Why Cosmology should care about the Milky Way

My talk at the 'Galaxy Coffee' at the MPIA in Heidelberg about various aspects of foreground modeling for Cosmology.

Transcript

1. Why Cosmology should care about the Milky Way in collaboration with O. Doré, B. Hensley, G. Lagache, P. Serra, P. Bull, A. Manzotti Daniel Lenz Galaxy Coffee @MPIA July 27 © 2017 California Institute of Technology. Government sponsorship acknowledged.
2. Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology Before we start… ❖ Talk to me about: ❖ Galactic HI ❖ Observational multiphase ISM ❖ CIB/CMB component separation, de-lensing ❖ Dust/Reddening ❖ HI intensity mapping (simulations) ❖ Machine learning, Bayesian models, python, Mac OS 2
3. Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology Before we start… ❖ Talk to me about: ❖ Galactic HI ❖ Observational multiphase ISM ❖ CIB/CMB component separation, de-lensing ❖ Dust/Reddening ❖ HI intensity mapping (simulations) ❖ Machine learning, Bayesian models, python, Mac OS 2
4. Which foregrounds do we care about? "(…) the name of the game is component separation, not noise reduction" H.K. Eriksen, 'Advances in Theoretical Cosmology in Light of Data 2017' Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 3
5. Which foregrounds do we care about? "(…) the name of the game is component separation, not noise reduction" H.K. Eriksen, 'Advances in Theoretical Cosmology in Light of Data 2017' ❖ Extinction for cosmological galaxy surveys ❖ Cosmic infrared background measurements ❖ De-lensing of CMB data for primordial gravitational waves ❖ Dust and synchrotron foregrounds in CMB data Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 3
6. Reddening
7. E(B-V) ❖ E(B-V) = Extinction in B band - Extinction in V band ❖ More dust => larger E(B-V) ❖ E(B-V) maps essential for correcting observations for Galactic reddening Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 5
8. Mapping E(B-V) Direct approach ❖ Find many sources with known spectrum (e.g. stars, passive galaxies) ❖ Measure spectra, attribute differences to dust ❖ E.g. Schlafly+ (2014), Green+ (2015) using Pan-STARRS ❖ Limited by modeling accuracy and sensitivity Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 6
9. Mapping E(B-V) Direct approach ❖ Find many sources with known spectrum (e.g. stars, passive galaxies) ❖ Measure spectra, attribute differences to dust ❖ E.g. Schlafly+ (2014), Green+ (2015) using Pan-STARRS ❖ Limited by modeling accuracy and sensitivity Indirect approach ❖ Measure dust optical depth, linearly related to E(B-V) ❖ Schlegel, Finkbeiner, Davis (SFD, 1998) still state-of-the art ❖ FIR emission may have contributions from Zodiacal Light and unresolved galaxies, also needs dust temperature correction Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 6
10. Mapping E(B-V) Direct approach ❖ Find many sources with known spectrum (e.g. stars, passive galaxies) ❖ Measure spectra, attribute differences to dust ❖ E.g. Schlafly+ (2014), Green+ (2015) using Pan-STARRS ❖ Limited by modeling accuracy and sensitivity Indirect approach ❖ Measure dust optical depth, linearly related to E(B-V) ❖ Schlegel, Finkbeiner, Davis (SFD, 1998) still state-of-the art ❖ FIR emission may have contributions from Zodiacal Light and unresolved galaxies, also needs dust temperature correction Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 6
11. HI emission as basis for E(B-V) ❖ Gas and dust are well-coupled in the ISM ❖ Perform an SFD-like analysis to convert HI emission to E(B-V) ❖ Resulting maps free from errors due to dust temperature, Zodi, and extragalactic emission ❖ Limited by non-HI gas along the line of sight Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 7
12. HI4PI Survey ❖ Merges data from Effelsberg and Parkes ❖ Replaces LAB as state-of-the-art full-sky HI survey ❖ Higher sensitivity & resolution, fewer systematics, full sampling 20 21 22 log(NHI [cm 2]) 180 135 90 45 0 315 270 225 180 60 30 0 30 60 HI4PI collaboration
 (2017) Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 8
13. 0 0.05 E(B V )model [mag] The E(B-V) map 40% sky coverage, 16.1’ resolution Lenz, Hensley, Doré (2017, submitted) Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 9
14. -0.01 0.01 SFD - PG10 [mag] Dust systematics ❖ Peek & Graves (2010) used SDSS passively evolving galaxies as "standard crayons" ❖ Correction to the SFD map at 4.5 deg Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 10
15. Dust systematics -0.01 0.01 SFD - PG10 [mag] -0.01 0.01 SFD - Model [mag] Based on extragalactic sources Based on galactic HI Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 11
16. The Cosmic Infrared Background What is the CIB? Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 12
17. The Cosmic Infrared Background What is the CIB? Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 12
18. The Cosmic Infrared Background Made up from dust in galaxies at z=1-3 Lagache+ (2002) Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 13
19. The Cosmic Infrared Background Made up from dust in galaxies at z=1-3 Lagache+ (2002) Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 13
20. The CIB as cosmological probe … of star formation history Planck collaboration (2013 XXX) ❖ Strong constraints on SFH up to z=2.5 ❖ Probe dust temperature across cosmic times ❖ Understand star formation in DM halos Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 14
