From: Serena Nobili (serena@physto.se)
Date: Fri Dec 26 2003 - 05:39:04 PST
Hi Alex,
here are the relevant answers to your comments:
On Thu, 18 Dec 2003, Alex Conley wrote:
> b) Why no sn1991bg in figure 1? You state that 91bg and 97cn aren't
> there, but don't explain why. If this means that you cut all
> subluminous objects from the data set, you should state that more
> clearly, justify it better, and eliminate the paragraph explaining
> how you estimated the error bars for 91bg. If you don't actually
> use it for anything, don't waste a bunch of space talking about
> it.
>
1991bg and 1997cn were simply showed in a separate plot, i.e. figure 5 in
your version; I put everything in Figure 1 continued now. For what
concerns the discussion about the uncertainties of 1991bg, I use that for
the lightcurve fitting, and I think it is fair to describe what I did, if
anyone wants to repeat the exercise.
>K-corrections:
>
>a) Do you have the right I band filter? First, you should give a source
> for standard Bessel I band. You probably have the transmission from
> Bessel PASP 102:1181 (1990). If that is the filter curve you are using,
> you need to be aware that the normalized passbands in Table 2 are NOT
> the dimensionless transmission R, but are actually lambda * R. That is,
> the filter is slightly bluer than you may think. This is not explained
> in the text of the paper.
>
> This is something that has been done incorrectly in all previous
> published SCP work. The effect is not large -- if you would like
> to see the size of the difference, take a look at
> http://panisse.lbl.gov/~aconley/kdiff.eps
>
> This was discovered due to a note in Jha's thesis. I ended up emailing
> Mike Bessel directly to confirm this. So you should look up Bessel
> (1990), compare it to your filter curve, and if it the one you are using
> you have a (minor) problem -- but one that isn't too hard to fix.
> Just divide your filter file by lambda, renormalize, and redo all your
> K-corrections.
>
I am correcting everything now. I get very small differences, of the order
of 0.01--0.05 mag. Thus the conclusions of the paper are not going to
change. It is just a slow-long-boring work redoing all the plots and
tables! I think it would be nice of you to send un email to deepnews
advertising this problem with Bessel's filters. If I knew it 1 month ago,
it would have been less of an headache ;)
>b) You say you use time information fro the B-band data. Is this
> available for all of the SNe? For example, I have never found a good
> lightcurve for sn1991bg with error bars in the B band.
>
In the case of 1991bg I have used the time if Bmax from
Filippenko et al. 1992AJ....104.1543F, which is estimated
as 2 days earlier than V-band maximum.
>The I-band Hubble diagram
>
>c) What is your source for the uncorrected Hubble diagram having
> sigma=0.4? That's higher than most quoted values, which are usually
> more like 0.3. Phillips '99 gives 0.24 for the B band without dm15
> correction.
>
I think I took it from Branch and Miller (1993) which quote 0.36 mag.
However, I will write "about 0.3 mag", since this is less controversial.
>4. High redshift supernovae
>
>a) In figure 11 you show a redshifted I band. Redshifted to what?
> Probably to 0.5, but I don't see that noted in the figure or it's
> caption.
>
It is red-shifted to z=0.543, i.e. the redshift of SN 2000fr, which is
the SN described in that section. I added a line in the caption to make it
clear.
>b) Did you use the I_{max} values from table 2 for the low-z SNe? I don't
> think that is a good idea because the high-z SNe were not fitted in the
> same way. I think you have little choice but to refit all of the low-z
> SNe with the template so that you treat the high and low redshift SNe
> the same way. You also can't stretch or host galaxy extinction correct
> the low-z sample when combining it with the high-z sample -- it isn't
> clear from the paper whether this was done or not.
>
I understand your point and in the paper I discuss the problems that could
be connected with the different fitting procedure applied to the low and
the high-z sample. However I don't really see the reason for fitting again
the low-z SNe. If I use the templates I built on them to fit the same
data, I would get the same values of the Imax. Moreover I take into
account differences in the fitting procedure of the high-z SNe, by adding
a systematic uncertainty to the results of the high-z sample. The Hubble
diagram up to z=0.5 has the main purpose of showing its feasibility.
> You also can't stretch or host galaxy extinction correct
> the low-z sample when combining it with the high-z sample -- it isn't
> clear from the paper whether this was done or not.
>
The low-z and high-z were treated in the same way in this section for what
concerns the stretch correction. Figure 15 shows the two cases,
not stretch corrected (middle panel) and stretch corrected (bottom panel).
However, the host galaxy correction was
applied to the low-z sample only. The main reason for that is the fact
that we do think the high-z SNe are not extincted in their host. To be
fair (and consistent with what I do later in the paper for the colors),
I will plot the Hubble diagram residuals in the case of no-extinction
corrected low-z, and add it to figure 15.
I will also make this more clear in the text.
Thanks again for your comments.
Merry Christmas and Happy New Year!
Serena
----------------------------------------------------------------------
www.physto.se/~serena
Tel +46 8 55378661
Give free food at:
http://www.porloschicos.com/
http://www.thehungersite.com/
This archive was generated by hypermail 2.1.4 : Fri Dec 26 2003 - 05:39:08 PST