage j
Saha equation for ground states of ionization stages j and j+1:
With Boltzmann formula we get occupation number of i-th level:
kTEeg
g
kTm
hnnn /
11j
1j
2/3
e
3
e1j1j1
jIon
22
1
+
+
=
π
kTEE
kTEkTE
eTCnng
gn
eg
gTCnne
g
gn
n
nn
/)(2/31e1j1
11j
ijij
/
11j
1j2/31e1j1
/
1j
ij1j
1j
ijij
ji
jIon
jIon
ji
−−+
+
+
−+
−
=⇒
==
Stellar Atmospheres: Emission and Absorption
64
More complex model atoms
Related to total number of particles in ionization stage j+1
Nj/Nj+1
kTEEkTEEeTCnN
U
gneTCnN
U
g
g
gn
NU
gn
U
g
N
n
U
g
n
n
N
n
/)(2/31e1j
1j
ijij
/)(2/31e1j
1j
11j
11j
ijij
1j
1j
11j11j
1j
11j
1j
11j
1j
11j
11j
1ij
1j
1ij
ji
jIon
ji
jIon
1 1i
−−+
+
−−+
+
+
+
+
+
++
+
+
+
+
+
+
+
+
+
+
=⇒=⇒
=→⋅== =
)(je/2/3
1e1j
j
1j
j
j/2/3
1e1j
1j
i
/ij
/2/31e
1j
1j
i
/)(2/31e1j
1j
ij
iijj
jIon
jIon
ji
jIon
ji
jIon
TneTCnU
U
N
N
UeTCnU
NegeTCn
U
N
eTCnNU
gnN
kTE
kTEkTEkTE
kTEE
Φ==
==
==
−
++
−
+
+−−
+
+
−−+
+
∑
∑∑
F
33
Stellar Atmospheres: Emission and Absorption
65
Ionization fraction
j j j 1 J-1
J j 1 j 2 J
J Jj J-1 J-1 J-2 J-1 1
j J Jj 1 j 1 J J J J-1 J 2
j j 1 J-1
j j j 1 j 2 JJ
J-1 J-1 J-2 J-1 1
J J J-1 J 2
J-1
e kj k j
e kk
1
1
( )
1 ( )
J
N N N N
N N N N
N N N N N NN N N N
N N N N N N
N N N
N N N N NN
N N N N NN N N
N N N N N
n TN
Nn T
+
+ +
= =
+
+ +
=
=
= ⋅ ⋅ ⋅
= = = + + ⋅ + +
⋅ ⋅ ⋅
= =+ + ⋅ + +
Φ
=
+ Φ
∑ ∑
∏
K
K L
K
K L
J-1J
m 1 m=∑∏
Stellar Atmospheres: Emission and Absorption
66
Ionization fractions
34
Stellar Atmospheres: Emission and Absorption
67
Summary: Emission and Absorption
F
Stellar Atmospheres: Emission and Absorption
68
Ɣ Line absorption and emission coefficients (bound-bound)
32 2low 0 low
lu lu low up lu lu up2up up
( ) 2
( ) ( ) ( )
= − =
g h ge ef n n f n
mc g c mc g
νπ πκ ν η νϕ ν ϕ ν
profile function, e.g., Voigtprofile( ) =ϕ ν2
2 2
1( , v)
(v )
∞ −
−∞
=− +∆ ∫
y
D
eV a dy
y aν π
Ɣ Continuum (bound-free)*
up h / kTlow up
low
n(v) n n e
n− ν
ν
κ = σ −
*
up h /kTup3
low
n2h( ) n e
c n− ν
ν ν
νη ν = σ
Ɣ Continuum (free-free), always in LTE
( )ff -h / kTff e k( ) ( )n n 1-e νκ ν = σ ν ff ff
e k( ) ( )n n B ( ,T)νη ν = κ ν ν
eeσκ n=Ɣ Scattering (Compton, on free electrons)e en Jν νη (ν) = σ
Total opacity and emissivity add up all contributions, then source function S ( )ν ν= η /κ ν
35
Stellar Atmospheres: Emission and Absorption
69
Excitation and ionization in LTE
( ) /− −= low upE E kTlow low
up up
n ge
n g
up low
3/2( ) /up upe
3low e low
22 − − =
E E kTn gm kT
en n h g
π
Boltzmann
Saha
1
Stellar Atmospheres: Hydrostatic Equilibrium
1
Hydrostatic Equilibrium
Particle conservation
Stellar Atmospheres: Hydrostatic Equilibrium
2
Ideal gas
P
P+dP
r
dr
dA
H
2 2
mass density
atomic weight
forces acting on volume element:
buoyancy:
r rg
P
P pressurek
P TAm
A
dV dAdr dm dV
GM dm GMdF dAdr
r r
dF dPdA
ρ ρ
ρ
ρ
=
= ⋅ =
=
= =
= − = −
= − (pressure difference * area)
2
Stellar Atmospheres: Hydrostatic Equilibrium
3
Ideal gas
In stellar atmospheres:
log g is besides Teff the 2nd
fundamental parameter of
static stellar atmospheres
2
2
2
-
mass of atmosphere negligible
thickness of atmosphere stellar radius
with surface gravity
usually written as log( / cm s
:
)
∗
∗
∗
∗
∗
∗
=
= <<
→ = − =
=
r
g
M M
r R
dF dAdr dAdrGM
gR
GMg
R
g
ρρ
4.0 .... 4.5
4.44
0 .... 1
~8
~15
3.0
Main sequence star
Sun
Supergiants
White dwarfs
Neutron stars
Earth
log gType
Stellar Atmospheres: Hydrostatic Equilibrium
4
Hydrostatic equilibrium, ideal gas
buoyancy = gravitational force:
example:
H
( )
( )( )
0
0
eliminate with ideal gas equation ( )(
)
:
+ =
− − =
= −
= −
P gdF dF
dPdA g dAdr
dPg
drr
A r mr P r
k
dPg
d rr T
ρ
ρ
ρ
0 H
0
H H
( )
0
( )/
0
/
H
( ) const , ( ) const (i.e., no ionization or dissociation)
1 ( )
solution:
P(r)
press
P(r )
P(r
: ure scale height
) P(r )
− −
− −
= = = =
= − ⇒ = −
=
=
= r r gAm
Hr
T
r
k
T r T A r A
Am AmdP dPg P r g
dr k
kTH
gA
T P d
m
r kT
e
e
3
Stellar Atmospheres: Hydrostatic Equilibrium
5
Atmospheric pressure scale heights
Earth:
Sun:
White dwarf:
Neutron star:
km 9
3log
K 300
)(N 28 2
=
=
≈
≈
H
g
T
A
km 180
44.4log
K 6000
)(H 1
=
=
≈
=
H
g
T
A
km 25.0
8log
K 15000
)(H 5.0
=
=
=
+= +
H
g
T
nA e
! mm 6.1
15log
K 10
)(H 5.06 =
=
=
+= +
H
g
T
nA e
H
=kT
HgAm
Stellar Atmospheres: Hydrostatic Equilibrium
6
Effect of radiation pressure
2nd moment of intensity
1st moment of transfer equation (plane-parallel case)
F
F
vR Kc
vPπ4
)( =
0
( )
4 with ( ) ( )
( )
4( )
integration over frequencies:
4( )
∞
=
= =
=
= ∫
vv
Rv
Rv
Rv
dKH
d v
dPH d v v dr
d v c
dPv H
dr c
dPv H dv
dr c
τ
πτ κ
τ
πκ
πκ
4
Stellar Atmospheres: Hydrostatic Equilibrium
7
Effect of radiation pressure
Extended hydrostatic equation
In the outer layers of many stars:
Atmosphere is no longer static, hydrodynamical equation
Expanding stellar atmospheres, radiation-driven winds
ef
0
eff rad
f
0
effective gravity
( (depth d
( ) ( )
( ) ( )
definition:
ependent4 1
: ( )(
!))
( )
)
4
∞
∞
= −
−
= −
=
= − =
∫
∫ v
Rv
dPv
g
dPg r g r
dr
g r r
g v H
H dvd
dv
c
gc
r
rr
g
ρ ρ
ρ
πκ
ρ
πκ
eff rad
0
4 10 i.e. ( )
( )
∞
< = >∫ vg g v H dv gc r
πκ
ρ
Stellar Atmospheres: Hydrostatic Equilibrium
8
The Eddington limit
Estimate radiative acceleration
Consider only (Thomson) electron scattering as opacity
(Thomson cross-section)
number of free electrons per atomic mass unit
Pure hydrogen atmosphere, completely ionized
Pure helium atmosphere, completely ionized
Flux conservation:
e)( σσ =v
=q
1=q
5.04/2 ==q
Hm
q
cdvH
m
q
cdvHn
qmncg
H
v
H
v
H
e
0
e
0
ee
e
e
rad
44
/
14 σπσ
πσ
π=== ∫∫
∞∞
4
eff4
H Tσ
π= F
e 2 44rad e effe
e eff 2
H H
4.51ee
H
414 1
m 4g m 4
/10
4 m /
g q R Tq MT G
c R c G M
q L LLq
c G M M M
σ πσσπ σ
π π
σ
π−
Γ = = =
Γ = =
5
Stellar Atmospheres: Hydrostatic Equilibrium
9
The Eddington limit
Consequence: for given stellar mass there exists a maximum
luminosity. No stable stars exist above this luminosity limit.
Sun:
Main sequence stars (central H-burning)
Mass luminosity relation:
Gives a mass limit for main sequence stars
Eddington limit written with effective temperature
and gravity
Straight line in (log Teff,log g)-diagram
4.51
max 10 1L L q M M−= ⋅ ⋅
1e <<Γ
( )3
max/ / 180L L M M M M≈ → =
0loglog4log12.15
1/10
eff
4
eff
12.15
e
=−++−
==Γ −
gTq
gqT
Stellar Atmospheres: Hydrostatic Equilibrium
10
The Eddington limit
Positions of analyzed
central stars of planetary nebulae
and
theoretical stellar evolutionary tracks
(mass labeled in solar masses)
6
Stellar Atmospheres: Hydrostatic Equilibrium
11
Computation of electron density
At a given temperature, the hydrostatic equation gives the
gas pressure at any depth, or the total particle density N:
NN massive particle density
The Saha equation yields for given (ne,T) the ion- and atomic
densities NN.
The Boltzmann equation then yields for given (NN,T) the
population densities of all atomic levels: ni.
Now, how to get ne?
We have k different species with abundances αkParticle density of species k:
gasP NkT=
atoms ions e N eN N N n N n= + + = +
( )K
N e N
k 1
, and it is k k k kN N N n N Nα α=
= = − =∑
Stellar Atmospheres: Hydrostatic Equilibrium
12
Charge conservation
Stellar atmosphere is electrically neutral
Charge conservation electron density=ion density * charge
Combine with Saha equation (LTE)
by the use of ionization fractions:
We write the charge conservation as
Non-linear equation, iterative solution, i.e., determine zeros of
e
1 1
, density of j-th ionization stage of species k jkK
jk jk
k j
n j N N= =
= ⋅ =∑∑
∑∏
∏
= =
=
+
==jk
m
jk-
ml
lk
jk-
jl
lk
k
jk
jk
(T)ĭn
(T)ĭn
N
Nf
1
1
e
1
e
1
)(),()(
),()(),(
e
1 1
eee
1 1
ee
1
e
1
e
nFTnfjnNn
TnfjnNTnfNjn
K
k
jk
j
jkk
K
k
jk
j
jkk
K
k
jk
j
jkk
=⋅−=
⋅−=⋅=
∑ ∑
∑ ∑∑∑
= =
= == =
α
α
0)( ee =− nnF use Newton-Raphson, converges after 2-4 iterations;yields ne and fij, and with Boltzmann all level populations
7
Stellar Atmospheres: Hydrostatic Equilibrium
13
Summary: Hydrostatic Equilibrium
Stellar Atmospheres: Hydrostatic Equilibrium
14
Summary: Hydrostatic Equilibrium
Hydrostatic equation including radiation pressure
Photon pressure: Eddington Limit
Hydrostatic equation → N
Combined charge equation + ionization fraction → ne
→ Population numbers nijk (LTE) with Saha and Boltzmann
equations
0
4( ) ( ) ( )R
v
dPdPg r g r v H dv
dr dr c
πρ ρ κ
∞
= − = − ∫
8
Stellar Atmospheres: Hydrostatic Equilibrium
15
3 hours
Stellar
Atmospheres……per day… …is too much!!!
1
Stellar Atmospheres: Radiative Equilibrium
1
Radiative Equilibrium
Energy conservation
Stellar Atmospheres: Radiative Equilibrium
2
Radiative Equilibrium
Assumption:
Energy conservation, i.e., no nuclear energy sources
Counter-example: radioactive decay of Ni56 →Co56 →Fe56 in
supernova atmospheres
Energy transfer predominantly by radiation
Other possibilities:
Convection e.g., H convection zone in outer solar layer
Heat conduction e.g., solar corona or interior of white dwarfs
Radiative equilibrium means, that we have at each location:
Radiation energy absorbed / sec
=
Radiation energy emitted / sec
integrated over all
frequencies and
angles
2
Stellar Atmospheres: Radiative Equilibrium
3
Radiative Equilibrium
Absorption per cm2 and second:
Emission per cm2 and second:
Assumption: isotropic opacities and emissivities
Integration over dω then yields
Constraint equation in addition to the radiative transfer
equation; fixes temperature stratification T(r)
∫ ∫∞
π
κω4 0
)( vIvdvd
∫ ∫∞
π
ηω4 0
)(vdvd
( ) 0)( )()(000
=−⇒= ∫∫∫∞∞∞
dvSJvvdvJvdv vvv κηκ
Stellar Atmospheres: Radiative Equilibrium
4
Conservation of flux
Alternative formulation of energy equation
In plane-parallel geometry: 0-th moment of transfer equation
Integration over frequency, exchange integration and
differentiation:
( )vvv SJ
dt
dH−= κ
( )0 0
4
eff
0
4
eff
0 0
0 because of radiative equilibrium
const for all depths. Alternatively written:4
(1st moment of transfer equat4
∞ ∞
∞
∞ ∞
= − =
⇒ = = =
= =
∫ ∫
∫
∫ ∫
v v v
v
v
v
dH dv J S dv
dt
H H dv T
dv T dvdK
Hd
κ
σ
π
σ
π τ
4
eff
0
ion)
( ) (definiton of Eddington factor)
4
∞
⇒ =∫ v vddv T
d
f J σ
τ π
F
F
3
Stellar Atmospheres: Radiative Equilibrium
5
Which formulation is good or better?
I Radiative equilibrium: local, integral form of energy
equation
II Conservation of flux: non-local (gradient), differential form
of radiative equilibrium
I / II numerically better behaviour in small / large depths
Very useful is a linear combination of both formulations:
A,B are coefficients, providing a smooth transition between
formulations I and II.
( ) 0)(
00
=
−⋅+
−⋅ ∫∫
∞∞
Hdvd
JfdBdvSJA vv
vv τκ
Stellar Atmospheres: Radiative Equilibrium
6
Flux conservation in spherically symmetric geometry
0-th moment of transfer equation:
( ) ( )
( )
HRLLdvHr
dvJSrdvHrr
JSHrrr
v
vvv
vvv
22
2
0
2
0
2
0
2
2
2
16 because 16
1const
0
1
ππ
κ
κ
===
=−=
∂
∂⇒
−=∂
∂
∫
∫∫∞
∞∞
4
Stellar Atmospheres: Radiative Equilibrium
7
Another alternative, if T de-couples from radiation field
Thermal balance of electrons
∑
∑
∑ ∫
∑ ∫
∑ ∫
∑ ∫
=
=
+
−=
−=
+=
=
=−
∞−
∞
∞−
∞
ml
lmlml
C
ml
lmlmm
H
kl
kThv
vlk
vlkk
C
kl
lkvlkl
H
j
kThv
vjj
C
j
vjj
H
CH
hvTqnnQ
hvTqnnQ
dvec
hvJ
v
vJvnQ
dvv
vJvnQ
dvec
hvJTvNnQ
dvJTvNnQ
,
ec
,
ec
, 0
2
3
bf,bf
, 0
bf,bf
0
2
3
ff,eff
0
ff,eff
)(
)(
21)( 4
1)( 4
2
),( 4
),( 4
0
απ
απ
απ
απ
Stellar Atmospheres: Radiative Equilibrium
8
The gray atmosphere
Simple but insightful problem to solve the transfer equation
together with the constraint equation for radiative equilibrium
Gray atmosphere: κκν =
( ) ( )
( ) ( )
( )
Moments of transfer equation
with
Integration over frequency
Radiative equilibrium ( ) ( ) 0
and because
= − = =
= − =
− = − = − =
=⇒∫ ∫
v vv v v
v v v v
dH dKI J S II H dt
d d
dH dKI
J
J S II Hd d
J S dv J S dv J
S
S
I
τ κτ τ
τ τ
κ κ
( ) ( )1 2 1 2
2
2
of conservation of flux 0
from follows , see below0
=
⇒ ⇒ == + = =
dH
d
dKII K c c II c H
d
dc
K
dτ
τ
ττ
5
Stellar Atmospheres: Radiative Equilibrium
9
The gray atmosphere
Relations (I) und (II) represent two equations for three
quantities S,J,K with pre-chosen H (resp. Teff)
Closure equation: Eddington approximation
Source function is linear in τ
Temperature stratification?
In LTE:
( )2K 1 3J J 3K IIIS 3H 3c= → = = = τ +
( )
( )
4
4
2
4
eff
4 4
eff 2 2
( ) ( ( ))
insert into : 3 3
with we get: 4
3( ) 3 is now determined from boundary condition ( =0)
4
= =
= +
=
= +
S B T T
III T H c
H T
T T c IV c
στ τ
πσ
τπ
σ
πσ
τ σ τ τπ π
F
Stellar Atmospheres: Radiative Equilibrium
10
Gray atmosphere: Outer boundary condition
Emergent flux:
( )
( )
[ ]
3
23 ,
3
2
4
3
3
2
2
1
3
1
2
3)0(
)(12
1)( and
!)(with
)()( 2
3
)(332
1
from with )()(2
1)0(
4
eff
4
22
12
0
0
22
0
2
0
22
0
2
+=
+=⇒
=→
+=
−−
=+
=
′′+′′′=
′′+′=
′′′=
−∞
∞∞
∞
∞
∫
∫∫
∫
∫
ττ
τττττ
τττ
τττ
HSTT
HccHH
ttEetEnl
ldttEt
dEcdEH
dEcH
IIISdESH
t
n
l
F
F
from (IV): (from III)
6
Stellar Atmospheres: Radiative Equilibrium
11
Avoiding Eddington approximation
Ansatz:
Insert into Schwarzschild equation:
Approximate solution for J by iteration (“Lambda iteration“)
( )
))((4
3)(
function Hopf )(
oftion generaliza ))((3)(
4
eff ττπ
στ
τ
τττ
qTJ
q
IIIqHJ
+=
=
+=
F
( ) ( )∫∞
′−′′+′=+⇒
Λ=Λ=
0
1)(2
1)(
Jfor equation integral )(
τττττττ
τ
dEqq
JSJ
( )
+−+=+Λ=Λ=
+=
)(2
1)(
3
1
3
23)32(3
)32(3
32
)1()2(
)1(
ττττ
τ
EEHHJJ
HJ
F
(*) integral equation for q, see below
i.e., start with Eddington approximation
(was result for linear S)
Stellar Atmospheres: Radiative Equilibrium
12
At the surface
At inner boundary
Basic problem of Lambda Iteration: Good in outer layers, but
does not work at large optical depths, because exponential
integral function approaches zero exponentially.
