On 6/23/25 01:22, Waldek Hebisch wrote:
ATM I did not look at conditions and categories.  However,
I have noticed a different issue.  Namely, Interval is
supposed to be an implementation of interval artithmetic.
Which means that interval obtained as result of
interval arithmetic is supposed to contain interval
computed using infinitely precise artithmtic.  Since
operations in Float are approximate it means that
results of arithmetic operations on intervals need to be
enlarged to compensate for effect of rounding.  This
is done by use of roundUp and roundDown in qinterval.

Right. That is why qInterval has no default implementation. And Interval does not inherit anything from AInterval only from AIntervalCategory.
AInterval is for the exact case, Interval for Floating point intervals.

I did not check if roundUp and roundDown _always_ lead
to correct result, but without them results are clearly
incorrect.

Yes. But roundUp and roundDown are implemented / used in Interval and there is no default package for IntervalCategory.

So, most of operations in Interval should stay the same.

Yes, when I first designed AIntervalCategory, I had in mind that the semantics of Interval should not change. Of course, I could have made a mistake, but then it would have been unintentional.

Maybe I should come up with a number of tests?

Ralf

--
You received this message because you are subscribed to the Google Groups "FriCAS - 
computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/fricas-devel/3ceacdae-7dba-43f0-b143-08949797f0e3%40hemmecke.org.

Reply via email to