Barton Willis wrote:
> but this is more of a mess than a simplification, I think. If it's OK
> with you all, I propose that we (we = I) eliminate the asksign stuff
> from atan2. We'll still get things like
>
It's ok with me. But I kind of like getting asked because sometimes I
forget about the cases, and answering the question is a nice quick way
to temporarily make some assumption.
> assume(x > 0); atan2(x,x) --> %pi / 4.
>
> A better atan2 function should pave the way for a better carg
> function.
>
> By the way, even with the assumption x # 0,
>
> atan2(x,x) = signum(x) * %pi / 4
>
> is a problem when atan2 is extended to nonreal arguments, I think.
>
>
How is atan2 defined for non-real arguments?
Ray