Re: [boost] arithmetic type wrapper and physical constants

From: boost (boost_at_[hidden])
Date: 2001-04-24 05:56:06

Salut,

On Tuesday 24 April 2001 00:20, you wrote:
> I have concluded that we MUST have a way of storing with the units
> the uncertainties (for the measurement-minded)
> (or intervals to the mathematically-minded).
>
> (For a few, uncertainties are zero - by definition conversion of inch to mm
> is 25.4 exactly.)
>
> But for most physical constants and conversion factors
> they are + something - and often - something different.
> This makes it more complicated!
[...]
> And are limits 95% confidence or what?

I dont't think that Boost needs to provide the errors of the
physical constants. Those who need them will have to be very
careful anyway. In addition ther are relative and systematic errors
which makes everythings more complicated. Providing COORDA values
of the constants should be fine. In my previous posts I just
wanted to point out that these values may change with time,
e.g. '86 and '98 defintions.

For mathematical constants the error is a well-defined number.

Best wishes,
Peter