maml.utils package

Utilities package.

class maml.utils.ConstantValue(value: float, **kwargs)

Bases: ValueProfile

Return constant value.

get_value()

Return constant value.

class maml.utils.DataSplitter()

Bases: MSONable

Data splitter base class.

split(mat_ids, **kwargs)

Split the mat_ids, optionally one can provide targets. This is useful in stratified split.

• Parameters mat_ids (list) – list of material ids

Returns: (train_ids, val_ids, test_ids) or

(train_ids, test_ids)

class maml.utils.DummyScaler()

Bases: MSONable

Dummy scaler does nothing.

classmethod from_training_data(structures: list[StructureOrMolecule], targets: VectorLike, is_intensive: bool = True)

• Parameters

  • structures (list) – list of structures/molecules

  • targets (list) – vector of target properties

  • is_intensive (bool) – whether the target is intensive

Returns: DummyScaler.

static inverse_transform(transformed_target: float, n: int = 1)

return as it is :param transformed_target: transformed target :type transformed_target: float :param n: number of atoms :type n: int

• Returns transformed_target.

static transform(target: float, n: int = 1)

• Parameters

  • target (float) – target numerical value

  • n (int) – number of atoms

• Returns target.

class maml.utils.LinearProfile(value_start: float, value_end: float = 0.0, max_steps: int = 100, **kwargs)

Bases: ValueProfile

LinearProfile by setting starting value and the rate of value change. The profile can be initialized either by [value_start, value_end, max_step] or [value_start, rate].

get_value()

Get LinearProfile value Returns: float.

class maml.utils.MultiScratchDir(rootpath: str | Path, n_dirs: int = 1, create_symbolic_link: bool = False, copy_from_current_on_enter: bool = False, copy_to_current_on_exit: bool = False)

Bases: object

Creates a “with” context manager that automatically handles creation of temporary directories (utilizing Python’s build in temp directory functions) and cleanup when done. The main difference between this class and monty ScratchDir is that multiple temp directories are created here. It enables the running of multiple jobs simultaneously in the directories The way it works is as follows:

1. Create multiple temp dirs in specified root path.

2. Optionally copy input files from current directory to temp dir.

3. User loops among all directories

4. User performs specified operations in each directories

5. Change back to original directory.

6. Delete temp dir.

SCR_LINK(_ = ‘scratch_link_ )

class maml.utils.Scaler()

Bases: MSONable

Base Scaler class. It implements transform and inverse_transform. Both methods will take number of atom as the second parameter in addition to the target property.

inverse_transform(transformed_target: float, n: int = 1)

Inverse transform of the target :param transformed_target: transformed target :type transformed_target: float :param n: number of atoms :type n: int

• Returns target.

transform(target: float, n: int = 1)

Transform the target values into new target values :param target: target numerical value :type target: float :param n: number of atoms :type n: int

• Returns scaled target.

class maml.utils.ShuffleSplitter(ratios: str = ‘80/10/10’, delim: str = ‘/’, random_seed: int | None = None)

Bases: DataSplitter

Randomly shuffe the material ids and split the ids into given ratios.

split(mat_ids, **kwargs)

Randomly split the mat_ids :param mat_ids: material ids :type mat_ids: list

Returns:

class maml.utils.StandardScaler(mean: float = 0.0, std: float = 1.0, is_intensive: bool = True)

Bases: Scaler

Standard scaler with consideration of extensive/intensive quantity For intensive quantity, the mean is just the mean of training data, and std is the std of training data For extensive quantity, the mean is the mean of target/atom, and std is the std for target/atom .. method:: transform(self, target, n=1)

standard scaling the target and.

classmethod from_training_data(structures: list[StructureOrMolecule], targets: VectorLike, is_intensive: bool = True)

Generate a target scaler from a list of input structures/molecules, a target value vector and an indicator for intensiveness of the property :param structures: list of structures/molecules :type structures: list :param targets: vector of target properties :type targets: list :param is_intensive: whether the target is intensive :type is_intensive: bool

Returns: new instance.

inverse_transform(transformed_target: float, n: int = 1)

