AT MetaNumerics?
Locked
 
|
Hello, does anybody use the AT MetaNumerics package? If so, would you
please let me know how you use it? TIA! Andres. _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
>From time to time I use Infinite (positive and negative) in matrices, intervals, constraints. But that's the only thing I've used in MetaNumerics. Uses: XHTML+CSS rendering, Constraint based layout, 3D modeling.
Michael
On 23 December 2010 09:27, Andres Valloud <[hidden email]> wrote:
_______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Andres Valloud-6
We just use +Infinity as the upper bound in a Range, so we can make
ranges to represent common cardinalities like 0..N or 1..N. When we later get a number of elements that should match that cardinality, we do a simple "less than" check against the upper bound. Steve > -----Original Message----- > From: [hidden email] [mailto:[hidden email]] On > Behalf Of Andres Valloud > Sent: 23. joulukuuta 2010 0:28 > To: [hidden email] > Subject: AT MetaNumerics? > > > Hello, does anybody use the AT MetaNumerics package? If so, would you > please let me know how you use it? TIA! > > Andres. _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Michael Lucas-Smith-2
Do you use Infinite together with other floating point values? If so,
could you simply use the floating point object for INF? On 12/22/2010 2:36 PM, Michael Lucas-Smith wrote: > >From time to time I use Infinite (positive and negative) in matrices, > intervals, constraints. But that's the only thing I've used in > MetaNumerics. Uses: XHTML+CSS rendering, Constraint based layout, 3D > modeling. > > Michael > > On 23 December 2010 09:27, Andres Valloud <[hidden email] > <mailto:[hidden email]>> wrote: > > Hello, does anybody use the AT MetaNumerics package? If so, would you > please let me know how you use it? TIA! > > Andres. > _______________________________________________ > vwnc mailing list > [hidden email] <mailto:[hidden email]> > http://lists.cs.uiuc.edu/mailman/listinfo/vwnc > > vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
It's usually unspecified, as in, I don't have to think about what kind of numbers they are. Since everything coerces up to floats/doubles anyway, the floating point INF object would work just as well. It was not available previously though.
On 23 December 2010 10:33, Andres Valloud <[hidden email]> wrote: Do you use Infinite together with other floating point values? If so, _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
What was not available when? Can you be a bit more specific? If you
mean the floating point INFs, they have been available for a while, at least since VW 7.5 IIRC when the IEEE floating point primitives were introduced. Since then, the base image was amended to handle IEEE special values gracefully. Even now, you can create a floating point INF and the image will handle it correctly even when the IEEE primitives are turned off (of course, the non-IEEE primitives will continue to fail if the answer is a special value like INF or NaN). Thus, I am curious about whether you need a meta object for INF, or perhaps the IEEE INF is a better match. From the emails I've received so far, it looks like not all infinities should be floating point. Is anybody using the meta-NaNs? Those are a bit of a pain because, if my memory serves me right, they do not behave the way an IEEE NaN must behave... On 12/22/2010 4:11 PM, Michael Lucas-Smith wrote: > It's usually unspecified, as in, I don't have to think about what kind > of numbers they are. Since everything coerces up to floats/doubles > anyway, the floating point INF object would work just as well. It was > not available previously though. > > On 23 December 2010 10:33, Andres Valloud <[hidden email] > <mailto:[hidden email]>> wrote: > > Do you use Infinite together with other floating point values? If so, > could you simply use the floating point object for INF? > > On 12/22/2010 2:36 PM, Michael Lucas-Smith wrote: > > >From time to time I use Infinite (positive and negative) in > matrices, > > intervals, constraints. But that's the only thing I've used in > > MetaNumerics. Uses: XHTML+CSS rendering, Constraint based layout, 3D > > modeling. > > > > Michael > > > > On 23 December 2010 09:27, Andres Valloud <[hidden email] > <mailto:[hidden email]> > > <mailto:[hidden email] <mailto:[hidden email]>>> wrote: > > > > Hello, does anybody use the AT MetaNumerics package? If so, > would you > > please let me know how you use it? TIA! > > > > Andres. > > _______________________________________________ > > vwnc mailing list > > [hidden email] <mailto:[hidden email]> > <mailto:[hidden email] <mailto:[hidden email]>> > > http://lists.cs.uiuc.edu/mailman/listinfo/vwnc > > > > > _______________________________________________ > vwnc mailing list > [hidden email] <mailto:[hidden email]> > http://lists.cs.uiuc.edu/mailman/listinfo/vwnc > > vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
Well in the grand scheme of things, 7.5 is quite recent. But more specifically than that, there's no API for creating INF directly, not without loading a separate package. So that leaves it in the same boat as MetaNumerics. Given that we can do it, we probably should have an API for it. Squaro has an API for it that we support in our extensions for Seaside.