Exact solution of (*) for Hopf function, e.g., by Laplace
transformation (Kourganoff, Basic Methods in Transfer Problems)
Analytical approximation (Unsöld, Sternatmosphären, p. 138)
( )358.034
1
3
1
3
23
2
1)0( , 1)0( , 0
)2(
32
+=
+−+=
===
ττ
τ
HHJ
EE
2 3
(2)
, ( ) 0 , ( ) 0
23
3
E E
J H
τ
τ
= ∞ ∞ = ∞ =
= +
ττ 972.11167.06940.0)( −−≈ eq
exact: q(0)=0.577….
7
Stellar Atmospheres: Radiative Equilibrium
13
Gray atmosphere: Interpretation of results
Temperature gradient
The higher the effective temperature, the steeper the
temperature gradient.
The larger the opacity, the steeper the (geometric) temperature
gradient.
Flux of gray atmosphere
τκ
τ
ττ
d
dT
dt
dT
Td
dT
Td
dTTT
d
d
=
==
4
eff
4
eff
34
~
4
34
[ ]
2 2
0
1/ 4
eff eff
LTE: ( ( ))
1 1( ) ( ( )) ( ) ( ( )) ( )
2 2
with , 3 4 ( ( )) ( ) ( )
and
v v
v v v
v
S B T
H B T E t dt B T E t dt
hv kT T T q p hv kT p
H d H dv H
τ
τ
α
τ
τ τ τ τ τ
α τ τ τ α τ
σα
∞
=
= − − −
= = + = → =
= =
∫ ∫
3 3 3
2 3 3
4
eff
43eff 2 2
4 3 2
eff 0
1 2 4
2
4
( ) ( )4 4( ) /
exp( ( )) 1 exp( ( )) 1
vv
hv k k
hc h v
T
H kT E t E tdv kH H H dt dt
H d T h h c p p
τ
α
τ
π α
σ
π
τ τπ πτ α
α σ σ α τ α τ
∞ − −→ = = = −
− − ∫ ∫
Stellar Atmospheres: Radiative Equilibrium
14
Gray atmosphere: Interpretation of results
Limb darkening of total radiation
i.e., intensity at limb of stellar disk smaller than at center by
40%, good agreement with solar observations
Empirical determination of temperature stratification
Observations at different wavelengths yield different T-
structures, hence, the opacity must be a function of
wavelength
4 4
eff
3 2I( 0, ) S( ) B(T( )) T ( ) T
4 3
I(0, ) 2 / 3 2 3(1 cos )
I(0,1) 1 2 / 3 5 2
σ σ τ = µ = τ = µ = τ = µ = τ = µ = µ +
π π µ µ +
→ = = + ϑ+
TTBSSI →=→=→= ))(()()( ),0( measure ττµτµτ
F
8
Stellar Atmospheres: Radiative Equilibrium
15
The Rosseland opacity
Gray approximation (κ=const) very coarse, ist there a good mean value ? What choice to make for a mean value?
For each of these 3 equations one can find a mean , with which the equations for the gray case are equal to the frequency-integrated non-gray equations.
Because we demand flux conservation, the 1st momentequation is decisive for our choice: → Rosseland mean of opacity
transfer equation
0-th moment
1st moment
non-graygray
)( ISdz
dI−= κµ ))(( vv
v ISvdz
dI−= κµ
))(( vvv JSv
dz
dH−= κ
vv Hv
dz
dK)(κ−=
0)( =−= JSdz
dHκ
Hdz
dKκ−=
κ
κ
Stellar Atmospheres: Radiative Equilibrium
16
The Rosseland opacity
Definition of Rosseland mean of opacity
3
0
340
0
00
4
)(
1
1
4 and with
)(
1
1
: LTE and 3/1ion approximatEddington with )(
1
1
1
)(
1const
T
dvdT
dB
v
dz
dTTT
dz
d
dz
dB
dz
dT
dT
dB
dz
dB
dz
dB
dvdz
dB
v
BJJK
dz
dK
dvdz
dK
v
dz
dKdv
dz
dK
vdvH
v
R
vv
v
R
v
R
R
vv
π
σ
κ
κ
π
σ
π
σκ
κ
κ
κ
κκ
∫
∫
∫
∫∫
∞
∞
∞
∞∞
=
=
===
===
===
9
Stellar Atmospheres: Radiative Equilibrium
17
The Rosseland opacity
The Rosseland mean is a weighted mean
of opacity with weight function
Particularly, strong weight is given to those frequencies, where the radiation flux is large.
The corresponding optical depth is called Rosseland depth
For the gray approximation with is very good,
i.e.
Rκ
1
)(
1
vκ dT
dBv
Rκ1>>Rossτ
∫ ′′=z
RRoss zdzz0
)()( κτ
))((4
3)(
4
eff
4
RossRossRoss qTT τττ +=
Stellar Atmospheres: Radiative Equilibrium
18
Convection
Compute model atmosphere assuming
• Radiative equilibrium (Sect. VI) → temperature stratification
• Hydrostatic equilibrium → pressure stratification
Is this structure stable against convection, i.e. small
perturbations?
• Thought experiment
Displace a blob of gas by ∆r upwards, fast enough that no heat
exchange with surrounding occurs (i.e., adiabatic), but slow
enough that pressure balance with surrounding is retained (i.e.
<< sound velocity)
10
Stellar Atmospheres: Radiative Equilibrium
19
Inside of blob outside
Stratification becomes unstable, if temperature gradient
rises above critical value.
( ), ( )T r rρ
ad ad
ad ad
( )
( )
T T T r r
r rρ ρ ρ
+ ∆ = + ∆
+ ∆ = + ∆
r∆
( ), ( )T r rρ
rad rad
rad rad
( )
( )
T T T r r
r rρ ρ ρ
+ ∆ = + ∆
+ ∆ = + ∆
ad rad
ad rad
ad rad
a
sta
( ) ( ) further buoyancy,
( ) ( ) gas blob falls back,
i.e.
with ideal gas equation p= and pressure balance
ble
stable
unstable
unsta
ble
+ ∆ < + ∆ →
+ ∆ > + ∆ →
<
>
H
r r r r
r r r r
d d
dr dr
kT
Am
ρ ρ
ρ ρ
ρ ρ
ρ ρ d ad rad rad
ad radunstable
s
=
table
>
<
T T
dT dT
dr dr
ρ
addT dr
Stellar Atmospheres: Radiative Equilibrium
20
Alternative notation
Pressure as independent depth variable:
Schwarzschild criterion
Abbreviated notation
eff eff
eff
eff eff
ad rad
hydrostatic equation: ( )
(ln )
(ln )
(ln ) (ln )
(ln ) (
ide
unstabl
al gas
stabln
e
le )
= − = −
→ = −
= − = −
<
>
H
H
H H
dp g dr g dr
kTdpAm g p
Am AmdT dT T d Tg g
k dp p k d p
d T d T
d
d
Am p
k T
p d
r
dr
p
ρ
ad rada
ad d
d rad
ra
(ln ) (ln );
(ln ) (l
sta
n
b
)
le
∇ = ∇ =
∇ > ∇
d T d T
d p d p
11
Stellar Atmospheres: Radiative Equilibrium
21
The adiabatic gradient
Internal energy of a one-atomic gas excluding effects of
ionisation and excitation
But if energy can be absorbed by ionization:
Specific heat at constant pressure
V V
0 (no heat exchange)
(1st law of therm
intern 0 (al
ody
energy
namic
*)
s)
=
= +
= ⇒ + =
dQ
dQ pdV
c d
dE
dE c dT T pdV
V
3 3
2 2E NkT c Nk= → =
V
3
2c Nk>>
p V V
t t
p V
( )
p cons p cons
c c
Q dE dV d NkT p Nkc p c p c p
T dT dT dT
Nk
p= =
∂
−
= = + = + = +∂
=→
Stellar Atmospheres: Radiative Equilibrium
22
The adiabatic gradient
( )
p V
p VV
p V
V
p
p V
V
V
p
V
p
V
Ideal gas:
(**)
/
from(*) with (**) 0
0
0
(ln ) (l
pV
n )
defi
= ⇒ + = =
+=
−
+−→ + =
⋅−
−+ + =
+ =
= −
−pV NkT Vdp pdV dT dT
Vdp pdVdT
c c
V pc cc pdV
c cc
c cdV
p V c
cdp dV
dp d
p V c
cd V d p
V
d
Nk c
p d
V
c
c
V
p
V
(ln ) 1nition: :
(ln )= = −c d V
c d pγ
γ
12
Stellar Atmospheres: Radiative Equilibrium
23
The adiabatic gradient
ad
ad
rad
(ln )needed:
(ln )
/
ln ln ln ln( )
(ln ) (ln )1
(ln ) (ln )
(ln ) 1 1
1 st
1(ln )
b e
1
a l
=
= + −
=
−∇ <
+
−= − =
−∇ =
d T
d p
T pV Nk
T p V Nk
d T d V
d p d p
d T
d p
γ
γ γ
γ
γ
γ
γ
Schwarzschild criterion
Stellar Atmospheres: Radiative Equilibrium
24
The adiabatic gradient
• 1-atomic gas
• with ionization
• Most important example: Hydrogen (Unsöld p.228)
V p V
ad
3 2 5 2
5 3 2 5 0.4
c Nk c c Nk Nk
γ
= = + =
= ∇ = =
ad1 0 convection starts effect−→ ∇ → γγ
( ) ( )
( ) ( )
2
Ion
ad 22
Ion
2
2 5 2
5 5 2
( ) ( ) ( )with ionization degree
2 2
+ − +∇ =
+ − +
= − + +
x x E kT
x x E kT
f T f T f Tx
N N NF
13
Stellar Atmospheres: Radiative Equilibrium
25
The adiabatic gradient
( )( )
( ) ( )
2
Ion
ad 22
Ion
2
2 5 2
5 5 2
( ) ( ) ( )
2 2
x x E kT
x x E kT
f T f T f Tx
N N N
+ − +∇ =
+ − +
= − + +
Stellar Atmospheres: Radiative Equilibrium
26
Example: Grey approximation
( )( ) ( )
( )( )( )
4 4
eff
4
eff
1
3 2( )4 3
3 24ln ln ln4 3
2ln3
4
hydrostatic equation: Ansatz: ( here a mass absorption coefficient)
(ln ) 1
243
1 integrate
1
+
= +
= + +
+= =
= =
→ = → = →+
+
b
b b
b
T T
T T
d
d
dp gAp
d
g g gp p
A b A
d T
Ap
dp
d
d
τ τ
τ
τ
τ
κ κτ κ
τ τ
ττ
( )
1
1
rad
rad
1
( 1)
1 1
( 1)
243
becomes large, if opacity strongly increases with depth (i.e
(ln ) 1
(
. exponent b large).
The absolute value of is
1)
l
not
l
essenti
n
n
+
+
=+
= = = =
+= =
+
∇
+
∇
b b
b
g g
p p
d T
A
d
d p
d b
d p
p
p
d p
d
d
A
b
τ
ττ
τ
τ τ
τ
κ
τ
rad ad
a
l but
large
the change o
(> ): c
f with depth (g
onvection starts,
radient
-Ef ekt
)
f∇ ∇
κ
κ
14
Stellar Atmospheres: Radiative Equilibrium
27
Hydrogen convection zone in the Sun
κ-effect and γ-effect act together
Going from the surface into the interior: At T~6000K ionization of
hydrogen begins
∇ad decreases and κ increases, because a) more and more
electrons are available to form H−
and b) the excitation of H is
responsible for increased bound-free opacity
In the Sun: outer layers of atmosphere radiative
inner layers of atmosphere convective
In F stars: large parts of atmosphere convective
In O,B stars: Hydrogen completely ionized, atmosphere radiative;
He I and He II ionization zones, but energy transport by
convection inefficient
Video
Stellar Atmospheres: Radiative Equilibrium
28
Transport of energy by convection
Consistent hydrodynamical simulations very costly;
Ad hoc theory: mixing length theory (Vitense 1953)
Model: gas blobs rise and fall along distance l (mixing length).
After moving by distance l they dissolve and the surrounding
gas absorbs their energy.
Gas blobs move without friction, only accelerated by buoyancy;
detailed presentation in Mihalas‘ textbook (p. 187-190)
( ) = pressure scale height
mixing length parameter
=0.5 2
=
L
l H r Hα
α
α
15
Stellar Atmospheres: Radiative Equilibrium
29
Transport of energy by convection
Again, for details see Mihalas (p. 187-190)
For a given temperature structure
4
rad eff
conv
conv
ad rad
( )
compute ( )
flux conservation including convective flux
new temperature stratification ( )
wit
(
h
)
= −
∇ < ∇ < ∇
→
→
→
F r T F r
F r
T r
σ
πiterate
Stellar Atmospheres: Radiative Equilibrium
30
Summary: Radiative Equilibrium
16
Stellar Atmospheres: Radiative Equilibrium
31
Radiative Equilibrium:
Schwarzschildt Criterion:
Temperature of a gray Atmosphere
( ) 0)(
00
=
−⋅+
−⋅ ∫∫
∞∞
Hdvd
JfdBdvSJA vv
vv τκ
4 4
eff
3 2
4 3T T τ
= +
ad rad(ln ) (ln )
(
unst
stable ln ) (
bl
ln
e
)
ad T d T
d p d p
>
<
Stellar Atmospheres: Radiative Equilibrium
32
3 hours
Stellar
Atmospheres……per day… …is too much!!!
1
Stellar Atmospheres: Non-LTE Rate Equations
1
The non-LTE Rate Equations
Statistical equations
Stellar Atmospheres: Non-LTE Rate Equations
2
Population numbers
LTE: population numbers follow from Saha-Boltzmann equations, i.e. purely local problem
Non-LTE: population numbers also depend on radiation field. This, in turn, is depending on the population numbers in all depths, i.e. non-local problem.
The Saha-Boltzmann equations are replaced by a detailed consideration of atomic processes which are responsible for the population and de-population of atomic energy levels:
Excitation and de-excitation
by radiation or collisions
Ionization and recombination
),(**
eii nTnn =
),,( JnTnn eii =
2
Stellar Atmospheres: Non-LTE Rate Equations
3
Statistical Equilibrium
Change of population number of a level with time:
= Sum of all population processes into this level
- Sum of all de-population processes out from this level
One such equation for each level
The transition rate comprises radiative rates
and collision rates
In stellar atmospheres we often have the stationary case:
These equations determine the population numbers.
∑∑≠≠
−=ij
iji
ij
jiji PnPnndt
d∑
∑
≠
≠
−
=
ij
iji
ij
jij
i
Pn
Pn
ndt
d
0 hence for all levels ≠ ≠
= =∑ ∑i j ji i ij
j i j i
dn n P n P i
dt
ijPijR
ijC
Stellar Atmospheres: Non-LTE Rate Equations
4
Radiative rates: bound-bound transitions
Two alternative formulations:a) Einstein coefficientsb) Line absorption coefficients advantage a): useful for analytical expressions with simplified
model atomsadvantage b): similar expressions in case of bound-free
transitions: good for efficient programmingNumber of transitions i→j induced by intensity Iν in frequency
interval dν und solid angle dω
Integration over frequencies and angles yields
Or alternatively
jijiij ABB
ij( )σ ν
/ 4 (absorbed Energy / )i ij v vn B I dvd hvϕ ω π
∫∞
=0
dvJBnRn vvijiiji ϕ
∫∞
==0
)(4 4/)()(with dvJ
hv
vnRnhvvBv v
ij
iijiijij
σππϕσ
F
3
Stellar Atmospheres: Non-LTE Rate Equations
5
Radiative rates: bound-bound transitions
In analogy, number of stimulated emissions:
Number of spontaneous emissions:
Total downwards rate:
∫
∫∫∞
∞∞
=′
==′
0
00
4 dvJhvg
gnRn
dvJg
gBndvJBnRn
v
ij
j
ijjij
vv
j
iijjvvjijjij
σπ
ϕϕ
∫
∫∫∞
∞∞
=′′
==′′
0
2
3
0
2
3
0
24
2
dvc
hv
hvg
gnRn
dvc
hvBndvAnRn
ij
j
ijjij
vjijvjijjij
σπ
ϕϕ
( )
[ ]
3
2
0
* *3
2
0
24
24
ijij ji j ji ji j v
j
ij hv kTi ij ji j ji j v
j j
g hvn R n R R n J dv
g hv c
n n hvn R n R n J e dv
n n hv c
σπ
σπ
∞
∞−
′ ′′= + = +
= = +
∫
∫
Stellar Atmospheres: Non-LTE Rate Equations
6
Radiative rates: bound-free transitions
Also possible: ionization into excited states of parent ion
Example C III:
Ground state 2s2 1S
Photoionisation produces C IV in ground state 2s 2S
C III in first excited state 2s2p 3Po
Two possibilities:
Ionization of 2p electron ĺ C IV in ground state 2s 2S
Ionization of 2s electron ĺ C IV in first excited state 2p 2P
C III two excited electrons, e.g. 2p2 3P
Photoionization only into excited C IV ion 2p 2P
4
Stellar Atmospheres: Non-LTE Rate Equations
7
Radiative rates: bound-free transitions
Number of photoionizations = absorbed energy in dν, divided by
photon energy, integrated over frequencies and solid angle
Number of spontaneous recombinations:
∫∫ ∫∞∞
=→00
)(4 dvJ
hv
vnRndvdIpn v
ij
iijivvi
σπω FF
3
2
0 0
3
2
0
*3
2
*
0
2(v) (v) v v 4 (v)
2
(v
4 (v)
( ) 24
)
1
(v)
j e j ji j e
j ji j e
ij hv kTij ji j
hv k
v
e
j
T i
j
G
nmp e
hv hn n F d d n R n n dv
c m
hv hn R n n dv
c m
vn hvn R n e dv
n hv c
h n n
ω π
π
σπ
∞ ∞
∞
∞−
−
→ =
=
=
∫ ∫ ∫
∫
∫
F
Stellar Atmospheres: Non-LTE Rate Equations
8
Radiative rates: bound-free transitions
Number of induced recombinations
Total recombination rate
∫
∫
∫∫ ∫
∞−
∞−
∞∞
=
=
=→
0
*
0
*
00
)(4
v)(
1v)(4
)v(v)(4vv)v(v)(
dveJhv
v
n
nnRn
dvm
hJ
nn
ne
h
mpnnRn
dvm
hJGnnRnddIGnn
kThv
v
ij
j
ijjij
v
ej
ikThv
vejjij
vejjijvej
σπ
π
πω
∫∞
−
+
=
0
2
3**
2)(4 dveJ
c
hv
hv
v
n
nnR
n
nn
kThv
v
ij
j
ijji
j
ij
σπ
5
Stellar Atmospheres: Non-LTE Rate Equations
9
Radiative rates
Upward rates:
Downward rates:
Remark: in TE we have
∫∞
−
+=
0
2
3
*
2)(4
with
dveJc
hv
hv
vR
Rn
nn
kThv
v
ij
ji
ji
j
ij
σπ
∫∞
=0
)(4
with
dvJhv
vR
Rn
v
ij
ij
iji
σπ
*
**
=⇒=⇒=
j
i
j
ijiijvv
n
n
n
nRRBJ
Stellar Atmospheres: Non-LTE Rate Equations
10
Collisional rates
Stellar atmosphere: Plasma, with atoms, ions, electrons
Particle collisions induce excitation and ionization
Cool stars: matter mostly neutral ⇒ frequent collisions with neutral hydrogen atoms
Hot stars: matter mostly ionized ⇒ collisions with ionsbecome important; but much more important becomeelectron collisions
Therefore, in the following, we only consider collisions of atoms and ions with electrons.
