323 Python capability source modules and 13 Rust workspace crates. The generated capability inventory is the source of truth for package, test, documentation, and workspace counts.
| Layer | Modules | Capability |
|---|---|---|
| Core Physics | 118 | Grad-Shafranov, transport (1.5D + QLKNN + FNO), GK three-path, MHD stability (7 criteria), neoclassical, disruption chain, ELM/MARFE/L-H, runaway electrons, pellet injection, plasma-wall, 3D equilibrium |
| Control | 54 | PID, H∞, NMPC-JAX, SNN (Petri net compiler), gain-scheduled, fault-tolerant, safe RL (PPO), free-boundary tracking, burn control, RZIP, RWM feedback, μ-synthesis, detachment, density, volt-second, EKF |
| I/O | 15 | IMAS/OMAS adapter, GEQDSK, tokamak archive, logging |
| SCPN Compiler | 12 | Petri net structure, compiler, contracts, safety interlocks, artifact packaging |
| Phase Dynamics | 10 | Kuramoto UPDE solver, adaptive Knm coupling, GK-to-UPDE bridge, plasma Knm, Lyapunov guard, real-time monitoring, WebSocket streaming |
| Diagnostics | 5 | Synthetic sensors, tomographic inversion, forward models |
| Nuclear | 5 | Blanket neutronics, PWI erosion, wall interaction |
| Engineering | 4 | Balance of plant, CAD raytrace, thermal hydraulics |
The curated layers above explain stable responsibilities; they are not an additive inventory ledger. The generated v4.0.0 manifest counts 323 Python capability source modules across the complete tree, including runtime, validation-harness, and shared-utility surfaces.
| Crate | Purpose |
|---|---|
| fusion-core | Grad-Shafranov kernel (0.52 µs PID; SOR/multigrid GS solve), vacuum field |
| fusion-physics | Native physics operators (transport, MHD, neoclassical) |
| fusion-control | Real-time control kernels (PID, reduced-order loops) |
| fusion-math | Numerical primitives and linear algebra |
| fusion-ml | ML inference (QLKNN, FNO surrogates) |
| fusion-gpu | GPU acceleration paths |
| fusion-diagnostics | Synthetic diagnostics and forward models |
| fusion-engineering | Balance-of-plant and engineering models |
| fusion-nuclear | Blanket neutronics and PWI |
| fusion-polyglot | Multi-language acceleration dispatch |
| fusion-types | Shared config and state types |
| fusion-python | PyO3 bindings (cdylib) |
| fusion-phase | Phase-dynamics and coherence-control kernels |