pydsm.relab.shiftdim

pydsm.relab.shiftdim(x, n=None, nargout=2)

Shift dimensions a la Matlab

When n is provided, shiftdim shifts the axes of x by n. If n is positive, it shifts the axes to the left, wrapping the leading axes with non unitary length to the end. When n is negative, it shifts the axes to the right, inserting n leading axes with unitary length. When n is not provided or None, it shifts the axes to the left, reducing the number of dimensions and removing all the leading axes with unitary length.

Parameters:

x : array like

multi-dimensional array to operate upon

n : int or None, optional

amount to shift. Defaults to None, which means automatic computation

nargout : int

number of output values

Returns:

y : ndarray

the result of the axes shift operation

n : int

the actual shift

Examples

>>> from numpy.random import rand
>>> a = rand(1, 1, 3, 1, 2)
>>> b, n = shiftdim(a)
>>> b.shape
(3, 1, 2)
>>> n
2
>>> c = shiftdim(b, -n, nargout=1)
>>> np.alltrue(c == a)
True
>>> d = shiftdim(a, 3, nargout=1)
>>> d.shape
(1, 2, 1, 1, 3)

>>> b, n = shiftdim([[[1]]])
>>> b, n
(array([[[1]]]), 0)