A more interesting variation on fibonacci might be illustrated by this
version

   fibphi=: 3 : '(%:5)%~(p^y)-(-.p)^y [ p=.-:>:%:5'

which uses the golden ratio in a slightly different fashion than "f3" on the
J wiki fibonacci sequences page.

This version is interesting in spite of the fact that it quickly gives the
same answers as the others in this thread because it is not restricted to a
domain or range of positive real integers.

Try

   load 'plot'
   plot fibphi i:5j99

and

   'surface' plot fibphi j./~i:3j99

to see why this might be a tad more interesting than the others.

BTW, I noticed that, contrary to what it says on the fibonacci sequences
page, "f3" is only good up to about n=70 (as is fibphi).  One interesting
feature of the recursive memoized version (mentioned earlier in this thread)

   fibM0=: 3 : ' if. 1>:y do. y else. (fib y-1)+fib y-2 end.'M.

is that it takes extended arguments we can use to check the other versions'
accuracy, e.g.

   (0j0&":)|:(f3,fibphi,fibM0)&>70x 71x
190392490709135 308061521170131
190392490709135 308061521170131
190392490709135 308061521170129

I'm assuming the last one is correct.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to