aim2dat.ml.transformers¶

Scikit learn Transformer classes extracting features from crystals or molecules.

Module Contents¶

Classes¶

class aim2dat.ml.transformers.StructureACSFTransformer(r_cut=7.5, g2_params=None, g3_params=None, g4_params=None, g5_params=None, elements=None, periodic=False, sparse=False, dscribe_n_jobs=1, dscribe_only_physical_cores=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseDscribeTransformer

Extract ACSF descriptor as defined in doi:10.1063/1.3553717. This transformer class is based on the implementations of the dscribe python package.

Variables:¶

r_cut : float¶

Cutoff value.

g2_params : np.array¶

List of pairs of eta and R_s values for the G^2 functions.

g3_params : np.array¶

List of kappa values for the G^3 functions.

g4_params : np.array¶

List of triplets of eta, zeta and lambda values for G^4 functions.

g5_params : np.array¶

List of triplets of eta, zeta and lambda values for G^5 functions.

elements : list¶

List of atomic numbers or chemical symbols.

periodic : bool¶

Whether to consider periodic boundary conditions.

sparse : bool¶

Whether to return a sparse matrix or a dense array.

dscribe_n_jobs : int¶

Number of jobs used by dscribe to calculate the interaction matrix.

dscribe_only_physical_cores : bool¶

Whether to only use physicsl cores.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureChemOrderTransformer(r_max=15.0, max_shells=3, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseStructureTransformer

Extract Warren Cowley like order parameters for each element as defined in doi:10.1103/PhysRevB.96.024104.

Variables:¶

r_max : float (optional)¶

Cut-off value for the maximum distance between two atoms in angstrom.

max_shells : int (optional)¶

Number of neighbour shells that are evaluated.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureCompositionTransformer(distinguish_kinds=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseStructureTransformer

Extract fractional concentrations of elements or kinds.

Variables:¶

distinguish_kinds : bool (optional)¶

Whether to use kinds instead of elements.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureCoordinationTransformer(r_max=15.0, method='minimum_distance', min_dist_delta=0.1, n_nearest_neighbours=5, econ_tolerance=0.5, econ_conv_threshold=0.001, voronoi_weight_type='rel_solid_angle', voronoi_weight_threshold=0.5, feature_types=('nrs_avg', 'nrs_stdev', 'nrs_max', 'nrs_min', 'distance_avg', 'distance_stdev', 'distance_max', 'distance_min'), n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseStructureTransformer

Extract coordination numbers and distances between elements or kinds.

Variables:¶

r_max : float (optional)¶

Cut-off value for the maximum distance between two atoms in angstrom.

method : str (optional)¶

Method used to calculate the coordination environment. The default value is 'minimum_distance'.

min_dist_delta : float (optional)¶

Tolerance parameter that defines the relative distance from the nearest neighbour atom for the 'minimum_distance' method.

n_nearest_neighbours : int (optional)¶

Number of neighbours that are considered coordinated for the 'n_neighbours' method.

econ_tolerance : float (optional)¶

Tolerance parameter for the econ method.

econ_conv_threshold : float (optional)¶

Convergence threshold for the econ method.

okeeffe_weight_threshold : float (optional)

Threshold parameter to distinguish indirect and direct neighbour atoms for the 'okeeffe'.

feature_types : tuple or str (optional)¶

Tuple of features that are extracted. Supported options are: 'nrs_avg', 'nrs_stdev', 'nrs_max', 'nrs_min', 'distance_avg', 'distance_stdev', 'distance_max' and 'distance_min'.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

feature_types	Feature types that are included.
precomputed_properties	Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property feature_types¶

Feature types that are included.

Type:¶

tuple or str

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureDensityTransformer(distinguish_kinds=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseStructureTransformer

Extract density of each element or kind.

Variables:¶

distinguish_kinds : bool (optional)¶

Whether to use kinds instead of elements.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureFFPrintTransformer(r_max=15.0, delta_bin=0.005, sigma=10.0, distinguish_kinds=False, add_header=False, use_weights=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseStructureTransformer

Extract the F-fingerprint for each element-pair as defined in doi:10.1103/PhysRevB.96.024104.

