From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Tue May 06 2003 - 01:08:36 PDT
There will be a few changes in the draft I release tomorrow.
The nutshell is that in trying to track down the ground/space
differences, I found that I had used the wrong aperture to normalize the
zeropoints of the HST data. I'd used a 0.5" diameter aperture, whereas
I should have used a 0.5" radius aperture. Fixing that was a 7%
effect. I really feel pretty much like an idiot at this point, so be
nice.
The ground/space offset isn't *gone*, but it's smaller. To see the
current lightcurves, look at:
http://brahms.phy.vanderbilt.edu/~rknop/scp/hst/hstltcvs_p1.ps
http://brahms.phy.vanderbilt.edu/~rknop/scp/hst/hstltcvs_p2.ps
I've spent a lot of time going back and redoing everything to fix this.
In the mean time, I also went back and iterated once more on my
uberspectrum. (It takes an uberspectrum to build an uberspectrum, since
K-corrections go into it.) The result of *that* is that the H96 set no
longer has a mean E(B-V) of -0.02, but only -0.005+-0.004 (which
hopefully is close enough to 0 for anybody). That cleans up some of the
mess in the paper; the relevant discussions on that will just be
shortened or deleted.
These color changes-- plus some color changes coming from fixing the HST
problems-- do mean the fits are all a bit different, but only a bit in
the end. For a while, I was worried that the extinction corrected fits
were blobbing out a lot, but ultimately what was doing that was the vast
*over*correction we see on a small number of reddened supernovae with
good error bars (e.g. 9855). Greg noted before that that one gets
overcorrected. There are a couple of others.
What I ended up doing was pointed out by Tony: Tony said, hey, why not
just do extinction correction fits on your primary subset? In fact,
that is now what I am doing. This will simplify the "subsets" business
in the paper as well. There are now only two relevant subsets:
everything (throwing out only the P99 stuff without any color
measurement), and the primary subset (throwing things out for all sorts
of reasons). There are *three* reddened HST SNe-- 9855, which was
always grossly reddened, 98122, the one that was previously *right* on
the border, and now 9878, the low stretch candidate whose color was more
affected by the change in the lightcurve than anything else. Between
9878 and 98104 (the two the spectral guys have complained about), 9878
is the one I'd more likely doubt as a Ia given the lightcurve. Since
it's out of our primary subset now, I'm not too worried about it.
(98104 stays in, and only goes out for the "contamination" systematic
estimation.) Despite this, the confidence intervals still look pretty
good.
Using only a single subset for all fits regardless of whether or not
E(B-V) corrections and error bars are explicitly applied saves us from
getting screwed up by overcorrected very reddened supernovae; more
mildly reddened ones won't throw things off so much if you overestimate
the correction, and the grossly reddened ones are not in the sample.
I *will* do a fit to "everything", to show it's consistent (which it is,
even if the chisquare there sucks rocks).
The conclusions and text, and even the limits on W, don't change
appreciably, so this really in the end will be a perturabitve change,
however much pain, work, and lack of sleep it cost me.
Look for the results of this (plus any comments sent me) tomorrow
(Tuesday). This will change the supernova subsets section, the table
which lists our fits (since I'll add a "fit to everything" there), and
maybe one or two other small things. (For example, it will be worth
noting in the "dust evolution" systematic section that a smaller R_V
could explain the overcorrection we're seeing for some of our
high-redshift reddened supernovae.)
Other new features, if I manage to get them finished, will be a
Malmquist fit (and addition to the systematic figure) and a dust
evolution fit (using a different RV for the high-redshift supernovae),
also adding to the systeamtic figure.
-Rob
-- --Prof. Robert Knop Department of Physics & Astronomy, Vanderbilt University robert.a.knop@vanderbilt.edu
This archive was generated by hypermail 2.1.4 : Tue May 06 2003 - 01:08:40 PDT