From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Wed May 21 2003 - 05:59:01 PDT
On Tue, May 20, 2003 at 11:48:41PM -0700, Greg Aldering wrote:
Re: describing the Hatano distribution without introducing panic and
mayhem: any suggestions?? I'll try something in the draft I send out
today, but I fear this is going to be a sticking point.
I think there are two main issues: Don's issue and the deception issue.
Don's issue is that by modifying the E(B-V) and error bars, you are
confusing the difference between an intrinsic dispersion and a
dispersion due to measurment uncertainties; this is where the bias comes
from. Our approach isn't sensitivie to this. The deception issue is
that they use this prior to kill their E(B-V) error bars and then claim
to have done an extinction correction-- whereas in fact the real
uncertainties in their measured colors have not gone into the
uncertainties on the cosmological parameters.
Our approach may be similarly motivated, but it doesn't even claim to
propogate in color errors. Additionally, as you note, because it's not
modifying points it doesn't introduce any new bias.
Unfortunately, this is a mess, because the discussion of the evil prior
and our ridgeline methods probably needs to be merged, but they're in
different places right now.
> p22 figure 5: A few things to think about - the HST points, not the rejected points
> should be highlighted. Also, the zoom box misses two of our SNe. The
> totally reject points don't seem like they should be shown. All the
> P99 data should be lighter or something, since otherwise all the noisy
> points get the attention (binning would be another approach.)
>
The reason I didn't highlight the HST points is that there's a separate
figure which has just the HST data-- that seemed to me like enough of an
HST highlight.
> p3 : it would be better if ""Fit E" of P99" and Riess references were swapped
> at the end of the last full paragraph in the left column. I understand
> that \cite might not offer this option.
\cite does offer the option-- in fact, it's easier that way. The real
issue is that next time Saul sees it, he will edit it to put Fit E first
again.... (I think Fit E used to be second.) He seems to think that we
need to take first credit for the prior method, and has systematically
edited the text to suggest that.
-Rob
-- --Prof. Robert Knop Department of Physics & Astronomy, Vanderbilt University robert.a.knop@vanderbilt.edu
This archive was generated by hypermail 2.1.4 : Wed May 21 2003 - 05:59:06 PDT