[SciPy-Dev] scipy.sparse spdiags/dia_matrix additional calling mode

Denis Laxalde denis.laxalde@mcgill...
Thu Oct 6 09:41:19 CDT 2011


On Thu, 06 Oct 2011 15:26:14 +0200,
Pauli Virtanen wrote:
> Adding a new function for new semantics, rather than using a keyword
> argument, feels to me like a clearer design choice here.
> 
> I think adding a new function called `diags` could be justified here.
> Numpy already has a function called `diag`, and with the new semantics
> one would have
> 
>     diags([a, b, c], [i, j, k]) == diag(a, i) + diag(b, j) + diag(c, k)
> 
> which hangs together nicely.

However this trailing 's' is not so consistent with other sparse matrix
constructors (e.g. eye, kron, identity) which follow NumPy names.

-- 
Denis Laxalde


More information about the SciPy-Dev mailing list