Event-horizon-scale Imaging of M87* under Different Assumptions Via Deep Generative Image Priors

Reconstructing images from the Event Horizon Telescope (EHT) observations ofM87*, the supermassive black hole at the center of the galaxy M87, depends on aprior to impose desired image statistics. However, given the impossibility ofdirectly observing black holes, there is no clear choice for a prior. Wepresent a framework for flexibly designing a range of priors, each bringingdifferent biases to the image reconstruction. These priors can be weak (e.g.,impose only basic natural-image statistics) or strong (e.g., impose assumptionsof black-hole structure). Our framework uses Bayesian inference withscore-based priors, which are data-driven priors arising from a deep generativemodel that can learn complicated image distributions. Using our Bayesianimaging approach with sophisticated data-driven priors, we can assess howvisual features and uncertainty of reconstructed images change depending on theprior. In addition to simulated data, we image the real EHT M87* data anddiscuss how recovered features are influenced by the choice of prior.