Adaptive proposals for nonparametric equation-of-state inference with RIFT: iterative Gaussian-process refinement in a compressed sequence representation

Abstract

Nonparametric equation-of-state (EOS) inference evaluates the likelihood on a finite ensemble drawn from a prescribed functional prior. Reweighting that ensemble suffices for the binary neutron stars detected so far, and loses effective sample size only once the likelihood concentrates --- which needs third-generation catalogues, or extra electromagnetic and nuclear-theory information folded into the same target. We develop and validate an adaptive-proposal procedure that leaves the base prior f…