Modules & Rust Crates — SCPN-Control

Python Subpackages

Package	Modules	Description	Key Contents
scpn	13	Top-level API, SPN → SNN compiler, contract system	compiler contracts bitstream
core	74	Physics solvers: GS equilibrium, MHD stability, transport, gyrokinetics, edge/pedestal	grad_shafranov gk_transport mhd neoclassical
control	50	10 controller types, digital twin, disruption prediction, actuator models	pid mpc hinf ppo snn
phase	9	Kuramoto-Sakaguchi dynamics, K_nm, UPDE, Lyapunov guard, WebSocket streaming	kuramoto knm upde
cli	1	Click-based command-line interface: demo, benchmark, validate, live	click

Crate	Modules	Description	Key Exports
control-types	5	Shared type definitions: state vectors, matrix wrappers, error types	StateVec KnmMatrix
control-math	14	Numerical primitives: Kuramoto kernel, RK4 stepper, Lyapunov exponent, matrix ops	kuramoto_step rk4 lyapunov
control-core	21	Physics solvers in Rust: GS iteration, transport coefficients, MHD criteria	gs_solve mercier ballooning
control-control	13	Controller implementations: PID, MPC, Riccati solver, gain scheduler	pid_step mpc_solve dare
control-python	6	PyO3 entry point: Python module registration, type conversions	PyModule

vizOptional

matplotlib, plotly — equilibrium contours, phase portraits, profile plots

dashboardOptional

streamlit, plotly — interactive Streamlit dashboard for live monitoring

hdf5Optional

h5py — HDF5 state checkpointing and GEQDSK I/O

wsOptional

websockets — async WebSocket phase streaming server

mlOptional

torch, scikit-learn — QLKNN surrogate, disruption predictor, neural equilibrium

jaxOptional

jax, jaxlib — JAX-differentiable GS solver, JIT compilation, GPU autodiff

neuroOptional

snnTorch — SNN compilation backend, LIF neuron models

rlOptional

stable-baselines3, gymnasium — PPO agent, tokamak environment

benchmarkOptional

pytest-benchmark — Python-side performance regression testing

fullMeta

All optional groups combined — equivalent to pip install "scpn-control[all]"