Phantom types for robust hierarchical models with typegeist

Shared by Gary on Nov 1, 2025 • 12:55 AM UTC

Authors:

Daniel O’Hanlon

Abstract

In Bayesian hierarchical models, group-level parameter arrays must be mapped to the observationaxis, oftenusingexplicitindexing. Incomplexmodelswithnumerousincompatible data and parameter sets, this introduces the potential for bugs, as indexing with the incorrect indices typically fails silently. H...

Subject

phantom types

typegeist

hierarchical models

Highlights

-Phantom types are introduced to enforce data-parameter-index correspondences in hierarchical models.

-Implemented in the typegeist Python package and compatible with Pyro, NumPyro, PyMC, TensorFlow Probability.

Subject

Highlights

Strengths & Weaknesses

Commercial Applications

Societal Benefits & Applications

Comments