QUIET Q/U Imaging ExperimenT

1. QUIETQ/U Imaging ExperimenT Osamu Tajima (KEK) QUIET collaboration

2. History of The Universe Planck scale Grand Unified Theory GUT scale 1016GeV ？ Today Foregronds CMB First star Happen to be same order !? Standard Model Begin of universe Energy scale of inflation Inflation Big bang reconbination reionization Dark age Galaxies V1/4 ≈ 0.01 Standard cosmology model ？ Age 10-36sec 380 Kyr1 Myr13.8 Gyr Inflation potential r 1/4 × 1016 GeV Parameterized with “ r “ : tensor-scalar ratio (T/S)

3. Targets of QUIET Model predictions of B-modes from the inflation E-modes Lensing B-modes l(l+1)Cl /2p (uK2) “Inflationary B-modes” Primordial B-modes QUIET 0.2°〜7° Wide multipole range should be covered for ``Inflationary B-modes’’

4. Toward Inflationary B modes • Good systematic error control • Inflationary BB power is less than 1/1,000,000 of TT, 1/10~1/100 or less of EE • Understanding of Foregrounds • Mitigation of experimental systematics • Large fields observation • Inflationary BB is significant more than 1o scale • Should be free from experimental 1/f noise QUIET is designed to fulfill these requirements

5. The QUIET Collaboration 5 countries, 14 institutes, ~50 scientists QUIET observation: Oct. 2008 – Dec. 2010 at Atacama, Chile (5,080m)

6. Thus far useful for demonstration  Observation Patches QUIET(43 GHz) Stokes, U Stokes, Q WMAP (5-year) Visible region along earth rotation ~20o 4 CMB patches were chosen (~3% of full sky) Galaxy observation when CMB patches are not visible

7. CMB QUIET Telescope CMB QUIET polarization module 90 sets for 95 GHz observation Receiver ( detector array inside) ~30cm

8. Constraint on Foregrounds with multi-frequency observations 95 GHz 43 GHz Other experiments QUIET QUIET’s 43GHz data is important to understand effects of Synchrotron radiations

9. QUIET observation at Atacama, Chile 5,080m ~30cm ~30cm > 11,000 H 19 detectors at 43 GHz array sensitivity 69uKs1/2 90 detectors at 95 GHz array sensitivity ~87uKs1/2 ~8 months ~1.5 years

10. Essence of tiny 1/f knee & good systematic error control QUIET polarization detector array CMB Detector array for 95 GHz Septum Polarizer Circuit module 3cm Yield of usable detectors: 95%

11. CMB QUIET’s detector Septum polarizer R L Antenna to pick up “L”, “R” LNA (HEMT) Phase switch phase flip modulation ( 4kHz & 50Hz ) Double Mod. ±1 1 180 Coupler (±1) D4 D1 90 Coupler (±i) W-band module D2 D3

12. CMB QUIET’s detector Stable ! No fluctuation !! Septum polarizer D1= +gAgB× Q D2= -gAgB× U D3= +gAgB× U D4= -gAgB× Q R L Tiny spurious polarization Imperfection of waveguide components makes tiny fake-pol. However, it doesn’t fluctuate, i.e., could be calibrated very well Precise polarization angle f = ½ tan-1(U/Q) ½ tan-1(D3/D1) Each diode response LNA (HEMT) gB gA Phase switch phase flip modulation ( 4kHz & 50Hz ) Double Mod. ±1 1 gA , gBResponsivity of LNA Simultaneous detection of Stokes Q and U！ 180 Coupler (±1) -Q D4 +Q D1 90 Coupler (±i) +U D2 -U D3

13. Very small 1/f knee Observing data under Chilean sky fknee << fscan Double demodulation suppressed 1/f noise !!

14. Very small 1/f knee Scan freq. E-modes B-modes Noise propertyof experiment Measurement range QUIET is free from effects of 1/f noise !!