Do we need a meta object for it? I'd say probably not. You can do more with an IEEE INF. It takes less coercing to do magnitudes against it or math operations of any kind. For my purposes, it doesn't matter what the form of the INF is, so long as I have it easily available to me. Michael On Dec 23, 2010, at 11:27 AM, Andres Valloud wrote: > What was not available when? Can you be a bit more specific? If you > mean the floating point INFs, they have been available for a while, at > least since VW 7.5 IIRC when the IEEE floating point primitives were > introduced. Since then, the base image was amended to handle IEEE > special values gracefully. Even now, you can create a floating point > INF and the image will handle it correctly even when the IEEE primitives > are turned off (of course, the non-IEEE primitives will continue to fail > if the answer is a special value like INF or NaN). Thus, I am curious > about whether you need a meta object for INF, or perhaps the IEEE INF is > a better match. > > From the emails I've received so far, it looks like not all infinities > should be floating point. > > Is anybody using the meta-NaNs? Those are a bit of a pain because, if > my memory serves me right, they do not behave the way an IEEE NaN must > behave... > > On 12/22/2010 4:11 PM, Michael Lucas-Smith wrote: >> It's usually unspecified, as in, I don't have to think about what kind >> of numbers they are. Since everything coerces up to floats/doubles >> anyway, the floating point INF object would work just as well. It was >> not available previously though. >> >> On 23 December 2010 10:33, Andres Valloud <[hidden email] >> <mailto:[hidden email]>> wrote: >> >> Do you use Infinite together with other floating point values? If so, >> could you simply use the floating point object for INF? >> >> On 12/22/2010 2:36 PM, Michael Lucas-Smith wrote: >>>> From time to time I use Infinite (positive and negative) in >> matrices, >>> intervals, constraints. But that's the only thing I've used in >>> MetaNumerics. Uses: XHTML+CSS rendering, Constraint based layout, 3D >>> modeling. >>> >>> Michael >>> >>> On 23 December 2010 09:27, Andres Valloud <[hidden email] >> <mailto:[hidden email]> >>> <mailto:[hidden email] <mailto:[hidden email]>>> wrote: >>> >>> Hello, does anybody use the AT MetaNumerics package? If so, >> would you >>> please let me know how you use it? TIA! >>> >>> Andres. >>> _______________________________________________ >>> vwnc mailing list >>> [hidden email] <mailto:[hidden email]> >> <mailto:[hidden email] <mailto:[hidden email]>> >>> http://lists.cs.uiuc.edu/mailman/listinfo/vwnc >>> >>> >> _______________________________________________ >> vwnc mailing list >> [hidden email] <mailto:[hidden email]> >> http://lists.cs.uiuc.edu/mailman/listinfo/vwnc >> >> > _______________________________________________ > vwnc mailing list > [hidden email] > http://lists.cs.uiuc.edu/mailman/listinfo/vwnc _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
It's true, we do not have an API to create INFs. And, now that I think
about this further, we do not even have an API to enable / disable the IEEE primitives easily... well, I think it should be fixed, so I will look into this. But back to INF. Fortunately, it's not that hard to create an expression that answers INF, for example LimitedPrecisionReal class>>positiveInfinity ^self unity timesTwoPower: self emax + 1 However, you'd need to turn on the IEEE prims for that to work. Would you expect an INF not to cause you trouble if the IEEE prims were turned off? With the exception of comparisons, the vast majority of operations would fail because the answer would be INF or NaN. On 12/22/2010 5:24 PM, Michael Lucas-Smith wrote: > Well in the grand scheme of things, 7.5 is quite recent. But more specifically than that, there's no API for creating INF directly, not without loading a separate package. So that leaves it in the same boat as MetaNumerics. Given that we can do it, we probably should have an API for it. Squaro has an API for it that we support in our extensions for Seaside. > > Do we need a meta object for it? I'd say probably not. You can do more with an