Convolution

diffqc.nn.Convolution(op, kernel_func: Callable[[Array, Array], Array], kernel_shape: Tuple[int], slide: Tuple[int], padding: Tuple[int]) → Callable[[Array, Array], Array]

Create Convolution Function

Parameters

• op – dense or sparse

• kernel_func (Callable) – kernel function of f(x, w)

• kernel_shape (tuple of ints) – kernel shape. len(kernel_shape) == 2.

• slide (tuple of ints) – slides. len(slide) == 2.

• padding (tuple of ints) – padding. len(padding) == 2

Returns

convoluted funcion of F(x, w)

Return type

Callable