agencitylab 0.1.11

Scientific framework for the theory of Agencity

AgencityLab

<<<<<<< HEAD

AgencityLab is a scientific Python framework dedicated to the study and quantification of Agencity — an observable designed to measure a system’s ability to actively structure its own evolution.

Unlike classical metrics (variance, entropy, energy), Agencity captures dynamic organization and decision-like behavior within a system, independent of domain-specific assumptions.

Agencity can be analyzed:

• at a single scale
• across multiple scales via the Agencity spectrum ( b(\tau) )

---

👤 Author

Gilbert BEMWIZ
Conceptor of the Agencity theory
Engineer and developer of the AgencityLab framework

---

🧠 Vision

AgencityLab aims to provide:

• a rigorous computational implementation of Agencity theory
• a domain-independent analysis tool applicable to:
  • physical systems
  • biological signals
  • economic dynamics
  • computational processes
• an experimental platform to explore links with:
  • Shannon information theory
  • Landauer’s principle
  • informational physics hypotheses (e.g. Vopson)

---

🔬 Core Concept

Agencity is not a measure of randomness or order.

It quantifies:

the capacity of a system to select and sustain a trajectory among those allowed by its underlying dynamics

This leads to a hierarchy of observables:

u → u* → X* → A* → M → O → β → b

Where:

• u : raw signal
• u* : normalized signal
• X* : activation
• A* : activity
• M : memory (causal correlation)
• O : organization
• β : structured agency
• b : Agencity (final observable)

---

⚙️ Installation

pip install agencitylab

Optional features:

pip install agencitylab[viz]
pip install agencitylab[app]
pip install agencitylab[ml]
pip install agencitylab[all]

---

🚀 Quick Start

import numpy as np
from agencitylab.api import compute_agencity

xi = np.linspace(0, 10, 200)
u = np.sin(xi)

result = compute_agencity(xi, u)

print(result.summary())

---

📊 Visualization

from agencitylab.api import visualize_agencity

visualize_agencity(result, kind="timeseries")

Available visualizations:

• timeseries → full pipeline signals
• spectrum → frequency structure of b
• phase → dynamic trajectories
• components → internal decomposition

---

📈 Multi-Scale Analysis

Agencity can be analyzed across scales:

from agencitylab.analysis.multi_scale import agencity_spectrum

taus = [0.5, 1, 2, 5, 10]
spec = agencity_spectrum(xi, u, taus)

This produces a multi-scale signature of the system.

---

🧬 Agencity Signature

Each system exhibits a characteristic Agencity signature:

• constant → inactive
• sinus → structured dynamics
• noise → high variability, low organization
• multi-scale → emergent organization

This enables:

• system classification
• clustering
• regime detection

---

🧪 Example Systems

AgencityLab can analyze:

• periodic signals
• stochastic processes
• hybrid signals (signal + noise)
• multi-scale dynamics

---

📊 Comparison with Classical Metrics

Agencity complements traditional measures:

Metric	Captures	Limitation
Variance	dispersion	ignores structure
Entropy	randomness	ignores dynamics
Energy	amplitude	ignores organization
Autocorr	temporal dependence	local only
Agencity	dynamic structuring ability	multi-scale, causal, intrinsic

---

🧠 Interpretation

• A zigzag trajectory is not noise if imposed by constraints
• A system is not “ordered” or “disordered” absolutely
• Agencity is contextual and intrinsic to system dynamics

---

📁 Project Structure

agencitylab/
├── core/            # mathematical operators
├── analysis/        # spectrum, signatures, clustering
├── api/             # user-facing interface
├── visualization/   # plotting tools
├── examples/        # usage examples

---

🔭 Roadmap

• Advanced multi-scale theory
• Real-world datasets integration
• GPU acceleration
• Interactive dashboards
• Scientific publication support

---

📜 License

MIT License

---

🤝 Contributions

Contributions are welcome.

You can:

• open issues
• propose improvements
• extend analysis modules
• test on new domains

---

⚠️ Disclaimer

Agencity is an emerging theoretical framework. This library provides a computational implementation intended for:

• research
• experimentation
• conceptual exploration

---

🌍 Perspective

AgencityLab is part of a broader effort to build:

a unified science of dynamic systems combining physics, information, and structure ======= gencityLab** is a scientific Python framework dedicated to the study and quantification of Agencity — an observable designed to measure a system’s ability to actively structure its own evolution.

7fdb362e5a99ca437937b5eb65e42c67b4ba6f62