Mixture modeling with PixelPop and parametric methods for GW population inference on GWTC-4

The origins of merging compact binaries observed by the LIGO–Virgo–KAGRA (LVK) gravitational-wave (GW) detectors remain uncertain, with multiple formation channels likely contributing to the overall merger rate. Current understanding of GW sources relies heavily on simplified parametric models, which are inspired by astrophysical expectations, but make strong assumptions about the population and are prone to bias due to model misspecification. Recent advances have introduced more flexible methods, like PixelPop – a high-resolution Bayesian nonparametric population model designed to infer joint distributions and parameter correlations with minimal assumptions. However, the increased flexibility comes at the expense of larger statistical uncertainties and reduced astrophysical interpretability. In this work, we leverage the benefits of PixelPop and the strong models in tandem and build a mixture model framework, which we apply to GWTC-4.