Micromagnetic parameter reference

This page lists every micromagnetic input parameter, its Matlab name, its Python name and its default value. The Matlab names are the field names of the struct that is handed to Fortran, as defined by getProblemFieldnames in MagTenseMicroMagIO.f90.

In Matlab, parameters that map onto an internal integer flag are set through a set... method rather than by assigning the number; those methods are named in the tables. Integer-valued fields must be assigned as int32.

In Python, most parameters are constructor arguments of MicromagProblem; those that are not are marked “attribute” and are assigned after construction.

Grid and geometry

Matlab	Python	Default	Description
`grid_n`	`res`	required	Number of cells along x, y and z. For unstructured meshes use `[ntot,1,1]`.
`grid_L`	`grid_L`	`[500e-9,125e-9,3e-9]`	Size of the rectangular domain in m. Only used for a uniform grid.
`grid_type`	`grid_type`	`uniform`	`uniform` (1), `tetrahedron` (2) or `unstructuredPrisms` (3). Matlab: `setMicroMagGridType`.
`grid_pts`	`grid_pts`	`0`	`(ntot,3)` cell centres for the two unstructured grid types.
`grid_abc`	`grid_abc`	`0`	`(ntot,3)` full side lengths of the prisms, `unstructuredPrisms` only.
`grid_nod`	`grid_nod`	`0`	Node coordinates of a tetrahedral mesh.
`grid_ele`	`grid_ele`	`0`	Node indices of each tetrahedral element.
`grid_nnod`	`grid_nnod`	`0`	Number of nodes in the tetrahedral mesh.

See Grids and meshes.

Material and anisotropy

Matlab	Python	Default	Description
`Ms`	`Ms`	`8e5`	Saturation magnetization [A/m], per cell.
`A0`	`A0`	`1.3e-11`	Exchange stiffness [J/m], per cell.
`K0`	`K0`	`0`	Uniaxial anisotropy constant [J/m^3], per cell.
`K1`, `K2`	`K1`, `K2`	`0`	Cubic anisotropy constants [J/m^3], per cell.
`K0_arr`	`K0_arr`	`0`	`(ntot,6,3)` general anisotropy expansion [J/m^3].
`CrysAxis`	`CrysAxis`	identity	`(ntot,3,3)` local crystal coordinate system.
`u_ea`	`u_ea`	`0`	`(ntot,3)` uniaxial easy axis. Python: attribute.
`alpha`	`alpha`	`4.42e3`	Damping constant [m/(A s)]. Zero switches to the `alphat` table.
`gamma`	`gamma`	`2.21e5`	Precession constant [m/(A s)]. Zero removes precession.
`alphat`, `nt_alpha`	`t_alpha`, `alpha_fct`	`0`	Tabulated time-dependent damping. Matlab: `setAlpha`.
`MaxT0`	`max_T0`	`2`	Legacy, not used by the current solver.
`m0`	`m0`	random	`(ntot,3)` initial reduced magnetization. Normalised on input.

See Material parameters and Magnetocrystalline anisotropy.

Solver, time and applied field

Matlab	Python	Default	Description
`solver`	`solver`	`Dynamic`	`Explicit` (1), `Dynamic` (2), `Implicit` (3, not implemented). Matlab: `setMicroMagSolver`.
`ProblemMod`	`prob_mode`	`new`	`new` (1) or `old` (2). Matlab: `setMicroMagProblemMode`.
`t`, `nt`	`t_end`, `nt`	`linspace(0,1,1000)`	Output times. Matlab: `setTime`. In Python these are arguments of `run_simulation`.
`Hext`, `nt_Hext`	`fct_h_ext`, `nt_h_ext`	zero field	Applied field table of `[t,Hx,Hy,Hz]` rows [A/m]. Matlab: `setHext`.
`tol`	`tol`	`1e-4`	Relative tolerance of the ODE solver.
`thres`	`thres`	`1e-6`	Threshold below which a solution component is treated as zero.
`t_conv`, `nt_conv`	`nt_conv`	`0`, `1`	Convergence-check times. Matlab: `setConvergenceCheckTime`.
`conv_tol`	`conv_tol`	`1e-4`	Convergence criterion.
`setTimeDis`	`setTimeDis`	`10`	Progress is reported every n’th step.
`useCVODE`	`cvode`	`0`	Use CVODE instead of RKSuite. Matlab: `setUseCVODE`.
`useCuda`	`cuda`	`0`	Evaluate the demagnetization field on a GPU. Matlab: `setUseCuda`.

