On Thu, 24 Oct 2002, _brian_d_foy wrote:

>> >>   modid:       Statistics::ErrorPropagation
>
>> >is this for basic arithmetic with numbers with uncertainities?
>
>> >does it define objects for those sorts of numbers?  a module to
>> >represent that might be live in another namespace.
>
>> That's about it, yes. I'm open to suggestions about a more appropriate
>> namespace.
>
>I think this might belong in Math::, since numbers with uncertainities
>are not necessarily statistical values.  a single scientific measurement,
>for instance, has an attached uncertainity, and you might have to add
>that to another, similar number.
>
>As far as I know the arithmetic rules are the same either way.

The rule I've implemented (which can be found in a basic stats textbook)
is based on combining the variances of data, so stricly, it does rather
assume that the data is statistical in nature.
But it's true, you could also apply the rules to numbers where the error
is a systematic experimental uncertainty rather than a statistical one, so
perhaps making these these objects more generally "mathematical" in
nature might be more useful.
Anyway, I'm happy to put this module in Math::
Need I submit another namespace registration?

---------------------------------------------------------------------------
Z. Sroczynski         telephone : +44 (0)151 794 3377
                            WWW : www.theorie.physik.uni-wuppertal.de/~zs/
                      snailmail : Theoretical Physics Division
                                  Dept. of Mathematical Sciences
                                  University of Liverpool
                                  Liverpool L69 3BX, U.K.
---------------------------------------------------------------------------

Reply via email to