On Thu, Jul 12, 2001 at 06:09:07PM -0400, Dan Sugalski wrote:
> Depending on what you do with them, precision (or, rather,
> significant digits) is a useful concept for integers as well. Just
> because you have, for example, an integer with 43 digits doesn't
> mean that all 43 are actually useful or trustable--you may only have
> 2 or 3 that mean anything.
I digress...
They went through Great Pains in Engineering class to hammer into our
head that 43 != 43.0 != 4.3 x 10 in the land of significant figures
and taught us all these special considerations for preserving sig figs
through mathematical operations.
A Math::SigFigs might be interesting, if I ever found myself needing
them I might write it.
For those of you whose branes haven't been loaded down with this bit
of engineering baggage, the idea is to express the accuracy of a
measurement through a number. If you've got a normal ruler and
measure the length of something, you're probably going to get accurate
it down to the centimeter. 43 cm, say. What you really mean is 43 cm
plus or minus 0.5, and that's what "43 cm" means in significant figures.
Say you then measure something else with a highly accurate tool and
get 20.5432 cm. Again, what you mean is 20.5432 cm +/- 0.00005 What
is the total length of those two things? 63.5432 cm, right? Not in
the land of sig figs. Because the first measurement is so much more
innacurate than the first, its innaccuracies swamp the measurement.
The answer is 63 (or 64, I forget which way the rounding goes) to
reflect the inaccuracies involved. 63 cm +/- 0.5 cm.
I have no idea if this is what Dan was thinking.
--
Michael G. Schwern <[EMAIL PROTECTED]> http://www.pobox.com/~schwern/
Perl6 Quality Assurance <[EMAIL PROTECTED]> Kwalitee Is Job One
If you got the wax out of your ears you could hear the twister picking up
the trailer park of your future!
