GBT 21 cm intensity mapping collaboration Academia Sinica (Tzu-Ching Chang, Victor Yu-wei Liao) Beijing (Xuelei Chen, Yi-Chao Li) Carnegie Mellon University (Aravind Natarajan, Jeff Peterson, Tabitha Voytek) CITA/UToronto (Nidhi Banavar, Liviu Calin, Adam Lewis, Kiyo Masui, Ue-Li Pen, Richard Shaw, Eric Switzer) McGill (Kevin Bandura)
ARAA review article May 2010 (Morales and Wyithe) comparing to z ~ 10: Two months later:
IM: now and then z~1~10 Science goalBAOEoR Signal (mK)0.110 Thermal noise (K)30300 Foreground spatial fluctuation (K) Angular scale~10’+ (non-linear scale) ~10’- (bubble scale) First proposed ’s? First detected2010>2012 Strategysingle dishInterferometers
RFI Liu and Tegmark 2011
The real data
Signal-only simulations 20 Mpc Ω HI = 10 -3
Relaxed quadratic estimation Most general: Inverse covariance, down-weight foreground in both angle and frequency. Imperfect spectral baseline – need non-parametric removal: SVD on cross maps Liu & Tegmark MNRAS, Volume 419, Issue 4, pp Projecting (idempotent) LOS modes is “good enough” vs. optimal downweighting. These suggest that our method is slightly sub-optimal (in theory), but ideal in practice because we use the covariance of the real data!
SVD eigenvalues Modes falling off rapidly: few DOF to describe most of the foregrounds. Mode number (sorted) Eigenvalue
Real data, cleaned (25 modes removed here.)
Cross-pairs: GBT power ( * ( ) ) * × Map A Weight B Map B Weight A
Cross-pairs: WiggleZ crosspower ( * ( ) ) * × Map A Selection function Galaxy catalog Weight A
Preliminary
Fisher-matrix predictions
Conclusions End-to-end demonstration: calibration, ionosphere, point sources, galaxy, spectral response, etc. No show stoppers. Lessons learned for EoR. Foreground subtraction under control: ready for z~1 21cm intensity mapping. Consistent autonomous 21cm maps and power spectra. Proposal for GBT 0.5<z<1 BAO survey using new ASIAA 9 pixel receiver array starting Retires major risk for CHIME!