Inverse transform of the target :param transformed_target: transformed target :type transformed_target: float :param n: number of atoms :type n: int

• Returns original target.

transform(target: float, n: int = 1)

Transform numeric values according the mean and std, plus a factor n :param target: target numerical value :type target: float :param n: number of atoms :type n: int

• Returns scaled target.

class maml.utils.Stats()

Bases: object

Calculate the stats of a list of values. This is particularly helpful when you want to convert lists of values of different lengths to uniform length for machine learning purposes.

supported

allowed_stats(_ = [‘max’, ‘min’, ‘range’, ‘mode’, ‘mean_absolute_deviation’, ‘mean_absolute_error’, ‘moment’, ‘mean’, ‘inverse_mean’, ‘average’, ‘std’, ‘skewness’, ‘kurtosis’, ‘geometric_mean’, ‘power_mean’, ‘shifted_geometric_mean’, ‘harmonic_mean’_ )

static average(data: list[float], weights: list[float] | None = None)

Weighted average.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: average value

static geometric_mean(data: list[float], weights: list[float] | None = None)

Geometric mean of the data.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: geometric mean of the distribution

static harmonic_mean(data: list[float], weights: list[float] | None = None)

harmonic mean of the data.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: harmonic mean of the distribution

static inverse_mean(data: list[float], weights: list[float] | None = None)

inverse mean.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: average value

static kurtosis(data: list[float], weights: list[float] | None = None)

Kurtosis of the distribution.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: Kurtosis of the distribution

static max(data: list[float], weights: list[float] | None = None)

Max of value Args:31

data (list): list of float data weights (list): optional weights.

Returns: maximum value

static mean(data: list[float], weights: list[float] | None = None)

Weighted average.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: average value

static mean_absolute_deviation(data: list[float], weights: list[float] | None = None)

mean absolute deviation.

• Parameters

  • data (list) – list of float data

  • weights (list) – optional weights

Returns: mean absolute deviation

static mean_absolute_error(data: list[float], weights: list[float] | None = None)

mean absolute error.

• Parameters

  • data (list) – list of float data

  • weights (list) – optional weights

Returns: mean absolute error

static min(data: list[float], weights: list[float] | None = None)

min of value :param data: list of float data :type data: list :param weights: optional weights. :type weights: list

Returns: minimum value

static mode(data: list[float], weights: list[float] | None = None)

Mode of data, if multiple entries have equal counts, compute the average of those.

• Parameters

  • data (list) – list of float data

  • weights (list) – optional weights

Returns: mode of values, i.e., max - min

static moment(data: list[float], weights: list[float] | None = None, order: int | None = None, max_order: int | None = None)

Moment of probability mass function.

order = 1 means weighted mean order = 2 means standard deviation order > 2 corresponds to higher order moment to

the 1./order power

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data points

  • order (int) – moment order

  • max_order (int) – if set, it will overwrite order

Returns: float or list of floats

static power_mean(data: list[float], weights: list[float] | None = None, p: int = 1)

power mean https://en.wikipedia.org/wiki/Generalized_mean.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

  • p (int) – power

Returns: power mean of the distribution

static range(data: list[float], weights: list[float] | None = None)

Range of values :param data: list of float data :type data: list :param weights: optional weights. :type weights: list

Returns: range of values, i.e., max - min

static shifted_geometric_mean(data: list[float], weights: list[float] | None = None, shift: float = 100)

Since we cannot calculate the geometric means on negative or zero values, we can first shift all values to positive and then calculate the geometric mean afterwards, we shift the computed geometric mean back by a shift value.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

  • shift (float) – shift value

Returns: geometric mean of the distribution

static skewness(data: list[float], weights: list[float] | None = None)

Skewness of the distribution.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: Skewness of the distribution

static std(data: list[float], weights: list[float] | None = None)

Standard deviation.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: Standard deviation

class maml.utils.ValueProfile(max_steps: int | None = None, **kwargs)

Bases: object

Base class for ValueProfile. The base class has the following methods .. method:: increment_step(self)

add one to step

get_value(self)

abstract method that return the value.

get_value()

abstract method that returns the current value Returns: value float.

increment_step()

Increase step attribute by one.

maml.utils.check_structures_forces_stresses(structures: list[Structure], forces: list | None = None, stresses: list | None = None, stress_format: str = ‘VASP’, return_none: bool = True)

Check structures, forces and stresses. The forces and stress are dependent on the lattice orientation. This function will rotate the structures and the corresponding forces and structures to lammps format [[ax, 0, 0], [bx, by, 0], [cx, cy, cz]].

