abs & asinh



>>>>> "Henry" == Henry Baker <hbaker1 at pipeline.com> writes:

    Henry> Raymond: Are you talking below about something other than
    Henry> minimal changes?

The current fix is very minimal and just brings those functions in
line with what maxima already knows about sinh.  Nothing really fancy.

    Henry> If there is a project that is doing something more
    Henry> fundamental about these issues, please let me know more
    Henry> about it.  I'm looking into various things, including:

    Henry> 1.  Propagation of type/range (see Common Lisp
    Henry>     declarations) info across argument/function results,
    Henry>     which would take care of most of these issues below.
    Henry> 2.  Partial function info, including functions whose domain
    Henry>     is not the entire complex plane or the entire real
    Henry>     line.

Ah, these would be very cool to have.  Long ago, I did something
like that for CMUCL's compiler so that it could derive the types for
basic arithmetic operators and some of the special functions.

    Henry> 3.  if-then-else simplification, which allows the extension
    Henry>     of the less-than-total domains of #2 -- e.g., the
    Henry>     if-then-else nature of atan2.

Hmm.  Isn't there some contrib package that can do some kinds of
if-then-else simplification?

Ray