Metadata-Version: 2.4
Name: serie
Version: 0.0.1
Summary: Stress-test time-series forecasters: known dynamical systems, controlled shocks, calibrated noise titration (the AURA protocol). Placeholder release; 0.1.0 in preparation.
Author: Qilin Wang
License-Expression: MIT
Project-URL: Homepage, https://github.com/QilinWang
Project-URL: Repository, https://github.com/QilinWang
Keywords: time-series,forecasting,benchmarking,dynamical-systems,robustness,chaos
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Science/Research
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Scientific/Engineering
Requires-Python: >=3.12
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

# serie

**Stress-test your forecaster before reality does.**

`serie` will implement the **AURA protocol** (Artificial Uncertainty and
Regime Alteration): take a clean reference series — a simulated dynamical
system with exact ground truth, or a clean recording treated as the
system — apply controlled non-stationary shocks, and inject calibrated
Gaussian observation noise at chosen levels. Measure exactly *when and how*
a forecaster breaks, instead of ranking models on one recorded history.

```python
import serie

stock = serie.prepare("lorenz", seed=42)
for sigma in (0.0, 0.25, 1.0, 2.0):
    split = serie.titrate(stock, sigma=sigma)
    ...
```

**This is a 0.0.1 name reservation — the API above ships in 0.1.0.**
Watch [github.com/QilinWang](https://github.com/QilinWang) for the release.

MIT license. Author: Qilin Wang.
