[Cfp-interest 3090] Re: csinh(x + i y) - G.6.3.5 - 5th bullet point of special cases
Damian McGuckin
damianm at esi.com.au
Sun Apr 7 20:07:31 PDT 2024
On Sun, 7 Apr 2024, Jim Thomas wrote:
> Conjugate, odd, and even were used where it seemed the extra bullets
> required to do otherwise would expand the specification unduly. Using
> them was not an absolute principle. For example, a better grouping of
> the bullets might have seemed more important. There might be
> inconsistencies here that would be important to remove, but we need to
> ask whether a considered change would lead to an overall significantly
> clearer specification?
As an example
csinh(x + i INF) returns NaN + i NaN for positive finite x
And because it is odd
csinh(-x + i INF) returns NaN + i NaN
But this could be written as
csinh(x + i INF) returns NaN + i NaN for finite non-zero x
Interesting, this is exactly what happens in ctanh
ctanh(x + i INF) returns NaN + i NaN for finite non-zero x
It could be written the same was as csinh (but it is not).
I chose to write 'csinh' the same way as ctanh() for consistency.
That also is the same as ccosh() (but then ccosh() is even).
Rather than looking at the various functions, I have instead taken a view
to look at each special case argument across all the functions and made
them as consistent as I can.
- Damian
More information about the Cfp-interest
mailing list