1. RE: Digest for EUforum at topica.com, issue 5452
> Subject: Re: Calculating with imprecise values
>
>
> posted by: Juergen Luethje <j.lue at gmx.de>
>
> Matt Lewis wrote:
>
> <snip>
>
> > It sounds like fuzzy numbers and fuzzy arithmetic. I've
> thought about
> > writing something to do this, but haven't yet...
>
> I'm not sure whether I understand the meaning of the
> expressions "fuzzy
> numbers" and "fuzzy arithmetic" correctly. 
> I do not mean what is called "Fuzzy Logic" (according to
> Lotfi Zadeh and
> others) -- although I considers that very useful. 
>
> What I mean ATM is something that I think has got to do with "error
> propagation". Please see my reply to "jxliv7" for details.
>
> Thanks,
> Juergen
>
I think I can see what you mean: operating not on real numbers, but on reals
surrounded by a small fuzzy ball.
Building such a library wouldn't be too difficult, since all functions Eu
natively supports have derivatives - power() being possibly a problem.
How could one input the error arguments in add_with_error() or whatever
function that takes an error term? The information is not widely available
in most practical cases.
Plus, there are at least two sorts of error: absolute (=additive) and
relative (=multiplicative). Both should be taken into account, but this
makes the error determination question I mentioned above all the harder.
In a nutshell: it is obviously possible to code a library to perform such
"operations/functions with error", but using it requires usually unavailable
quantities. Its interest would be imho more theoretical than practical.
CChris
2. RE: Digest for EUforum at topica.com, issue 5452
Hi Christian, thanks for your reply.
>> Subject: Re: Calculating with imprecise values
>>
>> posted by: Juergen Luethje <j.lue at gmx.de>
>>
>> Matt Lewis wrote:
>>
>> <snip>
>>
>>> It sounds like fuzzy numbers and fuzzy arithmetic. I've
>> thought about
>>> writing something to do this, but haven't yet...
>>
>> I'm not sure whether I understand the meaning of the
>> expressions "fuzzy
>> numbers" and "fuzzy arithmetic" correctly.
>> I do not mean what is called "Fuzzy Logic" (according to
>> Lotfi Zadeh and
>> others) -- although I considers that very useful.
>>
>> What I mean ATM is something that I think has got to do with "error
>> propagation". Please see my reply to "jxliv7" for details.
>
>
> I think I can see what you mean: operating not on real numbers, but on reals
> surrounded by a small fuzzy ball.
> Building such a library wouldn't be too difficult, since all functions Eu
> natively supports have derivatives - power() being possibly a problem.
>
> How could one input the error arguments in add_with_error() or whatever
> function that takes an error term? The information is not widely available
> in most practical cases.
> Plus, there are at least two sorts of error: absolute (=additive) and
> relative (=multiplicative). Both should be taken into account, but this
> makes the error determination question I mentioned above all the harder.
>
> In a nutshell: it is obviously possible to code a library to perform such
> "operations/functions with error", but using it requires usually unavailable
> quantities. Its interest would be imho more theoretical than practical.
I think I understand what you are writing. However, I read the website
that Elliot had mentioned:
http://www.upscale.utoronto.ca/PVB/Harrison/ErrorAnalysis/index.html
and I think it covers the topic pretty fine -- even with a good sense of
humor.
Regards,
Juergen
