A correction to my email just now: Curiously, this reminds me of a method I discussed in a paper I once wrote on arithmetic (eprints.utas.edu.au/2003) that would allow inexact arithmetic to be carried out on very large and very small numbers with overflow or underflow *NOT* a concern. Any use in J ?
---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
