f(x) = log(log(x))
Try with n = 10000, 20000, 40000, ...
output / notes
Increasing "n" improves the precision, but if "n" is too large the precision gets worse
(observing the outputs you can see when they no longer tend to converge)