scipy：如何获取每一行的所有非零值索引？

Question

I can't find more info about scipy.sparse indexing except SciPy v0.11 Reference Guide, which says that

The lil_matrix class supports basic slicing and fancy indexing with a similar syntax to NumPy arrays.

. I have read numpy document about index, but I didn't understand it clearly, for example,

 Asp = sparse.lil_matrix((3,3)) Asp.setdiag(zeros(3)) Asp[0, 1:3] = 10 print Asp.todense() 

1. why the output is

 [[ 0. 10. 10.] \n [ 0. 0. 0.] \n [ 0. 0. 0.]]  

what does [0,1:3] meaning? if I use

 Asp[0, 1:2,3] = 10 

there's a error:

 IndexError: invalid index  

, I don't know the reason.

2.what's the fastest way to get all non-zero values for each row?

Answer 1

For your second question, use the nonzero() method. I had to dig through the source to find it, since I couldn't find it in any of the reference documentation.

def nonzero(self):
    """nonzero indices

    Returns a tuple of arrays (row,col) containing the indices
    of the non-zero elements of the matrix.

    Examples
    --------
    >>> from scipy.sparse import csr_matrix
    >>> A = csr_matrix([[1,2,0],[0,0,3],[4,0,5]])
    >>> A.nonzero()
    (array([0, 0, 1, 2, 2]), array([0, 1, 2, 0, 2]))

    """

Answer 2

what does [0,1:3] mean?

That means: row 0, elements 1 to 3 (exclusive). Since Numpy and Scipy use zero-based indices, row 0 is the first row and 1:3 denotes the first and second column.

Asp[0, 1:2,3] is invalid because you've got three indices, 0 , 1:2 and 3 . Matrices only have two axes.

This is all standard Numpy stuff; read any good tutorial on that package.