nviebig/LorenzParameterEstimation: LorenzParameterEstimation.jl v0.1.0
Fuente:
Zenodo
Guardado en:
| Autor principal: | |
|---|---|
| Formato: | Recurso digital |
| Publicado: |
Zenodo
2026
|
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866901128211333120 |
|---|---|
| author | Niklas Viebig |
| author_facet | Niklas Viebig |
| contents | <p>First public release of <code>LorenzParameterEstimation.jl</code>, a Julia package for gradient-based parameter estimation in the Lorenz-63 chaotic dynamical system. Developed as part of a master thesis on parameter estimation in chaotic systems.</p> <h3>Features</h3> <ul> <li><strong>Two paradigms</strong> for parameter recovery:<ul> <li><em>Weather approach</em> — windowed trajectory loss with teacher forcing to handle chaotic divergence over the Lyapunov time</li> <li><em>Climate approach</em> — statistical matching via mean, PDF, and Wasserstein distance</li> </ul> </li> <li><strong>Automatic differentiation</strong> via <a href="https://github.com/EnzymeAD/Enzyme.jl">Enzyme.jl</a> for exact gradient computation through the ODE integrator</li> <li><strong>Modern training API</strong> (<code>modular_train!</code>) with mini-batch SGD, early stopping, and configurable window/stride parameters</li> <li><strong>Multiple optimizers</strong> — Adam, AdamW, SGD, Adagrad via <a href="https://github.com/FluxML/Optimisers.jl">Optimisers.jl</a></li> <li><strong>Recovery of all three Lorenz-63 parameters</strong> σ, ρ, and β simultaneously</li> <li>Comprehensive test suite and example notebooks</li> </ul> <h3>Package info</h3> <ul> <li>Julia ≥ 1.9</li> <li>UUID: <code>3ba9dec7-690f-4f7f-8c1b-5ed383f34073</code></li> <li>License: MIT</li> </ul> <h2>What's Changed</h2> <ul> <li>Fuse train statistics by @nviebig in https://github.com/nviebig/LorenzParameterEstimation/pull/1</li> </ul> <h2>New Contributors</h2> <ul> <li>@nviebig made their first contribution in https://github.com/nviebig/LorenzParameterEstimation/pull/1</li> </ul> <p><strong>Full Changelog</strong>: https://github.com/nviebig/LorenzParameterEstimation/commits/v0.1.0</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_19052322 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | nviebig/LorenzParameterEstimation: LorenzParameterEstimation.jl v0.1.0 Niklas Viebig <p>First public release of <code>LorenzParameterEstimation.jl</code>, a Julia package for gradient-based parameter estimation in the Lorenz-63 chaotic dynamical system. Developed as part of a master thesis on parameter estimation in chaotic systems.</p> <h3>Features</h3> <ul> <li><strong>Two paradigms</strong> for parameter recovery:<ul> <li><em>Weather approach</em> — windowed trajectory loss with teacher forcing to handle chaotic divergence over the Lyapunov time</li> <li><em>Climate approach</em> — statistical matching via mean, PDF, and Wasserstein distance</li> </ul> </li> <li><strong>Automatic differentiation</strong> via <a href="https://github.com/EnzymeAD/Enzyme.jl">Enzyme.jl</a> for exact gradient computation through the ODE integrator</li> <li><strong>Modern training API</strong> (<code>modular_train!</code>) with mini-batch SGD, early stopping, and configurable window/stride parameters</li> <li><strong>Multiple optimizers</strong> — Adam, AdamW, SGD, Adagrad via <a href="https://github.com/FluxML/Optimisers.jl">Optimisers.jl</a></li> <li><strong>Recovery of all three Lorenz-63 parameters</strong> σ, ρ, and β simultaneously</li> <li>Comprehensive test suite and example notebooks</li> </ul> <h3>Package info</h3> <ul> <li>Julia ≥ 1.9</li> <li>UUID: <code>3ba9dec7-690f-4f7f-8c1b-5ed383f34073</code></li> <li>License: MIT</li> </ul> <h2>What's Changed</h2> <ul> <li>Fuse train statistics by @nviebig in https://github.com/nviebig/LorenzParameterEstimation/pull/1</li> </ul> <h2>New Contributors</h2> <ul> <li>@nviebig made their first contribution in https://github.com/nviebig/LorenzParameterEstimation/pull/1</li> </ul> <p><strong>Full Changelog</strong>: https://github.com/nviebig/LorenzParameterEstimation/commits/v0.1.0</p> |
| title | nviebig/LorenzParameterEstimation: LorenzParameterEstimation.jl v0.1.0 |
| url | https://doi.org/10.5281/zenodo.19052322 |