Systematics from sky-pixel splitting and recombination in spectral analyses

Investigate in detail whether splitting and recombining gravitational-wave posterior samples across sky pixels in spectral-siren analyses introduces systematic effects or shifts in the inferred Hubble-constant posterior.

Background

The sampled line-of-sight redshift-prior method handles spectral-siren analyses differently from the earlier gridded-prior method. In the gridded implementation, the sky is divided into pixels and the likelihood contributions are summed across them, whereas the sampled implementation effectively treats the full sky as one pixel for spectral events. This implementation difference could affect the resulting posterior even though no systematic shift is expected a priori.

The authors identify a dedicated systematics study of this splitting-and-recombining procedure as still outstanding. Establishing its size is important for determining whether discrepancies between the sampled and gridded spectral results arise from numerical implementation choices rather than physical or statistical differences.