aspire.transforms ================= .. py:module:: aspire.transforms Attributes ---------- .. autoapisummary:: aspire.transforms.logger Classes ------- .. autoapisummary:: aspire.transforms.BaseTransform aspire.transforms.IdentityTransform aspire.transforms.CompositeTransform aspire.transforms.FlowTransform aspire.transforms.PeriodicTransform aspire.transforms.BoundedTransform aspire.transforms.ProbitTransform aspire.transforms.LogitTransform aspire.transforms.AffineTransform aspire.transforms.FlowPreconditioningTransform Module Contents --------------- .. py:data:: logger .. py:class:: BaseTransform(xp, dtype=None) Base class for data transforms. :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: xp .. py:attribute:: dtype :value: None .. py:method:: fit(x) :abstractmethod: Fit the transform to the data. .. py:method:: forward(x) :abstractmethod: .. py:method:: inverse(y) :abstractmethod: .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:method:: save(h5_file, path = 'data_transform') Save config + any fitted state into an HDF5 file. .. py:method:: load(h5_file, path = 'data_transform', strict = False) :classmethod: Reconstruct transform from file. :param h5_file: The HDF5 file to load from. :type h5_file: :py:class:`h5py.File` :param path: The path in the HDF5 file where the transform is stored. :type path: :py:class:`str`, *optional* :param strict: If True, raise an error if the class in the file does not match cls. If False, load the class specified in the file. Default is False. :type strict: :py:class:`bool`, *optional* .. py:class:: IdentityTransform(xp, dtype=None) Bases: :py:obj:`BaseTransform` Identity transform that does nothing to the data. .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) .. py:method:: inverse(y) .. py:class:: CompositeTransform(parameters, periodic_parameters = None, prior_bounds = None, bounded_to_unbounded = True, bounded_transform = 'probit', affine_transform = True, device=None, xp = None, eps = 1e-06, dtype = None) Bases: :py:obj:`BaseTransform` Base class for data transforms. :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: parameters .. py:attribute:: periodic_parameters :value: [] .. py:attribute:: bounded_to_unbounded :value: True .. py:attribute:: bounded_transform :value: 'probit' .. py:attribute:: affine_transform :value: True .. py:attribute:: eps :value: 1e-06 .. py:attribute:: device :value: None .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) .. py:method:: inverse(x) .. py:method:: new_instance(xp=None, dtype = None) .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:class:: FlowTransform(parameters, prior_bounds = None, bounded_to_unbounded = True, bounded_transform = 'probit', affine_transform = True, device=None, xp=None, eps=1e-06, dtype=None) Bases: :py:obj:`CompositeTransform` Subclass of CompositeTransform that uses a Flow for transformations. Does not support periodic transforms. .. py:method:: new_instance(xp=None) .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:class:: PeriodicTransform(lower, upper, xp, dtype=None) Bases: :py:obj:`BaseTransform` Base class for data transforms. :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: name :type: str :value: 'periodic' .. py:attribute:: requires_prior_bounds :type: bool :value: True .. py:attribute:: lower .. py:attribute:: upper .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) .. py:method:: inverse(y) .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:class:: BoundedTransform(lower, upper, xp, dtype = None) Bases: :py:obj:`BaseTransform` Base class for bounded transforms. Maps from [lower, upper] to [0, 1] and vice versa using a linear scaling. If the interval [lower, upper] is too small, it will shift by the midpoint. Must be subclassed to implement specific transforms (e.g., Probit, Logit). :param lower: The lower bound of the interval. :type lower: :py:class:`Array` :param upper: The upper bound of the interval. :type upper: :py:class:`Array` :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: name :type: str :value: 'bounded' .. py:attribute:: requires_prior_bounds :type: bool :value: True .. py:attribute:: lower .. py:attribute:: upper .. py:method:: to_unit_interval(x) Map from [lower, upper] to [0, 1]. :param x: The input array to be mapped. :type x: :py:class:`Array` :returns: A tuple containing the mapped array and the log absolute determinant Jacobian. :rtype: :py:class:`tuple[Array`, :py:class:`Array]` .. py:method:: from_unit_interval(y) Map from [0, 1] to [lower, upper]. :param y: The input array to be mapped. :type y: :py:class:`Array` :returns: A tuple containing the mapped array and the log absolute determinant Jacobian. :rtype: :py:class:`tuple[Array`, :py:class:`Array]` .. py:method:: interval_check(lower, upper) Check if the interval [lower, upper] is too small .