Description
Future gravitational wave observatories, such as the Einstein Telescope, will detect thousands of mergers between compact objects annually. By detecting the effects of tidal deformation in binary neutron star mergers, these detectors will be the dominant way to study the equation of state of ultradense nuclear matter in the coming decades. However, this requires a scalable Bayesian hierarchical inference framework that can be easily updated as soon as new detections are made. We show that sequential Monte Carlo is a natural framework for analyzing such growing datasets. Paired with GPU accelerators, this allows us to infer the equation of state of a mock catalog of one month of binary neutron star mergers detected by the Einstein Telescope within a few hours. We project that the extension to a full year of detectable sources would take our sequential Monte Carlo sampler just over a couple of days on modern hardware.