Spin Disorder

Group: Magnetism | Class: SpinDisorderCard

What the card does

Spin Disorder starts from existing moments or an element map and generates intermediate magnetic states at selected disorder fractions. It covers discrete flips, randomized directions, and cone disorder between FM/AFM and PM-like limits, keeping this discrete semantics separate from continuous small-angle Magmom Rotation.

Example workflow

Scenario: The model sees only FM/AFM and fully random PM-like endpoints

An FeCo set includes FM, AFM, and fully random PM-like states but lacks intermediate states with 10-70% local flips, leading to unstable finite-temperature energy ordering.

Input: Structures initialized by Magnetic Order or Set Magnetic Moments. Objective: Sample flip fractions 0.1,0.3,0.5,0.7. Parameters: mode=Flip fraction, those four fractions, samples_per_fraction=3, and use_seed enabled. Output: Twelve states per input. Verification: After retraining, energy and magnetic-force errors at intermediate fractions should no longer greatly exceed endpoint errors.

Parameters

Disorder model

Mode（mode）

str，默认 'Flip fraction'。Flip fraction 保持共线轴只翻转符号，适合 FM/AFM 到 PM 的离散无序梯度；Randomize fraction 把选中自旋方向在完整球面上随机化；Cone disorder 保持围绕参考方向的有限温非共线扰动。这三个值与下拉框和序列化参数完全一致。

Fractions（fractions）

str，默认 '0.1,0.3,0.5,0.7'。被翻转或随机化的自旋比例，必须是逗号分隔且位于 (0, 1] 的有限数值；非法文本和超范围值会明确报错，不会被忽略或截断。0.1/0.3/0.5/0.7 可覆盖 FM/AFM 到 PM 之间的无序梯度。

Samples Per Fraction（samples_per_fraction）

int, default 1. Independent site selections per fraction. One traces a path; 3-10 sample local variability.

Cone Angle（cone_angle）

float, default 30.0 degrees. Maximum cone deviation. About 10-30 degrees stays near order; values approaching 90 degrees represent strong disorder.

Used when the mode or direction model selects cone/noncollinear randomization.

Moment magnitudes

Magnitude Source（magnitude_source）

str, default 'Existing initial magmoms'. Reuse spin:R:3 when present and accept legacy initial_magmoms only when spin is absent. If neither exists, construct magnitudes from magmom_map and default_moment; never replace known element moments with a generic default.

Magmom Map（magmom_map）

str, default ''. Provide known element-resolved local moments explicitly, for example Fe:2.2,Ni:0.6. Do not invent a prior for an unknown element through a default value.

Default Moment（default_moment）

float, default 0.0. Used only when an element is absent from magmom_map. List important magnetic elements explicitly; nonmagnetic elements should normally remain at 0.

Lift Scalar（lift_scalar）

bool, default True. Enable when scalar moments must be promoted to noncollinear vectors. Do not promote again when the input already contains directional information.

Axis（axis）

list[float] | tuple[float, float, float], default (0.0, 0.0, 1.0). This is a directional reference, not an ordinary scalar setting: changing it changes a layering direction, surface normal, or moment direction. Confirm the cell orientation and intended physical direction first.

Used by modes that depend on a direction, layering, a surface, or vector initialization.

Apply Elements（apply_elements）

str, default ''. Limit processing to listed magnetic elements. Explicitly list them in alloys or interfaces to avoid assigning moments to a nonmagnetic host. Empty means all eligible atoms.

Randomness and output budget

Use Seed（use_seed）

bool, default False. Enable to reproduce with a fixed seed. It may be disabled for large exploration, but individual outputs then cannot be reproduced.

Seed（seed）

int, default 0. Identical input, parameters, and seed produce the same candidates.

Active only when use_seed=True.

Max Outputs（max_outputs）

int, default 100. Output is approximately number of fractions times samples per fraction. Keep a cap when many upstream frames could let magnetic variants overwhelm structural diversity.

Recommended presets

Collinear flip-fraction path

{
  "class": "SpinDisorderCard",
  "params": {
    "mode": "Flip fraction",
    "fractions": "0.1,0.3,0.5,0.7",
    "samples_per_fraction": 3,
    "cone_angle": 30.0,
    "magnitude_source": "Existing initial magmoms",
    "magmom_map": "",
    "default_moment": 0.0,
    "lift_scalar": true,
    "axis": [0.0, 0.0, 1.0],
    "apply_elements": "",
    "use_seed": true,
    "seed": 42,
    "max_outputs": 100
  }
}

Scan local flip fractions from an FM or AFM reference.

Noncollinear cone disorder

{
  "class": "SpinDisorderCard",
  "params": {
    "mode": "Cone disorder",
    "fractions": "0.25,0.5,0.75",
    "samples_per_fraction": 2,
    "cone_angle": 20.0,
    "magnitude_source": "Existing initial magmoms",
    "magmom_map": "",
    "default_moment": 0.0,
    "lift_scalar": true,
    "axis": [0.0, 0.0, 1.0],
    "apply_elements": "Fe,Co",
    "use_seed": true,
    "seed": 7,
    "max_outputs": 50
  }
}

Add noncollinear directional fluctuations near an ordered finite-temperature state.

Recommended combinations

Magnetic Order -> Spin Disorder: establish an FM/AFM reference, then generate intermediate disorder fractions.
Set Magnetic Moments -> Spin Disorder -> Small-Angle Spin Tilt: define magnitudes, apply discrete disorder, then add small angular deviations.
Spin Disorder -> Geometry Filter: clean geometry if magnetic variants also underwent strong structural perturbation.

Common questions

No eligible magnetic moments. The input lacks usable moments or apply_elements matches none with nonzero magnitude. Initialize with Set Magnetic Moments or Magnetic Order.

The flipped count is not the exact decimal fraction. Atom counts are discrete; the fraction is converted to an integer between one and the eligible count.

Cone disorder does not look PM-like. It samples only near reference directions. Use Randomize fraction for full-sphere randomness.

Output labels

SpinDis(f={fraction},n={count},mode={flip|rand|cone},s={seed},a={cone_angle}); s appears only with use_seed=True, and a only for cone output.

All exported outputs use spin:R:3; an ASE vector-valued initial_magmoms alias is kept in sync internally.

Reproducibility

With use_seed, site selection and directions are determined by the seed, input identity, fraction index, and sample index.