While we are talking about numerical precision, here is a simple example which illustrates some pitfalls of floating-point arithmetic. There are no transcendental functions here, just - and % .
Let f(x,y)=108-(815-1500%y)%x and let g(n+1)=g(n,n-1), with g(0)=4, g(1)=4.25. Then g(n)->L as n goes to infinity. The question is, what is L? Or even what is g(80)? [Hint: neither of these is 100.] Best wishes, John ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
