Do zero entries in numpy.array occupy memory?

Question

Suppose I use numpy arrays (e.g. numpy.ndarray) to store large, sparse matrices (i.e., most of the entries are 0): Do the zero entries actually occupy memory? Does numpy support sparse arrays, and if yes, which storage format is used?

score 5 · Accepted Answer · edited May 23 '17 at 12:04

5

I think that what you are really looking for is the scipy sparse module, which supports multiple formats.

SCIPY SPARSE DOCUMENTATION http://docs.scipy.org/doc/scipy/reference/sparse.html

The stackoverflow link below gives a great example of using scipy sparse matrices as well. Scipy sparse... arrays?

edited May 23 '17 at 12:04

Community

1
1

answered May 08 '12 at 16:17

aquil.abdullah

3,059
3
21
40

NPE · Answer 2 · 2012-05-08T17:04:18.193

3

Do the zero entries actually occupy memory?

With ndarray, they do.

There are, however, separate classes in SciPy that support a variety of sparse storage formats. See the documentation.

edited May 08 '12 at 17:04

answered May 08 '12 at 16:08

NPE

486,780
108
951
1,012

So sparse matrices are only in SciPy, not pure numpy? – clstaudt May 08 '12 at 16:09

Do zero entries in numpy.array occupy memory?

2 Answers2