Talk:Extreme floating point values

Why Draft?

Because I am no expert in this field and might not be using the right terms in the description or might have left something out. --Paddy3118 19:39, 15 July 2010 (UTC)

Unfortunate task name

These values are not extreme. They are not numbers, and of course, not floating point. Technically they are called ideals and used to make operations like +,-,*,/ closed. Another, often better, example of ideals are numeric exceptions. They too close operations. For example, + is closed in the set of real numbers filled up by overflow-exception ideal. --Dmitry-kazakov 21:37, 15 July 2010 (UTC)

Hi Dmitry, I have never heard of 'Ideals' with respect to floating point; and indeed, it is not in the quoted reference What Every Computer Scientist Should Know About Floating-Point Arithmetic. After reading the reference again, it seems to call them 'Special Quantities' which would lead to a task title something like "Special quantities in floating point values" which would read better than say "Ideal values in floating point values". If, however, ideal is a common term amongst floating point specialists, then maybe we should use the latter? --Paddy3118 21:52, 15 July 2010 (UTC)

They are values of the IEEE floating point type. --71.141.138.49 04:33, 16 July 2010 (UTC)

Do you know if the IEEE spec' uses a blanket term for them? I had hoped to leave the description loose enough for those using other floating point spec's to adapt accordingly and form their own entries. As the very informative Ada entry has managed to do. ( Yeah!) --Paddy3118 05:43, 16 July 2010 (UTC)