array_split.array_split

array_split.array_split(ary, indices_or_sections=None, axis=None, tile_shape=None, max_tile_bytes=None, max_tile_shape=None, sub_tile_shape=None, halo=None)[source]

Splits the specified array ary into sub-arrays, returns list of numpy.ndarray.

Parameters:
  • ary (numpy.ndarray) – Array which is split into sub-arrays.
  • indices_or_sections (None, int or sequence of int) – If an integer, indicates the number of elements in the calculated split array. If a sequence, indicates the indicies (per axis) at which the splits occur. See Splitting by number of tiles examples.
  • axis (None, int or sequence of int) – If an integer, indicates the axis which is to be split. If a sequence integers, indicates the number of slices per axis, i.e. if axis = [3, 5] then axis 0 is cut into 3 slices and axis 1 is cut into 5 slices for a total of 15 (3*5) rectangular slices in the returned (3, 5) shaped split. See Splitting by number of tiles examples and Splitting by per-axis cut indices examples.
  • tile_shape (None or sequence of int) – When not None, specifies explicit shape for tiles. Should be same length as array_shape. See Splitting by tile shape examples.
  • max_tile_bytes (None or int) – The maximum number of bytes for calculated tile_shape. See Splitting by maximum bytes per tile examples.
  • max_tile_shape (None or sequence of int) – Per axis maximum shapes for the calculated tile_shape. Only relevant when max_tile_bytes is specified. Should be same length as array_shape. See Splitting by maximum bytes per tile examples.
  • sub_tile_shape (None or sequence of int) – When not None, the calculated tile_shape will be an even multiple of this sub-tile shape. Only relevant when max_tile_bytes is specified. Should be same length as array_shape. See Splitting by maximum bytes per tile examples.
  • halo (None, int, sequence of int, or (len(ary.shape), 2) shaped numpy.ndarray) – How tiles are extended per axis in -ve and +ve directions with halo elements. See The halo parameter examples.
Return type:

list

Returns:

List of numpy.ndarray elements, where each element is a slice from ary (potentially an empty slice).