The lattice are formed by the row vectors.

• Parameters

  • structures (list) – list of structures

  • forces (list) – list of force matrixs (m, 3)

  • stresses (list) – list of stress vectors

  • stress_format (str) – stress format, choose from “VASP”, “LAMMPS”, “SNAP”

  • return_none (bool) – whether to return list of None for forces and stresses

Returns: structures [forces], [stresses]

maml.utils.convert_docs(docs, include_stress=False, **kwargs)

Method to convert a list of docs into objects, e.g., Structure and DataFrame.

• Parameters

  • docs ([dict]) – List of docs. Each doc should have the same format as one returned from .dft.parse_dir.

  • include_stress (bool) – Whether to include stress components.

  • **kwargs – Passthrough.

• Returns A list of structures, and a DataFrame with energy and force data in ‘y_orig’ column, data type (‘energy’ or ‘force’) in ‘dtype’ column, No. of atoms in ‘n’ column sharing the same row of energy data while ‘n’ being 1 for the rows of force data.

maml.utils.cwt(z: np.ndarray, widths: np.ndarray, wavelet: str | Callable = ‘morlet2’, **kwargs)

The scalogram of the signal :param z: 1D signal array :type z: np.ndarray :param widths: wavelet widths :type widths: np.ndarray :param wavelet: wavelet name :type wavelet: str

Returns: 2D scalogram.

maml.utils.feature_dim_from_test_system(describer)

Get feature size from a test system.

• Parameters describer (BaseDescriber) – describers instance

maml.utils.fft_magnitude(z: ndarray)

Discrete Fourier Transform the signal z and return the magnitude of the coefficients :param z: 1D signal array :type z: np.ndarray

Returns: 1D magnitude.

maml.utils.get_describer_dummy_obj(instance)

For a describers, get a dummy object for transform_one. This relies on the type hint.

• Parameters instance (BaseDescriber) – describers instance

maml.utils.get_full_args(func: Callable)

Get args from function.

• Parameters func (callable) – function to determine the args

maml.utils.get_full_stats_and_funcs(stats: list)

Get expanded stats function name str and the corresponding function callables.

• Parameters stats (list) – a list of stats names, e.g, [‘mean’, ‘std’, ‘moment:1:None’]

Returns: list of stats names, list of stats callable

maml.utils.get_lammps_lattice_and_rotation(structure: Structure, origin=(0, 0, 0))

Transform structure to lammps compatible structure. The lattice and rotation matrix are returned.

• Parameters

  • structure (Structure) – pymatgen structure

  • origin (tuple) – origin coordinates

Returns: new lattice, rotation symmetry operator, rotation matrix

maml.utils.get_sp_method(sp_method: str | Callable)

Providing a signal processing method name return the callable :param sp_method: name of the sp function :type sp_method: str

Returns: callable for signal processing.

maml.utils.njit(func: Callable)

Dummy decorator, returns the original function :param func: function to be wrapped. :type func: Callable

Returns: decorated function

maml.utils.pool_from(structures, energies=None, forces=None, stresses=None)

Method to convert structures and their properties in to datapool format.

• Parameters

  • structures ([Structure]) – The list of Pymatgen Structure object.

  • energies ([float]) – The list of total energies of each structure in structures list.

  • forces ([np.array]) – List of (m, 3) forces array of each structure with m atoms in structures list. m can be varied with each single structure case.

  • stresses (list) – List of (6, ) virial stresses of each structure in structures list.

• Returns ([dict])

maml.utils.spectrogram(z: np.ndarray, return_time_freq: bool = False)

The spectrogram of the signal :param z: 1D signal array :type z: np.ndarray :param return_time_freq: whether to return time and frequency :type return_time_freq: bool

Returns: 2D spectrogram.

maml.utils.stats_list_conversion(stats_list: list[str])

Convert a list of stats str into a fully expanded list. This applies mainly to stats that can return a list of values, e.g., moment with max_order > 1.

