Page 1
Searching for primordial features from CMB and LSS surveys
Bin HuLorentz Institute, Leiden University
The primordial Universe after Planck, IAP, Paris, Dec. 2014
collab. with A. Achucarro, V. Atal, P. Ortiz, J. Torrado
See also J.Torrado Poster
[PRD 89 (2014) 103006] [PRD 90 (2014) 023511]
[arXiv:1410.4804]
Page 2
Outline
1. Observational hints of oscillatory features
2. Models with a transient reduction of the speed of sound
3. Search with CMB map
4. Search with LSS survey
5. Conclusion
Page 3
Planck-2013: Great success of base-LCDM & single-field slow-roll inflationary model
[Planck-2013: XVI]
[Planck-2013: XXII]
[Planck-2013: XXIV]
Page 4
1. Observational hints of oscillatory features
TT spectrum residual from best-fit LCDM model
[Planck-2013: XXII]
Page 5
l ∈(500,1200)
Appears in all channels
Spectrum residual from best-fit LCDM model
Page 6
2. CMB bispectrum
Observational hints of oscillatory features
The best-fit template to the reconstructed CMB bisp
~ detection3σ
[Planck-2013: XXIV]
Page 7
1. A transient reduction of sound speed generically gives primordial oscillatory features.
2. It could produce sizeable and distinguishable features in CMB spectrum, bispectrum and matter spectrum.
3. Planck-2013 and WiggleZ data shows a coincidence in the best-fit mode.
4. The statistical significance is not big enough to claim a detection.
5. Based on our best-fit mode from power spectra, we have a specific prediction on the bispectrum, and we are waiting for Planck-2014(5) test.
Main results
Page 8
2. Models with a transient reduction of the speed of sound
Assumption: 1 light & 1 heavy fields
Time
cs2
Two field model:
[C. Cheung et. al. JHEP 0803 (2008) 014] [S. Weinberg Phys.Rev. D77 (2008) 123541] [A. Achucarro et. al. JHEP 1205 (2012) 066]
EFT for inflation:
light adiabatic heavy isocurvature
derivative coupling, e.g. => a turn
Page 9
After Integrating out heavy field
effective action for light field: slow roll sound speed
Primordial sprectrum: sub-leading
Primordial bispectrum: leading
Do NOT interrupt slow roll condition!
Page 10
Oscillatory features in the transient sound speed reduction models— Power spectrum
keep slow roll condition
[A.Achucarro et. al. PRD 89 (2014) 103006]Gaussian reduction in e-folds
Page 13
2. Primordial Bispectrum (leading order)
removing 1k13k2
3 +1
k13k3
3 +1
k23k3
3K = 0.19 K = 0.21
[Miranda et al. Phys.Rev. D86 (2012)], [Park et al. Phys.Rev. D85 (2012)]Step in sound speed:
[Adshead et al. PhysRevD.84.043519], [Bartolo et al. JCAP 1310 (2013) 038]
[Adshead et al. PhysRevD.84.043519], [Nakashima et al. Prog.Theor.Phys. 125 (2011)][Bean et al. JCAP 0803 (2008) 026], [Cannone et al. Phys.Rev. D89 (2014)]
Page 14
squeezedl1 = 4,l3 = l 2+4
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
0 100 200 300 400 500 600 700 800 900 1000
l 24 b4,l 2,l 2+4×10
8 [µK3 ]
l2
ISWLlocal (fNL=10)Sound speed model (|B|=0.1, log(-τ0)=5.55, log(β)=7.15)
sin model (fNL=106, kc=0.01, φ=0)
single field slow roll: fNL
local ~ 0.01
preliminary
Page 15
equilateral
-15
-10
-5
0
5
10
15
20
25
0 100 200 300 400 500 600 700 800 900 1000
l4b l,l,l×10
3 [µK3 ]
l
local (fNL=1)Sound speed model (|B|=0.1, log(-τ0)=5.55, log(β)=7.15)
sin model (fNL=103, kc=0.01, φ=0)
l1 = l2 = l3
preliminary
Also see Munchmeyer’s & Van Tent’s talks
Page 16
Fergusson et al. 1410.5114
Page 17
A
B
C
D
3. Search with CMB map—TT spectrum
degeneracy of featured and
vanilla parameters is
negligible
Planck+WP
profile likelihood
Also see Meerburg’s talk
Page 19
4. Search with LSS survey—WiggleZ
features shows around k~(0.1,0.2)
Search up to k=0.2
10
100
1000
10000
100000
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
P(k)
[h-3
Mpc
3 ]
k [h/Mpc]
|B|=0.1, log(-τ0)=5.55, log(β)=7.15
z=0z=0.5z=1z=1.5z=2
Page 20
Independent search with different data
Planck+WP WiggleZ
Page 21
Planck+WP WiggleZ
Independent search with different data
Two coincident modes including the best-fit mode
Page 22
Combine Planck and WiggleZ
get better constrained in Planck+WiggleZ
Page 23
Bayesian Evidence
posterior evidenceBeyesian
ratio
Jeffreys’s criterion (1<R<3): Barely worth mentioning!
