Detecting B-modes from the ground David Alonso COSMO19 – Aachen, Sept. 4th 2019 CMB polarization E-modes: Gradient-like, even parity B-modes: Curl-like, Seljak & Zaldarriaga 9609169 odd parity Kamionkowski et al. 9609132 CMB polarization Thomson scattering Unpolarized, isotropic source Unpolarized output e e’ Unpolarized, anisotropic source Partially polarized output Kosowski 9501045 Cabella & Kamionkowski 0403392 Kamionkowski & Kovetz 1510.06042 CMB polarization Thomson scattering Generated intensity fluctuations Scalar perturbation k This can only generate E-modes CMB polarization Thomson scattering Generated intensity fluctuations Tensor perturbation k This can generate both E-modes and B-modes Power spectra r ~ O(10-2-10-3) is a well-motivated target (e.g. R2-type models). 3 orders of More importantly: it’s within magnitude the realm of possibility! Reionization Recombination Power spectra 3 orders of magnitude Reionization Recombination Measurement challenges: lensing At low-l, lensing is effectively white noise with sN ~ 5 mK arcmin. -3 This limits sensitivity on the recombination bump to s(r) ~ 10 Challenges: lensing E Solution: delensing ● Obtain a template of the CMB lensing potential F(q). ● Combine it with a map of E-modes to get a template of lensing B-modes. ● Subtract it from you maps to reduce the variance F from lensing B-modes. ● Similar to multi-tracer methods of cosmic variance reduction. B Challenges: lensing Solution: delensing ● Obtain a template of the CMB lensing potential F(q). ● Combine it with a map of E-modes to get a template of lensing B-modes. ● Subtract it from you maps to reduce the variance from lensing B-modes. ● Similar to multi-tracer methods of cosmic variance reduction. Delensing is not cheap! ● You can get delens up to ~30-50% using large-scale structure and CIB data. ● To go beyond that, you need to reconstruct F with your own CMB data. ● You need very high-sensitivity data in order to go S4 collaboration – arXiv:1610.02743 beyond ~75%. ● Eventually you need to invest more detector time into delensing than into actually observing B-modes. Challenges: foregrounds For the first time in CMB’s history, the CMB is a perturbation on top of foregrounds, and not the other way around. Thankfully we (mostly) have to deal with Galactic foregrounds only. Challenges: foregrounds Cleaning strategies are based on the Synchrotron different spectra of different components. Main components: AME ● CMB: black-body spectrum with T=2.755 K. ● Dust: modified BB spectrum. Dominates at high frequencies. CMB ● Synchrotron: curved power law. Dust Dominates at low frequencies. Thorne et al. (inc. DA) 1608.02841 ● DA et al. 1608.00551 Other: AME (?), CO lines. Torne et al. (inc. DA) 1905.08888 Ali-Haimoud et al. 0812.2904, Puglisi et al. 1701.07856 Challenges: foregrounds Cleaning strategies are based on the different spectra of different components. Main components: ● CMB: black-body spectrum with T=2.755 K. ● Dust: modified BB spectrum. Dominates at high frequencies. ● Synchrotron: curved power law. Dominates at low frequencies. ● Other: AME (?), CO lines. Ward et al. (inc. DA) 1803.07630 Atmosphere limits the range of frequencies on which we can actually observe. Water vapour is the main culprit (e.g. prevents high-frequency observations). Challenges: foregrounds Foreground challenges: ● Spatial variability. ● Frequency decoherence (intrinsic and caused by 3D structure). b ,T b1,T1 3 3 b5,T5 b2,T2 b4,T4 Vansyngel et al. 1611.02577 Chluba, Hill, Abitbol 1701.00274 BICEP2/Keck et al. 1810.05216 Challenges: foregrounds Foreground challenges: ● Spatial variability. ● Frequency decoherence (intrinsic and caused by 3D structure). Foreground removal methods: ● Parametric: assume a model for all components and marginalize over free foreground parameters. ● Non-parametric: only assume CMB spectrum. FG residuals still need to be modelled (so not quite model-independent). Errard et al. 1509.06770 DA et al. 1608.00551 Torne et al. (inc. DA) 1905.08888 Errard & Stompor 1811.00479 Challenges: foregrounds Foreground challenges: ● Spatial variability. ● Frequency decoherence (intrinsic and caused by 3D structure). Foreground removal methods: ● Map-based: model the contribution of each component in each pixel, at each