• Parameters stats_list (list) – list of stats str

Returns: list of expanded stats str

maml.utils.stress_format_change(stress: np.ndarray | list[float], from_format: str, to_format: str)

Convert stress format from from_format to to_format :param stress: length-6 stress vector :type stress: list of float :param from_format: choose from “VASP”, “LAMMPS”, “SNAP” :type from_format: str :param to_format: choose from “VASP”, “LAMMPS”, “SNAP”. :type to_format: str

Returns: list of float stress vector

maml.utils.stress_list_to_matrix(stress: np.ndarray | list[float], stress_format: str = ‘VASP’)

convert a length-6 stress list to stress matrix 3x3.

• Parameters

  • stress (list* of *float) – list of stress

  • stress_format (str) – Supported formats are the follows VASP: xx, yy, zz, xy, yz, xz LAMMPS: xx, yy, zz, xy, zx, yz SNAP: xx, yy, zz, yz, xz, xy

Returns: 3x3 stress matrix

maml.utils.stress_matrix_to_list(stress_matrix: ndarray, stress_format: str = ‘VASP’)

Stress matrix to list representation :param stress_matrix: stress matrix 3x3 :type stress_matrix: np.ndarray :param stress_format: stress list format :type stress_format: str

Returns: list of float stress vector.

maml.utils.to_array(x)

Convert x into numerical array :param x: x can be a dataframe, a list or an array

return np.ndarray.

maml.utils.to_composition(obj: Composition | Molecule | Structure | str)

Convert str/structure or composition to compositions.

• Parameters obj (str/structure/composition) – object to convert

• Returns Composition object

maml.utils.write_data_from_structure(structure: Structure, filename: str, ff_elements: list[str] | None = None, significant_figures: int = 6, origin: tuple = (0, 0, 0))

Write structure to lammps data file, this is to speed up pymatgen LammpsData.

Args:a

structure (Structure): pymatgen structure
filename (str): filename
ff_elements (list of str): elements to be considered
significant_figures (int): significant figures of floats in output
origin (tuple): origin coordinates

maml.utils.wvd(z: np.ndarray, return_all: bool = False)

Wigner Ville Distribution calculator :param z: signal 1D :type z: np.ndarray :param return_all: whether to return time and freq info, default

only return the wvd information

Returns: NxN wvd matrix.

maml.utils._data_conversion module

Convert data list to docs or pool existing data lists for training.

maml.utils._data_conversion.convert_docs(docs, include_stress=False, **kwargs)

Method to convert a list of docs into objects, e.g., Structure and DataFrame.

• Parameters

  • docs ([dict]) – List of docs. Each doc should have the same format as one returned from .dft.parse_dir.

  • include_stress (bool) – Whether to include stress components.

  • **kwargs – Passthrough.

• Returns A list of structures, and a DataFrame with energy and force data in ‘y_orig’ column, data type (‘energy’ or ‘force’) in ‘dtype’ column, No. of atoms in ‘n’ column sharing the same row of energy data while ‘n’ being 1 for the rows of force data.

maml.utils._data_conversion.doc_from(structure, energy=None, force=None, stress=None)

Method to convert structure and its properties into doc format for further processing. If properties are None, zeros array will be used.

• Parameters

  • structure (Structure) – Pymatgen Structure object.

  • energy (float) – The total energy of the structure.

  • force (np.array) – The (m, 3) forces array of the structure where m is the number of atoms in structure.

  • stress (list/np.array) – The (6, ) stresses array of the structure.

• Returns (dict)

maml.utils._data_conversion.pool_from(structures, energies=None, forces=None, stresses=None)

Method to convert structures and their properties in to datapool format.

• Parameters

  • structures ([Structure]) – The list of Pymatgen Structure object.

  • energies ([float]) – The list of total energies of each structure in structures list.

  • forces ([np.array]) – List of (m, 3) forces array of each structure with m atoms in structures list. m can be varied with each single structure case.