21. ❖ CMB lensing and CIB match great in z and MHalo ❖ Ideal probe of relation between dark and luminous matter … of large scale structure to cross-correlate with lensing Planck collaboration (2014 XVIII) The CIB as cosmological probe Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 15
22. ❖ CMB lensing and CIB match great in z and MHalo ❖ Ideal probe of relation between dark and luminous matter … of large scale structure to cross-correlate with lensing Planck collaboration (2014 XVIII) The CIB as cosmological probe Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 15
23. How to obtain CIB maps? ❖ Galactic thermal dust and CIB dust dominate on large scales at ~200 to 1000 GHz ❖ How to disentangle them? Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 16
24. How to obtain CIB maps? A. Fit different frequency channels with modified blackbody spectra B. Utilize the different angular power spectra of these components C. Use template maps of Galactic dust (e.g. HI-based) ❖ Galactic thermal dust and CIB dust dominate on large scales at ~200 to 1000 GHz ❖ How to disentangle them? Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 16
25. Correlation of dust and gas HI Dust ❖ Linear relation to first order, but better model required to get to CIB levels Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 17
26. Modeling dust foregrounds • • Velocity separation difficult for complex structures and large scales Radial Velocity HVC IVC LVC I = ✏HVC NHVC + ✏IVC NIVC + ✏LVC NLVC Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 18
27. Modeling dust foregrounds • Generalised linear model (GLM) • Radial Velocity I = X i ✏iTi B Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 19
28. Modeling dust foregrounds • Generalised linear model (GLM) • • Regularised: • • Accounts for all features along line of sight I = X i ✏iTi B Radial Velocity | Datai Modeli |2 + ↵ · |✏i | Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 20
29. CIB: Access large scales ❖ Sliding window is moved across the sky ❖ Model is evaluated for each position, yields map of parameters and CIB values Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 21
30. CIB: Access large scales ❖ Sliding window is moved across the sky ❖ Model is evaluated for each position, yields map of parameters and CIB values Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 21
31. CIB: Galactic poles Total FIR intensity Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 22
32. CIB: Galactic poles CIB Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 22
33. Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology De-lensing the CMB 23
34. Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology The challenge ❖ Lensing of CMB E- modes leads to apparent B-modes ❖ De-lensing of this effect through internal algorithms or tracers of the large-scale structure 24 Courtesy A. Challinor
35. The CIB as cosmological probe … of large scale structure to de-lens CMB maps Manzotti+ (2017) Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 25
36. ❖ Herschel 500 micron as CIB template ❖ "No lensing" excluded at 7 sigma … of large scale structure to de-lens CMB maps Manzotti+ (2017) The CIB as cosmological probe Daniel Lenz, Caltech/JPL Foregrounds in observational Cosmology 26
37. Thank you! ❖ Talk to me about: ❖ Galactic HI ❖ Observational multiphase ISM ❖ CIB/CMB component separation, CMB de-lensing ❖ Dust/Reddening ❖ HI intensity mapping (simulations!) ❖ Machine learning, Bayesian models, python, Mac OS 27
38. Thank you! ❖ Talk to me about: ❖ Galactic HI ❖ Observational multiphase ISM ❖ CIB/CMB component separation, CMB de-lensing ❖ Dust/Reddening ❖ HI intensity mapping (simulations!) ❖ Machine learning, Bayesian models, python, Mac OS 27
39. Backup slides 28
40. Modeling dust foregrounds Model Residual Standard GLM GLM Standard Lenz+ (2016) 29
41. 1020 1021 1022 NHI [cm 2] 10 3 10 2 10 1 100 E(B V ) [mag] 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 NHI [cm 2] ⇥1020 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 E(B V ) [mag] E(B V ) [mag] = 1.216+0.009 0.009 ⇥ NHI [1022 cm 2] + 0.015+0.0002 0.0002 [mag] = 0.02406+0.00006 0.00006 100 101 102 103 104 # data points 0 10 20 30 40 50 60 70 80 # data points 1020 1021 1022 NHI [cm 2] 10 3 10 2 10 1 100 E(B V ) [mag] 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 NHI [cm 2] ⇥1020 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 E(B V ) [mag] E(B V ) [mag] = 1.113+0.002 0.002 ⇥ NHI [1022 cm 2] + 0.000+0.0001 0.0001 [mag] = 0.00570+0.00001 0.00001 100 101 102 103 104 # data points 0 100 200 300 400 # data points The E(B-V)/NHI ratio Pan-STARRS E(B-V), Schlafly+ (2014) SFD E(B-V) Star-based Dust-based 30
42. Current CIB maps: GNILC ❖ Planck collaboration (2016 XLVIII), focus on removing CIB from Galactic dust maps ❖ Using the angular power spectra of the two components ❖ Does not agree that well on a pixel-to-pixel basis 31
43. 0.3 0.2 0.1 0.0 0.1 0.2 0.3 Planck XXX 0.3 0.2 0.1 0.0 0.1 0.2 0.3 GNILC 100 101 102 103 Current CIB maps: GNILC ❖ Planck collaboration (2016 XLVIII), focus on removing CIB from Galactic dust maps ❖ Using the angular power spectra of the two components ❖ Does not agree that well on a pixel-to-pixel basis 32
44. Current CIB maps: GNILC ❖ Cross correlation with Planck CMB lensing ❖ Missing CIB power, especially on the largest scales 33
45. Current CIB maps: Planck (2013 XXX) ❖ Limited sky coverage, hard to access large scales 34