Anisotropic empTerature Correlation Function · Anisotropic empTerature Correlation Function (Work...
Transcript of Anisotropic empTerature Correlation Function · Anisotropic empTerature Correlation Function (Work...
Anisotropic Temperature Correlation Function
(Work in progress)
Thiago Pereira1,2 Raul Abramo2 Armando Bernui 3
1Instituto de Física Teórica - Unesp
2Instituto de Física - USP
3Instituto de Pesquisas Espaciais
Challenges New Physics in Space
April 2009
Campos do Jordão
What is the challenge we are working with?
What is the challenge we are working with?
Here it is:
...Precisely, we'd like to �nd out more about some of its anomalies,like:
• The low multipole alignments
• North/South and the Cold Spot asymmetries
• The low power in the quadrupole C2
• Any statistically signi�cant deviation from isotropy and/orgaussianity
Why is this a challenge?
In its simplest realization, i.e, single �eld in�ation, the temperatureof the universe is a gaussian random event... So any deviation of itcould tell us more about the primordial universe.
...Precisely, we'd like to �nd out more about some of its anomalies,like:
• The low multipole alignments
• North/South and the Cold Spot asymmetries
• The low power in the quadrupole C2
• Any statistically signi�cant deviation from isotropy and/orgaussianity
Why is this a challenge?
In its simplest realization, i.e, single �eld in�ation, the temperatureof the universe is a gaussian random event... So any deviation of itcould tell us more about the primordial universe.
...Precisely, we'd like to �nd out more about some of its anomalies,like:
• The low multipole alignments
• North/South and the Cold Spot asymmetries
• The low power in the quadrupole C2
• Any statistically signi�cant deviation from isotropy and/orgaussianity
Why is this a challenge?
In its simplest realization, i.e, single �eld in�ation, the temperatureof the universe is a gaussian random event... So any deviation of itcould tell us more about the primordial universe.
What does that exactly mean?
What does that exactly mean?
The temperature and its two-point function are de�ned by
∆T = Σ`,m a`,mY`,m, C (n1,n2) = 〈∆T (n1)∆T (n2)〉
The universe is statistically isotropic if and only if
C (n1,n2) = Σ`(2`+ 1)C`P`(n1 · n2)/4π , C` =1
2`+ 1Σ`|a`,m|2
What does that exactly mean?
The temperature and its two-point function are de�ned by
∆T = Σ`,m a`,mY`,m, C (n1,n2) = 〈∆T (n1)∆T (n2)〉
The universe is statistically isotropic if and only if
C (n1,n2) = Σ`(2`+ 1)C`P`(n1 · n2)/4π , C` =1
2`+ 1Σ`|a`,m|2
If it is not isotropic, what can it be?
Virtually anything! So it is important to analyze the problem in asmuch an independent manner as possible. For example:
• Pullen & Kamionkowski (2008) introduced a simpler version ofit
C (n1,n2) : S2 → R
• Hajian & Souradeep (2003) have considered the correlationfunction in its full form
C (n1,n2) : S2 × S2 → R
These approaches are either too complicated or too simple in orderto search for sub-structures in the morphology of CMB...
If it is not isotropic, what can it be?
Virtually anything! So it is important to analyze the problem in asmuch an independent manner as possible. For example:
• Pullen & Kamionkowski (2008) introduced a simpler version ofit
C (n1,n2) : S2 → R
• Hajian & Souradeep (2003) have considered the correlationfunction in its full form
C (n1,n2) : S2 × S2 → R
These approaches are either too complicated or too simple in orderto search for sub-structures in the morphology of CMB...
If it is not isotropic, what can it be?
Virtually anything! So it is important to analyze the problem in asmuch an independent manner as possible. For example:
• Pullen & Kamionkowski (2008) introduced a simpler version ofit
C (n1,n2) : S2 → R
• Hajian & Souradeep (2003) have considered the correlationfunction in its full form
C (n1,n2) : S2 × S2 → R
These approaches are either too complicated or too simple in orderto search for sub-structures in the morphology of CMB...
If it is not isotropic, what can it be?
Virtually anything! So it is important to analyze the problem in asmuch an independent manner as possible. For example:
• Pullen & Kamionkowski (2008) introduced a simpler version ofit
C (n1,n2) : S2 → R
• Hajian & Souradeep (2003) have considered the correlationfunction in its full form
C (n1,n2) : S2 × S2 → R
These approaches are either too complicated or too simple in orderto search for sub-structures in the morphology of CMB...
If it is not isotropic, what can it be?
Virtually anything! So it is important to analyze the problem in asmuch an independent manner as possible. For example:
• Pullen & Kamionkowski (2008) introduced a simpler version ofit
C (n1,n2) : S2 → R
• Hajian & Souradeep (2003) have considered the correlationfunction in its full form
C (n1,n2) : S2 × S2 → R
These approaches are either too complicated or too simple in orderto search for sub-structures in the morphology of CMB...
Angular-planar correlation function
Motivated by recent analysis (Copi et.al 2005), which suggest thatthe ecliptic and/or the galactic planes may be important in CMBanalysis, we consider
C (n1,n2) = C (n1 · n2) + C (n1 × n2)
which can be expanded as
C (n1,n2) = Σ`,l ,m C lm` P`(cosϑ)Yl ,m(θ, φ)
In this way we can de�ne a rotationally invariant angular-planarpower spectrum statistics
Bl
` =1
2l + 1Σl |C lm` |2 .
