>python -c "import numpy as np; from scipy import stats; stats.nct._sf(np.nan, 125.531223551, 2.82324307702)"
never comes back
I was tracking an rootfinding that never came back that had bad starting values.
scipy 0.9.0 and scipy 0.11.0
Not a stats bug; the non-underscored sf returns nan as expected.
It is a special bug that nan input to nctdtr hangs. Doesn't seem to depend on the values of argument 2 and 3, this hangs as well:
special.nctdtr(np.nan, 2, 1)
my code comment to justify using the private method
# use private methods, generic methods return nan with negative d
So is the nan from negative d a bug that needs fixing then?
As far as I remember it's already fixed in scipy master.