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

Re: min / max and empty intervals



On Jun 7 2011, Dan Zuras Intervals wrote:

I'd consider this an ill-formed constant, returning canonically Empty_ill. Whereas if any of the arguments x_i is Empty (with decoration emp or ill), the returned value is Empty, with the worst decoration of all the empty arguments. This is consistent and allows for straightforward debugging.

	Oh, I think that's a bit draconian, don't you?

That's what provable RAS is all about!  No pain, no gain :-(

	Actually, John touched on the reasonable application for
	NaNs in a matrix.  That of as yet unknown data.

Those aren't NaNs, as specified in IEEE 754 - those are missing values,
as used in statistical software since the 1960s.  They have different
semantics, though USUALLY max(<missing>,<value>) should be <missing>;
the only case when it should be <value> is when the max is a reduction
and not a normal operation.
	So, I don't know if this applies to the interval world.
	In an interactive application, are you interested in
	providing assured partial results or waiting until the
	input is completed even if its not all needed?

Either - but you don't do that by abusing NaNs or empty intervals.
You do that by coding it properly in terms of operations and reductions,
where the latter allow for the possibility of taking a snapshot.  That
is essentially what sequential sampling is in statistics.


Regards,
Nick Maclaren.