Welcome to the Generalized Notation Notation (GNN) documentation hub!
The doc/ directory is vast, encompassing a 25-step script pipeline, theoretical foundations, deep-dive implementation guides, cognitive phenomena models, and integrations across 7 computational frameworks.
This document serves as your Master Table of Contents, helping you quickly locate exactly what you need.
If you are a...
- Newcomer: Start with the GNN Overview and build your first model in 15 minutes with the Quickstart Tutorial.
- Researcher: Read the core theories in About GNN and explore Cognitive Phenomena.
- Developer/Architect: Dive straight into the Pipeline Architecture and learn about the "Thin Orchestrator" that governs the 25 processing steps.
- System Integrator: See the Framework Integration Guide and explore the 131 available tools in the MCP Hub.
Here is how the doc/ directory is organized at a high level. Use these links to jump directly to specific domain areas.
The doc/gnn/ subdirectory contains the immediate specifications, guides, and internal workings of the GNN modeling language itself. Start at the GNN README.
- Tutorials (
doc/gnn/tutorials/): Step-by-step guides and model progressions. - Reference (
doc/gnn/reference/): Strict technical specifications (syntax, architecture, DSL manuals, type systems). - Integration (
doc/gnn/integration/): Code generation, framework pipelines, and multi-format export. - Advanced (
doc/gnn/advanced/): Multi-agent systems, LLM/Neurosymbolic combinations, and ontology processing. - Operations (
doc/gnn/operations/): Resource metrics, troubleshooting, tooling, and improvement analysis.
Documents explaining how to run, scale, and fix the processing pipeline.
- Execution (
doc/execution/): Framework availability checks and multi-environment orchestration. - Troubleshooting (
doc/troubleshooting/): Master FAQ, error taxonomy, and resolutions for specific script warnings. - Dependencies (
doc/dependencies/): Core vs. optional libraries, and Julia/Python interoperability.
Deep-dive implementations for how GNN translates to executable code in specific libraries.
- PyMDP (
doc/pymdp/): Python-based POMDP solvers. - RxInfer (
doc/rxinfer/): Julia-based reactive message passing. - ActiveInference.jl (
doc/activeinference_jl/): Julia-based Active Inference solvers. - JAX (
src/render/jax/): GPU-accelerated render templates (see also integration guide). - DisCoPy (
doc/discopy/): Category theory and string diagrams. - PyTorch (
src/render/pytorch/): Deep learning render templates. - NumPyro (
src/render/numpyro/): Probabilistic programming render templates. - CatColab (
doc/catcolab/): Categorical compositional intelligence.
- MCP Hub (
doc/mcp/): 131 Model Context Protocol tools to interface with Claude/Cursor/LLMs. - GUI & Visualization (
doc/visualization/): Advanced multi-layer 3D network graphing, matrix heatmaps, and frontend interfaces. - LLM Processing (
doc/llm/): Workflows detailing how GNN operates in a neurosymbolic pipeline.
- Cognitive Phenomena (
doc/cognitive_phenomena/): Theoretical maps to concepts like attention, emotion, learning, depression, and meta-cognition. - Templates (
doc/templates/): Quick-start boilerplate models for building your own GNN specifications.
If you only read three documents in this entire repository, read these:
- GNN Syntax Reference: Master the domain-specific markdown to write expressive models.
- Pipeline Orchestrator: Understand how
src/main.pyprocesses your text file through 25 distinct scientific steps. - GNN Implementations: Explore how those 25 steps ultimately produce real Python/Julia files that evaluate mathematically rigorous Active Inference simulations.
Welcome to GNN.