IEEE INF. It takes less coercing to do magnitudes against it or math operations of any kind. For my purposes, it doesn't matter what the form of the INF is, so long as I have it easily available to me. > > Michael > > On Dec 23, 2010, at 11:27 AM, Andres Valloud wrote: > >> What was not available when? Can you be a bit more specific? If you >> mean the floating point INFs, they have been available for a while, at >> least since VW 7.5 IIRC when the IEEE floating point primitives were >> introduced. Since then, the base image was amended to handle IEEE >> special values gracefully. Even now, you can create a floating point >> INF and the image will handle it correctly even when the IEEE primitives >> are turned off (of course, the non-IEEE primitives will continue to fail >> if the answer is a special value like INF or NaN). Thus, I am curious >> about whether you need a meta object for INF, or perhaps the IEEE INF is >> a better match. >> >> From the emails I've received so far, it looks like not all infinities >> should be floating point. >> >> Is anybody using the meta-NaNs? Those are a bit of a pain because, if >> my memory serves me right, they do not behave the way an IEEE NaN must >> behave... >> >> On 12/22/2010 4:11 PM, Michael Lucas-Smith wrote: >>> It's usually unspecified, as in, I don't have to think about what kind >>> of numbers they are. Since everything coerces up to floats/doubles >>> anyway, the floating point INF object would work just as well. It was >>> not available previously though. >>> >>> On 23 December 2010 10:33, Andres Valloud<[hidden email] >>> <mailto:[hidden email]>> wrote: >>> >>> Do you use Infinite together with other floating point values? If so, >>> could you simply use the floating point object for INF? >>> >>> On 12/22/2010 2:36 PM, Michael Lucas-Smith wrote: >>>>> From time to time I use Infinite (positive and negative) in >>> matrices, >>>> intervals, constraints. But that's the only thing I've used in >>>> MetaNumerics. Uses: XHTML+CSS rendering, Constraint based layout, 3D >>>> modeling. >>>> >>>> Michael >>>> >>>> On 23 December 2010 09:27, Andres Valloud<[hidden email] >>> <mailto:[hidden email]> >>>> <mailto:[hidden email]<mailto:[hidden email]>>> wrote: >>>> >>>> Hello, does anybody use the AT MetaNumerics package? If so, >>> would you >>>> please let me know how you use it? TIA! >>>> >>>> Andres. >>>> _______________________________________________ >>>> vwnc mailing list >>>> [hidden email]<mailto:[hidden email]> >>> <mailto:[hidden email]<mailto:[hidden email]>> >>>> http://lists.cs.uiuc.edu/mailman/listinfo/vwnc >>>> >>>> >>> _______________________________________________ >>> vwnc mailing list >>> [hidden email]<mailto:[hidden email]> >>> http://lists.cs.uiuc.edu/mailman/listinfo/vwnc >>> >>> >> _______________________________________________ >> vwnc mailing list >> [hidden email] >> http://lists.cs.uiuc.edu/mailman/listinfo/vwnc > > vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Michael Lucas-Smith-2
Would anyone think a NaN API is necessary?
On 12/22/2010 5:24 PM, Michael Lucas-Smith wrote: > there's no API for creating INF directly _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Andres Valloud-6
Andres
We use Infinity in many places. Most of its usage is to represent a count or size. Terry =========================================================== Terry Raymond Crafted Smalltalk 80 Lazywood Ln. Tiverton, RI 02878 (401) 624-4517 [hidden email] <http://www.craftedsmalltalk.com> =========================================================== > -----Original Message----- > From: [hidden email] [mailto:[hidden email]] On Behalf Of Andres Valloud > Sent: Wednesday, December 22, 2010 5:28 PM > To: [hidden email] > Subject: [vwnc] AT MetaNumerics? > > Hello, does anybody use the AT MetaNumerics package? If so, would you > please let me know how you use it? TIA! > > Andres. > _______________________________________________ > vwnc mailing list > [hidden email] > http://lists.cs.uiuc.edu/mailman/listinfo/vwnc _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
«
Return to VisualWorks
|
1 view|%1 views
| Free forum by Nabble | Edit this page |