/Probabilistic_Programming_CDT

Probabilistic Programming Lectures for the CDT Workshop September 13-17 2021 Durham

Primary LanguageJupyter NotebookMIT LicenseMIT

Durham University STFC CDT Data Science Workshop

Probabilistic Programming Lectures for the STFC CDT Workshop (September 13-17 2021) hosted by Durham University.

These lectures use the probabilistic programming language PyAutoFit (https://github.com/rhayes777/PyAutoFit).

Probabilistic Programming 1: Fitting a Model to Data

In this lecture, you will learn how to compose a probabilistic model and perform inference on data by defining a likelihood function, which is sampled using a non-linear search algorithm.

The slides for this section can be viewed at the following link (part 1 is covered between the first slide and the slide title "Cosmology: Strong Gravitational Lensing"):

https://github.com/Jammy2211/Probabilistic_Programming_CDT/blob/main/PyAutoFitCDT.pdf

Lecture materials can be found using the link below (which run in a web browser without any user setup required):

https://mybinder.org/v2/gh/Jammy2211/Probabilistic_Programming_CDT/main?filepath=lecture_1.ipynb

Probabilistic Programming 2: Application To Astronomy

This lecture uses strong gravitational lensing to demonstrate how adopting a probabilistic programming language can benefit scientific research and shows advanced model composition using multi-level models can create an extensible and easy to maintain code base.

The slides for this section can be viewed at the following link (part 2 is covered between the slide title "Cosmology: Strong Gravitational Lensing" and "PyAutoFit: Advanced Features"):

https://github.com/Jammy2211/Probabilistic_Programming_CDT/blob/main/PyAutoFitCDT.pdf

Lecture materials can be found using the links below (which run in a web browser without any user setup required):

https://mybinder.org/v2/gh/Jammy2211/Probabilistic_Programming_CDT/main?filepath=lecture_2_part_1.ipynb

(There is no part 2)