ModeLocking.jl

Reduced-order ODE model and analysis tools for tearing-mode locking, resistive-wall-mode coupling, and error-field penetration physics. Originally developed as part of the FUSE ActorLocking mode-locking actor and extracted into a standalone package so that the physics/ODE core, NN-based locking classifier, persistence, and plotting utilities can be developed, tested, and reused independently of FUSE.

Contents

• ODEparams, RHS functions, and solve_ODEs / solve_grid / solve_and_classify: the reduced-order ODE model (rational-surface tearing mode +/- resistive-wall-mode +/- ideal-wall coupling), including normalization and analytic bifurcation bounds.
• NNparams, LockingNNModel, train_locking_nn, tune_locking_nn, transfer_learn_locking_nn: a small Flux-based classifier that predicts locking probability as a function of the two control parameters.
• save_ode_results / load_ode_results, save_locking_nn / load_locking_nn: BSON-based persistence of ODE grids and trained NN models.
• plot_sols, plot_scatter, plot_phase_diagrams, plot_probability, plot_time_traces, simulate_one_case: plotting utilities for grid scans and single-case time traces.

Usage

ModeLocking is intended to be used as a dev dependency of FUSE, where the ActorLocking actor (src/actors/stability/locking_actor.jl) builds an ODEparams from dd and calls into this package for all the heavy lifting.

Online documentation

For more details, see the online documentation.