  • stresses (list) – List of (6, ) virial stresses of each structure in structures list.

• Returns ([dict])

maml.utils._data_conversion.to_array(x)

Convert x into numerical array :param x: x can be a dataframe, a list or an array

return np.ndarray.

maml.utils._data_split module

Data split.

class maml.utils._data_split.DataSplitter()

Bases: MSONable

Data splitter base class.

split(mat_ids, **kwargs)

Split the mat_ids, optionally one can provide targets. This is useful in stratified split.

• Parameters mat_ids (list) – list of material ids

Returns: (train_ids, val_ids, test_ids) or

(train_ids, test_ids)

class maml.utils._data_split.ShuffleSplitter(ratios: str = ‘80/10/10’, delim: str = ‘/’, random_seed: int | None = None)

Bases: DataSplitter

Randomly shuffe the material ids and split the ids into given ratios.

split(mat_ids, **kwargs)

Randomly split the mat_ids :param mat_ids: material ids :type mat_ids: list

Returns:

maml.utils._dummy module

Dummy test systems.

maml.utils._dummy.feature_dim_from_test_system(describer)

Get feature size from a test system.

• Parameters describer (BaseDescriber) – describers instance

maml.utils._dummy.get_describer_dummy_obj(instance)

For a describers, get a dummy object for transform_one. This relies on the type hint.

• Parameters instance (BaseDescriber) – describers instance

maml.utils._inspect module

Inspect function args.

maml.utils._inspect.get_full_args(func: Callable)

Get args from function.

• Parameters func (callable) – function to determine the args

maml.utils._inspect.get_param_types(func)

Get param and type info.

• Parameters func (callable) – function to determine the arg types

maml.utils._jit module

Simple numba utility. Some functions can excelerated substantially with numba.

maml.utils._jit.njit(func: Callable)

Dummy decorator, returns the original function :param func: function to be wrapped. :type func: Callable

Returns: decorated function

maml.utils._lammps module

LAMMPS utility.

maml.utils._lammps._get_atomic_mass(element_or_specie: str)

Get atomic mass from element or specie string.

• Parameters element_or_specie (str) – specie or element string

Returns: float mass

maml.utils._lammps._get_charge(element_or_specie: str | Element | Species)

Get charge from element or specie.

• Parameters element_or_specie (str* or Element or *Species) – element or specie

Returns: charge float

maml.utils._lammps.check_structures_forces_stresses(structures: list[Structure], forces: list | None = None, stresses: list | None = None, stress_format: str = ‘VASP’, return_none: bool = True)

Check structures, forces and stresses. The forces and stress are dependent on the lattice orientation. This function will rotate the structures and the corresponding forces and structures to lammps format [[ax, 0, 0], [bx, by, 0], [cx, cy, cz]].

The lattice are formed by the row vectors.

• Parameters

  • structures (list) – list of structures

  • forces (list) – list of force matrixs (m, 3)

  • stresses (list) – list of stress vectors

  • stress_format (str) – stress format, choose from “VASP”, “LAMMPS”, “SNAP”

  • return_none (bool) – whether to return list of None for forces and stresses

Returns: structures [forces], [stresses]

maml.utils._lammps.get_lammps_lattice_and_rotation(structure: Structure, origin=(0, 0, 0))

Transform structure to lammps compatible structure. The lattice and rotation matrix are returned.

• Parameters

  • structure (Structure) – pymatgen structure

  • origin (tuple) – origin coordinates

Returns: new lattice, rotation symmetry operator, rotation matrix

maml.utils._lammps.stress_format_change(stress: np.ndarray | list[float], from_format: str, to_format: str)

Convert stress format from from_format to to_format :param stress: length-6 stress vector :type stress: list of float :param from_format: choose from “VASP”, “LAMMPS”, “SNAP” :type from_format: str :param to_format: choose from “VASP”, “LAMMPS”, “SNAP”. :type to_format: str

Returns: list of float stress vector

maml.utils._lammps.stress_list_to_matrix(stress: np.ndarray | list[float], stress_format: str = ‘VASP’)

convert a length-6 stress list to stress matrix 3x3.

