tnmf.backends._Backend

A module that defines a common interface for all backends.

Module Contents

Classes

Backend

The base class for all backends.

Attributes

sliceNone

tnmf.backends._Backend.sliceNone

class tnmf.backends._Backend.Backend(reconstruction_mode: str = 'valid')

Bases: abc.ABC

The base class for all backends.

It exists to define a generic functional interface that is shared by all backends through inheritance, which allows a clear separation between algorithmic steps and their actual numerical implementation.

Some common functionality between all backends is already defined inside this class but can be overridden if necessary.

initialize(self, V: numpy.ndarray, atom_shape: Tuple[int, Ellipsis], n_atoms: int, W: Optional[numpy.ndarray] = None, axes_W_normalization: Optional[Union[int, Tuple[int, Ellipsis]]] = None) → Tuple[numpy.ndarray, numpy.ndarray]

abstract static to_ndarray(arr) → numpy.ndarray

static normalize(arr: numpy.ndarray, axis: Optional[Union[int, Tuple[int, Ellipsis]]] = None)

abstract static convolve_multi_1d(arr: numpy.ndarray, kernels: Tuple[numpy.ndarray, Ellipsis], axes: Tuple[int, Ellipsis]) → numpy.ndarray

abstract reconstruction_gradient_W(self, V: numpy.ndarray, W: numpy.ndarray, H: numpy.ndarray, s: slice = sliceNone) → Tuple[numpy.ndarray, numpy.ndarray]

abstract reconstruction_gradient_H(self, V: numpy.ndarray, W: numpy.ndarray, H: numpy.ndarray, s: slice = sliceNone) → Tuple[numpy.ndarray, numpy.ndarray]

abstract reconstruct(self, W: numpy.ndarray, H: numpy.ndarray) → numpy.ndarray

partial_reconstruct(self, W: numpy.ndarray, H: numpy.ndarray, i_atom: int) → numpy.ndarray

reconstruction_energy(self, V: numpy.ndarray, W: numpy.ndarray, H: numpy.ndarray) → float