Am
Am
e
H
ion
electron 43masselectron
massion
v
v2/12/1
≈
=
=
6
Stellar Atmospheres: Non-LTE Rate Equations
11
Electron collisional rates
Transition i→j (j: bound or free), σij (v) = electron collisioncross-section, v = electron speed
Total number of transitions i→j:
minimum velocity necessary for excitation (threshold)
velocity distribution (Maxwell)
In TE we have therefore
Total number of transitions j→i:
)(vv)v()v(
0v
TnndfnnCn ijeiijeiiji Ω== ∫∞
σ
v)v( df
0v
jiij CC *
j
*
i nn =
ij
j
iji C
n
nC
*
jj nn
=
Stellar Atmospheres: Non-LTE Rate Equations
12
Electron collisional rates
We look for: collisional cross-sections σij (v)
• experiments
• quantum mechanical calculations
Usually: Bohr radius πa02 as unit for cross-section σij (v)
σij (v) = πa02 Q ij
Q ij usually tabulated as function of energy of colliding electron
2
0
0
0
1 2
2 v 2 2e
v
1 2
2 11
0 0 0
u
0
m( ) (v) (v)v v with 1 2 v and (v) v 4 v v
2 kT
8(ukT) u with and 5.456 10
( ) ( ) wit
u:
(
E
h
m kT
ij ij
u
ij
e
u
ij ij ij
T f d m
k
E f d e d
kC T Q ue d C aT
m
T C T e T
σ ππ
ππ
∞−
∞− −
−
=
Ω = = =
= = = ⋅
Ω = Γ Γ
∫
∫
0 0 0
0
) ( xkT)( ) x, x: ( E - E )x
ijT Q E x u e d kT kT∞
−= + + =∫
7
Stellar Atmospheres: Non-LTE Rate Equations
13
Electron collisional rates
Advantage of this choice of notation:
Main temperature dependence is described by
only weakly varying function of T
Hence, simple polynomial fit possible
⇒ Important for numerical application
Now: examples how to compute the Cij
)()( 0
0 TeTCT ij
u
ij Γ=Ω −
0 kTEeT−
)(TijΓ
i ij i e ijn C n n (T)= Ω
Stellar Atmospheres: Non-LTE Rate Equations
14
Computation of collisional rates: Excitation
Van Regemorter (1962): Very useful approximate formula for allowed dipole transitions
There exist many formulae, made for particular ions and transitions, e.g., (optically) forbidden transitions between n=2 levels in He I (Mihalas & Stone 1968)
coefficients c tabulated for each transition
[ ]
=
=Γ==
Γ
= −
else 2.0
number quantum principal equal with levelsbetween tionsfor transi 7.0
)(276.0,max)( /
n transitioradiative ofstrength oscillator
energy ionizationhydrogen
)(5.14
010000
00
2
0
0
0
0
g
uEegukTEuhvE
f
E
ueuE
EfTnCC
u
ij
ij
H
uHijeij
( ) 2
210
0
logloglogwith
)(0
−−
−
++=Γ
Γ=
TcTcc
TeTnCCu
eij
8
Stellar Atmospheres: Non-LTE Rate Equations
15
Computation of collisional rates: Ionization
The Seaton formula is in analogy to the van-Regemorter formulain case of excitation. Here, the photon absorption cross-sectionfor ionization is utilized:
Alternative: semi-empirical formula by Lotz (1968):
For H und He specific fit formulae are used, mostly from Mihalas(1967) and Mihalas & Stone (1968)
>
=
=
=
=
⋅=−
2 charge with ionsfor 0.3
2 charge with ionsfor 0.2
1 charge with ionsfor 0.1
ionizationfor section -crossphoton threshold
1055.1
0
0
0
130
Z
Z
Z
g
u
e
T
ngC
u
eij
σ
σ
[ ]
atoms individual toadjusted ,quantities empirical
/)()(5.2
01
1110010
2
0
0
a,b,cc uu
uuEubeuEuE
EaTnCC cH
eij
+=
−
=
Stellar Atmospheres: Non-LTE Rate Equations
16
Autoionization and dielectronic recombination
b bound state, d doubly excited state, autoionization level
c ground state of next Ion
d → c: Autoionization. d decays into ground state of nextionization stage plus free electron
c → d → b: Dielectronic recombination. Recombination via a doubly excited state of next lower ionization stage. d auto-ionizes again with high probability: Aauto=1013...1014/sec! But sometimes a stabilizing transition d → b occurs, bywhich the excited level decays radiatively.
0
negative
positive
Energyb
d
c
ion I, e.g. He I
ion II, e.g. He II
ionization energy
9
Stellar Atmospheres: Non-LTE Rate Equations
17
Computation of rates
Number of dielectronic recombinations from c to b:
In the limit of weak radiation fields the reverse process can beneglected. Then we obtain (Bates 1962):
So, the number of dielectronic recombinations from c to b is:
n transitiogstabilizin sspontaneoufor y probabilit == ssdcbc AAnRn
tionautoionizafor y probabilitn transitio
)( with )/(/2/3
1
**dIon
=
Φ==+= −
a
cdec
kTE
ecdsaadd
A
TnneTCnnnAAAnn
)/()( saascdeccbc AAAATnnRn +Φ=
Stellar Atmospheres: Non-LTE Rate Equations
18
Computation of rates
There are two different regimes:
a) high temperature dielectronic recombination HTDR
b) low temperature dielectronic recombination LTDR
for the cases that the autoionizing levels are close to the ionization limit (b) or far above it (a)
a) Most important recombination process He II → He I in the solar corona (T~2⋅106K)
b) Very important for specific ions in photospheres (T< 105K) e.g. N III λ4634-40Å emission complex in Of stars
Reason: upper level is overpopulated, because a stabilizingtransition is going into it.
Because in case b) *
ddsa nnAA =→>>
10
Stellar Atmospheres: Non-LTE Rate Equations
19
LTDR
The radiation field in photospheres is not weak, i.e., the reverse process b ĺ d is induced
Recombination rate:
Reverse process:
These rates are formally added to the usual ionization and recombination rates and do not show up explicitly in the rate equations.
2
3( ) 1
2
mean intensity in stabilizing transition, i.e.,
given by continuum value (line very broad, because short lifetime)
c cb c e cd s
cn R n n T A J
hv
J
= Φ +
Jg
g
hv
cAnJBnRn
b
dsbbdbbcb 3
2
2==
Stellar Atmospheres: Non-LTE Rate Equations
20
Complete rate equations
For each atomic level i of each ion, of each chemical elementwe have:
In detail:
0i j
j i j i
jj iin n PP≠ ≠
− =∑ ∑
( )
( )
( )
( )
*
*
0
i
j i
j i
j
j i
j
j
iji ij
j
ji j
ij ij
j
i
i
j i
i
i
j
R C
nR
nR C
n
Cn
n
C
n
Rn
>
<
>
<
+
+
+
+
+
−
−
=
∑
∑
∑
∑
rates out of i
rates into i
excitation and ionization
de-excitation and recombination
de-excitation and recombination
excitation and ionization
11
Stellar Atmospheres: Non-LTE Rate Equations
21
Closure equation
One equation for each chemical element is redundant, e.g., the equation for the highest level of the highest ionization stage; to see this, add up all equations except for the final one: these rate equations only yield population ratios.
We therefore need a closure equation for each chemical species:
Abundance definition equation of species k, written for example as number abundance yk relative to hydrogen:
∑∑=
hydrogen of numbers population
k species of numbers populationky
Stellar Atmospheres: Non-LTE Rate Equations
22
Abundance definition equation
Notation:
Population number of level i in ionization stage l : nl,i
NLTE levels
LTE levels
E0
do not appear explicitly in the rate equations; populations depend on ground level of nextionization stage:
)(,1,1
*
, Tnnn illeil Φ= +
12
Stellar Atmospheres: Non-LTE Rate Equations
23
Abundance definition equation
Notation:
NION number of ionization stages of chemical element k
NL(l) number of NLTE levels of ion l
LTE(l) number of LTE levels of ion l
Also, one of the abundance definition equations is redundant, since abundances are given relative to hydrogen (otherdefinitions don‘t help) ⇒ charge conservation
Φ++=
Φ+
⇒
++=
+
∑∑∑ ∑∑
∑∑∑ ∑∑
=== =+
=
=== ==
)(
1
)(
1
,
1
)(
1
1,1
)(
1
,
)(
1
*)(
11
)(
1
*
,
)(
1
,
)(1)(HLTE
i
ieprotons
HNL
i
ilk
NION
l
lLTE
i
liel
lNL
i
il
protons
HLTE
i
i
HNL
i
ik
NION
l
lLTE
i
il
lNL
i
il
TnnnyTnnn
nnnynn
Stellar Atmospheres: Non-LTE Rate Equations
24
Charge conservation equation
Notation:
Population number of level i, ion l, element k: nkli
NELEM number of chemical elements
q(l) charge of ion l
e
NELEM
k
NION
l
lLTE
i
klielk
klNL
i
kli
e
NELEM
k
NION
l
klLTE
i
kli
klNL
i
kli
nTnnnlq
nnnlq
=
Φ+
⇒=
+
∑ ∑ ∑∑
∑ ∑ ∑∑
= = =+
=
= = ==
1 1
)(
1
1,1,
),(
1
1 1
),(
1
*),(
1
)()(
)(
13
Stellar Atmospheres: Non-LTE Rate Equations
25
Complete rate equations: Matrix notation
Vector of population numbers
One such system of equations per depth point
Example: 3 chemical elements
Element 1: NLTE-levels: ion1: 6, ion2: 4, ion3: 1
Element 2: NLTE-levels: ion1: 3, ion2: 5, ion3: 1
Element 3: NLTE-levels: ion1: 5, ion2: 1, hydrogen
Number of levels: NLALL=26, i.e. 26 x 26 matrix
( )1 2 NLALL, , , NLALL= total number of NLTE levels
rate equation in matrixA notation =
= L
n b
n n n n
Stellar Atmospheres: Non-LTE Rate Equations
26
1 1
2 2
charge c
ab
ion
undance def. 1 1 1 1 1 1 1 1 1
abundance def. 1 1 1 1 1 1
ons
1
i
erv
on 2
ion 1
ion
H
1
2
t
− −
− −
x x
x x
x x
x x
x x
x x
x x
x x
x x
z z z
z z
y y
y
x
x
x
x
x
x
x
x
x
x
xx x
y z
11 11 12 12 21 21 22 2
1
2
2
25
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
n. 0 0 0 0 0
0
0
0
0
=
p ez z z z zq q q q q q q q
n
n
n
n n
LTE contributions
abundances
Ionization into excited states
14
Stellar Atmospheres: Non-LTE Rate Equations
27
Elements of rate matrix
For each ion l with NL(l) NLTE levels one obtains a sub-matrix with the following elements:
( )
( )
( ) ( )
*
*
m<i 1
lower left
upper right
diagonal >
− + < = − + >
+ + + =
∑ ∑
ji ji
iij ji ij
j
km
im mi im im
mi
R C j i
nA R C j i
n
nR C R C j i
n
1 ( ) 1 ( ) highest level in parent ion,
into which ion can ionize; does not have to be ( ) 1 !
= =
= +
K K Ki NL l j NL l k k
l NL l
Stellar Atmospheres: Non-LTE Rate Equations
28
Elements corresponding to abundance definition eq.
Are located in final row of the respective element:
1
1
1,
( 1)
( )
1 1 ( ) except of ground state of excited ions
1 ground state of excited ions
=
=
−= −
=
= …
= + Φ =
∑
∑∑
NION
l
NION
l
ij e l m
m LTE l
i NL l
j NL l
A n j
[ ] [ ]
( )
- ( ) 1
1 =
− = … −
− + Φ =
∑
k
k e m
m LTE H
y j NLALL NL H NLALL
y n j NLALL
15
Stellar Atmospheres: Non-LTE Rate Equations
29
Elements corresponding to charge conservation eq.
Are located in the very final row of rate matrix, i.e., in
1,
( 1)
( ) 1 , except of
ground state of excited ions
( ) ( 1) else −= −
=
= …
=+ − Φ∑
ij
l m
m LTE
i NLALL
q l j NLALL
Aq l q l
Note: the inhomogeneity vector b (right-hand side of statistical equations) contains zeros except for the very last element (i=NLALL): electron density ne (from charge conservation equation)
Stellar Atmospheres: Non-LTE Rate Equations
30
Solution by linearization
The equation system is a linear system for and can be solved if, are known. But: these quantities are in general unknown. Usually, only approximate solutions within an iterative process are known.
Let all these variables change by e.g. in order to fulfill energy conservation or hydrostatic equilibrium.
Response of populations on such changes:
Let with actual quantities
And with new quantities
Neglecting 2nd order terms, we have:
An b= n
, ,e vn T J
, ,e vn T Jδ δ δ
nδ
, ,e vn T J
An b=
( )( ) ( )A A n n b bδ δ δ+ + = +
An b n n A bδ δ δ− = − − +
16
Stellar Atmospheres: Non-LTE Rate Equations
31
Linearization of rate equations
Needed: expressions for:
One possibility:
If in addition to the variables are introduced as unknowns, then we have the
→ Method of Complete Linearization
Other possibility: eliminates from the equation system by expressing through the other variables :
As an approximation one uses
(and iterates for exact solution)
,A bδ δ
n , ,e kn T J
1
NF
e k
ke k
A A AA T n J
T n Jδ δ δ δ
=
∂ ∂ ∂= + +∂ ∂ ∂∑
kJ,en TkJ
( ), ,k eJ f n T n=~ ( , , )d d
k k eJ S n T n
Jνννν discretized in NF frequency points
Stellar Atmospheres: Non-LTE Rate Equations
32
Linearization of rate equations
Method of approximate Λ-operators (Accelerated Lambda Iteration)
1
1
NF
e k
ke k
NLALLk k k
k e j
je j
A A AA T n S
T n S
S S SS T n n
T n n
δ δ δ δ
δ δ δ δ
=
=
∂ ∂ ∂= + +∂ ∂ ∂
∂ ∂ ∂= + +∂ ∂ ∂
∑
∑
( )1
analogous, :
0, ,0,=
∂=∂ ∂∂
+ + =∂ ∂∑ Le e
e
NF
k
k k
b
bb bT S
Tb n
Sn
n
δ
δ δδ δδ
17
Stellar Atmospheres: Non-LTE Rate Equations
33
Linearization of rate equations
Linearized equation for response as answer on changes
Expressions show the complex coupling of all variables. A change in the radiation field and, hence, the source function at any frequency causes a change of populations of all levels, even if a particular level cannot absorb or emit a photon at that very frequency!
1
1
1 1
NFk
k k
NFk
e
ke k e
NLALL NFk
j
j k k j
A A SAn b nA T n n
T S T
A A Sn n n
n S n
A Sn n
S n
δ δ
δ
δ
=
=
= =
∂ ∂ ∂− = − + − − ∂ ∂ ∂
∂ ∂ ∂+ − − ∂ ∂ ∂
∂ ∂+ −
∂ ∂
∑
∑
∑ ∑
, ,e vn T Jδ δ δnδ
1
NFk
k k j
A Sn
S n=
∂ ∂∂ ∂∑
Stellar Atmospheres: Non-LTE Rate Equations
34
Linearization of rate equations
In order to solve the linearized rate equations we need to compute these derivatives:
All derivatives can be computed analytically!
Increases accuracy and stability of numerical solution. More details later.
k, , , with respect to A,b,S∂ ∂ ∂ ∂∂ ∂ ∂ ∂e j kn T n S
18
Stellar Atmospheres: Non-LTE Rate Equations
35
LTE or NLTE?
When do departures from LTE become important?
LTE is a good approximation, if:
1) Collisional rates dominate for all transitions
2) Jv =Bv is a good approximation at all frequencies
*
solution of rate equations LTE
because =
<< → = + ≈
→
ij ij ij ij ij ij
ij i
ji j
R C P R C C
C n
C n
*
solution of rate equations LTE
=
→
→ =
ij ji
i i
j j
R R
n n
n n
Stellar Atmospheres: Non-LTE Rate Equations
36
LTE or NLTE?
When do departures from LTE become important?
LTE is a bad approximation, if:
1) Collisional rates are small
2) Radiative rates are large
3) Mean free path of photons is larger than that of electrons
Example: pure hydrogen plasma
Departures from LTE occur, if temperatures are high and densities are low
~ / , ij e e ij
C n T n T C↓ ↑ ⇒ ↓
~ , >1 ↑ ⇒ ↑ij ij
R T T Rα α
3/ 23/ 2 / /
~ 1/ (density of neutral H)
Saha: ~ ~
,
− ∆ −∆
∆
→ ∆
↓ ↑ ⇒ ∆ ↑
H
E kT E kT
H e p
e p
e
z n
Tn n n T e z e
n n
n T z
19
Stellar Atmospheres: Non-LTE Rate Equations
37
LTE or NLTE?
Stellar Atmospheres: Non-LTE Rate Equations
38
LTE or NLTE?
20
Stellar Atmospheres: Non-LTE Rate Equations
39
LTE or NLTE?
DA white dwarf, Teff= 60000K, log g= 7.5
Stellar Atmospheres: Non-LTE Rate Equations
40
LTE or NLTE?
