On Fri, Jun 17, 2016 at 10:33 AM, John Peterson <jwpeter...@gmail.com>
wrote:
>
>
> On Fri, Jun 17, 2016 at 8:16 AM, David Knezevic <
> david.kneze...@akselos.com> wrote:
>
>> We use TypeVector::relative_fuzzy_equals in a number of places in the
>> library. I'm not convinced that this function makes sense, since the
>> tolerance effectively depends on where the points happen to be located. I
>> can't think of any situation where this behavior is desirable.
>>
>> My preferred approach is to always use absolute_fuzzy_equals, and to use
>> a tolerance that is relative to some reference length, if a
>> reference length is available.
>>
>
> Isn't this what relative_fuzzy_equals already does, where the reference
> length in question is (||x||_{L1} + ||y||_{L1})? It is implemented in
> terms of absolute_fuzzy_equals...
>
> I guess you are saying that this reference length is rarely the
> appropriate choice?
>
Yeah, that's a bad choice. If both points (x and y) are near the origin,
then that tolerance will be super small. If both points are far from the
origin the tolerance will be super large.
As a result, my preference would be to remove relative_fuzzy_equals from
the library.
David
------------------------------------------------------------------------------
What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic
patterns at an interface-level. Reveals which users, apps, and protocols are
consuming the most bandwidth. Provides multi-vendor support for NetFlow,
J-Flow, sFlow and other flows. Make informed decisions using capacity planning
reports. http://sdm.link/zohomanageengine
_______________________________________________
Libmesh-devel mailing list
Libmesh-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmesh-devel
