matplotlib.pyplot.spy does not work correctly for sparse matrices with a large number of entries (>= 2**32). I would guess it has something to do with the value range of int32.

An example that reproduces the bug is:

import numpy as np
import scipy.sparse
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
n = 16
s = 2**n
A = scipy.sparse.coo_matrix(np.ones((s,s), dtype=np.int8))
plt.spy(A)
file = '/tmp/spy_{}.png'.format(n)
plt.savefig(file)

For n = 16, spy shows only some entries.

For n = 15 it works correctly.

I have used Matplotlib 1.5.1 installed over pip with Python 3.5.1 on Linux.