By the way is below known/interesting ? sqrt(e-Pi/2) + sqrt(e+Pi/2) ~= 3.142216467362486483529141519 ~= Pi gp > sqrt(exp(1)-Pi/2) + sqrt(exp(1)+Pi/2) = 3.142216467362486483529141519 At least it gives better approximation of Pi than sqrt(2) + sqrt(3) ~= 3.146264369941972342329135... Alexander R. Povolotsky