The GPry documentation

GPry was developed as a result of my master thesis and and can be seen as an alternative to established samplers like MCMC and Nested Sampling. It is targeted at a specific class of posterior distributions (with the initial goal of speeding up inference in cosmology) with the aim of creating an algorithm which can efficiently obtain marginal quantities from (computationally) expensive likelihoods.

Additionally this package is designed to communicate with the Cobaya-Package making for easy use in cosmological applications.

Deployed on PyPI

Source code on GitHub

Contents:

• Installation
  • Pre-requisites
  • Installing GPry
• Introductory example to using GPry
  • The Function
  • Building the model
  • Creating the Runner object
  • Running the model
  • Running the MCMC
  • Plotting with GetDist
  • Validation
• Advanced example to using GPry
  • Introduction
  • The Model
  • The GP Regressor
  • Acquisition
  • Convergence
  • Training parameters
  • Training
  • MCMC
  • Validation
  • Plotting with GetDist
• Running the Bayesian optimization loop
• Running with Cobaya
• Intermediate plotting, diagnostics etc.: The Callback function
  • Printing extra information
  • Intermediate MCMCs
  • MPI-aware callback function
• Modules
  • Top level module
  • Customising the BO loop
  • Tools

Indices and tables

• Index

• Module Index

• Search Page