DA
O w
ith lo
g g
= 6
.5
DO
with
log g
= 7
.5
21
Stellar Atmospheres: Non-LTE Rate Equations
41
Summary: non-LTE Rate Equations
Stellar Atmospheres: Non-LTE Rate Equations
42
Complete rate equations
For each atomic level i of each ion, of each chemical elementwe have:
In detail:
0i j
j i j i
jj iin n PP≠ ≠
− =∑ ∑
( )
( )
( )
( )
*
*
0
i
j i
j i
j
j i
j
j
iji ij
j
ji j
ij ij
j
i
i
j i
i
i
j
R C
nR
nR C
n
Cn
n
C
n
Rn
>
<
>
<
+
+
+
+
+
−
−
=
∑
∑
∑
∑
rates out of i
rates into i
excitation and ionization
de-excitation and recombination
de-excitation and recombination
excitation and ionization
1
Stellar Atmospheres: Solution Strategies
1
Solution Strategies
Stellar Atmospheres: Solution Strategies
2
All equations
Radiation Transport Iv(z), Jv(z), Hv(z), Kv(z)
Energy Balance T(z)
Hydrostatic Equilibrium ne(z)
Saha-Boltzmann / Statistical Equilibrium nijk(z)
Huge system with coupling over depth (RT) and frequency (SE)
Complete Linearisation (Auer Mihalas 1969)
Separate in sub-problems
2
Stellar Atmospheres: Solution Strategies
3
RT: Short characteristic method
Olson & Kunasz, 1987, JQSRT 38, 325
Solution along rays passing through whole plane-parallel slab
max
maxmax
0
( , , ) ( , , )exp ( )exp
( , , ) (0, , ) exp ( ) exp
Solution on a discrete depth grid , 1, with boundi
dI v I v S
dI v I v S
i ND
τ
τ
τ
τ τ τ τ ττ µ τ µ τ
µ µ µ
τ τ τ ττ µ µ τ
µ µ µ
τ
+ +
− −
′ ′ − −′= − + −
′ ′−
′= − + −
=
∫
∫
1
max
ary conditions:
( , ) (0, , )
( , ) ( , , )ND
I v I v
I v I v
µ µ
µ τ µ
− −
+ +
=
=
Stellar Atmospheres: Solution Strategies
4
Short characteristic method
Rewrite with previous depth point as boundary condition for
the next interval:
( )
( )
( )
1
1 1
1
1
( , , ) ( , , ) exp ( , , )
( , , ) ( , , ) exp ( , , )
with
using a linear interpolation for the spatial variation of
the intergrals can be evaluated
i i i i
i i i i
i i
i
i
I v I v I S v
I v I v I S v
S
I
τ µ τ µ τ µ
τ µ τ µ τ µ
τ ττ
µ
+ + ++
− − −− −
−
−
±
= −∆ + ∆
= −∆ +∆
−∆ =
∆
1 1
as
i i i i i i iI S S Sα β γ± ± ± ±− +∆ = + +
3
Stellar Atmospheres: Solution Strategies
5
Short characteristic method
Out-going rays:
( )
1 1
1
/ / / /
/ / / /
( , , ) exp exp exp
, ( ) exp( ) , , ,
1 11
1
i i
i i
i ii
ii i
a a b a
i a
a b b a
i b
d dI S v S S
x g x x a b
ew e e e e
w e e e e
τ τ
τ τ
µ µ µ µ
µ µ µ
τ τ ττ τ τµ
µ µ µ µ µ
τττ τ
µ µ
β
γ
+ +
+
+
−∆+ − − −
+ − − −
′ ′ ′ ′ −∆ = − = −
′ ∆
= = − = = ∆ =
− ⇒ = = + − = + ∆ ∆
⇒ = = − − −∆
∫ ∫
( ) 1eeµ
−∆−∆ −
= − − ∆
Stellar Atmospheres: Solution Strategies
6
Short characteristic method
In-going rays:
( )
11
11
/ / / /
/ /
( , , ) exp exp exp
, ( ) exp( ) , , ,
1 1
1
i i
i i
i ii
ii i
b a b a
i a
b b b
i b
d dI S v S S
x g x x a b
ew e e e e e
w e e e
τ τ
τ τ
µ µ µ µ
µ µ
τ τ ττ τ τµ
µ µ µ µ µ
τττ τ
µ µ
α
β
−
−
−
−−
−∆−− −∆
−−
′ ′ ′ ′−∆ = − = −
′ ∆
= = = = ∆ =
− ⇒ = = − + − = − + ∆ ∆
⇒ = = −∆
∫ ∫
( )/ / 11
a ee
µ µ−∆−
− = − ∆
4
Stellar Atmospheres: Solution Strategies
7
Short characteristic method
Also possible: Parabolic instead of linear interpolation
Problem: Scattering
Requires iteration
1
1
1 , ( )
2e e e e e e
n J I dκ σ η κ κ µ µ−
= = = ∫
Stellar Atmospheres: Solution Strategies
8
Solution as boundary-value problem
Feautrier scheme
Radiation transfer equation along a ray:
Two differential equations for inbound and outbound rays
Definitions by Feautrier (1964):
( )( ) ( )
pp:
sp:
vv v
dII S
d
dtd
d
d dZ
ττ τ
τ
τ κµ
τ κ
±±± = −
=
= −
( )
( )
symmetric, intensity-like
antisymmetric, flux-l
1
2
k1
i e2
+ −
+ −
= +
= −
u I I
v I I
5
Stellar Atmospheres: Solution Strategies
9
Feautrier scheme
Addition and subtraction of both DEQs:
One DEQ of second order instead of two DEQ of first order
2
2
( )( ) ( )
( )( )
(1)
(2
( )( )
)
( )
= −
=
⇒ = −
v
v
dvu S
d
duv
d
d uu S
d
ττ τ
ττ
ττ
ττ τ
τ
Stellar Atmospheres: Solution Strategies
10
Feautrier scheme
Boundary conditions (pp-case)
Outer boundary ... with irradiation
Inner boundary
Schuster boundary-value problem
0
(2
(
)
0) 0 ( 0) ( 0)
( ) ( 0)
−
=
= = → = = =
⇒ = =
I u v
duu
d τ
τ τ τ
ττ
τ
max max
max
max
max max max
max
( ) ( ) ( )
(( )
) )2 (
+ + +
+
=
= = → + =
⇒ = −
I I u v I
duI u
d
τ τ
τ
τ τ
τ τ τ τ
ττ
τ
0
0
0
( 0)
( )( 0)
I I u v I
duu I
d τ
τ
ττ
τ
− − −
−
=
= = → − =
⇒ = = −
6
Stellar Atmospheres: Solution Strategies
11
Finite differencesApproximation of the derivatives by finite differences:
( )
1 2 1 2
2
2
1 2 1
1 1
1 1
discretization on a scale
at interfirst derivative
second
mediate points:
1
2
( ) ( ) and
(
derivative
)
:
)
(
+ −
+ +
+ −
+ −
− = −
= +
− −≈ ≈
− −
=
i i
i
i i i
i i i i
i i i i
d uu S
d
u u u udu du
d d
du
dd du
d d
τ τ
τ
ττ
τ τ τ
τ τ
τ τ τ τ τ τ
τ
τ
τ τ
( )
1 2 1 2
1 2 1 2
1 1
2
1 1
2
1 1
( )
( )
1
2
+ −
+ −
+ −
+ −
+ −
−
−
− −−
− −≈
−
i i
i
i i
i i i i
i i i i
i i
du
d
u u u u
du
d
τ τ
τ
τ
τ τ
τ τ
τ τ τ ττ
τ τ τ
Stellar Atmospheres: Solution Strategies
12
Finite differences
Approximation of the derivatives by finite differences:
( )
( )( )
( ) ( )
2
2
1 1
1 1
1 1
1 1
1
1 1 1
1
1 1 1
discretisation on a scale
, 2 11
2
, 2 1
1
2
1
2
1
i i i i
i i i ii i
i i
i i i i i i i
i i i i i
i i i i i
i
d uu S
d
u u u u
u S i ND
Au B u C u S i ND
A
C
B
ττ
τ τ τ τ
τ τ
τ τ τ τ
τ τ τ τ
+ −
+ −
+ −
− +
−
− + −
−
+ + −
− = −
− −−
− −⇒ − = = −
−
⇒ − + − = = −
= − −
= − − =
L
L
i iA C+ +
7
Stellar Atmospheres: Solution Strategies
13
Linear system of equations
Linear system for ui
Use Gauss-Jordan elimination for solution
11 1 1 1
2 2 2 2 2
1 1 1
2
1
*
0
0
=
ND ND
ND
ND ND
ND ND ND ND
B C u W
A B C u W
C u
S
S
S
S
W
A B u W
−− − −
− − − = −
−
Stellar Atmospheres: Solution Strategies
14
Upper diagonal matrix
1st step:
( ) ( )
1 11
2 22
1 11
1 1
1 1 1 1 1 1
1 1
1 1
1
1 0
0
1
1
2
ND NDND
ND ND
i i i i i i i i i
u WC
u WC
u WC
u W
i C B C W B W
i ND C B AC C W B AC
− −−
− −
− −
− −
−
− = −
= = =
= = − = −
%%
%%
%%%
% %
% % %%L ( )1i i iW AW −+ %
8
Stellar Atmospheres: Solution Strategies
15
Back-substitution
2nd step:
Solution fulfils differential equation as well as both boundary
conditions
Remark: for later generalization the matrix elements are
treated as matrices (non-commutative)
11 1
ND ND
i i i i
i ND u W
i ND u W C u +
= =
= − = +
%
%%L
Stellar Atmospheres: Solution Strategies
16
Complete Linearization
Auer & Mihalas 1969
Newton-Raphson method in 5n
Solution according to Feautrier scheme
Unknown variables:
Equations:
System of the form:
[ ]1 , 1 , , , ,Ti
i i ND
i
Ji ND
nψ ψ ψ ψ ψ
= = =
rr r r r
L L Lr
, 1, , , , 1, , ( ) 0 NF transfer equations
( ) 0 NL equations for SE
− +− + − − =
− =
r
rr ri k i k i k i k i k i k i k i
i i i
A J B J C J S n
P J n b
, ( ) 0 , 1if NF NLα ψ α= = +L
9
Stellar Atmospheres: Solution Strategies
17
Complete Linearization
Start approximation:
Now looking for a correction so that
Taylor series:
Linear system of equations for ND(NF+NL) unknowns
Converges towards correct solution
Many coefficients vanish
0
, ( ) 0if α ψ ≠
0
, ( ) 0 , if iα ψ δψ α+ = ∀
0
0
, ,
, ,0
, , ,
1 1 1, ,
0 ( ) ( )
( )
i i
ND NF NLi i
i i k i l
i k li k i l
f f
f ff J n
J n
α α
α αα
ψ
ψ ψ δψ
ψ δ δ= = =
= = +
∂ ∂ = + + +
∂ ∂ ∑ ∑ ∑ L
, , , i k i lJ nδ δ
Stellar Atmospheres: Solution Strategies
18
Complete Linearization - structureOnly neighbouring depth points (2nd order transfer equation
with tri-diagonal depth structure and diagonal statistical
equations):
Results in tri-diagonal block scheme (like Feautrier)
, , 1 1( ) ( , , )i i i i if fα αψ ψ ψ ψ− +=r r r
1 1
, 1 ,
1
,
0 0
0
0 0
0 0
0
0
0
0 0
i i i i i i i
i k i i k i
i i
i k
A B C L
A J B J
n n
C
δψ δψ δψ
δ δ
δ δ
− +
−
−
− + − =
− +
−
rr r r
O O r r
O O
r r
O
O1
0
,
1
( )
i
i
i
J
f
n
α
δ
ψ
δ
+
+
= −
r
r
10
Stellar Atmospheres: Solution Strategies
19
Complete Linearization - structure
Transfer equations: coupling of Ji-1,k , Ji,k , and Ji+1,k at the
same frequency point:
→Upper left quadrants of Ai, Bi, Ci describe 2nd derivative
Source function is local:
→ Upper right quadrants of Ai, Ci vanish
Statistical equations are local
→ Lower right and lower left quadrants of Ai, Ci vanish
2
2
d J
dτ
Stellar Atmospheres: Solution Strategies
20
Complete Linearization - structureMatrix Bi:
( )( )
,
,
,
,
, ,1 ,
0
0
i k
i k
i l
i
NLi l m
i m i l lm i k
SB
n
B
Pn P
J
′
′= ′
∂ − ∂
= ∂
∂
∑
O M
L L
O M
M M
L L L L
M M
1 ... NF 1 ... NL
1 ... N
F1 ... N
L
11
Stellar Atmospheres: Solution Strategies
21
Complete Linearization
Alternative (recommended by Mihalas): solve SE first and
linearize afterwards:
Newton-Raphson method:
• Converges towards correct solution
• Limited convergence radius
• In principle quadratic convergence, however, not achieved
because variable Eddington factors and τ-scale are fixed
during iteration step
• CPU~ND (NF+NL)3 → simple model atoms only
– Rybicki scheme is no relief since statistical equilibrium not as
simple as scattering integral
1( ) 0 ( )i i i i i iP J n b n P J b−− = → =r rr rr r
Stellar Atmospheres: Solution Strategies
22
Energy Balance
→ Including radiative equilibrium into solution of radiative
transfer → Complete Linearization for model atmospheres
→Separate solution via temperature correction
+ Quite simple implementation
+ Application within an iteration scheme allows completely linear
system → next chapter
− No direct coupling
− Moderate convergence properties
12
Stellar Atmospheres: Solution Strategies
23
Temperature correction – basic scheme
0. start approximation for
1. formal solution
2. correction
3. convergence?
Several possibilities for step 2 based on radiative equilibrium
or flux conservation
Generalization to non-LTE not straightforward
With additional equations towards full model atmospheres:
• Hydrostatic equilibrium
• Statistical equilibrium
0( ) ( )T Tτ τ←
( )v v v
J S T= Λ
( ) ( ) ( )T T Tτ τ τ← +∆
Stellar Atmospheres: Solution Strategies
24
LTE
Strict LTE
Scattering
Simple correction from radiative equilibrium:
( ) ( ( ))v vS B Tτ τ=
( ) (1 ) ( ( )) ( )v e v e vS B T Jτ β τ β τ= − +
( )
[ ]( )
( )
0
0
( )0
( )0 0
( , ) ( , ) ( ( ), ) 0
( , ) ( , ) ) 0
0
( )(
∞
=
∞
∆
=
∞
==
∞ ∞
== =
− ≠
→ − + =
∂⇒ − − =
∆
∆ ∂
∂⇒ = −
∂∆
∫
∫
∫
∫ ∫
v v
v
v vT
v
v
v v
T Tv
vv v
T Tv v
v J v B T v dv
v J v B T dv
BJ B dv
T
BJ B dv dv
T
T
T
T
τ
τ
κ τ τ τ
κ τ τ τ
κ
κ
τ
κ
13
Stellar Atmospheres: Solution Strategies
25
LTE
Problem:
Gray opacity (κ independent of frequency):
deviation from constant flux provides temperature correction
( )( )0 0
independent of the temperature 0
vv v
T Tv v
v v
BT J B dv dv
T
J B T
τ
τ
κ κ∞ ∞
== =
→∞
∂∆ = −
∂
→ ⇒ ∆ →
∫ ∫
( )0
0.Moment equation
( ) ( )
( ) 0
( )
v v
v
v J B dv J B
J B B
J B B
dHB
dt
κ κ
κ
κ κ
κ
∞
=
− → −
→ − −∆ =
→ − = ∆
→ = ∆
∫
Stellar Atmospheres: Solution Strategies
26
Unsöld-Lucy correction
Unsöld (1955) for gray LTE atmospheres, generalized by
Lucy (1964) for non-gray LTE atmospheres
( )
0 0
0
0-th moment
1st m
:
, , ,
:
,
now new quantities , , fulfill
omen
ing ra
t
diat
∞ ∞
= =
∞
=
= −
→ = − = = =
=
→ = =
′ ′ ′
∫ ∫ ∫
∫ ∫
L
L
vv v v
JB v v J v v B
B v v
vv v
HH v v
B v
dHJ B
dt
dHdv J B B B dv J J dv d dt
d
dKH
dt
dKdv H H H dv
d
J H K
κ
κκ κ κ κ τ κ
τ κ
κ
κκ κ
τ κ
4
eff
radiative equi
ive equilibrium (local) and
flux conservation (non local)
: librium
flux conservation
0
: 4
′′ ′= − =
′′= =
J
B
H H
B B
dHJ B
d
dKH T
d
κ
τ κ
κ κ σ
τ κ κ π
14
Stellar Atmospheres: Solution Strategies
27
Unsöld-Lucy correction
Now corrections to obtain new quantities:
0
0 0 0 0
0
integrate (0)
, (0) (0) (0)
(0)
(0) (0)
(0) (0
) 1
′=
∞ ∞ ∞ ∞
′=
′∆ = −
′→ ∆ = ∆ + ∆
= = = = = =
∆′→ ∆ = + ∆ =
∆→ ∆ = − + +
∆= ∆
∆= −∆
∆
∆∆
∫
∫ ∫ ∫ ∫
∫
HH
B
J
B
v v v v v v
H
B
B
J
B
X X X
K K Hd
K K dv f J dv fJ H H dv h J dv hJ
f H
d H
d
K Hd fh
f HB
fh
d KH
d
d H
fB
J
Jd
τ
τ
τ
τ
τ
κτ
κ
κ
κ
τ κ
τ
τκ
κ
κ
κ
κ
κ
0
3
0
3
0
4 (0) (0) 1
(0) (0) 1
4
′=
′=
′=
′∆
∆
′∆ = ∆ = − + + ∆
∆′∆ = + + ∆
′+
−
∫
∫
∫J
B
H
B
J H
B B
J H
B B
Hd
T dH dH
d d
f HB T Hd
fh f
f HT HdB
fh fJ
T
τ
τ
τ
τ
τ
τ
τκ
κ κστ
π κ κ
κ κπτ
κ κ
κ
κσ
τ τ
Stellar Atmospheres: Solution Strategies
28
Unsöld-Lucy correction
„Radiative equilibrium“ part good at small optical depths but
poor at large optical depths
„Flux conservation“ part good at large optical depths but poor
at small optical depths
Unsöld-Lucy scheme typically requires damping
Still problems with strong resonance lines, i.e. radiative
equilibrium term is dominated by few optically thick
frequencies
3
0
(0) (0) 1
4
J J H
B B B
f HT J B Hd
T fh f
τ
τ
κ κ κπτ
σ κ κ κ′=
∆′∆ = − + + ∆
∫
J B→
0dH
dτ→
15
Stellar Atmospheres: Solution Strategies
29
NLTE Model Atmospheres
Radiation Transport and Sattistical Equilibrium are very
closely coupled
Simple separation (Lamda Iteration) does not work
Complete Linearization
Accelerated Lambda Iteration
Stellar Atmospheres: Solution Strategies
30
Lambda Iteration
Split RT and SE+RE:
• Good: SE is linear (if a separate T-correction scheme is used)
• Bad: SE contain old values of n,T (in rate matrix A)
Disadvantage: not converging, this is a Lambda iteration!
( )0
( , )
( , )
( , , ) ( , , ) 0
new old
new
v v
J S n T
A J T n b
v n T J S v n T dvκ∞
= Λ
=
− =∫
RT formal solution
SE
RE
16
Stellar Atmospheres: Solution Strategies
31
Accelerated Lambda Iteration (ALI)
Again: split RT and SE+RE but now use ALI
• Good: SE contains new quantities n, T
• Bad: Non-Linear equations → linearization (but without RT)
Basic advantage over Lambda Iteration: ALI converges!
( )
* *
0
( , ) ( , ) ( , )
( , )
( , , ) ( , , ) 0
new old old old new new new old old old
newnew new
new new new new new
v v
J S n T S n T S n T
A J T n b
v n T J S v n T dvκ∞
= Λ +Λ −Λ
=
− =∫
RT
SE
RE
Stellar Atmospheres: Solution Strategies
32
Example: ALI working on Thomson scattering problem
amplification factor
( )
( )( ) ( )( )
( ) ( )( )( )
* *
* *
*
*
:=formal solution on
solve for
1
1
1 1
⇒ = Λ −Λ +Λ
= Λ −Λ +Λ
= −
= −
− +
− −Λ − =
= − Λ −
⇒
+
− +
new
new new
e e
new new new
e e
e e
old old
old old old
new old
FS
FS
FS
e
old new new
e e
new
e
S B J
S S
B J B
S
B J
B J
J S
J
J
J J J J
J
B
β β
β β
β
β
β
β
β
β
( )
( )
1* *
1*
1 subtract on both sides
1
−
−
= − Λ −Λ
⇒ − = − Λ −
FS old old
e e
new old FS old
e
J J J
J J J J
β β
β
Interpretation: iteration is driven by difference (JFS-Jold) but: this difference
is amplified, hence, iteration is accelerated.
Example: βe=0.99; at large optical depth Λ* almost 1 ĺ strong amplifaction
source function with scattering, problem: J unknownĺiterate
17
Stellar Atmospheres: Solution Strategies
33
What is a good ȁ*?
