more of U-B color dispersion

From: Greg Aldering (aldering@panisse.lbl.gov)
Date: Wed Apr 16 2003 - 00:13:46 PDT

  • Next message: Don Groom: "Re: further thougths on CMB vs. Helio redshifts"

    Earlier in this series I advocated using an intrinsic U-B color
    dispersion of 0.04 mag, as this was most consistent with our own
    observations of 10 z > 0.63 SNe Ia (5 with HST colors). This value is
    also consistent with Peter's fit to the U-B versus stretch relation
    from Jha's thesis after elimination of two 3-sigma outliers.

    Since then Peter has provided me with the data on the two 3-sigma
    outliers. One is too blue, and there is no SN in our HST sample as
    blue as that object. Accounting for our measurement errors, our SN with
    the bluest U-B (in residual space) is 2.3-sigma redder than this blue
    outlier of Jha's. Our z > 0.63 SN with the reddest U-B (again, in
    residual space) is within 1.6-sigma of being as red as the red Jha
    outlier. So, there is a 5% chance that such a SN is among our 5 SNe
    (i.e., the expectation is for 0.25 of such a SN in our sample).
    Furthermore, the incidence of such a red SN is suppressed by an
    additional factor of 1.4 due to flux selection effects. (It is
    unusually faint in U, so is less likely to be found at the limits of
    our searches.)

    From this, I conclude that within our HST sample there is no sign of
    color outliers like those in Jha's thesis. Given that the outliers in
    Jha's thesis are so extreme, they really are not part of a
    Gaussian-distributed distribution as the term "dispersion" would
    imply. Taken together, I don't see a justification for a citing a
    dispersion of 0.095 mag (i.e., the value when the outliers are
    included), and again conclude that 0.04 mag is a choice that best
    describes our data.

    Although the choice of U-B color dispersion doesn't much affect the
    OM,OL and OM,w fits, as Rob has shown, it does affect how well we can
    measure the mean extinction in our highest redshift SNe. In particular,
    using a U-B color dispersion of 0.095 versus 0.04 is statistically
    equivalent to reducing our sample size at the highest redshifts by a
    factor of 2 - quite a hit given the work and resources that went into
    making these measurements. I feel that constraining the amount of
    ordinary extinction in the highest redshift SNe is of interest,
    independent of the cosmology fit parameters.

    Cheers,

    Greg



    This archive was generated by hypermail 2.1.4 : Wed Apr 16 2003 - 00:14:13 PDT