Lecture 4 — Reproducible Python environments

Thu 15 Oct 2026 · Boris Bolliet

Slides Notes Self-test Recording

Demonstration files

File	View	Download	Colab
README.txt
analyse.py
ask_llm.py
check_env.py
period.py
requirements-lock.txt
requirements.txt
setup_env.sh

curl -LO https://teaching.borisbolliet.io/c1-2026/materials/class04-demo.zip && unzip class04-demo.zip

The scripts are those of the public repository borisbolliet/attractor (MIT). The ten chemostat runs it analyses, C1.csv to C10.csv, are the experiments of Blasius, Rudolf, Weithoff, Gaedke and Fussmann, Long-term cyclic persistence in an experimental predator-prey system, Nature 577, 226–230 (2020), data set on figshare doi:10.6084/m9.figshare.10045976 (CC BY 4.0).

Content

1. Layers of an environment
   • The three layers of an environment
   • Virtual environments
   • Specification and snapshot
2. Package versions
   • Version specifiers
   • Breaking releases and interpreter-bound wheels
3. Configuration outside the code
   • Configuration in environment variables
   • The .env file and python-dotenv
   • A service address in the configuration
4. Rebuilding the environment
   • The setup script
   • The import search path
   • Jupyter kernels
   • uv, conda and Colab

Exercises with solutions