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

[Thomas D. Dean]

Am I missing something?

sinhx = __ieee754_sinh (__real__ x);
coshx = __ieee754_cosh (__real__ x);
den = (sinhx*sinhx + cosix*cosix);

In maple,
solve(den = 0, x):
evalf(%,20)

den == 0 in ONE case:
.78539816339744830962*I

which is not real and we are using the real functions from libm.

plotting den shows a min at zero,

diff(den(x), x);
  4 sinh(x) cosh(x)
solve(%=0,x)
  0, pi/2*I

indicating a min of den at x=0, the only real value

limit(den(x), x = 0)
  1

So, using real valued functions, den cannot be zero, I believe.