CANDEL

CANDEL (CAlibration and Normalization of the DistancE Ladder) is a JAX-based framework for peculiar-velocity inference and distance-ladder calibration.

CANDEL forward-models distance-indicator observables and redshift while marginalising over latent variables such as distance and absolute magnitude. Posterior sampling uses the No-U-Turn Sampler (NUTS) from NumPyro, with JAX providing automatic differentiation and JIT compilation throughout.

Contents

• Installation
• Supported models
  • Peculiar-velocity models
  • \(H_0\) inference
  • Package structure
• API Reference
  • candel
  • candel.model
  • candel.pvdata
  • candel.cosmo
  • candel.field
  • candel.redshift2real
  • candel.mock