Would a1 = cat(3, a) do what you want?
Den mandag den 11. januar 2016 kl. 09.03.57 UTC+1 skrev Doug Y'barbo: > > apologies if this topic has been asked and answered; lack of distinctive > terms gave me large result sets from my queries of this archive > > it's common for me to want to add an axis to a Julia array, usually as a > predicate to broadcasting, or more precisely to modify two arrays so that > their trailing dimensions are equal and therefore invoke the Julia > compiler's own broadcasting. > > to be clear, i'm not talking about a very specific transformation--one > that doesn't change the Array's length and that can be reversed by calling > *squeeze* > > this is how i do it just now: > > > a = rand(4, 3) > > a1 = reshape(a, 4, 1, 3) > > what i would like to do is something like NumPy (which of course also has > a reshape method, but it likely wouldn't be used for this purpose) > > a1 = [:,NP.newaxis,:] > > to me this is much better because a1 is a view, rather than a copy, and i > never had to pass in the values for the other dimensions, which is > inconvenient and error-prone. (The two expressions for a1 are both valid > NumPy and are equivalent). > > i would bet that Julia has a built-in similar to NumPy's to do what i > want, i just can't find it > > --doug > > > >
