numpy ufunc redurs的标识值

Question

I have a numpy array which I want to find the maximum element of, so I call:

x = foo.max()

The problem is that foo is sometimes an empty array, and the max function (understandably) throws:

ValueError: zero-size array to reduction operation maximum which has no identity

This leads to my question:

Is there a way to supply the identity value for the reduction operation? (Ideally, one that will work for arbitrary numpy ufuncs with reduce methods.) Or am I stuck doing:

if foo.size > 0:
  x = foo.max()
else:
  x = 0  # or whatever identity value I choose

Answer 1

Surely the "pythonic" way to go about this is:

def safe_max(ar,default=np.nan):
    try:
        return np.max(ar)
    except ValueError:
        return default

x = safe_max(foo,0)

Considering that the maximum of an empty sequence is not defined, this would seem to also be the most logical approach.