gwkokab.analysis.subpopulation

The sub-population analysis family.

A mixture over three mass families – a truncated power law (spl), a broken power law (bpl) and a truncated normal (gpl) – mixed at the level of the primary mass rather than of the whole component. The primary mass models are combined into a single mixture with branching fractions lambda_<i>, tapered by a shared smoothing window and normalised, and one overall log_rate is then split across the components in proportion to their weights.

So where multisource has one rate per component, this family has one rate for the population and a set of branching fractions – the natural parameterisation when the components are sub-populations of a single channel rather than separate channels.

Console scripts

  • discrete_subpopulation – infer the population from per-event posterior samples.

  • analytical_gwalk_subpopulation – infer it from per-event Gaussian summaries.

The sampler backend is chosen at runtime by sampler_cfg.json, not by the entry point.

See also

gwkokab.models.hybrids.SubPopulationModel

The population model this family fits.

gwkokab.analysis.multisource

The same component families, but mixed at the level of the whole component.

Submodules