On Mon, 2 Oct 2006 apfelmus@quantentunnel.de wrote:
*d) reconsider your definition of f, separate concerns * The fact that the computation of ck depends on the iteration count j makes me suspicious. If you are using j for convergence tests etc. only, then it's not good. The most elegant way is to separate concerns: first generate an infinite list of approximations
f :: Double -> Double -> Double f c0 ck = {c_{k+1}}
cs = iterate (f c0)
and then look for convergence
epsilon = 1e-12 takeUntilConvergence [] = [] takeUntilConvergence [x] = [x] takeUntilConvergence (x:x2:xs) = if abs (x - x2) <= epsilon then [x] else x:takeUntilConvergence (x2:xs)
Once more: http://www.cs.chalmers.se/~rjmh/Papers/whyfp.html