If a is a multi-dimensional array, a[..., j] gives an array of shape a.shape[:-1]. So, for example if a is a m-by-n matrix a[..., j] gives an m dimensional vector of shape (m,). What is the correct way to do the above-mentioned slicing such that the result has the same number of dimensions as a with the last dimension having size 1. For example a[<magical_slice>] for the m-by-n array should give a (column) vector of shape (m, 1).
Currently I am using
a[..., j].rehape(a.shape[:-1] + (1,))
To achieve that goal. However, that looks too complicated (and most probably inefficient) for what I would like to do.