The choice of ȁ* is in principle irrelevant but in practice it decides about the success/failure of the iteration scheme.
First (useful) ȁ* (Werner & Husfeld 1985):
A few other, more elaborate suggestions until Olson & Kunasz (1987): Best ȁ* is the diagonal of the ȁ-matrix(ȁ-matrix is the numerical representation of the integral operator ȁ)
We therefore need an efficient method to calculate the elements of the ȁ-matrix (are essentially functions of ττττνννν ).Could compute directly elements representing the ȁ-integral operator, but too expensive (E1 functions). Instead: use solution method for transfer
equation in differential (not integral) form: short characteristics method
( )*( )
, ' ( ')0
>Λ =
≤v
v v
SS
τ τ γτ τ τ
τ γ
Stellar Atmospheres: Solution Strategies
34
Towards a linear scheme
ȁ* acts on S, which makes the equations non-linear in the
occupation numbers
• Idea of Rybicki & Hummer (1992): use J=ǻJ+Ȍ*Șnew instead
• Modify the rate equations slightly:
( )*
0 0
*3
2
0
*3
*
2
0
4 4 ( )
24
24 ( )
ij ij
ij i i v
ijiji j j v
j
i
i
j
ji
j
R n n J dv J dvhv hv
n hvR n n J dv
n hv c
n hvJ dv
n
n n
nhv
nc
σ σπ π η
σπ
σπ η
∞ ∞
∞
∞
= = Ψ +∆
= +
= Ψ +∆ +
∫ ∫
∫
∫
18
Stellar Atmospheres: Solution Strategies
35
Stellar Atmospheres
This was the contents of our lecture:
Radiation field
Radiation transfer
Emission and absorption
Energy balance and Radiative equilibrium
Hydrostatic equilibrium
Solution Strategies for Stellar atmosphere models
Stellar Atmospheres: Solution Strategies
36
Stellar Atmospheres
This was the contents of our lecture:
Radiation field
Radiation transfer
Emission and absorption
Radiative equilibrium
Hydrostatic equilibrium
Stellar atmosphere models
The End
19
Stellar Atmospheres: Solution Strategies
37
Stellar Atmospheres
This was the contents of our lecture:
Radiation field
Radiation transfer
Emission and absorption
Radiative equilibrium
Hydrostatic equilibrium
Stellar atmosphere models
The End
Thank you for
listening !
1
Stellar Atmospheres: Non-LTE Stellar Atmospheres
1
Stellar Atmospheres in Non-LTE
Stellar Atmospheres: Non-LTE Stellar Atmospheres
2
Stellar Atmospheres in Non-LTE
radiation transfer Iv or Jv
radiative equilibrium solve consistently (∗) T
hydrostatic equation N
Parameters: Teff , log g, yk
LTE: strict LTE
including scattering
Population numbers by Saha-Boltzmann equations
NLTE: rate equations
(1 )
v v
v v v
S B
S J Bρ ρ
=
= + −
0idnAn b
dt= → =
2
Stellar Atmospheres: Non-LTE Stellar Atmospheres
3
Solution Methods
(∗) is a non-linear system of equations, we look for the solution vector:
Solution principle: Newton-Raphson iteration
Solution methods:
1. Complete Linearization (Auer & Mihalas 1969)
2. Multi-frequency/multi-gray (Anderson 1985)
3. ALI method (Werner, Husfeld 1985, 1986)
All methods have in common: linearization and iteration
For that, it is necessary to invert matrices (Jacobi matrix) with rank = number of equations
Numerical limit: matrix inversion limits rank to the order of
~ 100
( )( ) ( )
11, , , , , , , ,
with
=
=
L LNF
dd
NL e v v
d d d
n n N T n J J
M c
ψ
ψ ψ ψ
d=depth index
Stellar Atmospheres: Non-LTE Stellar Atmospheres
4
Complete Linearization
Linearizes all equations
Enabled break-through for first calculation of NLTE models,
quite robust method
Depth coupling by radiation transfer
⇒ Feautrier scheme
Disadvantage: Capacity limit quickly reached
e.g. model atom with hydrogen and helium:
20 NLTE levels
80 frequency points
(5 for each spectral line, 2 for each bound-free edge)
Only rather rudimentary representation of plasma
⇒ Number of equations must be reduced
100 equations
3
Stellar Atmospheres: Non-LTE Stellar Atmospheres
5
Anderson‘s method
Does not linearize the transfer equation with respect to all
frequency points. First: grouping of frequency points in
energy blocks. Then: linearization of these quantities.
Number of blocks determines the dimension of the system of
equations.
In some sense related to multi-grid methods.
Very clever method, BUT: requires physical motivation for
grouping of frequencies. Must be done manually, quite
cumbersome, much experience and physical insight by
user necessary. Was essentially used by inventor himself,
is not used any more.
Stellar Atmospheres: Non-LTE Stellar Atmospheres
6
ALI method
Accelerated Lambda Iteration
Eliminates the explicit inclusion of the transfer equation into
the linearization scheme by using instead an implicit
approximate solution for Jv:
Lambda iteration:
ALI:
( )( )
( )
(
( 1)1
()
(
)
) ii
v
i
v
i
v
i
J n
n
S
A bJ
− −= Λ
=
( ) ( ) ( )( )
( )
( 1) (( 1) * ( ) * ( 1)( )
( )
( )
( )
(
1)
)
* ( ) ( 1)
( )
i i i
v v v
i i
i
v
i
v
i i
i
v
i
i
i
v
v
n nnS S S
S
J
bJA
n JJ
n
− −− −
−
= Λ + Λ − Λ
= Λ + ∆
=
i=iteration counter
4
Stellar Atmospheres: Non-LTE Stellar Atmospheres
7
ALI method
Advantage: number of frequency points no longer appears in
dimension of equation system to be linearized (but
calculation of derivatives of ην,κν w.r.t. source function)
No explicit depth coupling, i.e. local linearized equations for
every depth point
Starting solution
Calculate correction
from linearized equation
Improved solution
( )
( )
( )
1
1
1
, , , , ,
, , , , ,
dd
NL e
dd
NL e
d d d
d d d
d d d
n n N T n
n n N T n
M c
M c
ψ
δψ δ δ δ δ δ
δψ
δψ
ψ δψ ψ
−
=
=
=
=
+ →
L
L
1
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
1
Radiation Transport as Boundary-Value
Problem of Differential Equations
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
2
Solution with given source function
• Formal Solution, applications:
– Strict LTE,
– Step within iterative method
• Numerical integration, short characteristics method
• Algebraic equation, Feautrier method
( )v vS B T=
2
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
3
Solution by numerical integration
Emergent intensity, plane-parallel geometry
Depth grid, ND depth points
• Geometrical depth
• Optical depth
z
ϑϑϑϑ
( 0, ), cosvI τ µ µ ϑ= =
10t =
Lt
maxNDt t=
1,L L NDt = L
0
1 ,, ( ) ( , )Lt
L L
t
L ND v v t dtτ τ κ′=
= ′ ′= ∫L
F
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
4
Emergent intensity
Numerical integration:
Trapezoidal rule,
naive approach:
Not very smart, systematic
summation of approximation
errors
max
maxmax
0
( 0, ) ( ) exp ( )expd
I S I
τ
ν ν ν
τ
ττ ττ µ τ τ
µ µ µ+ +
′=
′ ′ ′= = − + −
∫ F
1
( 0, )ND
L L
L
I S wν τ µ+
=
= =∑
Laτ
µ= 1Lb
τ
µ+=
( )2
a b
a b
b aI S e S e− − −
= +
3
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
5
Proper integration weights
Including weight function in integration weights is smart if:
( )
Problem: ( ) ( )
Interval: ,
Integrand: ( )
Weight function: ( )
b
a
I f x g x dx
a b
f x
g x
= ∫
( ) is less curved than ( ) ( )
( ) and ( ) have handy antiderivatives
f x f x g x
g x x g x
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
6
Determination of proper weights
( )f x
x
r1− 0 1
a m b
∆
( )2
2
2
( ) ( ) ( ) ( )( ) , even and odd
2 2
b a
b am
x m r
dx dr
f b f a f b f af x r
∆ = −
+=
∆= +
∆=
+ −= +
( )( ) ( )
( ) ( )( )
( ) ( )( )
1
1
( ) ( ) ( ) ( ) ( ) ( )24
like wise:
( ) ( ) ( )
1( )
2 22
1( )
2 22
b
a r
e o
e
o
I f x g x dx f b f a r f b f a g m r dr
g x g r g r
g r g m r g m r
g r g m r g m r
=−
∆ ∆= = + + − +
= +
∆ ∆= + + −
∆ ∆= + − −
∫ ∫
4
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
7
Determination of proper weights
Integration over symmetric interval leaves only even
integrands:
( ) ( )
( ) ( )
( ) ( )
1 1
1 1
1 1
0 0
( ) ( ) ( ) ( ) ( ) ( )4
( ) ( ) ( ) ( ) ( ) ( )2
( ) ( ) G ( ) ( ) ( ) ( )2
( )2
( )2
e o
r r
e o
r r
a b
a
b
I f b f a g r dr f b f a rg r dr
f b f a g r dr f b f a rg r dr
f b f a f b f a H w f a w f b
w G H
w G H
=− =−
= =
∆= + + −
∆
= + + −
∆= + + − = +
∆⇒ = −
∆⇒ = +
∫ ∫
∫ ∫
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
8
Examples
,6
2 6
6
( )
2
a
b
g x
G m H
w
w m
x
m
∆= =
∆ ∆ = −
∆ ∆+
=
=
2
22
22
22
,12 3
2 12 3
2 12 3
( )
a
b
g
G m H m
w
x
m
w m m
x
m
∆ ∆= + =
∆ ∆ ∆= + −
∆ ∆ ∆
+
=
= +
( )
( )
( )
( )
1
1 2
(
1
1
)
a b
a b a b
a a b
a
b a
b
x
b
G e e
H e e
g x e
e e
w e e e
w e e e
− −
− − − −
−
− − −
− − −
= −∆
= − − + −
∆ ∆
= −
−∆
=
−∆
= + −
( )
( )
( )
( )
1
(
2
)
1
1
1
a b
a b a b
a a b
a
b a b
x
b
G e e
H e e e e
w e e e
w e e
g e
e
x
−
= − +∆
= + + −
∆ ∆
= − − −∆
= + −∆
=
5
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
9
Short characteristic method
Olson & Kunasz, 1987, JQSRT 38, 325
Solution along rays passing through whole plane-parallel slab
max
maxmax
0
( , , ) ( , , )exp ( )exp
( , , ) (0, , ) exp ( ) exp
Solution on a discrete depth grid , 1, with boundi
dI v I v S
dI v I v S
i ND
τ
τ
τ
τ τ τ τ ττ µ τ µ τ
µ µ µ
τ τ τ ττ µ µ τ
µ µ µ
τ
+ +
− −
′ ′ − −′= − + −
′ ′−
′= − + −
=
∫
∫
1
max
ary conditions:
( , ) (0, , )
( , ) ( , , )ND
I v I v
I v I v
µ µ
µ τ µ
− −
+ +
=
=
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
10
Short characteristic method
Rewrite with previous depth point as boundary condition for
the next interval:
( )
( )
( )
1
1 1
1
1
( , , ) ( , , ) exp ( , , )
( , , ) ( , , ) exp ( , , )
with
using a linear interpolation for the spatial variation of
the intergrals can be evaluated
i i i i
i i i i
i i
i
i
I v I v I S v
I v I v I S v
S
I
τ µ τ µ τ µ
τ µ τ µ τ µ
τ ττ
µ
+ + ++
− − −− −
−
−
±
= −∆ + ∆
= −∆ + ∆
−∆ =
∆
1 1
as
i i i i i i iI S S Sα β γ± ± ± ±− +∆ = + +
6
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
11
Short characteristic method
Out-going rays:
( )
1 1
1
/ / / /
/ / / /
( , , ) exp exp exp
, ( ) exp( ) , , ,
1 11
1
i i
i i
i ii
ii i
a a b a
i a
a b b a
i b
d dI S v S S
x g x x a b
ew e e e e
w e e e e
τ τ
τ τ
µ µ µ µ
µ µ µ
τ τ ττ τ τµ
µ µ µ µ µ
τττ τ
µ µ
β
γ
+ +
+
+
−∆+ − − −
+ − − −
′ ′ ′ ′ −∆ = − = −
′ ∆
= = − = = ∆ =
− ⇒ = = + − = + ∆ ∆
⇒ = = − − −∆
∫ ∫
( ) 1eeµ
−∆−∆ −
= − − ∆
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
12
Short characteristic method
In-going rays:
( )
11
11
/ / / /
/ /
( , , ) exp exp exp
, ( ) exp( ) , , ,
1 1
1
i i
i i
i ii
ii i
b a b a
i a
b b b
i b
d dI S v S S
x g x x a b
ew e e e e e
w e e e
τ τ
τ τ
µ µ µ µ
µ µ
τ τ ττ τ τµ
µ µ µ µ µ
τττ τ
µ µ
α
β
−
−
−
−−
−∆−− −∆
−−
′ ′ ′ ′−∆ = − = −
′ ∆
= = = = ∆ =
− ⇒ = = − + − = − + ∆ ∆
⇒ = = −∆
∫ ∫
( )/ / 11
a ee
µ µ−∆−
− = − ∆
7
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
13
Short characteristic method
Also possible: Parabolic instead of linear interpolation
Problem: Scattering
Requires iteration
1
1
1 , ( )
2e e e e e e
n J I dκ σ η κ κ µ µ−
= = = ∫
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
14
Determination of mean intensity and flux
Discrete angular points
Solution along each ray
Angular integration
Gauss integration with 3 points sufficient for pp-RT
Alternative: numerical integration of moment equation
Problem: numerical approximation of
10t =
Lt
maxNDt t=
, 1j
j NAµ = L
1
weights depend on ND
j L Lj j
L
I S w µ±
=
=∑
1 1
1 1 , H
2 2
NA NA
j j j j j
j j
J I w I wµ= =
= =∑ ∑
F2 ( )E τ
8
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
15
Spherical geometry
Observer
Z
P
core
r=1
Pj Zi
rl
Z1
r=Rmax
1
1 max
, 1 ,
0, , inters
Impact parameter
ecting the core
1
s
, ,+
= = +
=
= =
L
L
L
j
NC
NC NP
P j NP NP ND NC
P P
P P R
1 max
, 1
, 1
Radii
l
ND
r l ND
r R r
=
= =
L
( )
2
i
2
2 2
2 2
, 1 intersecting t
Z
he core
, 1 1- ,
, 1 1- , 2 1
points
= + − =
= + − = + =
= − − = + = + − −
L
L
L
i l j
i l j
i l j
Z r P i ND
Z r P l NP j i l
Z r P l NP j i NP j l
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
16
Spherical geometry
Numerical integration on this grid, e.g.:
Optical depth
Emergent intensity
Emergent flux
( )1
2 2iZ
i i j
Z
r Z P dZτ κ= = +∫
maxmax
01
( )i
j l i
i
I S e d S wτ
τ
ττ τ′+ −
′==
′ ′= =∑∫
max
01
1 1( )
2 2
NPR
j jP
j
H I P PdP I w+ + +
==
= = ∑∫ F
9
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
17
Solution as boundary-value problem
Feautrier scheme
Radiation transfer equation along a ray:
Two differential equations for inbound and outbound rays
Definitions by Feautrier (1964):
( )( ) ( )
pp:
sp:
vv v
dII S
d
dtd
d
d dZ
ττ τ
τ
τ κµ
τ κ
±±± = −
=
= −
( )
( )
symmetric, intensity-like
antisymmetric, flux-l
1
2
k1
i e2
+ −
+ −
= +
= −
u I I
v I I
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
18
Feautrier scheme
Addition and subtraction of both DEQs:
One DEQ of second order instead of two DEQ of first order
2
2
( )( ) ( )
( )( )
(1)
(2
( )( )
)
( )
= −
=
⇒ = −
v
v
dvu S
d
duv
d
d uu S
d
ττ τ
ττ
ττ
ττ τ
τ
10
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
19
Feautrier scheme
Boundary conditions (pp-case)
Outer boundary ... with irradiation
Inner boundary
Schuster boundary-value problem
0
(2
(
)
0) 0 ( 0) ( 0)
( ) ( 0)
−
=
= = → = = =
⇒ = =
I u v
duu
d τ
τ τ τ
ττ
τ
max max
max
max
max max max
max
( ) ( ) ( )
(( )
) )2 (
+ + +
+
=
= = → + =
⇒ = −
I I u v I
duI u
d
τ τ
τ
τ τ
τ τ τ τ
ττ
τ
0
0
0
( 0)
( )( 0)
I I u v I
duu I
d τ
τ
ττ
τ
− − −
−
=
= = → − =
⇒ = = −
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
20
Feautrier scheme
Boundary conditions (spherical)
Core rays:
Like pp-case
Non-core rays:
Restrict to one quadrant (symmetry) inner boundary at Z=0:
max
max max max( ) ( ) ( ) 0
( )0
I I v
du
d τ τ
τ τ τ τ τ
τ
τ
+ −
=
= = = → =
⇒ =
max 0τ τ τ= =
max 0τ τ τ= =
11
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
21
Finite differencesApproximation of the derivatives by finite differences:
( )
1 2 1 2
2
2
1 2 1
1 1
1 1
discretization on a scale
at interfirst derivative
second
mediate points:
1
2
( ) ( ) and
(
derivative
)
:
)
(
+ −
+ +
+ −
+ −
− = −
= +
− −≈ ≈
− −
=
i i
i
i i i
i i i i
i i i i
d uu S
d
u u u udu du
