Radio pulsar population synthesis with consistent flux measurements using simulation-based inference

The properties of isolated Galactic radio pulsars can be inferred by modelling their evolution, from birth to the present, through pulsar population synthesis. This involves simulating a mock population, applying observational filters, and comparing the resulting sources to the limited subset of detected pulsars. We specifically focus on the magneto-rotational properties of Galactic isolated neutron stars and provide new insights into the intrinsic radio luminosity law. To better constrain the intrinsic radio luminosity, for the first time in pulsar population synthesis studies, we incorporate data from the Thousand Pulsar Array program on MeerKAT, which contains the largest unified sample of neutron stars with consistent flux measurement to date. In particular, we employed a simulation-based inference technique called Truncated sequential neural posterior estimation (TSNPE) to infer the parameters of our pulsar population model. This technique trains a neural density estimator on simulated pulsar populations to approximate the posterior distribution of underlying parameters. This method efficiently explores the parameter space by focusing on regions most likely to match the observed data, significantly reducing the required training dataset size. We find that adding flux information as an input to the neural network significantly improves the constraints on the pulsars’ radio luminosity and improves the estimates on other input parameters. Moreover, we demonstrate the efficiency of TSNPE over standard neural posterior estimation, as we achieve robust inferences of magneto-rotational parameters consistent with previous studies while using only around 4% of the simulations required by NPE approaches.