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

Re: Motion to finalise interval literals




Michael,

Thanks for your query. I cannot find "degenerate mathematical interval" anywhere in the Interval Programming Reference manual. Please point me to the place you find unclear.

It might be helpful to you to see John's earlier off line query to me and my answers. See the attached. Nevertheless, I invite you, either publicly, of off line, to send me any queries you have about the Sun documentation. In the Index, the words "degenerate interval" appear on pages 42 and 130, and "degenerate interval representation" appears on page 17.

John,

In your answer to Michael, you mention "Fatemaan meaning". Please elaborate.

If by that you mean entering an undecorated decimal number, such as 2.34500, or even the integer 23 (as can be seen on page 17 in of Sun's Interval Reference) in response to a program requesting an interval input value, then absolutely, I intend for neither to be internally represented either as a degenerate (zero width) interval, or as a 1-ulp width interval.

I believe it is important for interval I/O to be transparent so users realize that in practical applications degenerate infinitely precise intervals should be very rare events.

Cheers,

Bill


Michael,
On 6/5/13 2:10 PM, Michel Hack wrote:
I'm still curious however what Sun's documentation means by "degenerate mathematical interval" -- assuming that they have no notion of Levels. Michel. ---Sent: 2013-06-05 21:22:48 UTC

--- Begin Message ---

John,

I neglected to add at the end of the first paragraph that

More accuracy can be specified with trailing zero digits.  That is,
".123000E-10", is interpreted as the infinitely precise interval 
"[122999, 124001]/10000000000000".

-------- Original Message --------
Subject: Re: Motion 44.01 PLEASE VOTE - I vote No
Date: Tue, 04 Jun 2013 06:59:49 -0700
From: G. William (Bill) Walster <bill@xxxxxxxxxxx>
Reply-To: bill@xxxxxxxxxxx
To: John Pryce <j.d.pryce@xxxxxxxxxx>


Thanks for your kind words, John.

What I wanted to do is to eliminate the default practice that now 
exists, in which users are misled by interpreting their string inputs as 
infinitely precise.  To do this, I let the appearance of a decimal point 
in a numerical string be the trigger for precision to be determined by 
the last decimal digit in the string.  The lack of a decimal point 
requires the ability to scale integers so that 123E-10 is interpreted as 
the infinitely precise rational number "123/10000000000".  The converse, 
".123E-10", is interpreted as the infinitely precise interval 
"[122,124]/10000000000".

Conversion from these strings to internal floating-point intervals is 
done with absolutely rigorous directed rounding, but only when 
required.  The same is true for conversion from internal floating-point 
intervals to strings for output.  Rigorous I/O was one of the most 
difficult parts of an interval compiler to implement.  I had done one 
before Sun's in CDC's Compass assembly language in the '70s on the 
University of Minnesota's MNF Fortran compiler. 
<http://www.comp.tc-ieee.org/archive/FORTRAN.html>

You, or somebody else will have to read the Sun Interval User Reference 
to compare it to the Vienna proposal.

Regards,

Bill


On 6/1/13 9:34 AM, John Pryce wrote:
> <I first sent this to George by mistake, and to the group>
> Bill
>
> On 31 May 2013, at 03:14, Corliss, George wrote:
> On  May 30, 2013 6:29:08 PM CDT, Bill Walster wrote:
>> I am unable (perhaps it is me) to determine if the scheme implemented in Sun's implementation of string to interval and interval to string conversion will be standard conforming or not.  The Sun Fortran 95 implementation explicitly deals with strings as infinitely precise decimal numbers versus strings in which interval width is determined by the last decimal digit in a string.
>>
>> If the Sun Fortran implementation is standard conforming, or if the draft can be updated to allow the sun string conversion implementation to be standard conforming, I will change my vote to Yes.
> I have much respect for the Sun Fortran implementation and we could probably learn from it about this issue. Could you just remind us of the details of its scheme that you mention above? And also how much it overlaps with the scheme in the Vienna proposal, which has a similar feature of "infinitely precise decimal numbers versus ... interval width determined by the last decimal digit"?
>
> John Pryce




--- End Message ---