A novel method using normalizing flows to efficiently estimate the Bayesian evidence and its numerical uncertainty from a set of posterior samples.
