Configuration Reference¶

All simulation parameters are set via the SimConfigFA dataclass in gyrojax/simulation_fa.py.

from gyrojax.simulation_fa import SimConfigFA
cfg = SimConfigFA(Npsi=32, N_particles=200_000, beta=0.01)

Grid Parameters¶

Field	Type	Default	Description
`Npsi`	int	32	Radial (ψ) grid points
`Ntheta`	int	64	Poloidal (θ) grid points
`Nalpha`	int	32	Binormal (α) grid points

Field	Type	Default	Description
`N_particles`	int	200000	Number of ion markers

Field	Type	Default	Description
`n_steps`	int	500	Number of time steps
`dt`	float	0.05	Time step size (normalized to R₀/vti)

Field	Type	Default	Description
`R0`	float	1.0	Major radius (normalized)
`a`	float	0.18	Minor radius
`B0`	float	1.0	On-axis magnetic field
`q0`	float	1.4	Safety factor at axis: q(r) = q₀ + q₁·(r/a)
`q1`	float	0.5	Safety factor shear coefficient

Field	Type	Default	Description
`Ti`	float	1.0	Ion temperature (normalized)
`Te`	float	1.0	Electron temperature (normalized)
`mi`	float	1.0	Ion mass (normalized)
`rho_star`	float	1/180	ρᵢ/a — gyroradius-to-machine-size ratio (CBC: 1/180)
`e`	float	1000.0	Charge-to-mass ratio factor (CBC: Ωᵢ = 1000)
`vti`	float	1.0	Ion thermal velocity (normalized)
`n0_avg`	float	1.0	Average background density

Field	Type	Default	Description
`R0_over_LT`	float	6.9	Ion temperature gradient R₀/LT (CBC: 6.9)
`R0_over_LTe`	float	6.9	Electron temperature gradient R₀/LTe (TEM driver)
`R0_over_Ln`	float	2.2	Density gradient R₀/Ln (CBC: 2.2)

Field	Type	Default	Description
`vpar_cap`	float	4.0	Parallel velocity cap (multiples of vti)
`use_global`	bool	False	Global radial profiles (True) vs flux-tube (False)

Field	Type	Default	Description
`pert_amp`	float	1e-2	Initial perturbation amplitude (use 1e-4 for multi-mode runs)
`zonal_init`	bool	False	Seed zonal flow (k_θ=0) for Rosenbluth-Hinton/GAM tests
`k_mode`	int	1	Binormal mode number n for ITG seed: sin(2θ + n·α)
`single_mode`	bool	False	Project φ to ±k_mode after each Poisson solve (linear benchmark)
`k_alpha_min`	int	0	Zero out α modes 1..k_alpha_min−1 (suppress aliasing)

Field	Type	Default	Description
`canonical_loading`	bool	False	Apply canonical Pφ weight correction at t=0
`use_pullback`	bool	False	Periodic f₀ pullback transformation
`pullback_interval`	int	50	Steps between pullbacks (0 = disabled)
`nu_soft`	float	0.0	Soft amplitude-dependent weight damping rate (0 = off)
`w_sat`	float	2.0	Saturation weight for soft damping
`soft_damp_alpha`	int	2	Power exponent for soft damping

Field	Type	Default	Description
`collision_model`	str	`'none'`	Collision operator: `'none'` \| `'krook'` \| `'lorentz'` \| `'dougherty'`
`nu_krook`	float	0.01	Krook (BGK) damping rate
`nu_ei`	float	0.01	Electron-ion collision frequency (Lorentz pitch-angle scattering)
`nu_coll`	float	0.01	Dougherty collision frequency

Field	Type	Default	Description
`use_weight_spread`	bool	False	Enable periodic weight smoothing onto grid
`weight_spread_interval`	int	10	Spread every N steps
`zonal_preserving_spread`	bool	True	Preserve zonal flow component during spreading

Field	Type	Default	Description
`electron_model`	str	`'adiabatic'`	`'adiabatic'` (Boltzmann) \| `'drift_kinetic'` (kinetic TEM)
`me_over_mi`	float	1/1836	Electron-to-ion mass ratio
`subcycles_e`	int	10	Electron subcycles per ion step
`N_electrons`	int	0	Number of electron markers (0 = same as N_particles)

Field	Type	Default	Description
`implicit`	bool	False	Use Crank-Nicolson + Picard iteration
`picard_max_iter`	int	4	Maximum Picard iterations per step
`picard_tol`	float	1e-3	Convergence tolerance on φ (L∞ norm)
`semi_implicit_weights`	bool	False	Semi-implicit CN weight update (unconditionally stable)
`use_cn_weights`	bool	False	Alias for `semi_implicit_weights`

Field	Type	Default	Description
`absorbing_wall`	bool	False	Absorbing BC: zero weights of escaped particles. False = hard clamp (legacy)

Field	Type	Default	Description
`beta`	float	0.0	Plasma β: 0.0 = electrostatic, >0 = electromagnetic (KBM)
`gyroaverage_scatter`	bool	True	Apply √Γ₀(b) to δn before Poisson solve (symmetric gyroaveraging)
`use_radial_gaa`	bool	True	Use radially-resolved g^αα(ψ) in Γ₀ operator (clamped+tapered)
`fused_rk4`	bool	True	Fuse particle push + weight update into single RK4 (3.78× speedup)