[Cfp-interest 3270] Re: Specification of complex operators

[Damian McGuckin]

Hi Paul,

On Thu, 12 Sep 2024, Paul Zimmermann wrote:

>> Shouldn't these yield a number whose real and imaginary components are
>> both an infinity?
>
> yes this is what we get:
>
> libm gives (-inf,-nan)
> mpc gives (-inf,-inf)
>
> thus we consider the error is infinite in this case.
> A complex multiply should never yield NaN (if no NaN in the inputs).

I whipped up a quick example in Chapel. I use a Chapel compiler which uses 
the CLANG/LLVM backend:

 	// p, q and t are complex

 	proc main64 // done in 64-bit floating point
 	{
 		const p = -0x1.80a55ep+0 + 0x1.1bfep+80i;
 		const q = 0x1.ap+64 - 0x1.54c296p+127i;
 		const t = p * q;

 		writeln(p);
 		writeln(q);
 		writeln(t);
 	}

 	proc main32 // done in 32-bit floating point
 	{
 		const p = -0x1.80a55ep+0:real(32) + 0x1.1bfep+80:imag(32);
 		const q = 0x1.ap+64:real(32) - 0x1.54c296p+127:imag(32);
 		const t = p * q;

 		writeln(p, " ", p.type:string);
 		writeln(q, " ", q.type:string);
 		writeln(t, " ", t.type:string);
 	}

 	proc main
 	{
 		main64;
 		main32;
 	}

The value of t (coming from main32 using 32-bit floating point) is

 	inf + inf i

which is Chapel-speak for

 	INFINITY + INFINITY * I

The Chapel compiler can be built with a GCC backend. I will ask somebody 
who has done this what result they see with Chapel in that case.

Thanks - Damian