Variables:¶

r_max : float (optional)¶

Cut-off value for the maximum distance between two atoms in angstrom.

delta_bin : float (optional)¶

Bin size to descritize the function in angstrom.

sigma : float (optional)¶

Smearing parameter for the Gaussian function.

distinguish_kinds : bool (optional)¶

Whether different kinds should be distinguished e.g. Ni0 and Ni1 would be considered as different elements if True.

add_header : bool¶

Add leading entries that describe the weights and composition for the ffprint kernels.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureMBTRTransformer(geometry={'function': 'inverse_distance'}, grid={'min': 0, 'max': 1, 'n': 100, 'sigma': 0.1}, weighting={'function': 'exp', 'scale': 1.0, 'threshold': 0.001}, normalize_gaussians=True, normalization='l2', elements=None, periodic=False, sparse=False, dscribe_n_jobs=1, dscribe_only_physical_cores=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseDscribeTransformer

Extract MBTR descriptor as defined in doi:10.1088/2632-2153/aca005. This transformer class is based on the implementations of the dscribe python package.

Variables:¶

geometry : dict¶

Setup the geometry function.

grid : dict¶

Setup the discretization grid.

weighting : dict¶

Setup the weighting function and its parameters.

normalize_gaussians : bool¶

Whether to normalize the gaussians to an area of 1.

normalization : str¶

Method for normalizing. Supported options are 'none', 'l2', 'n_atoms', 'valle_oganov'.

elements : list¶

List of atomic numbers or chemical symbols.

periodic : bool¶

Whether to consider periodic boundary conditions.

sparse : bool¶

Whether to return a sparse matrix or a dense array.

dscribe_n_jobs : int¶

Number of jobs used by dscribe to calculate the interaction matrix.

dscribe_only_physical_cores : bool¶

Whether to only use physicsl cores.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureMatrixTransformer(matrix_type='coulomb', n_atoms_max=None, enforce_real=False, permutation='eigenspectrum', sigma=None, seed=None, sparse=False, ewald_accuracy=1e-05, ewald_w=1, ewald_r_cut=None, ewald_g_cut=None, ewald_a=None, dscribe_n_jobs=1, dscribe_only_physical_cores=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseStructureTransformer

Extract features based on interaction matrices as defined in doi:10.1002/qua.24917. This transformer class is based on the implementations of the dscribe python package.

Variables:¶

matrix_type : str¶

Matrix type. Supported options are 'coulomb', 'ewald_sum' or 'sine'.

permutation : str¶

Defines the output format. Options are: 'none', 'sorted_l2', 'eigenspectrum' or 'random'.

sigma : float¶

Standar deviation of the Gaussian distributed noise when using 'random' for permutation.

seed : int¶

Seed for the random numbers in case 'random' is chosen for the permutation attibute.

sparse : bool¶

Whether to return a sparse matrix or a dense 1D array.

ewald_accuracy : float¶

Accuracy threshold for the Ewald sum.

ewald_w : int¶

Weight parameter.

ewald_r_cut : float or None¶

Real space cutoff parameter.

ewald_g_cut : float or None¶

Reciprocal space cutoff parameter.

ewald_a : float or None¶

Parameter controlling the width of the Gaussian functions.

dscribe_n_jobs : int¶

Number of jobs used by dscribe to calculate the interaction matrix.

dscribe_only_physical_cores : bool¶

Whether to only use physicsl cores.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructurePRDFTransformer(r_max=15.0, delta_bin=0.005, distinguish_kinds=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseStructureTransformer

Extract the partial radial distribution function for each element-pair as defined in doi:10.1103/PhysRevB.89.205118.

Variables:¶

r_max : float (optional)¶

Cut-off value for the maximum distance between two atoms in angstrom.

delta_bin : float (optional)¶

Bin size to descritize the function in angstrom.

distinguish_kinds : bool (optional)¶

Whether different kinds should be distinguished e.g. Ni0 and Ni1 would be considered as different elements if True.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.

class aim2dat.ml.transformers.StructureSOAPTransformer(r_cut=7.5, n_max=8, l_max=6, sigma=1.0, rbf='gto', weighting=None, compression={'mode': 'off', 'species_weighting': None}, average='off', elements=None, periodic=False, sparse=False, dscribe_n_jobs=1, dscribe_only_physical_cores=False, n_procs=1, chunksize=50, verbose=True)[source]¶

Bases: _BaseDscribeTransformer

Extract SOAP descriptor as defined in doi:10.1103/PhysRevB.87.184115. This transformer class is based on the implementations of the dscribe python package.

