NB. I stumbled across this trying to help my daughter with her NB. homework of finding a zero for the following function. I attempted to NB. cheat by graphing it but was puzzled by what I got until I realized NB. that, applied to negatives, this function returns complex numbers and NB. the plot package treats these as 2-D point pairs. NB. I actually solved it for one root by inspection.
ff=: 13 : '((3*y.^%2)-(5*y.^%4))-2' ff2=: 13 : '((3*%:y.)-5*%:%:y.)-2' NB. These should be the same but there's a tiny precision difference (ff-:ff2)i:16j1000 NB. Not the same over this domain 0 (ff,ff2) _16 NB. Negative arguments give complex results _9.0710678j4.9289322 _9.0710678j4.9289322 >./"1 | 9 11 o./ (ff-ff2)i:16j1000 NB. So compare real and imaginary parts 1.7763568e_15 2.6645353e_15 NB. separately->tiny differences... plot (ff-ff2) i: 16j1000 NB. Gives an interesting pattern. -- Devon McCormick ^me^ at acm. org is my preferred e-mail ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
