[SciPy-user] math operations on sparse matrices

[josef.pktd at gmail.com]

On Sat, Feb 21, 2009 at 6:43 PM, Stéfan van der Walt <stefan at sun.ac.za> wrote:
> 2009/2/21  <josef.pktd at gmail.com>:
>> I have a (dok) sparse distance matrix and I would like to take the
>> exponential of the distances
>
> I guess you could just as well switch to dense matrices then, since
> exp(0) is no longer zero.
>
> If you just want to change the non-zero values, you can use
>
> x.data = np.exp(x.data)
>
> Regards
> Stéfan

That was also my first guess, however

>>> M
<50x50 sparse matrix of type '<type 'numpy.float64'>'
	with 208 stored elements in Dictionary Of Keys format>
>>> M.data
Traceback (most recent call last):
  File "<pyshell#78>", line 1, in <module>
    M.data
  File "\Programs\Python25\Lib\site-packages\scipy\sparse\base.py",
line 429, in __getattr__
AttributeError: data not found


For now this seems to work pretty fast

Mexp = M.copy()
Mexp.update(((k,exp(-v)) for k,v in M.iteritems()))

But I'm not sure I know what I'm doing.

What I'm trying to do is something like OLS with a sparse X'X matrix
(kernel rigdge regression).
The next step are:
alpha = sparse.linalg.minres(M,y)
yhat = M1.matmat(alpha[0])

>From the graphical results it seems to work, but since this is my
first try with scipy.sparse.linalg, I'm not sure what the methods to
in detail.

Josef