bayesian analysis of interferometric data (arXiv: ) paul m. sutter benjamin wandelt, siddarth malu paris institute of astrophysics university of illinois at urbana-champaign

mock observations 2 signal primary beam uv-plane data

gibbs sampling 3 extract variance draw power spectrum realization construct Wiener- filtered map add fluctuations consistent with noise and spectrum

advantages 4  fast and scalable – O(n p log n p )  joint analysis of spectrum and signal  full exploration of uncertainties  automatically accounts for beam  free Wiener-filtered maps  trivial marginalization  straightforward foreground removal

results – power spectrum 5

results - map 6 posterior mean signal

results - map 7 posterior mean dirty map

results – marginalized posteriors 8

other applications 9 primary beam signal posterior mean

other applications 10

future work 11  multiple frequencies, polarization implemented  working on curved sky, foregrounds  developing point- and extended- source analysis  extending to 21 cm