Save multiple arrays to the same npz file

[Pedro Augusto Sampaio Messias Ribeiro]

It would be nice to save multiple arrays to a single npz file, just like in numpy.

[James Webber]

Is this actually possible with scipy.sparse matrices? In the sparse context, save_npz already needs to save multiple arrays for each sparse matrix npz

e.g. saving a scipy.sparse.csr_matrix saves to the following arrays:

['indices', 'indptr', 'format', 'shape', 'data']

I could imagine doing that multiple times to save more than one array (indices_0 indptr_0 ... indices_1 ...) but I don't think scipy does that at the moment.

[Pedro Augusto Sampaio Messias Ribeiro]

Sorry, i meant like numpy.savez or numpy.savez_compressed

[Hameer Abbasi]

While an NPZ file can hold many NumPy arrays, one sparse array contains multiple NumPy arrays, so it can only hold one sparse array unless tricks like ones mentioned in #549 (comment) are used.

[James Webber]

Sorry, i meant like numpy.savez or numpy.savez_compressed

NB: if you try to use these methods on a scipy.sparse matrix it will save it as a pickle, which tends to take up more space and be slower to read. That's why scipy.sparse.save_npz is a separate function.

[Hameer Abbasi]

Closing as this is complicated and out of scope.