dynamodels

A Model couples a governing equation to a pre-allocated state history and a pluggable time integrator. Every physical model in the package — three low-order oscillators, two spatially extended PDEs, and the Lorenz systems — implements the same interface:

• time_integrate(Nt) advances the state,
• get_observable_hist() returns what a sensor would measure, and
• visualize_history() plots both.

The package itself depends only on numpy, scipy, matplotlib and typeguard.

Where to go

• Getting started — install, the five-line quickstart, the Model interface.
• Example usage — the Model interface end to end in an executed notebook: integration, the history buffer, observables, ensembles and integrators.
• Available Models — one page per physical model, each with the governing equations and a figure of its time evolution: Lorenz63, Lorenz96, Van der Pol, Rijke tube, Annular combustor, Kuramoto-Sivashinsky (1-D), Kuramoto-Sivashinsky (2-D), Kuznetsov oscillator.
• API reference — Model, Integrator, HistoryTracker, utils.
• Learn more — Tutorials on thermoacoustics, the Rijke tube and annular combustor notebooks; Characterizing a model, nonlinear time-series diagnostics with ntsa, the sibling package built on this one.

Ecosystem

• ntsa — nonlinear time-series analysis (Lyapunov exponents, delay embeddings, regime classification, bifurcation sweeps) for any model that follows the dynamodels protocol.
• romda — bias-aware ensemble data assimilation built on top of both.

Citing

If a model in this package is central to your work, please cite its original reference, listed on that model's page. To cite the package itself:

@software{novoa2026dynamodels,
  author    = {N{\'o}voa, Andrea},
  title     = {dynamodels -- dynamical systems models},
  year      = {2026},
  version   = {v0.1.0},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.21843589},
  url       = {https://doi.org/10.5281/zenodo.21843589},
}