Follow-up Comment #1, bug #51000 (group gsl): As far as I can see the Airy function oscillates with a period of 2*pi at negative arguments. If the argument is -5e+102, this is about 7*101 times Pi which would require 101 digits to represent the input value to even get a hint where that function actually is. Because the usual IEEE representation of doubles has only like 16 decimals of precision, there is no way to compute with a straight C-implementation such a thing. The only thing one could do here is to clearly flag that the value is NaN due to complete loss of digits.
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/bugs/?51000>
_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
signature.asc
Description: PGP signature