See Solvers and time integration.

Adaptive hysteresis parameters

Matlab	Python	Default	Description
`adaptiveHext`	`hysteresis_solver`	`0` / `static`	`1` / `adaptive` enables adaptive field stepping.
`maxHextSteps`	`max_steps`	`1000`	Maximum number of accepted field steps.
`H_start`, `H_end`	`H_start`, `H_end`	`[0 0 0]`	Start and end applied field [A/m].
`dH_initial`	`dH_initial`	`1e3`	First field step [A/m].
`dH_min`, `dH_max`	`dH_min`, `dH_max`	`1e1`, `1e5`	Smallest and largest allowed field step [A/m].
`dH_grow`	`dH_grow`	`1.25` / `1.5`	Growth factor. The Matlab default is 1.25, the Python method default is 1.5.
`dH_shrink`	`dH_shrink`	`0.5` / `0.75`	Shrink factor. The Matlab default is 0.5, the Python method default is 0.75.
`dM_min`	`dM_min`	`1e-3`	Grow the step below this magnetization change.
`dM_target`	`dM_target`	`1e-2`	Shrink the step above this magnetization change.
`dM_reject`	`dM_reject`	`5e-2`	Reject and retry the step above this magnetization change.
`switch_refdH`, `use_sw_ref`	`switch_refine_dH`	`0`, `0`	Largest step accepted across a magnetization sign change [A/m].

See Adaptive hysteresis.

Exchange

Matlab	Python	Default	Description
`exch_meth`	`exch_meth`	`DirectLaplacianNeumann`	`DirectLaplacianNeumann` (1) or `GGNeumann` (2). Matlab: `setMicroMagExchMethod`.
`exch_intpn`	`exch_intpn`	`Extended`	`Extended` (1) or `Compact` (2). Matlab: `setMicroMagExchInterpn`.
`exch_weigh`	`exch_weigh`	`8`	Exponent of the inverse-distance face weighting.
`passExch`	`passexch`	`0`	Use an externally supplied exchange matrix. Matlab: `setMicroMagpassExch` or `setExchangeMatrixCOO`.
`exch_nval`, `exch_nrow`, `exch_ncol`	`exch_nval`, `exch_nrow`, `exch_ncols`	`0` / `1`	Size of the supplied exchange matrix.
`exch_val`, `exch_rows`, `exch_cols`	`exch_val`, `exch_rows`, `exch_cols`	`0`	The supplied exchange matrix in COO form, 1-based indices.
-	`exch_presize`	`12`	Python only. Length of the returned exchange-matrix arrays as `exch_presize * ntot`.
`exchPBC`	`exchPBC`	`[0 0 0]`	Periodic exchange boundaries along x, y and z.

See Exchange interaction and Periodic exchange boundaries.

Demagnetization

Matlab	Python	Default	Description
`useDemag`	`usedemag`	`1`	Include the demagnetization field. Matlab: `setUseDemag`.
`useAvgN`	`useavgn`	`1`	Use the cell-averaged prism tensor rather than the tensor at the cell centre.
`N_ave`	`N_ave`	`[1 1 1]`	Numerical sub-grid averaging of the tensor over the receiving cell, `unstructuredPrisms` only.
`dem_appr`	`demag_approx`	`none`	`none` (1), `threshold` (2), `fft_thres` (3), `threshold_fraction` (4), `fft_threshold_fraction` (5). Matlab: `setMicroMagDemagApproximation`.
`dem_thres`	`dem_thres`	`0`	Cut-off used by the threshold approximations.
`CV`	`cv`	`0`	Coefficient of variation of a random error added to the demagnetization field.
`N_ret`, `N_file_out`	`filename`	`1`, `'t'`	Return or write the demagnetization tensor. Matlab: `setReturnNFilename`.
`N_load`, `N_file_in`	`filename`	`1`, `'t'`	Load a stored demagnetization tensor. Matlab: `setLoadNFilename`.
`nThreads`	`n_threads`	`1`	OpenMP threads used when building the tensor.
`usePres`	`precision`	`0`	Accepted but not used; the tensor is always single precision.
`demigstp`	`demigstp`	`0`	Accepted but not used by the current solver.

