Tutorials

Below are example notebooks demonstrating how to use PZFlow. Each contains a link to open the notebook on Google Colab, as well as a link to the source code on Github.

Basic

• Introduction to PZFlow - using the default flow to train, sample, and calculate posteriors
• Conditional Flows - building a conditional flow to model conditional distributions
• Convolving Gaussian Errors - convolving Gaussian errors during training and posterior calculation
• Flow Ensembles - using FlowEnsemble to create an ensemble of normalizing flows

Intermediate

• Customizing the flow - Customizing the bijector and latent space
• Modeling Variables with Periodic Topology - using circular splines to model data with periodic topology, e.g. positions on a sphere

Advanced

• Marginalizing Variables - marginalizing over missing variables during posterior calculation
• Convolving Non-Gaussian Errors - convolving non-Gaussian errors during training and posterior calculation