Thread Links | Date Links | ||||
---|---|---|---|---|---|
Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
Vincent Lefevre wrote:
On 2010-10-15 06:42:58 -0400, Michel Hack wrote:This discussion again suffers from the one-sided view that intervals represent uncertain numbers. When intervals are used to express ranges of values, the concept of a midpoint frequently doesn't even make sense.But one wouldn't normally use mid-rad to express ranges of values; inf-sup is more suited for that.
inf-sup is indeed more suited for _most_ purposes, except for extremely high precision, where the triplex format is the best.
But I thought the discussion was about whether a multiprecision arithmetic could be the _only_ supported interval type.
Then one _must_ use it for ranges; so it must support ranges well.
I'm just saying that an implementation should have the choice, depending on the application. Just like IEEE-754 leaves the choice of the radix to the implementation (and possibly both binary and decimal could be implemented, but this is not required).