Re: [Phys-L] numerical methods

[David Bowman <David_Bowman@georgetowncollege.edu>]

Regarding:

> > If I was trying really hard to be miserly with the clock cycles
> > I might try something else that didn't require the evaluation
> > of 2 different transcendental functions.
>
> I disagree, because exp() and sinh() are both transcendental
> functions. One exp() and one sinh() is no worse than two exp()s.

True, but I was kind of hoping that, if needed, one could dispense with at least one or, preferably, both of the trancendental evaluations regardless of what they may be named--presumably using some sort of specialized DIY evaluations.

> ...
> BTW FWIW I prefer the name sinhc(x) = sinh(x)/x  which is
> perhaps not perfect, but it is memorable, in analogy to the
> established name sinc(x) = sin(x)/x.

I like it.  It's a lot shorter than compsinc.  I didn't know there already was a good/better name for it.

Dave Bowman