frequency. Pro: optimal, makes use of phase information. Con: it is hard to model the effects of many effects (bandpass uncertainties, filtering etc.). ● Cl-based: compute all auto- and cross-spectra between different frequencies and propagate foreground model into power spectrum. Pro: we know how to compute unbiased Cls in the presence of systematics. Con: may be lossy, harder to account for spatial variations. Errard et al. 1509.06770 DA et al. 1608.00551 Torne et al. (inc. DA) 1905.08888 Errard & Stompor 1811.00479 Challenges: atmosphere Atmosphere: ● Atmosphere emits on CMB frequencies. ● Source of non-white noise. Dominant on large scales. ● Mostly unpolarized, but leaks into polarization too. ● Prevents B-mode detection from reionization bump. Challenges: atmosphere Atmosphere: ● Atmosphere emits on CMB frequencies. ● Source of non-white noise. Dominant on large scales. ● Mostly unpolarized, but leaks into polarization too. ● Prevents B-mode detection from reionization bump. Li et al. 1710.03047 Challenges: systematics Half-wave plates ● Rotating HWPs change the polarization angle as a function of time. ● Use rotating speed to separate polarized and unpolarized components. ● More reliable at mitigating I→P leakage. Vital to reduce atmospheric noise. ● They have their own systematics! Emission from HWP (cryogenics), reflections, f(n), demodulation in mapmaking... Challenges: systematics Polarization angle calibration ● Mis-calibration leads to E-B mixing. ● Very good calibration can be achieved by nulling EB power spectrum. ● This is dangerous in the presence of foregrounds! Max Abitbol ● Calibration through external sources (e.g. artificial sources – drones, astrophysical sources with known polarization). Minami et al. 1904.12440 Challenges: systematics Bandpass calibration ● Uncertainties in bandpass or gains will cause foreground leakage. ● Calibration needed at the level of <1%. ● This can be relaxed by marginalising over these uncertainties Max Abitbol leading to self-calibration (CMB has a very well known spectrum!). Ward et al. (inc. DA) 1803.07630 Challenges: systematics Scan-synchronous signal ● A.K.A. ground. Pick-up from ground and other structures. ● Mitigated by good shielding. ● Removed through filtering or ground template removal. ● Direct effect on low-l measurements. Challenges: systematics Half-wave plates Polarization angle calibration Bandpass calibration Scan-synchronous signal Beam calibration Pointing I→ P leakage ... The Simons Observatory ● Seed fund by the Simons Foundation ● Join the efforts of ACT and POLARBEAR in Chile. ● >250 members, >40 institutions, 10 countries. The Simons Observatory ● Seed fund by the Simons Foundation ● Join the efforts of ACT and POLARBEAR in Chile. ● >250 members, >40 institutions, 10 countries. The Simons Observatory ● Seed fund by the Simons Foundation ● Join the efforts of ACT and POLARBEAR in Chile. ● >250 members, >40 institutions, 10 countries. The Simons Observatory ● Two surveys: Small-Aperture Telescope: 3 telescopes, B-modes. Large-Aperture Telescope: lensing, x-correlations, delensing. ● 6 frequencies (30-280 GHz). LAT SAT m 5 1 x3 The Simons Observatory ● Two surveys: Small-Aperture Telescope: 3 telescopes, B-modes. Large-Aperture Telescope: lensing, x-correlations, delensing. ● 6 frequencies (30-280 GHz). The Simons Observatory ● 5-year observations ● >80K detectors ● Planned extensions LAT Telescope construction (VERTEX) Science observations Receiver Receiver Testing, shipping, design manufacture install. 18 19 20 21 26 20 20 20 20 20 platform platform Testing SAT design fabrication shipping, install camera Camera Science observations 1st SAT on sky design manufacture The Simons Observatory GW forecasts s(r) x 103 SO et al. 1808.07445 The Simons Observatory GW forecasts SO et al. 1808.07445 Summary ● Tensor perturbations generate B-modes in the CMB polarization. ● Much smaller amplitude. Target: s(r)<10-3 ● Observational challenges: ● Foregrounds ● CMB lensing ● Atmosphere ● Calibration (HWPs, bandpass, gain, pol. angle…) ● Simons Observatory: ● LAT (lensing, delensing, cross-correlations) ● SAT (B-modes) ● Observations 2021-2026 ● Forecasts: s(r)=O(2-3) 10-3 ● Future experiments (SO+, S4, LiteBIRD) will break 10-3 threshold. Thanks a lot! .