π§ Brainβ5D
Sparse 5D SpikingβNeural Simulation with Observable Plasticity
Brainβ5D is an experimental spikingβneural simulation platform operating in a sparse 5D coordinate space.
It combines spiking dynamics, spikeβtimingβdependent plasticity (STDP), homeostatic regulation, structural selfβorganization, deterministic persistence, and an operatorβfacing dashboard.
Research status
Brainβ5D is an engineering and research project. The current implementation does not claim AGI, consciousness, sentience or biological equivalence.
π― Key Features
| Feature | Description |
|---|---|
| 𧬠Spiking Core | Izhikevich regularβspiking neurons in a sparse 5D coordinate space |
| π STDP Learning | Spikeβtimingβdependent plasticity with signed eligibility traces |
| βοΈ Homeostasis | Target firingβrate regulation, adaptive thresholds, energy homeostasis |
| π Structural Plasticity | Controlled neuron/synapse creation, pruning, and persistent undo |
| π§ SelfβOrganization | Policyβdriven structural proposals with safety validation |
| πΎ Deterministic Persistence | .b5d snapshots, state delta journal, structural journal, runtime checkpoints |
| π Operator Dashboard | Realβtime visualization, heatmaps, structural approval, live projection |
| π¬ Scientific Evidence | Builtβin evidence framework (B5DβSEF) for reproducible experiments |
ποΈ Architecture Overview
βββββββββββββββββββββββ
β Operator Dashboard β
β (127.0.0.1:8765) β
ββββββββββββ¬βββββββββββ
β HTTP / Bridge
ββββββββββββΌβββββββββββ
β RuntimeController β
ββββββββββββ¬βββββββββββ
β
ββββββββββββββββββββββΌβββββββββββββββββββββ
βΌ βΌ βΌ
βββββββββββββββββββ βββββββββββββββββββ βββββββββββββββββββ
β NeuralNetwork β β HomeostasisEng. β β LearningEngine β
β (5D sparse) β β (rate/threshold)β β (STDP/eligibility)β
ββββββββββ¬βββββββββ βββββββββββββββββββ βββββββββββββββββββ
β
βΌ
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Persistence Layer β
β .b5d Snapshot + Delta Journal + Structural Journal + CP β
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
π Quick Start
Installation
git clone https://huggingface.co/<your-org>/Brain-5D
cd Brain-5D
python -m venv .venv
source .venv/bin/activate # Linux/Mac
# .venv\Scripts\Activate.ps1 # Windows
pip install -e ".[dev]"
Run the Simulation
# With dashboard (default)
python -m src.main --config configs/poc_config.yaml
# Headless mode
python -m src.main --config configs/poc_structural_live.yaml --no-dashboard --ticks 500
Open the Dashboard
Navigate to http://127.0.0.1:8765
π¦ Repository Structure
Brain-5D/
βββ src/
β βββ core/ # Neural network and neuron/synapse core
β βββ controller/ # Operator/runtime controller
β βββ learning/ # STDP, eligibility and reward learning
β βββ homeostasis/ # Firing-rate / threshold / energy regulation
β βββ self_organization/ # Policy, coordinator, approval, plasticity
β βββ manipulation/ # Controlled mutation boundary
β βββ storage/ # Snapshots, journals, checkpoint, recovery
β βββ dashboard/ # Local operator console and API
β βββ ...
βββ tests/ # Test suite (379+ tests)
βββ configs/ # YAML configuration files
βββ docs/ # Documentation and roadmap
βββ research/ # Scientific evidence framework
βββ artifacts/ # Simulation snapshots and journals
βββ scripts/ # Utility and verification scripts
π Documentation
| Resource | Description |
|---|---|
| docs/ROADMAP.md | Development roadmap and milestones |
| docs/changelog.md | Detailed changelog |
| docs/todo.md | Current TODO and gate status |
| docs/ARCHITECTURE.md | Architecture documentation |
| docs/B5D_FORMAT.md | .b5d snapshot format specification |
| research/README.md | Scientific evidence framework |
π§ͺ Testing
# Fast regression suite
pytest -v -m "not slow"
# Full suite
pytest -v
# With coverage
pytest --cov=src --cov-report=term-missing
Current baseline: 379 passed, 0 failed β
π License
MIT License β see LICENSE.
Copyright Β© 2025-2026 Thomas Heisig and contributors.
π Citation
@software{heisig2026brain5d,
author = {Thomas Heisig},
title = {Brain-5D: Sparse 5D Spiking-Neural Simulation with Observable Plasticity},
year = {2026},
version = {0.5.0-alpha.5},
url = {https://huggingface.co/<your-org>/Brain-5D},
license = {MIT}
}
π€ Contributing
Contributions are welcome! Please see CONTRIBUTING.md and our Code of Conduct.
Current focus: Alpha.5 closure (restore determinism experiments) β Alpha.6 (morphological self-regulation).
π Links
- GitHub: Thomas-Heisig/Brain-5D
- Issues: GitHub Issues
- Changelog: CHANGELOG.md