[HACK] Implementing a differentiable Halo Occupation Distribution

[EiffL]

Implementing a differentiable Halo Occupation Distribution

This hack aims at building a TensorFlow Probability implementation of the Zheng et al. (2007) model. The goal is to have a differentiable model to populate simulations with galaxies.

Contacts: @EiffL @changhoonhahn @modichirag
Participants:

Goals and deliverable

• : Implement the Zheng et al. model in TFP
• : Sample a mock galaxy catalog from a halo catalog
• : Demonstrate variational inference of the model parameters

The deliverable is a notebook demonstrating the practicality of the method.

Resources needed

What's need is some TensorFlow knowledge, some experience with HODs, and halotools, and a lot of enthusiasm :-)

Detailed description

Prototype for the project is being hosted here: https://github.com/DifferentiableUniverseInitiative/DHOD

Chat about the project is here: