Large negative real number arguments return NaN for expintx
weymouth opened this issue · comments
expintx(complex(-x)) returns NaN for x more than around 700. It's almost like this is being handled by -exp(-x)*expinti(x) and we're getting an overflow error.
It seems to be about -709.782712894 at which point the result goes from roughly -0.00141087244 + 1e-308*im to -Inf + NaN*im.