• Parameters

  • stress (list* of *float) – list of stress

  • stress_format (str) – Supported formats are the follows VASP: xx, yy, zz, xy, yz, xz LAMMPS: xx, yy, zz, xy, zx, yz SNAP: xx, yy, zz, yz, xz, xy

Returns: 3x3 stress matrix

maml.utils._lammps.stress_matrix_to_list(stress_matrix: ndarray, stress_format: str = ‘VASP’)

Stress matrix to list representation :param stress_matrix: stress matrix 3x3 :type stress_matrix: np.ndarray :param stress_format: stress list format :type stress_format: str

Returns: list of float stress vector.

maml.utils._lammps.write_data_from_structure(structure: Structure, filename: str, ff_elements: list[str] | None = None, significant_figures: int = 6, origin: tuple = (0, 0, 0))

Write structure to lammps data file, this is to speed up pymatgen LammpsData.

Args:a

structure (Structure): pymatgen structure
filename (str): filename
ff_elements (list of str): elements to be considered
significant_figures (int): significant figures of floats in output
origin (tuple): origin coordinates

maml.utils._material module

Materials utils.

maml.utils._material.to_composition(obj: Composition | Molecule | Structure | str)

Convert str/structure or composition to compositions.

• Parameters obj (str/structure/composition) – object to convert

• Returns Composition object

maml.utils._preprocessing module

Target preprocessing.

class maml.utils._preprocessing.DummyScaler()

Bases: MSONable

Dummy scaler does nothing.

classmethod from_training_data(structures: list[StructureOrMolecule], targets: VectorLike, is_intensive: bool = True)

• Parameters

  • structures (list) – list of structures/molecules

  • targets (list) – vector of target properties

  • is_intensive (bool) – whether the target is intensive

Returns: DummyScaler.

static inverse_transform(transformed_target: float, n: int = 1)

return as it is :param transformed_target: transformed target :type transformed_target: float :param n: number of atoms :type n: int

• Returns transformed_target.

static transform(target: float, n: int = 1)

• Parameters

  • target (float) – target numerical value

  • n (int) – number of atoms

• Returns target.

class maml.utils._preprocessing.Scaler()

Bases: MSONable

Base Scaler class. It implements transform and inverse_transform. Both methods will take number of atom as the second parameter in addition to the target property.

inverse_transform(transformed_target: float, n: int = 1)

Inverse transform of the target :param transformed_target: transformed target :type transformed_target: float :param n: number of atoms :type n: int

• Returns target.

transform(target: float, n: int = 1)

Transform the target values into new target values :param target: target numerical value :type target: float :param n: number of atoms :type n: int

• Returns scaled target.

class maml.utils._preprocessing.StandardScaler(mean: float = 0.0, std: float = 1.0, is_intensive: bool = True)

Bases: Scaler

Standard scaler with consideration of extensive/intensive quantity For intensive quantity, the mean is just the mean of training data, and std is the std of training data For extensive quantity, the mean is the mean of target/atom, and std is the std for target/atom .. method:: transform(self, target, n=1)

standard scaling the target and.

classmethod from_training_data(structures: list[StructureOrMolecule], targets: VectorLike, is_intensive: bool = True)

Generate a target scaler from a list of input structures/molecules, a target value vector and an indicator for intensiveness of the property :param structures: list of structures/molecules :type structures: list :param targets: vector of target properties :type targets: list :param is_intensive: whether the target is intensive :type is_intensive: bool

Returns: new instance.

inverse_transform(transformed_target: float, n: int = 1)

Inverse transform of the target :param transformed_target: transformed target :type transformed_target: float :param n: number of atoms :type n: int

• Returns original target.

transform(target: float, n: int = 1)

Transform numeric values according the mean and std, plus a factor n :param target: target numerical value :type target: float :param n: number of atoms :type n: int

• Returns scaled target.

maml.utils._signal_processing module

Signal processing utils.

maml.utils._signal_processing.cwt(z: np.ndarray, widths: np.ndarray, wavelet: str | Callable = ‘morlet2’, **kwargs)

