1. title 21 cm Intensity Mapping for BAO and CHIME

2. Temperature - the imprint of BAO is visible in the co-added degree-scale hot (left) & cold (right) spots. Polarization - The expected radial/tangential polarization pattern around these extrema, due to Thompson scattering is clearly seen. Average hot spotAverage cold spot BAO have been observed in the CMB, and set the acoustic scale: l A = 302.35 ± 0.65 @ z ∗ =1091. (Bennett et al, 1212.5225) (Hinshaw et al, 1212.5226)

3. BAO now detected in multiple large-scale structure surveys. 107 h -1 Mpc = 148 Mpc h = 0.72

4. BAO measurements converted to distance measurements, with r s,fid = 153.19 Mpc, compared to WMAP prediction (with 1-σ uncertainty). Co-moving radial distance to redshift z ? (Anderson et al, 1203.6594)

5. (Eisenstein et al., ApJ, 2005) The BAO scale schematically superposed on the SDSS galaxy distribution.

6. 6 Hydrogen Intensity Mapping To measure BAO in LSS, no need to resolve individual galaxies. Map intensity of HI gas using redshifted 21 cm emission along the line of sight. For 15 Mpc spatial resolution at z~1-2, need ~15-25’ angular resolution. This requires observations with a ~100 m aperture and a fast mapping speed in a nearly continuous 400-800 MHz band.

7. Simulated 21 cm signal, tracer of ρ HI (θ, ϕ ;z) : ΔT ~ 1 mK left - over full sky at 400 MHz, right - over 50 MHz at 1 declination Main goal is to measure BAO structure in P HI (k), less concerned about amplitude of P HI (k). Hydrogen intensity signal - simulated

8. Modeled Galaxy signal, ΔT ~ 700 K! left - over full sky at 400 MHz, right - over 50 MHz at 1 declination - spectrally smooth The (large) foreground signal: mostly synchrotron

9. BAO Intensity Mapping Experiments

10. The Canadian Hydrogen Intensity Mapping Expt. (CHIME) Frequency: 400-800 MHz (21 cm radiation, z~0.8 - 2.5) Coverage: 20,000 sq. deg. sky; drift scanning (no moving parts); 5% of the co-moving volume of the observable universe Resolution: 1 MHz bandwidth; 13-26 FWHM beam Receivers: 1280 dual polarization feeds, cosmic variance limited in ~2 years It will consist of five 20 m x 100 m N-S cylindrical reflectors with receivers along each 100 m focal line.

11. Penticton BC

13. Beam response of a single feed in a single cylinder Blue: broad feed response w/o cylinder

14. Synthesized response of N feeds in one cylinder N → 256 in full cylinder

15. Synthesized response of N feeds in five cylinders

16. System diagram

17. Pipeline diagram (to analyze ~1 Tb/s)

18. Simulated 21 cm signal, tracer of ρ HI (θ, ϕ ;z) : ΔT ~ 1 mK left - over full sky at 400 MHz, right - over 50 MHz at 1 declination Main goal is to measure BAO structure in P HI (k), less concerned about amplitude of P HI (k). Hydrogen intensity signal

19. Foreground rejection must account for variable beam width and other spectral systematics. “True” foreground (model) Observed foreground over 50 MHz band with beam response

20. Frequency analysis Spatial analysis

21. Foreground signalHI signal ~equal brightness in filtered data Keep HI/F > 10 -1

22. Foreground signalHI signal F ~ 3× dimmer Keep HI/F > 10 -1

23. Foreground signalHI signal F ~ 5× dimmer Keep HI/F > 1

24. Foreground signalHI signal F ~ 10× dimmer Keep HI/F > 10

25. Foreground signalHI signal F ~ 70× dimmer Keep HI/F > 10 2

26. BAO in Fourier space and CHIME sensitivity - with foregrounds top left - projected CHIME sensitivity to BAO feature in matter power spectrum. top right - projected distance errors, D v (z)/r s right - CHIME dark energy constraints, blue: Planck + Stage II, green: add Euclid/BAO, black: add CHIME/BAO w0w0 wawa 0.5% - 1% per bin

27. This will consist of two 35m long cylindrical sections with 64 dual polarization feeds per cylinder (256 data channels), feeding one channelizer-correlator. The CHIME Pathfinder

28. 28 Construction of the CHIME Pathfinder begins, 23 Jan 2013!

30. CHIME Pathfinder Hardware - Analog Feed Amplifier Feed Analog

31. CHIME Pathfinder Hardware - Digital

32. Here, the dishes are shown tilted a bit North so Cas A passes through the beam every day.

33. CHIME is funded by the Canada Foundation for Innovation (CFI)

34. The End

35. 35 Visibility for a representative pair of feeds. Sky signal is periodic, T = 1 day. No mode mixing of m modes (in celestial coordinates). Penticton BC

36. K-L decomposition of the mode spectrum showing HI to foreground ratio in mode space: (l,ν) vs. m.

37. Pen et al. have GBT data over ~20 deg 2, ~700-900 MHz. Raw data shown on a 3 K scale @ 900 MHz. Foregrounds: GBT Study - I

38. Foregrounds: GBT Study - II Simulated large-scale structure signal (CDM) @ z=0.579. Shown on a ~1 mK scale @ 900 MHz.