d d
du
dd du
d d
τ τ
τ
ττ
τ τ τ
τ τ
τ τ τ τ τ τ
τ
τ
τ τ
( )
1 2 1 2
1 2 1 2
1 1
2
1 1
2
1 1
( )
( )
1
2
+ −
+ −
+ −
+ −
+ −
−
−
− −−
− −≈
−
i i
i
i i
i i i i
i i i i
i i
du
d
u u u u
du
d
τ τ
τ
τ
τ τ
τ τ
τ τ τ ττ
τ τ τ
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
22
Finite differences
Approximation of the derivatives by finite differences:
( )
( )( )
( ) ( )
2
2
1 1
1 1
1 1
1 1
1
1 1 1
1
1 1 1
discretisation on a scale
, 2 11
2
, 2 1
1
2
1
2
1
i i i i
i i i ii i
i i
i i i i i i i
i i i i i
i i i i i
i
d uu S
d
u u u u
u S i ND
Au B u C u S i ND
A
C
B
ττ
τ τ τ τ
τ τ
τ τ τ τ
τ τ τ τ
+ −
+ −
+ −
− +
−
− + −
−
+ + −
− = −
− −−
− −⇒ − = = −
−
⇒ − + − = = −
= − −
= − − =
L
L
i iA C+ +
12
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
23
Discrete boundary conditions
Outer boundary: first order
[ ]
1
2 11 1
2 1
2 11
2 1
1 1 1 2
1
1 2 1 1 1
0
0
, 1
u uduu u
d
u uu
B u C u
C B C
ττ τ τ
τ τ
τ τ−
−= → =
−
−→ − =
−
⇒ − =
= − = +
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
24
Discrete boundary conditions
Numerically better is a second-order condition:
Taylor expansion of u(τ) around τ1:
( ) ( )
( ) ( ) ( )
( )
[ ] [ ] [ ]
1 1
2
2 1 2 1 2 1
2
2 1 2 1 2 1
1 1 21 1 1 1 1 2 12
2 1 2 1
2 1 2
1 2 1 1 2 1 2 1
2
2
1 11
1
2
1
2
boundary condit
2 2
2 ,
ion
1 2
D
2
EQ
− − −
= + − + −
= + − + −
−+ + = ⇒ − =
− −
= − = + − + −
−
u u
u u
u u
d
uu S
d u
d
u S
B u C u S
u
B
d
C
u
τ τ
τ τ τ τ
τ τ τ τ
τ τ τ τ
τ τ τ
τ
τ τ
τ
τ
13
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
25
Discrete boundary conditions
Inner boundary: first order
[ ]
max
1
1
1
1
*
1
1
1
pp or core rays( )
0 non-core rays 0
0 0
1 ,
ND NDND ND
ND ND
ND ND ND
ND ND
ND ND ND ND ND
ND
ND ND ND ND
ND
I u I uu udu
d
u Iu u
A u B u S
AA B
A
τ τ
τ
τ τ τ
τ τ
τ τ
+ +
−
= −
+
−
−
−
−
−
− −−= → =
−
−→ + =
− ⇒ − + =
+= − = * ,
0ND
IS
+=
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
26
Discrete boundary conditionsOuter boundary: second order
( ) ( )
( )
2
1 1 1
1 1
2
2
boundary condition
1
2
0 non-core
c
DE
r
Q
o e
− − −
+
− −
= + − + −
= − −
+
−
ND ND
ND ND ND ND ND ND
ND ND ND
D
D
N
N
du d u
d
u uI u
du u
τ τ
τ τ τ ττ τ
τ τ
( ) ( )
( )
[ ] [ ] [ ]
[ ]
2
1
* *112
1
2 2
1
1
1 1
1*
1
1
1
2
2
2 , 1 2
2
2
2
−
−−
−
− −
− −
− +
+
−
−
−
−
− −
−+ = ⇒ − =
−−
−
= − = +
−
+ − −
= + −
ND ND ND ND
ND NDND ND ND ND ND ND ND
ND ND
ND ND ND ND ND ND
ND ND ND
ND
N
ND ND
ND N
D
D
u S
u uu S B u A u S
A B
S S I
I u
τ τ
τ τ
τ τ τ τ
τ
τ
τ τ
τ
τ
14
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
27
Linear system of equations
Linear system for ui
Use Gauss-Jordan elimination for solution
11 1 1 1
2 2 2 2 2
1 1 1
2
1
*
0
0
=
ND ND
ND
ND ND
ND ND ND ND
B C u W
A B C u W
C u
S
S
S
S
W
A B u W
−− − −
− − − = −
−
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
28
Upper diagonal matrix
1st step:
( ) ( )
1 11
2 22
1 11
1 1
1 1 1 1 1 1
1 1
1 1
1
1 0
0
1
1
2
ND NDND
ND ND
i i i i i i i i i
u WC
u WC
u WC
u W
i C B C W B W
i ND C B AC C W B AC
− −−
− −
− −
− −
−
− = −
= = =
= = − = −
%%
%%
%%%
% %
% % %%L ( )1i i iW AW −+ %
15
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
29
Back-substitution
2nd step:
Solution fulfils differential equation as well as both boundary
conditions
Remark: for later generalization the matrix elements are
treated as matrices (non-commutative)
11 1
ND ND
i i i i
i ND u W
i ND u W C u +
= =
= − = +
%
%%L
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
30
Solution with linear dependent source function
Coherent scattering:
General form (complete redistribution)
Thomson scattering
Results in coupling of equations for all directions
1 1
1 0
1 , ( ) ( )
2v v v v vS J J I d u d
µ µ
α β µ µ µ µ=− =
= + = =∫ ∫
( )
,
1 ,
e e e e e
tot e e e e e e
e e e e e e
tot ee e e
e
n J
S S J
S S J
κ σ η κ
η η κ η κ κ κ κκ η
κ κ κ κ κ κ κ κ κ κ κ κ
κβ β β
κ κ
= =
+ + −⇒ = = + = +
+ + + + +
= − + =+
( ) ( )( )
[ ]21
2
0
,, ( ) , 1 ( )e e
d uu u d S
dµ
τ µτ µ β τ τ µ µ β τ
τ=
− − = −∫
16
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
31
Discretization
Generalization of Feautrier scheme to a block-matrix scheme:
Angular discretization:
Depth discretization as previously:
( )1
10
,NA
ji j
j
u d u wµ
τ µ µ==
≈∑∫
1 1
equations for vector
− +− + − =i i i i i i i
i
Au B u C u W
u
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
32
Discretization
( ) ( )( )
[ ]21
2
0
1 1
1
,, ( ) , 1 ( )
equations for vector
0 0
, , C
0 0
0
0
=
− +
− − = −
→ − + − =
= = =
=
∫
e e
i i i i i i i
i
i i
i j i i i i
NA i i
i
i i
i
d uu u d S
d
Au B u C u W
u
u A C
u u A A C
u A C
B
B B
B
µ
τ µτ µ β τ τ µ µ β τ
τ
( )
( )
( )
1
1
1 ( )
( ) , 1 ( )
1 ( )
− − = −
−
NA e i
e i e i
NA e i
w w i S
i W i S
w w i S
β
β β
β
Identical lines
pp-case
identical in all
depths
17
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
33
1;1,1 1;1, 1
1; ,1 1; , 1
2 2;1,1 2;1, 2
2 2; ,1 2; , 2
1;1,1 1;1, 1
1; ,1 1; , 1
;1,1 ;1,
; ,1 ; ,
NA
NA NA NA
NA
NA NA NA
ND ND NA ND
ND NA ND NA NA ND
ND ND ND NA
ND ND NA ND NA NA
B B C
B B C
A B B C
A B B C
B B C
B B C
A B B
A B B
− − −
− − −
− − − −
− −
−
−
−
−
1;1 1;1
1; 1;
2;1 2;1
2; 2;
1;1 1;1
1; 1;
;1 ;1
; ;
NA NA
NA NA
ND ND
ND NA ND NA
ND ND
ND NA ND NA
u W
u W
u W
u W
u W
u W
u W
u W
− −
− −
=
( )
( ) ( )
;
1
; 1
1 ( ) , 1 1
1 ( ) 2
i j e i
ND j e ND ND ND j
W i S i ND
W ND S I
β
β τ τ− +
−
= − = −
= − + −
L
Block matrix
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
34
Feautrier scheme
Thermal source function
– Decoupled equation for each direction
– Separate Feautrier scheme for each ray
– ND•5 multiplications
Thermal source function + coherent scattering (e.g. Thomson)
– Coupled equations for all directions
– Feautrier scheme in block matrix form
– ND•NA3 multiplications (matrix inversions)
Spherical geometry
– Radius dependent angular integration
– Block matrix size depends on radius
– ~ND4 multiplications
12 2
, 21 , 1 ( ) , ( ) 1-
j
j l
l
Pj JMAX l JMAX l NP l
rµ
= − = = +
L
18
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
35
Solution with line scattering
Two-level atom, complete redistribution
Photon conservation within a spectral line
Approximately realized for resonance lines
The frequency independent line source function ( ) is a
weighted mean of the mean intensity
ground state
1st excited state
( ) ( ) , ( ) ( )
photon conservation:
( ) ( ) ( )
( )
( ) (generali scattering intergralzed)
= =
=
=
= = +
∫ ∫
∫
∫
L L
v
Line Line
L L v v
Line
L v v L L
Line
v v v v
v dv v J v dv
J v dv
S J v dv S J
κ κ ϕ η η ϕ
η κ
η κ ϕ
ϕ α β
F
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
36
Solution with line scattering
(non-coherent scattering)
Transfer equation:
Each one equation for each angular and frequency point
Each equation contains intensities from all other points
⇒coupling of all transfer equations to an integro-differential-
equation
This system is linear with respect to the intensity
⇒solution with Gauss-Jordan elimination
1
1
( , , )( , , ) ( )
( )
( , , ) 1( , , ) ( ) ( , , )
( ) 2
vv L
vv v
Line
dI vI v S
d v
dI vI v v I v d d
d v µ
µ τµ µ τ τ
τ
µ τµ µ τ ϕ µ τ µ τ
τ
+
=−
= −
= − ∫ ∫
19
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
37
Solution with line scattering
Discretization:
Analogous solution as with coherent scattering
Scattering integral:
With appropriate weights including the line profile function
(e.g. Gauss function)
1 2
2
1
( , , )( , , ) ( ) ( ) ( , , ) ( )
( )Line
d u vu v v u v d dv
d vµ
µ τµ τ α τ ϕ µ τ µ β τ
τ
+
=−
− − =∫ ∫
0
01
( ) ( )
v v NF
L k k
kv
J J v v dv J wν
ϕ+∆
=−∆
′= →∑∫
1 1
NA NF
L jk j k
j k
J u w w= =
′→∑∑
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
38
Linear system of equations
Linear system for ui
Use Gauss-Jordan elimination for solution
1 1 1 1
2 2 2 2 2
1 1 1
0
0 ND ND ND
ND ND ND ND
B C u W
A B C u W
C u W
A B u W
− − −
− − − = −
−
20
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
39
Block matrix
,11
,
,
1 1 1
0 0
= , , C
0 0
0
( )
0
j ijk ijk
j k ijk ijki j j i i
j NF ijk ijkNA
ijk NF
ijki
ijk
u A Cu
u A Cu u u A
u A Cu
B w w w w
BB i
B
α
= = =
′ ′ = −
1
1 1 1 1
( )
, ( )
( )
1
NA NA NF
i
NF NA NA NF
ijk jk ijk
w w w w i
W i
w w w w w w w w i
B A C
β
β
β
′ ′ = ′ ′ ′ ′
= + +
L
no longer identical, depend on k but not on j
Identical lines
( )
1 1
1 1
1 11 1
1
2
i i i i
NA NFi i i i
i i jk j k i
j ki i
u u u u
u u w wτ τ τ τ
α βτ τ
+ −
+ −
= =+ −
− −−
− −′− − =
−∑∑
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
40
Solution with line scattering
The modified Feautrier scheme is not very efficient
pp: inversion of matrices of rank NA•NF ~ 3 •5
ND•NA3 •NF3 multiplications
sp: inversion of matrices of rank NA•NF ~ 70 •5 \
ND•ND3 •NF3 multiplications
Repeated calculation of identical scattering integrals at each
frequency point
21
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
41
Rybicki scheme
Rybicki (1971)
Coherent scattering, e.g. Thomson scattering
Again: Feautrier transfer equation:
Discretization:
But now: additional variable
With additional equations
Use alternative grouping
( )1tot
v e v e vS S Jβ β= − +
( )( )
[ ]2
2
,, ( ) ( ) 1 ( ) ( )e e
d uu J S
d
τ µτ µ β τ τ β τ τ
τ− − = −
, 1, , , , 1,1
1 (pp) , 1 (sp)
(1),2 1, ( )
, , like in usual Feautrier scheme
e e
i j i j i j i j i j i j i i i i
i i i
A u B u C u J S
j NA j JMAX
i ND ND
A B C
β β− + − + − − = −
= =
= −
L L
L
FiJ
, ,
j
=i i j i j
J u w∑
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
42
Rybicki scheme( )
( )
( )
1 11
1
1
e
e
j i i
e
ND ND
S
K S
S
β
β
β
−
= − −
r
,
,
,
0
0
i j
i jj
ND j
w
wW
w
=
1,
,
,
j
i jj
ND j
u
uu
u
=
r
11 1 1
1
1
0
0
1 0 0
0 1 0
jj j j
NANA NA NA
j NA
ND
uT U K
uT U K
uT U K
J
W W W
J
=
− −
rr
rr
rr
1, 1, 1
, , ,
, ,
2
2
0 0
0 0
represents , J describes for the ray J
Scattering integral is repre
− − − −= = −
− −
− − → + =
r r rr r
ej j
ei j i j i jj j i
eND j ND j ND
j j j e j j j j
B C
A B CT U
A B
d uT u u U J j T u U K
d
β
β
β
βτ
1
sented by last block line 0=
− =∑rrNA
j j
j
W u J
22
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
43
Rybicki scheme
1st step: diagonal elements to unit matrixes
Corresponds to the solution of NA tri-diagonal systems of
equations of rank ND → cpu-time ~ NA•ND2
111
1
1
1 1
1
0
1
0
1
1 00
0 10
,
jj j
NANANA
j NA
ND
j j j j j j
KuU
uU K
uUK
J
W W W
J
U T U K T K− −
= − −
= =
rr %%
rr% %
r r%
%
r r% %
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
44
Rybicki scheme
2nd step: Wj to zero
Corresponds to cpu-time ~ NA•ND2
111
1 1
1
0
1
0
1
0 0 0
1 ,
jj j
NANANA
NA NA
j j j j
j j
KuU
uU K
uU KJW
Q
W W U Q W K= =
=
= − − = −∑ ∑
rr %%
rr% %
r r% %r
r
rr% %
23
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
45
Rybicki scheme
3rd step: solve for J
Corresponds to cpu-time ~ ND3
Finished if only J is required
4th step: solve for Feautrier variables u:
1WJ Q J W Q−= → =r rr r
1 1
j j j j ju T K T U J− −= −
r r rr
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
46
Comparison Rybicki vs. Feautrier
Thomson scattering
Few angular points: take Feautrier
Many angular points: take Rybicki
C1 NP ND2+ C 2ND3C ND4Spherical
C1 NA ND2+ C 2ND3C NA3 NDPlane-parallel
RybickiFeautrier
24
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
47
Solution with line scattering
(non-coherent scattering)
Two-level atom or Compton scattering
Each one block line
describes transfer equation for each ray j,k
(direction and frequency dependent!)
→ Huge system of equations
L LS Jα β= +
Jjk jk jk jkT u U K+ =r rr
11 1 1
1
1
0
0
1 0 0
0 1 0
jkjk jk jk
NANFNANF NANF NANF
jk NANF
ND
uT U K
uT U K
uT U K
J
W W W
J
=
− −
rr
rr
rr
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
48
Solution with line scattering
(non-coherent scattering)
1
jk i
ND
K
β
β
β
=
r
1,
,
j
ijjk
ND j
u
uu
u
=
r
,
,
,
0
0
i j k
i j kjk
ND j k
w w
w wW
w w
′ ′= ′
1
0
0
jk i
ND
U
α
α
α
− = − −
( ) ( )i v iJ J v dvτ ϕ= ∫
11 1 1
1
1
0
0
1 0 0
0 1 0
jkjk jk jk
NANFNANF NANF NANF
jk NANF
ND
uT U K
uT U K
uT U K
J
W W W
J
=
− −
rr
rr
rr
1, 1,
, , ,
, ,
0
0
jk jk
i jk i jk i jkjk
ND jk ND jk
B C
A B CT
A B
− − −= −
25
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
49
Comparison Rybicki vs. Feautrier
Line scattering or non-coherent scattering, e.g. Compton
scattering
Few frequency points: take Feautrier or Rybicki
Many frequency points: take Rybicki
Spherical symmetry: take Rybicki
C1 NP NF ND2+ C 2ND3C NF3 ND4Spherical
C1 NA NF ND2+ C 2ND3C NA3 NF3 NDPlane-parallel
RybickiFeautrier
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
50
Variable Eddington factors
0-th moment
1st moment
2nd moment
Eddington factor:
0-th moment of RT (pp)
1st moment of RT (pp)
1 1
1 0
1( ) ( )
2J I d u d
µ µ
µ µ µ µ+
=− =
= =∫ ∫1 1
1 0
1( ) ( )
2H I d v d
µ µ
µ µ µ µ µ µ+
=− =
= =∫ ∫1 1
2 2
1 0
1( ) ( )
2K I d u d
µ µ
µ µ µ µ µ µ+
=− =
= =∫ ∫
/f K J= F
( , )( , ) ( , )
( )
vv v
dH vJ v S v
d v
ττ τ
τ= −
( , )( , )
( )
vv
dK vH v
d v
ττ
τ=
26
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
51
Variable Eddington factors
Plane-parallel
With variable Eddington factor
With given f and S 2nd-order DEQ for J
Outer boundary:
2
2
( , )( , ) ( , )
( )
vv v
d K vJ v S v
d v
ττ τ
τ= −
[ ]2
2
( , ) ( , )( , ) ( , )
( )
v
v v
d f v J vJ v S v
d v
τ ττ τ
τ= −
[ ]
1 1
0 0
0
( 0) ( 0) / ( 0)
( 0) ( 0) ( 0)
( 0) ( ) ( )
( , ) ( , )( 0) ( , 0)
( )
v
v
h H J
I u v
h u d u d
d f v J vh J v
d v
µ µ
τ
τ τ τ
τ τ τ
τ µ µ µ µ µ
τ ττ τ
τ
−
= =
=
= = = =
= → = = =
→ = =
= = =
∫ ∫
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
52
Variable Eddington factors
Inner boundary:
2nd-order boundary conditions from Taylor series
[ ]
max
max max max
1 1
max
0 0
!
