Presentation #315.03 in the session Cosmology.
Measuring the power spectrum of 21 cm emission from neutral hydrogen would probe large-scale structure across a large redshift range, providing insight into cosmic evolution through the Dark Ages, Cosmic Dawn, Epoch of Reionization, and into the modern universe. These measurements would constrain early galaxy formation, Dark Matter interactions, and the universe’s expansion history. The promise of 21 cm cosmology has driven development of several radio interferometric experiments in recent years, but none have successfully measured the emission’s power spectrum. These experiments have the needed sensitivity but are limited by their ability to control measurement systematics that mix the 21 cm signal with the bright intervening foreground emission. We explore the capacity of next-generation 21 cm instruments to overcome systematic error and achieve the promise of 21 cm cosmology.
Uncertainties in the instrument response constitute a dominant source of systematics in 21 cm analyses. Precision signal reconstruction involves unraveling the instrument’s response to the incident signal, but this requires exquisite understanding of the instrument. Analyses must know to great accuracy both the complex, direction-independent antenna gains, typically constrained in calibration, and the direction-dependent beam responses. Any errors in these quantities introduce spectrally variant errors in the reconstructed signal, precluding the spectral filtering required to differentiate the 21 cm signal from the foregrounds. We show that pseudo-random arrays with many antennas—called “large-N arrays”—have reduced sensitivity to calibration and beam errors.
Novel data storage and processing techniques for radio interferometry currently enable larger arrays with up to thousands of fully cross-correlated elements. We explore the impact of these arrays on 21 cm measurements using high-fidelity simulations from which we quantify the relationship between uv plane sampling density and measurement systematics. We explore calibration error, beam modeling error, and mode-mixing for various array configurations, showing that large-N arrays are comparatively resilient to measurement systematics.