Variables:¶

r_cut : float¶

Cutoff value.

n_max : int¶

The number of radial basis functions.

l_max : int¶

The maximum degree of spherical harmonics.

sigma : float¶

The standard deviation of the gaussians.

rbf : str¶

The radial basis functions to use. Supported options are: 'gto' or 'polynomial'.

weighting : dict¶

Contains the options which control the weighting of the atomic density.

compression : dict¶

Feature compression options.

average : str¶

The averaging mode over the centers of interest. Supported options are: 'off', 'inner' or 'outer'.

elements : list¶

List of atomic numbers or chemical symbols.

periodic : bool¶

Whether to consider periodic boundary conditions.

sparse : bool¶

Whether to return a sparse matrix or a dense array.

dscribe_n_jobs : int¶

Number of jobs used by dscribe to calculate the interaction matrix.

dscribe_only_physical_cores : bool¶

Whether to only use physicsl cores.

n_procs : int (optional)¶

Number of parallel processes.

chunksize : int (optional)¶

Number of structures handed to each process at once.

verbose : bool (optional)¶

Whether to print a progress bar.

Overview

Properties¶

precomputed_properties

Precomputed properties given as list of tuples consisting of input parameters and

Methods¶

add_precomputed_properties(parameters, structure_operations)	Add precomputed properties.
clear_precomputed_properties()	Clear all precomputed properties.
fit(X, y)	Fit function that determines the number of features.
fit_transform(X, y, **fit_params)	Fit to data, then transform it.
get_feature_names_out(input_features)	Get feature names.
get_metadata_routing()	Get metadata routing of this object.
get_params(deep)	Get parameters for this estimator.
precompute_parameter_space(param_grid, X)	Precompute and store structural properties to be reused later e.g. for a grid search.
set_output(*None, transform)	Set output container.
set_params(**params)	Set the parameters of this estimator.
transform(X)	Transform structures to features.

property precomputed_properties¶

Precomputed properties given as list of tuples consisting of input parameters and StructureOperations object.

Type:¶

list

add_precomputed_properties(parameters, structure_operations)¶

Add precomputed properties.

Parameters:¶

• parameters (dict) – Dictionary of input parameters.

• structure_operations (StructureOperations) – StructureOperations object storing the properties according to the input parameters.

clear_precomputed_properties()¶

Clear all precomputed properties.

fit(X, y=None)¶

Fit function that determines the number of features.

Parameters:¶

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

• y (list (optional)) – list of target property.

Returns:¶

self – Transformer object.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:¶

• X (array-like of shape (n_samples, n_features)) – Input samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

• **fit_params (dict) – Additional fit parameters.

Returns:¶

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names_out(input_features=None)¶

Get feature names.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:¶

routing (MetadataRequest) – A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:¶

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:¶

params (dict) – Parameter names mapped to their values.

precompute_parameter_space(param_grid, X)¶

Precompute and store structural properties to be reused later e.g. for a grid search.

Parameters:¶

• param_grid (list or dict) – Dictionary or list of dictionaries of input parameters.

• X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

set_output(*, transform=None)¶

Set output container.

See sphx_glr_auto_examples_miscellaneous_plot_set_output.py for an example on how to use the API.

Parameters:¶

transform ({“default”, “pandas”}, default=None) – Configure output of transform and fit_transform.

• “default”: Default output format of a transformer

• “pandas”: DataFrame output

• “polars”: Polars output

• None: Transform configuration is unchanged

New in version 1.4: “polars” option was added.

Returns:¶

self (estimator instance) – Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:¶

**params (dict) – Estimator parameters.

Returns:¶

self (estimator instance) – Estimator instance.

transform(X)¶

Transform structures to features.

Parameters:¶

X (list or aim2dat.strct.StructureCollection) – List of structures or StructureCollection.

Returns:¶

numpy.array – Nested array of features.