Use Cases¶
Fusion Reactor Control¶
Tokamak Real-Time Plasma Control¶
Deploy scpn-control as the real-time controller for experimental or commercial tokamaks. The 11.9 us P50 control loop enables:
- Shape control at 10--30 kHz (vs. 4 kHz on current DIII-D systems)
- Disruption prediction with ML inference under 1 ms
- SPI mitigation with halo current and runaway electron physics
- Profile control via neuro-cybernetic dual R+Z SNN feedback
Target users: Fusion startups (Commonwealth Fusion Systems, TAE, Tokamak Energy, Zap Energy), national labs (DIII-D, JET, KSTAR, EAST).
ITER / DEMO Integration¶
The formal verification layer (contract-based pre/post-condition checking) directly addresses ITER nuclear safety requirements. The SNN controller provides fail-safe operation: if the Lyapunov guard detects instability, the system halts to safe state within one control cycle.
Digital Twin & Commissioning¶
Offline Plant Commissioning¶
Before first plasma, use the digital twin to:
- Commission control algorithms against synthetic plasma scenarios
- Train operators on disruption response procedures
- Validate sensor/actuator configurations
- Stress-test control logic with adversarial perturbations
The digital twin runs in real-time on a single CPU core, enabling rapid iteration without dedicated HPC resources.
Hardware-in-the-Loop Testing¶
The HIL harness (scpn-control hil-test) replays reference shot data
through the full controller pipeline, verifying end-to-end latency
and control fidelity against experimental baselines.
Research & Education¶
Neuro-Symbolic Control Research¶
scpn-control is the only open-source implementation of:
- Stochastic Petri Net to SNN compilation
- Kuramoto-Sakaguchi phase dynamics with formal Lyapunov stability
- Phase-amplitude coupling (PAC) between oscillator populations
These are active research frontiers in both fusion plasma physics and computational neuroscience.
Graduate-Level Course Material¶
The codebase includes:
- 5 tutorial notebooks (Jupyter) with step-by-step walkthroughs
- A live Streamlit dashboard for interactive exploration
- 1969 tests demonstrating expected behavior and edge cases
- Full competitive analysis against state-of-the-art codes
Edge & Embedded Deployment¶
No GPU Required¶
The neural equilibrium kernel achieves P-EFIT-class speed (0.39 ms) on CPU only. This enables deployment on:
- ARM-based edge controllers
- Radiation-hardened embedded systems
- Air-gapped control networks (no cloud dependency)
Rust Native Backend¶
The 5-crate Rust workspace provides:
- Zero-copy PyO3 bindings for Python interop
- No runtime garbage collection (deterministic latency)
- Rayon parallelism for multi-core scaling
- Criterion benchmarks for regression testing
Comparison Matrix¶
| Use Case | scpn-control | TORAX | FUSE | FreeGS |
|---|---|---|---|---|
| Real-time control | 11.9 us | No | No | No |
| Disruption prediction | ML-based | No | No | No |
| SPI mitigation | Yes | No | No | No |
| Digital twin | Real-time | No | No | No |
| Neural equilibrium | 0.39 ms (CPU) | No | No | No |
| SNN controller | Yes | No | No | No |
| Formal verification | Contracts | No | No | No |
| Edge deployment | Yes (Rust) | No (JAX) | No (Julia) | Partial |
| Autodifferentiation | No | Yes (JAX) | Yes (Julia) | No |
| GPU transport | No | Yes | Yes | No |
Get Started¶
For commercial licensing: protoscience@anulum.li