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) :abstractmethod: .. py:method:: inverse(y) :abstractmethod: .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:class:: ProbitTransform(lower, upper, xp, eps=1e-06, dtype=None) Bases: :py:obj:`BoundedTransform` Base class for bounded transforms. Maps from [lower, upper] to [0, 1] and vice versa using a linear scaling. If the interval [lower, upper] is too small, it will shift by the midpoint. Must be subclassed to implement specific transforms (e.g., Probit, Logit). :param lower: The lower bound of the interval. :type lower: :py:class:`Array` :param upper: The upper bound of the interval. :type upper: :py:class:`Array` :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: name :type: str :value: 'probit' .. py:attribute:: requires_prior_bounds :type: bool :value: True .. py:attribute:: eps :value: 1e-06 .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) .. py:method:: inverse(y) .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:class:: LogitTransform(lower, upper, xp, eps = 1e-06, dtype = None) Bases: :py:obj:`BoundedTransform` Base class for bounded transforms. Maps from [lower, upper] to [0, 1] and vice versa using a linear scaling. If the interval [lower, upper] is too small, it will shift by the midpoint. Must be subclassed to implement specific transforms (e.g., Probit, Logit). :param lower: The lower bound of the interval. :type lower: :py:class:`Array` :param upper: The upper bound of the interval. :type upper: :py:class:`Array` :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: name :type: str :value: 'logit' .. py:attribute:: requires_prior_bounds :type: bool :value: True .. py:attribute:: eps :value: 1e-06 .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) .. py:method:: inverse(y) .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:class:: AffineTransform(xp, dtype=None) Bases: :py:obj:`BaseTransform` Base class for data transforms. :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: name :type: str :value: 'affine' .. py:attribute:: requires_prior_bounds :type: bool :value: False .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) .. py:method:: inverse(y) .. py:method:: config_dict() Return the configuration of the transform as a dictionary. .. py:class:: FlowPreconditioningTransform(parameters, flow_backend = 'zuko', prior_bounds = None, bounded_to_unbounded = True, bounded_transform = 'probit', affine_transform = True, periodic_parameters = None, device=None, xp=None, eps=1e-06, dtype=None, flow_matching = False, flow_kwargs = None, fit_kwargs = None) Bases: :py:obj:`BaseTransform` Base class for data transforms. :param xp: The array API namespace to use (e.g., numpy, torch). :type xp: :py:class:`Callable` :param dtype: The data type to use for the transform. If not provided, defaults to the default dtype of the array API namespace if available. :type dtype: :py:class:`Any`, *optional* .. py:attribute:: parameters .. py:attribute:: periodic_parameters :value: [] .. py:attribute:: prior_bounds :value: None .. py:attribute:: bounded_to_unbounded :value: True .. py:attribute:: bounded_transform :value: 'probit' .. py:attribute:: affine_transform :value: True .. py:attribute:: eps :value: 1e-06 .. py:attribute:: device :value: 'cpu' .. py:attribute:: flow_backend :value: 'zuko' .. py:attribute:: flow_matching :value: False .. py:attribute:: flow_kwargs .. py:attribute:: fit_kwargs .. py:attribute:: flow :value: None .. py:method:: fit(x) Fit the transform to the data. .. py:method:: forward(x) .. py:method:: inverse(y) .. py:method:: new_instance(xp=None, dtype = None) .. py:method:: save(h5_file, path='data_transform') :abstractmethod: Save config + any fitted state into an HDF5 file.