Re: Same .m file: different results with different versions of Octave

[Judd Storrs]

On Wed, Apr 21, 2010 at 8:43 AM, Thomas D. Dean <address@hidden> wrote:
>        t = sinh(arg)/cosh(arg);

The problem here is that this represents a complex division.

(a + bi)/(c + di) = [ (ac+bd) + (bc-ad)i ]/(c^2 + d^2)

The problems are the ac+bd and bc-ad parts. If ac ~ -bd or bc ~ ad
then the additions and subtractions will lose many digits of
precision. The idea is that if you work out the real and imaginary
components mathematically, you can formulate an expression that is
equivalent but will have nearly full (if not full) precision in those
cases.


--judd