Re: [Numpy-discussion] How do I make a diagonal matrix?

[Alan G I. <ai...@am...>]

On Fri, 23 Jun 2006, Sven Schreiber apparently wrote:=20
>>>> help(n.diag)=20
> Help on function diag in module numpy.lib.twodim_base:=20
> diag(v, k=3D0)=20
>     returns the k-th diagonal if v is a array or returns a array=20
>     with v as the k-th diagonal if v is a vector.=20

That is pretty damn obscure.
Apparently Travis's new doc string did not survive?
The Numpy book says:
diag (v, k=3D0)
    Return the kth diagonal if v is a 2-d array, or returns=20
    an array with v as the kth diagonal if v is a 1-d array.

That is better but not great.
I think what is wanted is:
diag (v, k=3D0)
    If v is a 2-d array:
    return a copy of the kth diagonal of v (as a 1-d array).
    If v is a 1-d array:
    return a 2-d array with a copy of v as the kth diagonal=20
    (and zeros elsewhere).

fwiw,
Alan Isaac

PS As a response to the question, it might be worth noting=20
the following.
>>> y=3DN.zeros((5,5))
>>> values=3DN.arange(1,6)
>>> indices=3Dslice(0,25,6)
>>> y.flat[indices]=3Dvalues
>>> y
array([[1, 0, 0, 0, 0],
       [0, 2, 0, 0, 0],
       [0, 0, 3, 0, 0],
       [0, 0, 0, 4, 0],
       [0, 0, 0, 0, 5]])

Generalizing we end up with the following (from pyGAUSS):

def diagrv(x,v,copy=3DTrue):
=09if copy: x =3D numpy.matrix( x, copy=3DTrue )
=09else: x =3D numpy.matrix( x, copy=3DFalse )
=09stride =3D 1 + x.shape[1]
=09x.flat[ slice(0,x.size,stride) ] =3D v
        return x