The notebooks are stored with their outputs, so the results can be viewed here on GitHub without running anything. Each one also opens in Colab, where its first cell installs PyBDR.
| notebook | content | runtime | |
|---|---|---|---|
| colab_demo | collision verification: can a car changing lanes from a non-convex set of initial states hit an obstacle? |  |
1 min |
| asb2008cdc | nonlinear systems, conservative linearization (14 cases) | |
11 min |
| alth2013hscc | nonlinear systems, conservative polynomialization (5 cases) | |
4 min |
| gira2005hscc | linear systems with zonotopes (7 cases) | |
12 min |
| alk2011hscc | linear systems with uncertain inputs | |
30 s |
| xse2016cav | under-approximation of backward reachable sets | |
30 s |
The runtimes were measured on an Apple Silicon Mac; Colab (2 CPUs) is slower.
python scripts/run_checks.pyruns the tests and all notebooks (about 30 minutes), saves the outputs of the notebooks that run
through and writes reports/summary.md. Notebooks that fail are saved to reports/failed/ with the
error, the stored version stays unchanged. Pass notebook paths to run only those, or
--skip-notebooks for the tests only.