Re: complex arithmetic in lcc-Win32

John Smith wrote:
jacob navia wrote:

John Smith wrote:

jacob navia wrote:

Wht are your experiences with the complex number arithmetic?

jacob




With a little more testing I found that your hyperbolic arctan function returns nan nan for the inputs 1.0+0.0i and 0.0+1.0i. The correct return values for these inputs are .7854+0.0i and 0.0+.7854i.

John Smith



No.

gcc gives 0 and 0, Ibm-C gives INF + 0.

The C standard specifies:
catanh(+1 + i0) returns +Inf + i0 and raises the divide-by-zero
floating-point exception.
Annex G, G.6.2.3: The catanh function, page 473

There is obviously an overflow in there.


You're right. My error. Thanks for making me aware of it. BTW, do you only reply when you are defending yourself from criticism?

JS

???

This is a no win for me. If I reply, I confirm your thesis.

If I don't I confirm your thesis implicitely...

:-)
.