On 4/25/07, Henry Rich <[EMAIL PROTECTED]> wrote:
Are the computations carefully arranged so that numbers
that are known to approach close to 0 are allowed as
operands only to a restricted set of functions that are
known not to push them over the boundary?

I think the specific issue is not branch cuts as a whole,
but the complex domain of ^. when negative real numbered
values are used.

If other arithmetic operations retain the sign of the
complex zero, then the complex part of the result of ^.
can be pi or -pi, as appropriate.

For example, ^.^ in ^.^0j_1p1 could be replaced by an
identity.

And, I'm not sure what we have to offer as a replacement
for exponentials in the complex domain that better
handles this issue.

--
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to