primordial)inhomogeneies )in)the)cmb)spectrum ... · primordial)inhomogeneies...
TRANSCRIPT
Primordial inhomogenei.es in the CMB spectrum. Re-‐analysis of 4-‐year COBE-‐DMR data
Group 2
Eskil Varenius, Thøger Rivera-‐Thorsen, Livia Vallini
Goals
• Check if there are large scale fluctua.ons in the CMB radia.on
• Check the slope of the CMB spectrum • Compare with similar analysis
Theory: CMB
Classic BB theory à T0 Infla.on à Fluctua.ons What scales?
Theory: CMB
l = 2
l = 3 l = 20 l = 50
l = 100
= +
+
+
+
+ ...
Theory: Likelihood analysis 1) We need a model:
signal (Gaussian and correlated)
noise (Gaussian and uncorrelated)
foregrounds: • instrumental monopole • dipole (our movement) • galaxy emission
2) Covariance matrix:
3) Harrison Zel’dovic power spectrum:
Q
n-‐1
Theory: Likelihood analysis
Bayes theorem
Theory: Likelihood analysis Model is Gaussian:
(Gaussian)
(Gaussian)
(ignored)
Model depends only on (Q, n) Likelihood func.on for our model:
Data: COBE – DMR features
Differen.al Microwave Radiometer
Data: HEALpix A “smart” choice: HEALpix format
Data
COBE raw data
Data
COBE raw data smoothed to the beam of 7°
Data
COBE raw data smoothed to the beam of 7° COBE mask to get rid of Galac.c foregrounds
Data
OUR proposed mask to get rid of ALL bad pixels
Data II
• Likelihood analysis: implementa.on fixed grid
101 points in Q direc.on 51 points in n direc.on
Scan the whole plane
Data II
• Likelihood analysis: implementa.on MCMC
“Intelligent” random walk
Results
Results
Results
Results
• Comparison with Gorski et al. 1994 analysis
Conclusions
• Maximum-‐likelihood analysis of the COBE-‐DMR data ① The results are consistent with the Harrison-‐Zel’dovich spectrum
with n = 1 favored by infla.onary models ② The clear non-‐zero value of Q strongly suggests that there is a large
scale structure in the fluctua.ons of the CMB ③ Good agreement with results achieved by Gorski et al. (1994) in an
analogous study of the 2-‐year COBE-‐ DMR dataset.
• See you at Nobel Prize Ceremony! ;-‐)
Power Spectrum