mars.tensor.dstack#

mars.tensor.dstack(tup)[source]#

Stack tensors in sequence depth wise (along third axis).

This is equivalent to concatenation along the third axis after 2-D tensors of shape (M,N) have been reshaped to (M,N,1) and 1-D arrays of shape (N,) have been reshaped to (1,N,1). Rebuilds arrays divided by dsplit.

This function makes most sense for arrays with up to 3 dimensions. For instance, for pixel-data with a height (first axis), width (second axis), and r/g/b channels (third axis). The functions concatenate, stack and block provide more general stacking and concatenation operations.

Parameters: tup (sequence of tensors) – The tensors must have the same shape along all but the third axis. 1-D or 2-D arrays must have the same shape.
Returns: stacked – The array formed by stacking the given tensors, will be at least 3-D.
Return type: Tensor

See also

stack: Join a sequence of tensors along a new axis.
vstack: Stack along first axis.
hstack: Stack along second axis.
concatenate: Join a sequence of arrays along an existing axis.
dsplit: Split tensor along third axis.

Examples

>>> import mars.tensor as mt

>>> a = mt.array((1,2,3))
>>> b = mt.array((2,3,4))
>>> mt.dstack((a,b)).execute()
array([[[1, 2],
        [2, 3],
        [3, 4]]])

>>> a = mt.array([[1],[2],[3]])
>>> b = mt.array([[2],[3],[4]])
>>> mt.dstack((a,b)).execute()
array([[[1, 2]],
       [[2, 3]],
       [[3, 4]]])