Adding a numpy array to a scipy.sparse.dok_matrix

I have a scipy.sparse.dok_matrix (dimensions m x n), wanting to add a flat numpy-array with length m.

开发者_JAVA百科

for col in xrange(n):
    dense_array = ...
    dok_matrix[:,col] = dense_array

However, this code raises an Exception in dok_matrix.__setitem__ when it tries to delete a non existing key (del self[(i,j)]).

So, for now I am doing this the unelegant way:

for col in xrange(n):
    dense_array = ...
    for row in dense_array.nonzero():
        dok_matrix[row, col] = dense_array[row]

This feels very ineffecient. So, what is the most efficient way of doing this?

Thanks!

---

I'm surprised that your unelegant way doesn't have the same problems as the slice way. This looks like a bug to me upon looking at the Scipy code. When you try to set a certain row and column in a dok_matrix to zero when it is already zero, there is be an error because it tries to delete the value at that row and column without checking if it exists.

In answer to your question, what you are doing in your inelegant way is exactly what the __setitem__ method does currently with your elegant method (after a couple of isinstance checks and what not). If you want to use the elegant way, you can fix the bug I mentioned in your own Scipy package by opening up dok.py in Lib/site-packages/scipy/sparse/ and changing line 222 from

if value==0:

to

if value==0 and self.has_key((i,j)):

Then you can use the elegant way and it should work just fine. I went to submit a bug fix, but this it is already fixed for the next version and this is the way that it was fixed.

---

I think that this bug has been fixed in Scipy 0.8.0