pinnicle

pinnicle.pinn module

class pinnicle.pinn.PINN(params={}, loadFrom='')

Bases: object

a basic PINN model

check_path(path, loadOnly=False)

check the path, set to default, and create folder if needed

compile(opt=None, loss=None, lr=None, loss_weights=None, decay=None)

compile the model

load_model(path='', epochs=-1, subfolder='pinn', name='model', fileformat='')

laod the neural network from saved model

load_setting(path='', filename='params.json')

load the settings from file

plot_history(path='')

plot training history

plot_predictions(path='', filename='2Dsolution.png', **kwargs)

plot model predictions

Parameters:

• path (Path, str) – Path to save the figures

• filename (str) – name to save the figures, if set to None, then the figure will not be saved

• X_ref (dict) – Coordinates of the reference solutions, if None, then just plot the predicted solutions

• u_ref (dict) – Reference solutions, if None, then just plot the predicted solutions

• cols (int) – Number of columns of subplot

save_history(path='')

save training history

save_model(path='', subfolder='pinn', name='model')

save the neural network to the hard disk

save_setting(path='', subfolder='pinn')

save settings from self.params.param_dict

setup()

setup the model according to self.params from the constructor

train(iterations=0)

train the model

update_callbacks(params=None)

update callback functions for the training according to the settings in params

update_parameters(params)

update self.params according to the input params. If key already exists, update the value; if not, add the pair

update_training_data(training_data)

update data set used for the training, the order follows ‘output_variables’

pinnicle.parameter module

class pinnicle.parameter.DataParameter(param_dict={})

Bases: ParameterBase

list of all data used

check_consistency()

check consistency of the parameter data

set_default()

default parameters

update()

convert dict to class SingleDataParameter

class pinnicle.parameter.DomainParameter(param_dict={})

Bases: ParameterBase

parameters of domain

check_consistency()

need to provide start and end time if solving a time dependent problem

set_default()

set default values

class pinnicle.parameter.EquationParameter(param_dict={})

Bases: ParameterBase

parameter of equations

check_consistency()

check consistency of the parameter data

classmethod create(equation_type, **kwargs)

set_default()

set default values

set_parameters(pdict: dict)

overwrite the default function, so that for ‘scalar_parameters’, only update the dict

subclasses = {'DUMMY': <class 'pinnicle.physics.dummy.DummyEquationParameter'>, 'MC': <class 'pinnicle.physics.continuity.MCEquationParameter'>, 'MOLHO': <class 'pinnicle.physics.stressbalance.MOLHOEquationParameter'>, 'Mass transport': <class 'pinnicle.physics.continuity.ThicknessEquationParameter'>, 'SSA': <class 'pinnicle.physics.stressbalance.SSAEquationParameter'>, 'SSA Taub': <class 'pinnicle.physics.stressbalance.SSATauEquationParameter'>, 'SSA_SHELF': <class 'pinnicle.physics.iceshelf.SSAShelfEquationParameter'>, 'SSA_SHELF_VB': <class 'pinnicle.physics.iceshelf.SSAShelfVariableBEquationParameter'>, 'SSA_VB': <class 'pinnicle.physics.stressbalance.SSAVariableBEquationParameter'>, 'Time_Invariant': <class 'pinnicle.physics.timeinvariant.TimeInvariantConstraintParameter'>, 'Weertman': <class 'pinnicle.physics.friction.WeertmanFrictionParameter'>}

class pinnicle.parameter.LossFunctionParameter(param_dict={})

Bases: ParameterBase

parameter of customize loss function

check_consistency()

check consistency of the parameter data

set_default()

set default values

class pinnicle.parameter.NNParameter(param_dict={})

Bases: ParameterBase

parameters of nn

check_consistency()

check consistency of the parameter data

is_input_scaling()

if the input boundaries are provided, or fourier feature transform is used

is_output_scaling()

if the output boundaries are provided

set_default()

default values:

set_parameters(pdict: dict)

find all the keys from pdict which are avalible in the class, update the values

update()

update the input_size for fourier feature transform

class pinnicle.parameter.ParameterBase(param_dict)

Bases: ABC

Abstract class of parameters in the experiment

abstractmethod check_consistency()

check consistency of the parameter data

has_keys(keys)

if all the keys are in the class, return true, otherwise return false

abstractmethod set_default()

set default values

set_parameters(pdict: dict)

find all the keys from pdict which are avalible in the class, update the values

update()

after set_parameter, make some necessary update of the parameters

class pinnicle.parameter.Parameters(param_dict={})

Bases: ParameterBase

parameters of the pinn, including domain, data, nn, and physics

check_consistency()

check consistency of the parameter data

set_default()

set default values

set_parameters(param_dict)

find all the keys from pdict which are avalible in the class, update the values

update()

update parameters according to the input

class pinnicle.parameter.PhysicsParameter(param_dict={})

Bases: ParameterBase

parameter of physics

check_consistency()

check consistency of the parameter data

set_default()

set default values

setup_equations()

translate the input dict to subclass of EquationParameter(), and save back to the values in self.equations

class pinnicle.parameter.SingleDataParameter(param_dict={})

Bases: ParameterBase

parameters of a single data file

check_consistency()

check consistency of the parameter data

set_default()

default settings

update()

update name_map according to data_size

class pinnicle.parameter.TrainingParameter(param_dict={})

Bases: ParameterBase

parameter of training

check_callbacks()

check if any of the following variable is given from setting

check_consistency()

check consistency of the parameter data

has_EarlyStopping()

check if param has the min_delta or patience for early stopping

has_MiniBatch()

check if param has mini batch

has_ModelCheckpoint()

check if param has checkpoint=True for checkpointing

has_PDEPointResampler()

check if param has the period for resampler

set_default()

set default values

update()

convert dict to class LossFunctionParameter