See Demagnetization field.

Macrogeometry and shape correction

Matlab	Python	Default	Description
`n_macro`	`n_macro`	`[0 0 0]`	Number of domain copies on each side along x, y and z.
`shiftVec`	`shiftVec`	`[0 0 0]`	Spacing between neighbouring copies [m].
`macroShape`	`macroShape`	`[1 1 1]`	Side lengths of the prism approximating the macrogeometry [m].
`sampleShape`	`sampleShape`	`[1 1 1]`	Side lengths of the prism approximating the sample [m].

See Periodic boundaries and macrogeometry.

Thermal parameters

Matlab	Python	Default	Description
`temperature`	`T`	`0`	Per-cell temperature [K].
`rng_seed`	`rng_seed`	`0`	Random seed: `0` compiler default, `>0` deterministic, `<0` from the clock.

See Thermal fluctuations.

FMM demagnetization

Matlab	Python	Default	Description
`use_fmm`	`use_fmm`	`0`	Enable the FMM demagnetization path. Requires `USE_FMM3D=1`.
`fmm_cells`	`fmm_cells_per_node`	`100`	Maximum cells in a leaf node before splitting.
`fmm_eps`	`fmm_eps`	`1e-4`	Requested FMM accuracy, which sets the plane-wave expansion order.
`fmm_nterms`	`fmm_nterms`	`-1`	Multipole expansion order. Negative means derive it from `fmm_eps`.
`ifunif`	`ifunif`	`1`	Tree type. Must be 1 (uniform tree).
`nlmin`, `nlmax`	`nlmin`, `nlmax`	`1`, `5`	Minimum and maximum level of the octree.
`fmm_short`	`allow_fmm_short_circuit`	`1`	Allow falling back to the direct calculation for small problems.
`fmm_min_n`	`fmm_min_n`	`20000`	Cell-count threshold above which FMM is used.

See Demag field - FMM.

Output, tracing and timing

Matlab	Python	Default	Description
`ReturnHall`	`usereturnhall`	`0`	Return the individual effective-field terms.
`dummy_run`	`dummy_run`	`0`	Build everything but skip the time integration.
`log_dir`, `N_log_dir`	`log_dir`	`'logs'`	Directory for the trace and timing logs. Matlab: `setLogDirFilename`.
`timer_log`, `N_timer_log`	`timer_log_file`	`'timing.log'`	Timing log file. Matlab: `setTimerLogFilename`.
`trace_log`, `N_trace_log`	`trace_log_file`	`'trace.log'`	Trace log file. Matlab: `setTraceLogFilename`.
`window_ena`	`window_enabled`	`1`	Flush timing statistics at intervals rather than only at the end.
`window_int`	`window_interval`	`30.0`	Timing window length [s].
`trace_ena`	`trace_enabled`	`0`	Enable the execution trace. Significant performance cost.
`flush_each`	`flush_each`	`1`	Flush the trace file after every entry.
`trace_verb`	`trace_verbose`	`1`	Only log trace events with at least this verbosity.

See Micromagnetic output and Performance, Timing & Trace Logging.

Matlab-only convenience fields

The following fields exist on the Matlab problem object but are not passed to Fortran. They are used by the example scripts to steer plotting and saving:

SaveTheResult, ShowTheResult, SolverType, DirectoryFilename, SimulationName, FileName, HextFct, FFTdims, ExternalMesh, MeshType, ExternalMeshFileName, DemagTensorFileName and exch_mat.