Re: Alex's comments

From: Serena Nobili (serena@lpnhep.in2p3.fr)
Date: Thu Aug 26 2004 - 07:53:04 PDT

  • Next message: Alex Kim: "Re: Comments on July21 draft of I-band Hubble diagram paper."

    Hi Alex,

    thanks for your comments. I try to answers below to some of your
    questions. Please, let me know if something is still unclear
    Ciao

           Serena

    On Mon, 23 Aug 2004, Alex Kim wrote:

    > 2
    > You say "Implicitly, we have thus assumed that the rising part of the
    > lightcurve in the I-band is the same as in B-band". Technically, this
    > is true only if on average if s_I=1, which isn't quite true when I look
    > at Fig 6. The assumption is that the relative rise and decline are the
    > same.
    >

    I think this is a delicate point. I believe we have never claimed the
    stretch method to work in the I-band before. It is, therefore, not obvious
    to me that the rise and decline are the same, nor that the s_I=1 has any
    meaning at this moment of the paper (in fact the relation between s_I and
    s_B is not well defined). The template used for fitting the I-band is a
    double B-band template (i.e. s_B=1), thus, I would assume this could
    correspond to a s_I=1. However, as I show later in the paper (e.g. fig6)
    the relation is not obvious at all. My intention with that sentence is to
    warn the reader that we are trying to describe the rise of I-band
    lightcurve with a B-band template, and that this could be not optimal, as
    in fact happens to be (see end of section 2.2).

    > 2.2
    >
    > Wondering why you didn't plot a histogram for t_Imax-t_sec. This may
    > give a tighter dispersion that t_Bmax
    >

    Unfortunately, it does not.

    > 2.3
    > SNe 97br and 98ab have the most plateau-like light curves. This may
    > have to do with why some Monte Carlo realizations the don't give good
    > chi-squared. Is there something obvious going on with the light curves
    > in these cases?
    >

    What it looks like from the MonteCarlo is that the first peak is not well
    constrained (there is only one point). Sometimes the fitting procedure
    tries to put the first peak to a too earlier epoch, and screw it up.
    Although this could be worsen by the "plateau-like light curves", the real
    problem is in the sampling of the lightcurve, I believe.

    >
    > 5.
    >
    > In Table 9, do the delta chi-squares convey any meaning in terms of
    > precision of the determination of Omega_M? The large delta chi-squares
    > makes it look like, contrary to your sentence, that the high-redshift
    > sample is sufficient to make strong conclusions on cosmological
    > parameters.
    >

    I am not sure what you mean by delta chi-squares.
    The reduced chisq in Table 9 are all very different from 1. The reason for
    that is mainly 1999Q which is quite far from any of the models considered.
    This, while making the chisq larger than the degrees of freedom, does
    push toward the concordance model, because is the model closer to 1999Q.
    My conclusion is simply that trying to make a statistical analysis based
    on three data points is extremely dangerous.

    > I am curious whether, in your sample of three supernovae, if there is
    > any correlation between the magnitude residuals from the concordance
    > cosmology of the high-z SNe and their corresponding low-z template.
    >

    I have tried to check for correlations in this sense. There is a
    correspondence only if the magnitudes are stretch corrected. In any case,
    I would not trust any conclusions on this, given the large systematic
    uncertainty on the fit of the high-z SNe.

    ----------------------------------------------------------------------
    www.physto.se/~serena
    Tel +33 1 44 27 75 82
                                                                                    
    Give free food at:
    http://www.porloschicos.com/
    http://www.thehungersite.com/
                                                                                    



    This archive was generated by hypermail 2.1.4 : Thu Aug 26 2004 - 07:53:28 PDT