39. Empirical component separation employs principal component analysis to reject dominant source of variance. Foregrounds: GBT Study - III SVD mode spectrum Data cube (schematic) Mode number w/w max

40. Foregrounds: GBT Study - IV Cross-correlate with WiggleZ as a function of SVD rejection.

41. Foregrounds: GBT Study - V Mass variance vs. |k| for 3 mode cuts - fairly robust limit. GBT team claims consistent spectrum from auto- correlation.

42. We have installed CHIME feeds and amplifiers on a two 8 m dish system at the DRAO to gain experience with real signals and real RFI. The dishes are coupled to a 4-input digital correlator (out of view) via 50 m cables.

43. The baseline design for a feed for CHIME is a four-square dual polarization system, made of circuit board materials. Here is an early test of our polarized beam pattern.

44. Feed pattern measured by Meling Deng, UBC Feed pattern calculated by Ivan Lima-Padilla, McGill The feed patterns are broad, and will fill a cylinder with modest edge taper. The model and data below are contributions from 2 undergraduates.

45. The development of low-noise, room temperature transistors has made CHIME possible. The transistor in this circuit costs $3 - much less than the box, filters, connectors…. Amplifier design and test: Greg Davis, M.sc., UBC

46. A prototype 8-channel ADC board for CHIME, designed and tested at McGill, based on 1.25 GS/s ADC chips. We alias sample at 850 Ms/s which allows us to use lower speed analog-to-digital converters than would be required otherwise. (See next slide.)

47. (~1 μs integration) The DRAO site provides an excellent RFI environment. Over the CHIME pass-band, the only RFI visible above our noise is the synthesis telescope LO, which we can turn off. This spectrum was obtained with a CHIME feed and amplifier, with considerable integration time.

48. We alias sample the signal (in the second Nyquist zone.) When we sample near the top frequency of our filter, the entire pass band appears folded into the range 0 to f sample /2. Here we sample at 3 different frequencies to test whether out-of- band lines are folded into our band. They are not.

49. Cas A fringes from the 2-dish system over the full CHIME band. These demonstrate: - That our prototype digital correlator works. It is based on a similar ADC and firmware planned for the full CHIME. - That our band-defining filters have very well-matched phase delays, as demonstrated by the flat phase curves at null.

50. Tilting the dishes a bit more each day, we made small drift scan maps at each of 1000 frequencies. Noise Budget at 600 MHz Amplifier 35K Feed losses* 24K Mismatch 3K CMB 3K Beam Spill* 15K Total 80K Measured 80K (a coincidence!) Goal 50 K *Items targeted for further improvement: - We have built new Teflon feeds whose modeled noise is 7K. - The full CHIME cylinders are level with the feeds, reducing ground spill.

51. 51 Conclusions: - CHIME is a “digital” radio telescope optimized for fast mapping speed over the band ~400-800 MHz, to map HI brightness from z ~ 0.8-2.5 - CHIME can produce a Dark Energy FoM comparable to Euclid/BAO at a fraction of the cost. - We have a 2-dish prototype in operation at DRAO and we’re developing a 2-cylinder pathfinder now. - CHIME was the highest-rated medium scale ($5- 30M)project in Canada’s 2011 LRP. A proposal to CFI is pending.

52. title people CHIME - the Canadian Hydrogen Intensity Mapping Experiment Mark Halpern Gary Hinshaw Kris Sigurdson Tom Landecker Dick Bond Ue-Li Pen Keith Vanderlinde Matt Dobbs David Hanna UBC DRAO Toronto McGill

53. title people CHIME - the Canadian Hydrogen Intensity Mapping Experiment UBC McGill Greg Davis Mandana Amiri Meiling Deng Mike Sitwell Don Wiebe Ze Fu Yin-Zhe Ma Richard Shaw Kyoshi Masui Keith Vanderlinde Kevin Bandura Ivan Lima-Padilla Adam Gilbert Jean Francois Cliché Juan Mena Perra Nicholas Lebel-Buchanon Toronto

54. The measured RF Interference at the proposed CHIME site at the DRAO in Penticton looks promising. Measured December 2009 Measurement resolution is 10 kHz. The cell phone band at 850 GHz defines CHIME’s top frequency. Test noise floor is 400K. Peterson/Landecker/Bandura

55. Figure of merit for an HI measurement from z min to z max Pushing BAO measurements to low redshift drives the ultimate sensitivity of a survey, down to redshifts reachable from existing optical surveys. The FoM for z max saturates above z~2.5 and site RFI starts to dominate at low frequencies.

56. We have built dual polarization feeds using printed circuit techniques and measured their response and coupling parameters.

57. A waterfall plot of the real part of one cross product for the time that Cas A is in the beams.

58. Real and imaginary parts of the cross product at one frequency. (This is a vertical slice through the previous graph.) So, the interferometer works, but this signal is too large to see our noise.

59. A “null” cross product,. Notice that our noise, the variance of this null product, is time dependent. The measurement noise is dominated by the sky temperature, not our amplifiers.

60. Composition - and hence expansion rate - changes with time (redshift), depending on equation of state of constituents, including Dark Enrgy BAO as a probe of Dark Energy

61. 61 BAO the Movie: evolution of a density peak Credit: SDSS Collaboration (co-moving coordinates)

62. Animation: SDSS Collaboration