Re: Cosmic acceleration rediscovered
From: Greg Hennessy (greg.hennessy_at_tantalus.cox.net)
Date: 01/29/05
- Next message: CNC Area: "Men on the moon?"
- Previous message: Sam Wormley: "Sizing Up Complex Webs: Close or far, many networks look the same"
- In reply to: greywolf42: "Re: Cosmic acceleration rediscovered"
- Messages sorted by: [ date ] [ thread ]
Date: Sat, 29 Jan 2005 04:57:25 +0000 (UTC)
In article <6FDKd.6794$VA5.5384@fe07.usenetserver.com>,
greywolf42 <mingstb@marssim-ss.com> wrote:
> > > LOL! Figure 2 has nothing to do with the measurements (as I described
> > > below -- and you snipped).
> >
> > The text of figure 2 is:
> > Monte Carlo simulations of the probability density function of \te\,
> > for the value of $N$(\ion{C}{ii}$^\ast$)/$N$(\ion{C}{ii}) =
> > $3.8\times10^{-3}$. The mean value is \te\, = $12.1^{+1.7}_{-3.2}$ K
> > [the $\frac{1}{2}(1-p)$ and $\frac{1}{2}(1+p)$ quantiles were used to
> > estimate the uncertainty interval at $p = 0.95$]. The \tr from the
> > standard Big Bang cosmological model is marked with a vertical dashed
> > line.
>
> Try translating your cut-and-pastes.
Well, since you had a copy of the paper I thought it would be obvious,
but if you insist "Monte Carlo simulations of the probability density
function of Teff for the value of N(CII*)/N(CII)= 3.8E-3.
It shows the probability of a the measured ratio being caused by a
range of values for Teff. Which proves you incorrect when you claimed
the figure had nothing to do with the measurments.
> LOL! But it *doesn't* have anything to do with measurements.
Since if you change the ratio of N(CII*)/N(CII) the derived value of
Teff changes then it in fact has something to do with the
measurements.
> > Molaro provides two data points and error values for those points.
>
> No. Molaro provides two experimental values, with standard deviations for
> those values. Individual data points *never* have "error values."
Of course they can. Now if you try to obtain the error value by
examining several values and calculating a variance of the data points
and then a standard deviation, in that subset of possible ways to
determine error values you are correct, you can't have a standard
deviation for a individual data points, however if you calculate a
column density by a fit to a spectrum, then your fitting routine can
deliver a single column density and an error for that single column
density.
> > There are multiple velocity structures in the quasar, and the
> > different components will have different strengths. The measurements
> > are independant, even if scaling factors
>
> Even if scaling factors "what?"
"are used".
> 5 to 6 standard deviations is a bit more than "somewhat" discrepant. Now,
> what does your statistics 101 text say about discrepant results?
That scientists almost always underestimate the errors.
The two values differ by about 35%. Is it really worth arguing over?
Expecially since the the temperature depends on the log of ratio, not
the ratio. If we forget about the P&W data entirely, the calculated
value for Teff chages from about 12.1K to 10.2K. The value predicted
by the Standard Big Bang model is about 11 degrees anyway, so the
temperature inferred by the experiment is STILL consistent with the
SBB model.
> > Classic special plead.
>
> LOL! Again, if I had provided no support that would be a
> proof-by-assertion, not a special plead. Did you ever take a logic course?
Yes. Have you ever taken a science course?
http://en.wikipedia.org/wiki/Special_pleading
Special pleading is a form of spurious argumentation. Special pleading
for a position in a dispute introduces favorable details or excludes
unfavorable details by alleging a need to apply additional
considerations without proper criticism of these considerations
themselves
You introduce the additional consideration that the values have to be
"consistent". They don't.
> > The formulas can in fact be used with inconsistent results.
>
> Hmmm. Not according to my statistics 101 book (Young, p 109):
>
> "In using equations (14.13) and (14.16)*, one should keep in mind that the
> variance associated with each x_i also provides a means of testing whether
> the values are CONSISTENT in a statistical sense.
What you quote (and the text I snipped for brevity does not contradict
my point) in no way claims that the formulas cannot be used with
inconsistent data.
> Excuse me, but no "data" was actually provided in this paper. Only an
> experimental value and a standard deviation.
We were provided with a measure of N(CII*) and an error value for that
datum, a measure of N(CII) and an error value for that datum, we were
provided with the redshift of the quasar (3.025) without an error
value, we were provided with a N(HI) density and the error value, the
abundance F(H2) and an error value for the datum, and a measured rate
of photo absorbtion beta.
Thus data was provided.
> > And if you use a the weighted mean of 1.92+- 0.08 then the two values
> > are within about 4 sigma.
>
> LOL! They are beyond 5 sigma of *each other*. The "mean" was not measured.
Mean's aren't measured, means are calculated.
> > > Let's assume for the sake
> > > of simple demonstration that the error bars are the result of two
> > > measurements, each, in Molaro and P&W.
> >
> > Why would I assume something dumb like that? The error bars came from
> > a fit to a single spectrum.
>
> How many measurements were in that "single spectrum?" ;)
Measurements of what? Of N(CII*)? One.
> > We do not. That is not how Molaro came up with the error value.
>
> How do you know? Molaro didn't tell us how he came up with the value.
I know how you measure a column density given a spectrum.
> They deal specifically with UV flux (paragraph 3), infrared photons from
> dust (paragraph 4), particulate collisions (paragraph 5), and electron
> collisions (paragraph 6).
Yes, and you claimed they didn't deal with "other" sources of
energy. Those are other sources.
- Next message: CNC Area: "Men on the moon?"
- Previous message: Sam Wormley: "Sizing Up Complex Webs: Close or far, many networks look the same"
- In reply to: greywolf42: "Re: Cosmic acceleration rediscovered"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
