ANUBIS - Astrophysical Nonparametrically Upgraded Bayesian Inference of Subpopulations

ANUBIS implements the inference of a multivariate probability density given a set of samples and one or more parametric models. The Augmented Mixture Model, or AMM, is a finite mixture of parametric models augmented with a DPGMM to capture the potential features missed by the parametric models:

$$p(x) \simeq \sum_i^N w_i f_i(x|\Lambda_i) + \big(1-\Sigma_i^N w_i \big) \sum_j^\infty \phi_j \mathcal{N}(x|\mu_j,\sigma_j)$$

Please install this code with pip install anubis (stable version) or pip install . after cloning this repository (development version, potentially unstable).