It is known that numerical computation of confluent hypergeometric function with large argument is challenging.
I found a strange behavior of scipy special function hyp1f1.
For example, scipy.sepcial.hyp1f1(-40.5,0.5,64.0) gives
-1.1979592767053608e+16 in Linux
and 65755449963894.45 in Windows.
Compared the result with the independent fortran code
written by Shanjie Zhang, Jianming Jin,
it is consistent with -1.1979592767053608e+16.
(The fortran code gives the same result in both OS.)
How can I understand this? Scipy version difference?