The scalogram of the signal :param z: 1D signal array :type z: np.ndarray :param widths: wavelet widths :type widths: np.ndarray :param wavelet: wavelet name :type wavelet: str

Returns: 2D scalogram.

maml.utils._signal_processing.fft_magnitude(z: ndarray)

Discrete Fourier Transform the signal z and return the magnitude of the coefficients :param z: 1D signal array :type z: np.ndarray

Returns: 1D magnitude.

maml.utils._signal_processing.get_sp_method(sp_method: str | Callable)

Providing a signal processing method name return the callable :param sp_method: name of the sp function :type sp_method: str

Returns: callable for signal processing.

maml.utils._signal_processing.spectrogram(z: np.ndarray, return_time_freq: bool = False)

The spectrogram of the signal :param z: 1D signal array :type z: np.ndarray :param return_time_freq: whether to return time and frequency :type return_time_freq: bool

Returns: 2D spectrogram.

maml.utils._signal_processing.wvd(z: np.ndarray, return_all: bool = False)

Wigner Ville Distribution calculator :param z: signal 1D :type z: np.ndarray :param return_all: whether to return time and freq info, default

only return the wvd information

Returns: NxN wvd matrix.

maml.utils._stats module

Utils for describers.

class maml.utils._stats.Stats()

Bases: object

Calculate the stats of a list of values. This is particularly helpful when you want to convert lists of values of different lengths to uniform length for machine learning purposes.

supported

allowed_stats(_ = [‘max’, ‘min’, ‘range’, ‘mode’, ‘mean_absolute_deviation’, ‘mean_absolute_error’, ‘moment’, ‘mean’, ‘inverse_mean’, ‘average’, ‘std’, ‘skewness’, ‘kurtosis’, ‘geometric_mean’, ‘power_mean’, ‘shifted_geometric_mean’, ‘harmonic_mean’_ )

static average(data: list[float], weights: list[float] | None = None)

Weighted average.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: average value

static geometric_mean(data: list[float], weights: list[float] | None = None)

Geometric mean of the data.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: geometric mean of the distribution

static harmonic_mean(data: list[float], weights: list[float] | None = None)

harmonic mean of the data.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: harmonic mean of the distribution

static inverse_mean(data: list[float], weights: list[float] | None = None)

inverse mean.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: average value

static kurtosis(data: list[float], weights: list[float] | None = None)

Kurtosis of the distribution.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: Kurtosis of the distribution

static max(data: list[float], weights: list[float] | None = None)

Max of value Args:31

data (list): list of float data weights (list): optional weights.

Returns: maximum value

static mean(data: list[float], weights: list[float] | None = None)

Weighted average.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: average value

static mean_absolute_deviation(data: list[float], weights: list[float] | None = None)

mean absolute deviation.

• Parameters

  • data (list) – list of float data

  • weights (list) – optional weights

Returns: mean absolute deviation

static mean_absolute_error(data: list[float], weights: list[float] | None = None)

mean absolute error.

• Parameters

  • data (list) – list of float data

  • weights (list) – optional weights

Returns: mean absolute error

static min(data: list[float], weights: list[float] | None = None)

min of value :param data: list of float data :type data: list :param weights: optional weights. :type weights: list

Returns: minimum value

static mode(data: list[float], weights: list[float] | None = None)

Mode of data, if multiple entries have equal counts, compute the average of those.

• Parameters

  • data (list) – list of float data

  • weights (list) – optional weights

Returns: mode of values, i.e., max - min

static moment(data: list[float], weights: list[float] | None = None, order: int | None = None, max_order: int | None = None)

Moment of probability mass function.

order = 1 means weighted mean order = 2 means standard deviation order > 2 corresponds to higher order moment to

the 1./order power

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data points

  • order (int) – moment order

  • max_order (int) – if set, it will overwrite order

Returns: float or list of floats

static power_mean(data: list[float], weights: list[float] | None = None, p: int = 1)

power mean https://en.wikipedia.org/wiki/Generalized_mean.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

  • p (int) – power

Returns: power mean of the distribution

static range(data: list[float], weights: list[float] | None = None)

Range of values :param data: list of float data :type data: list :param weights: optional weights. :type weights: list