1 1 1
max
0 0 0
max max max
( ) ( ) / ( )
( ) ( ) ( )
: ( ) ( ) ( )
( , ) ( , )( ) ( ) ( , )
( )
H J
v
v v
h H J
h u d u d
H v d I d u d H hJ
d f v J vH h J v
d v
µ µ
µ µ µ
τ τ
τ τ τ τ τ τ
τ τ µ µ µ µ µ
τ τ µ µ µ µ µ µ µ µ µ
τ ττ τ τ τ τ τ
τ
= =
≠ =
+ +
= = =
+
=
= = = =
= =
= = = − = −
= = − = =
∫ ∫
∫ ∫ ∫
1442443 14243
F
27
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
53
Solution
Discretization → algebraic equation
Tri-diagonal system, solution analogous to Feautrier scheme
Thomson scattering:
Possible without extra costs
[ ]2
2
1 1
( , ) ( , )( , ) ( , )
( )
v
v v
i i i i i i i
d f v J vJ v S v
d v
A J B J C J S
τ ττ τ
τ
− +
− =
→ − + − =
F
[ ]2
2
1 1
(1 )
( , ) ( , )(1 ) ( , ) (1 ) ( , )
( )
( ) (1 )
tot
e e
v
e v e v
e e
i i i i i i i i i
S S J
d f v J vJ v S v
d v
A J B J C J S
β β
τ τβ τ β τ
τ
β β− +
= − +
→ − − = −
→ − + − − = −
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
54
Variable Eddington factors
But the Eddington factors are unknown → iteration
1. Formal solution u with S=B
2. Start value
3. Solution of the moment equation for J→S tot
4. Formal solution u with given S → K
5. New Eddington factors f
6. Converged?
1 3if ≡
, ,
1 1
2
contains , contains
NA NA
i i j j i j j
j j
j j
f u w u w
w wµ µ
= =
′=
′
∑ ∑
28
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
55
Variable Eddington factors
0-th moment
1st moment
2nd moment
0-th moment of RT (sp)
1st moment of RT (sp)
1 1
2
1 0
1( ) ( ) ,
2J I d u d J r J
µ µ
µ µ µ µ+
=− =
= = =∫ ∫ %
1 1
2
1 0
1( ) ( ) ,
2H I d v d H r H
µ µ
µ µ µ µ µ µ+
=− =
= = =∫ ∫ %
1 1
2 2 2
1 0
1( ) ( ) ,
2K I d u d K r K
µ µ
µ µ µ µ µ µ+
=− =
= = =∫ ∫ %
2
( , )( , ) ( , ) ( , )v
v v
r S
dH v rv r S v r J v r
drκ
=
= −
%% %123
[ ]( , ) 1
3 ( , ) ( , ) ( , ) ( , )vv v v
dK v rK v r J v r v r H v r
dr rκ+ − = −
F
F
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
56
Variable Eddington factors
1st moment of RT (sp)
Eddington factor
Introduction of the sphericality factor (Auer 1971),
which corresponds to the integrating factor for the DEQ
[ ]( , ) 1
3 ( , ) ( , ) ( , ) ( , )vv v v
dK v rK v r J v r v r H v r
dr rκ+ − = −
[ ]( )
3 1d fJ J
f Hdr r
κ+ − = −
F
( )2
2 2
1
( )3 ( ) 1 3 ( ) 1( ) exp , ( )
( ) ( )
r d r q rf r f rr q r dr r q r
r f r dr rf r
′ − −′= =
′ ′ ∫
( )( ) ( ) , ( ) exp ( )y f x y g x M x f x dx′ + = = ∫
29
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
57
Variable Eddington factors
1st moment equation:
( ) ( )
( )
2 2
2
2 2
( ) ( )( )
3 1 ( )( ) ( )
d r q r K d r q r dKK r q r
dr dr dr
f d fJr q r fJ r q r H
rf drκ
= +
−= + = −
( ) ( )
( )
( )( )
( )
2
2
2
2
( )
( )
1
1
dx q dr
d qfJd r q r KH H
r q r dr q dr
dHJ S
d qfJdx qJ S
dx qd qfJH
dx
κ
κκ
=−
= − → = −
→
= −
= −
=
%%
%%%
%%%
%%
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
58
Solution
Iteration of Eddington factors like in pp case
Additional integration of sphericality factor
f=1/3 is a bad starting point, f→1 for r →∞
Computational demand much smaller than for formal solution
2 3 1exp , include weights for
NDl
i i l l
l i l
f drq r w w
f r
−
=
−=
∑
30
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
59
Non-coherent scattering and moment equation
Two-level atom or Compton scattering
For each frequency point one moment equation of 2nd order
for mean intensity and Eddington factor
Coupled by frequency integral
pp
sp
LS Jα β= +
( ) , ( , )v k ik i k
J v f vτ
1
( )NF
v k k
k
J J v dv J J wϕ=
= → =∑∫
( )2
2
( , ) ( , )( , ) ( ) ( )
( )
v
v vv
d f v J vJ v J v dv
d v
τ ττ α ϕ β τ
τ− − =∫
( )2
2
( , ) ( , ) ( , )( , ) ( , ) ( )
( )
( )
v
v vv
d q r v f r v J r vJ r v q r v J v dv
dx v
r
α ϕ
β
− −
=
∫%
% %
%
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
60
Non-coherent scattering and moment equation
Feautrier:
cpu-time~ND•NF3
Rybicki:
cpu-time~NF•ND2+ND3
[ ]1
1 1
2
21
, , , , , 1
( )
T
i i ik iNF
i i i i i i i
NF
i ik k
k
J J J J i ND
AJ B J C J
d fJJ J w
d
β
τ
− +
=
= =
− + − =
=∑
rL L L
rr r r
[ ]1
1
1
, , , , , 1
, , , ,
, 0
T
k k ik NDk
T
i ND
NF
k k k k k k
k
J J J J k NF
J J J J
T J U J K W J J=
= =
=
+ = − =∑
rL L L
L L
r rr r
31
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
61
Multi-level atom
Atmospheric structure assumed to be given
(or accounted for by iteration)
Two sets of equations:
Radiative transfer equation for mean intensity
Statistical equilibrium
Both equations are coupled via radiative rates in matrix P
( )2
2
sum over all bb-, bf-, and ff-transitions
( , ) ( , )( , ) ( , )
( )
( , ) ( )
− =
=∑ ∑
v
v v
lu lu
v v
lu lu
d f v J vJ v S v
d v
S v v
τ ττ τ
τ
τ η κ
[ ]1( ) , T
v NLP J n b n n n= =rr r
L
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
62
Multi-level atomCoupling of J twofold:
• Over frequency via SE
• Over depth via RT
→ Simultaneous solution
→ non-linear in J
Lambda Iteration:
0. start approximation for n
1. formal solution with given S
2. solution of statistical equilibrium with given J
3. converged?
Not convergent for large optical depths F
32
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
63
Complete Linearization
Auer & Mihalas 1969
Newton-Raphson method in 5n
Solution according to Feautrier scheme
Unknown variables:
Equations:
System of the form:
[ ]1 , 1 , , , ,Ti
i i ND
i
Ji ND
nψ ψ ψ ψ ψ
= = =
rr r r r
L L Lr
, 1, , , , 1, , ( ) 0 NF transfer equations
( ) 0 NL equations for SE
− +− + − − =
− =
r
rr ri k i k i k i k i k i k i k i
i i i
A J B J C J S n
P J n b
, ( ) 0 , 1if NF NLα ψ α= = +L
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
64
Complete Linearization
Start approximation:
Now looking for a correction so that
Taylor series:
Linear system of equations for ND(NF+NL) unknowns
Converges towards correct solution
Many coefficients vanish
0
, ( ) 0if α ψ ≠
0
, ( ) 0 , if iα ψ δψ α+ = ∀
0
0
, ,
, ,0
, , ,
1 1 1, ,
0 ( ) ( )
( )
i i
ND NF NLi i
i i k i l
i k li k i l
f f
f ff J n
J n
α α
α αα
ψ
ψ ψ δψ
ψ δ δ= = =
= = +
∂ ∂ = + + +
∂ ∂ ∑ ∑ ∑ L
, , , i k i lJ nδ δ
33
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
65
Complete Linearization - structureOnly neighbouring depth points (2nd order transfer equation
with tri-diagonal depth structure and diagonal statistical
equations):
Results in tri-diagonal block scheme (like Feautrier)
, , 1 1( ) ( , , )i i i i if fα αψ ψ ψ ψ− +=r r r
1 1
, 1 ,
1
,
0 0
0
0 0
0 0
0
0
0
0 0
i i i i i i i
i k i i k i
i i
i k
A B C L
A J B J
n n
C
δψ δψ δψ
δ δ
δ δ
− +
−
−
− + − =
− +
−
rr r r
O O r r
O O
r r
O
O1
0
,
1
( )
i
i
i
J
f
n
α
δ
ψ
δ
+
+
= −
r
r
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
66
Complete Linearization - structure
Transfer equations: coupling of Ji-1,k , Ji,k , and Ji+1,k at the
same frequency point:
→ Upper left quadrants of Ai, Bi, Ci describe 2nd derivative
Source function is local:
→ Upper right quadrants of Ai, Ci vanish
Statistical equations are local
→ Lower right and lower left quadrants of Ai, Ci vanish
2
2
d J
dτ
34
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
67
Complete Linearization - structureMatrix Bi:
( )( )
,
,
,
,
, ,1 ,
0
0
i k
i k
i l
i
NLi l m
i m i l lm i k
SB
n
B
Pn P
J
′
′= ′
∂ − ∂
= ∂
∂
∑
O M
L L
O M
M M
L L L L
M M
1 ... NF 1 ... NL1 ... N
F1 ... N
L
Stellar Atmospheres: Radiation Transport as Boundary-Value Problem
68
Complete Linearization
Alternative (recommended by Mihalas): solve SE first and
linearize afterwards:
Newton-Raphson method:
• Converges towards correct solution
• Limited convergence radius
• In principle quadratic convergence, however, not achieved
because variable Eddington factors and τ-scale are fixed
during iteration step
• CPU~ND (NF+NL)3 → simple model atoms only
– Rybicki scheme is no relief since statistical equilibrium not as
simple as scattering integral
1( ) 0 ( )i i i i i iP J n b n P J b
−− = → =r rr rr r
1
Stellar Atmospheres: Temperature Correction Schemes
1
Temperature Correction Schemes
Stellar Atmospheres: Temperature Correction Schemes
2
Motivation
Up to now: Radiation transfer in a given atmospheric
structure
No coupling between radiation field and temperature included
→ Including radiative equilibrium into solution of radiative
transfer → Complete Linearization for model atmospheres
(next chapter)
→ Separate solution via temperature correction
+ Quite simple implementation
+ Application within an iteration scheme allows completely linear
system → next chapter
− No direct coupling
− Moderate convergence properties
2
Stellar Atmospheres: Temperature Correction Schemes
3
Temperature correction – basic scheme
0. start approximation for
1. formal solution
2. correction
3. convergence?
Several possibilities for step 2 based on radiative equilibrium
or flux conservation
Generalization to non-LTE not straightforward
With additional equations towards full model atmospheres:
• Hydrostatic equilibrium
• Statistical equilibrium
0( ) ( )T Tτ τ←
( )v v vJ S T= Λ
( ) ( ) ( )T T Tτ τ τ← + ∆
Stellar Atmospheres: Temperature Correction Schemes
4
LTE
Strict LTE
Scattering
Simple correction from radiative equilibrium:
( ) ( ( ))v vS B Tτ τ=
( ) (1 ) ( ( )) ( )v e v e vS B T Jτ β τ β τ= − +
( )
[ ]( )
( ) ( )
0
0
( )0
( )0 0 0 0
( , ) ( , ) ( ( ), ) 0
( , ) ( , ) ( ) 0( )
0
∞
=
∞
∆
=
∞
==
∞ ∞ ∞ ∞
== = = =
− ≠
→ − + =
∂⇒ − − = ∂
∂⇒ = − = − ∆
∂
∆
∆
∆
∫
∫
∫
∫ ∫ ∫ ∫
v v
v
v vT
v
v
v v
T Tv
vv v v v v
T Tv v v v
v J v B T v dv
v J v B T dv
BJ B dv
T
BJ B dv
T
T
T dv J B dv B dvT
τ
τ
κ τ τ τ
κ τ τ τ
κ κ
τ
κ
κ κ
3
Stellar Atmospheres: Temperature Correction Schemes
5
LTE
Strict LTE
Scattering
Simple correction from radiative equilibrium:
( ) ( ( ))v vS B Tτ τ=
( ) (1 ) ( ( )) ( )v e v e vS B T Jτ β τ β τ= − +
( )
[ ]( )
( )
0
0
( )0
( )0 0
( , ) ( , ) ( ( ), ) 0
( , ) ( , ) ) 0
0
( )(
∞
=
∞
∆
=
∞
==
∞ ∞
== =
− ≠
→ − + =
∂⇒ − − =
∆
∆ ∂
∂⇒ = −
∂∆
∫
∫
∫
∫ ∫
v v
v
v vT
v
v
v v
T Tv
vv v
T Tv v
v J v B T v dv
v J v B T dv
BJ B dv
T
BJ B dv dv
T
T
T
T
τ
τ
κ τ τ τ
κ τ τ τ
κ
κ
τ
κ
Stellar Atmospheres: Temperature Correction Schemes
6
LTE
Problem:
Gray opacity (κ independent of frequency):
deviation from constant flux provides temperature correction
( )( )0 0
independent of the temperature 0
vv v
T Tv v
v v
BT J B dv dv
T
J B T
τ
τ
κ κ∞ ∞
== =
→∞
∂∆ = −
∂
→ ⇒ ∆ →
∫ ∫
( )0
0.Moment equation
( ) ( )
( ) 0
( )
v v
v
v J B dv J B
J B B
J B B
dHB
dt
κ κ
κ
κ κ
κ
∞
=
− → −
→ − − ∆ =
→ − = ∆
→ = ∆
∫
4
Stellar Atmospheres: Temperature Correction Schemes
7
Unsöld-Lucy correction
Unsöld (1955) for gray LTE atmospheres, generalized by
Lucy (1964) for non-gray LTE atmospheres
( )
0 0
0
0-th moment
1st m
:
, , ,
:
,
now new quantities , , fulfill
omen
ing ra
t
diat
∞ ∞
= =
∞
=
= −
→ = − = = =
=
→ = =
′ ′ ′
∫ ∫ ∫
∫ ∫
L
L
vv v v
JB v v J v v B
B v v
vv v
HH v v
B v
dHJ B
dt
dHdv J B B B dv J J dv d dt
d
dKH
dt
dKdv H H H dv
d
J H K
κ
κκ κ κ κ τ κ
τ κ
κ
κκ κ
τ κ
4
eff
radiative equi
ive equilibrium (local) and
flux conservation (non local)
: librium
flux conservation
0
: 4
′′ ′= − =
′′= =
J
B
H H
B B
dHJ B
d
dKH T
d
κ
τ κ
κ κ σ
τ κ κ π
Stellar Atmospheres: Temperature Correction Schemes
8
Unsöld-Lucy correction
Now corrections to obtain new quantities:
0
0 0 0 0
0
integrate (0)
, (0) (0) (0)
(0)
(0) (0)
(0) (0
) 1
′=
∞ ∞ ∞ ∞
′=
′∆ = −
′→ ∆ = ∆ + ∆
= = = = = =
∆′→ ∆ = + ∆ =
∆→ ∆ = − + +
∆= ∆
∆= − ∆
∆
∆∆
∫
∫ ∫ ∫ ∫
∫
HH
B
J
B
v v v v v v
H
B
B
J
B
X X X
K K Hd
K K dv f J dv fJ H H dv h J dv hJ
f H
d H
d
K Hd fh
f HB
fh
d KH
d
d H
fB
J
Jd
τ
τ
τ
τ
τ
κτ
κ
κ
κ
τ κ
τ
τκ
κ
κ
κ
κ
κ
0
3
0
3
0
4 (0) (0) 1
(0) (0) 1
4
′=
′=
′=
′∆
∆
′∆ = ∆ = − + + ∆
∆′∆ = + + ∆
′+
−
∫
∫
∫J
B
H
B
J H
B B
J H
B B
Hd
T dH dH
d d
f HB T Hd
fh f
f HT HdB
fh fJ
T
τ
τ
τ
τ
τ
τ
τκ
κ κστ
π κ κ
κ κπτ
κ κ
κ
κσ
τ τ
5
Stellar Atmospheres: Temperature Correction Schemes
9
Unsöld-Lucy correction
„Radiative equilibrium“ part good at small optical depths but
poor at large optical depths
„Flux conservation“ part good at large optical depths but poor
at small optical depths
Unsöld-Lucy scheme typically requires damping
Still problems with strong resonance lines, i.e. radiative
equilibrium term is dominated by few optically thick
frequencies
3
0
(0) (0) 1
4
J J H
B B B
f HT J B Hd
T fh f
τ
τ
κ κ κπτ
σ κ κ κ′=
∆′∆ = − + + ∆
∫
J B→
0dH
dτ→
Stellar Atmospheres: Temperature Correction Schemes
10
Unsöld-Lucy correction
Generalization for scattering
All the rest is the same
Difficulties for scattering dominated regions: weak coupling
between radiation field and temperature
( ) ( )
0 0
:
(1 ) , (1 ) ,
(0-th moment
1 )
∞ ∞
= =
= =
→ = −
= − = −
− − − −
=
∫
∫ ∫
L
M
v
tvv v v v
J
B
B e v v J e v v
o
B
e v e
v
v
v
tdHJ J
dt
dHdv J B
d
B B dv J J dv d d
S B J
t
κ κ
κ
τ κ
κ β κ κ β κ τ κ
β β
0
1 0
0
B
e J
B
J
d H
d
κ
β κ
τ
∆ →→ ⇒ ∆ →
∆ →
6
Stellar Atmospheres: Temperature Correction Schemes
11
Unsöld-Lucy correction
Generalization to non-LTE (Werner & Dreizler 1998, Dreizler 2003)
All the rest is the same
should contain only terms which couple directly to the
temperature, i.e. bf and ff transitions
Depth dependent damping (need to play with parameters c):
( )
0 0
0 :
, ( ) ,
-th
moment
∞ ∞
= =
= =
→ = −
= = − =
− − −
∫
∫ ∫
%
M
%
L
v v v v v
v
vv
v
v v v
J
B
B v J v v B
v v
dHJ J
dt
dHdv J B
d
B B dv J J dv d dt
S B Jκ κ
κ
τ κ
κ
κ
τ
γ
κκ γκ κ
vκ%
0
0 0
3
0
// / 3
1 2
(1 )(1
()
0) (0)
4
−−
=
−
′
∆′∆ = − + + ∆
−− ∫J J H
B B B
f HT J B Hd
c ec e c
he
T f f
τ ττ τ
τ
τ
τ τκ κ κπτ
σ κ κ κ
Stellar Atmospheres: Temperature Correction Schemes
12
Stellar Atmospheres
This was the contents of our lecture:
Radiation field
Radiation transfer
Emission and absorption
Radiative equilibrium
Hydrostatic equilibrium
Stellar atmosphere models
7
Stellar Atmospheres: Temperature Correction Schemes
13
Stellar Atmospheres
This was the contents of our lecture:
Radiation field
Radiation transfer
Emission and absorption
Radiative equilibrium
Hydrostatic equilibrium
Stellar atmosphere models
The End
Stellar Atmospheres: Temperature Correction Schemes
14
Stellar Atmospheres
This was the contents of our lecture:
Radiation field
Radiation transfer
Emission and absorption
Radiative equilibrium
Hydrostatic equilibrium
Stellar atmosphere models
The End
Thank you for
listening !
