python - Scipy sparse csr matrix returns nan on 0.0/1.0 -

i spotted unexpected behavior in scipy.sparse.csr_matrix, seems bug me. can confirm not normal? not expert in sparse structures may misunderstanding proper usage.

>>> import scipy.sparse >>> a=scipy.sparse.csr_matrix((1,1)) >>> b=scipy.sparse.csr_matrix((1,1)) >>> b[0,0]=1 /home/marco/anaconda3/envs/py35/lib/python3.5/site-packages/scipy/sparse/compressed.py:730: sparseefficiencywarning: changing sparsity structure of csr_matrix expensive. lil_matrix more efficient.   sparseefficiencywarning) >>> a/b matrix([[ nan]]) 

on other hand, numpy handles this:

>>> import numpy np >>> a=np.zeros((1,1)) >>> b=np.ones((1,1)) >>> a/b array([[ 0.]]) 

thanks

for sparse matrix/sparse matrix,

scipy/sparse/compressed.py

    if np.issubdtype(r.dtype, np.inexact):         # eldiv leaves entries outside combined sparsity         # pattern empty, must filled manually.         # nan, matrix full.         out = np.empty(self.shape, dtype=self.dtype)         out.fill(np.nan)         r = r.tocoo()         out[r.row, r.col] = r.data         out = np.matrix(out) 

the action explained in section.

try larger matrices

in [69]: a=sparse.csr_matrix([[1.,0],[0,1]]) in [70]: b=sparse.csr_matrix([[1.,1],[0,1]]) in [72]: (a/b) out[72]:  matrix([[  1.,  nan],         [ nan,   1.]]) 

so ever a has 0s (no sparse values), division nan. it's returning dense matrix, , filling in nan.

without code, sparse element element division produces sparse matrix 'empty' off diagonal slots.

in [73]: a._binopt(b,'_eldiv_') out[73]:  <2x2 sparse matrix of type '<class 'numpy.float64'>'     2 stored elements in compressed sparse row format> in [74]: a._binopt(b,'_eldiv_').a out[74]:  array([[ 1.,  0.],        [ 0.,  1.]]) 

the inverse might instructive

in [76]: b/a out[76]:  matrix([[  1.,  inf],         [ nan,   1.]]) in [77]: b._binopt(a,'_eldiv_').a out[77]:  array([[  1.,  inf],        [  0.,   1.]]) 

it looks combined sparsity pattern determined numerator. in further test looks after eliminate_zeros.

in [138]: a1=sparse.csr_matrix(np.ones((2,2))) in [139]: a1 out[139]:  <2x2 sparse matrix of type '<class 'numpy.float64'>'     4 stored elements in compressed sparse row format> in [140]: a1[0,1]=0 in [141]: a1 out[141]:  <2x2 sparse matrix of type '<class 'numpy.float64'>'     4 stored elements in compressed sparse row format> in [142]: a1/b out[142]:  matrix([[  1.,  nan],         [ inf,   1.]])