Workflows Overview

End-to-end automated pipelines for analyzing chaotic systems.

The workflows module provides high-level orchestrators. Instead of manually calling solvers, generating seeds, and classifying trajectories, you can use these workflows to run complete experiments with a single command.

Most workflows can be run either from the command line interface (CLI) or programmatically via Python.

Core Workflows

1. Unified Chua Workflow: The primary pipeline for fractional-order Chua circuits. It generates seeds, integrates trajectories, classifies basins, and saves JSON summaries.
2. Robustness Overlay: Sweeps a single parameter (like $\alpha$) and plots the distance from the attractor to the equilibria over time, tracking the emergence and collapse of hidden attractors.
3. Integer Lur’e Workflow: A generic $q=1$ pipeline for systems supplied in Lur’e form. The non-smooth Chua reference case uses the classical describing-function branch and is documented with promoted validation evidence.

The validated reference for this integer workflow is Chua Integer q=1.

Secondary Workflows

These workflows are typically run after a primary workflow has identified a HIDDEN_CANDIDATE.

1. Sphere Controls: Densely samples a sphere around an equilibrium to ensure no narrow basin connection exists.
2. Refined Basin: Takes a known hidden attractor and tests a dense grid of points between the attractor and the nearest equilibrium.
3. Strict Refinement: Verifies the topological robustness of a candidate using high-accuracy ABM integration instead of EFORK.

Output Standard

All workflows output their results using a standardized JSON schema. This allows secondary scripts (like plotting utilities or summary generators) to easily parse the output of any workflow run, regardless of which specific pipeline generated the data.