Returns: range of values, i.e., max - min

static shifted_geometric_mean(data: list[float], weights: list[float] | None = None, shift: float = 100)

Since we cannot calculate the geometric means on negative or zero values, we can first shift all values to positive and then calculate the geometric mean afterwards, we shift the computed geometric mean back by a shift value.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

  • shift (float) – shift value

Returns: geometric mean of the distribution

static skewness(data: list[float], weights: list[float] | None = None)

Skewness of the distribution.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: Skewness of the distribution

static std(data: list[float], weights: list[float] | None = None)

Standard deviation.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

Returns: Standard deviation

maml.utils._stats._add_allowed_stats(cls)

Decorate to add allowed_stats to the Stats class.

• Parameters cls – Stats class

Returns: Stats class with allowed_stats attributes

maml.utils._stats._convert_a_or_b(v: str, a=<class ‘int’>, b=None)

maml.utils._stats._moment_symbol_conversion(moment_symbol: str)

maml.utils._stats._root_moment(data, weights, order)

Auxiliary function to compute moment.

• Parameters

  • data (list) – list of float data

  • weights (list* or *None) – weights for each data point

  • order (int) – order of moment

Returns: moment of order

maml.utils._stats.get_full_stats_and_funcs(stats: list)

Get expanded stats function name str and the corresponding function callables.

• Parameters stats (list) – a list of stats names, e.g, [‘mean’, ‘std’, ‘moment:1:None’]

Returns: list of stats names, list of stats callable

maml.utils._stats.stats_list_conversion(stats_list: list[str])

Convert a list of stats str into a fully expanded list. This applies mainly to stats that can return a list of values, e.g., moment with max_order > 1.

• Parameters stats_list (list) – list of stats str

Returns: list of expanded stats str

maml.utils._tempfile module

Temporary directory and file creation utilities. This file is adapted from monty.tempfile.

class maml.utils._tempfile.MultiScratchDir(rootpath: str | Path, n_dirs: int = 1, create_symbolic_link: bool = False, copy_from_current_on_enter: bool = False, copy_to_current_on_exit: bool = False)

Bases: object

Creates a “with” context manager that automatically handles creation of temporary directories (utilizing Python’s build in temp directory functions) and cleanup when done. The main difference between this class and monty ScratchDir is that multiple temp directories are created here. It enables the running of multiple jobs simultaneously in the directories The way it works is as follows:

1. Create multiple temp dirs in specified root path.

2. Optionally copy input files from current directory to temp dir.

3. User loops among all directories

4. User performs specified operations in each directories

5. Change back to original directory.

6. Delete temp dir.

SCR_LINK(_ = ‘scratch_link_ )

tempdirs(: list[str )

maml.utils._tempfile._copy_r_with_suffix(src: str, dst: str, suffix: Any | None = None)

Implements a recursive copy function similar to Unix’s “cp -r” command. Surprisingly, python does not have a real equivalent. shutil.copytree only works if the destination directory is not present.

• Parameters

  • src (str) – Source folder to copy.

  • dst (str) – Destination folder.

maml.utils._typing module

Define several typing for convenient use.

maml.utils._value_profile module

ValueProfile return values according to certain settings. For example, one can design a linearly increasing value profile, a sinusoidal profile or a constant profile, depending on the step, and previous values.

class maml.utils._value_profile.ConstantValue(value: float, **kwargs)

Bases: ValueProfile

Return constant value.

get_value()

Return constant value.

class maml.utils._value_profile.LinearProfile(value_start: float, value_end: float = 0.0, max_steps: int = 100, **kwargs)

Bases: ValueProfile

LinearProfile by setting starting value and the rate of value change. The profile can be initialized either by [value_start, value_end, max_step] or [value_start, rate].

get_value()

Get LinearProfile value Returns: float.

class maml.utils._value_profile.ValueProfile(max_steps: int | None = None, **kwargs)

Bases: object

Base class for ValueProfile. The base class has the following methods .. method:: increment_step(self)

add one to step

get_value(self)

abstract method that return the value.

get_value()

abstract method that returns the current value Returns: value float.

increment_step()

Increase step attribute by one.