What is the fastest way to slice a scipy.sparse matrix?

Question

I normally use ``` matrix[:, i:] ``` It seems not work as fast as I expected.

Accepted Answer

To obtain a sparse matrix as output the fastest way to do row slicing is to have a `csr` type, and for columns slicing `csc`, as detailed here. In both cases you just have to do what you are currently doing:

```
matrix[l1:l2, c1:c2]
```

If you want a `ndarray` as output it might be faster to perform the slicing directly in the `ndarray` object, which you can obtain from the sparse matrix using the `.A` attribute or the `.toarray()` method:

```
matrix.A[l1:l2, c1:c2] 
```

or:

```
matrix.toarray()[l1:l2, c1:c2]
```

As mentioned in the comment below, converting the sparse array to a dense array might lead to memory errors if the array is big enough.

What is the fastest way to slice a scipy.sparse matrix?

Solution

Problem