Re: Low-z stability test

From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Fri Mar 14 2003 - 07:10:14 PST

  • Next message: Gerson Goldhaber: "Re: "Error floor""

    On Thu, Mar 13, 2003 at 08:58:03PM -0800, Alex Conley wrote:
    > sn1992bo -- solid. Rob has this as one that requires floating
    > offsets.

    This one always gets thrown out, so I haven't thought about it. The
    reason for using the floating offsets was the huge improvement in the
    chisquare: 116/37 - 29.6/35.

    > sn1993b -- pretty solid. Very minor (3rd decimal place) shifts in the
    > error bars when you tweak. Rob has this as an SNe which require
    > floating offsets, but I don't really see it. Nonetheless, fits IV and
    > V show what floating offsets do. Floating offsets make a substantial
    > difference

    The reason for using the floating offests on this was the twofold.
    First, the improvement in chi2/dof. Second, those late time points in Vhad
    a systematically high residual; see:

        http://brahms.phy.vanderbilt.edu/~rknop/scp/hst/#fltzero-xpoints

    > sn1992bp -- solid. Another floating offset. This time the effect is a
    > little larger. As before I-III are fixed, IV and V are floated.

    Note that the affect on the colors is fairly important here.

    > sn1996bo -- solid, both fixed and floating. However, the floating fit is
    > worse. It seems to miss the peak. Again, using floating offsets has
    > ruined a perfectly good fit.

    Again, it was a matter of the chi2/dof improving substantially, from
    890/22 to 316/20. Still bad, but that first "fixed-0" fit I find very
    difficult to describe as "perfectly good" given the change in the chi2.

    This one also gets thrown out from all fits, and is hugely host galaxy
    reddened.

    ---
    

    In general, where I saw an alarmingly large chisquare, I went back and asked if a floating 0 would help it a lot. If it did, then I considered using the floating 0. If not, then I just stayed with the fixed 0.

    -Rob

    -- --Prof. Robert Knop Department of Physics & Astronomy, Vanderbilt University robert.a.knop@vanderbilt.edu



    This archive was generated by hypermail 2.1.4 : Fri Mar 14 2003 - 07:10:15 PST