[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Zero Welch-Stetson Indices found



Doug Welch wrote:
> 
> On Thu, 13 Jun 2002, jg wrote:
> 
> > Doug Welch wrote:
> > >
> > > John,
> > >
> > > You are correct that stars need to meet a certain set of criteria
> > > to make into the output. However, the index=0.000 case will make it
> > > through, as far as I can see. Why do you think otherwise?
> > >
> > > Doug
> >
> >
> > AhAh! Don't know perl at all but the following from the wsv3.pl on the
> > CD suggests an output of 0.000 is not going to happen :-
> >
> > >   if ($ws >  0.000 && $ws < 0.001) { $ws =  0.001; }
> > >   if ($ws > -0.001 && $ws < 0.000) { $ws = -0.001; }
> >
> > (immediately followed by the printf output statement)
> >
> > So the 0.000 ones are being reported, but not noted as such?  Needs
> > 0.000 to +/- 0.00049999 conditionals to get $ws = 0?
> >
> 
> John,
> 
> The conditionals are not >= or <=, so 0.000 makes it through
> unscathed. The only reason I put those statements in was to
> avoid floats being printed. This was probably unnecessary -
> I didn't check what PERL actually does in those circumstances.
> 
> Still, 0.000 should make it out. It just is not all that likely
> a value.
> 
> Cheer,
> Doug


Okay, first of all I should note to Andrew Bennett that the 94,500+
figure is wrong, only roughly 33,500 figures have more than 30 pairs of
observations (as determined by wsv3.pl from the CD) so roughly 61,000 of
the roughly 94,500 objects are excluded simply for not having enough
pairs.  Sorry about that.  So the negative index values are relative to
the roughly 33,500 figure.  Mind you, though tripling the percentage of
negative objects, still a small percentage relative to the whole as you
noted.

Doug,

Now, the conditionals.  I couldn't see anything prior to the "if tests"
which looked like it set the numerical precision, and was pretty certain
that the default perl precision would be greater than the rounded output
of seven significant figures with three decimals, ie never 0, but
0.0000000001 or something, so still <>0.  So I took the conditionals out
and re-ran the wsv perl script against collected.big.

This produced some zero (0.000) value WS indices.

Fourteen of them in total ;)

But there are some there at least ;)

Cheers

John G.