Is this intended behaviour?
b = sparsevec([1], [1.0], 10)
typeof(b * b')
# Array{Float64,2} instead of SparseMatrixCSC
Is this intended behaviour?
b = sparsevec([1], [1.0], 10)
typeof(b * b')
# Array{Float64,2} instead of SparseMatrixCSC
I don’t know if you ever filed an issue for this, but it looks like this works correctly on master, now that we have RowVector
s and sparse broadcast has seen a major overhaul. Behavioral tests for this kind of thing would be welcome, it might not be covered by existing tests.
Thanks for the update. I don’t think Infiles an issue.