Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: As simple as it is now, I am still against motion 24.03...



> Subject: Re: As simple as it is now, I am still against motion 24.03...
> From: John Pryce <prycejd1@xxxxxxxxxxxxx>
> Date: Sat, 11 Jun 2011 12:03:41 +0100
> To: stds-1788 <stds-1788@xxxxxxxxxxxxxxxxx>
> 
> Michel, P1788
> 
> On 10 Jun 2011, at 15:50, Michel Hack wrote:
> > . . .
> > 
> > A standard that only addresses group (1) leaves a lot of uncertainty for
> > groups (2) and (3), and therefore makes it more difficult to achieve what
> > is needed by group (1).
> 
> I find Michel's argument very persuasive. I think P1788 should have some
> SHALLs in it directed towards group (2), and perhaps group (3).
> 
> Motion 24.03 does that, but I am uneasy with it for at least 2 reasons:
> 
> . . .
> 
> I suggest that on 754-compliant systems the operations be parameterised
> by the output interval format, like a 754 "formatOf" operation, and that
> interval format shall be of infsup type only. So one can input data of
> any FP format having the *same radix* as the output format. (Recall this
> eliminates the "double rounding" problem, while being fairly easy to
> implement.)
> 
> . . .
> 
> An advantage of (point,point) -> interval is that it allows a P1788-oriented
> resolution of "what is the result when some input is not a finite real?"
> Whereas, if we take Ulrich's route we may find ourselves struggling with
> Qs like
> 
>   "What is the result of Inf-Inf in round-away from zero mode?"
> 
> which is definitely not what P1788 was set up for. 
> 
> Michel, Vincent, other people who work at the lower level: Would such a set
> of operations be the right sort of compromise between too low-level and too
> high-level?
> 
> I aim to propose a motion on these lines, and offer it as a friendly
> amendment to 24.03. Suggestions on wording welcomed.
> 
> Regards


	John,

	I find I like this approach as a compromise.

	If you look in Clause 9.4 of 754-2008, we tried to provide
	a good set of such vector --> scalar operations in the
	form of reductions.  They include a vector sum, dot product,
	sum of squares, & sum of absolute values.  They also include
	scaled multiplication reductions like product, product of
	sums, & product of differences.  (These are used in the
	evaluation of something called Racah coefficients.  These,
	in turn, are used in quantum mechanics & string theory
	simulations.)

	Now, one problem with these as 754 operations is exactly the
	sort of thing you mentioned.  Things like inf - inf, 0*inf,
	& the like.  So the details of the definitions may not be
	exactly what you have in mind.  But the list might be a good
	place to start to find a way to define a useful set of
	vector real --> scalar interval operations.

	The other problem is that they are optional in 754.

	But I think a mandatory (under shall) set of such operations
	defined for use by the 1788 library writer might be just the
	thing.  At least *I* could support it.

	Ulrich, would such a thing be sufficient to make you happy?


				   Dan