9.s916 Course Materials

This repository contains the course materials for 9.s916, a course on probabilistic programming and Bayesian methods taught by Vikash Mansinghka at MIT.

Course Resources

MIT students can access course videos, assignments, and the full syllabus on the course Canvas page.

Setup

This repository uses pixi for environment management. To get started:

1. Clone and navigate to the repository:

git clone https://github.com/ChiSym/9.s916-course-materials.git
cd 9.s916-course-materials

2. Run scripts/bootstrap-env.sh to install pixi and create the Python virtual environment
3. Restart your terminal (or re-source your shell config) to ensure pixi is on your PATH

IDE Setup

If using VS Code or Cursor:

1. Run the "Python: Select Interpreter" command (Ctrl/Cmd + Shift + P)
2. Select .pixi/envs/default/bin/python as your Python interpreter
3. When opening Jupyter notebooks, select the "default" kernel

Repository Contents

The src directory contains the following materials:

Localization Tutorial

A comprehensive tutorial on probabilistic robot localization implemented in JAX and Gen:

• localization-tutorial.py - The source file in Jupytext percent format
• localization-tutorial.ipynb - The Jupyter notebook generated from the .py file

The tutorial demonstrates:

• Modeling robot motion and sensor observations
• Implementing particle filtering for state estimation
• Visualizing robot paths and particle distributions
• Working with probabilistic programming concepts in JAX/Gen

The notebook can be run interactively using JupyterLab or any Jupyter-compatible editor:

# Run JupyterLab on CPU
pixi run lab

# Run JupyterLab on GPU
pixi run -e gpu lab

License

The course materials are licensed under the MIT license. See the LICENSE file for details.