The `array_split.split` Module¶

Defines array splitting functions and classes.

Classes and Functions¶

`shape_factors`(n[, dim])	Returns a `numpy.ndarray` of factors `f` such that `(len(f) == dim) and (numpy.prod(f) == n)`. The returned factors are as square (cubic, etc) as possible. For example::.
`calculate_num_slices_per_axis`(...[, ...])	Returns a `numpy.ndarray` (`return_array` say) where non-positive elements of the `num_slices_per_axis` sequence have been replaced with positive integer values such that `numpy.prod(return_array) == num_slices` and.
`calculate_tile_shape_for_max_bytes`(...[, ...])	Returns a tile shape `tile_shape` such that `numpy.prod(tile_shape)*numpy.sum(array_itemsize) <= max_tile_bytes`.
`convert_halo_to_array_form`(halo, ndim)	Converts the `halo` argument to a `(ndim, 2)` shaped array.
`ShapeSplitter`(array_shape[, ...])	Implements array shape splitting.
`shape_split`(array_shape, args, *kwargs)	Splits specified `array_shape` in tiles, returns array of `slice` tuples.
`array_split`(ary[, indices_or_sections, ...])	Splits the specified array `ary` into sub-arrays, returns list of `numpy.ndarray`.

array_split.split.ARRAY_BOUNDS = <property object>¶: Indicates that tiles are always within the array bounds, resulting in tiles which have truncated halos. See The halo parameter examples.

array_split.split.NO_BOUNDS = <property object>¶: Indicates that tiles may have halos which extend beyond the array bounds. See The halo parameter examples.

`is_scalar`(obj)	Returns `True` if argument `obj` is a numeric type.
`is_sequence`(obj)	Returns `True` if argument `obj` is a sequence (e.g. a `list` or `tuple`, etc).
`is_indices`(indices_or_sections)	Test for the `indices_or_sections` argument of `ShapeSplitter.__init__()` to determine whether it is specifying total number of tiles or sequence of cut indices.
`pad_with_object`(sequence, new_length[, obj])	Returns `sequence` `list` end-padded with `obj` elements so that the length of the returned list equals `new_length`.
`pad_with_none`(sequence, new_length)	Returns `sequence` `list` end-padded with `None` elements so that the length of the returned list equals `new_length`.