8
Stellar Atmospheres: Temperature Correction Schemes
15
Avrett-Krook methodIn case that flux conservation and radiative equilibrium is not
fulfilled, Unsöld-Lucy can only change the temperature
Change of other quantities, e.g. opacity, is not accounted for
→ Avrett & Krook (1963)
strict LTE assumed, generalization straightforward
Current quantities:
Does not fulfill flux conservation and radiative equilibrium
New quantities:
( )
0
0 00 0 0 0 0
0 0( ( )) with some kind of mean opacity
v
v vv v
dII B T
dχ
κµ τ κ
τ κ=
= −
( )( ( )) with mean opacity
v
v vv v
dII B T
dχ
κµ τ κ
τ κ=
= −
Stellar Atmospheres: Temperature Correction Schemes
16
Avrett-Krook methodLinear Taylor expansion of the new quantities from old ones:
Radiative transfer equation:
10 1 0 1 0 1 0 1 4
eff0 0
0 1 0 1 0 1 0 1
0 0
1 ( ) 4v v v v v
v vv v v v v v v v
d dT T T I I I H H H dv T
d d
d dBB B B B T
d dT
τ ττ τ τ σ π
τ τ
χχ χ χ χ τ
τ
= + → = + = + = + = + =
= + = + = + = +
∫
( )
( )( )
( ) ( )( )
( ) ( )
( )
0 10 1 0 1 0 1
0 0 0
1 10 0 0 0 1 0 1 0 1
0 0
1 10 0 0 0 1 0 0 1 0 1
0 0
10 1 1
0
( ( ))
1
vv v v
v vv v v v v v
vv v v v v v v v v
vv v v v v v v v v v
vv v v
dII B T
d
dI dI dI I B B
d d d
dI dI B I I B B
d d
dI dI B I I B B
d d
dII B
d
µ χ ττ
τµ µ χ χ
τ τ τ
τχ µ χ χ
τ τ
τχ µ χ χ χ
τ τ
µ χ χτ
= −
+ = + + − −
− + = + + + − −
− + = + + + − −
= − + ( )1
1 0 0 0
0v v v v
dI B
d
τχ
τ
+ −
9
Stellar Atmospheres: Temperature Correction Schemes
17
Avrett-Krook method
1st moment:
( ) ( )1 1
0 1 1 1 0 0 0
0 0
1 1 10 1 1 0 0 0 1 1 0 0
0 0 0 0
010 1 1
0 0 0
010 1 0
ef0 0 0
0
: 0
4
vv v v v v v v
v vv v v v v v v v v
v vv v
v
v v
v
dI dI B I B d
d d
dK dd dH H H H
d d d d
d HdH H dv
d d
d HdH dv H T
d d
τµ χ χ χ µ µ
τ τ
χτ τχ χ χ χ τ χ
τ τ τ τ
χττ
τ τ χ
χτ στ
τ τ χ π
∞
= − + + −
→ = + + = + + =
→ + = −
→ + = −
∫
∫
∫
L
L
0
4
f
4 01 eff
0 0 0 0 0
0 0 0
, linear DEQ of first order
1 1( ) 1 , ( ) exp
( ) 4 ( )
x
v v
v
T d HM x dx M x dy dv
M H x d H
τ σ χτ
τ π χ τ
∞ ⇒ = − =
∫ ∫ ∫
F
Stellar Atmospheres: Temperature Correction Schemes
18
Avrett-Krook method
Outer boundary:1 1
4 41 4 0 0 0eff eff
eff 0 0
0
40 1eff
0
0 0
4 01 eff
0
11
0
(0) (0)
(0) (0) 1 (0) 1 (0)4 4 (0) 4 (0)
1 (0) (0)4 (0)
(0)(0) 1
4 (0)
: 0 (
v v v
v
v v v
vv
v
vv v
H h J dv
T TH T H H H dv
H H
TH dv h J dv
H
T HJ
H h
dKf J
d
σ σσ
π π π
σ
π
σ
π
τ
∞
∞ ∞
=
= − = − = −
→ − =
→ = −
= ⇒
∫
∫
∫ ∫
L
4 00 1 eff
0
4 01 0 eff
0
(0)) const (0) 1
4 (0)
(0)( ) 1
4 (0)
vv v
v
vv
v v
T Hf J
H h
T HJ
H f h
στ
π
στ
π
= = = −
⇒ = −
10
Stellar Atmospheres: Temperature Correction Schemes
19
Avrett-Krook method0-th moment:
( ) ( )
( ) ( )
( )
1 10 1 1 1 0 0 0
0 0
1 10 1 1 1 0 0 0
0 0
1 10 1 1 1 0 0 0
0 0 0 0
1 00 1 1 0
0 0
0
vv v v v v v v
vv v v v v v v
v v vv v v v v
vv v v
dI dI B I B d
d d
dH dJ B J B
d d
dH dB d dJ T J B dv
d dT d d
dBdH dHJ dv T
d d dT
τµ χ χ χ µ
τ τ
τχ χ χ
τ τ
χ τχ τ χ
τ τ τ
χ χτ τ
∞
= − + + −
→ = − + + −
→ = − + + −
→ = − = −
∫
∫
∫
L
L
( )
( )
1 01 0 0
0 0 0 0
0 0
1 0
0
0
4 01 01 0 0 0eff
0 0 0 0
0 0
1
(0)1 1
4 (0)
vv v
vv
v vv v v
v v
d d dHdv J B dv
d d d
dBT dv
dT
d T Hd dHJ B dv dv
d d d H f h
χ ττ
τ τ τ
χ
χ σττ χ
τ τ τ π
∞ ∞
∞
∞ ∞
+ − +
⇒ =
− + + + −
∫ ∫
∫
∫ ∫
Stellar Atmospheres: Temperature Correction Schemes
20
Radiative equilibrium and Complete Linearization
(LTE)Simultaneous solution of RT and RE radiation transfer:
( )
2
2
1, , , 1
0 0 0 0 0 0
0 0
( , )( , ) ( , ( )) 0
( )
( , ) 0 , , , , ( , , , , )
, ( , ) 0 correction , ( , ) 0
Taylor expansion: ( , )
vv v
ik k k k i k ND k i ND
k ik k k ik k k
ik k k i
d J vJ v B v T
d v
f J T J J J J T T T T
J T f J T J T f J J T T
f J J T T
ττ τ
τ
δ δ δ δ
δ δ
− − =
→ = = =
≠ ⇒ → + + =
+ +
=
r r r rL L L L
r r r r r r r r r r
r r
( )
( )
0 0
1 1
1 1
0 0
0 ( , )
: tri-diagonal with usual , ,
( , ): diagonal
( , )
ik ik ik ikik k i k ik i k i
i k ik i k i
k k k k
k ik ik ik
v k ik k ii
i
k ik ki
f f f ff J T J J J T
J J J T
T J U T K
T A B C
B v TU U
T
K f J T
δ δ δ δ
δ δ
− +
− +
∂ ∂ ∂ ∂= + + + +
∂ ∂ ∂ ∂
→ + =
− −
∂= −
∂
= −
r r
r r r
r r r
F
11
Stellar Atmospheres: Temperature Correction Schemes
21
Radiative equilibrium and Complete Linearization
(LTE)Simultaneous solution of RT and RE radiative equilibrium:
( )
( ) ( )
0
, 1 ,1 , ,
1
( , ) ( , ) ( , ) 0
, , , , , ( , ) 0
i v i v i
v
NF
i NF i i k i NF i k ik v k i
k
v J v B v T dv
f J J J T w J B v T
κ τ τ∞
=
+=
− =
→ = − =
∫
∑L L
( )
( )
( )
( )
0 0
, 11,
NF, 1 , 10 0
, 1
1 k=1
1
0
Taylor expansion: ,
0 ( , )
: diagonal
( , ): diagonal
i NF k k ik NF
NFi NF i NF
i NF k ik i k k
k ik i
k k kii
NFv k i
kiik i
k iki
f J J T T
f ff J T J T W J D T L
J T
W W w
B v TD D w
T
L w J B
δ δ
δ δ δ δ
+=
+ +
+=
′′
′=
+ +
∂ ∂= = + + → + =
∂ ∂
=
∂= −
∂
= − −
∑ ∑
∑
r r
r r r r r
r( )0
1
( , )NF
v k i
k
v T=∑
Stellar Atmospheres: Temperature Correction Schemes
22
Radiative equilibrium and Complete Linearization
(LTE)Together: Rybicki scheme:
RE takes the part of the scattering integral
Instead of solve for temperature corrections
Non-linear → iteration
During the iteration: new opacities, Eddington factors
1 1 1 1
1
0
0
k k k k
NF NF NF NF
k NF
T U J K
T U J K
T U J K
W W W D T L
δ
δ
δ
δ
=
r r
r r
r r
r r
J
12
Stellar Atmospheres: Temperature Correction Schemes
23
Radiative equilibrium and Complete Linearization
(NLTE)Direct generalization at least problematic due to weak
coupling of NLTE source function to the temperature
Take into account the change of the population numbers
→ Add RE to the Complete Linearization scheme
(Auer & Mihalas 1969)
Radiative equilibrium in NLTE:
Linearization:
F
( )
( ) ( )
0
, 1 ,1 , ,1 ,
1
( , ) ( , ) ( , ) 0
, , , , , , ( , ) ( , ) 0
i v i v i
v
NF
i NF NL i i NF i i NL i k k i ik v k i
k
v J v S v dv
f J J n n T w v J v
κ τ τ τ
κ τ η τ
∞
=
+ +=
− =
→ = − =
∫
∑L L
0
, 1
, 1 , 1 , 10
, 1
1 1
( )
0 ( )
i NF NL
NF NLi NF NL i NF NL i NF NL
i NF NL ik il i
k lik il i
f
f f ff J n T
J n T
ψ δψ
ψ δ δ δ
+ +
+ + + + + +
+ += =
+ =
∂ ∂ ∂= + + +
∂ ∂ ∂∑ ∑
r r
r
k kk ik
il il
w Jn n
κ η ∂ ∂−
∂ ∂
Stellar Atmospheres: Temperature Correction Schemes
24
Radiative equilibrium and Complete Linearization
(NLTE)Together with RT and SE: F
0
,
, , ,0
, , ,
1 1 1, ,
1( )
0 ( )
i
ND NF NLi i i
i i k i l i
i k li k i l i
NF NLf
f f ff J n T
J n T
α
α α αα
αψ δψ
ψ δ δ δ= = =
= ++ =
∂ ∂ ∂ = + + +
∂ ∂ ∂ ∑ ∑ ∑
L
1 1
1, ,
1
1
0 0
0
0 0
0 0
0 0
i i i i i i i
i ii k i k
i i
i i
A B C L
J JA B
n n
T T
δψ δψ δψ
δ δ
δ δ
δ δ
− +
−
−
−
− + − =
− +
rr r r
O O r r
O O
r r
1,
0
,
1
0
0
0
( )
0 0
0 0
ii k
i
i
i
JC
f
n
T
α
δ
ψ
δ
δ
+
+
− = −
O r
O
r
13
Stellar Atmospheres: Temperature Correction Schemes
25
Radiative equilibrium and Complete Linearization
(NLTE)Matrix Bi: F
( )( )
( )
, ,
,
,
, ,
, ,1 ,
1 1 1
, ,
0
0
i k i k
i k
i l i
iNL
i il m l m
i m i il lm i k i
iNF NL iNF NL iNF NL
i k i l i
S SB
n T
BP P
n P TJ T
f f f
J n T
′
′= ′
+ + + + + +
′ ′
∂ ∂ − − ∂ ∂ = ∂ ∂
∂ ∂
∂ ∂ ∂ ∂ ∂ ∂
∑
O M
L L
O M
M M
L L L L
M M
1 ... NF 1 ... NL T1 ... N
F1 ... N
L T
1
Stellar Atmospheres: Accelerated Lambda Iteration
1
Accelerated Lambda Iteration
Stellar Atmospheres: Accelerated Lambda Iteration
2
Motivation
Complete Linearization provides a solution scheme, solving
the radiation transfer, the statistical equilibrium and the
radiative equilibrium simultaneously.
But, the system is coupled over all depths (via RT) and all
frequencies (via SE, RE) → HUGE!
Abbreviations used in this chapter:
RT = Radiation Transfer equations
SE = Statistical Equilibrium equations
RE = Radiative Equilibrium equation
F
2
Stellar Atmospheres: Accelerated Lambda Iteration
3
Multi-frequency / multi-gray
Ways around:
Multi-frequency / multi-gray method by Anderson (1985,1989)
• Group all frequency points according to their opacity into
bins (typically 5) and solve the RT with mean opacities of
these bins. → Only 5 RT equations instead of thousands
• Use a Complete Linearization with the reduced set of
equations
• Solve RT alone in between to get all intensities, Eddington-
factors, etc.
• Main disadvantage: in principle depth dependent grouping
Stellar Atmospheres: Accelerated Lambda Iteration
4
Lambda Iteration
Split RT and SE+RE:
• Good: SE is linear (if a separate T-correction scheme is used)
• Bad: SE contain old values of n,T (in rate matrix A)
Disadvantage: not converging, this is a Lambda iteration!
( )0
( , )
( , )
( , , ) ( , , ) 0
new old
new
v v
J S n T
A J T n b
v n T J S v n T dvκ∞
= Λ
=
− =∫
RT formal solution
SE
RE
3
Stellar Atmospheres: Accelerated Lambda Iteration
5
Accelerated Lambda Iteration (ALI)
Again: split RT and SE+RE but now use ALI
• Good: SE contains new quantities n, T
• Bad: Non-Linear equations → linearization (but without RT)
Basic advantage over Lambda Iteration: ALI converges!
( )
* *
0
( , ) ( , ) ( , )
( , )
( , , ) ( , , ) 0
new old old old new new new old old old
newnew new
new new new new new
v v
J S n T S n T S n T
A J T n b
v n T J S v n T dvκ∞
= Λ + Λ − Λ
=
− =∫
RT
SE
RE
Stellar Atmospheres: Accelerated Lambda Iteration
6
Example: ALI working on Thomson scattering problem
amplification factor
( )
( )( ) ( )( )
( ) ( )( )
( )
* *
* *
*
*
:=formal solution on
solve for
1
1
1 1
⇒ = Λ − Λ + Λ
= Λ − Λ + Λ
= −
= −
− +
− −Λ − =
= − Λ −
⇒
+
− +
new
new new
e e
new new new
e e
e e
old old
old old old
new old
FS
FS
FS
e
old new new
e e
new
e
S B J
S S
B J B
S
B J
B J
J S
J
J
J J J J
J
B
β β
β β
β
β
β
β
β
β
( )
( )
1* *
1*
1 subtract on both sides
1
−
−
= − Λ − Λ
⇒ − = − Λ −
FS old old
e e
new old FS old
e
J J J
J J J J
β β
β
Interpretation: iteration is driven by difference (JFS-Jold) but: this difference
is amplified, hence, iteration is accelerated.
Example: βe=0.99; at large optical depth Λ* almost 1 ĺ strong amplifaction
source function with scattering, problem: J unknownĺiterate
4
Stellar Atmospheres: Accelerated Lambda Iteration
7
What is a good ȁ*?
The choice of ȁ* is in principle irrelevant but in practice it
decides about the success/failure of the iteration scheme.
First (useful) ȁ* (Werner & Husfeld 1985):
A few other, more elaborate suggestions until Olson &
Kunasz (1987): Best ȁ* is the diagonal of the ȁ-matrix
(ȁ-matrix is the numerical representation of the integral operator ȁ)
We therefore need an efficient method to calculate the
elements of the ȁ-matrix (are essentially functions of ττττνννν ).
Could compute directly elements representing the ȁ-integral operator, but
too expensive (E1 functions). Instead: use solution method for transfer
equation in differential (not integral) form: short characteristics method
( )*( )
, ' ( ')0
>Λ =
≤v
v v
SS
τ τ γτ τ τ
τ γ
Stellar Atmospheres: Accelerated Lambda Iteration
8
In the final lecture tomorrow, we will learn two important methods to
obtain numerically the formal solution of the radiation transfer
equation.
1. Solution of the differential equation as a boundary-value problem
(Feautrier method). [can include scattering]
2. Solution employing Schwarzschild equation on local scale (short
characteristics method). [cannot include scattering, must ALI iterate]
The direct numerical evaluation of Schwarzschild equation is much too
cpu-time consuming, but in principle possible.
5
Stellar Atmospheres: Accelerated Lambda Iteration
9
Olson-Kunasz ȁ*
Short characteristics with linear approximation of source
functionmax
maxmax
0
( , , ) ( , , ) exp ( )exp
( , , ) (0, , ) exp ( ) exp
dI v I v S
dI v I v S
τ
τ
τ
τ τ τ τ ττ µ τ µ τ
µ µ µ
τ τ τ ττ µ µ τ
µ µ µ
+ +
− −
′ ′ − −′= − + −
′ ′−
′= − + −
∫
∫
( )
( )
( )
1
1 1
1
1
( , , ) ( , , ) exp ( , , )
( , , ) ( , , ) exp ( , , )
with
using a linear interpolation for the spatial variation of
the intergrals can be evalu
i i i i
i i i i
i i
i
i
I v I v I S v
I v I v I S v
S
I
τ µ τ µ τ µ
τ µ τ µ τ µ
τ ττ
µ
+ + ++
− − −− −
−
−
±
= −∆ + ∆
= −∆ + ∆
−∆ =
∆
1 1
ated as
i i i i i i iI S S Sα β γ± ± ± ±− +∆ = + +
Stellar Atmospheres: Accelerated Lambda Iteration
10
Olson-Kunasz ȁ*
Short characteristics with linear approximation of source
function
0
11
1
i
i
i
e
ee
α
β
γ
+
−∆+
−∆+ −∆
=
−= +
∆
−= − −
∆
1
11
0
i
i
i
ee
e
α
β
γ
−∆− −∆
−∆−
−
−= − +
∆
−= −
∆
=
( ) ( )
( ) ( )
1 1
0 0
1 1
2 2
use 0, ,1, ,0 for 0, , , ,0 to project colums of T
J I I d S S d
S i
µ µµ µ+ − + −= + = Λ + Λ
= Λ
∫ ∫L L L L
6
Stellar Atmospheres: Accelerated Lambda Iteration
11
Inward
( )
1
1 1
1
2
1
1 1
1 1 1
1 1
0
ˆ
ˆ ˆ0 exp( ) 0
ˆ ˆexp( )
ˆ
0
0 exp( ) 0
exp( ) exp( )
ˆ exp( )
i
i i i
i i i
k i ND
i
i i i
i i i i i
k k
i
I
I I
I I
I
i
I
τ
τ
γ
γ τ β
γ τ β τ α
τ
−
−
− −
− −
− −
+
−
= +
−
−
− −− −
− − −
− − +
−
− −
∆ =
−∆ + ∆ −∆ + ∆
−∆ + −∆ + −∆ +
−∆
L
M
M
M
M
M
M
M
M
1
1 1
0
0
0 0
exp( )
ˆ exp( )
i
i i i
k k
i
I
β
β τ α
τ
−
− −+
−− −
= −∆ +
−∆
M
M
M
M
Stellar Atmospheres: Accelerated Lambda Iteration
12
Outward
( )
0 2
1 1
1
1
1 1
1 1 1
1
1
ˆ
ˆ ˆexp( )
ˆ ˆ0 exp( ) 0
ˆ
0
ˆ exp( )
exp( ) exp( )
0 exp( ) 0
0
k i
i i i
i i i
i
k k
i i i i i
i i i
i
i
I
I I
I I
I
i
I
τ
τ
τ
α τ β τ γ
α τ β
α
+
= −
+ +
− −
+ ++
+
+
+
+ −
+ + +
+ − −
+ +
+
+
+
−∆ + ∆
= −∆ + ∆
∆
−∆ −∆ + −∆ +
−∆ +
L
M
M
M
M
M
M
M
M
1 1
1 1
ˆ exp( )
exp( )
0 0
0
0
k k
i i i
i
i
I τ
β τ γ
β
+
+ −
+ +
− −
+
−∆ −∆ +
=
M
M
M
M
7
Stellar Atmospheres: Accelerated Lambda Iteration
13
ȁ-Matrix
( )
( )
( )
( )
1
1 1
0
1
1 1
0
1
*,
0
1
1
0
1
1 1
0
1 ˆ exp( )2
1exp( )
2
10 0
2
1exp( )
2
1 ˆ exp( )2
k k
i i i
i i i
i i i
k k
d I
d
d
d
d I
µ τ
µ β τ γ
µ β β
µ β τ α
µ τ
+
+ −
+ +− −
+ −
− −+
−
− −
−∆ −∆ +
Λ = +
−∆ +
−∆
∫
∫
∫
∫
∫
M
M
M
M
Stellar Atmospheres: Accelerated Lambda Iteration
14
Towards a linear scheme
ȁ* acts on S, which makes the equations non-linear in the
occupation numbers
• Idea of Rybicki & Hummer (1992): use J=ǻJ+Ȍ*Șnew instead
• Modify the rate equations slightly:
( )*
0 0
*3
2
0
*3
*
2
0
4 4 ( )
24
24 ( )
ij ij
ij i i v
ijiji j j v
j
i
i
j
ji
j
R n n J dv J dvhv hv
n hvR n n J dv
n hv c
n hvJ dv
n
n n
nhvn
c
σ σπ π η
σπ
σπ η
∞ ∞
∞
∞
= = Ψ + ∆
= +
= Ψ + ∆ +
∫ ∫
∫
∫