Evidence:
: Sound speed modelM1M 0 : Base-LCDM model
R>1: data faver M1R<1: data faver M0
Page 24
1. A transient reduction of the speed of sound generically gives primordial oscillatory features.
2. It could produce sizeable and distinguishable features in CMB spectrum, bispectrum and matter spectrum.
3. Planck-2013 and WiggleZ data shows a coincidence in the best-fit mode.
4. The statistical significance is not big enough to claim a detection.
5. Based on our best-fit mode from power spectra, we have specific prediction on the bispectrum, and we are waiting for Planck-2014(5) test.
Conclusion-2.5
-2
-1.5
-1
-0.5
0
0.5
1
0 100 200 300 400 500 600 700 800 900 1000
l 24 b4,l 2,l 2+4×10
8 [µK3]
l2
ISWLlocal (fNL=10)Sound speed model (|B|=0.1, log(-τ0)=5.55, log(β)=7.15)
sin model (fNL=106, kc=0.01, φ=0)
-15
-10
-5
0
5
10
15
20
25
0 100 200 300 400 500 600 700 800 900 1000
l4b l,l,l×10
3 [µK3]
l
local (fNL=1)Sound speed model (|B|=0.1, log(-τ0)=5.55, log(β)=7.15)
sin model (fNL=103, kc=0.01, φ=0)
Page 25
Thanks for your attention! Merry Xmas to Planck!
Page 27
Log(β ) = 7.2
Two mode with the same frequency but with different location (red) (green)
Log(−τ 0 ) = 5.5Log(β ) = 6.3 Log(β ) = 7.2
Page 28
1.9e-09
1.95e-09
2e-09
2.05e-09
2.1e-09
2.15e-09
2.2e-09
2.25e-09
2.3e-09
2.35e-09
2.4e-09
0.01 0.1
Log tau0=5.5, Log beta=6.3Log tau0=5.5, Log beta=7.2
Primordial power spectrum
Page 29
0
5e+06
1e+07
1.5e+07
2e+07
2.5e+07
1e-06 1e-05 0.0001 0.001 0.01 0.1 1 10
’./test_cs_transfer_out.dat’ u 1:7
Transfer function
Page 30
After convolving with transfer function they looks similar, due to the damping effect on small scale
0
5e-10
1e-09
1.5e-09
2e-09
2.5e-09
0.01 0.1
Log tau0=5.5, Log beta=6.3/before filteringLog tau0=5.5, Log beta=7.2/before filtering
Log tau0=5.5, Log beta=6.3/after filteringLog tau0=5.5, Log beta=7.2/after filtering
Page 34
A
B
C
D
3. Search with CMB map—TT spectrum
degeneracy with vanilla
parameter is negligible
Planck+WPprofile likelihood
CoV Mat
Page 35
Search with CMB map—Zoom in best-fit
Need to consider look-elsewhere effect!
Enlarge the parameter space
Page 36
2. Models with a transient reduction of the speed of sound
integrating out heavy field
Time
cs2
turn
sound speed reduced
A.Achucarro et. al. JHEP 1205 (2012) 066 light adiabatic heavy isocurvature
effective action:
Page 37
4. Search with LSS survey—WiggleZ
1
10
100
1000
10000
100000
0.0001 0.001 0.01 0.1 1
P(k)
[h-3
Mpc
3 ]
k [h/Mpc]
|B|=0.1, log(-τ0)=5.55, log(β)=7.15
z=0z=0.5z=1z=1.5z=2
features shows around k~(0.1,0.2)
Search up to k=0.2