On Tue, 30 Nov 1999, Peter Shenkin wrote:
> What other things? Well, maybe someone would want to write
> a module to do propagation of statistical error based on inputs
> of value +/- sigma, just to revive my own example.
Well I did write one some time ago (though not very comprehensive). I
didn't find it very useful, actually, probably because the equations for
error propagation assume Gaussian distributions and no correlation between
errors in the variables, neither of which is a safe assumption in many
practical situations. I'd happily make the code available, if only I
could remember what I called it :-(. Like interval arithmetic, it may be
more of academic than practical interest.
--
Clive Page,
Dept of Physics & Astronomy,
University of Leicester.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|