RuntimeWarning: invalid value encountered in divide

Solution 1:

I think your code is trying to "divide by zero" or "divide by NaN". If you are aware of that and don't want it to bother you, then you can try:

import numpy as np
np.seterr(divide='ignore', invalid='ignore')

For more details see:

  • http://docs.scipy.org/doc/numpy/reference/generated/numpy.seterr.html

Solution 2:

Python indexing starts at 0 (rather than 1), so your assignment "r[1,:] = r0" defines the second (i.e. index 1) element of r and leaves the first (index 0) element as a pair of zeros. The first value of i in your for loop is 0, so rr gets the square root of the dot product of the first entry in r with itself (which is 0), and the division by rr in the subsequent line throws the error.

Solution 3:

To prevent division by zero you could pre-initialize the output 'out' where the div0 error happens, eg np.where does not cut it since the complete line is evaluated regardless of condition.

example with pre-initialization:

a = np.arange(10).reshape(2,5)
a[1,3] = 0
print(a)    #[[0 1 2 3 4], [5 6 7 0 9]]
a[0]/a[1]   # errors at 3/0
out = np.ones( (5) )  #preinit
np.divide(a[0],a[1], out=out, where=a[1]!=0) #only divide nonzeros else 1

Solution 4:

You are dividing by rr which may be 0.0. Check if rr is zero and do something reasonable other than using it in the denominator.