Angular-planar correlation function
Motivated by recent analysis (Copi et.al 2005), which suggest thatthe ecliptic and/or the galactic planes may be important in CMBanalysis, we consider
C (n1,n2) = C (n1 · n2) + C (n1 × n2)
which can be expanded as
C (n1,n2) = Σ`,l ,m C lm` P`(cosϑ)Yl ,m(θ, φ)
In this way we can de�ne a rotationally invariant angular-planarpower spectrum statistics
Bl
` =1
2l + 1Σl |C lm` |2 .
Angular-planar correlation function
Motivated by recent analysis (Copi et.al 2005), which suggest thatthe ecliptic and/or the galactic planes may be important in CMBanalysis, we consider
C (n1,n2) = C (n1 · n2) + C (n1 × n2)
which can be expanded as
C (n1,n2) = Σ`,l ,m C lm` P`(cosϑ)Yl ,m(θ, φ)
In this way we can de�ne a rotationally invariant angular-planarpower spectrum statistics
Bl
` =1
2l + 1Σl |C lm` |2 .
Angular-planar correlation function
Motivated by recent analysis (Copi et.al 2005), which suggest thatthe ecliptic and/or the galactic planes may be important in CMBanalysis, we consider
C (n1,n2) = C (n1 · n2) + C (n1 × n2)
which can be expanded as
C (n1,n2) = Σ`,l ,m C lm` P`(cosϑ)Yl ,m(θ, φ)
In this way we can de�ne a rotationally invariant angular-planarpower spectrum statistics
Bl
` =1
2l + 1Σl |C lm` |2 .
Some statistics...
The �rst model we would like to test with the angular-planarestimator is the gaussian and isotropic model itself. Using...
C lm` =∑l1m1
∑l2m2
〈al1m1al2m2
〉 × geometrical terms
we can construct histograms for B l
`...
We have constructed these histograms for 40.000 simulations ofrandom, gaussian and isotropic alm's, for l ∈ [2, 12] and ` ∈ [2, 12]
Some statistics...
The �rst model we would like to test with the angular-planarestimator is the gaussian and isotropic model itself. Using...
C lm` =∑l1m1
∑l2m2
〈al1m1al2m2
〉 × geometrical terms
we can construct histograms for B l
`...
We have constructed these histograms for 40.000 simulations ofrandom, gaussian and isotropic alm's, for l ∈ [2, 12] and ` ∈ [2, 12]
Here is a sample:
0 2 4 6 80.0
0.1
0.2
0.3
0.4
0.5
B62�103�HmKL4
0 2 4 6 8 100.0
0.1
0.2
0.3
0.4
0.5
B43�103�HmKL4
0.00 0.02 0.04 0.06 0.08 0.10 0.120
5
10
15
20
25
30
35
B86�103�HmKL4
0.000 0.005 0.010 0.0150
50
100
150
200
B127�103�HmKL4
Now, we would like now to compare it to the Wmap 5yr data...
0 2 4 6 80.0
0.1
0.2
0.3
0.4
0.5
B62�103�HmKL4
0 2 4 6 8 100.0
0.1
0.2
0.3
0.4
0.5
B43�103�HmKL4
0.00 0.02 0.04 0.06 0.08 0.10 0.120
5
10
15
20
25
30
35
B86�103�HmKL4
0.000 0.005 0.010 0.015 0.020 0.025 0.0300
50
100
150
200
B127�103�HmKL4
Rigorously, we need to consider the CMB measurement errors asrandom variables:
∆Tmap(n) = ∆TCMB(n) + statistical and istrumental errors
In order to estimate the errors, we took the mean and deviation ofthe following various data set
• Third year data: ILC and HILC with di�erent sky-cuts (KQ75and KQ85)
• Five year data: ILC and HILC with di�erent sky-cuts (KQ75and KQ85)
So, we will suppose that statistical and instrumental errors aregaussian random variables, with means and variances de�ned bythese data set.
0 1 2 3 4 50
1
2
3
4
B62�103�HmKL4
0 1 2 3 4 5 6 70.0
0.2
0.4
0.6
0.8
B43�103�HmKL4
0.00 0.02 0.04 0.06 0.080
20
40
60
80
100
120
140
B86�103�HmKL4
0.000 0.005 0.010 0.015 0.020 0.025 0.0300
50
100
150
200
B127�103�HmKL4
The probabilities of observing such values are
P<(B62 ) = 14.2%, P<(B8
6 ) = 29%, P>(B43 ) = 31%, P>(B12
7 ) = 0.037%
Partial Conclusions
We have build an estimator able of quantifying angular and planarcorrelations of the CMB, which is:
• unbiased, minimum-variance and rotationally invariant
• model independent
This work is still in progress. In particular, we haven't checked thee�ect of di�erent masks in our �nal probabilities.
...the exact role of gaussianity and statistical isotropy in the CMBanomalies is still unknown, and a well de�ned distinction betweenthese two properties is still a major challenge.
Acknowledgements
I would like to thank
• Professors R. Opher, R. Rosenfeld and all the organizers
• Fundação de Amparo à pesquisa do Estado de São Paulo.
and the audience.
Thank you!