15. Tiny spurious polarization Total power response as a function of time q Elevation nods DQ Calibration was scheduled every a few hours (~0.3% precision for each) DI We also performed cross calibration by using astronomical objects, e.g., Jupiter Median of all channels (95 GHz band): 0.2% ±0.2% (syst. error dominant)

16. (cross check for relative) Absolute Angle calibration: TauA x sparse-wires Relative(cross check for absolute) Taurus Tpol. = 5mK, αsky=149.9±0.2° Orientation of sinusoidal curve determines detector angle Q Measured angle of ``standard detectors’’ calibration everyday unless it was invisible Yellow bar: precision of single calibration No angle fluctuation !! U dangle: 0.5deg (catalog uncertainty is 0.2deg)

17. (cross check for relative) Absolute Angle calibration: TauA x sparse-wires Relative(cross check for absolute) Artificial calibrator, ``sparse wires’’ determined relative angles Systematic error for relative angle: 0.8o

18. Analysis Strategy Calibration, Data Selection E-modes Filter / Map Making This is simulation This is simulation Stokes U map Stokes Q map Validation Tests B-modes This is simulation B-mode, E-mode spectra Multipolel (=180o/q)

19. “Robust” Analysis Strategy Calibration, Data Selection Blind Analysis Framework Systematic Error Check ✓ Filter / Map Making ✓ Validation Tests B-mode, E-mode spectra “Box Open” Un-blinding the results

20. Analysis Validation: Null Tests • Divide data set into two maps, difference them. • Calculate “null” power spectrum • Perform 42 data divisions for 43 GHz (32 divisions for 95 GHz receiver) • Q vs. U channels • weather conditions • cryostat temperature (CMB+NoiseA) - (CMB+NoiseB) (NoiseA-NoiseB) Null Power Spectrum

21. Passed null tests ? YES ! No bias was detected ! • Zero-consistent mean shift +0.02 ±0.02 (-0.02±0.02) for 43 GHz (95 GHz) • Distribution is consistent with MC  validation of statistical error ●　data ーMC w/o any systematics = Cl / sl Bias estimator : 43 GHz band receiver Mean shifts  bias detection Width  statistical error validation

22. One of the source of detected bias by the validation tests ``Far-sidelobes’’ induced ground pickup 43 GHz receiver Characterized by using the Sun 95 GHz receiver UGS solves Far-sidelobes 43 GHz observation 95 GHz observation

23. Remove effects of ground pickupby far-sidelobes Motion of each patch Take cross-correlation 10 divisions for Azimuth X 6 divisions of boresight rotations x 6 different angles

24. QUIET’s E-modes 95 GHz band receiver 43 GHz band receiver Two independent analysis pipeline obtained consistent results. (Calibrations are not common partially)

25. QUIET’s B-modes 95 GHz band receiver 43 GHz band receiver Zero-consistent power observed

26. Upper limit for B-modes Upper bounds at 95% C.L. 43 GHz band: r < 2.2 95 GHz band: r < 2.7

27. Systematic error for B modes The smallest syst. error to date: δr<0.01 Major inflation models could be covered with large statistics

28. Foreground receiver did its task WMAP 30GHz QUIET(43GHz)WMAP(30GHz) cross-correlation QUIET 43GHz (~1/3 of EE from LCDM) F.G. for E-modes QUIET 95GHz F.G. for B-modes r = 0.02 One of four patches (CMB-1) at 1stbin (l=25–75) b= –3.1 for extrapolation • Real data shows “Foreground receiver” is important !! • Good estimator for effects of Synchrotron radiation

29. Summary • QUIET’s target: B-modes from the inflation • Designed to minimize systematics • Having Foreground receiver • Very good systematic error control • Very low 1/f noise • First experiment Japanese institution joined • One of the best CMB polarization spectrum measurements to date. • In particular E modes “spectrum” • The lowest systematic error to date: dr < 0.01 • Published papers • Results with 43 GHz receiver: ApJ, 741, 111 (2011). • Results with 95 GHz receiver: ApJ, 760, 145 (2013). • About Instruments: ApJ, 768, 9 (2013).

30. Referee report for 95 GHz receiver results Let me congratulate the QUIET team for this impressive piece of work! The control of all systematics down to r of 0.01 is absolutely spectacular. I found the paper clearly written, and a model for future polarization based CMB papers…