From: Greg Aldering (aldering@panisse.lbl.gov)
Date: Mon Sep 06 2004 - 18:34:01 PDT
Hi Serena,
On my last flight to Hawaii I was able to read your July 21st draft of the
I-band Hubble diagram paper. This paper is steadily improving and you are
to be commended for that.
During this reading session in particular, I was able to formulate a few
long-standing concerns, many of which center around 99Q or for which 99Q
is a good example case. Of course Chris' reanalysis might drastically
improve the situation for 99Q - I hope so! But, so that you don't have to
wait entirely for reanalysis of 99Q photometry, I am sending my
mark-up and these summary comments now:
a) There are issues with the choice of the +40 day cut-off. First, it
removes a critical point from the 99Q lightcurve later in the analysis.
This in turn leads to a large systematic uncertainty in the 99Q peak
brightness. Second, about half of the nearby SNe obviously deviate from
the template at late times and well before +40 days. Therefore,
although the reliability of the fits at late-time might be drawn into
question, there is no reason which justifys +40 days as a cut off.
Given this, I think you should not exclude the last 99Q point. If
nothing else, you need to address whether or not that point would have
helped address the ambiguity between the 99ac and 89B templates.
b) The systematic deviations from the fits need to be addressed head-on.
6 of 42 SN have deviant rising fits while 7 or 42 have reasonable
rising fits (the rest lack data). So, the fitting method works only
half the time when there is rising data. Of those with a
well-constrained 2nd peak, 2 fit poorly while 16 are well-fit.
In particular, the poor chi^2 for 94D is due to a systematic failure
of the model, not to poor NIR photometry or underestimated error bars
(as the paper now suggests). As no high-z SNe have rising data, you
may simply have to cut out the rising part - at least as a test of
whether these systematic errors matter.
c) I would like to see extinction applied for the I-band Hubble diagram.
99ff has extinction, as the later analysis shows, and 99Q might have
extinction (depending on how all the photometry works out).
For instance, for 99Q Vitaliy measures E(B-V)_Host = 0.095+-0.051, and
if he uses a time of max constrained to that given by Tonry, he gets
E(B-V)_Host = 0.200 +- 0.038. (My opinion is that applying Tonry's
constraint to our fit seems unnecessary now.) This extinction-corrected
Hubble diagram should be one of the comparisons shown in Fig 12.
(The statment in the text that these SN do not have extinction should
be modified, as that is based on the analysis of others which
apparently is inconsistent with our own).
To conclude, addressing a few overarching themes -- is the analysis
reasonable, justified and fair, and does the Hubble diagram look good --
will strengthen this paper considerable. Then people will listen to what
else the paper has to say.
Cheers,
Greg
This archive was generated by hypermail 2.1.4 : Mon Sep 06 2004 - 18:34:30 PDT