From: Chris Lidman (clidman@eso.org)
Date: Mon Dec 08 2003 - 13:57:51 PST
Hi Serena (with cc to Rob, Ariel and Tony),
Thanks for the paper. I add comments, replies and questions below.
Do we have an e-mail archive for your paper yet?
Cheers, Chris.
On Wed, 2003-12-03 at 07:34, Serena Nobili wrote:
> Dear Chris,
>
> I have implemented your (and Rob's) suggestions in the paper, and it is
> now in a better form. Sections 5 and 6 underwent major changes due to
> the
> exclusion of SN 1999fn (see my email of November 25), but also other
> sections have been modified slightly. In particular note the
> discussion at
> the end of Section 3 about differences in the low-z samples. I updated
> the
> paper in the web page:
>
> www.physto.se/~snova/internal/papers/iband/
>
>
> (in your last email you said that you found difficult to read the web
> page, I tried it in different browsers and it worked well in all of
> them,
> can you tell me what kind of problems you have so that I can try to fix
> it.)
>
It's OK now. It must have been my web browser on Paranal.
> The are only a couple of issues I did not address. These are:
>
>>
>> I think that you misunderstand what I had meant. Here is a recipe of
>> what I meant.
>>
>> - Take one nearby SNe.
>>
>> - Do the light curve fit to this SNe as you would normally do (section
>> 2.2). From this we get an estimate of I_max.
>>
>> - Throw away points until just three are left. Perhpas this is a bit
>> unfair since we do not space our observations in a random manner. As
>> an
>> alternative you can divide the light curve into three sections, one
>> around maximum light, another around the minimum and a third around
>> the
>> second maximum. Then you delete points until one point is left in each
>> section.
>>
>> - Incease the error bars to match the errors in the high z SNe.
>>
>> - Use the method that you use for the high z SNe (section 4.5) to fit
>> the
>> lightcurve of this artifically created data set, but do not include
>> the
>> one nearby SNe that was used to create the artificial data set. This
>> gives you a second estimate of I_max.
>>
>> - How does the second estimate compare to the first one?
>>
>> - Repeat this exercise for all nearby SNe.
>>
>
> This could be an interesting test. However since I don't think is going
> to change anything in the conclusions, I will do it while the
> paper is in the review pipeline of the collaboration.
>
>
Good. This will be an important test.
>> I had a look at plot callled read_fitlog.ps and this is very
>> interesting.
>> One can use it in several ways. You have used in the following way.
>>
>> - Take the SNe with the lowest chisq to determine the best fit IMAX.
>> - Take those with chisq < chisq_min + 3 to determine the error.
>>
>> One could use these plots in a slightly different way. For example,
>> the
>> distribution for 00fr is highly assymetric and the I_max that has been
>> determined from the best fit local SNe is brighter than the I_max of
>> other SNe.
>>
>> As an alternative to what you have done, you can
>>
>> - Take the fits with chisq < chisq_min+3 to determine the mean or
>> median
>> I_max. - Use these same SNe to determine the error bars, which will be
>> assymmetric. The error bars can be determined from the 1st and 4th
>> quartiles.
>>
>> I do not think that this will change your results significantly, but
>> the
>> estimation of the maximum light will be more robust as it is based
>> average fit of several SNe and not a single one.
>
> I have now updated the plot in the web page:
>
> http://www.physto.se/~snova/internal/papers/iband/figs/read_fitlog.ps
>
> The bottom plots for each SN are now extended to include all the
> templates
> up to chisq_min+3. As you can see there, the strong asymmetry evident
> in
> the overall plot (top one) including all the fits, is not as clear
> when
> looking at the bottom plot.
>
> Moreover given the large difference in chisq between the best fit one
> and
> the next (at least for 2000fr and 1999ff), I do not think is really
> fair
> to average together the results up to chisq_min+3.
> As you said, this will not change the results anyway, but I don't think
> averaging results from "worse" fits, would give a more robust estimate
> of
> the maximum.
>
I'm not sure and I'd like to explore this point further. Ask yourself
the following question. Why would you use one sample to compute the
first moment of a distribution and then a larger sample to compute a
higher order moment. This is what you have done. You have used one
point (the one with the lowest chi-sqaure) to compute the "average",
but you use a larger sample (the ones with chisq_min+3) to compute the
RMS. I'd argue that you should the same sample to do both. I.e
compute the mean, median, standard deviation) from the same sample.
>
>> The SOFI J band filter is different from the ISAAC Js filter. I have
>> a copy of the filter transmission if you need it.
>>
>
> Discussing this with Ariel, it turned out that we are using the Keck J
> band also for the one epoch observed with SOFI instead. Can you send us
> the SOFI filter so that we can check the difference in the
> K-corrections?
>
>
I've attached the SOFI J band filter. Note that the red edge of the SOFI
filter is defined by the atmosphere, so you'll have to use the filter
curve and a model of the Earth's atmosphere to compute K corrections.
The latter is also attached.
> Hi Chris,
>
> one thing I forgot to mention in my previous email is the Hubble
> diagram
> for the high-z SNe. At some point you asked me to have it with all the
> SNe, but 1999Q, stretch corrected. I put this plot in the web page
> now.
> The r.m.s. for the high-z SNe does not improve significantly, since
> one of
> them (99Q) will still be deviant. Note that the fit of script M with
> and
> without stretch correction is different, e.g. the models in the two
> plots
> are shifted.
> I still think it is better the way it is now in the paper, i.e not
> stretch corrected for all the SNe, especially since we do not reach
> any
> conclusions about the cosmology. This is just used to show the method.
> I hope you agree with me.
> Cheers
>
>
> Serena
>
I like the Stretch corrected plot better. As you point out, we are not
doing cosmology with the plot, so why not show it?
Here are some other random comments.
Section 3.
You have calculated the dispersions of the CFA, CfA2 and Calan.Tololo
samples separately. Have these been calculated before or after the
correction for B-band stretch? One infers from the text that they are
calculated after correction. You should explicitly state this in
the text. It may also be useful to see both the corrected and
uncorrected dispersions of both samples in table 4.
Section 4.
Perhaps you can state why the 4th high-z SN has been rejected
from the analysis.
Section 4.1.
Rather than saying that the errors are dominated by the uncertainty in
the ZP, I think the following is more direct.
- the uncertainty in the k-correction is dominated by the uncertainty
in relating IR and optical photometric systems, which we estimate to be
of the order of 0.05 magnitudes.
It may be worth concentrating the discussion about the systematic errors
in the photometry of high-z SNe into one section, perhaps section 6.1.
Currently, it is spread throughout the paper - section 4.1, 5.1 and 6.1.
Section 5.1
The paragraph which starts with "A general problem ..." is a bit vague.
I will think about this and I'll come up with an alternative.
Section 6.1
The second point of 99bf is on the mean B-I curve and not below it.
For a readable explanation on how the average luminosity of a type Ia
might vary with redshift, I encourage you to read K. Nomoto's
article that was presented at last years conference in Garching. The
title of the book is "From twilight to highlight: The physics of
SNe" - ESO Astrophysics Symposia. There are two effects -
metallicity and age.
99Q - J=23.83 is an extremely faint target for SOFI. According to the
ETC, they would need a 25,000 second exposure. Some day, I am going
to grab the data from the archive and check their analysis.
Cheers, Chris.
This archive was generated by hypermail 2.1.4 : Mon Dec 15 2003 - 17:06:43 PST