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

Re: More on trits & tetrits... (long)



P1788

On 22 Apr 2010, at 22:09, Michel Hack wrote:
> Ian McIntosh wrote:
>> It should be possible to always compute correct results without using
>> decorations.  This situation is the sort of issue that led me to conclude
>> that 754 should have distinguished, and ideally 1788 should distinguish,
>> between Infinity versus Overflow.
> 
> Would be nice, I agree.  ...  The only sensible
> decoration for a float is in fact a decorated interval!

I agree. On the theory side, Siegfried Rump spent some time (was it Jan 2008 or Jan 2009?) trying to devise a self-consistent arithmetic with notions of exact/inexact infinity & zero VERY similar to what's being discussed now. After a while he went silent on this; if he couldn't manage it, probably no such system exists.

On the practical side, we are trying to make a standard that can be implemented efficiently. We have to take 754 arithmetic as it is. 

It may be useful to form a P1788 subgroup to provide input to the next 754 revision, but consensus may not be easy. I had some brilliant tweaks to make 754 work better with cset arithmetic, but found that other people's proposed tweaks, for equally valid purposes, were incompatible with mine.

Regards

John