Taylor Model Enabled Reachability Analysis Framework.
TERA is a Python-native, open-source Taylor Model (TM) reachability framework for computing rigorous flowpipe enclosures of nonlinear dynamical systems under uncertainty. It unifies continuous, hybrid, and stochastic analysis within a single symbolic–numeric workflow to support rapid prototyping without sacrificing enclosure guarantees.
- Validated TM flowpipe computation for nonlinear ODEs
- Hybrid reachability
- Stochastic reachability
-
Visualization utilities
- Plotting/projection tools for deterministic flowpipes and
$\delta$ -PRS enclosures, plus comparisons against Monte Carlo samples.
- Plotting/projection tools for deterministic flowpipes and
- Examples and notebooks for continuous, hybrid, and stochastic workflows and benchmarks.
-
TERA/TMCore— core algebra: intervals, polynomials, Taylor models, TM vectors. -
TERA/TMFlow— continuous reachability: flowpipe construction, remainder handling, preconditioning (single-step + left–right). -
TERA/Hybrid— hybrid automata: modes, invariants, guard intersection, resets, worklist orchestration. -
TERA/Stochastic— stochastic reachability and$\delta$ -PRS workflows; sampling utilities for comparison. -
TERA/Workbench— visualization and result helpers. -
Examples— Jupyter notebooks for end-to-end usage.
See INSTALL.txt for up-to-date installation instructions. SageMath is required for core functionality, and installation is currently documented as an editable install inside a SageMath environment.
Note: this project is not currently pip-installable; please use the SageMath environment workflow described in INSTALL.txt.
Run one of the example notebooks in Examples/:
-
Continuous_*.ipynb
-
Hybrid_*.ipynb
-
Stochastic_*.ipynb
For full configuration details, see:
Docs/Configuration_Ref.md and Docs/Workbench_API.md.
Research prototype. APIs and configuration knobs may evolve as the engine matures. Current focus is consolidating hybrid + stochastic components into a unified SHS formulation. CI tests are planned for a future release (not yet enabled).
Notebook size: example notebooks in Examples/ include saved outputs. If repo size matters, clear notebook outputs before pushing.
This project is licensed under the GNU General Public License v2.0 (GPL-2.0). TERA is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License, version 2 (GPL-2.0). A copy of the license is provided in LICENSE.GPL2.




