Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> comp.databases.theory -> Re: Possreps and numeric types
On Mar 28, 5:27 am, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
> Marshall wrote:
>
> > Has anyone seen it? :-(
>
> Not yet. Is it in a 'Sent Items' folder of any kind?
Crappity crap. Nope, no sent folder.
Well, for some years now it's been true that I've needed to read news at multiple sites, and that stopped being true Monday. Maybe I should get a "real" newsreader now.
Roughly, I said my primary interest is in correctness, and that proofs of properties of code are the best tool I know of in that direction, and that code tends to manipulate a lot of numbers, and that proofs depend on algebraic properties, and the rationals have many desirable algebraic properties and floats don't. Furthermore if we declare that floats do have those properties (because they *almost* do) then we have rendered our proof system unsound.
I also claimed that epsilon was a floating-point specific idea that had no bearing on the rationals, no more so than the integers, and that epsilon does not belong to the domain of discourse of any actual application nor of any mathematical theory.
Except there was, like, actual argument and examples and stuff, whereas the above is just a bunch of unsupported claims.
Marshall Received on Wed Mar 28 2007 - 13:25:10 CDT